EDITED BY : Nilesh B. Patel, Vivienne A. Russell and Nouria Lakhdar-Ghazal PUBLISHED IN : Frontiers in Neuroanatomy, Frontiers in Human Neuroscience, Frontiers in Pharmacology, Frontiers in Neurology, Frontiers in Neuroscience and Frontiers in Behavioral Neuroscience

#### Frontiers Copyright Statement

© Copyright 2007-2019 Frontiers Media SA. All rights reserved. All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.

The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.

Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.

Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.

As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.

All copyright, and all rights therein, are protected by national and international copyright laws.

The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use. ISSN 1664-8714 ISBN 978-2-88945-879-0 DOI 10.3389/978-2-88945-879-0

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

## Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

Frontiers in Neuroanatomy 1 June 2019 | Neuroscience in Africa

# NEUROSCIENCE IN AFRICA

Topic Editors: Nilesh B. Patel, University of Nairobi, Kenya Vivienne A. Russell, University of Cape Town, University of KwaZulu-Natal, South Africa Nouria Lakhdar-Ghazal, Mohammed V University, Morocco

Images:

"Africa outlined with neurons" - Giuseppe Bertini, University of Verona, Italy. "Elephant, Mara" - Nilesh B. Patel, University of Nairobi, Kenya.

This Research Topic covers some of the latest research on brain and behavior in health and disease in Africa. With its untapped resources and unique situations, "Neuroscience in Africa" has the potential to contribute to a better understanding of human brain function both in health and disease. The diverse African fauna display a range of specializations in brain structure/function relationships as a result of adaptations to the environment. Exploration of these may lead to insights into coping strategies which could be extrapolated to humans. Africa's unique flora is being investigated for anti-inflammatory, antinociceptive, antioxidant, antiepileptogenic and neuroprotective properties to determine its potential for use in the treatment of human brain disorders. There is also research on neurodegenerative and infectious diseases, not only common to the global world, but also neglected tropical diseases and conditions which provide unique avenues of investigations in basic and translational neuroscience on highly debilitating disorders - and on the effects of pathogens and environmental toxins.

Citation: Patel, N. B., Russell, V. A., Lakhdar-Ghazal, N., eds. (2019). Neuroscience in Africa. Lausanne: Frontiers Media. doi: 10.3389/978-2-88945-879-0

# Table of Contents

## CHAPTER 1

REVIEWS


Sharon L. Juliano and Julius J. Lutwama

## CHAPTER 2

## COMPARATIVE NEUROANATOMY AND NEUROPHYSIOLOGY


Adhil Bhagwandin, Mark Haagensen and Paul R. Manger

*56 Putative Adult Neurogenesis in Old World Parrots: The Congo African Grey Parrot* (Psittacus erithacus) *and Timneh Grey Parrot* (Psittacus timneh)

Pedzisai Mazengenya, Adhil Bhagwandin, Paul R. Manger and Amadi O. Ihunwo

*71 Sociality Affects REM Sleep Episode Duration Under Controlled Laboratory Conditions in the Rock Hyrax,* Procavia capensis Kuan-Yu Lin and Na-Sheng Lin

## CHAPTER 3

## ETHNOPHARMACOLOGY


Antoine K. Kandeda, Germain S. Taiwe, Fleur C. O. Moto, Gwladys T. Ngoupaye, Gisele C. N. Nkantchoua, Jacqueline S. K. Njapdounke, Jean P. O. Omam, Simon Pale, Nadege Kouemou and Elisabeth Ngo Bum


Houria Manouze, Otmane Bouchatta, A. Chemseddoha Gadhi, Mohammed Bennis, Zahra Sokar and Saadia Ba-M'hamed

## CHAPTER 4

## PARASITIC AND INFECTIOUS DISEASES


Shayne Mason, Carolus J. Reinecke and Regan Solomons


Jadrana T. F. Toich, Paul A. Taylor, Martha J. Holmes, Suril Gohel, Mark F. Cotton, Els Dobbels, Barbara Laughton, Francesca Little, Andre J. W. van der Kouwe, Bharat Biswal and Ernesta M. Meintjes

*196 Perinatal HIV Infection or Exposure is Associated With Low*  N*-Acetylaspartate and Glutamate in Basal Ganglia at Age 9 but Not 7 Years*

Frances C. Robertson, Martha J. Holmes, Mark F. Cotton, Els Dobbels, Francesca Little, Barbara Laughton, André J. W. van der Kouwe and Ernesta M. Meintjes

*206 Larger Subcortical Gray Matter Structures and Smaller Corpora Callosa at Age 5 Years in HIV Infected Children on Early ART* Steven R. Randall, Christopher M. R. Warton, Martha J. Holmes, Mark F. Cotton, Barbara Laughton, Andre J. W. van der Kouwe and Ernesta M. Meintjes

*218 Working Memory Profiles in HIV-Exposed, Uninfected and HIV-Infected Children: A Comparison With Neurotypical Controls* Robyn Milligan and Kate Cockcroft

## CHAPTER 5

## ALCOHOL

*231 Reductions in Corpus Callosum Volume Partially Mediate Effects of Prenatal Alcohol Exposure on IQ*

Stevie C. Biffen, Christopher M. R. Warton, Nadine M. Lindinger, Steven R. Randall, Catherine E. Lewis, Christopher D. Molteno, Joseph L. Jacobson, Sandra W. Jacobson and Ernesta M. Meintjes

*243 Altered Parietal Activation During Non-symbolic Number Comparison in Children With Prenatal Alcohol Exposure* Keri J. Woods, Sandra W. Jacobson, Christopher D. Molteno,

Joseph L. Jacobson and Ernesta M. Meintjes

*257 Changes in the Cholinergic, Catecholaminergic, Orexinergic and Serotonergic Structures Forming Part of the Sleep Systems of Adult Mice Exposed to Intrauterine Alcohol*

Oladiran I. Olateju, Adhil Bhagwandin, Amadi O. Ihunwo and Paul R. Manger

## CHAPTER 6

## PARKINSON'S DISEASE, DEMENTIA, AND BRAIN TRAUMA


Mounia Rahmani, Maria Benabdeljlil, Fouad Bellakhdar, Mustapha El Alaoui Faris, Mohamed Jiddane, Khalil El Bayad, Fatima Boutbib, Rachid Razine, Rachid Gana, Moulay R. El Hassani, Nizar El Fatemi, Meryem Fikri, Siham Sanhaji, Hennou Tassine, Imane El Alaoui Balrhiti, Souad El Hadri, Najwa Ech-Cherif Kettani, Najia El Abbadi, Mourad Amor, Abdelmjid Moussaoui, Afifa Semlali, Saadia Aidi, El Hachmia Ait Benhaddou, Ali Benomar, Ahmed Bouhouche, Mohamed Yahyaoui, Abdeslam El Khamlichi, Abdessamad El Ouahabi, Rachid El Maaqili, Houyam Tibar, Yasser Arkha, Adyl Melhaoui, Abdelhamid Benazzouz and Wafa Regragui

*293 Both Reaction Time and Accuracy Measures of Intraindividual Variability Predict Cognitive Performance in Alzheimer's Disease* Björn U. Christ, Marc I. Combrinck and Kevin G. F. Thomas

*304 Non-Motor Symptoms of Parkinson's Disease and Their Impact on Quality of Life in a Cohort of Moroccan Patients* Houyam Tibar, Khalil El Bayad, Ahmed Bouhouche,

El Hachmia Ait Ben Haddou, Ali Benomar, Mohamed Yahyaoui,

Abdelhamid Benazzouz and Wafa Regragui

	- Rufus O. Akinyemi, Louise M. Allan, Arthur Oakley and Rajesh N. Kalaria

Robyn Human, Michelle Henry, W. Jake Jacobs and Kevin G. F. Thomas

## CHAPTER 7

## ENVIRONMENTAL POLLUTANTS


Mariam Sabbar, Claire Delaville, Philippe De Deurwaerdère, Nouria Lakhdar-Ghazal and Abdelhamid Benazzouz

*430 Prenatal Exposure to Paint Thinner Alters Postnatal Development and Behavior in Mice*

Hanaa Malloul, Ferdaousse M. Mahdani, Mohammed Bennis and Saadia Ba-M'hamed

# Notes on the Recent History of Neuroscience in Africa

#### Vivienne A. Russell 1,2 \*

<sup>1</sup>Department of Human Biology, Faculty of Health Sciences, University of Cape Town, Observatory, Cape Town, South Africa, <sup>2</sup>School of Laboratory Medicine and Medical Sciences, College of Health Sciences, University of KwaZulu-Natal, Durban, South Africa

Neuroscience began with neuroanatomy and neurosurgery in Egypt more than 5000 years ago. Knowledge grew over time and specialized neurosurgery centers were established in north Africa in the eleventh century. However, it was not until the twentieth century that neuroscience research became established in sub-Saharan Africa. In most African countries, clinical research focused on understanding the rationale and improving treatment of epilepsy, infections, nutritional neuropathies, stroke and tumors. Significant advances were made. In the twenty-first century, African knowledge expanded to include all branches of neuroscience, contributing to genetic, biochemical and inflammatory determinants of brain disorders. A major focus of basic neuroscience research has been, and is, investigation of plant extracts, drugs and stress in animal models, providing insight and identifying potential novel therapies. A significant event in the history of African neuroscience was the founding of the Society of Neuroscientists of Africa (SONA) in 1993. The International Brain Research Organization (IBRO) supported SONA conferences, as well as workshops and neuroscience training schools in Africa. Thanks to their investment, as well as that of funding agencies, such as the National Institutes of Health (NIH), International Society for Neurochemistry (ISN), World Federation of Neurosurgical Societies (WFNS), World Federation of Neurology (WFN) and the International League Against Epilepsy (ILAE), neuroscience research is well-established in Africa today. However, in order to continue to develop, African neuroscience needs continued international support and African neuroscientists need to engage in policy and decision-making to persuade governments to fund studies that address the unique regional needs in Africa.

#### Edited by:

Jackson Cioni Bittencourt, University of São Paulo, Brazil

#### Reviewed by:

Marina Bentivoglio, University of Verona, Italy Lazaros C. Triarhou, University of Macedonia, Greece

> \*Correspondence: Vivienne A. Russell vivienne.russell@uct.ac.za

Received: 30 August 2017 Accepted: 16 October 2017 Published: 07 November 2017

#### Citation:

Russell VA (2017) Notes on the Recent History of Neuroscience in Africa. Front. Neuroanat. 11:96. doi: 10.3389/fnana.2017.00096 Keywords: brain, nervous system, neuroanatomy, neurosurgery, neurology, neuropsychiatry, neurophysiology

## INTRODUCTION

The earliest evidence of neuroscience research dates back more than 5000 years. Egyptian embalmers were the first to obtain knowledge of human anatomy through the mummification process. However, they had very little regard for the brain and did not try to preserve it (Cappabianca et al., 2007; Elhadi et al., 2012). The oldest record of neuroscience research, the Edwin Smith Papyrus, is believed to have been written around 2620 BC by the Egyptian physician and architect, Imhotep who was also a high priest of the sun god Ra (Elhadi et al., 2012). The Edwin Smith Papyrus reports traction as the first recorded neurosurgical procedure, used to reverse a paralyzing spinal injury in an Egyptian leader around 3000 BC (Filler, 2007). There is also evidence of early knowledge of the association between cerebral lesions and loss of movement on the contralateral side of the body, as well as fractures of the cervical spine being associated with neck rigidity, limb paralysis and conjugate eye deviation (El-Gindi, 2001). Trepanation, the process of creating a burr hole in the skull to access the brain, was widely used to relieve pressure after head injury, and is used to this day in the diagnosis and treatment of patients with traumatic brain injury in many African countries, especially those where modern techniques such as computerized tomography scans are not available (Eaton et al., 2017).

For more than 3000 years, Egypt was the center of knowledge of human brain structure and function. Early physicians studied neuroanatomy at the medical school in Memphis (Elhadi et al., 2012). However, religious conflicts restricted the study of human anatomy and myths replaced scientific research for centuries (Elhadi et al., 2012). It was not until Alexander the Great conquered Egypt in 332 BC and founded the city of Alexandria that significant advances were made (Elhadi et al., 2012). His conquest of the Persian Empire opened communication and promoted the exchange of knowledge across a world of previously hostile nations (Elhadi et al., 2012). A revolution in the study of functional anatomy followed. Great men such as Galen, Herophilus, Erasistratus and Rufus were prominent physicians who studied at the medical school in Alexandria (Elhadi et al., 2012). The Library of Alexandria became known for its impressive collection of recorded knowledge which attracted philosophers, scientists and teachers to Alexandria to study, debate and conduct scientific investigations (Elhadi et al., 2012). Egyptians were known for their skill in the practice of medicine. Their lively spirit of enquiry lead to discoveries in what was regarded as the science or philosophy of the day (Elhadi et al., 2012). These early neuroscientists carefully documented their detailed systematic dissections of human and animal nervous systems, tracing nerves to the brain, debating the function of nerve fibers originating in the brain stem and spinal cord, differentiating between sensory and motor nerves and attributing motor function to these nerves (Elhadi et al., 2012). Alexandria continued to be the center of neuroscience research until 30 BC when the Romans conquered Egypt and laws were passed that prohibited human dissections thereby preventing further progress in acquiring knowledge of the anatomy and physiology of the human nervous system for the next 1500 years (Elhadi et al., 2012).

## PROGRESS IN THE TWENTIETH AND TWENTY-FIRST CENTURY

Close proximity to Middle East and European training schools encouraged north African neuroscientists to further their studies and thereby contribute to the advancement of neuroanatomy and neurosurgery in these countries (El-Fiki, 2010). Knowledge grew over time and specialized neurosurgery centers were established in north Africa in the eleventh century (El Khamlichi, 1998). Modern neurosurgery was subsequently introduced to many countries in Africa in the twentieth century, mainly as a result of colonization by France and Britain (El Khamlichi, 1998). Departments of Neurosurgery and Neurology were established in African cities but these were initially staffed by foreigners (El Khamlichi, 1998). Unfortunately, the first African generation of neuroscientists who were trained in foreign countries did not stay in Africa (El-Fiki, 2010). They were unhappy with the lack of equipment and failure of existing equipment, largely due to poor maintenance, as well as the poor working conditions and absence of basic research facilities (El-Fiki, 2010). This situation began to change as some African neuroscientists returned to their home countries during the twentieth century and the African diaspora began to contribute significantly to the development of neuroscience research in Africa. Strong ties to foreign Universities helped to establish neuroscience research which developed rapidly in the latter part of the twentieth century. Spinal surgery, for example, advanced from decompression to spinal reconstruction and internal stabilization as a result of the introduction of computerized tomography and magnetic resonance imaging, and basic neuroscience research became established in many African countries (Loots et al., 1975; Shanley et al., 1975; Bengelloun et al., 1976; Wangai et al., 1978; Hattingh et al., 1979; Kimani and Mungai, 1983; Anderson et al., 1985; Nurse et al., 1985; Lakhdar-Ghazal et al., 1986; Bennis and Versaux-Botteri, 1995; McDonnell, 2004).

In most countries in Africa, clinical neuroscience research focused on understanding the rationale and improving treatment of neurological disorders, including epilepsy, infections (predominantly cerebral malaria, meningitis, encephalitis, poliomyelitis, leprosy, tetanus), nutritional neuropathies, stroke, tumors, motor disorders and the adverse effects of snake venom (Lambo, 1956; Osuntokun et al., 1968; Dada et al., 1969; Kramer et al., 1971; Wangai et al., 1978; Anderson et al., 1985; Tazir and Geronimi, 1990; Bhigjee et al., 1993; Vallat et al., 1993; Hentati et al., 1994; Ogunniyi, 2010). Significant advances were made. In the twenty-first century, African knowledge expanded rapidly to embrace all branches of neuroscience. Research covered genetic, biochemical and inflammatory determinants of brain disorders as well as basic neuroscience research on local fauna and animal models of brain disorders. An increasing focus of basic neuroscience research in Africa was in the field of neuropharmacognosy. Many countries have been, and still are, investigating the disease-modifying benefits of administering plant extracts to experimental animals (Ojewole and Amabeoku, 2006; Bum et al., 2009; Oboh et al., 2012; Eduviere et al., 2015; Akinrinmade et al., 2017; Manouze et al., 2017; Ngoupaye et al., 2017). African neuroscientists have the advantage of access to unique ecosystems of high biodiversity as well as critical knowledge of traditional medicine which has great potential to lead to the discovery of novel bioactive compounds (Karikari and Aleksic, 2015). African neuroscientists are also beginning to use invertebrate model organisms such as Drosophila melanogaster as powerful low-cost alternatives

to animal models for testing natural products which will strengthen their ability to identify bioactive plant extracts with therapeutic potential (Karikari and Aleksic, 2015; Akinyemi et al., 2017).

It is impossible to do adequate justice to all of the neuroscience research that has been carried out in Africa. This mini-review will therefore trace a few of the major trends based on a PubMed search of neuro-related publications by African neuroscientists in the eight African countries listed as having an H index of 20 or greater, in the SCImago Journal & Country Rank Report (2016)<sup>1</sup> . These countries have at least 20 neuroscience publications that have earned at least 20 citations each. They are, from north to south; Tunisia, Algeria, Morocco, Nigeria, Cameroon, Kenya, Tanzania and South Africa (**Figure 1**).

## NORTH AFRICA

In the twentieth and twenty-first centuries, neuroscience research in Tunisia and Algeria involved mostly clinical studies on neurogenetics and movement disorders (Tazir and Geronimi, 1990; Hentati et al., 1994; Gouider-Khouja et al., 2000; Younes-Mhenni et al., 2007). Clinical research in Morocco included stroke, Alzheimer's disease, addiction and fMRI studies of brain plasticity, to mention a few (El Kadmiri et al., 2014; Souirti et al., 2014; Mohamed and Kissani, 2015; Chtaou et al., 2016; Zarrouq et al., 2016; Boujraf et al., 2017). Basic neuroscience research was introduced to Morocco in the 1970s with the study of behavioral consequences of brain lesions and nutritional deficiency in rats (Bengelloun et al., 1976). In the 1980s and more recently, the focus of researchers across the country expanded to include not only brain lesions and malnutrition but also stress, drugs of abuse, neurotoxins, sensory systems and biological

<sup>1</sup> SCImago (2007) http://www.scimagojr.com/worldreport.php?area=2800

rhythms in laboratory animals and local fauna (Lakhdar-Ghazal et al., 1986; Bennis and Versaux-Botteri, 1995; Sansar et al., 2012; Said et al., 2015). Recent years have seen a marked increase in the number of publications by African neuroscientists (**Figure 2**).

## SUB-SAHARAN AFRICA

The earliest knowledge of neurological disorders in sub-Saharan Africa is attributed to Yoruba Traditional healers in Nigeria, in the seventeenth century (Ogunniyi, 2010). However, it was only in the middle of the 20th century that neuroscience research began to flourish in sub-Saharan Africa. In the 1950s, the first black African neuropsychiatrist introduced a community–based system of treatment of psychiatric patients in Nigeria, a form of treatment that remains relevant to this day (Lambo, 1956; Ogunniyi, 2010). Several articles describing disorders of the nervous system were published in the 1960s (Odeku, 1965; Osuntokun, 1968; Dada et al., 1969; Osuntokun et al., 1969b; Okubadejo et al., 2006; Ogunniyi et al., 2015). In 1969, the tropical ataxic neuropathy, konzo, was attributed to the cyanogenic glycosides present in cassava (Osuntokun et al., 1969a). In the 1970s and early 1980s, neuroepidemiological studies dominated tropical neurology (Osuntokun, 1978; Ogunniyi, 2010). More recently, research has focused on neuroprotective properties of indigenous plant extracts, neurotoxicity of environmental factors, neurogenomics, stroke, epilepsy and neurodegenerative diseases, including Parkinson's disease and dementia, amongst others (Okubadejo et al., 2006; Akinyemi et al., 2014, 2016; Lekoubou et al., 2014; Mkenda et al., 2016; Ekong et al., 2017; Folarin et al., 2017; Ilesanmi et al., 2017; Ojagbemi et al., 2017).

Neuropharmacognosy is an important component of basic neuroscience research in many countries in Africa, where native African traditional medicines are scientifically analyzed with a view to validating their benefits and thereby offer potential novel therapies (Amos et al., 2001; Ojewole and Amabeoku, 2006; Bum et al., 2009; Bisong et al., 2010; Ishola et al., 2013; Ogunniyi et al., 2015; Qulu et al., 2016; Adebesin et al., 2017; Elufioye et al., 2017; Ngoupaye et al., 2017).

In Kenya, the earliest studies on the nervous system were largely descriptive and focused on brain size rather than function (Vint, 1934). As in many African countries that were colonized by Britain, the development of modern neurosurgical procedures in Kenya occurred in the late 1940s, as a result of the two world wars (Qureshi and Oluoch-Olunya, 2010). The last quarter of the twentieth century saw neurosurgery develop to its present level, with Kenyan neurosurgeons engaged in both practice and research (Qureshi and Oluoch-Olunya, 2010). The earliest reports on basic neuroscience research in Kenya appeared in the late 1970s and early 1980s (Wangai et al., 1978; Kimani and Mungai, 1983; Anderson et al., 1985). More recently, research has focused on morphine-induced aggression and antinociceptive effects in the naked mole-rat, antinociceptive and anti-inflammatory effects of plant extracts in mice as well as the mechanism of action of cathinone (active ingredient of khat, Catha edulis) an addictive psychostimulant grown in abundance in Kenya and impairment of executive function in children with malaria, to name a few (Kanui and Hole, 1990; Kanui et al., 1993; Patel, 2000, 2015; Kariuki et al., 2013, 2014; Jørgensen et al., 2016; Kimani et al., 2016). In Tanzania and Cameroon, a dominant focus of neuroscience research has been epilepsy, infectious diseases, stroke and assessment of the potential therapeutic value of indigenous plant extracts (Matuja et al., 2001; Njamnshi et al., 2006, 2009, 2012; Bum et al., 2009; Levira et al., 2017; Ngoupaye et al., 2017).

The earliest publication on the nervous system, from South Africa, examined the embryonic history of the segmented mesoderm and neural tube (Dart, 1924). Basic and clinical neuroscience research emerged in the 1950s and 1960s with the neurophysiology of the spinal cord and research on Cannabis sativa (Ames, 1958; Holemans et al., 1966). In the 1970s research expanded to include neurological disorders, including porphyria, psychiatric disorders, nociception and thermoregulation in different animal species (Kramer et al., 1971; Loots et al., 1975; Shanley et al., 1975; Woolf et al., 1977; Hattingh et al., 1979). In the 1980s and more recently, in collaboration with other countries in Africa, attempts were made to understand the success of traditional healers in the treatment of patients with psychiatric disorders (Wessels, 1985; Gureje et al., 2015). Research has expanded in all branches of neuroscience to include neuroimaging studies of e.g., children with fetal alcohol spectrum disorder which is highly prevalent in the Western Cape region of South Africa, as well as neurodevelopmental outcomes of children with tuberculous meningitis and hydrocephalus, which is also highly concentrated in this region, as well as neuropsychiatric genomics, psychiatric disorders such as post-traumatic stress disorder, HIV-associated neurocognitive disorders, schizophrenia, bipolar disorder, drug addiction, the neuropsychology of emotional experience, neurology, neurosurgery, functional neuroanatomy including neurogenesis and understanding neural disturbances in animal models of brain disorders, to name a few (Howells et al., 2016; Qulu et al., 2016; Rohlwink et al., 2016; Sterley et al., 2016; van Wyk et al., 2016; Dallé et al., 2017; Kilian et al., 2017; Mazengenya et al., 2017; Panksepp et al., 2017; Uys et al., 2017; Womersley et al., 2017; du Plessis et al., 2017).

This mini-review cannot provide a comprehensive overview but seeks to highlight some of the research carried out in those countries in Africa that have achieved recognition in terms of the number of neuroscience publications that have been cited by other researchers, according to the SCImago Journal & Country Rank Report (2016)<sup>2</sup> . However, there is no mention of all the excellent research that has been, and is being carried out in countries not listed here, for example, research on konzo in the Democratic Republic of Congo, nodding syndrome and cerebral malaria in Uganda, to mention a few (Bumoko et al., 2015; Bangirana et al., 2016; Idro et al., 2016).

## TRAINING IN NEUROSCIENCE IN AFRICA

Over the years, science-based non-profit organizations such as the International Brain Research Organization (IBRO) have supported African neuroscientists and helped to build scientific capacity for sustainable education and research by sponsoring training courses and workshops across Africa. Training programmes in Neurosurgery, Neurology and basic Neuroscience, supported by, the World Federation of Neurosurgical Societies (WFNS), the World Federation of Neurology (WFN), the Society of French Speaking Neurosurgeons (SNCLF) the International League Against Epilepsy (ILAE), the National Institutes of Health (NIH), IBRO and the International Society of Neurochemistry (ISN), amongst others, have led to the production of many generations of neuroscientists in Africa. IBRO and ISN, in particular, have engaged in basic neuroscience with a strong translational approach to brain disorders in Africa. IBRO African Centres for Advanced Training in Neuroscience have been established in Morocco and South Africa.

<sup>2</sup>http://www.scimagojr.com/worldreport.php?area=2800

## REFERENCES


WFNS-recognized Centres of Excellence for training neurosurgeons have been established in Nigeria, Ivory Coast, Senegal, Kenya, Zimbabwe and South Africa (Dechambenoit, 2010).

A significant event in the history of African neuroscience is the founding of the Society of Neuroscientists of Africa (SONA) by James Kimani in 1993. IBRO generously supported SONA conferences, as well as several workshops and neuroscience training schools in more than 14 African countries. Thanks to their investment, as well as that of the other funding agencies, neuroscience research is well-established in Africa. However, in order to continue to develop, African neuroscience needs continued international support and African neuroscientists need to engage in policy and decision-making to persuade governments to fund studies of unique regional needs in Africa (Bentivoglio et al., 2014). Basic neuroscience is not a luxury but a means to address the challenges of specific needs in disease-endemic regions (addiction, toxic pollutants, infections) by investigating the pathogenic mechanisms (in neglected tropical diseases and conditions) and taking advantage of unique models (the brain and behavior of African fauna) and unique populations (children with fetal alcohol spectrum disorder) to better understand adaptations to environmental conditions and susceptibility to disease (Bentivoglio et al., 2014). For instance, a better understanding of the nervous system of the mosquito or the tsetse fly can lead to better methods to control these disease vectors (Kristensson et al., 2010; Sparks and Dickens, 2016). African problems need African solutions, goal-directed basic neuroscience research is needed in Africa where there is a dire need to combat regionally-specific devastating brain disorders (Bentivoglio et al., 2014).

## AUTHOR CONTRIBUTIONS

VAR wrote the mini-review.

## ACKNOWLEDGMENTS

The author would like to thank Professors Nouria Lakhdar-Ghazal, Adesola Ogunniyi and Nilesh Patel for their very helpful comments and their valued information on countries in north, west and east Africa.

African stroke survivors: the CogFAST Nigeria study. J. Neurol. Sci. 346, 241–249. doi: 10.1016/j.jns.2014.08.042


of oxidative stress and acetylcholinesterase activity in mice. Life Sci. 132, 20–26. doi: 10.1016/j.lfs.2015.04.007


(Heterocephalus glaber). J. Comp. Physiol. A Neuroethol. Sens. Neural Behav. Physiol. 202, 7–15. doi: 10.1007/s00359-015-1048-x


experience. Neurosci. Biobehav. Rev. 76, 187–215. doi: 10.1016/j.neubiorev. 2016.09.010


**Conflict of Interest Statement**: The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Russell. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Angelina Kakooza-Mwesige1 \*, Abdul H. Mohammed2 , Krister Kristensson3 , Sharon L. Juliano4 and Julius J. Lutwama5*

*1Department of Paediatrics and Child Health, Makerere University College of Health Sciences and Mulago Hospital, Kampala, Uganda, 2Department of Psychology, Linnaeus University, Växjö, Sweden, 3Department of Neuroscience, Karolinska Institutet, Stockholm, Sweden, 4Anatomy, Physiology and Genetics, Uniformed Services University of the Health Sciences, Bethesda, MD, United States, 5Arbovirology Laboratory, Uganda Virus Research Institute, Entebbe, Uganda*

The global public health concern is heightened over the increasing number of emerging viruses, i.e., newly discovered or previously known that have expanded into new geographical zones. These viruses challenge the health-care systems in sub-Saharan Africa (SSA) countries from which several of them have originated and been transmitted by insects worldwide. Some of these viruses are neuroinvasive, but have been relatively neglected by neuroscientists. They may provide experiments by nature to give a time window for exposure to a new virus within sizeable, previously non-infected human populations, which, for instance, enables studies on potential long-term or late-onset effects on the developing nervous system. Here, we briefly summarize studies on the developing brain by West Nile, Zika, and Chikungunya viruses, which are mosquito-borne and have spread worldwide out of SSA. They can all be neuroinvasive, but their effects vary from malformations caused by prenatal infections to cognitive disturbances following perinatal or later infections. We also highlight Ebola virus, which can leave surviving children with psychiatric disturbances and cause persistent infections in the non-human primate brain. Greater awareness within the neuroscience community is needed to emphasize the menace evoked by these emerging viruses to the developing brain. In particular, frontline neuroscience research should include neuropediatric follow-up studies in the field on long-term or late-onset cognitive and behavior disturbances or neuropsychiatric disorders. Studies on pathogenetic mechanisms for viral-induced perturbations of brain maturation should be extended to the vulnerable periods when neurocircuit formations are at peaks during infancy and early childhood.

Keywords: developing nervous system, emerging viruses, Ebola virus, Chikungunya virus, West Nile virus, Zika virus, neurological disorders, sub-Saharan Africa

## INTRODUCTION

Exposure to infections during the first part of fetal life, the so-called teratogenic window, can cause severe brain malformations. To the established human neuroteratogenic pathogens (*Toxoplasma gondii*, Other pathogens, Rubella, Cytomegalovirus, and Herpes simplex virus; TORCH), Zika virus (ZIKV) is now added (1–3). Perinatal, infant, and childhood infections may also disturb the developing

#### *Edited by:*

*Nilesh Bhailalbhai Patel, University of Nairobi, Kenya*

#### *Reviewed by:*

*Richard S. Nowakowski, Florida State University College of Medicine, United States Berlin L. Londono-Renteria, Kansas State University, United States*

#### *\*Correspondence:*

*Angelina Kakooza-Mwesige akakooza246@gmail.com, angelina\_kakooza@yahoo.co.uk*

#### *Specialty section:*

*This article was submitted to Neurodegeneration, a section of the journal Frontiers in Neurology*

*Received: 20 September 2017 Accepted: 05 February 2018 Published: 23 February 2018*

#### *Citation:*

*Kakooza-Mwesige A, Mohammed AH, Kristensson K, Juliano SL and Lutwama JJ (2018) Emerging Viral Infections in Sub-Saharan Africa and the Developing Nervous System: A Mini Review. Front. Neurol. 9:82. doi: 10.3389/fneur.2018.00082*

**Abbreviations:** CHIKV, Chikungunya virus; CNS, central nervous system; CSF, cerebrospinal fluid; EVD, Ebola virus disease; SSA, sub-Saharan Africa; WNV, West Nile virus; ZIKV, *Zika* virus.

nervous system during the peaks of neurocircuit formations and possibly cause more subtle changes in brain maturation. Thus, cognitive impairments and behavioral disturbances in children born to HIV-infected mothers (4) or subjected to childhood malaria (5) have been described in SSA; unprovoked late-onset epilepsy may also occur following the latter infection (6).

Of particular concern to African neuroscience are emerging viral infections. They can reveal associations between infections and rare sequelae in human populations, as poliomyelitis once did for "infantile paralysis" (7). Early life viral infections may, in one way or the other, be implied in the pathogenesis of cognitive and neuropsychiatric disturbances: the "neurodevelopmental hypothesis" for late-onset brain dysfunctions [e.g., Ref. (8)]. By comparing four neuroinvasive infections originating from sub-Saharan Africa (SSA), i.e., ZIKV, West Nile virus (WNV), Chikungunya virus (CHIKV), and Ebola virus (EBOV), we find that they attack the human brain at various stages of development. Thus, time windows for viral invasions, given by occurrence of the emerging epidemics, may reveal unique associations between viral infections and late-onset human brain disturbances.

## DISCOVERY OF THE VIRUSES AND THE MAGNITUDE OF THE PROBLEM

The emerging viruses dealt with in this review are indigenous to and were first identified in Africa (**Figure 1**). WNV was isolated from a woman, who had a mild febrile illness in the West Nile Region of Uganda in 1937 (9). ZIKV was first isolated from a Rhesus monkey, placed on a platform as bait for mosquitoes in studies on yellow fever in the Zika forest, in Uganda in 1947 (10). CHIKV was first isolated in 1953 at the Uganda Virus Research Institute from samples collected in Tanganyika (11). In 1976, the investigation of concurrent outbreaks of a hemorrhagic fever syndrome in Zaire (currently Democratic Republic of Congo) and Sudan (currently Republic of South Sudan) led to isolation of two viruses now referred to as EBOV and Sudan virus, respectively. Another member of the EBOV, Bundibugyo Virus was identified in Uganda (12) (**Figure 1**).

All these viruses, which are endemic in tropical SSA, have had outbreaks also in other African regions, e.g., WNV in north and west Africa, in South Africa, and in Madagascar during the last decade (13). Currently, there is an outbreak of CHIKV in Kenya. The mosquito-borne WNV, ZIKV, and CHIKV have spread worldwide out of Africa. For instance, WNV is the most common cause of encephalitis in the US, enhanced by the bird reservoir hosts (14). CHIKV has caused extensive epidemics on islands in the Indian Ocean, in particular, La Reunion (15–17). The spread of these emerging viruses into previously unaffected regions can be attributed to more frequent and distant travels, evolution of mutant viral strains with altered virulence, and changes in climate and local ecosystems.

## SENSORY CUES IN MOSQUITO–HUMAN INTERACTION

Neuroscientific research is important to design specific, repellant molecules that reduce mosquito's attraction to humans. Three of

#### Figure 1 | The map of Uganda showing the sites and times for the discoveries of the emerging viral infections by Ebola, West Nile, and Zika viruses. Chikungunya virus (not depicted on this map) is also widespread in Uganda. It was first isolated at the Uganda Virus Research Institute in Entebbe from blood samples obtained from Tanzania (Tanganyika territory). The various viral outbreaks are denoted by the stars in different colors, while the respective country locations are indicated by the small black circles.

the emerging infections discussed are mosquito-borne, i.e., *Aedes aegypti* for CHIKV and ZIKV and *Culex pipiens* for WNV; *Aedes albopictus* can also host ZIKV (18). Neurophysiological and behavioral studies have revealed *A. aegypti* attraction cues emanating from humans [e.g., Ref. (19)]. Odor is one important cue of insect host-seeking behavior [e.g., Ref. (20)]. The mosquito's olfactory receptor neurons can detect various odor volatiles from the human host, which make some people more attractive than others to mosquitoes. Carbon dioxide (21), lactic acid (22), and visual cues (23, 24) can also attract mosquitoes to the host. Neuroanatomical studies have uncovered the organizations of olfactory centers in the mosquito brain (25), and multimodal integration of carbon dioxide, body odor, and temperature cues in the mosquito brain can enhance mosquito attraction to humans (26). Novel strategies that specifically block more than one sensory cue for attraction of insects are envisioned to contribute to control of emerging viral spread and outbreaks.

## ENTRY ROUTES FOR VIRUSES TO THE BRAIN: EFFECTS OF NEUROINVASIVE VIRAL INFECTIONS DURING DEVELOPMENT

Following the sting of an infected insect, neurotropic viruses enter through the skin, replicate in susceptible cells, and spread to the central nervous system (CNS) *via* peripheral nerve fibers and/or the bloodstream (27). WNV can directly spread by the former route (28), while viral spread *via* the bloodstream into the brain is hampered by barriers, i.e., the blood–CSF barrier and the blood–brain barrier (BBB), established early during embryonic life (29). Nevertheless, certain viruses pass across these barriers and target neural cells. Of the presently described viruses, the BBB may be crossed by WNV (14), ZIKV (see below), and probably also EBOV (30), while CHIKV can pass the blood–CSF barriers through the permeable choroid plexus to infect ependymal cells in a mouse model (31).

During pregnancy, viruses can enter the fetal brain through the bloodstream following placental transmission. Importantly, the placental structure as well as its expression of receptors used by various viruses and its innate immune response differ markedly between humans, mice, and ruminants, which hinders direct comparisons among animal species (3). Notably, human fetal immune adaptations provide early postnatal protection against extracellular pathogens but enhance the risk of virus-induced persistent infections (32).

The effects of an intrauterine infection depend on timing of the insult with highest vulnerability during peaks in neuronal cell proliferation and migration (33). In humans, simplified gyral patterns may result from disturbed cell migration at 8–16 weeks of gestation. While congenital microcephaly (reduced head circumference) is associated with various insults causing disturbed proliferation, migration, or destruction of cells at various periods of intrauterine life (34). Since the human brain increases about 3.5 times in weight after birth, when neuronal network formation and myelination are at their peaks, the question can be posed whether infant/childhood infections may cause less conspicuous, functional disturbances of the developing nervous system than malformations as long-term or late-onset effects [**Figure 2A**; (29)]. In fact, a register-based study has indicated that children with severe viral CNS infections at 0.5–8 years of age show enhanced risk of psychotic illness when they reach young adulthood (35).

## CLINICAL EFFECTS OF THE EMERGING VIRAL INFECTIONS ON THE DEVELOPING NERVOUS SYSTEM

## Zika Virus

Zika virus infections have been endemic for decades in SSA (36), but effects on the brain were not reported until the virus spread out of Africa. The infection is most often asymptomatic or associated with mild signs of disease such as maculopapular rash and non-purulent conjunctivitis (37). Ultrasonographic studies in Brazil of ZIKV-infected pregnant women carrying fetuses diagnosed with microcephaly, showed in addition, severe progressive ventriculomegaly, periventricular and basal ganglia calcifications, corpus callosal dysgenesis as well as posterior fossa abnormalities [e.g., Refs. (38, 39)]. Neuropathological findings in 11 newborns with congenital microcephaly include small brains with almost complete agyria, neuronal heterotopia, enlarged ventricles with or without aqueduct stenosis, and well-formed brains with calcifications (40, 41). Only two cases of perinatal infections, discharged in good health, and a limited number of children infected *via* mosquitoes, have been reported; they were mildly infected or asymptomatic-like adults (42).

Recent neuroimaging reports of infants with postnatal microcephaly or normal skull size, but with other signs of congenital Zika syndrome, indicate a spectrum with less severe brain changes (43, 44). Whether the virus can persist in the human infant brain during postnatal development remains to be clarified. Related to this, a critical question is whether disturbances in cognitive development may appear later in life even following asymptomatic infections in newborn, which may occur in the majority of children borne to ZIKV-infected mothers [cf. Ref. (45)]. Currently, no information is available on ZIKV-related nervous system disturbances in SSA.

## West Nile Virus

Most WNV infections are asymptomatic, but some are associated with flu-like illness and maculopapular rash. Less than 1% are neuroinvasive causing meningitis and encephalitis, which increases in incidence by age (14, 46); flaccid paralysis, movement disorders with tremor, and myoclonus are other rare complications (47). Adult patients recovering from WNV encephalitis can show low scoring in cognitive and other neuropsychological tests for various periods of time (48, 49).

Studies of effects of maternal WNV infections during pregnancy on early childhood development in SSA do not exist. Studies from the US initially indicated a risk of abnormalities such as microcephaly, while two later retrospective and prospective studies showed no signs of CNS malformations or short-term development disturbances when compared to controls (50, 51). However, the number of children in these studies was small (50), and the incidence of neuroinvasive WNV infections in children is probably underestimated (52, 53). More studies are essential including long-term follow-up to assess potential effects of neuroinvasive WNV disease during various neurodevelopmental vulnerable periods (54). For instance, it would be of interest to conduct research into disturbances in the development of cognitive functions in the reported several hundreds of children of different ages surviving this disease since 1999 (55).

## Chikungunya Virus

In contrast to the previous two arboviruses, the majority of CHIKV-infected individuals develop symptoms such as fever and arthralgia. Meningoencephalitis and fatal outcome are rare. Mother-to-child transmission of CHIKV is also a rare event that has been reported in large-scale outbreaks of the disease such as on La Reunion 2005–2006. Although maternal infections long before delivery showed no observable effects on the outcome and there is no evidence of congenital virus transmission (56), perinatal infections with nervous system involvement do occur (57). Of 30 neonates with acute neurological manifestations, 2 died and 5 showed abnormal MRI scans (high intensity in periventricular white matter and corpus callosum). Five had neurological sequelae at discharge and 6 months later. These sequelae included behavior and communication disorders, autism and echolalia, recurrent seizures; one child had microcephaly and strabismus (58). In another follow-up study from the island about 50% of perinatally infected children showed at 2 years of age delay in development of coordination and language skills as well as in sociability and movement performance. Five out of 12 newborns with neonatal encephalopathy developed postnatal microcephaly with severe reduction of the white matter visualized on MRI (59). More long-term consequences of these perinatal and childhood infections remain to be studied.

## Ebola Virus

Ebola virus, which has bats as intermediate hosts (60), causes rapidly progressive severe hemorrhagic fever with a very high lethality rate. Based on findings from Ebola virus disease (EVD) studies done in West Africa, complications occurring >10 days from disease onset include meningoencephalitis (61, 62). The particularly vulnerable patient populations include children <5 years of age, the elderly, and pregnant women.

Systematic longitudinal assessments of EVD survivors in Africa are scarce. However, new insights and understanding about the long-term effects of infection are currently being generated from the survivors of the largest ever epidemic of EVD to date that occurred in Guinea, Liberia, and Sierra Leone (63). Interim analysis of data from this multidisciplinary longitudinal study of 804 EVD survivors in Guinea (20% are children) reports that EVD survivors exhibit an array of neurological and psychiatric symptoms even after more than 1 year following discharge from the hospital (64). The majority of survivors experienced physical disorders such as psychosocial problems, depression, or ophthalmological problems.

Given the paucity of current data, there is need for systematic longitudinal assessments of EVD survivors to clarify the spectra of nervous system sequelae and the magnitude of the problem. Of paramount interest in this respect is that EVD can persist after recovery in the brain of non-human primates (65).

## OBTAINING AN EXPERIMENTAL MODEL OF ZIKV AND THE DEVELOPING NERVOUS SYSTEM

Spurred by the great awareness of ZIKV-induced microcephaly, data on molecular mechanisms underlying this teratogenic virus infection are now rapidly accumulating. Here, we briefly review some recent findings and indicate gaps-in-knowledge for neuroscience to fill not only for teratogenesis but also for potential postnatal developmental disturbances in this and the other emerging viral infections.

Nowakowski et al. (66) suggested that AXL could be a candidate as a receptor for endocytotic entry of ZIKV into the fetal brain. AXL is a member of the multifunctional TAM receptor protein tyrosine kinase family that among other things regulates phagocytosis and plays a pivotal role in innate immune responses (67). In the human fetal brain, AXL is expressed in neural stem cells that generate neurons populating the neocortex, radial glial cells, astrocytes, and endothelial cells [**Figure 2B**; (66)]. From an infected mother, ZIKV may have a privileged entry to the placenta as indicated by the finding that it infects human endothelial cells of the umbilical vein in culture by binding to AXL with a high efficiency compared to other flaviviruses, such as WNV (68). Additional evidence suggests that the ZIKV infects other cells of the placenta, e.g., macrophages and cytotrophoblasts (69). Chavali et al. (70) recently demonstrated that the ZIKV has a predilection in its RNA genome to bind to a specific RNAbinding protein (Musashi-1), which is liberally expressed in the radial glial cells. Since these progenitor cells are precursors for neurons and astrocytes that form the cerebral cortex (71), this binding capacity may directly influence formation of the cerebral cortex.

Additional experiments using various types of cell culture have been useful in furthering our knowledge of ZIKV infection, especially on neural progenitor cells. Direct infection of human tissue placed in culture shows special susceptibility of radial glial and epithelial stem cells, while demonstrating limited direct infection of neurons (72). Since human progenitor cells form three-dimensional organoids in culture, they demonstrate features of radial glial cells, which also show evidence of specific direct infection in culture in this model. Numerous other experiments using various cell culture models and direct infection of vertebrates also demonstrate involvement of radial glial-like cells and progenitor cells [for review, see Ref. (73)]. This apparent focus of infection specificity to neural progenitor cells may help to clarify the preferential effect of ZIKV on the fetus leading to microcephaly.

Development of animal models is clearly an essential route to more completely understand ZIKV effects on the brain and the nervous system. Several such models exist to study effects of ZIKV infections in offspring to pregnant animals, but none of them fully replicate the human situation [for review, see Ref.

(74)]. Viruses mostly need adaptations to a new animal species, the neurovirulence between viral strains differs, the placental immune response and receptors that viruses use differ between species (see, above), and even within the same species, such as mice, various strains differ markedly between their reactions to an infection. Nonetheless, offspring to ZIKV-infected immunedeficient mice and non-human primates have shown some abnormalities in the brains (74). Considerable efforts are needed for neuroscience–microbiology collaborations to develop suitable animal models of viral infections to disclose and validate not only neuroteratogenic viral effects but also more subtle alterations that may result from prenatal abortive or persistent infections.

## CONCLUSION AND PERSPECTIVES

We present evidence that various spectra of neurodevelopmental disturbances may be associated with the emerging viral infections: ZIKV with congenital microcephaly following intrauterine infections, but also postnatal brain alterations with or without microcephaly; CHIKV with postnatal microcephaly following perinatal infections, but also behavior and communication disturbances; WNV with cognitive dysfunctions following infections mostly in adults, but infant/childhood infections are underestimated and need further studies. EVD, which has flared up in waves in SSA, may also be a threat to the developing human nervous system. Important gaps in knowledge include risks of cognitive impairment, behavior disturbances, and neuropsychiatric diseases following infections during infancy and childhood with these emerging infections, when neuronal network formation and synaptogenesis are at the peak. The molecular mechanisms resulting in disrupted neural development need to be unraveled in more detail and the possibility addressed whether some of these emerging viruses can persist in the human brain.

Based on this situation, we foresee a number of activities to reduce the menace of these infections to nervous system maturation:

• Strengthening capacity of African virology and neuroscience research with enhanced worldwide collaboration to strongly benefit the community.


## REFERENCES


## AUTHOR CONTRIBUTIONS

The authors contributed expertise information and presented one section each at a symposium with the same title at the SONA conference in Entebbe, Uganda, June 11–14, 2017. All the authors approved the final version of the manuscript.

## FUNDING

AK-M was supported through the DELTAS Africa Initiative grant # DEL-15-011 to THRiVE-2. The DELTAS Africa Initiative is a funding scheme of the Accelerating Excellence in Science in Africa (AESA) with funding from the Wellcome Trust grant # 107742/Z/15/Z and the UK government.

pathogenesis and therapy. *Antiviral Res* (2013) 99:345–70. doi:10.1016/j. antiviral.2013.06.009


West Nile illness during pregnancy. *Birth Defects Res A Clin Mol Teratol* (2016) 106:716–23. doi:10.1002/bdra.23523


**Conflict of Interest Statement:** The authors declare that the writing of this article was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Kakooza-Mwesige, Mohammed, Kristensson, Juliano and Lutwama. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# The Suprachiasmatic Nucleus of the Dromedary Camel (Camelus dromedarius): Cytoarchitecture and Neurochemical Anatomy

Khalid El Allali<sup>1</sup> \*, Mohamed R. Achaâban<sup>1</sup> , Mohammed Piro<sup>2</sup> , Mohammed Ouassat<sup>1</sup> , Etienne Challet<sup>3</sup> , Mohammed Errami<sup>4</sup> , Nouria Lakhdar-Ghazal<sup>5</sup> , André Calas<sup>6</sup> and Paul Pévet<sup>3</sup>

<sup>1</sup> Comparative Anatomy Unit/URAC49, Department of Biological and Pharmaceutical Veterinary Sciences, Hassan II Agronomy and Veterinary Medicine Institute, Rabat, Morocco, <sup>2</sup> PMC-EC, Department of Medicine, Surgery and Reproduction, Hassan II Agronomy and Veterinary Medicine Institute, Rabat, Morocco, <sup>3</sup> Neurobiology of Rhythms UPR 3212 CNRS, Institute for Cellular and Integrative Neurosciences, University of Strasbourg, Strasbourg, France, <sup>4</sup> Department of Biology, Faculty of Science, Abdelmalek Essaâdi University, Tétouan, Morocco, <sup>5</sup> Unit of Research on Biological Rhythms, Neuroscience and Environment, Faculty of Science, Mohammed V-Agdal University, Rabat, Morocco, <sup>6</sup> IINS, CNRS UMR 5297, University of Bordeaux, Bordeaux, France

In mammals, biological rhythms are driven by a master circadian clock located in the suprachiasmatic nucleus (SCN) of the hypothalamus. Recently, we have demonstrated that in the camel, the daily cycle of environmental temperature is able to entrain the master clock. This raises several questions about the structure and function of the SCN in this species. The current work is the first neuroanatomical investigation of the camel SCN. We carried out a cartography and cytoarchitectural study of the nucleus and then studied its cell types and chemical neuroanatomy. Relevant neuropeptides involved in the circadian system were investigated, including arginine-vasopressin (AVP), vasoactive intestinal polypeptide (VIP), met-enkephalin (Met-Enk), neuropeptide Y (NPY), as well as oxytocin (OT). The neurotransmitter serotonin (5-HT) and the enzymes tyrosine hydroxylase (TH) and aromatic L-amino acid decarboxylase (AADC) were also studied. The camel SCN is a large and elongated nucleus, extending rostrocaudally for 9.55 ± 0.10 mm. Based on histological and immunofluorescence findings, we subdivided the camel SCN into rostral/preoptic (rSCN), middle/main body (mSCN) and caudal/retrochiasmatic (cSCN) divisions. Among mammals, the rSCN is unusual and appears as an assembly of neurons that protrudes from the main mass of the hypothalamus. The mSCN exhibits the triangular shape described in rodents, while the cSCN is located in the retrochiasmatic area. As expected, VIP-immunoreactive (ir) neurons were observed in the ventral part of mSCN. AVP-ir neurons were located in the rSCN and mSCN. Results also showed the presence of OT-ir and TH-ir neurons which seem to be a peculiarity of the camel SCN. OT-ir neurons were either scattered or gathered in one isolated cluster, while TH-ir neurons constituted two defined populations, dorsal parvicellular and ventral magnocellular neurons, respectively. TH colocalized with VIP in some rSCN neurons. Moreover, a high density of Met-Enk-ir, 5-HT-ir and NPY-ir fibers were observed within the SCN. Both the cytoarchitecture and

#### Edited by:

Jackson Cioni Bittencourt, University of São Paulo, Brazil

#### Reviewed by:

Paul Manger, University of the Witwatersrand, South Africa Marina Bentivoglio, University of Verona, Italy

> \*Correspondence: Khalid El Allali k.elallali@iav.ac.ma

Received: 30 July 2017 Accepted: 27 October 2017 Published: 16 November 2017

#### Citation:

El Allali K, Achaâban MR, Piro M, Ouassat M, Challet E, Errami M, Lakhdar-Ghazal N, Calas A and Pévet P (2017) The Suprachiasmatic Nucleus of the Dromedary Camel (Camelus dromedarius): Cytoarchitecture and Neurochemical Anatomy. Front. Neuroanat. 11:103. doi: 10.3389/fnana.2017.00103 the distribution of neuropeptides are unusual in the camel SCN as compared to other mammals. The presence of OT and TH in the camel SCN suggests their role in the modulation of circadian rhythms and the adaptation to photic and non-photic cues under desert conditions.

Keywords: suprachiasmatic nucleus, immunofluorescence, cytoarchitecture, neuropeptides, oxytocin, tyrosine hydroxylase, dromedary camel

## INTRODUCTION

fnana-11-00103 November 14, 2017 Time: 15:49 # 2

Rhythmicity is a ubiquitous property of all living organisms (Turek and Van Reeth, 1996). Biological rhythms in mammals are driven by a circadian clock located in the suprachiasmatic nucleus (SCN) of the hypothalamus (Moore and Eichler, 1972; Stephan and Zucker, 1972; Inouye and Kawamura, 1979; Green and Gillette, 1982; Groos and Hendriks, 1982; Lehman et al., 1987). The SCN is a strong autonomous oscillator cycling with a period close to, but different from, 24 h and entrained to exactly 24 h by environmental cues, the Zeitgeber (Refinetti, 2006). This nucleus is a complex structure containing several neuronal populations that have specific afferents and efferents. In most species, especially in rodents, the SCN is divided into two subdivisions: the dorsomedial SCN or 'shell' and the ventrolateral SCN or 'core' (Abrahamson and Moore, 2001; Morin et al., 2006). The dorsomedial SCN contains neuronal perikarya immunopositive for arginine-vasopressin (AVP) (Ibata et al., 1999; Moore et al., 2002; Morin et al., 2006; Nascimento et al., 2010) while the ventrolateral SCN contains neurons expressing vasoactive intestinal polypeptide (VIP), gastrin releasing peptide and peptide histidine isoleucine (Stopa et al., 1984; Card et al., 1988; Mikkelsen et al., 1991; Swaab et al., 1994; Smale and Boverhof, 1999; Moore et al., 2002; Nascimento et al., 2010). A large number of other neuropeptides have been described in the SCN of mammals with some interspecies variations. It was demonstrated that somatostatin perikarya are generally located in the intermediate region of SCN (Card et al., 1988; Ibata et al., 1999) while neurophysin, neurotensin, thyrotropinreleasing hormone, enkephalins (Enk), and angiotensin II were described in different parts of the nucleus (Block et al., 1988; Tillet et al., 1989; Abrahamson and Moore, 2001; Thomas et al., 2004). Moreover, calcitonin gene-related peptide (Park et al., 1993), galanin (Skofitsch and Jacobowitz, 1985; Abrahamson and Moore, 2001), substance P (Morin et al., 1992; Mikkelsen and Larsen, 1993; Abrahamson and Moore, 2001; Piggins et al., 2001) and the calcium-binding protein calbindin (Silver et al., 1996; Ikeda and Allen, 2003; Menet et al., 2003) have also been reported in this nucleus in different species. Additionally, gammaaminobutyric acid (GABA), an important neurotransmitter in the circadian system, is found in almost all neurons of the SCN (Okamura et al., 1989; Moore and Speh, 1993; Kalsbeek et al., 2000; Moore et al., 2002).

In addition to neuronal cell bodies, a typically dense innervation for various neuroactive agents exists in the SCN. Glutamate and pituitary adenylate cyclase activating peptide afferents originating from melanopsin retinal ganglion cells and forming the retinohypothalamic tract (RHT) are observed in the ventral region of the SCN containing VIP neurons (for review see Hannibal, 2002). Moreover, the ventrolateral SCN of rodents also contains serotonin (5-HT)-immunoreactive (ir) fibers that originate from the midbrain raphe nuclei. The core of the SCN also receives afferents originating from the intergeniculate leaflet (IGL) of the lateral geniculate complex containing neuropeptide Y (NPY), GABA and met-enkephalin (Met-Enk) (Moore and Eichler, 1972; Moore et al., 1978; Steinbush, 1981; van den Pol and Tsujimoto, 1985; Card and Moore, 1988, 1989; Tillet et al., 1989; Morin et al., 1992).

Functional relevance of the neurochemical composition of the master clock has not yet been understood fully. However, mechanisms and molecular processes underlying the synchronization of the SCN by environmental cues are well documented. It is known that in mammals, the light-dark cycle (LD) is the strongest Zeitgeber (Refinetti, 2006) imposing its period and phase to the circadian clock. The VIP neurons of the ventrolateral SCN, which are the target of the RHT (Tanaka et al., 1993), play an important role in the synchronization by light and transmit photic signaling to the pacemaker AVP neurons located dorsally in the dorsomedial SCN (Jacomy et al., 1999; Aton et al., 2005). In mammals, the molecular basis of clock function includes alternating activation and repression of gene expression by different proteins. These rhythmic self-regulating proteins are encoded by clock genes. The molecular machinery of the main clock is well known, and is based on two interdependent feedback loops (Reppert and Weaver, 2000, 2001; Shearman et al., 2000; Okamura et al., 2002; Takahashi et al., 2008; Mohawk et al., 2012).

In addition to photic entrainment by the LD cycle, a non-photic entrainment by other Zeitgebers and cues also exists. In particular, these include ambient temperature (Ta), food availability, pharmacological effect of melatonin and social cues. Experimental lesions have demonstrated that the geniculohypothalamic tract (Janik and Mrosovsky, 1994; Wickland and Turek, 1994; Challet et al., 1996; Miller et al., 1996; Schuhler et al., 1999; Goel et al., 2000) and serotoninergic afferents from the raphe nucleus (Cutrera et al., 1994; Miller et al., 1996; Challet et al., 1997) are involved in such non-photic entrainment. The relative importance of non-photic vs photic entrainment has not been widely studied.

Recently, we have demonstrated in the dromedary camel (one-humped camel, Camelus dromedarius) that a non-photic entrainment can be sufficiently strong to drive the circadian clock (El Allali et al., 2013). Indeed, in this species, environmental temperature cycles as well as the LD cycle synchronize (phase and period) melatonin and body temperature rhythms. The neural circuitry and neuropeptides involved in such entrainment are still

unknown. This raises several questions about the structure and functioning of the SCN in this species. However, there are no data on the organization of the camel hypothalamus and only limited data are available on the brain of this species. For these reasons, the present study aimed at investigating the organization of the SCN in the dromedary camel and intended to fill the existing gap in our understanding of neuroanatomy. We carried out a cartography of the nucleus and studied its cytoarchitecture. The distribution of relevant neuropeptides involved in the circadian system was also investigated. These include AVP, VIP, Met-Enk, NPY, and oxytocin (OT). The enzymes tyrosine hydroxylase (TH) and aromatic L-amino acid decarboxylase (AADC) and the neurotransmitter serotonin (5-HT) were also studied.

## MATERIALS AND METHODS

This study was conducted by sampling the hypothalamus of 31 camels, slaughtered to provide meat for public consumption, at two sites: the Eddakhla slaughterhouse in the south of Morocco (latitude: 23◦◦43<sup>0</sup> N, 15◦ 57<sup>0</sup> W) and the Rabat slaughterhouse in the north of the country (latitude: 34◦ 01<sup>0</sup> N, 6◦ 50<sup>0</sup> W). The animals included 13 females and 18 males of local varieties (Piro et al., 2011) aged between 4 and 12 years (the lifespan of a camel is 20–30 years). They were kept outdoors under natural environmental (photoperiod and temperature) conditions with free access to water. Slaughter was carried out at different periods of the year and almost at the same time of the morning: 06h00. Brain samples were dissected out as animals were slaughtered. The study was performed in conformity with the Hassan II Agronomy and Veterinary Institute of Rabat and Moroccan Ministry of Agriculture recommendations, which are in accordance with international ethical standards (Touitou et al., 2006).

The heads were perfused immediately after slaughter. The perfusion was performed through the two external carotids using, first, a washing heparin (2.5 UI/ml) solution in saline (0.9% sodium chloride, Fluka <sup>R</sup> ), followed by tissue fixation (5–6 l), using either formaldehyde (Sigma–Aldrich <sup>R</sup> ; 10% in water) for cytoarchitectonic studies or 4% paraformaldehyde (Scharlau <sup>R</sup> ) in phosphate-buffered saline (PBS; 0.2M, pH 7.4) for immunohistochemistry. Thirty to forty minutes after perfusion, the brain was carefully removed and the hypothalamus dissected and post-fixed at 4◦C for 3 days in the fixative solution.

## Histology

Fifteen hypothalami from both sexes were used for studying the SCN cytoarchitecture using Nissl and hematoxylin-eosin (H&E) stains of sections from paraffin-embedded tissue samples and frozen sections.

#### Paraffin Embedding

As mentioned above the brains were fixed using 10% formaldehyde solution. The hypothalamic samples containing the SCN were dehydrated, embedded in paraffin and cut into frontal or sagittal sections at a 6 or 9 µm thickness using a microtome (Shandon Hypercut <sup>R</sup> ). All sections were collected and every 10th consecutive section was processed for staining. Freely floating sections in 4% warmed (42◦C) gelatin solution were mounted on glass slides and allowed to dry for 20 min in an oven at 56◦C. Slides were then immersed in toluene (two times for 5 min) and then stained either with H&E or Nissl staining (cresyl violet or toluidine blue). The staining procedure involved sequentially dipping the slides in different solutions: toluene and ethanol; and then dye solutions: hematoxylin (5 min) and eosin (5 min), or cresyl violet for 1 min or toluidine blue for 2 min. The staining was followed by washing in distilled water for 1 min, dehydration in different baths of ascending ethanol (75, 90, and 100%) and clearing with toluene.

#### Frozen Sections

These sections were used to calculate the length of the SCN and its subdivisions. The fixed hypothalamic specimens containing the SCN were cryoprotected in 20% sucrose (Fluka <sup>R</sup> ) in distilled water until subsequent freezing in −30◦C isopentane (C5H12, Fluka <sup>R</sup> ) cooled with liquid nitrogen. Sections were cut in the coronal plane at a 20 µm thickness using a freezing microtome (Leica-3050 <sup>R</sup> ), mounted on gelatinized slides, and kept at −20◦C until being processed for Nissl or H&E staining.

In all cases, sections were coated with mounting medium: Eukitt <sup>R</sup> (Sigma–Aldrich) for cresyl violet and toluidine blue staining, and Canada balsam (Fisher Scientific) for H&E, and coverslipped. The sections were viewed under a light microscope (Topview 4500 <sup>R</sup> ) and images taken with Lumenera's Lw1130-1.4 megapixel CCD digital camera.

## Immunofluorescence

Sixteen hypothalami of both sexes, perfused with 4% paraformaldehyde as indicated above, were used for the study of SCN chemical neuroanatomy. After post-fixation, hypothalamic tissue samples containing the SCN were rinsed three times for 10 min in phosphate buffer (PB 0.1M/ pH7.4), and then cryoprotected in 20% sucrose (Fluka <sup>R</sup> ) in PB at 4◦C until subsequent freezing in −30◦C isopentane (C5H12, Fluka <sup>R</sup> ) cooled with liquid nitrogen. The tissue samples were cut into 20 µm-thick sections using a freezing microtome (Leica-3050 <sup>R</sup> ). The sections were mounted on gelatin or Super-frost <sup>R</sup> (Menzel Glaser) slides and kept at −20◦C until being processed for immunofluorescence. For this procedure, the sections were pre-incubated for 1 h at room temperature in 5% bovine serum albumin (BSA, Sigma-Aldrich) and 0.5% of Triton (Sigma-Aldrich <sup>R</sup> ) in 0.05M PBS. Incubation in the primary antibody was then carried out overnight at room temperature. The antibodies used are listed in **Table 1**.

After being washed three times for 10 min in buffer-2 solution (0.2% BSA in 0.05M PBS), sections were incubated in the secondary antibody diluted in the same buffer-2 for 2 h at room temperature.

For single labeling, the secondary antibodies were biotinylated anti-rabbit IgGs raised in goat (Vector; diluted 1:200). After washes in 0.05M PBS, the sections were incubated in streptavidin-Cy3 (or streptavidin - FITC) diluted 1:200 in 0.05M PBS for 2 h in darkness. Sections were finally rinsed 4 times for 10 min in 0.05M PBS.

For double labeling, the secondary antibodies used were coupled directly to fluorochrome: anti-rabbit-Cy3 antibody raised in sheep (Sigma; diluted 1:500), anti-mouse-FITC antibody raised in horse (Vector, diluted 1:500) or antirabbit IgG Alexa Fluor 488 antibody raised in goat (diluted 1:200; Molecular Probes <sup>R</sup> ). The sections were coverslipped in polyvinyl alcohol Moviwol <sup>R</sup> 4-88 (Sigma–Aldrich**)**. They were then examined using a fluorescence microscope (Leica <sup>R</sup> DM400B), inverted fluorescent microscope with motorized stage (Zeiss <sup>R</sup> Axiovert 200) or confocal microscope (Leica <sup>R</sup> TCS SP2 AOBS).

Specificity tests were based on the omission of the primary or secondary antibodies in some sections. In all cases, the immunolabelling was completely abolished.

## Cytoarchitectural Study

fnana-11-00103 November 14, 2017 Time: 15:49 # 4

The length of the SCN and its subdivisions were calculated as the sum of the cross-sectional surface area of each nucleus multiplied by the section thickness (20 µm) (Tessonneaud et al., 1994). Values were measured from frozen sections. Due to the large dimension of the camel hypothalamus, the frozen sectioning allowed a successful production of serial sections. Values from all specimens were averaged and expressed as a mean ± standard error mean (SEM).

The cell size, which corresponds to the widest diameter of the soma, was measured microscopically using two validated methods (Cassone et al., 1988; Kraus and Wolff, 2008; Mesaros et al., 2008; Suzuki et al., 2010): automatically by using Axovision <sup>R</sup> software of Zeiss microscopy, or manually by using a metric ruler. The cell size (d) was calculated using the formula d = f/N where (f) is the field of view and (N) the estimated number of cells which fit across the diameter of the field of view. The calculation was limited to neurons with a visible nucleus in the focal plane. A total of 30 neurons were randomly selected from each specimen and data for each part of the SCN were averaged and expressed as mean ± SEM.



## Images

Due to the large dimensions of the camel hypothalamus, data are illustrated by images at different magnifications or collating images to reconstruct the SCN and surrounding structures. The montage was achieved automatically when a motorized microscope stage was available.

## RESULTS

## SCN Morphology and Cytoarchitecture

The SCN of the camel could be clearly distinguished from the surrounding hypothalamic nuclei and areas using Nissl staining. The nucleus appeared as a bilateral, confined aggregate of neurons, relatively long and extending rostro-caudally for 9.55 ± 0.10 mm (**Figure 1**). Its shape and location were complex and changed along the rostrocaudal axis. Based on histological and immunofluorescence findings, the SCN showed a distinct topography (**Figures 1**, **2**), permitting us to subdivide the camel SCN into rostral or preoptic (rSCN), middle or main body (mSCN) and caudal or retrochiasmatic (cSCN) divisions.

Extending for 5.40 ± 0.12 mm, the rSCN is the largest division of SCN, and appears as a collection of neurons located outside the main mass of the hypothalamus, and was observed to protrude from the preoptic area (**Figures 1**, **2A,B,F,I,J,K**). The most rostral part of this division is represented by a thin horizontal band of tissue laying on the dorsal surface of the optic chiasm, 1 mm from the point where the optic nerves fuse into their chiasm (**Figures 1C**, **2A,B**). More caudally, the nucleus is located within and below the supraoptic recess (SoR) and ventral to the organum vasculosum of the lamina terminalis (OVLT). The rostral SCN division extends caudally and the supraoptic narrow tissue band becomes a dome-shaped tissue (Mds) at the midline (**Figures 1D,E**, **2F,G**). Proceeding caudally, this Mds cell group divides into two lateral suprachiasmatic swellings (Lsw) jutting on either side into the SoR (**Figures 1F,G**, **2I–K**). At this level, the OVLT is delineated ventrolaterally by two lateral expansions of the hypothalamus (Leh), which constitute a strand supporting the optic chiasm and contribute to form the SoR. Further caudally, the optic chiasm ascends dorsally and the supraoptic recess disappears (**Figures 1G,H**). The swellings of the rSCN become closer to the main mass of the hypothalamus, within which they gradually merge to fuse with the mSCN. Caudally and beyond these levels, the mSCN exhibits the characteristic triangular shape described in rodents (**Figures 1H,I**, **2L–N**). Its total length is 2.09 ± 0.10 mm. In this location, the SCN is found bilaterally in the basal hypothalamus and is formed by dense neuronal aggregates, separated by the third ventricle. A thin layer of tissue, which persists within the boundaries of the optic chiasm, connects the two sides of the mSCN.

The cSCN is observed in the retrochiasmatic areas posterior to the optic chiasm (**Figure 2T**) and at the beginning of the optic tract. It extends rostrocaudally for 2.29 ± 0.13 mm. At this level, the nucleus is located dorsally to the arcuate nucleus (Arc). It is formed by periventricular neurons at the base of the third ventricle (**Figures 1**, **2P–S**). Compared to the mSCN, the cell

SCN in coronal sections. The nucleus is represented in black; the optic nerve (on) optic chiasm (ox) and optic tract (ot) are in gray. The remaining areas are delineated by drawing lines representing the borders of the third ventricle, the preoptic area and the rest of the hypothalamus. The level of each section is indicated by the distance (mm) from the point of fusion of optic nerve (level 0: optic chiasm formation). Note that in the camel, the SCN is very long and divided into 3 parts: the rostral SCN (rSCN: A–G), the main SCN (mSCN: H,I) and retrochiasmatic or caudal SCN (cSCN: J). Note also that the rostral SCN is located outside the main mass of the hypothalamus lining the dorsal surface of the optic chiasm. 3V, third ventricle; Leh, lateral expansions of the hypothalamus; Lsw, lateral suprachiasmatic swellings of rSCN; Mds, median dome-shaped tissue of rSCN; OVLT, Organum vasculosum of the lamina terminalis; SoR, supraoptic recess.

density of the cSCN diminishes at these levels. More caudally, the borders of this nucleus become progressively indistinct.

Measurements of cell soma diameters reveal that the cytoarchitecture of the camel SCN is heterogeneous. The majority of cell bodies are small-sized but there are also a number of large-sized neurons (**Figures 2F,N,O**). The small-sized neurons are predominant in the main body of the nucleus and also dorsally (**Figure 2H**), whereas the large-sized cell bodies are less numerous and located ventrally (**Figures 2R,S**). This distribution was largely confirmed by our immunofluorescence studies. The small-sized cells of the camel SCN present a mean diameter of 10.1 ± 0.1 µm (range: 5–13 µm). They are generally elongated, triangular or round-shaped. The large-sized cells in the ventral part of the SCN show a mean diameter of 26.5 ± 1.5 µm (with a majority of diameters ranging from 15 to 35 µm). Their somata display various irregular shapes. Immunofluorescence investigations show that the large-sized neurons are less numerous than the small-sized ones and are observed in the narrow supraoptic tissue band of the rSCN laying directly on the optic chiasm. Overall, the dromedary SCN displayed several populations of neurons, with different sizes, shapes and neurochemical phenotypes.

## VIP Immunoreactivity Neuronal Cell Bodies

As expected, numerous neuronal cell bodies were VIP-ir in the camel SCN. The VIP-positive perikarya were confined within the SCN and were not found in the surrounding hypothalamus.

In the rSCN, the VIP-ir perikarya were clustered at the dorsal edge of the nucleus immediately below the SoR (**Figure 3**). These cells were round, with a mean diameter of 14.0 ± 0.8 µm (ranging 6–20 µm) and displaying a small soma with a narrow nucleus and a large cytoplasm (**Figures 3B,E**). These small-sized neurons occupied a dorsal position along the different levels of the rSCN. Thus, they were seen in the Mds and the Lsw SCN tissue. In these locations, the VIP-ir neurons gave rise to fibers directed toward the OVLT vessels. In the rostral levels of the mSCN (**Figures 4A–C**), the VIP-positive cells appeared first in the ventral region of the nucleus adjacent to the optic chiasm, and became concentrated to form a round, dense cluster of perikarya. More caudally, the VIP-ir-small-sized neurons were numerous and not restricted to the ventral surface (**Figures 4D–F**), and were widely distributed through a large area of the mSCN core body. The mSCN VIP-ir neurons were small-sized, with a mean diameter of 13.6 ± 0.5 µm (ranging from 9 to 18 µm), and displayed a small perikaryon, a relative narrow nucleus and a large cytoplasm. Their shape was predominantly multipolar. The VIP-ir neurons were sparsely distributed in the cSCN at this level (**Figure 5**).

Double immunostaining for VIP and the neuropeptides and enzymes we examined showed a colocalization only of VIP and TH in several neurons of the rSCN (**Figure 3**).

#### Fibers

The SCN of the camel showed numerous fine VIP-ir fibers with varicosities, generally uniform in size. These fibers were

FIGURE 2 | Images showing the cytoarchitecture and rostrocaudal organization of the camel SCN using classical histology staining of coronal sections. The rostral SCN (rSCN) is presented in (A–H), the main SCN (mSCN) in (I–O), the retrochiasmatic or caudal SCN (cSCN) in (P–T). (A): Low power view of toluidine blue-stained section at the level of preoptic area. The image shows the location of the rSCN. (B): Montage of 15 images showing the location of the rSCN lying on the dorsal surface of the optic chiasm (ox). Toluidine blue staining. (C): Montage of 12 images showing the peculiar morphology of the rSCN medially, juxtaposed to the dorsal surface of ox. Toluidine blue staining. (D): Image in another representative animal showing the same shape of rSCN tissue as in (C). Toluidine blue-stained section. (E): Higher magnification of the field boxed in (D), showing small-sized neurons of the rSCN stained with toluidine blue. (F): Morphology of cresyl violet-stained rSCN at the middle level of its rostro-caudal extent. The image shows the median dome shaped tissue (Mds) of rSCN under the supraoptic recess (SoR). Montage of 6 images. (G): Higher magnification of the field boxed in (F). Montage of 4 images. (H): Cresyl-violet-stained neurons of rSCN. Higher magnification of the field boxed in (G). (I): Low power view of cresyl-violet- stained section of the anterior region of the hypothalamus. The image shows the location of the rSCN at its caudal level. Note that at these levels, the Mds tissue of rSCN is split into two lateral swellings of tissue (Lsw) under the SoR and the Organum Vasculosum of the Lamina Terminalis (OVLT). (J,K): Images showing progressive changes of the Lsw from median to the lateral position. At their most lateral part, the Lsw fuse with the lateral expansions of the hypothalamus (Leh). This constitutes the beginning of the mSCN portion (Cresyl violet). (L): Low power view of cresyl violet- stained section of the (Continued)

#### FIGURE 2 | Continued

fnana-11-00103 November 14, 2017 Time: 15:49 # 7

tuberal region of the hypothalamus. The image shows the location of the bilateral mSCN occupying the classical location within the hypothalamus at the ventral edges of the third ventricle (3V). (M): Image of the mSCN from another representative camel (Cresyl violet). (N): Image of the ventral part of the mSCN stained with toluidine blue. (O): Higher magnification of the field boxed in (N) showing small-sized cells of the mSCN (Toluidine blue). (P): Montage of 4 images (Toluidine blue) showing the location of cSCN in the caudal part of tuberal region of the hypothalamus. Note that at this level of coronal sections, the boundaries of the cSCN are difficult to delimit from surrounding regions. (Q): Montage of 12 images (Toluidine blue) showing at higher magnification part of the image shown in (P). (R): Montage of 12 images (Toluidine blue) showing at higher magnification the field bordering the third ventricle in (Q). (S): Higher magnification of the field boxed in (R) (Toluidine blue) showing small-sized neurons of the cSCN. (T): Sagittal section showing the rostrocaudal extent of the SCN in the camel (H&E). Note the rostral position of the rSCN under the SoR and also the retrochiasmatic extent of the cSCN (its limits are indicated by the red arrows) caudal to the ox (black arrow).

rostral SCN (rSCN) of the camel. (A–F) Represents respectively two coronal sections corresponding to the levels of panel (E) in Figure 1. Merge shows colocalization (arrowheads) of VIP and TH. ox, optic chiasm; SoR, supraoptic recess.

observed within the nucleus but also running out of the SCN toward the surrounding structures. They were observed at the rSCN level coursing laterally to the optic chiasm and dorsally to the OVLT vessels. In the mSCN, the VIP-ir fibers were first directed toward the dorsal part of the nucleus (**Figures 4A,D**). At the rostral levels of the mSCN, VIP-ir fibers were also directed toward the supraoptic nucleus (SON) (**Figure 4C**), while in the caudal part of the mSCN, these fibers constituted a dense network leaving the nucleus toward the dorsal hypothalamic area (**Figure 4D**). These VIP-ir fibers were observed branching dorsally to the paraventricular nucleus of the hypothalamus (PVN) and a large part of them changed their direction from the third ventricle to deviate laterally toward the hypothalamic dorsomedial nucleus, within which they formed an oblique dense plexus (**Figure 4D**). Furthermore, other thick-caliber periventricular fibers were observed to course dorsally, probably toward the subthalamic region. More caudally, in the cSCN, only some VIP-positive fibers were seen in the area of the nucleus and at the supraependymal level along the third ventricle wall (**Figure 5**).

## AVP Immunoreactivity Neuronal Cell Bodies

AVP-ir neurons were observed at different levels of the rostrocaudal extent of the camel SCN. They were located at the ventral boundary of the rSCN (**Figures 6A,B**) and also at the dorsal edges and in the lateral parts of the mSCN (**Figure 6D**). In the remaining sections sampled from the mSCN, the AVP-ir perikarya were observed only in the dorsal portion of the nucleus above the ventral population of VIP-ir neurons. Caudally, only sparse AVP-ir neurons were observed.

The AVP-ir perikarya displayed an irregular shape, and a few of them were round-shaped (**Figures 6C,D**). They were the smallest neurons of the SCN, ranging from 5 to 15 µm in diameter with a mean diameter of 8.8 ± 0.4 µm. Their nuclei were large and they were mostly of the multipolar type (**Figure 6D**).

#### Fibers

Immunofluorescence revealed the presence of AVP-ir fibers in the camel SCN. Some of them were observed within the nucleus (**Figures 6C–E**), while the majority was seen to course from the SCN to surrounding hypothalamic areas. In addition to the SON, the densest network of AVP-ir fibers appeared to connect the SCN with the PVN. Other AVP-ir fibers were located along the periventricular edge of the third ventricle.

# OT Immunoreactivity

### Neuronal Cell Bodies

OT-ir perikarya within the camel SCN were observed at different levels of the mSCN in coronal sections but neither in the rSCN nor in the cSCN (**Figure 7**). The OT-ir neuronal cell bodies were either scattered throughout the mSCN (**Figure 7A**) or clustered in the ventrolateral portion of the nucleus (**Figure 7B**). They were exclusively of small size with a diameter ranging from 5 to 16 µm and a mean diameter of 9.2 ± 0.5 µm. Their somal shape was irregular, sometimes with a round shape containing a small nucleus and a large cytoplasm. They were predominantly multipolar (**Figure 7C**).

Double immunofluorescence revealed that there was no co-localization of OT with either AVP or with the other neuropeptides studied. Overall, the density of OT neurons was lower than that of AVP-ir or VIP-ir neurons.

FIGURE 4 | Immunofluorescent labeling of vasoactive intestinal polypeptide (VIP; red, A–C) and double immunolabelling (D–F) of VIP (red) and tyrosine hydroxylase (TH: green) in the main SCN (mSCN) of the camel. The coronal sections correspond to different levels of (H,I) of Figure 1. (B): higher magnification of the field boxed in (A). (D) Is a reconstruction of 18 images representing a merge of red (VIP) and green (TH) immunofluorescence in the camel hypothalamus. (E): higher magnification of the field boxed in (D). (F): higher magnification of the field boxed in (E), showing the VIP-ir neurons within the mSCN. Note that the VIP-ir cells form a cluster of small-sized neurons located ventrally in the rostral levels of mSCN (A–C), whereas at the middle levels of the mSCN (D–F), these neurons occupy almost the entire nucleus. VIP-ir fibers were branching up dorsally to dorsomedial (DMH) and paraventricular (PVN) nuclei and to other dorsal hypothalamic areas. ox, optic chiasm; 3V, third ventricle.

#### Fibers

OT-ir fibers were observed within the mSCN (**Figures 7A,B**), and also leaving the mSCN toward the surrounding area.

## TH Immunoreactivity

#### Neuronal Cell Bodies

Several neuronal cell bodies were TH-immunopositive in the camel SCN. These neurons exhibited some peculiarities. First, at the level of the rSCN, TH ir-neurons were observed in different parts, especially in the Mds and Lsw tissue (**Figures 8A,B,E,H**, **9A**). Indeed, in the Mds and especially in the Lsw (**Figure 8**), two populations of TH-ir neurons were revealed: a dorsal population which contained small neurons with a mean diameter of 16.1 ± 0.7 µm (ranging from 5 to 20 µm) juxtaposed to the SoR (**Figures 3**, **8D–I**, **9A**) and a ventral population which was instead formed by less numerous but more intensely immunopositive large-sized neurons with a mean diameter of 33.5 ± 6.1 µm (ranging from 25 to 39 µm) and lying on the dorsal surface of the optic chiasm (**Figures 8A,B**, **9A**). The small-sized neurons were mostly unipolar and the ventral larger neurons were mostly bipolar or multipolar. These ventral large and dorsal small TH-ir neurons were also observed within the mSCN. We have observed in different sections through the mSCN both the ventral population of TH-ir large neurons, with a mean diameter of 39.8 ± 2.0 µm (ranging from 30 to 50 µm), and the dorsal population of small neurons with a mean diameter of 15.0 ± 1.0 µm (ranging from 5 to 22 µm) (**Figures 9B–G**). The small-sized TH-ir neurons were found in the most superficial part of the mSCN, whereas the ventral TH-ir large-sized neurons

were located within the ventral extent of the mSCN in a medial and two lateral clusters of neurons (**Figures 9B–D**). The TH-ir ventral large neurons were of different shapes, mostly bipolar or multipolar but also of the unipolar and pseudo-unipolar types (**Figures 9D,F,G**), whereas most of the dorsal TH-ir small neurons were unipolar or bipolar (**Figure 9E**). More caudally, in the cSCN, the TH-ir perikarya were exclusively large-sized, with a mean diameter of 24.1 ± 1.5 µm (ranging from 11 to 28 µm), scattered and less numerous (**Figure 10**). The majority of these neurons were unipolar and located in a very medial and periventricular position.

Double immunofluorescence showed that TH-ir neurons did not express any immunoreactivities for the neuropeptides we examined (**Figures 4D–F**, **6**, **7**), except for VIP. In fact, only a few TH-ir dorsal small-sized neurons of the rSCN showed such colocalization (**Figures 3**, **8D–I**). TH-ir neurons within the SCN did not express AADC (**Figures 8C**, **10C**).

#### Fibers

The TH-ir fibers were observed at different levels of the SCN. In the rSCN, fibers were seen in Mds, the two Lsw, and in Leh-containing tissue (**Figures 8A–C**, **9A**). Fibers were abundant ventrally at the junction of the optic chiasm. Caudally, some of these fibers appeared to course laterally to penetrate the SON. In the mSCN, the TH-ir fibers formed a dense network in the region of TH-ir large-sized neurons but also in the dorsal region containing small-sized neurons. These TH-ir fibers were very dense in the ventral part of the mSCN, forming a horizontal plexus delineating the optic chiasm (**Figures 9B–D**). At these levels, some fibers appeared to course laterally toward the SON. Other fibers had a vertical path with a ventral orientation and a ventrally penetrating trajectory in the optic chiasm. Furthermore, the ventral TH-ir large-sized neurons also extended fibers toward the dorsal population of TH-ir small-sized neurons and the lateral area of the nucleus devoid of TH-ir cell bodies (**Figures 9C,D**). Double immunofluorescence did not identify the chemical phenotype of neurons within this area targeted by TH-ir fibers. Caudally, in the cSCN, TH ir-fibers were less dense (**Figure 10B**).

## AADC Immunoreactivity

Immunolabelling of AADC was performed to identify whether the TH-ir neurons within the camel SCN are dopaminergic. The labeling demonstrated the presence of only rare, sparse AADC-ir neurons in the three parts of the SCN. Few labeled fibers were also observed. These neurons were large-sized and did not coexpress TH (**Figures 8C**, **10C**).

## Met-Enk, 5-HT and NPY Immunoreactivities

A very dense network of Met-Enk-ir fibers was observed in the camel SCN. These fibers displayed numerous varicosities forming a plexus within the three divisions of the SCN and especially in the mSCN (**Figures 11A–C**). Furthermore, plexuses of Met-Enk-ir fibers were widely distributed at different levels in surrounding areas of the hypothalamus. Immunolabelling showed also that the camel SCN contains several Met-enk-ir neurons (**Figures 11B,C**).

Immunofluorescence did not reveal 5-HT-ir perikarya in the camel SCN. Instead, there were numerous 5-HT-ir fibers (**Figures 11D–F**), with a relatively thick caliber and non-uniform varicosities. Interestingly, these fibers were observed to descend laterally from the dorsal hypothalamic area and also lining the wall of the third ventricle toward the SCN (**Figures 11D,E**). These 5-HT-ir fibers were found to form a dense plexus within the mSCN (**Figure 11F**).

Examination of the different levels of the camel SCN showed that this nucleus did not contain NPY-ir neuronal cell bodies. However, a high density of NPY-ir fibers was found within the camel SCN (**Figures 11G–K**). These fibers were mainly

FIGURE 6 | Double immunofluorescence of arginine-vasopressin (AVP) and tyrosine hydroxylase (TH) in the SCN of the camel. (A–C): Merge of AVP (green) and TH (red) in the rostral SCN (rSCN): levels of panel (E) of Figure 1. (D,E): Merge of double labeling of AVP (red) and TH (green) in the main SCN (mSCN): levels of (I) of Figure 1. (A) is a montage of 9 images; (B): higher magnification of the field boxed in (A) showing AVP-ir perikarya in rSCN; (C): image from the hypothalamus of another camel showing AVP neurons and TH fibers in the rSCN; (D): distribution of AVP and TH neurons in the mSCN; (E): higher magnification of the field boxed in (D), showing AVP and TH neurons. Note the dorsal and lateral locations of AVP neurons in the mSCN and the lack of colocalization of AVP and TH in the same neurons. 3V, third ventricle; SoR, supraoptic recess.

varicose and formed a dense network in the mSCN, but were sparse in rSCN and cSCN. Compared to all the other neurochemically identified fibers observed in the present study, the most extensively distributed fibers in the SCN were those immunoreactive for NPY.

## DISCUSSION

This study is the first undertaken on camel SCN neuroanatomy including its cartography, cytoarchitecture and chemoarchitecture. The results show that the SCN of the camel displays peculiarities regarding its size, extent, morphology and cytoarchitecture. Based on histological and immunohistochemical criteria, we subdivided the camel SCN into rostral, main and caudal divisions (rSCN, mSCN and cSCN; respectively). The location of the rSCN, organized as a collection of neurons protruding from the preoptic area outside the main mass of the hypothalamus, is unusual. The SCN at these levels is located below the SoR, lining the dorsal surface of that portion of optic chiasm that is still detached at these levels from the ventral borders of the brain. Despite the fact that such a conformation is uncommon in most studied species, a similar organization has been described in sheep (Tillet et al., 1989; Tessonneaud et al., 1994), cow (Lignereux, 1986), and in humans (Swaab et al., 1985; Hofman et al., 1988; Hofman and Swaab, 1989). Similarities with these species were also demonstrated with regard to the rostrocaudal morphological variations of the SCN. Indeed, structures comparable to the median dome-shaped tissue (Mds), lateral suprachiasmatic swellings tissue (Lsw), and lateral expansions of the hypothalamus (Leh); have been observed in the rSCN of the above-mentioned species. Caudally, the mSCN exhibits the usual triangular shape and location at the base of the third ventricle, dorsally to the optic chiasm, as classically described in rodents and several other mammals (Spiegel and Zweig, 1917; Bleier et al., 1979; van den Pol, 1980; Cassone et al., 1988; Vrang et al., 1995).

Due to the existence of its two rostro-caudal subdivisions, the rSCN and cSCN, which are lacking in most species, the dromedary SCN seems to be the largest studied thus far. This can

be attributed to the fact that the SCN starts far rostrally outside the main mass of the hypothalamus and elongates caudally to the optic chiasm. The SCN extends in the camel for 9.55 mm, while in sheep and in humans (species which exhibit a similar SCN morphology), the total rostrocaudal length is much shorter (2.8–3.1 mm and 1.47 mm, respectively; Swaab et al., 1985; Swaab and Hofman, 1990; Tessonneaud et al., 1994). The rostrocaudal extent of the SCN in other mammals depends on the species but ranges between 0.60 and 2.15 mm (0.60 mm in the Syrian hamster, 0.9 mm in the rat, 0.76 mm in the guinea pig, 0.76 mm in the mouse, 0.80–1.0 mm in the squirrel monkey, 0.90 mm in the macaque monkey, 0.95–1.65 mm in five marsupial species, 1.74 mm in the domestic pig and 2.15 mm in the domestic cat) (van den Pol, 1980; Lydic et al., 1982; Cassone et al., 1988).

There are great differences in the general topography and cytoarchitecture of the SCN across species. It seems reasonable to argue that mammals can be divided in two groups concerning the morphology of the SCN: camel, sheep, cow and human (and maybe other species) with a relatively long and large SCN, protruding out of the hypothalamus in its rostral division, and other mammals (including rodents) with the SCN corresponding to more "classical" descriptions.

Based on retinal projections, size and shape criteria as well as neurochemical phenotypes, the SCN, especially in rodents, has been divided in two dorso-ventral subdivisions: the dorsomedial SCN or 'shell' and the ventrolateral SCN or 'core' (Abrahamson and Moore, 2001; Morin et al., 2006). In the present histological study, no such subdivisions were evident in the camel SCN. However, clear cytoarchitectural differences were identified with respect to the size of SCN neurons, with populations of dorsal parvicellular and ventral magnocellular neurons. This organization is largely confirmed by the immunofluorescence studies. In sheep, there is also no clear dorso-ventral subdivision of the SCN, but neurons are smaller in the ventral than in the

in the mSCN. 3V, third ventricle.

dorsal region of the SCN (Tessonneaud et al., 1994). Neurons of the camel SCN are the smallest cells among those observed within the hypothalamus and adjacent areas. Nevertheless, the camel SCN contains neurons larger than those reported in sheep and other species, especially in the ventral magnocellular part, in which neurons are larger than 20 µm in diameter and can reach up to 50 µm. The largest somal diameters in the SCN of many species (including the domestic pig, domestic cat, mouse, guinea pig, rat, hamster and five marsupial species) do not exceed 12.5 ± 3.1 µm (measured in the ventrolateral division of the cat SCN), while in other species the diameter of SCN neurons ranges between 7.0 and 10.7 µm (van den Pol, 1980; Cassone et al., 1988).

The neurochemical organization of the camel SCN is different from that of the other species studied so far. In camel SCN, two populations of TH neurons were identified, ventral magnocellular or dorsal parvicellular. Their location, number, size and immunoreactivity intensity, suggests that they could have different roles. In the absence of a precise additional phenotyping of these neurons, it is difficult to further speculate on their role (s). To our knowledge, no similar findings were reported in the SCN of other mammals. Very few TH-ir neurons were observed to be homogeneously distributed in the SCN of sheep (Tillet et al., 1994). Moreover, in most rodents, this nucleus either lacks or contains only a few TH neurons. For example, in the rat, a few TH neurons were reported to be present only transiently in the developing brain (Battaglia et al., 1995; Ugrumov, 2013). In the Syrian hamster, only sparse TH-ir round-shaped cells were observed in the SCN, mostly outside the nucleus and lining its borders (Vincent, 1988; Novak and Nunez, 1998; Strother et al., 1998).

Tyrosine hydroxylase is the rate-limiting enzyme for the biosynthesis of catecholamines (dopamine, epinephrine and norepinephrine) and catalyzes the transformation of the amino acid L-tyrosine into L-DOPA (L-3, 4 Dihydroxyphenylalanine), which is then converted to dopamine by AADC (also called Dopa-decarboxylase). Within various animals, the AADC-ir neurons in the SCN seem to be more abundant than the TH neurons (Sheep: Tillet et al., 1994; rat: Jaeger et al., 1984; Inatomi, 1994; cat: Kitahama et al., 1988; house-shrew: Karasawa et al., 1992).

Regarding the puzzle of the relatively high density of TH-ir neurons we observed in the camel SCN, we have investigated their possible dopaminergic phenotype by testing colocalization with AADC. The double immunostaining showed that AADC-ir neurons are very rare and sparse within the nucleus, largesized and never coexpress TH. This indicates that there are no

(Continued)

#### FIGURE 9 | Continued

fnana-11-00103 November 14, 2017 Time: 15:49 # 14

this level, as in the rostral part of rSCN, two TH-ir neuronal populations: dorsal small-sized and ventrolateral large neurons. (B): TH immunofluorescence in the mSCN [levels of (H) of Figure 1). The montage (of 16 images) shows TH- immunolabelling in the mSCN of the two sides. (C) Shows at higher magnification, the field boxed in (B), with TH-ir neurons in the mSCN. (D) Shows at higher magnification the field boxed in (C) with TH-ir neurons and fibers in the right mSCN. Note also at these levels the existence of two populations of TH-ir neurons: small-sized neurons (white arrowheads) which fill the mSCN except for the lateral and ventrolateral portions of the nucleus, and large-sized neurons (yellow arrowheads) in the ventral part of the mSCN with medial and lateral clusters of neurons. Large-sized neurons show intense immunofluorescence and are intermingled with a dense network of TH-ir fibers. (E): dorsal TH-ir small-sized neuronal population. (F,G): different types of the ventral large-sized TH-ir neurons.

FIGURE 10 | Immunofluorescence of tyrosine hydroxylase (TH) in the caudal SCN (cSCN). (A): montage of 25 images showing merge of TH (green) and VIP (red) in the caudal part of the tuberal region of camel hypothalamus and cSCN. The coronal section corresponds to the levels of panel (J) in Figure 1. (B): higher magnification of the field boxed in (A). Note that in the cSCN the TH-ir perikarya are exclusively small-sized neurons, scattered and not very numerous. (C): merge of TH (green) and AADC (red); showing rare, sparse AADC neurons in the cSCN of the camel and which do not coexpress TH. 3V, third ventricle; LH, lateral hypothalamus; ot, optic tract.

catecholaminergic, especially no dopaminergic, neurons in the SCN of the camel. The lack of AADC expression within the TH-ir neurons indicates the synthesis of L-DOPA, but not dopamine, in these neurons. This raises several questions. L-DOPA itself has a role in mediating the release of neurotrophic factors that are important for the growth, survival and differentiation of neurons (Lopez et al., 2008; Malenka et al., 2009; Zigmond et al., 2012; Hiroshima et al., 2014). In the camel SCN, given the relatively high number of TH-neurons, L-DOPA could be involved in the survival and differentiation of neurons. A possible involvement of TH neurons in mediating circadian activity is also suggested by the colocalization of TH with VIP especially in rSCN neurons.

The presence of OT-ir neurons in the SCN has been reported briefly by El May et al. (1987) when studying the hypothalamohypophyseal axis of the camel. However, their given localization of the SCN and their description as a large extension from the optic chiasm to the pituitary stalk attachment were not very precise. In the present work, the camel SCN cartography is performed at high spatial resolution and results confirmed clearly that OT neurons are found within the mSCN and are either scattered throughout the nucleus or grouped ventro-laterally. The presence of OT-ir neurons within the SCN is unusual, and represents a peculiarity of the nucleus in the camel. Sofroniew and Weindl (1980) studied the neuropeptidergic content of the SCN in 13 species belonging to 6 mammalian orders (marsupials, rodents, lagomorphs, artiodactyls, carnivores and primates). In all of these animals, the SCN lacks OT-ir neurons, as confirmed in other studies (Sofroniew and Glasmann, 1981; Reuss et al., 1989; Caba et al., 1996). For the camel, a possible role of OT neurons in the modulation of the circadian clock activity cannot be excluded. This could be of particular interest in species such as the dromedary, that faces the problem of adaptation to the harsh environment of the desert and which has to adapt its physiology and circadian rhythms to this biotope.

The existence of vasopressinergic neurons in the camel SCN is consistent with reports in other mammalian species (Vandesande et al., 1975; Vandesande and Dierickx, 1975; Sofroniew and Weindl, 1980; Card et al., 1981; Tillet et al., 1989; Kalsbeek and Buijs, 1992; Kikusui et al., 1997; Abrahamson and Moore, 2001; Kalsbeek and Buijs, 2002). However, the distribution of these neurons is different in the camel SCN compared to the animals studied thus far. They occupy different locations on the rostrocaudal extension of the nucleus and appear more caudally in a dorsomedial position forming a population equivalent to the "shell" described in rodents (Ibata et al., 1999; Moore et al., 2002; Morin et al., 2006; Nascimento et al., 2010). The distribution of AVP-ir neurons in the camel SCN, as well as their relatively high number, seem to be a peculiarity of a species living in arid and desert biotopes. In the jerboa (Jaculus orientalis), a semidesert rodent living in the same latitudes as the dromedary, AVP neurons in the SCN are intensely immunostained, and are located in the dorsomedial and ventromedial but also dorsal, dorsolateral and ventral portions of the SCN (Lakhdar-Ghazal et al., 1995a). Across all species studied to date, the AVP neurons located in the dorsomedial SCN play a crucial role in the development and distribution of circadian signals (for a review, see Reghunandanan and Reghunandanan, 2006) and this could also be the case for the camel SCN. Furthermore, studies have shown that the amount of vasopressin release in the SCN (Kalsbeek et al., 1995) and its mRNA levels (rat: Larsen et al.,

1994, mice: Smith and Carter, 1996, Siberian hamster: Duncan et al., 1995) show daily variations with a diurnal acrophase. This rhythmicity is maintained under constant conditions (Yamase et al., 1991; Cagampang et al., 1994), demonstrating its circadian origin.

A relatively large number of VIP-ir neurons were found in the camel SCN. In the present investigation, the highest density of these neurons within the camel hypothalamus was observed in the SCN. Similar observations have been made in other mammalian species (Card et al., 1981; Stopa et al., 1984; Cassone et al., 1988; Tessonneaud et al., 1994; Abrahamson and Moore, 2001; Moore et al., 2002). It is well known that the ventrolateral subdivision of the SCN receives direct retinal afferents (Moore, 1973; Cassone et al., 1988; Ibata et al., 1989; Abrahamson and Moore, 2001), contacting VIP neurons (Ibata et al., 1989). Moreover, VIP expression (mRNA and peptide) in the SCN exhibits a nycthemeral rhythm depending on the LD cycle (Albers et al., 1990; Ibata et al., 1993; Shinohara et al., 1993; Yang et al., 1993; Larsen et al., 1994). In the camel SCN, the most rostral VIP-ir neuronal population, forming a cluster of cells in the ventral part of the mSCN, is similar in its shape and location to the findings reported in other species. Although no data on retinal projections is available in the camel, due to its location this group of VIP neurons could serve as the target of such projections also in the camel.

In rodents, VIP neurons project to the dorsal AVP neurons in the SCN (Jacomy et al., 1999) to regulate the activity of the clock by light (Harmar et al., 2002). Our findings suggest that VIP neurons in the camel SCN also project to the AVP neurons located dorsally. We also observed a dense plexus of VIP-ir fibers directed to the hypothalamic area above the SCN, which seem

to reach the PVN and the dorsomedial nucleus, as reported in several other species (Kalsbeek et al., 1993; Saper et al., 2005). The neurophysiological significance of such dorsal projections could be related to the well-known modulatory role of VIP on AVP neurons (Watanabe et al., 1998; Jacomy et al., 1999; Maywood et al., 2006) and thus on clock activity. AVP neurons, in turn, would distribute the modulated circadian message to other structures in the brain.

In the jerboa, the SCN content of VIP shows seasonal variations (Lakhdar-Ghazal et al., 1992) which seems to be related to the photoperiod and the effect of sex hormones (Oukouchoud et al., 2003). Such mechanisms remain to be investigated in the camel. However, in view of data on the variation in the duration of melatonin secretion in this species (El Allali et al., 2005, 2008) and the existence of seasonal breeding activity, it seems reasonable to suppose that the photoperiod could modulate VIP expression in the camel SCN.

The present results show that the NPY-ir fibers were the densest in the camel SCN among the other neurochemically identified fibers investigated in this study. The NPY fibers form a dense plexus in different parts of the camel SCN, and especially in the mSCN. Such a high density of NPY fibers coming from the IGL has been previously described in the rodent SCN (Moore et al., 1984; Ueda et al., 1986; Sabatino et al., 1987; Card and Moore, 1989; Morin et al., 1992; Lakhdar-Ghazal et al., 1995b; Jacob et al., 1999; Menet et al., 2001; Abrahamson and Moore, 2001) and these fibers contact VIP neurons (Ibata et al., 1988; François-Bellan and Bosler, 1992). This innervation is reportedly involved in non-photic synchronization mechanisms (Challet et al., 1996, 1997; Juhl et al., 2007). The neuroanatomical pathways and the entrainment by the IGL are thus well demonstrated in rodents, but remain to be fully understood in the camel, sheep and most non-human primates. In non-human primates, a complex of NPY neurons, the pregeniculate nucleus, is equivalent to the IGL of rodents, but does not send efferents to the SCN (Moore, 1989; Chevassus-au-Louis and Cooper, 1998). Likewise, the SCN of the sheep harbors only sparse NPY-ir fibers (Tillet et al., 1989). The present findings show a dense plexus of NPY fibers in the camel SCN, but a combination of immunohistochemistry and tract tracing is necessary to address its origin from the IGL.

The camel has to adapt its physiology and to anticipate changes in its harsh environment by integrating the most important environmental cues, mainly the environmental temperature and LD cycles. The density of NPY-ir fibers in the camel SCN may reflect the integration of non-photic signals, correlated with an important non-photic entrainment of the circadian clock represented by the daily cycle of environmental temperature (El Allali et al., 2013).

The present findings also demonstrate a dense plexus of Met-Enk-ir fibers in the camel SCN. Met-Enk innervation of the SCN has also been reported in others species, including: mouse (Abrahamson and Moore, 2001), sheep (Tillet et al., 1989; Tessonneaud et al., 1994) and Syrian hamster (Morin et al., 1992). In this latter species, Met-Enk fibers originate from neurons located in the IGL (Morin and Blanchard, 1995) and participate in photic transmission and clock synchronization (Harrington and Rusak, 1986; Pickard et al., 1987; Edelstein and Amir, 1999; Juhl et al., 2007).

Our data for the camel SCN demonstrate that 5-HT immunopositive fibers cross the hypothalamus, especially the periventricular areas, toward the SCN. A high density of 5-HT fibers in the SCN has been reported for several species (Ueda et al., 1983; Abrahamson and Moore, 2001). These fibers constitute a third major set of afferents to the SCN and are involved in the transmission of non-photic stimuli to this nucleus. Serotonergic modulation of the SCN is well characterized in the Syrian hamster (Azmitia and Segal, 1978; Meyer-Bernstein and Morin, 1996; Hay-Schmidt et al., 2003; Vrang et al., 2003). Moreover, in some species (rat and cat), direct retinal projections to the raphe nuclei were identified. Photic information could therefore also reach the SCN indirectly through 5-HT fibers. The existence of a significant 5-HT innervation in the camel SCN could be related to the particular adaptation of this species to its biotope thus requiring pathways combining non-photic and photic entrainment.

In most mammals, the relative importance of photic versus non-photic entrainment of the circadian clock is not fully understood. In addition to LD entrainment, several non-photic factors can synchronize the circadian clock. These stimuli may be behavioral, dietary or other environmental factors such as environmental temperature, for which little information is available to date. It appears that the unusual morphology of the camel SCN (both in terms of its length and shape) and the uncommon existence of TH and OT neurons, in addition to the presence of a dense innervation of NPY, Met-Enk and 5-HT fibers, reflect collectively the importance of this nucleus in the circadian adaptation of the camel to its harsh biotope. The synchronization of the circadian clock in this species both by photic and non-photic cues deserves special study. Under experimental conditions (El Allali et al., 2013), we demonstrated that both entrainments occur in the camel: synchronization by the LD cycle and by the daily cycle of environmental temperature. Neuroanatomical interactions in the camel brain between different pathways for photic and nonphotic entrainment are also very likely and warrant further investigation.

## ETHICS STATEMENT

There were no animal sacrifices to carry out this study. The work was conducted by using camels's brains. Samples were taken as animals were slaughtered to provide meat for public consumption. The study was in conformation with the Hassan II Agronomy and Veterinary Institute of Rabat and Moroccan Ministry of Agriculture recommendations which are in accordance with international ethical standards (Touitou et al., 2006).

## AUTHOR CONTRIBUTIONS

KEA, AC, PP, and NL-G conceived and designed the work; KEA, MA, and MO performed brains sampling; KEA

performed immunohistochemical labeling experiments; KEA and MP performed image acquisition and quantification of immunofluorescence; KEA, EC, and PP prepared the manuscript; MA, MP, MO, EC, ME, NL-G, AC, and PP revised and approved the final review.

## FUNDING

Funding was provided by PRAD Programs no. 05-10 & 03-07; The Hassan II Agronomy and Veterinary Medicine Institute Program: PRFI IAV; The Moroccan Program "PROFERD Dromadaire"; BTC- Belgian Development Agency Program; European Doctoral College Program "Rosa Parks Class" and the

## REFERENCES


National Center of Scientific and Technical Research (CNRST: URAC-49), Morocco.

## ACKNOWLEDGMENTS

The authors are grateful to Dr. Mourad El Allouchi, Mr. Mohcine Hadaoui, Dr. Lahcen Boukbir, Dr. Hakim Chaaibi, Dr. Belakhal, Dr J. Malik, Dr. Aziz Marhaban, Dr. Johann Egginger, Dr. Caroline Parmentier, and Dr. Hélène Hardin-Pouzet for their help. The authors are thankful to referees of this journal for their constructive comments and would like also to thank Prof. Rachid Boukhliq and Prof. Arshad Khan for having improved the English language of the paper.





Zigmond, M. J., Cameron, J. L., Hoffer, B. J., and Smeyne, R. J. (2012). Neurorestoration by physical exercise: moving forward. Parkinsonism Relat. Disord. 18, S147–S150. doi: 10.1016/S1353-8020(11)70046-3

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 El Allali, Achaâban, Piro, Ouassat, Challet, Errami, Lakhdar-Ghazal, Calas and Pévet. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Brain of the Black (Diceros bicornis) and White (Ceratotherium simum) African Rhinoceroses: Morphology and Volumetrics from Magnetic Resonance Imaging

#### Adhil Bhagwandin<sup>1</sup> , Mark Haagensen<sup>2</sup> and Paul R. Manger<sup>1</sup> \*

<sup>1</sup> School of Anatomical Sciences, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa, <sup>2</sup> Department of Radiology, Wits Donald Gordon Medical Centre, University of the Witwatersrand, Johannesburg, South Africa

#### Edited by:

Nilesh Bhailalbhai Patel, University of Nairobi, Kenya

#### Reviewed by:

Bruno Cozzi, University of Padua, Italy Cristiano Bombardi, Università di Bologna, Italy Annamaria Grandis, Dipartimento di Scienze Mediche Veterinarie, Università di Bologna, Italy

> \*Correspondence: Paul R. Manger paul.manger@wits.ac.za

Received: 06 June 2017 Accepted: 15 August 2017 Published: 31 August 2017

#### Citation:

Bhagwandin A, Haagensen M and Manger PR (2017) The Brain of the Black (Diceros bicornis) and White (Ceratotherium simum) African Rhinoceroses: Morphology and Volumetrics from Magnetic Resonance Imaging. Front. Neuroanat. 11:74. doi: 10.3389/fnana.2017.00074 The morphology and volumetrics of the understudied brains of two iconic large terrestrial African mammals: the black (Diceros bicornis) and white (Ceratotherium simum) rhinoceroses are described. The black rhinoceros is typically solitary whereas the white rhinoceros is social, and both are members of the Perissodactyl order. Here, we provide descriptions of the surface of the brain of each rhinoceros. For both species, we use magnetic resonance images (MRI) to develop a description of the internal anatomy of the rhinoceros brain and to calculate the volume of the amygdala, cerebellum, corpus callosum, hippocampus, and ventricular system as well as to determine the gyrencephalic index. The morphology of both black and white rhinoceros brains is very similar to each other, although certain minor differences, seemingly related to diet, were noted, and both brains evince the general anatomy of the mammalian brain. The rhinoceros brains display no obvious neuroanatomical specializations in comparison to other mammals previously studied. In addition, the volumetric analyses indicate that the size of the various regions of the rhinoceros brain measured, as well as the extent of gyrification, are what would be predicted for a mammal with their brain mass when compared allometrically to previously published data. We conclude that the brains of the black and white rhinoceros exhibit a typically mammalian organization at a superficial level, but histological studies may reveal specializations of interest in relation to rhinoceros behavior.

Keywords: rhinoceros, Rhinocerotidae, Perissodactyla, central nervous system, surface anatomy

## INTRODUCTION

The fossil record of Perissodactyla places its earliest members in the upper Palaeocene (Radinsky, 1969), yet molecular studies suggest that the Perissodactyla diverged from Cetartiodactyla 97.5 – 83.4 million years ago (Eizirik et al., 2001; Price and Bininda-Emonds, 2009) or Carnivora approximately 80 million years ago (Springer et al., 2003). The order Perissodactyla is comprised

of the Rhinoceritidae, Tapiridae, and Equidae families. The Rhinoceritidae includes five extant species: the black (Diceros bicornis) and white (Ceratotherium simum) African rhinoceroses, the Indian rhinoceros (Rhinoceros unicornis), the Javan rhinoceros (Rhinoceros sondaicus) and the Sumatran rhinoceros (Dicerorhinus sumatrensis) (Wilson and Reeder, 1993). After much debate regarding the phylogenetic relationships of these living species, it has been established that the Asian and African lineages diverged approximately 26 million years ago (Tougard, 2001).

The black and white rhinoceros are iconic species of the African continent with the white rhinoceros being the second largest terrestrial mammal after the African elephant. Observational studies of the African rhinoceroses have revealed some interesting behaviors, including aggression (Leuthold, 1977), a complex social structure (Owen-Smith, 1971), ear movements signaling intimidation (Leuthold, 1977), acute olfactory capabilities (Cave, 1966), and differences in the duration of a sleep bout between males and females (Santymire et al., 2012). For the most part, these behavioral studies do not refer to the structure (and inferred functional capacities) of the rhinoceros brain, as the information required to make this sort of interpolation is not available. Studies of rhinoceros brains would begin to unlock the neural architecture subserving the observed behaviors and may provide clues leading behavioral studies in new directions, providing a deeper understanding of rhinoceros behavior.

Very little is known about the structure, and through comparative implication, functional capacities, of the rhinoceros brain, although substantially more is known about the closely related horse brain (e.g., Yoshikawa, 1968; Cozzi et al., 2014). Most recorded observations detail the anatomy of the skull and only mention the anatomy of the brain in passing (Sparrman, 1778; Burchell, 1817). To date, drawings of the superficial appearance of the Indian and Sumatran/Javan rhinoceros brain (Owen, 1850; Garrod, 1878; Beddard and Treves, 1887), a few images of a fetal rhinoceros brain (Davies, 1952), the occasional report of brain mass or other measures in these species (Manger et al., 2010; Manger, 2011), and the topography of the retinal ganglion cells and visual acuity (Pettigrew and Manger, 2008; Coimbra and Manger, 2017) is all that is presently known of the rhinoceros brain. In the current paper we begin to fill this gap in our knowledge by providing descriptions of the external and internal structure of the brain of both the black and white African rhinoceros using direct observation of the brain surface and magnetic resonance imaging. We also provide volumetric analyses of various regions of the brain for both species and compare them to data previously published for other mammals.

## MATERIALS AND METHODS

## Specimens

One brain of a black (D. bicornis) and a white (C. simum) African rhinoceros was used in the current study. A 40 year-old male black rhinoceros, obtained from the National Zoological Gardens, Pretoria, South Africa, was euthanized due to health complications following an intractable gastrointestinal illness, while the 3 year-old female white rhinoceros, obtained from Wildlife Assignments International, Hammanskraal, South Africa, was euthanized following irreparable damage to the right radial nerve after an attempted poaching incident. The animals were initially immobilized with a dose of etorphine (approximately 5 mg, i.m.), following which they were given an intravenous overdose of sodium pentobarbital (approximately 40 mg/kg, i.v.). Following cessation of the heartbeat, the head and neck was dissected free from the body and perfused using a gravity feed via the paired carotid arteries initially with a 20 l rinse of 0.9% saline to flush the remaining blood followed by 40 l of fixative (4% paraformaldehyde in 0.1M phosphate buffer) (Manger et al., 2009). The brains were then removed from the skull and post-fixed for 48 h in the fixative solution at 4◦C, after which they were placed in a 30% sucrose in 0.1M phosphate buffer solution until equilibrated (approximately 7 days). Following this, the brains were placed in an antifreeze solution for 7 days at 4◦C, before being stored at −20◦C until use (Manger et al., 2009). Both animals were treated according to the guidelines of the University of the Witwatersrand Animal Ethics Committee (clearance certificate number 2008/36/1), which parallel those of the NIH for the use of animals in scientific experiments.

## Magnetic Resonance Imaging

The brains of both rhinoceroses were scanned in coronal, sagittal, and horizontal planes. The specimens were scanned on a Phillips 1.5 Tesla Intera System (Eindhoven, The Netherlands), using all three elements of the head and neck coil. The brains were removed from their containers, drained of excess fluid and placed in the head coil wrapped in a dry sheet, thus being exposed directly to air, which also partly entered the ventricles. After testing different scan parameters the following sequence was selected as giving the best detail and the least artifact (especially at the air-fluid interfaces). The selected T1 weighted inversion recovery sequence consisting of 2 mm slices without gap, had a TR (time to repeat) of between 6.5 and 10.9 ms and a TI (time to invert) of 300 ms. The number of signal averages varied between 3 and 4 with a flip angle of 90◦ and an echo train of 10. The scan times varied between 15 and 25 min. The antifreeze liquid in which the brains were stored showed high signal on both T1 and T2 weighted sequences and the routine clinical T1 and T2 sequences produced very similar T2 like images of the brain specimens. This is possibly related to the lack of water in the tissues of the specimen secondary to the fixation and storage process (see above). The images were processed using the freely available open source software program Osirix (Rosset et al., 2004) 1 .

## Volumetrics

In the current paper, the MRI scans allowed us to determine volume measurements for a variety of identifiable brain structures. These included the mid-sagittal cross-sectional area of the corpus callosum, the volume of the hippocampus, amygdala,

<sup>1</sup>www.osirix-viewer.com

cerebellum, and lateral ventricles and the gyrencephalic index (GI) for both species of African rhinoceros. We used the methodology of Manger et al. (2010) to calculate the volume of the corpus callosum, Patzke et al. (2015) to calculate the volume of the hippocampus and amygdala, Maseko et al. (2012) to calculate cerebellar volume, Maseko et al. (2011) to calculate the volume of the ventricular system and of Zilles et al. (1989); Pillay and Manger (2007), and Manger et al. (2012) to calculate the GI of the cerebral cortex.

## RESULTS

The male black rhinoceros had a brain mass of 531 g while that of the female white rhinoceros was 536.5 g. Unfortunately, due to a lack of necessary heavy-duty equipment we were unable to obtain accurate measures of the body mass of each of these individuals. Previous reports indicate that the average body mass of a male black rhinoceros is 852 kg (Hitchins, 1968) and the average body mass of a female white rhinoceros is 1600 kg (Kirby, 1920). Using these average body masses, we could calculate the encephalization quotients (the relative mass of the brain compared to the mass of the body, using the regression provided in Manger, 2006) for each individual. The least squares regression provided by Manger (2006) was calculated using data from 271 mammal species, covering most mammalian orders, but excluded data from both primates and cetaceans, and did not use any phylogenetic corrections methods. Thus, the encephalization quotient for the male black rhinoceros was 0.469, while that for the female white rhinoceros was 0.299, both substantially below the expected encephalization quotient of 1 for mammals. The particularly low encephalization quotient for the female white rhinoceros may be due to the young age (3 years old) of this specimen. As we could not obtain a body mass from this specific specimen, we used the average adult female body mass to calculate the encephalization quotient as data on the average body mass of wild white rhinoceroses at different ages is not available. Thus, the encephalization quotients provided here must be interpreted cautiously when used in a comparative sense.

## Observable Features on the Surface of the Brain

The various subdivisions of the rhinoceros brain are clearly demarcated, and follow the organizational pattern generally observed in other mammals (**Figures 1**, **2**). Briefly, two moderately sized olfactory bulbs are found anterior to the two distinct and gyrencephalic cerebral hemispheres. Caudal to the hemispheres the cerebellum, with a distinctly asymmetrical vermal portion, overlies the brainstem (midbrain, pons, and medulla oblongata). In general, the superficial appearance of the black and white rhinoceros brains is very similar, but the black rhinoceros brain appears to be shorter rostrocaudally, wider mediolaterally, and taller dorsoventrally than the white rhinoceros brain (**Figures 1**, **2**). This gives the black rhinoceros brain a mild globular appearance in comparison to the while rhinoceros brain, which appears more elongated in the rostrocaudal dimension. We measured the cerebral hemisphere

of each rhinoceros specimen to compared this impression and found that the cerebral hemisphere in the black rhinoceros was 9.3 cm long in the rostrocaudal dimension, compared to 10.7 cm in the white rhinoceros. Dorsoventrally, the black rhinoceros cerebral hemisphere was 7.1 cm, while that of the white rhinoceros was 6.7 cm, and mediolaterally, the black rhinoceros hemisphere measured 10.4 cm, while that of the white rhinoceros measured 10 cm. Thus, these measurements confirm the qualitative impression. However, as we only have one specimen from each species, this may be an individual difference rather that a species-specific difference, but the variation between these two individuals is quite clear. Despite this slight variation in global appearance, both species show similar overall morphology, thus, the description provided herein applies to both species, except where noted.

The most rostral neural structures were the paired olfactory bulbs, which arch dorsally anterior to the rostral pole of the cerebral hemisphere (**Figures 1**, **2**). Unfortunately, the olfactory bulbs of the rhinoceroses are very difficult to remove together with the brain due to the presence of a thick and tough fold of the dura mater between the caudal aspect of the olfactory bulb and the rostral pole of the cerebral hemisphere. Indeed, the drawings of the brain of the Javan rhinoceros provided by

FIGURE 2 | Dorsal, lateral, and ventral views of the brain of the white rhinoceros (Ceratotherium simum). Left column of images unlabelled. Right column of images, duplicates of those on the left with a number of specific structures labeled. Note the typically mammalian appearance of the brain in terms of the major anatomical subdivisions. Also note the very asymmetrical appearance of the vermal portion of the cerebellum. Scale bar = 1 cm. h, hemispheric portion of cerebellum; lot, lateral olfactory tract; mb, mammillary bodies; mcp, middle cerebellar peduncle; mo, medulla oblongata; NEO, cerebral neocortex; OB, olfactory bulb; opt, optic tract; OT, olfactory tubercle; ot, olfactory tract; p, pons; PC, cerebral peduncle; PIR, piriform cortex; rs, rhinal sulcus; sc, spinal cord; sf?, potential sylvian fissure; tc, tuber cinereum; v, vermal portion of cerebellum.

Beddard and Treves (1887, plate 37) does not show any olfactory bulbs or even remnants of the bulbs and olfactory tract, with this region of the brain "smoothed" over in the drawings. The olfactory bulbs do not appear to be either overly large, or reduced in size. From the ventrocaudal aspect of the olfactory bulbs a wide (around 15 mm mediolateral) olfactory tract courses caudally toward the brain. This tract coalesces anterior and lateral to the olfactory tubercle (which evinces the distinct arterial spaces that give the name perforated substance) to form a distinct lateral olfactory tract. The lateral olfactory tract appears to invest into a clearly demarcated piriform lobe, which is separated from the neocortex by a distinct rhinal sulcus (which can be followed from the olfactory bulb caudally). The piriform cortex is observed lateral to the olfactory tract, expanding caudally, to form a distinct bulge on the ventral aspect of the cerebral hemisphere. Small sulci and gyri are observed throughout the piriform cortex. Although not quantified, it appears that the piriform lobe of the black rhinoceros is somewhat larger than that of the white rhinoceros, and is also seemingly more gyrencephalic (**Figures 1**, **2**).

The neocortex occupies the majority of the lateral, dorsal, and mesial surfaces of the brain. The neocortex has numerous sulci and gyri, and these are generally not symmetrical in appearance, i.e., the left and right hemispheres show very different patterns. The majority of the gyri have a width of around 5–7 mm and the sulci do not appear to form long continuous furrows in any plane. There does not appear to be any specifically distinct sulci or gyri that can be conclusively compared to those present in other species, even closely related species such as horses. In addition, our MR images did not assist in the determination of specific sulci and gyri, as many of the sulci are quite deep and do not appear to have enough continuity to allow a specific nomenclature to be applied to them for comparison (**Figures 3**–**6**). In order to avoid confusion, and the potential incorrect assignation of associated functional regions, we have thus not named the sulci and gyri present in the neocortex of the rhinoceros, which was also avoided in earlier descriptions of the brains of other rhinoceros species (Owen, 1850; Garrod, 1878; Beddard and Treves, 1887). While one may be tempted to label a sylvian fissure in the lateral view of the brain of both species (**Figures 1**, **2**) and a possible cruciate sulcus in the rostral dorsal aspect of the black rhinoceros brain (**Figure 1**), at best these identifications would be very tentative.

In our whole brain specimens only the most ventral aspect of the diencephalon is visible, as is typical for mammals (**Figures 1**, **2**). The rostral border of the diencephalon is demarcated by the presence of the optic chiasm and optic tracts. These are not large in size, commensurate with the low number of retinal ganglion cells in the rhinoceroses (Pettigrew and Manger, 2008; Coimbra and Manger, 2017); however, it should be noted that the left optic tract of the black rhinoceros (**Figure 1**) is smaller than the right optic tract, and this appears to be associated with the loss of an eye from this male black rhinoceros 5 years prior to euthanasia following a fight with another male black rhinoceros. Caudal to the optic tracts in the black rhinoceros a distinct tuber cinereum, with associated ventricular space can be observed (**Figure 1**), while in the white rhinoceros, the tuber cinereum appears to have been lost during removal of the brain and the third ventricle is apparent (**Figure 2**). Caudal to the tuber cinereum in the black rhinoceros is a distinct swelling demarcating the caudal aspect of the diencephalon, which appears to be the mammillary bodies (**Figure 1**). The mammillary bodies, while present, appear to be less distinct in the white rhinoceros (**Figure 2**).

Continuing caudally from the diencephalon on the ventral surface of the brain, the floor of the midbrain is represented by rostrocaudally elongated cerebral peduncles with a distinct and also elongated interpeduncular fossa clearly visible, as well as the ventral median sulcus (**Figures 1**, **2**). In keeping with the comparatively elongated nature of the white rhinoceros brain compared to the black rhinoceros brains, the cerebral peduncles are longer in the rostrocaudal direction in the white rhinoceros. At the caudal end of the cerebral peduncles the ventral aspect of the pons bulges ventrally from the surface and exhibits the typical mediolaterally oriented fiber pathways, or stria, typical of mammals. These stria coalesce laterally to form the distinct middle cerebellar peduncles which enter the ventral aspect of the cerebellum. Emerging from the lateral aspect of the pons is the root of the trigeminal nerve, while the caudal border of the pons is marked by the coalesced roots of the facial and vestibulocochlear nerves (**Figures 1**, **2**). Caudal to the pons, a broad medulla oblongata is observed, with visible pyramidal tracts and small

FIGURE 3 | A series of coronal structural MR images through the brain of the black rhinoceros (D. bicornis). (A) is the most rostral section and (V) is the most caudal section, with each section having a thickness of 2 mm and each section being 6 mm apart. Note the typically mammalian topography of the various regions and structures of the brain. Scale bar = 2 cm. Amyg, amygdaloid body; C, caudate nucleus; ca, cerebral aqueduct; Cb, cerebellum; cc, corpus callosum; DT, dorsal thalamus; EC, entorhinal cortex; f, fornix; Hip, hippocampus; Hyp, hypothalamus; IC, inferior colliculus; LV, lateral ventricle; N.Acc, nucleus accumbens; NEO, neocortex; mcp, middle cerebellar peduncle; mo, medulla oblongata; OB, olfactory bulb; Olf. Tub, olfactory tubercle; P, putamen; PC, cerebral peduncle; Pir, piriform cortex; py, pyramidal tract; S, septal nuclear complex; SC, superior colliculus; spc, cervical spinal cord; VPO, ventral pontine nucleus.

laterally placed bulges representing the inferior olivary nuclear complex. A distinct ventral median sulcus that continues into the spinal cord is also present. The medulla oblongata tapers toward the spinal cord, becoming less broad caudally until the spinomedullary junction is reached, although this junction is not clearly marked by the decussation of the pyramidal tracts. The cephalic end of the cervical spinal cord appears to maintain a consistent cross-sectional area for at least 20 cm caudal to the foramen magnum, which is where we sectioned through the spinal cord to remove the brain.

The dorsal surface of the brainstem is completely covered by the cerebellum (**Figures 1**, **2**). The cerebellum shows clear folia and is slightly broader mediolaterally in the white rhinoceros compared to the black rhinoceros. A distinct flocculus is not apparent, but the vermal and hemispheric portion of the cerebellum can be seen. Interestingly, in both species the vermis, while being located at the midline, is highly asymmetrical and forms asymmetrical borders with the cerebellar hemispheres. Indeed, the cerebellum of both species appears somewhat loosely and asymmetrically organized when compared to many other mammalian species.

## Internal Structures of the Brain Apparent in MR Images

The MR images reveal substantial detail regarding the internal organization of the brains of both species of rhinoceros and, as for the external surface of the brain, the organization of the internal aspects of the brain revealed with MR imaging are similar between the two species (**Figures 3**–**6**). Due to this similarity, the following description applies to both species unless otherwise stated. For the most part, the internal organization of the rhinoceros brain is what would be considered typically mammalian in that the regular subdivisions of the brain are apparent. The remnants of the olfactory bulbs are evident in the MR images, and within these remnants a distinct olfactory ventricle is found. At its caudal most aspect the olfactory ventricle communicates with the lateral ventricle via a small duct. Although the olfactory tract is not readily visible in the MR images, the olfactory tubercle and piriform cortex are apparent. The piriform cortex can be followed caudally where it expands ventrally to cover the ventral aspects of both the amygdaloid and hippocampal complexes.

The MR images reveal the depths of the numerous sulci, many of which can be up to 2 cm deep. The images confirm the difficulty in the precise identification of sulci and gyri by examining the surface of the brain and unfortunately do not provide additional information. Within the telencephalon, the corpus striatum, septal nucleus complex, basal forebrain, amygdaloid body, hippocampus, claustrum, and lateral ventricles were readily observed. The corpus striatum occupied the position typically found in mammals, with a clear head of the caudate nucleus being observed rostral to the splitting of this nucleus into the caudate and putamen by the internal capsule. The body of the caudate nucleus could be followed caudally along the dorsal lateral wall of the lateral ventricle (**Figures 3E–H**, **4E–J**, **5C,D**, **6C,D**). No clear globus pallidus was observed with the MR

images. The septal nuclear complex is apparent as a small gray matter bulge on the medial wall of the cerebral hemisphere at the level of the decussation of the anterior commissure (**Figures 3G**, **4G,H**). No clear subdivisions of the septal nuclear complex were evident in the MR images. Ventral to the septal nuclear complex, between it and the olfactory tubercle the region where the diagonal band of Broca should be found histologically was observed in the MR images, and immediately lateral to this was the nucleus accumbens (located just anterior and ventral to the head of the caudate nucleus). Immediately caudal and ventral to the posterior pole of the putamen, within the medial aspect of the temporal lobe, the amygdaloid complex could be observed. Both nuclear and cortical portions of the amygdaloid nucleus could be seen, but greater detail in terms of the subdivisions of the amygdaloid complex were not visible (**Figures 3H,I**, **4I,J**, **5E,F**, **6B**). The hippocampus occupied a typically mammalian position within the telencephalon (**Figures 3J–L**, **4K–M**, **5C–F**, **6B–E**), with the temporal pole of the hippocampus located caudal to the amygdaloid complex. The amygdaloid complex and the temporal horn of the hippocampus are separated by an incursion of the lateral ventricle. A distinct vertically oriented ventral portion of the hippocampus located lateral to the diencephalon, before turning forward to form the dorsal portion of the hippocampus lying over the diencephalon. From the most rostral part of the dorsal hippocampus the fornix was readily seen to turn ventral and invest into the hypothalamus and mammillary bodies. The lateral ventricles occupied a position typical of mammals within the telencephalon, having a frontal horn (connected to the olfactory ventricle), a body overlying the corpus striatum and diencephalon, a small temporal horn dorsal to the hippocampal formation in the temporal lobe, but no clear occipital horn could be observed. The ventricles of the white rhinoceros appear somewhat larger than those of the black rhinoceros (**Figures 3**–**6**).

The diencephalon is found in the caudal ventral half of the cerebral hemisphere, and is united across the midline by a large interthalamic adhesion. The rostral border of the diencephalon coincides with the optic chiasm, while the caudal border coincides with the appearance of the superior colliculi of the midbrain (**Figures 3H–M**, **4H–M**, **5A–D**, **6A–C**). The dorsal thalamus, hypothalamus and parts of the epithalamus and ventral thalamus are evident in the MR images. The dorsal thalamus appears as a continuous large cylinder of gray matter

various regions and structures of the brain. Scale bar = 2 cm. Amyg, amygdaloid body; C, caudate nucleus; ca, cerebral aqueduct; Cb, cerebellum; cc, corpus callosum; DT, dorsal thalamus; f, fornix; Hip, hippocampus; Hyp, hypothalamus; IC, inferior colliculus; NEO, neocortex; Mb, midbrain; mcp, middle cerebellar peduncle; mo, medulla oblongata; Olf. Tub, olfactory tubercle; P, putamen; Pir, piriform cortex; SC, superior colliculus; scp, superior cerebellar peduncle.

extending the entire rostrocaudal length of the diencephalon. No clearly demarcated internal medullary lamina was evident in the MR images, prohibiting further parcellation of the dorsal thalamus into component parts. The hypothalamus is found below the dorsal thalamus, and for the most part surrounding the lower portions of the third ventricle. The caudal border of the hypothalamus is marked by the presence of the mammillary bodies, and does not extend the full rostrocaudal extent of the diencephalon. The fornix can be observed in the hypothalamus, dividing the hypothalamus into approximately equal sized medial and lateral zones (**Figure 3I**). The only portions of the epithalamus apparent in the MR images are the two habenular complexes located dorsal and caudal to the dorsal thalamic nuclear mass.

The MR images reveal the three cerebellar peduncles (middle most clearly), as well as the extent of the foliation of the cerebellar cortex and the size of the cerebellar white matter, but the deep cerebellar nuclei are not apparent (**Figures 3P,Q**, **4P**, **5D**). The asymmetry of the vermal region of the cerebellum is clear in the MR images, but the size and/or extent of the flocculus is inconclusive. Several parts of the midbrain, including the superior colliculi, the inferior colliculi, the cerebral aqueduct, periaqueductal gray matter, midbrain tegmentum, and the cerebral peduncles were readily observed in the MR images (**Figures 3M,N**, **4M–O**, **5B,C**, **6A–C**). Vague gray regions dorsal and medial to the

the black rhinoceros (D. bicornis). (A) is the most ventral section and (H) is the most dorsal section, with each section having a thickness of 2 mm and each section being 6 mm apart. Note the typically mammalian topography of the various regions and structures of the brain. Scale bar = 1 cm. Amyg, amygdaloid body; C, caudate nucleus; Cb, cerebellum; DT, dorsal thalamus; Hip, hippocampus; Hyp, hypothalamus; IC, inferior colliculus; LV, lateral ventricle; NEO, neocortex; Olf. Tub, olfactory tubercle; P, putamen; PC, cerebral peduncle; Pir, piriform cortex; SC, superior colliculus.

cerebral peduncles may represent the ventral tegmental area and the substantia nigra, but this needs immunohistochemical staining to verify. Within the pons the pontine tegmentum and ventral pontine nuclei can be identified. Rostrally in the medulla oblongata, at the ventral midline, the pyramidal tracts are clearly seen, however, they become less obvious caudally in the medulla oblongata. Occasional patches of gray matter are observed in the medulla oblongata, but to ascribe these to specific structures is difficult without histological verification. Lastly, as the level of the spinomedullary junction, the decussation of the pyramidal tract is apparent (**Figures 3U,V**, **4V**).

## Volumetric Analyses

Using the methodology described in Zilles et al. (1989); Pillay and Manger (2007), and Manger et al. (2012), we calculated the GI of both rhinoceros. For the black rhinoceros we calculated a GI of 2.47, while for the white rhinoceros we found a GI of 2.44. We compared these values to a regression calculated in Manger et al. (2012) based on the GI of primates, carnivores

and artiodactyls, and found that both species of rhinoceros had GIs that fell slightly above this regression, but well within the 95% prediction intervals (**Figure 7**; Manger et al., 2012). Thus, the extent of gyrencephaly in the rhinoceros brain appears to be very close to what one would predict for a mammal based on their brain mass. The mid-sagittal cross-sectional area of the corpus callosum of the male black rhinoceros was found to be 2.52 cm<sup>2</sup> , while that of female white rhinoceros was 2.19 cm<sup>2</sup> . These values are very close to what would be predicated based on regressions derived from other non-primate mammals and fall well within the 95% prediction intervals based on the data for other mammals (**Figure 7**; Manger et al., 2010).

The volume of the hippocampus in the black rhinoceros was found to be 4.71 cm<sup>3</sup> , while that of the white rhinoceros was 3.97 cm<sup>3</sup> . The hippocampal volume in both rhinoceros species is very close to what would be predicted based on regressions derived from other mammals, with the black rhinoceros falling within the 95% confidence intervals of the mammalian regression, and both species falling within the 95% prediction intervals of the mammalian regression (**Figure 7**; Patzke et al., 2015). The volume of the amygdala in the black rhinoceros was found to be 3.10 cm<sup>3</sup> , which was markedly larger than that for the white rhinoceros, which had an amygdalar volume of 1.67 cm<sup>3</sup> . Despite this difference in size, the amygdalar volume of both species fell within the 95% prediction intervals based on the regression developed for other mammals, but both fell outside the 95% confidence intervals for this regression (**Figure 7**; Patzke et al., 2015). Interestingly, the black rhinoceros amygdalar volume is slightly larger than you would expect for a mammal with its brain mass, while the white rhinoceros has an amygdalar volume that is somewhat smaller than you would expect for its brain volume. This may be related to the age of the individual specimens, with the black rhinoceros being markedly older than the white rhinoceros, or it may reflect a true species difference.

The black rhinoceros was found to have a cerebellar volume of 37.73 cm<sup>3</sup> , and the white rhinoceros had a cerebellar volume of 44.50 cm<sup>3</sup> . Using a regression based on primate, megachiropteran, and insectivore cerebellar volumes (Maseko et al., 2012), it was found that the volume of the cerebellum for both species of rhinoceros fell within the 95% prediction intervals, but below the 95% confidence intervals (**Figure 7**; Maseko et al., 2012). Thus, in comparison to the species used to create the regression, the size of the rhinoceros cerebellum is on the low side, but within than the range that would be predicted for their brain mass. The total ventricular volume of the black rhinoceros brain was calculated to be 9.86 ml, while in the white rhinoceros the total ventricular volume was found to be 12.05 ml. When comparing these volumes to those obtained in other mammals (Maseko et al., 2011), it is clear that both rhinoceros have total ventricular volumes that fall within the ranges seen for other mammals, having neither a particularly large, or particularly small ventricular system for their respective brain masses (**Figure 7**; Maseko et al., 2011).

## DISCUSSION

The current description and analysis of the gross morphological features of the external and internal aspects of the brains of two species of rhinoceros indicate that the rhinoceros brain is typically mammalian in its general structure, organization, and relative topology of the component parts of the brain. In addition, volumetric analyses of specific aspects of the rhinoceros brains supports this overall conclusion, with no specific regions of the brain investigated being larger or smaller than predicted for a mammal of the brain masses exhibited by the rhinoceroses studied. There were some notable qualitative differences between the two species, with the black rhinoceros having a more globular appearance of the brain, larger mammillary bodies and a larger, more distinct, piriform lobe than the white rhinoceros, while the white rhinoceros had a larger ventricular system than the black rhinoceros. It should also be noted here that the black and white rhinoceros form the African branch of the Rhinocerotidae and despite exact dates not being available for the time of divergence between these two species (although more recent than 20 million years ago, Norman and Ashley, 2000), they are each other's closest extant relatives (Price and Bininda-Emonds, 2009). In this sense it is not surprising that the general appearance of the brains of these two species is very similar.

## The Typically Mammalian Organization of the Rhinoceros Brain

As outlined, the general appearance of the brain of the two rhinoceros individuals studied is very typically mammalian, which is not a surprising finding. In addition, no specific specializations of the brain were observable at the level of analysis undertaken, and major organizational changes, in regards to proportions of different brain regions, the topological organization and appearance, was limited to the lack of an identifiable cortical sulcal and gyral pattern and the asymmetrical vermal region of the cerebellum in the rhinoceroses. Due to the variance of the sulcal and gyral pattern across the four hemispheres available for study, variation supported by the MR images, we avoided labeling specific sulci and gyri as this might infer functional aspects of the cerebral cortex that may be incorrect. Indeed, Garrod (1878, p. 411) states: "So complicated and numerous are the convolutions that the general type-plan of their disposition is to a considerable extent disguised." This lack of a clear pattern in the anatomy of the sulci and gyri appears to be specific to the rhinoceroses, as several sulci and gyri that are apparently homologous to those in other mammalian species, are readily apparent on the horse brain (Edinger, 1948; Pascalau et al., 2015). In addition, the vermal region of the rhinoceros cerebellum is very asymmetrical, which is also not often seen across mammalian species, and horses show a distinctly symmetrical vermis (Bradley, 1899). This would mean that a full mid-sagittal section through the vermis of the rhinoceroses would not produce the classical arbor vitae appearance of the cerebellum. This vermal asymmetry is more marked in the black rhinoceros specimen than the white rhinoceros specimen. Interestingly, in their depictions of the

derived on data from other mammals previously reported. Note that for all parameters measured the relative size of these parameters are close to what would be predicted based on the data from other mammals. Thus, the relative proportions of the brain of the rhinoceroses appear to be typically mammalian.

Sondiac rhinoceros brain (Beddard and Treves, 1887, plate 37) and Sumatran rhinoceros brain (Garrod, 1878, plate LXX) the authors show a very symmetrical cerebellum, indicating that potentially the asymmetrical vermis is a derived feature of the African rhinoceroses, although as noted for the olfactory bulbs, these diagrams appeared to be somewhat enhanced compared to the anatomical reality of the rhinoceros brain. Despite these two variations, the remainder of the rhinoceros brains can be readily navigated by those with a basic understanding of the gross anatomy of larger mammalian brains.

## Qualitative Differences between Black and White Rhinoceros Brains

In the current analysis we noted four specific qualitative differences between the brain of the black and white rhinoceros investigated, including the more globular appearance of the black

rhinoceros brain, the larger mammillary bodies and piriform lobe of the black rhinoceros brain, and the larger ventricular system of the white rhinoceros brain. While we note these differences as differences between the species generally, the strength of the emphasis we place on these differences must be measured by the fact that we only have one brain of each species, therefore the differences might be those of individuals rather than species. However, given that the acquisition of these brains was truly random (in that there were no selection criteria for the specific individuals apart from veterinary necessity), it is likely that the individual brains examined are representative of the species as a whole, and therefore the differences are likely to be specific species differences.

It is clear that the brain of the black rhinoceros is shorter rostrocaudally, but broader mediolaterally and taller dorsoventrally than the white rhinoceros brain (**Figures 1**, **2**). This makes the black rhinoceros brain appear somewhat globular in comparison to the white rhinoceros brain. Whether this difference is due to changes occurring in the evolutionary history of the black or the white rhinoceros is unclear – we don't have a description of the brain of the last common ancestor of these two species to compare to the extant species. Despite this, the skull being shorter and broader in the black rhinoceros, and longer and narrower in the white rhinoceros are considered key indicators of the two genera (Ansell, 1974). Thus, the shape of the brains reflects the overall shape of the skulls, which in turn are thought to reflect the browsing (black rhinoceros) versus grazing (white rhinoceros) feeding habits of the two species. Palaeoneurological examination of fossil specimens of the two lineages may be able to determine whether the brain became shortened, or lengthened, or both during the evolution of the extant African rhinoceros species.

The mammillary bodies of the black rhinoceros formed distinct ventral bulges at the caudal end of the diencephalon, whereas, in the whole brain specimens the mammillary bodies were not easy to discern in the white rhinoceros. The larger size of the mammillary bodies in the black rhinoceros is commensurate with the slightly larger size of the hippocampus in the black rhinoceros (4.71 cm<sup>3</sup> ) compared to the white rhinoceros (3.97 cm<sup>3</sup> ), despite the black rhinoceros having a slightly smaller brain than the white rhinoceros. The mammillary bodies, which receive strong input from the hippocampus (via the fornix), are known to be heavily involved in the process of spatial memory and the relationship of head position to cognitive spatial maps (Dillingham et al., 2015). This indicates that, for at least the parts we have identified to date, the overall neural navigation system of the black rhinoceros is larger than that of the white rhinoceros. Male white rhinoceroses are territorial with home ranges of 0.75 to 13.9 km<sup>2</sup> , while female white rhinoceroses have larger home ranges of 3 to 45.23 km<sup>2</sup> (Skinner and Chimimba, 2005). In contrast, male black rhinoceroses are not strictly territorial and as a species have reported home ranges from as small as 0.5 km<sup>2</sup> to as large as 500 km<sup>2</sup> (Skinner and Chimimba, 2005). It appears then that the difference in home range size might account for the larger mammillary bodies in the black rhinoceros. In addition, the black rhinoceroses browse on a greater number of plant species, in a more enclosed environment, that vary across the seasons more dramatically than the grasses grazed by the white rhinoceros (Skinner and Chimimba, 2005). Therefore, in their larger home ranges in a denser wooded habitat, it would be more important for the black rhinoceroses to have a better cognitive map of the location and timing of available food sources than is needed for the grazing diet of the white rhinoceroses, and this may underlie the enlargement of the mammillary bodies in the black rhinoceros.

The piriform cortex of the black rhinoceros appears to be around 125% larger and more gyrencephalic than that of the white rhinoceros, indicating that the processing of olfactory information is of more relevance to the black rhinoceros than the white rhinoceros. Unfortunately, as mentioned earlier, the olfactory bulbs were difficult to remove intact along with the brain, meaning that we don't have a direct comparison of olfactory bulb size to make between the two species to support the difference observed in piriform cortex size. This potentially increased reliance on olfaction is likely linked to the diet, both in number of plants eaten and their seasonal variability (Skinner and Chimimba, 2005), of the black rhinoceros compared to the white rhinoceros.

The last potential interspecies difference noted was the larger size of the ventricular system in the white rhinoceros compared to the black rhinoceros. The ventricular system of the white rhinoceros is 2.19 cm<sup>3</sup> (or 122%) larger than that of the black rhinoceros, and this is most evident in the lateral ventricles. Again, this difference may be related to the specific diets of the species, with some of the grasses eaten by the white rhinoceros potentially containing enough toxins that would necessitate the need for larger ventricles to assist in the flushing of toxins from the central nervous system. While the ventricular systems of both species of rhinoceros are not larger than that seen in many other mammals, this interspecies difference is readily apparent.

## Future Studies of Rhinoceros Brains

While the brains of these two rhinoceroses are clearly mammalian, and certain differences are apparent between the two individuals at the gross level investigated herein, it is likely that examination at the microscopic level will reveal additional species-specific features, derived features specific to the Rhinocerotidae, and derived features specific to the Ceratomorpha, and/or Perissodactyla, and/or Laurasiatheria. For example, the presence of a parvocellular cluster of orexinergic neurons in the medial hypothalamus in Cetartiodactyls (Dell et al., 2012, 2017a,b) and African elephants (Maseko et al., 2013) is also likely to be present due to the dietary requirements and phylogenetic affinities of the rhinoceroses. Despite this, it is most likely that a large suite of features common to all mammals will be observed, although variations in the proportional and absolute sizes, as well as neuronal numbers are likely to be found. Thus, our intention is to section these two brains coronally and undertake basic neuroanatomical staining (Nissl and myelin) as well as a suite of immunohistochemical stains to reveal specific and non-specific neural systems and neural structures. It is hoped that by doing this we will be able to relate the neuroanatomy of the rhinoceroses to observed behaviors (as briefly touched

upon in the previous section), and perhaps make suggestions for future behavioral observations based on our findings. The overall aim of this process is to help develop a better understanding of the rhinoceros brain and behavior to provide information of relevance to both conservation and management of these iconic African species.

## AUTHOR CONTRIBUTIONS

AB and PM collected the brains. MH undertook the MR imaging of the brains. AB and PM prepared the draft of the manuscript, which was edited by MH.

## REFERENCES


## FUNDING

The research reported herein was funded by the National Research Foundation of South Africa (grant number 93610 to AB).

## ACKNOWLEDGMENTS

The authors are grateful to the National Zoological Gardens of South Africa (Pretoria, South Africa) and Wildlife Assignments International (Hammanskraal, South Africa) for allowing us to harvest the tissue from the two rhinoceros specimens.



focus on the fossil and extant faunas from Thailand. Palaeogeog. Palaeoclimatol. Palaeoecol. 168, 337–358. doi: 10.1016/S0031-0182(00) 00243-1


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Bhagwandin, Haagensen and Manger. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Putative Adult Neurogenesis in Old World Parrots: The Congo African Grey Parrot (Psittacus erithacus) and Timneh Grey Parrot (Psittacus timneh)

Pedzisai Mazengenya, Adhil Bhagwandin, Paul R. Manger and Amadi O. Ihunwo\*

School of Anatomical Sciences, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa

In the current study, we examined for the first time, the potential for adult neurogenesis throughout the brain of the Congo African grey parrot (Psittacus erithacus) and Timneh grey parrot (Psittacus timneh) using immunohistochemistry for the endogenous markers proliferating cell nuclear antigen (PCNA), which labels proliferating cells, and doublecortin (DCX), which stains immature and migrating neurons. A similar distribution of PCNA and DCX immunoreactivity was found throughout the brain of the Congo African grey and Timneh grey parrots, but minor differences were also observed. In both species of parrots, PCNA and DCX immunoreactivity was observed in the olfactory bulbs, subventricular zone of the lateral wall of the lateral ventricle, telencephalic subdivisions of the pallium and subpallium, diencephalon, mesencephalon and the rhombencephalon. The olfactory bulb and telencephalic subdivisions exhibited a higher density of both PCNA and DCX immunoreactive cells than any other brain region. DCX immunoreactive staining was stronger in the telencephalon than in the subtelencephalic structures. There was evidence of proliferative hot spots in the dorsal and ventral poles of the lateral ventricle in the Congo African grey parrots at rostral levels, whereas only the dorsal accumulation of proliferating cells was observed in the Timneh grey parrot. In most pallial regions the density of PCNA and DCX stained cells increased from rostral to caudal levels with the densest staining in the nidopallium caudolaterale (NCL). The widespread distribution of PCNA and DCX in the brains of both parrot species suggest the importance of adult neurogenesis and neuronal plasticity during learning and adaptation to external environmental variations.

Keywords: doublecortin, proliferating cell nuclear antigen, old world parrots, Congo African grey parrot, Timneh grey parrot, adult neurogenesis, cell proliferation, cell migration

## INTRODUCTION

Adult neurogenesis encompasses the birth and maturation of new neurons that become incorporated into existing circuitry or replace old and damaged neurons under normal physiological and or pathological conditions (Lindsey and Tropepe, 2006). The process of adult neurogenesis was confirmed in a variety of species from different taxa ranging from insects to humans (for reviews, see Cayre et al., 2002; García-Verdugo et al., 2002; Lindsey and Tropepe, 2006;

#### Edited by:

Nilesh Bhailalbhai Patel, University of Nairobi, Kenya

#### Reviewed by:

Basilis Zikopoulos, Boston University, United States Pavel Nemec, Charles University in Prague, Czechia Marco Aurelio M. Freire, University of the State of Rio Grande do Norte, Brazil

> \*Correspondence: Amadi O. Ihunwo

amadi.ihunwo@wits.ac.za

Received: 31 July 2017 Accepted: 15 January 2018 Published: 13 February 2018

#### Citation:

Mazengenya P, Bhagwandin A, Manger PR and Ihunwo AO (2018) Putative Adult Neurogenesis in Old World Parrots: The Congo African Grey Parrot (Psittacus erithacus) and Timneh Grey Parrot (Psittacus timneh). Front. Neuroanat. 12:7. doi: 10.3389/fnana.2018.00007

Chapouton et al., 2007). Although this process has been confirmed in wide spectrum of species, the animals studied to date represent only a small fraction of the extant species, and studying further species will allow a broader understanding of the potential diversity of the process and function of adult neurogenesis (Barnea and Pravosudov, 2011). Across many species similarities in the process of adult neurogenesis have been observed, which include the origin of neurons, the phenotype of stem cells and their proliferative mechanisms, the migration and differentiation of neurons (for a review, see Doetsch and Scharff, 2001); however, they are also important differences among species in the spatial distribution of adult born neurons, their mode of migration and phenotypic diversity (Doetsch and Scharff, 2001; Ngwenya et al., 2017).

Numerous studies have revealed widespread adult neurogenesis in non-mammalian species such as fish, amphibians, reptiles, and birds when compared to mammalian species (García-Verdugo et al., 2002; Marchioro et al., 2005; Ghosh and Hui, 2016; Macedo-Lima et al., 2016; Ngwenya et al., 2017). Generally, the rate of adult neurogenesis decreases with age across vertebrate taxa (Kuhn et al., 1996; Knoth et al., 2010; Amrein et al., 2011; LaDage et al., 2011; Ngwenya et al., 2017), and the process is affected by factors such as genetics, endogenous and exogenous factors, and seasonal variation (for a review, see Barnea and Pravosudov, 2011).

**Abbreviations:** A, arcopallium; AL, ansa lenticularis; An, nucleus angularis; APH, area parahippocampalis; Bas, nucleus basorostralis; CA, commissura anterioris; Cb, cerebellum; CbL, nucleus cerebellaris lateralis; CbM, nucleus cerebellaris medialis; OC, optic chiasm; CP, commissura posterior; CS, nucleus centralis superior (Bechterew); DBC, brachium conjunctivum decendens; DCX, doublecortin; DLL, nucleus dorsolateralis anterior thalami pars lateralis; DLP, nucleus dorsolateralis posterior thalami; DMA, nucleus dorsomedialis anterior thalami; DMM, magnicellular nucleus of the dorsomedial thalamus; DMP, nucleus dorsomedialis posterior thalami; E, entopallium; EM, nucleus ectomammillaris; EPL, external plexiform layer of the olfactory bulb; FPL, fasciculus prosencephali lateralis; GCt, substantia grisea centralis; GL, glomerular layer of olfactory bulb; GP, Globus pallidus; HA, hyperpallium apicale; HD, hyperpallium densocellulare; HI, hyperpallium intercalatus; HP, hippocampus; IGrL, internal granular layer of the olfactory bulb; Imc, nucleus isthmi, pars magnocellularis; IP, nucleus interpeduncularis; Ipc, nucleus isthmi pars parvocellularis; L1, lamina 1 of Field L; L2a, lamina 2a of Field L; La, nucleus laminaris; ll, lateral layer of the ventral hippocampus; LLV, ventral nucleus of the lateral lemniscus; LoC, locus coeruleus; LSt, lateral striatum; LV, lateral ventricle; M, mesopallium; MCL, mitral cell layer of the olfactory bulb; ml; medial layer of ventral hippocampus; MPV, nucleus mesencephalicus profundus; MSt, medial striatum; N, nidopallium; NC, nidopallium caudale; NI, nidopallium intermedium; nIII, nucleus oculomotorius; NIVL, nidopallium intermedium pars ventrolateralis; NLc, central nucleus of the lateral nidopallium; nV, nucleus motorius nervi trigemini; nX, vagus nucleus; nXIIts, nucleus nervi hypoglossis pars tracheosyringealis; OI, nucleus olivaris inferioris; OM, tractus occipitomesencephalicus; ON, olfactory nerve; OV, nucleus ovoidalis; PCNA, proliferating cell nuclear antigen; PL, nucleus pontis lateralis; PM, nucleus pontis medialis; POA, nucleus preopticus anterior; POM, nucleus preopticus medialis; PrV, nucleus sensorium principalis nervi trigemini; RP, nucleus reticularis pontis; Rt, nucleus rotundus; SCv, nucleus subcoeruleus ventralis; SGC, stratum griseum centrale; SGF, stratum griseum et fibrosum; SL, lateral septum; SLu, nucleus semilunaris; SM, medial septum; SNc, substantia nigra pars compacta; SPC, nucleus superficialis parvocellularis; SpM, nucleus spiriformis medialis; SpL, nucleus spirformis lateralis; TeO, Optic tectum; tn, nucleus taenia of the amygdala tr, triangular area of ventral hippocampus; TrO, Optic tract; TTd, nucleus et tractus descendens nervi trigemini; Tu, nucleus tuberis; VMH, ventromedial hypothalamus nucleus; VSt, ventral striatum.

African grey parrots are stocky, short tailed birds with an average body mass of 400 g. The Congo African grey parrot (Psittacus erithacus) is larger and lighter grey in color compared to the smaller and darker grey Timneh grey parrot (Psittacus timneh). The African grey parrots live a long life span ranging between 20 and 50 years in the wild, but this can increase to nearly 100 years in captivity (Carey and Judge, 2001; Schmid et al., 2006). African grey parrots lead a complex social life, coupled with long term monogamous relationships (Seibert, 2006; Emery et al., 2007). In addition to displaying complex social interactions, African grey parrots are thought to exhibit potentially advanced cognitive capabilities comparable to Apes and young humans (Pepperberg et al., 1999; Emery and Clayton, 2004; Iwaniuk et al., 2004; Pepperberg, 2006). Their relative brain size is comparable to that of primates (Olkowicz et al., 2016). Parrots feature very high neuronal densities and high total neuronal numbers particularly in the forebrain when compared primates with larger brain sizes (Olkowicz et al., 2016). These parrots display vocal learning and mimicry abilities that rival those of young humans (Iwaniuk et al., 2004; Pepperberg, 2006) and these characteristics are valuable in studies examining and comparing vocal learning and cognitive abilities in avians and primates. Parrots show cooperative problem solving and the ability to discriminate discrete and continuous variables (Al Aïn et al., 2009; Péron et al., 2011). Here we investigated the generation and maturation of new neurons throughout the brains of the two subspecies of African grey parrots, the Congo African grey parrot (Psittacus erithacus), and the Timneh grey parrot (Psittacus timneh) using PCNA and DCX immunohistochemistry.

## MATERIALS AND METHODS

## Animals and Tissue Processing

Two brains of adult male Congo African grey (Psittacus erithacus) and two brains of adult male Timneh grey parrots (Psittacus timneh) were used in the current study. The birds were purchased from a local breeder in South Africa, and were sacrificed and perfused in November 2013. The birds were treated and used according to the guidelines of the University of the Witwatersrand Animal Ethics Committee (clearance no: 2013/05/02B), which parallel those of the NIH for the care and use of animals in scientific experimentation. Five minutes prior to being euthanized, both birds were given an intramuscular dose of heparin, 2,500 units (0.5 ml) to prevent blood clotting. Animals were then injected with an intraperitoneal dose of Euthapent (1 ml/kg) and body mass recorded. The average body mass of the two Congo African grey parrots was 453.75 g, and the two Timneh grey parrots was 286.36 g. All animals were transcardially perfusion-fixed, initially with a rinse of 0.9% saline, followed by 4% paraformaldehyde in 0.1M phosphate buffer (PB, pH 7.4). The brains were carefully removed from the skull, and post-fixed overnight in 4% paraformaldehyde in 0.1M PB. The average brain mass of the Congo African grey parrots was 10.30 g and that of the Timneh grey parrot was 7.85 g. Before sectioning, the tissue was allowed to equilibrate in a 30% sucrose in 0.1M PB solution at 4◦C for 4 days. The brains were then frozen in dry

ice and sectioned in the coronal plane, 50 µm thick sections, on a sliding microtome. A one in ten series of sections were taken and three series stained for Nissl substance, PCNA and DCX. The remaining series of sections were placed in an antifreeze solution and are stored at −20◦C for future use. The series of sections used for Nissl staining were mounted on 0.5% gelatine coated slides, dried overnight, cleared in a 1:1 mixture of 100% ethanol and 100% chloroform and stained with a 1% cresyl violet solution.

## Immunohistochemistry for PCNA and DCX

The series of sections used for free floating PCNA and DCX immunohistochemistry were initially treated for 30 min at room temperature under gentle shaking with an endogenous peroxidase inhibitor (49.2% methanol: 49.2% 0.1M PB: 1.6% of 30% H2O<sup>2</sup> = 0.48% H2O2), followed by three 10 min rinses in 0.1M PB. To block non-specific binding sites the sections were then preincubated for 2 h, at room temperature under gentle shaking, in a blocking buffer solution consisting of 3% normal horse serum (NHS) for PCNA sections or 3% normal rabbit serum (NRS) for DCX sections, 2% bovine serum albumin, and 0.25% Triton X-100 in 0.1M PB. Following preincubation the primary antibodies were added to the blocking buffer solution and the sections were incubated for 48 h at 4◦C under gentle shaking (PCNA – 1:500 dilution of mouse anti-PCNA, NCL-L-PCNA Leica Biosystems, Newcastle, United Kingdom; DCX – 1:300 dilution of goat anti-DCX antibody, C-18, Santa Cruz Biotechnology, Dallas, TX, United States) under gentle agitation. The primary antibody incubation was followed by three 10-min rinses in 0.1M PB and the sections were then incubated in a secondary antibody solution (PCNA sections – 1:1000 dilution of biotinylated anti-mouse IgG [BA-2001, Vector Labs] in 3% NHS and 2% bovine serum albumin in 0.1M PB; DCX sections – 1:1000 dilution of anti-goat IgG [BA-5000, Vector Labs] in 3% NRS and 2% bovine serum albumin in 0.1M PB) for 2 h at room temperature. This was followed by three 10 min rinses in 0.1 M PB, after which sections were incubated for 1 h in an avidin biotin solution (1:125 in 0.1M PB; Vector Labs, Burlingame, CA, United States), followed by three 10-min rinses in 0.1M PB. Sections were then transferred to a solution consisting of 0.05% diaminobenzidine tetrahydrochloride in 0.1M PB for 5 min at room temperature, after which 3.3 µl of 30% H2O2/ml of solution was added. With the aid of a low power stereomicroscope the progression of the staining was visually followed and allowed to continue until a level was reached where the background staining could assist in architectonic matching to the Nissl stained sections without obscuring the immunopositive structures. The tissue was then rinsed twice more in 0.1M PB before being mounted on glass slides coated with 0.5% gelatine and allowed to dry overnight. Once dry, the slides were placed in a solution of 70% ethanol for 2 h and then dehydrated, cleared in xylene and coverslipped with Depex. To test for non-specific staining of the immunohistochemical protocol, the primary and secondary antibodies were omitted


TABLE 1 | Summary of qualitative distribution and density of PCNA and DCX immunoreactive cells in the brain of the Congo African grey parrot and Timneh grey parrot.

−, absent; +, low; ++, moderate; +++, high density of PCNA and DCX immunoreactive structures.

from random sections and no staining was evident. The observed immunostaining patterns support the specificity of the antibodies and are compatible with observations made in pigeons, parakeets, and quails (Charvet and Striedter, 2008, 2009; Melleu et al., 2013, 2015).

## Analysis

The Nissl stained sections were examined with a low power stereomicroscope and the architectonic borders traced using a camera lucida. The PCNA and DCX immunostained sections were then matched to the drawings from the Nissl stained sections and the location of immunopositive soma marked on the drawings. Selected drawings were then scanned and redrawn using the Canvas 8 Software (Deneba Software, Miami, FL, United States). High power microscopic observation allowed for the determination of the relative densities of stained structures throughout the various regions of the brain. The relative densities of immunostained structures were visually compared and recorded on a scale ranging from low (+) to moderate (++) to high (+++). A second observer was used to eliminate observer bias.

The brain regions were identified and named in accordance with the stereotaxic atlas of the brain of the budgerigar (Brauth et al., 2011) using the nomenclature recommended by the Avian Brain Nomenclature Forum (Reiner et al., 2004). Digital photomicrographs were captured using a digital camera (Axio Cam HRc, Zeiss, South Africa) mounted on the light microscope (Axioskop 2 plus, Ziess, South Africa) and operating on the ZEN 2010 computer software (Zeiss, South Africa). No pixilation adjustments or manipulation of the captured images were undertaken, except for the adjustment of contrast, brightness, and levels using Adobe Photoshop 7.

## RESULTS

## General Observations

In the present study, we examined putative adult neurogenesis throughout the brains of the Congo African grey parrot (Psittacus erithacus) and the Timneh grey parrot (Psittacus timneh) using immunohistochemical techniques for the endogenous markers PCNA and DCX. The distribution of PCNA and DCX immunoreactivity was almost identical in both subspecies, but a few minor differences were observed (**Table 1**). Due to this extensive similarity we depict only the mapping of the distribution of the PCNA and DCX immunoreactive cells in the Timneh grey parrot (**Figures 1**, **2**). In both parrots, PCNA and DCX immunoreactivity was observed in the layers of the olfactory bulbs (OBs), the subventricular zone (SVZ) of the lateral, third and fourth ventricles and the cerebral aqueduct, subdivisions of the pallium (Hp, HA, HI, HD, M, N, E, and A), subpallium (MSt, LSt, SM, and SL), diencephalon, mesencephalon, and rhombencephalon. Generally, the telencephalic regions had a higher density of PCNA and DCX immunoreactive cells than other brain regions in both subspecies of parrots examined. In the majority of the telencephalic regions, the density of PCNA and DCX

list for abbreviations.

FIGURE 2 | Diagrammatic reconstructions of a series of evenly spaced coronal sections through one half of the brain of the African grey Parrot, illustrating the distribution of DCX-immunoreactive cells. A single solid black star indicates a single cell body and a single open star represents a weakly stained cell. Drawing (A) represent the most rostral section, (P) the most caudal, and each drawing is approximately 1,500 µm apart. For example distance between A,B is 1,500 µm. The same applies for B–P. See list for abbreviations.

immunoreactive cells increased from rostral to caudal in both parrot subspecies. DCX immunoreactivity was more intense in fibers than in cell bodies and the majority of DCX immunopositive cells included small rounded cells, fusiform unipolar and bipolar cells.

## Distribution of PCNA Immunoreactivity Olfactory Bulb

Proliferating cell nuclear antigen immunoreactive cells were observed at high density in the IGrL, MCL, and EPL layer, while they were found in low density in the GL and ON layer (**Figures 3A,C**).

## Subventricular Zone

In both parrot species the SVZ of the lateral ventricle exhibited a high density of PCNA immunoreactivity with occasional cells clustering. There was evidence of proliferative hot spots as described by Alvarez-Buylla et al. (1990) in the dorsal and ventral poles of the lateral ventricle in the Congo African grey parrots at rostral levels, but only a dorsal accumulation of PCNA immunoreactive cells was observed in the Timneh grey parrots (**Figures 4A,C**). The third and fourth ventricles and the cerebral aqueduct of both parrot species exhibited a medium density of PCNA immunoreactivity without cell clustering. There was a low density of PCNA immunoreactivity in the SVZ of the tectal portion of the cerebral aqueduct in both parrot species.

## Pallial Regions

In both parrot subspecies examined the HA, HD, HP, and APH exhibited a moderate density of PCNA immunoreactive cells (**Figure 5A**). In the ventral hippocampus the ml showed a low density of PCNA immunoreactive cells compared to the tr and ll which presented a moderate density (**Figure 5C**). The HI showed a low density of PCNA immunoreactive cells in the medial and core regions but an increase in the density of PCNA immunoreactive cells was observed in its lateral regions.

The mesopallium (M) exhibited a homogenous moderate density of PCNA immunoreactive cells at rostral levels, but at caudal levels the density of PCNA immunoreactive cells were reduced to low density in core regions, but moderate density in the medial and lateral regions. The entopallium exhibited a moderate density of PCNA immunoreactive cells and their distribution decreased markedly in caudal sections in both species of African grey parrots. There was a high density of PCNA immunoreactive cells in the nidopallium frontale (NF) of both parrot subspecies. The core region of NF of the Congo African grey parrot exhibited a large cluster of PCNA immunoreactive cells. At caudal levels, the NI and NC exhibited a moderate density of PCNA immunoreactive cells in the lateral regions, while the medial and core regions showed a low density of PCNA immunoreactive cells in both parrot species. The arcopallium exhibited a moderate density of PCNA immunoreactive cells which were distributed in higher density in the ventral regions encompassing the Tn than the dorsal regions, which showed a low density to almost no PCNA immunoreactive cells.

power images on inserts were taken from consecutive sections to low power images. Scale bar in (D) = 500 µm and applies to all. The scale bar in the inset in

Subpallial Regions

In the striatum of both parrot species, a moderate density of PCNA immunoreactive cells was observed in MSt and LSt. PCNA immunoreactive cells were distributed in higher density in ventral regions of the MSt and also in regions adjacent to the ventrolateral wall of the lateral ventricle. No detectible PCNA immunoreactive cells were observed in the regions corresponding to the GP. Generally in the striatum, there was a rostro-caudal decline in the density of PCNA immunoreactive cells. In the septal complex of both parrots, the SM exhibited a low density of PCNA immunoreactive cells while the SL showed a moderate density of PCNA immunoreactive cells.

(D) = 50 µm and applies to all insets. See list for abbreviations.

#### Diencephalon

The diencephalon exhibited a high density of PCNA immunoreactive cells in the paraventricular nuclei, including POA and POM, the dorsal margin in the DMA, DLL, and DLP and the lateral margin in the nucleus rotundus (Rt), SPC, and SpL. In the core regions of the diencephalon, the African grey parrot exhibited a moderate density of PCNA immunoreactive cells in OV. The VMH in the ventral hypothalamus of both parrots exhibited the highest density of PCNA immunoreactive cells. The OC showed PCNA immunoreactive cells ventral to the inferior pole of the third ventricle in both species (**Figures 6A,C**).

#### Mesencephalon, Rhombencephalon, and Cerebellum

Most of the regions of the midbrain exhibited a moderate density of PCNA immunoreactive cells, including the nIII, GCt, IP and the isthmic region in the SLu, Ipc, and Imc. In the optic tectum in both parrots a higher density of PCNA immunoreactive cells was observed in the SGF and SGC.

In the cerebellum, a moderate density of PCNA immunoreactive cells was observed in the Purkinje cell layer and the CbM and CbL (**Figures 7A,C**). Occasional PCNA immunoreactive cells were observed in the molecular and granule cell layers. In the pons and the medulla oblongata, a

the brains of the two parrot species – (A,B) – African grey parrot (Psittacus erithacus), (C,D) – Timneh grey parrot (Psittacus timneh). PCNA immunoreactivity revealed aggregates of proliferating cells (hot spots) in the SVZ of the dorsal pole of the lateral ventricle in the African grey parrot (Psittacus erithacus) (A), and ventral pole of the lateral ventricle in Timneh grey parrot (Psittacus timneh) (C). Insets in (A,C) show a higher magnification images of PCNA immunoreactive cells. DCX immunoreactivity revealed aggregates of DCX-immunoreactive cells and fibers in the SVZ of the dorsal pole of the lateral ventricle in the African grey parrot (Psittacus erithacus) (B), and in the ventral pole of the lateral ventricle in the Timneh grey parrot (Psittacus timneh) (D). Insets in (B,D) show a higher magnification of DCX immunoreactive cells and fibers. All high power images on inserts were taken from consecutive sections to low power images. In all images dorsal is to the top and medial to the right. Scale bar in (D) = 500 µm and applies to all. The scale bar in the inset in (D) = 50 µm and applies to all insets. LV, lateral ventricle.

moderate density of PCNA immunoreactive cells was observed in some regions not limited to the LoC, medial and lateral pontine nuclei (PM and PL), PrV, TTd, and nucleus raphe (R).

# Distribution of DCX Immunoreactivity

Olfactory Bulb

The OB was the region that contained the highest density of DCX immunoreactive structures in both parrots. DCX immunoreactive cells and fibers were seen in high density in the inner layers of the olfactory bulb, but in lower density in the outer layers (**Figures 3B,D**).

## Subventricular Zone

In both parrot subspecies, the SVZ of the lateral ventricle exhibited a high density of DCX immunoreactive cells and fibers. The majority of the DCX immunoreactive cells were orientated parallel to the walls of the lateral ventricles, depicting tangential migration, while a few DCX immunoreactive cells were oriented perpendicular to the walls of the lateral ventricle, suggesting radial migration. At some rostral levels in the Congo African grey parrot there were occasional accumulations of DCX immunoreactive cell clusters in the dorsal and ventral poles of the lateral ventricle, similar to the dorsal and ventral 'hot spots' suggested by Alvarez-Buylla et al. (1990) (**Figures 4B,D**). Such accumulations of DCX immunoreactive cells were only observed in the dorsal pole of the lateral ventricle of the Timneh grey parrots. The SVZ of the third and fourth ventricles and that of the cerebral aqueduct showed a low density of DCX immunoreactive fibers with no cells clearly discernable.

## Pallial Regions

The pallial portion of the telencephalon exhibited the highest density of DCX immunoreactive cells and fibers, although the density of DCX immunopositive structures varied across different pallial regions in both parrots. Moderate densities of DCX immunoreactive cells and fibers were present in the HA, HD, medial and lateral NF, nidopallium (N), and NI. The medial and lateral regions of the HI and the M showed a low density of DCX immunoreactive cells and fibers. There was almost no DCX immunoreactivity detected in the

medial to the right. Scale bar in (D) = 500 µm and applies to (A–D). The scale bar in the inset in (D) = 50 µm and applies to all insets. See list for abbreviations.

core of HI and M. In the entopallium a low density DCX immunoreactive cells and fibers was observed. At rostral levels of these telencephalic pallial regions, DCX immunoreactive cells and fibers were present in a slightly higher density laterally than medially. Generally there was a rostro-caudal decrease in the amount of DCX immunoreactive material in these pallial regions. The NC and the nidopallium caudolaterale (NCL) exhibited a much higher density of DCX immunoreactive cells and fibers than any other pallial region, particularly in the ventral and lateral aspects. The arcopallium exhibited a very low density of DCX immunoreactive cells and fibers in both parrot species. The HP and APH contained a moderate density of DCX immunoreactive cells and fibers (**Figure 5B**). In the ventral hippocampus, a moderate density DCX immunoreactive material was observed in the tr and ll while the ml showed a low density (**Figure 5D**).

#### Subpallial Regions

The medial and lateral striatum exhibited a moderate density of DCX immunoreactive cells and fibers at rostral levels, but the density of these cells and fibers decreased noticeably at levels caudal to the anterior commissure (CA). In the MSt, the DCX immunoreactive cells and fibers were observed in higher density in the medial and ventral aspect than in the dorsal and lateral aspects. A low density of DCX immunoreactive cells and fibers was observed in the SM and SL; however, at caudal levels in both parrots the SL showed a moderate density of DCX immunoreactive cells and fibers. No detectible DCX immunoreactive structures were observed in the regions corresponding to the GP.

#### Diencephalon

The diencephalon exhibited a very low density of DCX immunoreactive cells and fibers. A weak DCX immunoreactivity was observed in the paraventricular nuclei, including POA and POM of the medial hypothalamus and in the VMH of the ventral hypothalamus and in the dorsolateral group of nuclei of the diencephalon, which include the DMA, DLL, DLP, nucleus rotundus (Rt), SPC, and SpL. The OC showed DCX immunoreactive fibers (**Figures 6B,D**), while no staining was detected in CA.

#### Mesencephalon, Rhombencephalon, and Cerebellum

A low density of DCX immunoreactive fibers was observed in the layers of the optic tectum (SGF and SGC). Some mesencephalic nuclei and tracts also exhibited weak DCX immunoreactive fibers and vesicles including the GCt, nIV, ICo, and the fasciculus longitudinalis medialis (FLM). In the cerebellum, the Purkinje cell layer was weakly stained, with only fibers being observed (**Figures 7B,D**). In the pons and medulla oblongata weakly stained fibers and vesicles in very low density were observed in some nuclei including, but not limited to, the LLV, PM, PL, La, raphe nucleus (R), nXIIts, and nX.

## DISCUSSION

## General Considerations

In the current study we examined putative adult neurogenesis in the brains of the two species of African grey parrots the Congo African grey parrot (Psittacus erithacus) and Timneh grey parrot (Psittacus timneh) using PCNA and DCX immunohistochemistry. The PCNA immunoreactive cells were distributed heterogeneously in all brain regions including the olfactory bulb, telencephalon, diencephalon, mesencephalon and rhombencephalon, while DCX immunoreactive structures were dense in the olfactory bulb, telencephalon and some regions of the diencephalon, but far less dense in some parts of the diencephalon, the mesencephalon and rhombencephalon in both African grey parrot species. To a large extent our results conform and are comparable to the wide spread distribution of adult neurogenesis reported in other birds (Kim et al., 2006; Boseret et al., 2007; Balthazart et al., 2008; Vellema et al., 2014; Mazengenya et al., 2017). Despite the overall similarity in the distribution of PCNA and DCX in both African grey parrot species, region specific differences, particularly associated with the density of PCNA and DCX immunoreactive structures, were observed in the telencephalon. Such region specific differences might highlight the relationship between adult neurogenesis and brain function, particularly in regions responsible for certain behavioral repertoires in different species.

## Validity of PCNA and DCX Antibodies in Determining Adult Neurogenesis in Birds

The PCNA antibody used in the current study recognizes the 36 kDa nuclear protein subunit of the DNA polymerase that

FIGURE 7 | Photomicrographs of PCNA – (A,C) and DCX (B,D) – immunostained coronal sections through the cerebellar cortex in the brains of the two parrot species – (A,B) – African grey parrot (Psittacus erithacus), (C,D) – Timneh grey parrot (Psittacus timneh). A clear similarity in the distribution of PCNA immunoreactive cells in the layers of the cerebellar cortex in the African grey parrot (A) and Timneh grey parrot (C) is observed. A similar distribution and staining intensity of DCX immunoreactive cells and fibers in the layers of the cerebellar cortex in the African grey parrot (Psittacus erithacus) (B) and Timneh grey parrot (Psittacus timneh) (D) is also observed. Note large and deeply stained PCNA immunoreactive cells (arrows) in the Purkinje cell layer (pcl) in both parrots. In all images dorsal is to the top and medial to the right. Scale bar in (D) = 100 µm and applies to all. See list for abbreviations.

is essential for DNA replication and is expressed selectively in proliferating cells (Hall et al., 1990; Balthazart and Ball, 2014). The PCNA antibody is present during the various stages of the cell cycle except the G<sup>0</sup> stage (Kurki et al., 1988; Hall et al., 1990; Balthazart and Ball, 2014). In birds, the PCNA antibody has been characterized in developing quail, parakeets, zebra finches (Charvet and Striedter, 2008, 2009) and chickens (Hannan et al., 1999; Capes-Davis et al., 2005). The regions where PCNA immunoreactivity was observed in the African grey parrots correspond with those reported in quails after BrdU injection (Charvet and Striedter, 2008). Apart from labeling proliferating cells, the PCNA antibody also labels cells repairing their DNA (Hall et al., 1990; Shivji et al., 1992; Essers et al., 2005) and remains detectable several days after cells exit the cell cycle (Balthazart and Ball, 2014). Such characteristics can affect the reliability of the PCNA antibody as a marker of adult neurogenesis in birds since they lead to an overestimation of proliferating cells.

Doublecortin is a microtubule associated protein that plays a key role in the migration of neurons during development and in post mitotic neurons undergoing migration, remodeling of their dendritic processes, and synaptogenesis in adulthood (Francis et al., 1999; Gleeson et al., 1999; Nacher et al., 2001; Brown et al., 2003; Yang et al., 2004; Capes-Davis et al., 2005; Couillard-Despres et al., 2005). DCX expression levels have been found to decline with age in mammals (Brown et al., 2003; Couillard-Despres et al., 2005) and in birds (Ling et al., 1997; Hannan et al., 1999; Capes-Davis et al., 2005; Kim et al., 2006). In rats DCX expression is first observed 1 day after cell division and last for about 2–3 weeks until the neurons begin to express markers of mature neurons such as NeuN (Brown et al., 2003). In birds, DCX expression has been found to last for up to 30 days post mitosis (Balthazart et al., 2008). The DCX antibody was recommended for adult neurogenesis (Rao and Shetty, 2004; Couillard-Despres et al., 2005) and its expression has been consistent in areas known to recruit new neurons in mammals (Brown et al., 2003). In contrast, DCX expression in birds has been found in locations not associated with adult neurogenesis. In adult canaries BrDU immunoreactive cells were found to co-express the DCX antibody in the subtelencephalic regions (Vellema et al., 2014). The DCX antibody used in the current study has been characterized in the following birds: canaries

(Boseret et al., 2007; Balthazart et al., 2008; Yamamura et al., 2011; Vellema et al., 2014), chickadees (Fox et al., 2010), zebra finches (Kim et al., 2006), sparrows (LaDage et al., 2010), starlings (Hall and MacDougall-Shackleton, 2012), pigeons (Melleu et al., 2013, 2015; Mazengenya et al., 2017) Japanese quail (Balthazart et al., 2010) and chickens (Hannan et al., 1999; Capes-Davis et al., 2005; Mezey et al., 2012).

## Adult Neurogenesis in African Grey Parrots Compared to Other Birds

Adult neurogenesis has been observed in budgerigars (Melopsittacus undulatus), which are small parrots (Nottebohm, 2011), but the current study is the first to report the presence of adult neurogenesis in the brains of the larger African grey parrots. There are remarkable similarities and as well as some variations in the pattern of adult neurogenesis across the avian species studied to date. In the avian brain, new neurons are generated in the SVZ of the lateral ventricle, with the largest density of proliferating cells reported in its dorsal and ventral poles. These cells migrate to different target regions of the brain (Alvarez-Buylla et al., 1994). In the present study proliferating cells were observed in high density in the SVZ of the lateral ventricle and throughout its rostrocaudal extent. The Congo African grey parrot exhibited an accumulation of proliferating cells in the dorsal and ventral poles of the lateral ventricle, at the level of CA, similar to the proliferative "hot spots" described by Alvarez-Buylla (1990). These aggregates of proliferating cells were only observed in the dorsal pole of the lateral ventricle in the Timneh grey parrot. The proliferative hot spots phenomenon was also reported in other adult birds, including pigeons (Melleu et al., 2013), canaries (Alvarez-Buylla, 1990; Alvarez-Buylla et al., 1998; Mazengenya et al., 2017), chickens (Mezey et al., 2012), and marsh tits (Poecile palustris) (Patel et al., 1997); however, they were absent in the adult ring dove (Streptopelia risoria) (Ling et al., 1997). The identity of the proliferating cells in these germinal zones remains contentious in both mammals and non-mammals. In the song bird, Serinus canaria the proliferative hot spots were associated with large accumulations of radial glial cells (Alvarez-Buylla and Nottebohm, 1988; Alvarez-Buylla et al., 1988), suggesting that radial glial cells were the primary cells undergoing mitosis in the germinal zones (Alvarez-Buylla et al., 1990; Goldman, 1990; Vellema et al., 2010). In contrast, cell proliferation, including in the present study, was observed in the brain parenchyma (Alvarez-Buylla et al., 1988; Vellema et al., 2010; Mazengenya et al., 2017), the SVZ of the cerebral aqueduct, third, fourth, and tectal ventricles, albeit in low density, and in the midbrain, hindbrain and cerebellum (Vellema et al., 2010). This localized proliferative activity suggests that certain neurons may be produced and integrate locally without the need for cell migration, and or thus proliferating cells can be glia or endothelial cells (Vellema et al., 2010).

The recruitment of new neurons is widespread in the telencephalon of adult birds and affected by various factors including age, environmental complexity, seasonal variation, hormones, stress and social complexity (for review, see Barnea and Pravosudov, 2011). In the African grey parrots, the distribution of DCX immunoreactive cells and fibers exhibited a similar pattern to that observed in the adult canaries (Boseret et al., 2007; Balthazart et al., 2008; Vellema et al., 2014), starlings (Absil et al., 2003) and zebra finches (Kim et al., 2006); however, DCX immunoreactivity was reported to be restricted to the telencephalon in other studies on adult canaries (Alvarez-Buylla and Nottebohm, 1988; Kirn et al., 1994; Vellema et al., 2010), pigeons (Melleu et al., 2013), ring doves (Ling et al., 1997), and chickens (Mezey et al., 2012).

Neuronal recruitment exhibited regional differences in the telencephalic subdivisions of adult canaries, zebra finches, domestic pigeons, and rock pigeons (Kim et al., 2006; Boseret et al., 2007; Balthazart et al., 2008; Melleu et al., 2013; Mazengenya et al., 2017). In the adult canaries and zebra finches, the highest densities of DCX immunoreactive cells were observed in the M, NC, and striatum. Intermediate DCX expression was observed in the hyperpallium and rostral nidopallium, while a low density of DCX immunopositive cells was noted in the hippocampus, arcopallium and subtelencephalic regions (Kim et al., 2006; Boseret et al., 2007; Vellema et al., 2014). We observed similar trends in the adult African grey parrots, but the hippocampus and parahippocampal regions exhibited moderate to high density of DCX immunoreactive cells and fibers. In addition, DCX immunoreactive neurons and fibers showed some area specific distribution in certain regions of the telencephalon and the brain stem nuclei in the two species of African grey parrots. For example, rostral nidopallium exhibited a moderate density DCX immunoreactive cells and fibers, whereas the caudal nidopallium exhibited the densest DCX immunoreactive cells and fibers. In addition, the distribution of DCX immunoreactive cells was not uniform in most telencephalic regions, where DCX immunoreactive cells and fibers were found in higher density in the medial and lateral margins compared to the core regions. The area of specific distribution of DCX immunoreactive cells and fibers observed in these parrot species was also reported in adult canaries (Vellema et al., 2010, 2014) and might highlight brain areas where plastic changes are necessary for the maintenance of behavioral repertoires in different species. In both song and non-song birds, the arcopallium and the entopallium exhibited absent to low densities of DCX immunoreactive cells (Goldman and Nottebohm, 1983; Vellema et al., 2010; Melleu et al., 2013; Mazengenya et al., 2017). Similarly, the two species of African grey parrots showed a low density of DCX immunoreactive cells in the entopallium and the arcopallium. In the arcopallium, the DCX immunoreactive cells and fibers were distributed only in the ventrocaudal margins encompassing the nucleus taenia (Tn).

The morphology of DCX immunoreactive cells varied in the different regions of the brain in the two species of African grey parrots. In regions closer to the lateral ventricle particularly the ventral striatum the DCX immunoreactive cells exhibited elongated somata with unipolar or bipolar morphology. In other regions of the telencephalon elongated DCX immunoreactive cells were mixed with cells with round and multipolar morphology, although the former were more abundant. Elongated unipolar or bipolar cells represent migrating young neurons, usually emanating from the SVZ, whereas the round and multipolar neurons characterize mature neurons (Alvarez-Buylla, 1990). Migration of neurons in the telencephalon of

birds follows two defined modes including tangential and radial migration (Alvarez-Buylla et al., 1988; Alvarez-Buylla, 1990). Vellema et al. (2010) further described the random migration of neuroblasts, whereby migrating immature neurons follow undefined routes within the telencephalon. In birds, the most pronounced mode of migration is radial migration, whereby migrating neuroblasts are guided on cellular processes of radial glial cells (Alvarez-Buylla and Nottebohm, 1988; Alvarez-Buylla, 1990; Vellema et al., 2010). Under this mode of migration, young neurons migrate from the SVZ of the lateral ventricle preferentially toward the lateral and ventrocaudal aspects of the telencephalon (Alvarez-Buylla, 1990; Vellema et al., 2010). In birds including in the current study, proliferative hot spots, both dorsal and ventral (PCNA immunoreactive as in the present study) coincide with accumulations of immature and migrating neurons (DCX immunoreactive cells and fibers) particularly in the hyperpallium and the striatum, respectively (Alvarez-Buylla and Nottebohm, 1988; Alvarez-Buylla, 1990). According to Alvarez-Buylla (1990), radial glia processes extend laterally for about 2 mm into the parenchyma of the adult avian telencephalon and these radial cells are concentrated in proliferative hot spots suggesting that migration through scaffolds of the radial glia processes is only possible for very short distances. In the current study, the Congo African grey parrot showed accumulation of a higher density of DCX immunoreactive cells and fibers in regions of the telencephalon corresponding to the dorsal and ventral hot spots whereas in the Timneh grey parrot a higher density of DCX immunoreactive structures was observed in the dorsal regions of the telencephalon.

This preferred direction of migration may explain the presence of the high densities of DCX immunoreactive cells and fibers in the lateral and caudal regions of the telencephalon when compared to the medial and rostral regions. Migration of immature neurons furthest from the extends of the radial glia processes becomes random and neuroblasts migrate and differentiate in regions that are topographically disjointed from the proliferative hot spot indicating that there is no SVZ specification for the final destination and position where new neurons differentiate and integrate (Alvarez-Buylla and Nottebohm, 1988; Alvarez-Buylla, 1990). This may also help to explain the variations in the topography of the proliferative hot spots observed in the current study and in other birds examined to date. According to Melleu et al. (2013), the majority of round and multipolar cells expressing the microtubule marker DCX do not co-express with the adult neuronal marker NeuN, suggesting that they are immature neurons even though they show the morphology of mature neurons.

Recruitment of new neurons in the diencephalon, mesencephalon and rhombencephalon of adult birds has been reported in experimental animals (Cao et al., 2002; Chen et al., 2006), but no studies support adult neuronal recruitment under normal physiological conditions. Despite this, reports in adult canaries (Boseret et al., 2007; Balthazart et al., 2008; Vellema et al., 2014) and zebra finches (Kim et al., 2006) indicate the presence of DCX immunoreactive cells in these regions. In the species of the African grey parrots studied herein, similar, but low density, distributions of DCX immunoreactive cells and fibers were observed. In agreement with Kim et al. (2006) and Boseret et al. (2007) the low density expression of DCX immunoreactivity in the brain stem may not signify adult recruitment of new neurons, but represent cellular plasticity of adult neurons (Nacher et al., 2001; Brown et al., 2003).

## Functional Implications of Adult Neurogenesis Related to Certain Behaviors of the African Grey Parrots

The role of adult neurogenesis in brain function and behavior in various species is still contentious. In mammals adult hippocampal neurogenesis is generally associated with acquisition of new and clearance of old memories (Kempermann et al., 2004; Zhao et al., 2008); however in birds, neuronal recruitment is wide spread in the telencephalon, making it difficult to understand the precise functional implications of this widespread characteristic. In addition, birds live comparably longer than mammals of similar body mass, and adult neurogenesis in avian species may be adaptive, promoting continuous learning by updating and renewing memories (Nottebohm, 1991; Gahr et al., 2002). The effect of learning on adult neurogenesis is bidirectional. Learning has been found to increase the survival, maturation and the response to stimuli of adult born neurons (Kee et al., 2007; Tronel et al., 2010; Lemaire et al., 2012). In contrast, adult neurogenesis facilitates learning in new conditions, which may include new habitats, new members of social groups, or new sources of food. In avian species adult neurogenesis is coupled with cell death (Kirn and Nottebohm, 1993; Rasika et al., 1994; Yamamura et al., 2011) and has been found to be task-dependant in various species (Barker et al., 2011). According to Nottebohm et al. (1986) increased singing in canaries leads to increased neuronal recruitment in the high vocal centre (HVC). In food storing species increased hippocampal neurogenesis was observed during the peak periods of food caching in autumn, whereas in non-storing species no addition of hippocampal neurons was recorded during the same period (Barnea and Nottebohm, 1996; Sherry and Hoshooley, 2009). Moreover in migratory species, which rely on spatial memory, species with larger home ranges exhibited increased hippocampal neurogenesis compared to related species that inhabit smaller territories (Cristol et al., 2003; Hoshooley and Sherry, 2007; LaDage et al., 2009). The olfactory bulbs recruit new neurons in adults of almost all studied species, except for a few studies of canaries and zebra finches (Barker et al., 2011; Melleu et al., 2013). In the current study we observed a high density of PCNA and DCX immunoreactive structures in the inner layers of the olfactory bulbs in both African grey parrots. The functional implications of olfactory neurogenesis remain elusive, but the olfactory system is involved in reproduction, the monitoring of environmental changes, the maintenance of survival by detecting presence of predators, the identification of clan members and the location of food (Jones and Roper, 1997; Patzke et al., 2010; Barker et al., 2011).

Apart from adult hippocampal and olfactory neurogenesis, neuronal recruitment in the NC has been shown to increase under various states of environmental enrichment. In a study

conducted on zebra finches, adult neurogenesis in the NC was found to increase in adults introduced to communal living compared to individuals living in pairs (Lipkind et al., 2002). The reason associated with this increase is the possibility that individuals introduced to communal living need to identify new members and establish new social relationships (Lipkind et al., 2002; Adar et al., 2008). Similar changes in neuronal recruitment in the NC were identified during the reproductive cycle, when adult individuals of zebra finches were feeding their young (Barkan et al., 2007). In the present study, the NC and the NCL exhibited the densest DCX immunoreactivity in both species of African grey parrots. The NC is associated with reproductive behavior (Melleu et al., 2013), whereas the NCL, a proposed analog of the mammalian prefrontal cortex, participates in higher order cognitive functions, such as decision making, speech and planning of behavior (Güntürkün, 2005).

African grey parrots live in large colonies of approximately 10,000 individuals (Parr and Juniper, 2010). Although few studies have examined the behavior of African grey parrots in the wild, results show that their social ecology is comparable to that of primates (Emery et al., 2007). The African grey parrots lead a complex social life coupled with long

## REFERENCES


term monogamous relationships (Seibert, 2006; Emery et al., 2007). Shultz and Dunbar (2010) postulated that species in monogamous relationships develop larger brain to body mass ratios to preserve the stable pair bonded relationships. The African grey parrots may be useful models to assess the impact of changing social environments on adult neurogenesis.

## AUTHOR CONTRIBUTIONS

AI and PRM designed the study, analyzed the data, and reviewed the final manuscript for submission. PM and AB collected and processed tissue and carried out the initial analysis of data. PM prepared the initial manuscript draft.

## FUNDING

This work was supported by the Swiss South African Joint Research Programme (SSAJRP09) to AI and PRM and the South African National Research Foundation (NRF) research Grant No. CSUR13082730945 to AI.


Chapouton, P., Jagasia, R., and Bally-Cuif, L. (2007). Adult neurogenesis in nonmammalian vertebrates. Bioessays 29, 745–757. doi: 10.1002/bies.20615


migration and pattern formation in diverse regions of the developing chick brain. J. Neurosci. Res. 55, 650–657. doi: 10.1002/(SICI)1097-4547(19990301) 55:5<650::AID-JNR12>3.0.CO;2-S



adult dentate gyrus. Eur. J. Neurosci. 19, 234–246. doi: 10.1111/j.0953-816X. 2003.03123.x


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Mazengenya, Bhagwandin, Manger and Ihunwo. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Sociality Affects REM Sleep Episode Duration Under Controlled Laboratory Conditions in the Rock Hyrax, Procavia capensis

Nadine Gravett <sup>1</sup> \*, Adhil Bhagwandin<sup>1</sup> , Oleg I. Lyamin2,3 , Jerome M. Siegel 2,3 and Paul R. Manger <sup>1</sup>

<sup>1</sup>School of Anatomical Sciences, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa, <sup>2</sup>Department of Psychiatry, School of Medicine, University of California, Los Angeles, Los Angeles, CA, United States, <sup>3</sup>Brain Research Institute, Neurobiology Research, Sepulveda VA Medical Centre, Los Angeles, CA, United States

The rock hyrax, Procavia capensis, is a highly social, diurnal mammal. In the current study several physiologically measurable parameters of sleep, as well as the accompanying behavior, were recorded continuously from five rock hyraxes, for 72 h under solitary (experimental animal alone in the recording chamber), and social conditions (experimental animal with 1 or 2 additional, non-implanted animals in the recording chamber). The results revealed no significant differences between solitary and social conditions for total sleep times, number of episodes, episode duration or slow wave activity (SWA) for all states examined. The only significant difference observed between social and solitary conditions was the average duration of rapid eye movement (REM) sleep episodes. REM sleep episode duration was on average 20 s and 40 s longer under social conditions daily and during the dark period, respectively. It is hypothesized that the increase in REM sleep episode duration under social conditions could possibly be attributed to improved thermoregulation strategies, however considering the limited sample size and design of the current study further investigations are needed to confirm this finding. Whether the conclusions and the observations made in this study can be generalized to all naturally socially sleeping mammals remains an open question.

#### Keywords: REM, NREM, SWA, rock hyrax, Hyracoidea, Afrotheria

## INTRODUCTION

Sleep is a homeostatically regulated process and it is characterized by its easy reversibility, immobility and reduced responsiveness to sensory stimuli. Species specific sleep postures and sleep sites, as well as closure of the eyes, are typically regarded as signs of behavioral sleep. Mammalian sleep is divided into non rapid eye movement sleep (NREM), which is often but not always synonymous to slow wave sleep (SWS), and rapid eye movement sleep (REM; Nicolau et al., 2000; Zepelin, 2005; Cirelli and Tononi, 2008; Lesku et al., 2008; Siegel, 2008). In addition to the classic mammalian sleep states an additional state, unique to the rock hyrax, termed somnus innominatus (meaning sleep without a name, SI) has been identified. It is currently not known if SI is form of low-voltage SWS or REM sleep as it is characterized by a low-voltage, high frequency electroencephalogram (EEG), an electromyogram (EMG) that is similar in amplitude to the preceding SWS episode and a mostly regular heart rate (see Gravett et al., 2012).

#### Edited by:

Nouria Lakhdar-Ghazal, Mohammed V University, Morocco

#### Reviewed by:

Marina Bentivoglio, University of Verona, Italy William Wisden, Imperial College London, United Kingdom Sidarta Ribeiro, Federal University of Rio Grande do Norte, Brazil

#### \*Correspondence:

Nadine Gravett nadine.gravett@wits.ac.za

Received: 25 July 2017 Accepted: 03 November 2017 Published: 16 November 2017

#### Citation:

Gravett N, Bhagwandin A, Lyamin OI, Siegel JM and Manger PR (2017) Sociality Affects REM Sleep Episode Duration Under Controlled Laboratory Conditions in the Rock Hyrax, Procavia capensis. Front. Neuroanat. 11:105. doi: 10.3389/fnana.2017.00105

Many studies performed in mammals in laboratories to date have been conducted on animals housed in solitary, or asocial, conditions, even though that particular animal may be naturally social (McNamara et al., 2008). It has been suggested that animals in a social setting sleep less, exhibit more fragmented sleep patterns, and have lower NREM and REM quotas (Capellini et al., 2008a,b, 2009). A possible reason for this is that species that sleep socially can enter deeper stages of sleep, as they have the security of sleeping in a group, and are thus able to sleep more efficiently and acquire the benefits of sleep in a shorter time frame. It has also been hypothesized that social species have to invest more time in social interactions and relationships, which in effect leaves them with less time to sleep (Capellini et al., 2009).

In primates, it has been proposed that hierarchy may play a role in the manifestation of the sleep patterns that are observed (Noser et al., 2003). Male Gelada baboons show no correlation between sleep duration and social rank, whereas females and juveniles exhibit an increase in sleep duration with decreasing rank (Noser et al., 2003). Furthermore, dominant male Gelada baboons have an increased amount of transitional sleep, which indicates that increasing rank leads to a decreased amount of relaxed sleep, which may in turn lead to an increased degree of vigilance, enabling these animals to react swiftly to nocturnal dangers. This study also indicated that no correlation existed between sleep fragmentation and social rank (Noser et al., 2003). It has also been shown that group or network size in primates does not correlate with sleep times (Nunn et al., 2010); however, when Drosophila was placed in a socially enriched environment they exhibited an increase in daytime, but not night time, sleep (Ganguly-Fitzgerald et al., 2006). Studies by Meerlo et al. (1997, 2001) have also shown that in rodents, social conflict affects NREM sleep by increasing electroencephalographic slow wave activity (SWA) during the subsequent sleeping bout. Sleep duration was not affected by the social stimulus, but the rodents appeared to compensate for the socially induced sleep debt by increasing SWA during NREM. It was also noted that this increase in SWA was not only the result of the length of the preceding waking episode, but also the social nature of the preceding waking episode.

Despite these previous studies, to the authors' knowledge, no comparative physiologically monitored sleep studies have been undertaken on the same animal under freely interacting social and solitary conditions. Thus, in the present study sleep was telemetrically recorded in the rock hyrax, Procavia capensis, under both solitary and freely interacting social conditions, extending our previous study of sleep in the rock hyrax (Gravett et al., 2012). Rock hyraxes are naturally diurnal mammals that live in colonies on rocky outcrops with crannies and crevices in which they shelter. The structure of the colonies is hierarchical consisting of a dominant male and female. The size of the colonies varies and is dependent on habitat and food availability but typically ranges between four and eight individuals. Rock hyraxes are often seen basking in the sun in the wild and have been reported to be poor thermoregulators. It has also been reported that rock hyraxes frequently heap and huddle together in captivity in an effort to conserve energy whilst in the wild many postures are adopted while resting, the most common being sitting with the head held into the body (Smithers, 1983).

The aim of the present study was thus to determine whether any significant differences existed in sleep and wake states between the social and solitary conditions to test hypotheses regarding the effect of sleeping socially.

## MATERIALS AND METHODS

A total of five adult rock hyraxes, with body masses ranging between 1.74 kg and 4.3 kg (**Table 1**; body mass was used to identify adult individuals based on data provided in Skinner and Chimimba, 2005), were used in the present study. Permits from the Limpopo and Gauteng Provincial Governments were obtained for the capture and transport of the animals from the wild. All animals were treated and used according to the guidelines of the University of the Witwatersrand Animal Ethics Committee (clearance number: AESC 2005/8/5), and the study was approved by the University of the Witwatersrand Animal Ethics Screening Committee which parallel those of the NIH for the care and use of animals in scientific experimentation. The methods describe below are the same as those reported in Gravett et al. (2012).

The animals were captured at random in groups of three from wild populations and thereafter allowed to acclimatize for a period of 1 month to the recording enclosures that had a 12:12 lighting schedule (light intensity 420 lx, measured with a digital lux meter) with temperature maintained between 19◦C and 21◦C. Following acclimatization one animal in the group of three that were housed together was selected randomly and implanted with a telemetric recording device (Data Sciences International) that allowed EEG, EMG and electrocardiography (ECG) recording without cables or restraint. The remaining two animals were moved into a neighboring enclosure, within the same room, that was identical to the recording/acclimatized enclosure which allowed for the recording of sleep in a solitary setting from the implanted animal continuously for 72 h (see Gravett et al., 2012). Following the solitary recording phase, the non-implanted animals were moved back into the recording enclosure with the implanted animal. The animals were then allowed an additional 3 days of acclimatization to the social setting after which sleep was recorded continuously for 72 h from the same implanted animal under social conditions. The animals were disturbed only once a day for approximately 5 min at the same time during each of the recording days for feeding.

The enclosure in which recording occurred was 1.8 × 1.5 m with a painted concrete surface that was covered with straw. The height of the chamber was approximately 1.5 m and steel mesh was placed over the top of the enclosure to prevent the animals from escaping. A wooden box (90 × 90 × 30 cm) with a Perspex roof and two entrances was placed inside the chamber and food (combinations of cucumber, tomato, sweet potato, pumpkin, apples and rabbit pellets as a source of roughage) and fresh water were supplied daily. Behavior was recorded with a low light CCD digital camera connected to a DVD recorder.



## Surgical Procedure

After acclimatization, surgical implantation of the telemetric recording device was performed. The animals were weighed before surgery and anesthetized with weight-appropriate doses of a 2:1 mixture of ketamine and xylazine (Anaket-V and Chanazine 2% Injection, Bayer HealthCare). The head and neck, left thoracic (two 2 × 1 cm) and abdominal (10 × 10 cm) regions were shaved and cleaned with chlorhexidine disinfectant (CHX, 0.5% chlorhexidine digluconate in 75% alcohol, Kyron Laboratories Pty Ltd.) before surgery commenced. These areas correspond to the regions where the EEG, EMG and ECG electrodes and telemeter would be implanted. The animal was placed on a heat blanket in order to maintain a constant body temperature throughout the surgery and the head was placed in a stereotaxic frame to prevent movement and allow for the accurate placement of the EEG and EMG electrodes. During the surgical procedure the animal was kept under a constant state of anesthesia by means of isoflurane ventilation (1%–2% in an oxygen/70% nitrous oxide mixture, isoflurane inhalation anesthetic, Safe Line Pharmaceuticals Pty Ltd.). The animal's heart rate, body temperature and percentage oxygen saturation were monitored throughout the surgery.

Under aseptic conditions, a midsagittal incision was made over the skull and the skin and temporal muscle were reflected to expose the part of the skull overlying the motor cortex. Using a dental drill, three 2-mm-diameter holes were made in the cranial vault to expose the underlying dura mater. The first hole was drilled anterior to the olfactory bulbs for the placement of the indifferent electrode, while two holes were drilled approximately 5 mm apart just lateral to the sagittal sinus over the left motor cortex for the placement of the stainless-steel recording electrodes (gauge of electrode 0.457 mm, silastic outside diameter 0.9 mm and inside diameter 0.508 mm, PhysioTel<sup>r</sup> Multiplus Transmitter, Data Sciences International). The electrodes were placed in such a manner that the tips rested firmly on the surface of the cortex and were secured in place with dental cement. Two of the stainless-steel EMG electrodes (1.5 cm apart) were sutured into the dorsal nuchal musculature, while two ECG electrodes (3 cm apart) were sutured into the subdermal tissue overlying the left thoracic region. A subcutaneous pocket was created (10 × 10 cm) over the left abdominal region for the implantation of the telemetry unit. All skin incisions were sutured following implantation. After surgery was complete, the animal was given an intramuscular analgesic (0.1 ml Tamgesic, Schering-Plough, mixed with 0.9 ml sterile water, 1 ml mixture/kg) and returned to the recording enclosure. Recovery was monitored every half hour until it could be established that the animal was able to move freely and eat and drink normally.

## Sleep Recording

After the surgical procedure the animal was allowed a recovery period of 1 week before the recording of sleep commenced. The animals were housed in the same enclosure (i.e., the enclosure they were acclimatized to prior to surgery and experimentation) within a sound-attenuating room for recovery as well as recording. A receiver was mounted and secured to one wall of the enclosure while a low light CCD digital camera

 as mean was mounted above the enclosure. The telemetric recording system (Data Sciences International, DSI, PhysioTel Multiplus Transmitter, model TL10M3-D70-EEE implant—this particular model provided the strongest signal in our recording enclosure, smaller transmitters were unable to transmit a signal to the receiver as the distance over which it had to operate was too long) consisted of a DEM multiplex interface to which the receiver was connected. The signal from the implanted transmitter (round, 13 cm<sup>2</sup> with stainless steel electrodes, weight 37 g, volume 25 ml, 3 channels) detected by the receiver was relayed to the input amplifier of the Data Sciences computer system, after which it was digitally recorded (in DSI format) for analysis. After the recording was completed, data digitally saved in the DSI format was converted to text format and these files were in turn converted into the appropriate format needed for recognition and analysis by the Spike 2 computer program (version 4.2, Cambridge Electronic Design).

## Data Analysis

Version 4.2.2 of the Spike 2 software (Cambridge Electronic Designs, UK) was used in order to convert the recorded data into the appropriate format, i.e., Spike 2 data format, for offline analysis. The EEG data was scored in 5-s epochs as: (1) wake, characterized by low-voltage, high-frequency EEG and high-voltage EMG; (2) SWS, characterized by high-voltage, low-frequency EEG and EMG lower in amplitude than waking; (3) REM, characterized by low-voltage, high-frequency EEG, an almost atonic EMG and irregular ECG, or (4) somnus

innominatus (SI), characterized by a low-voltage, high-frequency EEG, EMG amplitude characteristic of SWS and a regular ECG (Gravett et al., 2012; **Figure 1**). An epoch was only assigned to a particular state if the state occupied at least 50% of the epoch. The modal state per minute was calculated from the 5-s epoch data and was used in all further analyses. The power spectrum for each of the defined states was calculated with the Spike 2 computer program (Hanning window, FFT number 512, sampling frequency 500 Hz, segment length 1.024 s; see Gravett et al., 2012) and converted from

episodes (middle set of graphs) and episode duration (right set of graphs) for solitary and social conditions for the 24 h (left set of graphs on each individual graph), dark (middle set of graphs on each individual graph) and light (right set of graphs on each individual graph) periods. No significant difference was noted for total state time, number of episodes and episode duration for each of the defined states for the 24 h, light and dark periods, except for REM episode duration, which was greater during social conditions for the 24 h and dark periods. Significant differences between solitary and social conditions are indicated by a star and the respective p-value is indicated on the graph (dependent t-test, p > 0.05, d.f. = 4, please refer to results section for respective t—values). For each condition, there was also no difference between the light and dark periods for total state time, number of episodes and episode duration for each of the defined states (dependent t-test, p > 0.05, d.f. = 4). The mean is shown by the circles and the median by the horizontal bar within each box.

mV<sup>2</sup> to mV which was used in all subsequent statistical analyses.

Behavior was scored in 1-min epochs as: (1) immobile—animal was completely immobile for >30 s; (2) quiet waking—animal was immobile and only moving its head or made minor movements in the same place for >30 s; (3) active waking—animal was actively moving around for >30 s (this state included exploratory and grooming behavior); or (4) eating/drinking—animal was eating and/or drinking for >30 s.

Data was tested for normality prior to statistical analysis (Shapiro-Wilk's W test for normality, p > 0.05). All data was normally distributed and there was no need for data transformation. A t-test for dependent variables was used in all statistical analyses and a significant difference was obtained in all cases with p < 0.05. Statistical tests were performed to determine whether significant differences existed between the solitary and social settings with regard to total state times, number of episodes, and episode duration for the 24 h, light and dark periods. The dependent t-test was also used to determine whether significant differences existed between the solitary and social setting with regard to sleep cycle length, SWA during SWS and during all states as well as the behaviorally defined states for the 24 h, light and dark periods. The Pearson correlation test was used to determine the degree of correlation between the behavior of the implanted and non-implanted animals (significant correlation, p < 0.05). Microsoft Excel, Minitab 17 and GraphPad Prism 6 were used in the scoring, analysis and graphing of the data.

## RESULTS

The physiologically measurable parameters of sleep as well as the associated behaviors were recorded in a total of five hyraxes continuously for a period of 72 h under solitary conditions (Gravett et al., 2012). This was followed by the introduction of one or two other non-implanted hyraxes to the enclosure of the implanted hyrax, followed by 72 h of continuous recording under social conditions. The polygraphic data was scored in 5 s epochs as wake, SWS, SI or REM (**Figure 1**) and the modal state per minute was determined and used in all further analyses (Gravett et al., 2012). There was no significant difference between solitary and social conditions for total state times, number of episodes, episode duration or SWA for all states. A significant difference was however observed for REM sleep duration. REM sleep duration was on average 20 s longer daily and 40 s longer during the dark period under social conditions (**Table 1**). For brevity, only the social condition results are reported in the sections that follow. Please refer to **Tables 1, 2** for a complete summary of all mean ± SER values for both solitary and social conditions for all periods.

## Time Budgets

On a daily basis, under social conditions, the percentage of time spent awake was on average 66.8% (± 16.0 h), SWS 25.7% (± 6.2 h), SI 2.8% (± 40.3 min), and REM sleep 0.6% (± 8.6 min). During the light- and dark periods, 68.7% (± 8.2 h) and 64.0% (± 7.7 h) of time was spent awake, 26.7% (± 3.2 h) and 25.3% (± 3.0 h) in SWS, 2.7% (± 19.4 min) and 2.9% (± 20.9 min) in SI, and 0.5% (± 3.6 min) and 0.64% (± 4.6 min) in REM sleep, respectively. These values did not differ significantly from the daily, light- or dark period results obtained under solitary conditions (dependent t-test, p > 0.05, d.f. = 4 in all instances; **Table 1**, **Figures 2**, **3**).

## Number of Episodes

The daily average number of wake-, SWS-, SI- and REM sleep episodes under social conditions amounted to 122, 113, 26 and 5, respectively. During the light period, the average number of episodes for wake was 62, 60 for SWS, 12 for SI, and 3 for REM sleep. On average, the dark period consisted of 60 wake-, 53 SWS-, 14 SI-, and 2 REM sleep episodes. Like total state times, the average number of episodes for each of the defined states, for all periods, under social conditions did not differ significantly from those observed under solitary conditions (dependent t-test, p > 0.05, d.f. = 4 in all instances; **Table 1**, **Figure 3**).

## Duration of Episodes

The average duration of wake episodes under social conditions remained relatively constant across the periods (daily—476 s, light period—478 s, dark period—478 s). SWS episode duration was similar daily (196 s) and during the light (191 s) period and increased marginally during the dark (201 s) period. SI

(A) and social conditions (B). The most common state transition pathway for both conditions is wake → SWS → wake. Under social conditions: (1) REM transitions more frequently to SWS and wake and less frequently to SI; (2) SI transitions more frequently to REM; (3) wake transitions to REM but not to SI. All other state transition probabilities are similar between the two conditions. The second most common state transition pathway during both conditions is wake → SWS → SI → wake.

episode duration was relatively constant (daily—94 s, light period—96 s, dark period—92 s), while REM sleep episode duration showed some variation across all three periods (daily—98 s, light period—84 s, dark period—114 s). For the majority of the defined states, across all periods, social sleep episode duration did not differ significantly from those under solitary conditions (dependent t-test, p > 0.05, d.f. = 4 in all instances), however a significant difference did exist between conditions for REM sleep episode durations, daily and during the dark period (dependent t-test, p = 0.009, t = 3.421 (daily); p = 0.047, t = −2.824 (dark period); **Table 1**, **Figure 3**). REM sleep duration was on average approximately 20 s longer daily and 40 s longer during the dark period under social conditions.

## State Transition Probabilities

The most common state transition pathway during both conditions was wake → SWS → wake. Under social conditions: (1) REM transitioned more frequently to SWS and wake and less frequently to SI; (2) SI transitioned more frequently to REM; (3) wake transitioned to REM but not to SI. All other state transition probabilities were similar between the two conditions. The second most common state transition pathway during both conditions was wake → SWS → SI → wake (**Figure 4**).

## Sleep Cycle Length and Architecture

Sleep cycle length was calculated for both conditions as the time between the onset of one REM sleep episode to the next. Due to the nature of the sleep architecture of the rock hyrax, as well as the limited amount of REM sleep observed in this species, the sleep cycle length was calculated with wake episodes included and excluded. The average sleep cycle length under social conditions with wake episodes included was 222.8 min and 66.7 min when wake episodes were excluded. No significant difference was noted between social and solitary conditions with regard to sleep cycle length, including or excluding wake episodes (dependent t-test, p > 0.05 and d.f. = 56 in all instances; **Table 2**, **Figures 5A1,B1**). The time of day during which the longest sleep cycle occurred was highly variable between days, individuals and experimental conditions.

The sleep cycle was further examined to determine the contribution of each of the defined states to the sleep cycle. When wake episodes were included as part of the sleep cycle, in the social condition, it accounted for 68.5% of the cycle composition, SWS 28%, SI 2.9% and REM sleep 0.9%. However, when wake episodes were excluded from the sleep cycle, it consisted of 87.5% SWS, 9.8% SI, and 3.2% REM sleep. No significant differences were noted between conditions with regard to the contributions of each of the defined states to the composition of the sleep cycle, but it is worth noting that the percentage contribution of REM sleep to the sleep cycle, including and excluding wake episodes increased in the social setting by 125% and 128%, respectively (**Table 2**, **Figures 5A2,3,B2,3**).

With regard to the states that most commonly preceded REM sleep in the sleep cycle, in the social condition SI preceded REM 50.8% of the time, followed by SWS (41.3%) and wake (7.9%), whereas in the solitary condition SWS preceded REM most often (51.2%) followed by SI (47%) and wake (1.5%). The states that most commonly followed REM in the sleep cycle were similar in the social and solitary conditions, with wake being most common, followed by SWS and SI (**Table 2**).

## Slow Wave Activity

The average SWA based on 2-h intervals was calculated for SWS and all other states. Under social conditions SWA was greatest during SWS (8.7 mV) and gradually declined in intensity from wake (4.6 mV) to SI (3.9 mV) to REM sleep (2.9 mV). These results were similar to those observed under solitary conditions and no significant difference was observed between conditions for all periods (dependent t-test, p > 0.05 and d.f. = 2 in all instances; **Table 1**, **Figure 6**).

## Instantaneous Heart Rate

Instantaneous heart rate (IHR) under social conditions was on average 174 bpm (SER ± 25.2) for wake, 155 bpm (SER ± 6.9) for SWS, 156 bpm (SER ± 9.2) for SI, and 110 bpm (SER ± 14.2) for REM sleep. Under solitary conditions IHR was on average 154 bpm (SER ± 7.8) for wake, 158 bpm (SER ± 11.0) for SWS, 153 bpm (SER ± 10.6) for SI, and 163 bpm (SER ± 27.3) for REM sleep. IHR during wake and REM sleep was significantly different between conditions (dependent t-test, p = 0.000 (both), t = 14.85 (wake) and −18.74 (REM sleep), d.f.= 4 for both). IHR during wake was greater and more variable under social conditions, whereas IHR during REM sleep was greater and more variable under solitary conditions.

## Light vs. Dark

Despite the rock hyrax being naturally a diurnal mammal, under controlled laboratory conditions this does not seem to

TABLE 2 | Average sleep cycle length (A) and individual state contributions to the sleep cycle (B) including and excluding wake episodes, and state occurrences before and after REM sleep episodes (C).


All values expressed as mean ± SER.

Social 7.9 41.3 50.8 69.8 28.6 1.6

and social (right) conditions. No significant differences were noted with regard to the contribution of each of the defined states to the sleep cycle (including and excluding wake episodes) between solitary and social conditions, however it is worth noting that the contribution of REM sleep to the sleep cycle, including and excluding wake episodes is increased in the social condition by 125% and 128%, respectively.

hold true. No significant difference existed between the light and dark periods for both conditions for total state times, number of episodes or episode durations (dependent t-test, p > 0.05 and d.f. = 4 in all instances; **Table 1**, **Figure 3**). A significant difference between the light and dark periods was however noted under social conditions for SWA during wake (p = 0.002, t = −20.21), SI (p = 0.003, t = −18.07) and REM sleep (p = 0.043, t = −4.67; dependent t-test, d.f. = 2 in all instances). SWA during the aforementioned states was greatest during the dark period compared to the light period (**Table 1**, **Figure 6**). Further significant light-dark differences were noted under solitary conditions for immobile behavior (p = 0.03, t = 3.28), and active wake behavior under both conditions (social: p = 0.029, t = −3.34; solitary: p = 0.017, t = −3.96; dependent t-test, d.f. = 4 in all instances). Animals were more immobile during the light period under solitary conditions, whilst more active wake occurred during the dark period under both conditions (**Table 1**, **Figure 7**).

## Behavioral Data

Behavior was recorded in concert with the physiological recordings and scored in 1 min epochs as immobile, quiet wake, active wake (which included exploratory and grooming behavior) or eating and drinking. Daily, under social conditions, approximately 71% (± 17.0 h) was spent immobile, 21% (± 5.0 h) in quiet wake, 3.3% (43.2 min) in active wake, and 3.3% (43.2 min) eating and drinking. No significant differences were noted for each of the behaviorally defined states between conditions daily (dependent t-test, p > 0.05 and d.f. = 3 in all instances), with the exception of active wake which was on average 30 min longer under solitary conditions (dependent t-test, p = 0.0017, t = 3.464 and d.f. = 3; **Table 1**, **Figure 7**). During the light and dark periods respectively, under social conditions, 75% (± 9.0 h) and 68% (8.2 h) of the time was spent immobile, 19% (2.3 h) and 23% (2.8 h) in quite wake, 2% (14.4 min) and 4% (28.8 min) in active wake, and 2% (14.4 min) and 4% (28.8 min) eating and drinking (**Table 1**, **Figure 7**). No significant differences were noted with regard

FIGURE 6 | Box plots showing slow wave activity (SWA; based on 2-h intervals) for all states for the 72 h recording period for solitary and social conditions. Data presented are the daily averages of three animals. No significant differences were observed between solitary and social conditions across all periods (dependent t-test, p > 0.05, d.f. = 2). Significant differences were noted between the light and dark periods for SWA during wake, SI and REM sleep under social conditions. For each of these sates SWA was greater during the dark period (dependent t-test, p < 0.005, d.f. = 2, please refer to results section for specific p- and t-values). The mean is shown by the circles and the median by the horizontal bar within each box.

FIGURE 7 | Box and whisker plots illustrating the percentage of time occupied by each behavioral state for the 24 h, light and dark periods under solitary and social conditions. All behavioral states, with the exception of active wake, showed no significant difference between solitary and social conditions. More active wake is observed under solitary conditions for the 24 h and dark periods. A star indicates significant differences between solitary and social conditions and the respective p-value is indicated on the graph (dependent t-test, p > 0.05, d.f. = 3, please refer to results section for respective t—values). Significant differences were also noted between the light and dark periods for immobility under solitary conditions and active wake under both conditions. In all instances, these behaviors were greater during the dark period (dependent t-test, p < 0.005, d.f. = 2, please refer to results section for specific p- and t-values). The mean is shown by the circles and the median by the horizontal bar within each box.

to the amount of time spent in each of the behaviorally defined states between conditions for the light and dark periods (dependent t-test, p > 0.05 and d.f. = 3 in all instances), with the exception of active wake which was on average 25 min longer during the dark period under solitary conditions (dependent t-test, p = 0.037, t = 3.174, and d.f. = 3; **Table 1**, **Figure 7**).

In the social setting, the behavior of the non-implanted animals was also scored and analyzed. The non-implanted animals spent similar amounts of time immobile (79.7%), and eating and drinking (3.4%) to that reported for the implanted animals in the social and solitary conditions, but they spent less time in the quiet wake (7.6%) state and showed increased amounts of active wake (3.4%). A significant correlation (Pearson's r = 0.3809, p < 0.0001) existed between the behavior of the implanted and non-implanted animal(s), and in all cases the probability of the behavior of the implanted vs. non-implanted animals being unrelated was found to approach zero. In addition, a 72.9% agreement was also observed with regard to the implanted and non-implanted animals entering the same behavioral state at the same time (**Figure 8**).

## DISCUSSION

The aim of the present study was to investigate the effect of sociality on sleep architecture in a naturally occurring social, diurnal species, the rock hyrax. Most studies that have reported on the possible effects of sociality on the evolution of sleep in mammals have compared solitary species to social species. It has been hypothesized that social species sleep less, have more fragmented sleep and lower NREM and REM sleep quotas (Capellini et al., 2008a,b, 2009). It is also believed that social species invest more time in social interactions, which in effect leave them with less time to sleep. According to these studies, social species can enter deeper stages of sleep and in doing so sleep more effectively in a shorter period of time (Capellini et al., 2009). As these hypotheses are primarily based on observations comparing solitary and social species and not solitary and social conditions within the same species, the results from the present study are discussed in relation to these hypotheses and cannot support or refute them.

## Total Sleep Times and Sleep Fragmentation

Our results revealed no significant difference in total sleep time or the average number of sleep episodes between conditions across all periods. In general, a negligible increase in total sleep time (combined SWS, SI and REM sleep time) was observed under social conditions while the number of sleep episodes (combined SWS, SI and REM sleep episodes) increased marginally under solitary conditions. We can thus conclude that under social conditions in a controlled laboratory environment, total sleep time does not become reduced and sleep does not become more fragmented.

## NREM and REM Sleep Quotas and Sleep Cycle Composition

NREM and REM sleep quotas are said to be reduced in social species (Capellini et al., 2009). In the current study, we did not see a significant difference between conditions across all periods for SWS and REM sleep quotas as they were only marginally increased under social conditions. The average duration of REM sleep episodes was however significantly different daily and during the dark period when the conditions were compared. The average duration of REM sleep episodes was approximately 20 s and 40 s longer under social conditions, daily and during the dark period respectively. Looking at the composition of the sleep cycle our results also revealed that the percentage contribution of REM sleep to the sleep cycle increased under social conditions by 125% when wake episodes were included, and 128% when wake episodes were excluded. It was also noted that the range of REM sleep episode duration within the sleep cycle was also greater under social conditions compared to solitary conditions.

It is possible that the increase in REM sleep duration and contribution to the sleep cycle under social conditions could be linked to improved thermoregulation. Thermoregulation in the rock hyrax involves both physiological and behavioral (i.e., basking and huddling) mechanisms. Laboratory based studies have shown that hyraxes are capable of maintaining a constant body temperature when exposed to a range of ambient temperatures that they would naturally encounter. In addition, these studies also reported that hyraxes become hypothermic at night and rely on morning solar basking as opposed to metabolic warming to increase body temperature (Taylor and Sale, 1969). Apart from solar basking, hyraxes also employ another behavioral strategy to conserve body heat, especially during the night—huddling. Energy expenditure is reduced during the sleep phase when animals huddle together as the surface area to which heat is lost to the environment is reduced (Weatherhead et al., 1985). Brown and Downs (2006) thus hypothesized that social behavior in the rock hyrax may play a significant role in thermoregulation.

It has been reported that a relationship exists between total REM sleep time and the temperature of an animal's environment. At thermoneutral temperatures total REM sleep time is maximal, whereas heat or cold stress results in a decline. Studies have also shown that a brief change in skin temperature toward thermoneutrality is capable of triggering REM sleep in cold stressed neonatal rats (Szymusiak et al., 1980; Szymusiak and Satinoff, 1981). Thus, it is possible that the increased contribution of REM sleep to the sleep cycle as well as the increase in the average REM sleep duration, especially during the dark period, could be accounted for by improved thermoregulation during social conditions. Huddling together at night could aid in maintaining a constant body temperature throughout this period and the skin-to-skin contact could possibly activate peripheral thermoreceptors, potentially resulting in more frequent REM episodes. REM sleep is not the only state affected by raised skin temperature. A number of studies have reported that increases in skin temperature also affect NREM sleep in both humans and animals. Deeper

stages of NREM sleep has been reported with as little as a 0.4◦C increase in skin temperature in young and old healthy and insomniac participants, while afternoon body heating in humans and whole-body heating during the last 4 h of the light period in rats have been associated with increased NREM quotas (Morairty et al., 1993; Raymann et al., 2008; Romeijn et al., 2012). Despite these reports, our results did not show a significant increase in total SWS time or duration, as the increases seen under social conditions were negligible. As body temperature was not measured in the current study, our interpretation of the increased REM sleep duration and contribution to the sleep cycle as a possible result of improved thermoregulation under social conditions, is highly speculative and requires further investigation.

## SWA and Sleep Efficiency

Another hypothesis related to sleep in social species proposes that socially sleeping species enter deeper stages of sleep and thus sleep more efficiently over shorter periods of time. This need to maximize sleep efficiency in a shorter time frame is believed to be due to a trade-off that exists between times devoted to social interactions and sleep—more time is invested in social interactions leaving less time for sleep (Capellini et al., 2009). SWA during NREM is considered a measure of sleep intensity and sleep debt, as well as an indicator of NREM sleep homeostasis (Borbély and Neuhaus, 1980; Daan et al., 1984; Tobler and Borbély, 1986; Meerlo et al., 1997, 2001; Borbély and Achermann, 1999). SWA is usually highest during the first NREM sleep episode and decreases during later NREM sleep episodes. The intensity of NREM SWA is reportedly also influenced by the duration as well as the nature of the prior wake experience (Bellesi et al., 2014). Sleep deprivation, intensive learning, social defeat and stress during the preceding wake experience can all lead to an increase in NREM SWA (Meerlo et al., 1997, 2001; Bellesi et al., 2014; Kamphuis et al., 2015). As our results showed no significant difference in SWA for all states between conditions across all periods we can conclude that sleep efficiency does not appear to be significantly enhanced under social conditions in the rock hyrax.

## CONCLUSION

The current study provides four results of interest regarding the effects of sleeping in a social setting for the rock hyrax: (1) total sleep times remain similar to solitary conditions; (2) sleep does not become more fragmented; (3) sleep efficiency remains unchanged; and (4) REM sleep episode duration and overall contribution to the sleep cycle increases. The

## REFERENCES


significant increase in REM episode durations and increased contribution to the sleep cycle could be a function of improved thermoregulation under social conditions, however this is highly speculative and requires further investigation. Based on these findings, it is possible to assume that sleep quality does not improved significantly under social conditions in the rock hyrax, however considering the limited sample size and design of the current study further investigations are needed to confirm this finding. Whether the conclusions and the observations made in this study can be generalized to all naturally socially sleeping mammals remains an open question.

## AUTHOR CONTRIBUTIONS

NG and PRM conceptualized the study. PRM obtained the funding. NG, PRM, AB and OIL performed the experimental procedures. NG analyzed the data and wrote the manuscript. PRM, AB, OIL and JMS contributed to the editing and improvement of drafts of the manuscript. All authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

## FUNDING

This work was supported by funding from the National Research Foundation of South Africa (PRM).

## ACKNOWLEDGMENTS

We gratefully acknowledge the assistance of the members of the Central Animal Services of the University of the Witwatersrand throughout this study.


**Conflict of Interest Statement**: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Gravett, Bhagwandin, Lyamin, Siegel and Manger. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Effects of Gladiolus dalenii on the Stress-Induced Behavioral, Neurochemical, and Reproductive Changes in Rats

David Fotsing1,2, Gwladys T. Ngoupaye<sup>3</sup> , Agnes C. Ouafo<sup>1</sup> , Stephanie K. J. Njapdounke<sup>2</sup> , Yongabi A. Kenneth<sup>4</sup> and Elisabeth Ngo Bum3,5 \*

<sup>1</sup> Department of Biological Sciences, Faculty of Science, University of Bamenda, Bambili, Cameroon, <sup>2</sup> Department of Biological Sciences, Faculty of Sciences, University of Ngaoundéré, Ngaoundéré, Cameroon, <sup>3</sup> Department of Animal Biology and Physiology, Faculty of Science, University of Dschang, Dschang, Cameroon, <sup>4</sup> Directorate of Research, Catholic University of Cameroon, Bamenda, Cameroon, <sup>5</sup> Institute of Mines and Petroleum Industries, University of Maroua, Maroua, Cameroon

#### Edited by:

Nilesh Bhailalbhai Patel, University of Nairobi, Kenya

#### Reviewed by:

Johanna Mahwahwatse Bapela, University of Pretoria, South Africa Yaxing Zhang, Zhongshan School of Medicine, Sun Yat-sen University, China

> \*Correspondence: Elisabeth Ngo Bum eli\_bum@yahoo.fr

#### Specialty section:

This article was submitted to Ethnopharmacology, a section of the journal Frontiers in Pharmacology

Received: 01 June 2017 Accepted: 13 September 2017 Published: 27 September 2017

#### Citation:

Fotsing D, Ngoupaye GT, Ouafo AC, Njapdounke SKJ, Kenneth YA and Ngo Bum E (2017) Effects of Gladiolus dalenii on the Stress-Induced Behavioral, Neurochemical, and Reproductive Changes in Rats. Front. Pharmacol. 8:685. doi: 10.3389/fphar.2017.00685 Gladiolus dalenii is a plant commonly used in many regions of Cameroon as a cure for various diseases like headaches, epilepsy, schizophrenia, and mood disorders. Recent studies have revealed that the aqueous extract of G. dalenii (AEGD) exhibited antidepressant-like properties in rats. Therefore, we hypothesized that the AEGD could protect from the stress-induced behavioral, neurochemical, and reproductive changes in rats. The objective of the present study was to elucidate the effect of the AEGD on behavioral, neurochemical, and reproductive characteristics, using female rats subjected to chronic immobilization stress. The chronic immobilization stress (3 h per day for 28 days) was applied to induce female reproductive and behavioral impairments in rats. The immobilization stress was provoked in rats by putting them separately inside cylindrical restrainers with ventilated doors at ambient temperature. The plant extract was given to rats orally everyday during 28 days, 5 min before induction of stress. On a daily basis, a vaginal smear was made to assess the duration of the different phases of the estrous cycle and at the end of the 28 days of chronic immobilization stress, the rat's behavior was assessed in the elevated plus maze. They were sacrificed by cervical disruption. The organs were weighed, the ovary histology done, and the biochemical parameters assessed. The findings of this research revealed that G. dalenii increased the entries and the time of open arm exploration in the elevated plus maze. Evaluation of the biochemical parameters levels indicated that there was a significant reduction in the corticosterone, progesterone, and prolactin levels in the G. dalenii aqueous extract treated rats compared to stressed rats whereas the levels of serotonin, triglycerides, adrenaline, cholesterol, glucose estradiol, follicle stimulating hormone and luteinizing hormone were significantly increased in the stressed rats treated with, G. dalenii, diazepam and in co-administration of the plant extract and diazepam treated rats. Moreover stressed rats showed significant changes in estrous cycle phases compared to vehicle control and these changes of the estrous cycle were less in the rats treated

**85**

with G. dalenii compared to the negative control rats. G. dalenii extract showed antagonizing effects on the stress-induced reproductive, behavioral, and neurochemical changes. These effects could be related to the bioactive molecules and secondary metabolites like alkaloids and flavonoids in the plant.

Keywords: biochemical parameters, estrous cycle, Gladiolus dalenii, neurochemical, restraint stress

## INTRODUCTION

fphar-08-00685 September 25, 2017 Time: 13:39 # 2

Individuals frequently face stressful conditions. Chronic stress consistently activates the hypothalamic–pituitary–adrenal (HPA) axis. Each individual component of the HPA axis exerts deleterious effect on the hypothalamic–pituitary–gonadal axis and subsequently leads to human reproductive failure (Nepomnaschy et al., 2004; Chatterjee et al., 2006). During stress induction, many behavioral, biochemical, and reproductive parameters are altered. The stress-induced alterations have been attributed to an imbalance in the neuroendocrine system (Kenjale et al., 2007). Therefore, assessment of some of the biochemical, endocrinal, and behavioral parameters will serve as an important basis for the evaluation of anti-stress activity (Rai et al., 2003). Biological responses to stress are known to suppress reproductive function across the human life course. For example, hypothalamic amenorrhea, a clinical condition without endocrine or systemic cause, is triggered by metabolic, physical, or psychological stress as well as high stress perception is a risk factor for severe premenstrual pain or ovarian dysfunction (Woods et al., 1998; Kaplan and Manuck, 2004; Genazzani et al., 2006). Impairment of reproductive outcomes is triggered by stress-inducing factors and is more established in women susceptible to a physiological stress response (Brotman et al., 2007). Unlike in the males, the level of corticotropin-releasing factor (CRF) in the female hypothalamus is very important (Fredericksen et al., 1991), therefore, the females HPA axis responds to stress more intensely than the males (Lund et al., 2004). In women in their working environment, persistent stimulation of the HPA axis has been shown to hamper the hypothalamic–pituitary–ovarian axis (Imaki et al., 1996; Fenster et al., 1999). Stress powerfully stimulates the hypothalamus and extra-hypothalamic sites for the release of CRF (Horrocks et al., 1990). Rats with normal estrous cycle restrained inside a cylindrical restrainers (stress induction) exhibit behavioral, neurochemical, and reproductive impairments (Bowman et al., 2001; Sivaprasad et al., 2015).

In the stress-related disorders, the treatments available include the anxiolytics like the benzodiazepines. These treatments have a broad numbers of side effects such as the muscle relaxation, the memory loss, and addiction (Lader and Morton, 1991; Czobor et al., 2010). These limits have developed more interest in the use of natural products to treat stress-related disorders. Several plants have proven anxiolytic-like effects in animal models. For example, Afrormosia laxiflora (Benth) Harms (Fabaceae), Chenopodium ambrosioides Linn (Chenopodiaceae), Microglossa pyrifolia Kuntze (Lam) (Asteraceae), Mimosa pudica Linn (Mimosaceae), Nelsonia canescens (Acanthaceae), and Gladiolus dalenii Van Geel (Iridaceae) (Ngo Bum et al., 2011; Ngoupaye et al., 2013a; Fotsing et al., 2016). G. dalenii is plant of the Iridaceae family generally used in Cameroon pharmacopoeia, especially in the West region to cure various ailments like schizophrenia, depression, and headaches. The research conducted by Adejuwon et al. (2013) revealed that uncontrolled consumption of the aqueous extract of G. dalenii (AEGD) had toxic effects in male rat's reproductive parameters notably the spermatozoa. Past studies indicated that G. dalenii corm crude extract had antifungal activity (Odhiambo et al., 2010), anticonvulsivant and sedative effects (Ngoupaye et al., 2013b), and antidepressant-like effects in epileptic mice (Ngoupaye et al., 2013a). The data indicated that the ability of G. dalenii to antagonize the PTZ-induced seizures could be attributed to its modulatory effects on the GABA<sup>A</sup> receptor neurotransmission. Also, the antidepressant activity could be mediated through the restoration of the HPA axis activities (Ngoupaye et al., 2013a). Therefore, this study focuses on the effects of the AEGD on the stress-induced behavioral and physiological changes in non-epileptic chronic stressed adult female rats. These effects were assessed using the elevated plus maze test, and measuring the stress markers and the reproductive parameters. The effects of G. dalenii were compared to those of diazepam, a benzodiazepine that binds to a specific subunit on the GABA<sup>A</sup> receptor, inducing anxiolytic effects (Campo-Soria et al., 2006).

## MATERIALS AND METHODS

## Plant and Aqueous Extract Preparation

The corms of G. dalenii were harvested in Babajou in the West region of Cameroon during the month of November 2013 and were identified at the national Herbarium of Yaoundé (number 25742/SRF/Cam). They were cleaned in water to remove dust and mud, dried under ambient air, and ground to get a powder that was used to prepare the AEGD (Fotsing et al., 2016). The aqueous extract was provided by macerating 250 g of the air dried powder of G. dalenii in 6 l of distilled water for 72 h at room temperature. The preparation was filtrated and the filtrate was evaporated to dryness in an oven at 35◦C and 24.6 g of a brown solid extract was obtained. The yield of the extraction was 9.84%.

## The Animals and Experimental Design

Adult female Wistar albino rats with normal estrous cycle were used in the study. They were provided from the animal unit of the University of Bamenda. These rats were raised in standard conditions (room temperature, 12/12 h light–dark cycle). They were supplied with pellets and water ad libitum. The study was carried out in accordance with the Cameroon National

Ethical Committee (Ref No. FW-IRB00001954, 22 October 1987 with an authorization number CEI-UDo/909/01/2017/T), and in conformation with the international regulation, minimizing the number of rats used and their suffering. The animals were organized into six groups containing five rats each. The vehicle control group was given distilled water and was kept unstressed, the negative control group and the positive group were respectively treated with the distilled water and the reference substance, diazepam (3 mg/kg) and were stressed. Two experimental groups received two doses of the extract (7.5 or 15 mg/kg), and were stressed. An experimental group was treated in co-administration with the plant extract (15 mg/kg) and diazepam (3 mg/kg) and was stressed. The doses used in the experiment were based on previous study (Ngoupaye et al., 2013a). The experimental rats were adapted in the new environment for 2 weeks and then were stressed. The restraint stress involved confining rats inside individual plastic cylindrical restrainers (21 cm in length × 6 cm in diameter) with ventilated sliding doors at ambient temperature (Bowman et al., 2001; Ngoupaye et al., 2013b; Sivaprasad et al., 2015). This restraint stress procedure was performed 3 h daily for 28 days. Right after the stress period, a vaginal smear was prepared to find out the consequent stage of the estrous cycle. Vaginal smears were obtained by placing a small drop of saline in the vagina with a blunted Pasteur pipette and removing a sample of vaginal cells which were immediately observed microscopically under low magnification (Baron and Brush, 1979; Saraswathi et al., 2010; Sivaprasad et al., 2015). The rats of the vehicle control group were taken to a different experimental room and kept in plasticbox cages unstressed. After the restraint stress sequences, the behavior was evaluated in the elevated plus maze before the rats were sacrificed by cervical disruption. The blood was collected in EDTA tubes and the biochemical parameters measured, while the organs were freed from connective tissues and weighed. The ovaries imbedded in paraffin were cut into 4 µm sections and stained with hematoxylin–eosin for histological analysis.

## Behavioral Assessment

The behavioral test was done with the elevated plus maze (Pellow et al., 1985). The elevated plus maze was placed in an isolated room, far from any irrelevant interference of scents, movement, or noises. The arms of the maze were approximately 90 cm above the floor which was covered by foam rubber. At the beginning of each session, the rat was placed in the central area facing the open arms of the maze and was allowed to explore the maze freely during 5 min. The entries and the time spent in the different arms were recorded. The data were used to calculate the percentage of entries and time spent for each arm. After the assessment of a rat's behavior, the maze was cleaned with alcohol.

## Biochemical Parameters

Blood was centrifuged at 3,000 rpm for 20 min at 4◦C, and 1 ml aliquot of plasma was transferred to 1.5 ml Eppendorf vials and kept at −20◦C. Plasma serotonin, adrenaline, triglycerides, glucose, cholesterol, estradiol, prolactin, follicle stimulating hormone (FSH), luteinizing hormone (LH), progesterone, and corticosterone were measured using commercially available immunoassay kits (Human Gesellschaft für Biochemica und Diagnostica mbH, Wiesbaden, Germany).

## Statistical Analysis

Statistical analysis was done using the software program Statgraphics 11.0. All data are presented as mean ± SEM. Analysis of variance (ANOVA) was done for different groups and means were separated by Newman–Keuls post hoc test and corrected with Bonferroni multiple comparisons test at 5% confident limit.

## RESULTS

## Effect of AEGD on Stress-Induced Behavioral Change

Our results indicated that in the elevated plus maze, the chronic restraint stress (CRS) caused a decrease (though not significant) of the number open arms entries and a significant (p ≤ 0.05) decrease of the open arm time. The AEGD (7.5 mg/kg) significantly increased the number of entries (p ≤ 0.05) and the time spent (p ≤ 0.001) in the open arms from 1.4 ± 0.55 and 16.75 ± 4.27 s in the negative control (CRS) to 6.20 ± 0.84 and 188.2 ± 13.59 s respectively (**Table 1**). As awaited, the diazepam caused a more significant increase in number of entries (p ≤ 0.001) and time spent (p ≤ 0.001) in the open arms. The co-administration of diazepam (3 mg/kg) and AEGD (15 mg/kg) also caused a significant increase in number of entries (p ≤ 0.05) in the open arms (**Table 1**).

The CRS caused a decrease (not significant) of the percentage of the open arm entries. The percentage of entries in the open arms significantly (p ≤ 0.05) increased from 30.33 ± 9.31% in the stressed group to 58.39 ± 5.81 and 63.79 ± 3.4% in the groups treated with AEGD at the dose of 15 and 7.5 mg/kg, respectively. As anticipated, diazepam significantly (p ≤ 0.001) increased the open arm entries percentage to 59.01 ± 7% (**Figure 1A**). The CRS caused a significant (p ≤ 0.005) decrease of the percentage of time spent in the open arm. After the treatments, the percentage of open arm time increased significantly (p ≤ 0.01) from 5.58 ± 1.42% in the negative control group to 59.33 ± 3.27 and 62.73 ± 4.53% in the diazepam and the AEGD at the dose of 15 mg/kg treated groups, respectively (**Figure 1B**). On the other hand, the CRS caused an increase (not significant) of the percentage of close arm time. The AEGD at the dose of 15 mg/kg and the diazepam (3 mg/kg) showed a significant (p ≤ 0.005) decrease of the percentage of close arm time to 37.27 ± 4.53 and 39.31 ± 3.1%, respectively (**Figure 1C**). This stress-induced behavioral change could be attributed to an imbalance in the neuroendocrine system (Rai et al., 2003; Kenjale et al., 2007).

## Effect of AEGD on Stress-Induced Neurochemical Changes

### Effects of AEGD on the Plasma Adrenaline and Serotonin Corticosterone and Prolactin Levels

Restraint stress group showed significant (p ≤ 0.01) decrease in the level of serum serotonin when compared with vehicle control.



Results are expressed as mean ± SEM for the number of entries in the open arm and in the close arm; the time spent in the open arms and in the close arm and the total number of entries. N = 5 per group. Data were analyzed using one-way ANOVA, followed by Newman–Keuls post hoc test and corrected with Bonferroni multiple comparisons test. <sup>∗</sup>p < 0.05, ∗∗p < 0.01 vs vehicle (distilled water); <sup>a</sup>p < 0.05, <sup>b</sup>p < 0.01 vs negative control (vehicle + CRS). CRS, chronic restraint stress; vehicle, distilled water (10 ml/kg, p.o.); GD, G. dalenii (7.5 and 15 mg/kg, p.o.); DZP, diazepam (3 mg/kg, i.p.).

Groups treated with AEGD (15 mg/kg) + diazepam (3 mg/kg) along with the restraint stress showed significant (p ≤ 0.05) increase in the levels of serotonin when compared with restraint stress group (**Table 2**). The significant decrease (p ≤ 0.05) in adrenaline level in negative control group (CRS) compared with vehicle control was significantly (p ≤ 0.05) increased in the groups that received AEGD (15 mg/kg), and diazepam + AEGD (15 mg/kg) along with the restraint stress when compared with restraint stress group (**Table 2**).

Restraint stress group showed significant (p ≤ 0.05) increase in the level of serum prolactin when compared with vehicle control. Group that received AEGD (15 mg/kg) + diazepam (3 mg/kg) along with the restraint stress showed significant (p ≤ 0.05) decrease in the levels of prolactin when compared with restraint stress group (**Table 2**). The significant (p ≤ 0.05) increased in the corticosterone level in negative control group (CRS) compared to vehicle control was significantly (p ≤ 0.05) reduced in the groups that received AEGD (15 mg/kg) and AEGD (15 mg/kg) + diazepam (3 mg/kg) along with the restraint stress (**Table 2**).

#### Effects of AEGD on the Plasma Progesterone, Estradiol, FSH, and LH Levels

Restraint stress group showed a significant (p ≤ 0.05) increase in the level of serum progesterone when compared with vehicle control. Groups that received AEGD (15 mg/kg) body weight, diazepam, and diazepam + AEGD (15 mg/kg) along with the restraint stress showed a decrease (not significant) in the levels of progesterone when compared with restraint stress group (**Table 3**).

The restraint stress group exhibited significant drop in the level of serum estradiol when compared with vehicle control. Groups that received diazepam (3 mg/kg) and diazepam + AEGD (15 mg/kg) along with the restraint stress showed significant (p ≤ 0.05) increase in the levels of estradiol when compared with restraint stress group (**Table 3**).

The assessment of the FSH levels reveals that restraint stress group showed a significant (p ≤ 0.05) decrease in the level of serum FSH when compared with vehicle control. AEGD (15 mg/kg), diazepam (3 mg/kg), and diazepam + AEGD (15 mg/kg) significantly (p ≤ 0.05) reversed the reduction of FSH induced by the CRS (**Table 3**).

The restraint stress group showed significant (p ≤ 0.05) decrease in the level of serum LH when compared with vehicle control. Groups that received AEGD (15 mg/kg), diazepam, and diazepam + AEGD (15 mg/kg) along with the restraint stress showed significant (p ≤ 0.05) increase in the levels of LH when compared with restraint stress group (**Table 3**).

#### Effects of AEGD on the Plasma Cholesterol, Triglycerides, and Glucose Levels

Assessment of the cholesterol levels revealed that CRS significantly (p ≤ 0.05) increased the levels of cholesterol compared with the vehicle control rats. Groups that received diazepam (3 mg/kg) and diazepam + AEGD (15 mg/kg) along with the restraint stress showed a significant increase in cholesterol levels when compared with restraint stress group (**Table 4**). Restraint stress group showed significant (p ≤ 0.05) decrease in the level of serum triglycerides when compared with vehicle control. Groups that received diazepam (3 mg/kg) and diazepam (3 mg/kg) + AEGD (15 mg/kg) along with the restraint stress showed significant (p ≤ 0.05) increase in the concentration of triglycerides when compared with restraint stress group (**Table 4**). Stress rats showed highly significant (p ≤ 0.01) decrease in the level of serum glucose when compared with vehicle control. Groups that received AEGD (15 mg/kg), diazepam (3 mg/kg), and diazepam + AEGD (15 mg/kg) along with the restraint stress showed significant (p ≤ 0.05) increase in the levels of glucose when compared with restraint stress group (**Table 4**). This stress-induced neurochemical imbalance could lead to reproductive impairments (Anderson et al., 1996).

## Effect of AEGD on Stress-Induced Reproductive Changes

#### Effect of AEGD on Stressed Rat's Estrous Cycle

Stressed animals exhibited changes in the mean duration of estrous cycle phases when compared to vehicle control. It was recorded a significant (p ≤ 0.05) increase in proestrus phase length and significant (p ≤ 0.05) decreases in estrous and metestrus phases duration. Groups that received AEGD (15 mg/kg), diazepam (3 mg/kg), and diazepam (3 mg/kg) + AEGD (15 mg/kg) along with the restraint stress showed a significant (p ≤ 0.05) restoration of proestrus, estrous, and metestrus duration compared to stressed groups (**Table 5**).

#### Effect of the AEGD on the Different Organs Weights

Restraint stress group showed significant (p ≤ 0.05) decrease in the adrenal glands weight when compared with vehicle control. Groups that received AEGD (7.5 mg/kg), AEGD (15 mg/kg), diazepam, and diazepam + AEGD (15 mg/kg) along with the CRS showed increase (not significant) in the weight of the adrenal gland when compared with restraint stress group. The CRS also caused a significant decrease in uteri weight compared to vehicle control. Diazepam (3 mg/kg), and diazepam + AEGD (15 mg/kg) significantly (p ≤ 0.05) reversed the uteri weight loss compared to stress rats (**Table 6**).

### Effect of AEGD on the Ovary Histological Analysis of the Treated Rats

The histological analysis of the ovaries revealed localized alterations of tissues in stressed rats (**Figure 2A**) with hyperchromatic nucleus, multiple follicular cysts and atretic follicles and corpus fibrosum compared to the vehicle control group (**Figure 2B**) that had a normal stroma with primary and secondary developing follicles and matured graafian follicle. In the stressed rats and treated with the AEGD at the dose 7.5 mg/kg (**Figure 2C**) and AEGD at the dose of 15 mg/kg (**Figure 2D**), the changes observed in the stress rats were reversed with a normal stroma, developing follicles, and matured graafian follicle.

## DISCUSSION

The results of this study indicate that in the open arms, the number of entries, the time spent and the respective percentages significantly increased in the chronic restraint stressed nonepileptic adult female rats, in the presence of the AEGD at doses of 7.5 and 15 mg/kg and were comparable to the effects of diazepam a recognized anxiolytic dose (3 mg/kg). On the contrary the AEGD significantly decreased the percentage of closed arms entries and time spent. Any increased activity in open arms indicates a decreased anxiety level (Lee and Rodgers, 1991; Rodgers and Dalvi, 1997; Holmes et al., 2000; Majchrzac, 2003). Also, a decrease of these behavioral parameters in the closed arms indicates a reduction of stress level (Lister, 1990; Ngo Bum et al., 2009). These results show the anxiolytic-like activity of the AEGD (Gomes et al., 2010; Souto-Maior et al., 2011). Diazepam is referred to as an anxiolytic in humans and causes decrease in anxiogenic-like. Several studies have reported that diazepam at anxiolytic dose facilitates exploratory behavior which is expressed as increased locomotion in the elevated plus maze (Bhattacharya and Mitra, 1991; Ramanathan et al., 1998). Our findings showed that the animals treated with the AEGD at the doses of 7.5 and 15 mg/kg caused increase in the opened arm entries without increasing the total number of entries thereby leading to not changes in locomotion of rats. In order to further corroborate the anxiolytic activity observed in the EPM test, we also assessed the stress markers levels. Our results showed that stressed rats exhibited anxiogenic behavior associated to reduction of plasma adrenaline and serotonin concentrations and the increased plasma corticosterone, progesterone, and prolactin levels. This reveals that the rats underwent stress and the alteration observed is similar to clinically related pathophysiology of anxiety (Saavedra et al., 1979; Saavedra and Torda, 1980; Heim and Nemeroff, 1999; Millan, 2003). Administration of the AEGD during stress period restored the exploratory behavior of rats. The increased exploratory behavior of rats was correlated with restoration of plasma adrenaline levels (Srinivasan et al., 2003). The results showed that stressed rats treated with the AEGD had corticosterone level significantly reduced almost to normal values (Kyrou and Tsigos, 2009). The HPA axis is made up of

p.o.); DZP, diazepam (3 mg/kg, i.p.).

#### TABLE 2 | Effects of the AEGD on the adrenaline, serotonin, corticosterone, and prolactin levels.


Results are expressed as mean ± SEM for the levels of adrenaline, serotonin, corticosterone, and prolactin. N = 5 per group. Data were analyzed using one-way ANOVA, followed by Newman–Keuls post hoc test and corrected with Bonferroni multiple comparisons test. <sup>∗</sup>p < 0.05 vs vehicle (distilled water); <sup>a</sup>p < 0.05 vs negative control (vehicle + CRS). CRS, chronic restraint stress; vehicle, distilled water (10 ml/kg, p.o.); GD, G. dalenii (7.5 and 15 mg/kg, p.o.); DZP, diazepam (3 mg/kg, i.p.).

TABLE 3 | Effects of the AEGD on the estradiol, FSH, LH, and progesterone hormones levels.


Results are expressed as mean ± SEM for the levels of estradiol, FSH, LH, and progesterone. N = 5 per group. Data were analyzed using one-way ANOVA, followed by Newman–Keuls post hoc test and corrected with Bonferroni multiple comparisons test. <sup>∗</sup>p < 0.05 vs vehicle (distilled water); <sup>a</sup>p < 0.05 vs negative control (vehicle + CRS). CRS, chronic restraint stress; vehicle, distilled water (10 ml/kg, p.o.); GD, G. dalenii (7.5 and 15 mg/kg, p.o.); DZP, diazepam (3 mg/kg, i.p.).

TABLE 4 | Effect of the AEGD on the cholesterol, triglycerides, and glucose levels.


Results are expressed as mean ± SEM for the levels of cholesterol, triglycerides, and glucose. N = 5 per group. Data were analyzed using one-way ANOVA, followed by Newman–Keuls post hoc test and corrected with Bonferroni multiple comparisons test. <sup>∗</sup>p < 0.05, ∗∗p < 0.01 vs vehicle (distilled water); <sup>a</sup>p < 0.05, <sup>b</sup>p < 0.01 vs negative control (vehicle + CRS). CRS, chronic restraint stress; vehicle, distilled water (10 ml/kg, p.o.); GD, G. dalenii (7.5 and 15 mg/kg, p.o.); DZP, diazepam (3 mg/kg, i.p.).

TABLE 5 | Effects of AEGD on mean numbers of days on different phases of estrous cycle (28 days).


Results are expressed as mean ± SEM for the mean number of days of proestrus, estrous, metestrus, and diestrus. N = 5 per group. Data were analyzed using one-way ANOVA, followed by Newman–Keuls post hoc test and corrected with Bonferroni multiple comparisons test. <sup>∗</sup>p < 0.05 vs vehicle (distilled water); <sup>a</sup>p < 0.05 vs negative control (vehicle + CRS). CRS, chronic restraint stress; vehicle, distilled water (10 ml/kg, p.o.); GD, G. dalenii (7.5 and 15 mg/kg, p.o.); DZP, diazepam (3 mg/kg, i.p.).


TABLE 6 | Effect of the AEGD on the different organs weights (g) in the restraint stressed rats.

Results are expressed as mean ± SEM for the weight of liver, ovaries, uterus, and adrenal gland. N = 5 per group. Data were analyzed using one-way ANOVA, followed by Newman–Keuls post hoc test and corrected with Bonferroni multiple comparisons test. <sup>∗</sup>p < 0.05 vs vehicle (distilled water); <sup>a</sup>p < 0.05 vs negative control (vehicle + CRS). CRS, chronic restraint stress; vehicle, distilled water (10 ml/kg, p.o.); GD, G. dalenii (7.5 and 15 mg/kg, p.o.); DZP, diazepam (3 mg/kg, i.p.).

an assembly of stress responses mediated by the brain, pituitary, and adrenal gland. The endocrine activity of the hypothalamus causes the production of the CRF, a compound that stimulates the production of adrenocorticotropic hormone (ACTH). ACTH is liberated into the circulatory system, and causes the adrenal cortex to secrete corticosteroid hormones, particularly cortisol. Cortisol increases the availability of refueling the body with substances necessary for the body's response to stress (Dornhorst et al., 1981). The results showed a significant drop in glucose and triglycerides levels in stressed rats when compared to vehicle control group. These substances levels were reversed and were returned to more normal value in AEGD treated stressed rats, this suggests that the AEGD showed anxiolytic properties. These findings are similar to the results obtained by Fotsing et al. (2016) in the analogous studies with N. canescens. The AEGD is rich in polyphenols, flavonoids, tannins, tripertenes, or other secondary metabolites that may support the anxiolytic activity of the plant (Harsha and Anilakumar, 2013a,b; Ngoupaye et al., 2013a). Anti-anxiety secondary metabolites can interfere with the serotonin and GABA systems; this may explain the similar effects of the AEGD and diazepam, a GABA benzodiazepine agonist, in relieving anxiety (Pravinkumar et al., 2007; Priprem et al., 2008).

To further establish the anxiolytic properties of the AEGD, we studied its effects on the reproductive parameters. Because stress can alter neurotransmitters and hormones involved in the regulation of reproductive physiology, it has been reported that stress affects reproductive function in female (Anderson et al., 1996). Chronic restrained rats showed a significant rise in the mean number of days in proestrus phase and decrease in estrous and metestrus phases (Brotman et al., 2007). This demonstrates the disruption of follicular development at the initial stages causing the non-maturation of follicles (Sivaprasad et al., 2015). The AEGD treated groups showed significant decrease duration of proestrus phase indicating the development of follicles. The treatment also causes a significant increase in the mean days of estrous, metestrus, and diestrus phases. These findings reveal the antagonizing effect of the AEGD against stress-induced estrous cycle changes. It also indicates the maturation of follicles, the formation of Graafian follicles and corpus luteum due to the increased secretion of either gonadotrophic, or steroidal hormones or both (Bhutani et al., 2004). The ovaries are made up of three endocrine tissues, the stroma, the follicle and the corpus luteum. Therefore, the net

weight of the ovaries is the sum of the weights of these tissues. Our study showed that there was a decrease in the ovarian weight in stressed rats. This undoubtedly indicated that there was no follicular development and consequently decreased activities of the stroma, the follicles, and the corpus luteum caused by nonavailability of either gonadotrophic hormones or the steroidal hormones or both (Shivalingappa et al., 2002). Concerning the effect of CRS on the pituitary–ovarian axis of the adult female albino rats, the present investigation showed a significant reduction in serum FSH, LH, and estradiol concentrations.

Moreover, ovarian histological changes were detected in the stressed rats as evidenced with the hyperchromatic nucleus, multiple follicular cysts, atretic follicles, and corpus fibrosum when compared with the distilled water treated rats. The treatment with the AEGD displayed protective effects on the ovaries that showed a normal stroma, developing follicles and matured graafian follicle. The chronic immobilization stress also caused a significant decrease in uterine weight and this was caused by the non-availability of hormones required for the development of the uterus (Fotsing et al., 2016). AEGD treated groups showed prevention in the loss of weight of uterus which may be due to uterotrophic effect of the plant. The significant increased weight of adrenal glands in stress rats is related to the active involvement of the HPA and sympathetic stimulation, which is fast to respond to stress.

## CONCLUSION

In this work, the effects of AEGD on the chronic immobilization stress-induced behavioral, neurochemical, and reproductive changes in female albino rats were assessed. The findings revealed that the AEGD significantly increased the number of entries and the time spent in the open arm of the EPM. The chronic immobilization stress-induced increased corticosterone, progesterone, and prolactin concentrations were antagonized by G. dalenii. Moreover, the decreases in reproductive hormones as well as the changes in estrous cycle duration caused by the chronic immobilization stress were normalized in the G. dalenii treated rats. AEGD displayed adaptogenic potential against chronic restraint model on experimental animals. Further studies may be carried out to identify and characterize the active principles and their mechanism of action.

## REFERENCES


## ETHICS STATEMENT

This study was carried out in accordance with the recommendations of the Cameroon National Ethical Committee (Ref No. FW-IRB00001954). The protocol was approved by the Comité d'Ethique Institutionnel de la Recherche pour la Santé Humaine with an authorization number CEI-UDo/909/01/2017/T.

## AUTHOR CONTRIBUTIONS

DF, ENB, SN, and YK made substantial contributions to the conception or design of the work. DF, ENB, SN, and YK contributed to the acquisition and analysis of the data. DF, ENB, SN, GN, and AO interpreted data for the work. DF, ENB, GN, and AO drafted the work. DF, ENB, YK, and GN critically revised for important intellectual content. DF, ENB, YK, and GN approved the final version to be published. DF, ENB, SN, YK, and AO agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy. DF, ENB, YK, and GN agreed to be accountable for integrity of any part of the work are appropriately investigated and resolved.

## ACKNOWLEDGMENTS

We thank Dr. Nantia Akono Edouard for his technical assistance during the biochemical analysis, Mr. Gangue Tiburce for his assistance during the statistical analysis, and Dr. Ebanga Echi Eyong Joan for her assistance with the language editing.



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Fotsing, Ngoupaye, Ouafo, Njapdounke, Kenneth and Ngo Bum. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Antiepileptogenic and Neuroprotective Effects of Pergularia daemia on Pilocarpine Model of Epilepsy

Antoine K. Kandeda1,2, Germain S. Taiwe<sup>3</sup> , Fleur C. O. Moto<sup>4</sup> , Gwladys T. Ngoupaye<sup>5</sup> , Gisele C. N. Nkantchoua<sup>2</sup> , Jacqueline S. K. Njapdounke<sup>2</sup> , Jean P. O. Omam2,4 , Simon Pale2,3, Nadege Kouemou2,3 and Elisabeth Ngo Bum2,6 \*

<sup>1</sup> Department of Animal Biology and Physiology, Faculty of Science, University of Yaoundé I, Yaoundé, Cameroon, <sup>2</sup> Department of Biological Sciences, Faculty of Science, University of Ngaoundere, Ngaoundere, Cameroon, <sup>3</sup> Department of Zoology and Animal Physiology, Faculty of Science, University of Buea, Buea, Cameroon, <sup>4</sup> Department of Biological Sciences, Higher Teachers' Training College, University of Yaounde I, Yaounde, Cameroon, <sup>5</sup> Department of Animal Biology, Faculty of Science, University of Dschang, Dschang, Cameroon, <sup>6</sup> Institute of Mining and Petroleum Industries, University of Maroua, Kaele, Cameroon

#### Edited by:

Nouria Lakhdar-Ghazal, Mohammed V University at Agdal, Morocco

#### Reviewed by:

Marina Bentivoglio, University of Verona, Italy Astrid Nehlig, Institut National de la Santé et de la Recherche Médicale (INSERM), France

> \*Correspondence: Elisabeth Ngo Bum eli\_bum@yahoo.fr

#### Specialty section:

This article was submitted to Ethnopharmacology, a section of the journal Frontiers in Pharmacology

Received: 31 March 2017 Accepted: 19 June 2017 Published: 30 June 2017

#### Citation:

Kandeda AK, Taiwe GS, Moto FCO, Ngoupaye GT, Nkantchoua GCN, Njapdounke JSK, Omam JPO, Pale S, Kouemou N and Ngo Bum E (2017) Antiepileptogenic and Neuroprotective Effects of Pergularia daemia on Pilocarpine Model of Epilepsy. Front. Pharmacol. 8:440. doi: 10.3389/fphar.2017.00440 In this study, we investigated antiepileptogenic and neuroprotective effects of the aqueous extract of Pergularia daemia roots (PDR) using in vivo and in vitro experimental models. In in vivo studies, status epilepticus caused by pilocarpine injection triggers epileptogenesis which evolves during about 1–2 weeks. After 2 h of status epilepticus, mice were treated during the epileptogenesis period for 7 days with sodium valproate and vitamin C (standards which demonstrated to alter epileptogenesis), or Pergularia daemia. The animals were then, 1 week after status epilepticus, challenged with acute pentylenetetrazole (PTZ) administration to test behaviorally the susceptibility to a convulsant agent of animals treated or not with the plan extract. Memory was assessed after PTZ administration in the elevated plus maze and T-maze paradigms at 24 and 48 h. Antioxidant and acetylcholinesterase activities were determined in the hippocampus after sacrifice, in vitro studies were conducted using embryonic rat primary cortical cultures exposed to L-glutamate. Cell survival rate was measured and apoptotic and necrotic cell death determined. The results showed that chronic oral administration of PDR significantly and dose-dependently increased the latency to myoclonic jerks, clonic seizures and generalized tonic–clonic seizures, and the seizure score. In addition, PDR at all doses (from 4.9 to 49 mg/kg) significantly decreased the initial and retention transfer latencies in the elevated plus maze. Interestingly PDR at the same doses significantly increased the time spent and the number of entries in T-maze novel arm. PDR significantly increased the activities of acetylcholinesterase and antioxidant enzymes superoxide dismutase, catalase, and total glutathione and proteins, and decreased malondialdehyde level. Furthermore, PDR increased viability rate of primary cortical neurons after L-glutamate-induced excitotoxicity, in a dose dependent manner. Altogether these results suggest that PDR has antiepileptogenic and neuroprotective effects, which could be mediated by antioxidant and antiapoptotic activities.

Keywords: antiepileptogenic, antioxidant, neuroprotective, Pergularia daemia, pilocarpine, status epilepticus

## INTRODUCTION

fphar-08-00440 June 30, 2017 Time: 16:41 # 2

Temporal lobe epilepsy (TLE) is a neurological disease that may originate from early precipitating events such as febrile seizures, head trauma, status epilepticus (SE), and infections (Loscher and Brandt, 2010; Kan et al., 2012). The activation of glutamate ionotropic receptors triggers neuronal injury or death predominantly mediated by excessive influx of calcium into neurons through ion channels (Emerit et al., 2004). Therefore, the latter events would be followed by a progressive latent phase of epileptogenesis, leading eventually to spontaneous recurrent seizures and which could also cause cognitive impairment (Marcangelo and Ovsiew, 2007). It was proposed that during epileptogenesis reactive oxygen species overproduction may cause an overwhelming intrinsic antioxidant scavenging capacity, resulting in the development of oxidative stress (Azam et al., 2010), as well as associated tissue injury and apoptotic processes (Todorova et al., 2004; Noor et al., 2015). Despite the high number of antiepileptic drugs currently available, pharmacological agents able to prevent epileptogenesis are lacking. In addition, a high percentage of TLE patients (40%) do not respond to conventional antiepileptic drugs (Kwan and Brodie, 2003; Loscher and Brandt, 2010). Thus, new antiepileptic drugs, possibly with antiepileptogenic properties, are needed. Medicinal plants represent a potential source of such compounds.

According to World Health Organization, about threequarters of the world population rely upon traditional remedies, mainly medicinal plants (Gilani and Rahman, 2005; Rahmati et al., 2013). Among these, the African and Asian tropical/subtropical plant Pergularia daemia (Forsk.) Chiov. (Asclepiadaceae) (P. daemia) is used in African and Indian traditional medicine to treat leprosy, poisoning, asthma, anemia, seizures, and mental disorders (Mittal et al., 1962; Karthishwaran and Mirunalini, 2010; Sravani et al., 2012; Sridevi et al., 2014). In Northern Cameroon and in Benin, traditional healers use decoctions of P. daemia roots to treat malaria, febrile seizures, epilepsy, mental, and inflammatory disorders (Arbonnier, 2002).

Phytochemically, alkaloids, flavonoids, saponins, triterpenes, tannins and steroidal compounds have been searched in P. daemia roots (Bhaskar and Balakrishnan, 2009; Sridevi et al., 2014). Phytochemicals like glucosides and cardenolides in seed, coroglaucigenin, corotoxigenin, uscharidin, and uzarigenin in stem have been identified (Bhaskar and Balakrishnan, 2009; Sridevi et al., 2014). Roots of P. daemia were reported to contain β-sitosterol, lupeol, lupeol acetate, and β-amyrin and its acetate (Bhaskar and Balakrishnan, 2009; Sridevi et al., 2014). Organic esters, fatty acids, and phenolic compounds were identified by analysis of the ethanolic extract of the plant (Bhaskar and Balakrishnan, 2009; Sridevi et al., 2014). Various pharmacological properties, including hepatoprotective, antidiabetic, anti-inflammatory, antioxidant, antipyretic, analgesic, and sedative activities have been reported in whole plant extracts (Wahi et al., 2002; Suresh and Mishra, 2008). Aqueous, ethanolic and petroleum ether extracts of P. daemia leaves exhibited significant analgesic, antioxidant, antipyretic activities, and antibacterial properties (Suresh and Mishra, 2008). Moreover, active compounds like kaempferol extracted from the roots demonstrated antiepileptic activities (Lokesh, 2009; Sravani et al., 2012).

In the present study, to assess antiepileptogenic effect of P. daemia extract, the pilocarpine-induced SE model was used. In this paradigm, SE was induced in mice by intraperitoneal pilocarpine injection. Animals that developed SE for 2 h were selected and received P. daemia extract for 7 days (i.e., during the epileptogenesis period). The effects of P. daemia were compared to those of sodium valproate (a widely used antiepileptic drug) and vitamin C (an exogenous antioxidant known to inhibit oxidative stress in the brain). These drugs are known to modify epileptogenesis process (Brandt et al., 2003; Xavier et al., 2007; Loscher and Brandt, 2010). During the epileptogenesis period, animals were challenged with pentylenetetrazole (PTZ) in order to assess the susceptibility of animals to seizures and behavioral alterations (Ilhan et al., 2005; Blanco et al., 2009). Effects of the extract on excitotoxicity induced by L-glutamate were assessed on primary cortical neurons in culture. To date, no published study assessed antiepileptogenic and neuroprotective properties of P. daemia extract. Therefore, the aim of this study was to assess the putative antiepileptogenic and neuroprotective effects of the aqueous extract of P. daemia roots prepared mimicking the traditional healer decoction.

## MATERIALS AND METHODS

## Drugs and Chemicals In Vivo Studies

Vitamin C, PTZ, scopolamine methyl nitrate, diethyl ether, pilocarpine hydrochloride, sodium valproate, Biuret reagent, acetylcholine iodide, 5<sup>0</sup> 5-dithiobis-(2-nitrobenzoic acid) (DNTB), adrenaline, acetic acid, dichromate, hydrogen peroxide (H2O2), Tris-Hcl, trichloroacetic acid, thiobarbituric acid, sodium phosphate buffer, Griess reagent were purchased from Sigma Chemical Co., St. Louis (United States), while diazepam was purchased from Roche, Neuilly sur-Seine, France. The minimal dose of chemoconvulsant at which 99% of the animals showed a convulsion was determined based on the doses used by other researchers and by a dose-percentage effect curve (Miller and Tainter, 1944; Ahmadiani et al., 2003). Vitamin C and sodium valproate were dissolved in distilled water. All solutions were prepared freshly in the day of the experiment and were administered intraperitoneally at a volume of 10 ml/kg, except for distilled water and aqueous extract of P. daemia administered per os at the same volume.

## In Vitro Studies

β-D-arabinofuranoside hydrochloride, Hoechst 33342, propidium iodide, 3-(4,5-dimethylthiazol-2-yl)-2,5 diphenyltetrazolium bromide (MTT), oxamate, dimethyl sulfoxide (DMSO), sodium bicarbonate, phosphate buffer saline (PBS), nicotinamide dinucleotide adenine (NAD), diaphorase, L-glutamic acid monosodium salt hydrate, poly-L-lysine, Dulbecco's modified Eagle's medium (DMEM), β-mercaptoethanol, lactate, L-glutamine were purchased from Sigma–Aldrich (St. Louis, MO, United States). Fetal bovine serum (FBS) and bovine serum albumin (BSA) were purchased from Gibo/Invitrogen (Carlsbad, CA, United States). Penicillin and streptomycin were purchased from Sanofi-Aventis (Guildford, United Kingdom).

## Plant

## Collection and Identification

fphar-08-00440 June 30, 2017 Time: 16:41 # 3

Fresh roots of P. daemia were harvested during the month of June 2012 in Mayo-Tsanaga division (Far-North Region, Cameroon). A voucher specimen has been deposited at the Yaoundé national herbarium on the number 7797/SRF/Cam.

## Preparation of the Aqueous Extract of P. daemia

The extract was prepared the day of the experiment, mimicking strictly the procedures used by the traditional healers. The roots were peeled-off, cut to pieces, and air dried at room temperature. Then, dried root samples were grounded into coarse powder. The powder was added to distilled water (5 g in 75 ml) and boiled for 20 min. Following cooling at room temperature, the solution obtained was filtered with Whatman N<sup>o</sup> 1 filter paper. The filtrate was considered as the stock solution. The amount of dry matter in the extract was determined by evaporating water in a drying oven (50◦C). A solid residue (0.37 g) was obtained. The yield of extraction was 7.34%, and the stock solution dose was 49 mg/kg. The other doses used in the study (24.5, 12.3, and 4.9 mg/kg) were obtained by dissolving the stock solution in distilled water at ratios of 1/2, 1/4, and 1/10, respectively.

## Animals

Ninety male or female Swiss albino mice weighting 18–29 g (37–48 days old) were used. They were obtained from Cameroon National Veterinary Laboratory (Lanavet, Garoua, Cameroon) and were housed and bred in the animal facility of the University of Ngaoundere (Ngaoundere, Cameroon). They were kept in a controlled environment, with ad libitum access to food and tap water. Animals were maintained on a 12 h/12 h light/dark cycle (lights on at 7:00 a.m.). Animals were acclimated to laboratory conditions before starting the experiments. All procedures were performed in conformance with the Cameroon National Ethical Committee directives (Ref No. FW-IRB00001954, October 22, 1987 under an authorization number CEI-UDo/907/01/2017/T). The study was also performed conforming to international regulations minimizing the number of animals used and avoiding their suffering.

## In Vivo Studies

## Experimental Design

Mice were randomly divided into eight groups of seven animals each. One control group received only distilled water (DW + DW group) and in the other seven groups SE was induced by a single injection of pilocarpine hydrochloride (360 mg/kg, i.p., Sigma–Aldrich). Two hours after SE induction, the following groups were formed: (i) disease group receiving distilled water (10 ml/kg; DW + PILO group); to control the eventual effect of distilled water; (ii) two groups administered either with sodium valproate (300 mg/kg, Sigma–Aldrich) or vitamin C (250 mg/kg, Sigma–Aldrich); (iii) four test groups receiving the doses of P. daemia extract (4.9, 12.3, 24.5, and 49 mg/kg) orally, through an intragastric feeding tube. Treatments were administered daily for 7 days. Twenty-four hours after the last administration of the treatments, mice were challenged with PTZ (Blanco et al., 2009). Then, memory was assessed using the following behavioral paradigms: elevated plus-maze (48-h after treatment) and T-maze (72-h after) (Mehla et al., 2010; Taiwe et al., 2015). Animal behavior was recorded by two blinded experimenters. Afterward, mice were sacrificed by decapitation under deep anesthesia with diethyl ether (8%, v/v, Sigma– Aldrich). The brain was dissected out and processed for the quantification of markers of oxidative stress and cholinergic status determination.

## SE Induction, Behavioral Observations, and Tests **SE induction and seizure evaluation**

Animals were subjected to epileptogenesis induction by a single intraperitoneal injection of pilocarpine (Turski et al., 1983). The minimal dose of chemoconvulsant at which 99 % of the animals showed seizures was determined based on the available reports (Miller and Tainter, 1944; Turski et al., 1983). This was verified by dose-percentage effect curves obtained in our laboratory; the survival rate was 90%. To prevent peripheral muscarinic stimulation, scopolamine (Sigma–Aldrich) was injected subcutaneously at a dose of 1 mg/kg, 30 min before injection of pilocarpine (Liu et al., 2010). About 30 min after pilocarpine injection, animals became hypoactive and displayed oro-facial movements, salivation, eye blinking, twitching of vibrissae and yawning. Generalized seizure and limbic SE were observed 40–80 min after pilocarpine injection. Only mice that displayed 2-h of SE were selected in this study (Goffin et al., 2007; Curia et al., 2008). SE was stopped after 2-h with an injection of diazepam (10 mg/kg, Roche) in order to prevent mortality. SE initiated by pilocarpine injection triggers epileptogenesis which progress during about 1–2 weeks (Cavalheiro, 1995; Curia et al., 2008). Two hours after SE, animals were treated for 1 week with P. daemia extract, sodium valproate and vitamin C. During this period (i.e., the epileptogenesis period) mice were challenged with a convulsant. The challenge was characterized by the acute PTZ administration (Ilhan et al., 2005; Blanco et al., 2009). The challenge with PTZ was used to assess behaviorally the sensibility to a convulsant agent of animals treated or not with the plan extract (Ilhan et al., 2005; Blanco et al., 2009). Then, mice were placed in a 30 cm × 30 cm chambers for 30 min observation. A progressive evolution of seizure activity was evaluated using a six phase scale (Erakovic et al., 2001): (i) 0 indicated no response; (ii) 1, ear and facial twitching (iii) 2, convulsive waves axially through the body; (iv) 3, myoclonic body jerks; (v) 4 generalized clonic seizures turn over into side position; (vi) 5, generalized seizures with tonic extension episode and SE; and (vii) 6, death (Khalili et al., 2011). The latency and duration of the first myoclonic jerk, clonic seizure and generalized tonic–clonic seizure were measured. Latencies to generalized tonic–clonic seizure were used to calculate the seizure score as follows: S = 1 – (control latency/drug seizure latency) (Mehla et al., 2010).

### **Elevated plus-maze paradigm**

fphar-08-00440 June 30, 2017 Time: 16:41 # 4

Cognitive function was assessed using an elevated plus maze. The apparatus is made up of two open arms (16 cm × 5 cm) and two closed arms (16 cm × 5 cm × 10 cm) that extend from a common central platform (5 cm × 5 cm). The entire maze is elevated to a height of 50 cm above the floor level (Bum et al., 2011). Procedures were performed as previously described (Taiwe et al., 2015). Briefly, in the first task, each animal was placed at the end of the open arm and the initial transfer latency, i.e., the latency to closed arm entry was recorded. A 60 s cut-off was set. The mouse was then allowed to move freely in the maze for another 10 s (Taiwe et al., 2015). Similarly, 24-h later the latency to closed arm entry, termed retention transfer latency, was determined. Mice which did not enter the enclosed arm within 60 s on the second trial were assigned a score of 60 s (Taiwe et al., 2015).

### **T-maze paradigm**

The T-shaped maze is made of wood and consists of a start arm and two choice arms. Each arm is 30 cm × 10 cm × 20 cm (length × width × height) (Taiwe et al., 2015). A recessed black plastic cup (3 cm in diameter, 1 cm in depth) containing food was placed on the floor at the end of each choice arm (Taiwe et al., 2015). A day before the experiment, each animal was placed in the start position (at the end of the start arm) for a 10 min exploration trial, one arm open and the other one closed and at the end, they were returned to their home cage. The following day, animals were reintroduced in the T-maze for a 5 min testing period (Taiwe et al., 2015). During the retrial (the two choice arms were opened), animals were placed in a start arm and the number of visits and the time spent in the two arms were recorded (Taiwe et al., 2015).

#### Biochemical Tests

Immediately after the animals were sacrificed, the brain hemispheres were quickly dissected out and cleaned with icecold saline (0.9%, w/v) to remove the hippocampus. After weighing the hippocampi, they were stored at −43◦C. To perform biochemical analyses, 10% (w/v) homogenates prepared with ice-cold 0.1 M phosphate buffer (pH 7.4) were centrifuged (10,000 × g, 15 min). Aliquots of the supernatant were collected and used for biochemical estimation of reduced glutathione (GSH), protein, nitric oxide (NO), and malondialdehyde (MDA) levels. Superoxide dismutase (SOD) and catalase (CAT) activities were also determined from these tissues. Acetylcholinesterase (AchE) activity was assessed in hippocampi dissected from the right hemisphere (Mehla et al., 2010; Taiwe et al., 2015).

#### **Total proteins quantification**

The protein amount was estimated using the Biuret method (Gornall et al., 1949). The BSA (Carlsbad) was used as standard. Briefly, 3 ml of Biuret reagent (Sigma–Aldrich) and 10 µl of homogenate were added into test tubes. The contents were mixed by inversion and the absorbance was measured at 590 nm after 2 min against blank (3 ml of NaCl 0.9% mixed with 3 ml of Biuret reagent). The weight of protein was plotted against the corresponding absorbance resulting in a standard curve used to determine the protein in unknown samples. The concentration of protein was expressed in mg/ml of protein in the tissue.

## **AchE activity**

The AchE activity was assessed by the Ellman method (Ellman et al., 1961). The assay mixture contained 0.05 ml of supernatant, 3 ml of sodium phosphate buffer (pH 8, Sigma–Aldrich), 0.1 ml of acetylthiocholine iodide (Sigma–Aldrich) and 0.1 ml of DNTB (Ellman reagent, Sigma–Aldrich). The change in absorbance was measured at 412 nm for 2 min, at 30 s intervals. Results were expressed in U/min/mg of protein in the tissue (1 U/min/mg of AchE was defined as the amount of enzyme that hydrolyzed 1 µmol of acetylthiocholine iodide).

### **SOD activity**

The SOD activity in the tissues was determined by the method of Misra and Fridovich (1972), where the autoxidation of adrenaline (Sigma–Aldrich) is followed in terms of the production of adrenochrome (maximum absorption at 480 nm). Tissue homogenates (134 µl) were introduced in a test tube and 1666 µl of phosphate buffer (0.05 M, pH 10.2) in a blank tube to equilibrate the spectrophotometer. The reaction was started by adding 200 µl of freshly prepared adrenaline (0.3 mM). Then, the mixture was quickly mixed. The increase in absorbance at 480 nm was recorded at 20 and 80 s against the blank. One unit (U) of SOD was defined as the quantity of SOD required to inhibit 50% of the oxidation of adrenaline in adrenochrome for 1 min. The activity of SOD was expressed in U/min/mg of protein in the tissue.

## **CAT activity**

The CAT activity was assayed following the method of Sinha (1972). In this method, dichromate (Sigma–Aldrich) in acetic acid (Sigma–Aldrich) is reduced to chromic acetate when heated in the presence of H2O2. The blue perchromic acid, an unstable intermediate is then formed. The reaction mixture consisted of 187.5 µl phosphate buffer (0.1 M, pH 7.5) and 12.5 µl of homogenate. The reaction was started by adding 50 µl of H2O<sup>2</sup> (50 mM, Sigma–Aldrich). After 1 min, the reaction was stopped by the addition of 500 µl of dichromate acetic acid reagent. The tubes were immediately kept in a boiling water bath at 100◦C for 10 min, and the green color developed during the reaction was read at 570 nm on a spectrophotometer against the blank. Blank tube, devoid of enzyme, was also processed in parallel. The amount of H2O<sup>2</sup> remaining was determined using a standard curve. The enzyme activity was expressed in mmol of H2O<sup>2</sup> consumed/min/mg of protein in brain tissue. The specific activity of CAT was calculated as follows: CAT activity = [(A of sample – A of blank) × f/(a × t × mi)]. Where A is the absorbance, f the dilution factor, a standard curve coefficient, t the time in minute and mi the weight of tissue processed.

### **GSH level**

Glutathione was measured using the method of Ellman (1959). Briefly, 1500 µl of DNTB and 500 µl of Tris-HCl (Sigma– Aldrich) buffer (50 mM, pH 7.4) were added to a blank tube containing 100 µl of Tris-HCl buffer (50 mM, pH 7.4) or to test tubes containing tissue homogenates (100 µl). The mixture solution was incubated for 1 h, and the absorbance was read against the blank at 412 nm. The GSH concentration was calculated using an extinction coefficient of 13600 mol−<sup>1</sup> cm−<sup>1</sup> .

The concentration of GSH was expressed as µmol/g of protein in the tissue.

## **MDA level**

The method of Wilbur et al. (1949) for MDA determination was used. Briefly, distilled water (250 µl) and homogenate (20 µl) were introduced in the control tube and in the test tubes, respectively. Then, 250 µl of Tris-HCl buffer (50 mM, pH 7.4), 500 µl of trichloroacetic acid (20%, Sigma–Aldrich) and 1000 µl of thiobarbituric acid (0.67%, Sigma–Aldrich) were added. The mixture solution was heated in a water-bath (90◦C, 10 min). After cooling at room temperature, the tubes were centrifuged (3000 rpm, 15 min). The absorbance of the pinkcolored supernatant was measured against the blank at 530 nm. The MDA concentration was calculated using an extinction coefficient of 1.56 × 10<sup>5</sup> mmol−<sup>1</sup> cm−<sup>1</sup> . MDA level was expressed in µmol/g of protein in the tissue.

## **NO level**

Nitric oxide content was assayed by the Griess method (Grand et al., 2001). NO is a compound with a short half-life that is rapidly converted to the stable end products nitrate (NO<sup>3</sup> −) and nitrite (NO<sup>2</sup> <sup>−</sup>). In this assay, the conversion of nitrate into nitrite is accompanied by color development in the presence of [0.1% N-(1-naphthyl) ethylenediamine dihydrochloride, 1% sulfanilamide and 2.5% phosphoric acid) in acidic medium] (Griess reagent, Sigma–Aldrich) (Grand et al., 2001). To estimate the quantity of NO, 200 µl of homogenate and 200 µl of Griess reagent were introduced in test tubes. The solution was mixed and the absorbance was read at 570 nm after 10 min. A standard curve NaNO<sup>2</sup> was established with a set of serial dilutions of nitrite. Linear regression was done by using the peak area from nitrite standard. The resulting equation was used to calculate the unknown sample concentrations. Results were expressed in mmol/g of protein in the tissue.

## In Vitro Studies

Primary cortical neuron cultures were prepared from the cerebral cortex of Wistar rat embryo of 17 days as described previously (Kim et al., 1998). Briefly, pregnant Wistar rats were anesthetized with sodium pentobarbital (30 mg/kg, i.p., Sigma–Aldrich) and sacrificed by cervical dislocation. The cerebral cortex of fetal rats was rapidly removed bilaterally and collected. Tissues were then gently minced using a sterile razor blade and digested in PBS (0.1 M, pH 7.4, Sigma–Aldrich) for 15 min. A Pasteur pipette was used for dissociation of cells (approximately 5–10 times). After centrifugation (200 × g for 3 min), cells were re-suspended in DMEM (Sigma–Aldrich) supplemented with FBS (15%, Carlsbad), L-glutamine (2 mM, Sigma–Aldrich), sodium bicarbonate (4.2 mM, Sigma–Aldrich), BSA (0.3 g/l, Sigma–Aldrich), β-mercaptoethanol (0.1 mM, Sigma–Aldrich), penicillin (1%, Sanofi Aventis), streptomycin (50 µg/ml, Sanofi Aventis) and grown on 0.1% poly-L-Lysine (Sigma–Aldrich) coated plates. Cultures were incubated at 37◦C in a humidified 5% CO<sup>2</sup> atmosphere. To prevent proliferation of non-neuronal cells, cytosine β-D-arabinofuranoside hydrochloride (10 µM, Sigma– Aldrich) was added 3 days after plating. In all experiments, 11 days mature cells were used.

## Cell Viability Assay by MTT

3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide test is based on the ability of viable cells to metabolize a tetrazolium salt to formazan blue in the mitochondria (Loveland et al., 1992). The formazan accumulation is proportional to the number of viable cells and inversely proportional to the degree of cytotoxicity (Berridge et al., 2005). Briefly, sample supernatants are inoculated in 96 wells. Cortical cell cultures were treated and incubated with 10 µl of aqueous extract of P. daemia (5, 10, 19, 40, 77, 153, 306, 615, 1225, 2450 µg/ml) for 1 h. Cultures were then exposed to L-glutamate (10 mM) and maintained for 24 h (Kim et al., 1998). After the incubation, culture medium was removed before adding 100 µl of solution of MTT (1 mg/ml, Sigma–Aldrich). The plates were incubated during 1 h at 37◦C. Excess MTT was removed and 100 µl of DMSO (0.1%, Sigma–Aldrich) were added to each well to dissolve formazan crystals (precipitates resulting from the conversion of MTT by the mitochondrial succinate dehydrogenase). The plates were vortexed for 5 min and read at 540 nm with a microplate reader. The percentage of cell viability was expressed according to the following formula: percentage of cell viability = 100 × [(optical density (OD) of L-glutamate + extract treated cultures) − (OD of L-glutamate treated cultures)/(OD of control cultures − OD of L-glutamate treated cultures)] (Koo et al., 2006).

## Cell Viability Assay by Lactate Dehydrogenase (LDH)

Lactate dehydrogenase (LDH) is a cytosolic enzyme present in many different cell types. Plasma membrane damage releases LDH into the cell culture media (Decker and Lohmann-Matthes, 1988). Extracellular LDH in the media can be quantified by a coupled enzymatic reaction in which LDH catalyzes the conversion of lactate to pyruvate via NAD<sup>+</sup> reduction to NADH (Decker and Lohmann-Matthes, 1988). Therefore, NADH is used to reduce a tetrazolium salt to a red formazan product that can be measured at 490 nm (Nachlas et al., 1960). For the assay 20 µl of lactate solution (36 mg/ml of 10 mM Tris buffer, pH 8.5, Sigma–Aldrich) were added to the samples in microliter wells, followed by 20 µl solution of MTT (2 mg/ml of PBS (0.1 M, pH 7.4) prepared from a 10-fold concentrated stock solution in DMSO). The enzymatic reaction was then started by addition of 20 µl of a solution containing NAD<sup>+</sup> (3 mg/ml, Sigma–Aldrich) and diaphorase (13.5 U/ml; BSA: 0.03%; sucrose: 1.2%; in PBS, Sigma–Aldrich) and allowed to proceed for 20 min (Decker and Lohmann-Matthes, 1988). The reaction was terminated by the addition of 20 µl of the LDH inhibitor oxamate (16.6 mg/ml of PBS, Sigma– Aldrich). Measurements were performed at 490 nm with a microplate reader. Percentage of cell viability was evaluated as above.

## Quantification of Apoptosis and Necrosis by Hoechst 33342 and Propidium Iodide Staining

The experiment was realized according to the method described by Syed et al. (2013). Briefly, cells were grown in tissue culture

dishes and treated with or without the aqueous extract of P. daemia at concentrations (5, 10, 19, 40, 77, 153, 306, 615, 1225, 2450 µg/ml). After 24 h of incubation in an incubator (37◦C in 5% CO2), the cells were harvested and washed with cold PBS (0.1 M, pH 7.4). The cells were suspended in Hoechst 33342 solution (10 µg/ml, Sigma–Aldrich) and were incubated (37◦C in 5% CO2) for 7 min (Syed et al., 2013). After incubation with Hoechst 33342, the cells were stained with propidium iodide (2.5 µg/ml, Sigma–Aldrich). The samples were maintained in the dark for 15 min. After staining, an aliquot of cell suspension was placed on a glass microscope slide. The slides were observed immediately under a fluorescence microscope and the fluorescence was measured at 630× magnification. Cells were counted and the numbers of each of the four cellular states were recorded and analyzed using fluorescence microscopy for quantification of apoptosis and necrosis (Moongkarndi et al., 2004; Syed et al., 2013). The experiment was conducted in triplicates. Hoechst 33342 was used to determine apoptotic nuclear morphology, while propidium iodide indicated dead cells by necrosis. Cells with fragmented or condensed nuclei were considered as apoptotic cells. After the exclusion of the positive apoptotic cells from Hoechst 33342, the propidium iodide positive cells were considered necrotic cells (Moongkarndi et al., 2004). The numbers of apoptotic or necrotic cells in the treatment groups were compared to the control. The percentages of apoptotic and necrotic cells were determined according to the following formula:


## Statistical Analysis

Inter-group differences were assessed using one-way analysis of variance (ANOVA), followed by Newman Keul's multiple comparisons post hoc test. The significance level was set at p < 0.05, with Mann–Whitney U test correction. Analyses were performed using Graph Pad Prism version 5.1 for Windows (Graph Pad Software, San Diego, CA, United States) and XLSTAT, 2007. Data were expressed as mean ± standard error of the mean (SEM) for in vivo tests and as percentage for in vitro tests.

## RESULTS

## Effects of P. daemia on Seizures Induced by PTZ Challenge Latency to Seizures

The mice treated with PTZ resulted in a classical pattern of limbic motor seizures culminating into generalized tonic–clonic seizures. A decreased myoclonic jerks latency was observed in DW + PILO group compared to DW + DW group (p < 0.05) (**Figure 1A**). P. daemia caused a two-fold increase (p < 0.05) in the latency to myoclonic jerks compared to DW + PILO group (37.87 ± 1.33 s in DW + PILO group against 77.8 ± 2.18 s in the group administered with P. daemia dose 24.5 mg/kg) [F(7,49) = 23.00, p < 0.0001] (**Figure 1A**). Sodium valproate induced an increase of this latency which did not reach statistical significance (**Figure 1A**).

Latency to clonic seizures was decreased in DW + PILO group compared to DW + DW group (p < 0.05) (**Figure 1A**). P. daemia increased the latency to clonic seizures compared to DW + PILO group [F(7,49) = 121.25, p < 0.001] (**Figure 1A**). The latency to clonic seizure increased and reached 165.97 ± 2.38 s in the group administered with P. daemia dose 24.5 mg/kg (p < 0.01) (**Figure 1A**). This effect was stronger than sodium valproate effect (159.92 ± 2.42 s, p < 0.05) (**Figure 1A**).

Similarly, a decreased generalized tonic–clonic seizure latency was observed in DW + PILO group compared to DW + DW group (p < 0.01) (**Figure 1A**). P. daemia (24.5 mg/kg) increased significantly and the latency to generalized tonic–clonic seizures [F(7,49) = 312.14, p < 0.001], compared to DW + PILO group in a dose dependent manner up to 205.08 ± 1.25 s (p < 0.01) in the group administered with P. daemia dose 24.5 mg/kg. This effect was stronger than sodium valproate (162.10 ± 2.97 s, p > 0.05) and vitamin C (173.12 ± 2.89 s, p < 0.05) effect (**Figure 1A**).

## Seizure Duration

Significant inter-group differences were observed in the duration of myoclonic jerks [F(7,49) = 1.33, p < 0.01], clonic seizures [F(7,49) = 2.45, p < 0.001] and generalized tonic–clonic seizures [F(7,49) = 7.66, p < 0.0001]. P. daemia decreased the duration of myoclonic jerks from 8.44 ± 1.03 s in DW + PILO group to 4.00 ± 0.48 s (p < 0.05) in the group administered with P. daemia dose 24.5 mg/kg (**Figure 1B**). The duration was slightly decreased from 8.44 ± 1.03 s in DW + PILO group to 4.98 ± 0.73 s (p > 0.05), 4.94 ± 0.80 s (p > 0.05) in the groups administered with sodium valproate and vitamin C, respectively (**Figure 1B**). Generalized tonic–clonic seizures duration was decreased from 12.30 ± 1.26 s in DW + PILO group to 9.09 ± 1.30 s (p < 0.05) and 9.14 ± 1.36 s (p < 0.05) in the groups administered with P. daemia doses 12.3 and 49 mg/kg, respectively (**Figure 1B**).

#### Seizure Score

A significant reduction in seizures score was observed in DW + PILO group compared to DW + DW group (p < 0.01) (**Figure 1C**). Overall, seizures score between the groups was also

significantly different [F(7,49) = 73.11, p < 0.001]. P. daemia increased the seizures score from 0 in DW + PILO group to 0.43 ± 0.02 (p < 0.01) and 0.43 ± 0.02 (p < 0.01), respectively, in groups treated with P. daemia doses 24.5 and 49 mg/kg (**Figure 1C**). The seizures score was also increased in groups treated with sodium valproate (0.28 ± 0.03, p < 0.05) and vitamin C (0.32 ± 0.03, p < 0.05) (**Figure 1C**).

## Effects of P. daemia on Memory Elevated Plus Maze

Significant inter-group differences were observed in the initial transfer latency in the elevated plus maze [F(7,49) = 6.04, p < 0.001]. P. daemia decreased initial transfer latency up to 26.35 ± 1.56 s at dose 24.5 mg/kg, against 39.84 ± 2.60 s in DW + PILO group (p < 0.05) (**Figure 2**). Intergroup differences were observed in the retention transfer latency [F(7,49) = 5.99, p < 0.01]. P. daemia induced dose-dependent decrease from 36.03 ± 2.13 s in DW + PILO group to 18.84 ± 2.05 s (p < 0.01) in the group administered with P. daemia dose 49 mg/kg (**Figure 2**). However, sodium valproate and vitamin C did not induce significant decreases in initial and retention transfer latencies (**Figure 2**).

#### T-maze

No inter-group difference was observed in the number of entries in the start arm (**Figure 3B**). P. daemia decreased the time spent in the familiar arm (14.92 ± 2.30 s at dose 24.5 mg/kg vs. 22.90 ± 1.44 s in DW + PILO group, p < 0.05) (**Figure 3A**). As sodium valproate and vitamin C, P. daemia increased the

time spent in the novel arm up to 23.16 ± 2.16 s at dose 24.5 mg/kg (against 9.08 ± 2.24 s in DW + PILO group, p < 0.01) (**Figure 3A**).

Similar, no inter-group difference was observed in the number of entries in the start arm (**Figure 3B**). The number of entries in the familiar arm was decreased up to 13.14 ± 1.35 in the group treated with P. daemia dose 24.5 mg/kg (against 22.71 ± 1.48 in DW + PILO group, p < 0.05) (**Figure 3B**). Conversely, the number of entries in novel arm was increased up to 23.57 ± 0.72 (p < 0.01) and 23.14 ± 1.22 (p < 0.01) in groups treated with P. daemia doses 24.5 and 49 mg/kg, respectively (against 11.71 ± 2.53 in DW + PILO group) (**Figure 3B**). Although in a lesser extent, sodium valproate and vitamin C also increased the number of entries in the novel arm (**Figure 3B**).

## Levels of Total Proteins, AchE, Antioxidant Enzymes, and Oxidative Stress Markers

#### Total Level of Protein

Significant inter-group differences were observed in hippocampus total proteins level [F(7,49) = 130.20, p < 0.001]. Pilocarpine significantly decreased the protein level up to 2.29 ± 0.00 mg/ml wet tissue in DW + PILO group, against 8.32 ± 0.00 mg/ml in the DW + DW group (p < 0.01) (**Figure 4A**). Treatment with P. daemia prevented such decrease in a dose-dependent manner. At dose 49 mg/kg, the extract resulted in protein level comparable to DW + DW group (7.45 ± 0.00, p < 0.01 vs. DW + PILO group) (**Figure 4A**). The well-established antioxidant vitamin C also prevented pilocarpine-induced protein decrease (6.60 ± 0.00, p < 0.05 vs. DW + PILO group) (**Figure 4A**).

## AchE Activity

Significant inter-group differences were observed in AchE activity [F(7,49) = 244.76, p < 0.001]. Pilocarpine decreased AchE activity compared to DW + DW group (1.59 ± 0.00 U/min/mg in DW + PILO group vs. 12.01 ± 0.00 U/min/mg, p < 0.001) (**Figure 4B**). P. daemia dose 49 mg/kg prevented the decrease in AchE activity caused by pilocarpine (7.10 ± 0.00 U/min/mg, p < 0.01 vs. DW + PILO group) (**Figure 4B**).

## Antioxidant Enzymes

The effects of P. daemia extract on activities of the antioxidant enzymes tested (CAT and SOD) in hippocampi of pilocarpineinjected mice is shown in **Table 1**. P. daemia treatment increased SOD (p < 0.05) and CAT (p < 0.01) activities compared to DW + PILO group. Vitamin C also increased CAT activity. Although to a lesser extent, sodium valproate also displayed some of these effects, particularly the marked decrease in CAT activity induced by pilocarpine (**Table 1**).

### Oxidative Stress Markers Level

The effects of P. daemia extract on levels of oxidative stress markers tested (GSH, MDA, and NO) in hippocampi of pilocarpine-injected mice is shown in **Table 1**. P. daemia treatment induced moderate increase (p < 0.05) in the GSH level compared to DW + PILO group. On the other hand, the extract decreased significantly the MDA level (p < 0.05) and, surprisingly, increased the estimated NO level. Although to a lesser extent, sodium valproate also displayed a non-significant decrease in GSH level and MDA level induced by pilocarpine (**Table 1**).

TABLE 1 | Effects of P. daemia on antioxidant enzymes and oxidative stress markers in hippocampi of pilocarpine-injected mice.


Data are mean ± SEM, N = 7 per group. Newman Keul's multiple comparisons post hoc test, with Mann–Whitney U test correction: (i) vs. control animals (DW + DW group) receiving only distilled water: <sup>∗</sup>p < 0.05, ∗∗p < 0.01, (ii) vs. disease control animals (DW + PILO group) receiving distilled water and pilocarpine: 0.05 <sup>a</sup>p < 0.05, <sup>b</sup>p < 0.01. DW, distilled water; PD, Pergularia daemia; PILO, pilocarpine; VIC, vitamin C; SVA, sodium valproate; MDA, malondialdehyde; GSH, reduced glutathione; SOD, superoxide dismutase; CAT, catalase; NO, nitric oxide.

## In Vitro Neuroprotective Effects of P. daemia

### Protective Effect of P. daemia Extract against L-Glutamate-Induced Neurotoxicity

In the MTT test, stimulation with L-glutamate alone resulted in a decrease in cell viability up to approximately 0.19% compared to control. Nevertheless, the different doses of the extract exhibited a significant decrease of L-glutamate-induced toxicity in a dose dependent manner. The highest concentration of the extract exhibited protective effect (85.92% vs. control) (**Figure 5A**). The protective effect of P. daemia was also revealed by LDH release assay. As shown in **Figure 5A**, cell viability decreased up to approximately 0.39% after exposure to L-glutamate. However, treatment with P. daemia resulted in a significant increase of this viability up to 73.01% at the highest concentration.

## Protective Effect of P. daemia against L-Glutamate-Induced Apoptosis and Necrosis

Results of Hoechst staining in control culture indicated that, after exposure to L-glutamate, cortical neurons exhibited high levels of condensed chromatin and apoptotic bodies, indicating an increase of apoptotic cells up to 93.00% (**Figure 5B**). Treatment with P. daemia resulted in a significant decrease of these apoptotic features up to 19.67% at the highest concentration (**Figure 5B**). Results of propidium iodine staining in control culture indicated that, after exposure to L-glutamate, cortical neurons culture exhibited high levels of degenerated neurons, indicating an increase of necrotic cells up to 89.67% (**Figure 5B**). P. daemia was not able to protect neurons against L-glutamate induced cell necrosis (**Figure 5B**).

## DISCUSSION

The aim of this study was to evaluate the antiepileptogenic and neuroprotective effects of the decoction of P. daemia roots. In vivo and in vitro experimental models were used. As the results show, the acute administration of PTZ in mice treated with distilled water (DW + PILO group) for 1 week (epileptogenesis period) after SE, induced an increase in the latency of seizures, and a decrease in the duration and score of seizures (Blanco et al., 2009). Remarkably, our results demonstrate that PTZ produces different effects when injected in epileptogenic and non-epileptogenic mice. This is a significant demonstration that the pharmacologic response outline of acute seizures contrasts from that of chronic seizures paradigms (Loscher et al., 1991; Blanco et al., 2009). In the present study, P. daemia reduced the severity of seizures induced by PTZ challenge on epileptogenic process. P. daemia extract also increased the seizure score dose dependently. Such reduction in seizure severity and in seizure susceptibility to a convulsant during epileptogenesis process suggests that the decoction antagonized or altered the epileptogenic process induced by pilocarpine (Pitkanen et al., 2005; Mehla et al., 2010; Pitkanen, 2010). Indeed, the PTZ (GABA<sup>A</sup> receptor complex antagonist) is known to increase the seizure threshold and therefore to induce more severe seizures in epileptogenic brain compared to non-epileptogenic brain (Ilhan et al., 2005; Blanco et al., 2009; Mehla et al., 2010; Taiwe et al., 2015). Thus, the challenge is used to test behaviorally the susceptibility to a convulsant agent with or without treatment with the plant extract (Ilhan et al., 2005; Blanco et al., 2009). The findings of the present study suggest therefore that the aqueous extract of P. daemia has antiepileptogenic effects in

Pergularia daemia (5, 10, 19, 40, 77, 153, 306, 615, 1225, 2450 µg/ml).

mouse model of TLE. The effects of P. daemia on seizures induced by PTZ were more marked than those of vitamin C (powerful antioxidant), which was previously reported to mitigate epileptogenesis by blocking the efflux, rather than influx, of calcium, and therefore it interferes with these mechanisms (Xavier et al., 2007; dos Santos et al., 2011). These observations suggest that the extract of P. daemia could interfere with the mechanisms of neurotransmitter release and/or uptake from neuronal terminals. The effects of P. daemia were also more marked than those of the antiepileptic drug sodium valproate. The main mechanisms of valproate include an increase in GABAergic activity, reduction in excitatory neurotransmission and modification of monoamines (Xavier et al., 2007; Loscher and Brandt, 2010; Rahmati et al., 2013; Taiwe et al., 2015). These observations suggest that P. daemia could have altered the epileptogenesis process by increasing GABAergic activity and by reducing excitatory neurotransmission. The antioxidant and anticonvulsant effects of the plant probably mediated by different molecules or mechanisms could have as result a synergic effect greater than the effect of Vitamin C or sodium valproate alone.

Furthermore, P. daemia extract improved cognitive processes as revealed by the elevated plus maze and T-Maze. Considering that cognitive impairment or decline can also be associated with epileptogenesis in TLE (Cha et al., 2002; Stafstrom, 2006; Kumar et al., 2008; Mehla et al., 2010; Pahuja et al., 2013), these findings further suggest that P. daemia extract has antiepileptogenic effects.

The loss of neurons in the hippocampus is the first event characterizing epileptogenesis. This loss of neurons is associated with a significant decrease in total proteins

(Dalby and Mody, 2001; Yamamoto and Mohanan, 2003; Patsoukis et al., 2005; Waldbaum and Patel, 2010). Given that P. daemia prevented the decrease in total proteins induced by pilocarpine, these results suggest that P. daemia has neuroprotective effects (Bahndari et al., 2008; Waldbaum and Patel, 2010). Interestingly, the drastic decrease in AchE activity, marker of neuronal loss (Cavazos and Sutula, 1990; Duysen et al., 2002; Veerendra and Gupta, 2002; Freitas et al., 2005; Niessen et al., 2005), was prevented by P. daemia. These results also suggest that P. daemia has neuroprotective effects. In addition, these effects were more marked than those of vitamin C and valproate sodium. These drugs are known to prevent neuronal loss by preventing oxidative stress (Xavier et al., 2007; dos Santos et al., 2011) and by increasing GABA neurotransmission (Brandt et al., 2003; Loscher and Brandt, 2010), respectively. Altogether, these observations suggest that P. daemia antiepileptogenic effects are mediated by neuroprotective effects.

To confirm the implication of antioxidant pathways in the realization of antiepileptogenic effects of P. daemia, the effects of the extract on antioxidant enzymes and oxidative stress markers in the hippocampus were assessed. In the present study, the activity of SOD enzyme, which protects cells against harmful superoxide radicals and the resulting oxidative stress (Agarwal et al., 2011; Shin et al., 2011), was drastically increased in groups receiving the extract. The activity of CAT enzyme, which eliminate H2O<sup>2</sup> and its toxic radicals resulting from the antioxidant action of SOD (Freitas et al., 2005; Karthishwaran and Mirunalini, 2012; Kiasalari et al., 2013), was also increased in groups treated with P. daemia, in dose dependent manner. These results suggest that the extract induced its antioxidant activities by increasing SOD and CAT activities (Karthishwaran and Mirunalini, 2012; Kiasalari et al., 2013).

Furthermore, decrease in level of MDA, a lipid peroxidation marker caused by free radicals (Dal-Pizzol et al., 2000; Ilhan et al., 2005; Balaji et al., 2013), was also observed. P. daemia treatment increased significantly the tissue levels of GSH, an endogenous antioxidant that reacts with free radicals and prevents the generation of hydroxyl radical (Schulz et al., 2000; Gupta et al., 2003; Agarwal et al., 2011). Altogether, these results also suggest that P. daemia antioxidant activity is mediated in part by the decrease in the MDA level and by the increase in the GSH level. These results are in agreement with a report by Bhaskar and Balakrishnan (2009) in which P. daemia decreased the MDA level and increased the GSH level. However, the level of NO, whose radicals cause oxidative damage via lipid peroxidation (Ilhan et al., 2005; Khadrawy et al., 2013), was not improved in the present study. This result suggests that P. daemia antioxidant activity is not mediated by the inhibition of NO production. This result is in agreement with the study reported by Balaji et al. (2013). Overall, changes in the levels of antioxidant enzymes and oxidative stress markers observed strongly suggest that P. daemia extract has antioxidant properties. These properties were more effective than those of vitamin C, a powerful antioxidant known to increase the SOD and CAT activities, and to decrease the MDA level by scavenging free radicals (Xavier et al., 2007; dos Santos et al., 2011). Taken together, these results suggest that P. daemia antiepileptogenic effects are partly mediated by its antioxidant properties.

Neuronal cell death is a pathophysiological consequence of many brain insults that induced epilepsy (Henshall and Engel, 2013). This event is implicated as a causal factor in epileptogenesis (Henshall and Engel, 2013). Overactivation of glutamate receptors under pathophysiological conditions leads to excitotoxic cell death (Meldrum, 2002; Brown and Bal-Price, 2003). The present findings show that P. daemia significantly protected cortical neurons against excitotoxicity induced by L-glutamate. These results suggest that P. daemia has neuroprotective effects mediated in part by antiapoptotic mechanisms (Narkilahti et al., 2003; Gandhi and Abramov, 2012). Thus, these results explain and confirm antiepileptogenic and neuroprotective effects of P. daemia extract in vivo.

## CONCLUSION

In this study, we investigated the antiepileptogenic and neuroprotective effects of aqueous extract of P. daemia using in vivo and in vitro approaches. In in vivo studies, oral administration of the extract resulted in reduction in the severity of seizures and cognitive impairment. The study of AchE activity and oxidative stress markers revealed that P. daemia extract may mediate its antiepileptogenic effects at least partly through its antioxidant properties. In in vitro studies P. daemia protected cells against death induced by L-glutamate. This effect may be mediated by antiapoptotic pathways. Taken together, these findings indicate that P. daemia has antiepileptogenic and neuroprotective effects. Future experiments aimed at characterizing further the antiepileptogenic properties of P. daemia extract should be designed, considering the therapeutic potential for TLE. This plant could exert some beneficial effect in the threshold of seizures and could be used as complementary treatment for epilepsy and other neurological diseases.

## AUTHOR CONTRIBUTIONS

AK, SP, and GT performed all behavioral studies, accomplished the data analysis and drafted the manuscript. AK, EN, SP, and GT designed the study. EN critically revised the manuscript for important intellectual content. FM, GNg, GNk, JN, JO, SP, and NK helped in in vivo studies. All authors have read and approved the final manuscript.

## ACKNOWLEDGMENTS

The authors are very thankful to Smartox Biotechnologies (France), the University of Ngaoundéré (Cameroon) and the University of Buea (Cameroon). The authors are also thankful Rosette Megnekou, Paul Desire Djomeni Dzeufiet, Paul Etet Seke and Danielle Bilanda for their kindly assistance.

## REFERENCES

fphar-08-00440 June 30, 2017 Time: 16:41 # 12


cytotoxicity and tumor necrosis factor (TNF) activity. J. Immunol. Methods 115, 61–69. doi: 10.1016/0022-1759(88)90310-9



epilepticus in rats. Epilepsia 46, 1021–1026. doi: 10.1111/j.1528-1167.2005. 60704.x



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Kandeda, Taiwe, Moto, Ngoupaye, Nkantchoua, Njapdounke, Omam, Pale, Kouemou and Ngo Bum. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Nootropic and Neuroprotective Effects of Dichrocephala integrifolia on Scopolamine Mouse Model of Alzheimer's Disease

Nadège E. Kouémou1,2, Germain S. Taiwe1,2, Fleur C. O. Moto<sup>3</sup> , Simon Pale1,2 , Gwladys T. Ngoupaye<sup>4</sup> , Jacqueline S. K. Njapdounke<sup>2</sup> , Gisèle C. N. Nkantchoua<sup>2</sup> , David B. Pahaye<sup>2</sup> and Elisabeth Ngo Bum2,5 \*

<sup>1</sup> Department of Zoology and Animal Physiology, Faculty of Science, University of Buea, Buea, Cameroon, <sup>2</sup> Department of Biological Sciences, Faculty of Science, University of Ngaoundéré, Ngaoundéré, Cameroon, <sup>3</sup> Department of Biological Sciences, Higher Teachers' Training College, University of Yaoundé I, Yaoundé, Cameroon, <sup>4</sup> Department of Animal Biology, Faculty of Science, University of Dschang, Dschang, Cameroon, <sup>5</sup> Institute of Mines and Petroleum Industries of Maroua, University of Maroua, Maroua, Cameroon

#### Edited by:

Nilesh Bhailalbhai Patel, University of Nairobi, Kenya

#### Reviewed by:

Zhimin Song, University of Michigan, United States Johanna Mahwahwatse Bapela, University of Pretoria, South Africa

> \*Correspondence: Elisabeth Ngo Bum eli\_bum@yahoo.fr

#### Specialty section:

This article was submitted to Ethnopharmacology, a section of the journal Frontiers in Pharmacology

Received: 09 June 2017 Accepted: 06 November 2017 Published: 21 November 2017

#### Citation:

Kouémou NE, Taiwe GS, Moto FCO, Pale S, Ngoupaye GT, Njapdounke JSK, Nkantchoua GCN, Pahaye DB and Bum EN (2017) Nootropic and Neuroprotective Effects of Dichrocephala integrifolia on Scopolamine Mouse Model of Alzheimer's Disease. Front. Pharmacol. 8:847. doi: 10.3389/fphar.2017.00847 Alzheimer's disease the most common form of dementia in the elderly is a neurodegenerative disease that affects 44 millions of people worldwide. The first treatments against Alzheimer's disease are acetylcholinesterase inhibitors; however, these medications are associated with many side effects. Dichrocephala integrifolia is a traditional herb widely used by indigenous population of Cameroon to treat and prevent Alzheimer's disease and for memory improvement. In this study, we evaluated the effect of the decoction prepared from leaves of D. integrifolia, on scopolamineinduced memory impairment in mice. Seven groups of six animals were used. The first two groups received distilled water for the distilled water and scopolamine groups. The four test groups received one of the four doses of the decoction of the plant (35, 87.5, 175 or 350 mg/kg p.o.) and the positive control group received tacrine (10 mg/kg), a cholinesterase inhibitor used in the treatment of Alzheimer's disease, during 10 consecutive days. Scopolamine (1 mg/kg), a cholinergic receptor blocker, administered 30 min after treatments, was used to induce memory impairment to all groups except the distilled water group on day 10 of drug treatment. The behavioral paradigms used to evaluate the effects of the treatment were the elevated plus maze for learning and memory, Y maze for spatial short-term memory, the novel object recognition for recognition memory and Morris water maze for the evaluation of spatial long-term memory. After behavioral tests, animals were sacrificed and brains of a subset were used for the assessment of some biomarkers of oxidative stress (malondialdehyde and reduced glutathione levels) and for the evaluation of the acetylcholinesterase activity. From the remaining subset brains, histopathological analysis was performed. The results of this study showed that, D. integrifolia at the doses of 87.5 and 350 mg/kg significantly (p < 0.01) improved spatial short-term and long-term memory, by increasing the percentage of spontaneous alternation in the Y maze and reducing the escape latency in the Morris water maze. Furthermore, the results of histopathological evaluation showed that D. integrifolia attenuated the neuronal death in the hippocampus induced

by scopolamine. The main finding of this work is that D. integrifolia improves learning capacities and counteracts the memory impairment induced by scopolamine. Thus, D. integrifolia can be a promising plant resource for the management of Alzheimer's disease and memory loss.

Keywords: Dichrocephala integrifolia, Alzheimer's disease, memory impairment, behavior, scopolamine, acetylcholinesterase inhibitor, oxidative stress

## INTRODUCTION

Alzheimer's disease (AD) the most common form of dementia in the elderly, is a neurodegenerative disease that is clinically characterized by progressive memory loss, cognitive dysfunction and reduction of learning capacities with increase age (Baulac et al., 2003; Cheng et al., 2011; Terry et al., 2011; Alzheimer's Association, 2016). The number of persons affected by dementia worldwide is estimated at 47 million and AD represents 60–80% of this number (Alzheimer's Association, 2016; Alzheimer's Disease International, 2016). Neurofibrillary tangles and neuritic plaques are the two main pathological hallmarks of AD. The AD brain is also characterized by a reduction in cholinergic neurotransmission and an increase in oxidative stress (Francis et al., 1999). The evidence of oxidative stress in the brains of AD patients is the increase of the end products of lipid peroxidation, like malondialdehyde and a reduction in antioxidant enzymes: glutathione and superoxide dismutase (Christen, 2000; Padurariu et al., 2013). The reduction in the cholinergic transmission appears to be the critical element producing dementia. The first treatments against AD are acetylcholinesterase inhibitors, which enhance the cholinergic neurotransmission by increasing the availability of acetylcholine in cholinergic synapse (Giacobini, 2000). Acetylcholinesterase inhibitors have failed in the treatment of AD because of their limited efficacy and bioavailability and because they are associated which many side effects such as hepatotoxicity (Kulkarni et al., 2011). Due to the growing population and extended lifespan, dementia of the type Alzheimer have become a major health concern in the elderly in Africa (Kalaria et al., 2008; Olayinka and Mbuyi, 2014). Dichrocephala integrifolia is a plant of the family Asteraceae that is widely used in traditional medicine in Cameroon to treat and prevent dementia and Alzheimer disease (Ngueguim et al., 2016). In west Cameroon, D. integrifolia is known as "Mbag'api" and the decoction of its leaves is used to treat dementia. The indigenous populations of central region of Cameroon call it "Ngninada Elokn" and use the infusion of the whole plant to treat memory impairment. In far north Cameroon where D. integrifolia is called "Ganki" the decoction prepared from its leaves is used in the treatment of Alzheimer disease (Personal communication). Despite the vast empirical knowledge about the uses of D. integrifolia in the treatment of dementia in Cameroon, pharmacological studies to validate its use in the treatment of dementia are scarce. Thus, the aim of the present study was to evaluate the effect of the decoction of the leaves of D. integrifolia on scopolamine mouse model of Alzheimer's disease.

## MATERIALS AND METHODS

## Animals

The animals used in this study were Swiss mice of either sex weighing 25–30 g. The mice were bred in the animal house of the Faculty of Science of the University of Buea, under a 12 h light/dark cycle. The mice, grouped 6 per cage had food and water available ad libitum. The mice were acclimatized to laboratory conditions for 24 h before the beginning of experimentations. The study was conducted in accordance with the Cameroon National Committee (Ref No. FW-IRB00001954), and was authorized under a number CEI-UDo/908/01/2017/T, and in conformation with the international regulation. All efforts were taken to minimizing the number of mice used and their suffering. Behavioral procedures were performed between 9 a.m. and 4 p.m.

## Plant Material

### Collection and Identification

Dichrocephala integrifolia leaves were harvested in April 2014 in the South–West Region of Cameroon. The harvesting coordinates are 4◦ 150 3000and 9◦ 250 4800. The botanical identification of the plant was done at the National Herbarium of Cameroon, where a voucher specimen was conserved under the reference number: 24276/SRFcam.

#### Preparation of the Decoction of D. integrifolia

The leaves of D. integrifolia were cleaned, shade-dried and ground. The decoction of D. integrifolia was prepared daily according to the instructions of the native doctor. Ten (10 g) of the leaves' powder of D. integrifolia were introduced in 75 ml of distilled water, the mixture was then boiled for 20 min. After cooling, the mixture was filtered using Whatman N◦ 1 filter paper. The filtrate constituted the stock solution. In another experiment, 20 ml of the stock solution was evaporated to dryness and the dry residue obtained was 700 mg. The corresponding concentration of the stock solution was 35 mg/ml. The stock solution was diluted 2; 4; and 10 times in distilled water for less concentrated solutions. All solutions were administered to mice in a volume of 10 ml/kg body weight. The corresponding dose for the stock solution as described by the traditional healer was 350 mg/kg. The doses of the different dilutions were 175; 87.5; and 35 mg/kg.

## Drugs and Chemicals

Tacrine (9-amino-1, 2, 3, 4-tetrahydroacridine hydrochloride), scopolamine hydrobromide, trichloroacetic acid and thiobarbituric acid were purchased from Sigma chemical, St. Louis, MO (United States). Acetylthiocholine iodide, and 5, 5<sup>0</sup> -dithiobis (2-nitrobenzoic acid) (Ellman reagent) were purchased from Biochemica (China).

## Study of the Effect of D. integrifolia on Memory Impairment Induced by a Single Dose of Scopolamine

This test was used to assess the effect of the decoction of D. integrifolia administered for 10 consecutive days against memory impairment induced by a single injection of scopolamine at the dose of 1 mg/kg i.p. The behavioral tasks used to evaluate the effect of the treatment were the Y maze, the elevated plus maze and the novel object recognition task.

## General Experimental Design

fphar-08-00847 November 20, 2017 Time: 17:21 # 3

In this part of the work, mice were randomly divided into seven groups of six mice each (three males and three females) and group as follow:

> Group I: The distilled water group; which received only distilled water (10 ml/kg) orally;

> Group II: Scopolamine group; which received distilled water (10 ml/kg) orally;

> Groups III–VI: Tests groups; which received one of the four doses of the decoction of D. integrifolia; 350, 175; 87.5, or 35 mg/kg, orally

> Groups VII: Tacrine group; which received tacrine at the dose of 10 mg/kg orally.

All these groups received the corresponding treatment for 10 consecutive days. On day 10, 30 min after the various treatments, scopolamine (1 mg/kg i.p.) was injected to all groups except the distilled water group that still received distilled water. The behavioral tests were performed 30 min after the injection of scopolamine.

## Behavioral Assessment

### **Y-maze test**

Y-maze test was used to evaluate short-term memory of mice by recording spontaneous alternation in a single session on day 10. The maze used in this study was a Y-maze made of polywood with three identical arms (35 cm length × 8 cm height × 15 cm width) mounted at 120 degrees to one another in a single piece. Each arm of the Y-maze was decorated with a different letter (A, B, or C) in order to be differentiated (Ma et al., 2007). One hour after the last treatment and 30 min after scopolamine injection (except for the distilled water group), each mice, previously naive to the maze, was placed at the end of one arm and were allow to move freely through the maze during 8 min. The number of arm entries was recorded for each mouse. An arm entry was noted when a mouse entered an arm of the maze with all its paws. Specific sequences of arm transitions (ABC, BCA, or CAB but not BAB or CAC or CBC) were recorded as a spontaneous alternation that reflects short-term memory. The total number of arm entries reflects general locomotor activity. The arms of the maze were cleaned between sessions with 10% ethanol. The percentage of spontaneous alternation was defined according to the following equation:percentage of spontaneous alternation = [(Number of alternations)/(Total arm entries − 2)] × 100 (Ma et al., 2007; Hritcu et al., 2012; Beppe et al., 2014).

### **Elevated plus-maze test**

Elevated Plus Maze (EPM) is an exteroceptive behavioral model used to evaluate learning and memory in rodents (Itoh et al., 1990; Sharma and Kulkarni, 1992; Kulkarni et al., 2011). The EPM was in plywood and comprised two open arms (30 cm × 5 cm) and two closed arms (30 cm × 5 cm × 15 cm) that extended from a common central platform (5 cm × 5 cm). The maze was elevated to a height of 50 cm above the floor level.

On the 1st day of the test (day 9 of drug treatment), 1 h after various treatments, each mouse was placed at the end of an open arm, facing away from the central platform for a learning trial. Transfer latency (TL) was taken as the time taken by the animal to move into any one of the covered arms with all its four limbs. The cutting time was 120 s after this time an animal that did not enter into one of the covered arms was gently push into one and the TL was assigned as 120 s. The mouse was allowed to explore the maze for another 2 min. On the 2nd day of the test (day 10 of drug treatment), during retention phase, retention transfer latency was recorded 1 h after the last treatment and 30 min after scopolamine injection (except for the distilled water group). Reduction in TL values of the 2nd day of the test in comparison to the 1st day test indicates improvement in memory (Itoh et al., 1990; Sharma and Kulkarni, 1992; Joshi and Parle, 2007; Kulkarni et al., 2011).

## **The novel object recognition test (NOR)**

The NOR test was used to evaluate recognition memory of mice. This test was performed in an open field apparatus consisted of square plywood of dimensions (40 cm × 40 cm × 25 cm). The day before the test (day 8 of drug treatment), 1 h after drug treatment, each mouse individually was familiarized with the apparatus for 5 min. On the 1st day of the test (day 9 of drug treatment) 1 h after drug treatment, two identical objects were presented to each mouse for a 5 min session of exploration. An animal explores an object when it touches the object or it directs its nose at a distance less than 2 cm to the object. The 2nd day of the test (day 10 of drug treatment), 30 min after scopolamine injection (except for the distilled water group), a new object replaced one of the objects presented in the 1st day. The time spent by the animal for exploring the new object (tB) and the familiar (tA) objects was recorded during 5 min (Kulkarni et al., 2011). The discrimination index (DI) was calculated as (tB/tB+tA) (Ennaceur and Delacour, 1988; Rajendran et al., 2014).

## Study of the Effect of D. integrifolia against Memory Impairment Induced by Repeated Doses of Scopolamine Experiment Design

To delineate the mechanism by which D. integrifolia exerts its protective effect, in a subsequent test, the mice were divided into seven groups of six as described above. Scopolamine (1 mg/kg i.p.) were injected to all the groups every day for 10 days 1 h after 30 min after drug administration except the distilled water which received injection of saline (10 ml/kg ip).

The behavioral task used to evaluate the effect of the treatment was Morris water maze task (MWM).

Behavioral Assessment: The Morris Water Maze Task

The MWM test was used to evaluate spatial long-term memory of mice. The MWM was performed as previously described by Morris in 1984 with little modifications (Morris, 1984; Parle and Singh, 2007). The MWM consisted of a brown circular pool (100 cm diameter, 50 cm height). The pool was located in a room with various visual cues (pictures, shelters, curtains, lamps, etc. . .). The position of the pool and that of the cues were maintained all the days of the experimentations. The pool was filled with water at the temperature of 25 ± 2 ◦C. The MWM was virtually divided into four equal quadrants: North, South, East, and West. A platform of white color (11 cm diameter and 16 cm height) was centered in the South–East quadrant 1 cm below the water surface. The water was whitened by addition of liquid milk so that the platform was invisible at water surface. The position of the platform was kept unaltered during the training session. The 1st day of the MWM test (day 6 of drug treatment), 1 h after drug administration and 30 min after scopolamine injection, each mouse received an acclimatization session during which, the mouse was placed inside the MWM for swimming for 60 s. During the acquisition phase (days 7–9 of drug treatment), 30 min after scopolamine injection, each mouse was released into the pool, head facing the wall. The cutting time for each trial was 120 s. each mouse that did not find the platform during the time was gently guide to it and allowed there for 10 s. Each animal had four training sessions per day of 5 min interval. After each trial each mouse was taken to its cage and was allowed to dry up under a 60 watt bulb.

During each trial session, the time taken to reach the platform (escape latency) was recorded with stopwatches. In the retention phase (day 10 of drug treatment), the platform was removed from the pool. Each mouse individually was placed into the MWM. The latency time taken to reach the place of the formal platform and the time spent in the target quadrant was recorded during 120 s using stopwatches by experimenter researchers.

#### Biochemical Assays

On day 11 following the MWM test, mice were decapitated under light ether anesthesia. In each group, the brain of a sub set of animal was used for histopathological analysis and the other for the dosage of brain malondialdehyde (a marker of lipid peroxidation), reduced glutathione (the principal antioxidant enzyme of the body) and acetylcholinesterase activity (which give an idea on brain acetylcholine level).

#### **Tissue preparation**

After sacrifices, the brains were immediately removed, from the skull, rinsed and weighed. Each brain was divided into two cerebral hemispheres. Brain homogenate was prepared from one half with 50 mM Tris/HCl buffer for the assessment of brain malondialdehyde (MDA) and reduced glutathione levels. For the assessment of acetylcholine esterase activity, the other hemisphere of the brains was used to prepare 10% homogenate with 50 mM Tris/HCl buffer containing 1% Triton –X.

## **Estimation of protein concentration**

The total protein of brains homogenate was determined by the method described by Bradford (1976). Five (5) µl of the brain homogenate was introduced in microplate wells and 250 µl of Bradford reagent was added. After agitation, the absorbance of the mixture was read using a microplate reader at 590 nm. The determination of the protein concentration was done using bovine serum albumin (BSA) as standard.

#### **Determination of brain acetylcholinesterase activity**

The determination of brain acetylcholinesterase activity was performs using acetylthiocholine iodide as artificial substrate based on the colorimetric method, of Ellman (Ellman et al., 1961). Briefly, 925 µL of 0.5 mM of Ellman reagent prepared in 100 mM Tris buffer (pH.8) was mixed with 50 µL of 20 mM acetylthiocholine iodide and 25 µL of supernatant. The change in absorbance was monitored using a spectrophotometer during 3 min at 30 s interval. The activity of AchE is expressed as AChE activity is expressed as micromoles of acetylthiocholine iodide hydrolyzed per milligram of protein per minute. Each assay was done in duplicate.

## **Brain reduced glutathione level**

Reduced glutathione (GSH) level was estimate in the brain supernatant according to the method of Ellman (1959). Twenty (20) µl of brain homogenates were mixed with 3 ml of Ellman reagent prepared in phosphate buffer (0.1 M pH 7.2) at room temperature. After 1 h, the absorbance of the mixture was read at 412 nm. The amount of glutathione was calculated with the formula of Beer Lambert using the extinction coefficient value of 13,600/M/cm (Fotio et al., 2009). Each assay was done in triplicate.

#### **Brain malondialdehyde level**

The brain malondialdehyde (MDA) level was measured in the supernatant using the thiobabituric assay. One (1 ml) of brain's supernatant was added to 0.5 ml of trichloroacetic acid (20%) and 1 ml of thiobarbituric acid (0.67%). The mixture was heated in a water bath at 100◦C for 60 min. After cooling, the mixture was centrifuged at 3000 rpm for 15 min. The absorbance of the supernatant was read at 530 nm. The amount of MDA was calculated with the formula of Beer Lambert using the extinction coefficient value of 1.56 × 10<sup>5</sup> M/cm. The concentration of MDA is expressed as µmol/g tissue (Nelson et al., 1994; Fotio et al., 2009). Each assay was done in triplicate.

#### Histopathological Studies

After sacrifices, the brains were fixed in 10% formol for a week. Fifty (50) µm coronally sections were made from the brain in the hippocampus region using the mouse brain Atlas with the following coordinate (Anterior/Posterior = −2.0 mm, Medial/lateral = −1.5 mm, and dorsal/ventral AP = −2.0 mm) (Paxinos and Franklin, 2001). The brains sections were collected in nine well plates. The dehydration of brain section consisted in introducing brain session in ascending concentration of ethanol and then followed by immersion in xylol and then embedding in paraffin. Paraffin sections of the brain were deparaffinized and rehydrated through washes in descending concentration series of

FIGURE 1 | Effect of Dichrocephala integrifolia on scopolamine-induced memory impairments in the Y-maze test. The spontaneous alternation percentage (A) and the numbers of arm entries (B). Each column represents mean ± SEM of six mice. Data analysis was performed using one way ANOVA followed by Tukey multi comparison test, <sup>∗</sup>P ≤ 0.01, ∗∗∗P ≤ 0.001 vs. scopolamine treated group (DW + Sco); <sup>c</sup>P < 0.001 vs. distilled water group. DW, distilled water; DI, D. integrifolia; Sco, scopolamine; Tac, tacrine.

alcohol. Brain sections were then stained using the Nissl stain. After drying overnight, the brain sections were photographed and images were captured using a digital camera attached to a light microscope.

## Statistical Analysis

Statistical analysis was done using the software Graphpad Instat for windows. The differences amongst groups were analyzed using One-Way Analysis of Variance (ANOVA). P-values ≤ 0.05 were considered significant. Tukey post hoc test were used for multiple comparisons.

## RESULTS

## Effects of D. integrifolia on Spontaneous Alternation of Scopolamine Treated Mice in the Y Maze

The results of spontaneous alternation show that there was a significant difference among all the treatments groups (P < 0.0001). Scopolamine reduced the mice spontaneous alternation. Ten days treatment with D. integrifolia at all doses significantly reversed the effect of scopolamine and increased the spontaneous alternation percentage (P < 0.0001), when compared to scopolamine-alone treated group. Tacrine 10 mg/kg pre-administration also reversed the reduction of spontaneous alternation induced by scopolamine (**Figure 1A**). Tacrine reversed the effects of scopolamine at the percentage of 78.59% and the decoction at the dose of 87.5 mg/kg at a percentage of 80.21%.

Dichrocephala integrifolia did not increased or impaired the locomotion of scopolamine treated mice in the Y maze (**Figure 1B**).

## Effects of D. integrifolia on Transfer Latencies in an Elevated Plus Maze of Scopolamine-Treated Mice

As shown in **Figure 2A**, there was a significant difference among groups concerning the initial transfer latency. The decoction of D. integrifolia at the doses of 87.5 and 350 mg/kg decreased the

initial transfer latency to 68.33 ± 3.33 s and 67.66 ± 3.04 s, respectively, against 97.5 ± 5.12 s in the distilled water group at day 9. Tacrine group had also a short initial transfer latency of 63.83 ± 3.39 s on day 9 when compared to distilled water group (P < 0.0001).

∗∗P < 0.01, ∗∗∗P < 0.001 vs. Scopolamine treated group (DW + Sco). DW, distilled water; DI, D. integrifolia; Sal, saline; Sco, scopolamine; Tac, tacrine.

A single administration of scopolamine significantly increased the retention transfer latency (RTL, recorded on the second testing day) of scopolamine alone treated group when compared to distilled water group. This RTL was 42.16 ± 2.07 s in the distilled water group against 76.16 ± 2.49 s in the scopolamine alone treated group. The decoction of D. integrifolia at all the doses significantly (P < 0.0001) reduced the RTL. Tacrine also reversed the retention deficit induced by scopolamine on day 10 of treatment (**Figure 2B**).

## Effects of D. integrifolia on Recognition Memory of Scopolamine Treated Mice

The NOR test was used to assess recognition memory of mice after a single injection of scopolamine. The administration of scopolamine before the retention phase of the test resulted in a reduction of the exploration of the novel object in comparison with the ancient object (**Figure 3**). The discrimination index which is 0.62 ± 0.03 in distilled water group was significantly reduced (P < 0.0001) to a value of 0.014 ± 0.04 in the scopolamine-alone-treated group.

Ten (10) days pretreatment of mice with the decoction of D. integrifolia at the dose of 87.5 mg/kg significantly (P < 0.01) increased the exploration of the novel object in comparison with the ancient object. Thus, the discrimination index rose from a value of 0.014 ± 0.04 in the scopolamine-alone group to 0.46 ± 0.10 at the dose 87.5 mg/kg of D. integrifolia. During the retention phase, tacrine significantly (p < 0.01) increased the exploration of the novel object in comparison with the ancient object presented during the acquisition phase. The discrimination index of the tacrine group was 0.51 ± 0.80 (**Figure 3**).

## Effects of D. integrifolia on Acquisition and Retention Parameters of Scopolamine-Treated Mice in a Morris Water Maze Task

#### Acquisition

Dichrocephala integrifolia at all doses significantly reduced the acquisition deficit caused by scopolamine starting from the 2nd day of the acquisition phase (day 7) (P < 0.0001) (**Table 1**). The time to find the invisible platform in the 2nd day of the acquisition was 69.91 ± 4.41 s in the scopolamine alone against 14.58 ± 2.58 s at the dose of 87.5 of D. integrifolia.

#### Retention

As shown in **Figure 4A**, D. integrifolia at all doses and also tacrine 10 mg/kg significantly reduced the latency time to the non-existing platform on the retention phase when compared to scopolamine-alone treated group (P < 0.0001). The latency time to reach the non-existing platform was 61.66 ± 9.83 s in the scopolamine alone treated group against 13.16 ± 2.48 s at the dose 87.5 mg/kg of D. integrifolia. Furthermore, D. integrifolia from the dose of 87.5 mg/kg and tacrine significantly increased the time spent in the target quadrant during the retention



Results are expressed as mean ± SEM for six mice, <sup>∗</sup>p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 compare to scopolamine alone treated group (DW + Sco). <sup>a</sup>p < 0.05, <sup>b</sup>p < 0.01, <sup>c</sup>p < 0.0001 compare to control group (DW + Sal). F(6,35) = 4.366; P = 0.0022 (Day 6); F(6,35) = 16.67; P < 0.0001 (Day 7); F(6,35) = 16.67; P < 0.0001 (Day 8); F(6,35) = 136.26; P < 0.0001 (Day 9). DW, distilled water; DI, D. integrifolia; Sal, saline; Sco, scopolamine; Tac, tacrine.

phase when compared to scopolamine–alone treated group (P < 0.0001) (**Figure 4B**).

## Effects of D. integrifolia on Brain Acetylcholinesterase Activity, Malondialdehyde and Reduced Glutathione Levels

The activity on AChE was significantly reduced by D. integrifolia at the dose of 87.5 mg/kg, Tacrine also inhibited the effect of scopolamine induced increased in the activity of AChE when compared to scopolamine-alone treated group (P < 0.0001) (**Table 2**).

The co-administration of D. integrifolia from the dose 87.5 mg/kg and scopolamine significantly reduced the level of brain malondialdehyde when compared to scopolamine-alone treated group (P < 0.0001) (**Table 2**). Tacrine also reduced the level of MDA (**Table 2**).

The level of GSH was lower in the scopolamine -alone -treated group. The administration of the decoction of D. integrifolia from the dose of 87.5 mg /kg significantly (P < 0.0001) reverses the reduction of GSH induced by scopolamine (**Table 2**).

## Results of the Histopathological Studies

The histopathological analysis show that the dentate gyrus of distilled water group of mice is normal without any sign of neurodegeneration or necrosis (**Figure 5A**). The hippocampal sections of scopolamine-treated group show a significant reduction in the density of cells of all the layers of the dentate gyrus associated with the presence of apoptotic cells (**Figure 5B**). The decoction of D. integrifolia at the doses of 87.5, 175, and 350 mg/kg show a normal architecture of the cells layer of the dentate gyrus (**Figures 5C–E**, respectively). Tacrine group shows a dentate gyrus without any sign of necrotic or apoptotic cells (**Figure 5F**).

## DISCUSSION

Alzheimer's disease is a deadly progressive neurodegenerative disorder of the elderly associated with loss of memory and cognitive dysfunctions (Behl, 2002; Madeo and Elsayad, 2013). Cumulative evidences have suggested that the cognitive symptoms of AD are a result of the impairment of adult neurogenesis in the hippocampus (Demars et al., 2010; Lazarov and Marr, 2013). AD has become a public health burden due to

TABLE 2 | Effects of Dichrocephala integrifolia on acetylcholinesterase activity, malondialdehyde and reduced glutathione levels of scopolamine treated mice.


Results are expressed as mean ± SEM for six mice, <sup>∗</sup>p < 0.05, ∗∗p < 0.001, ∗∗∗p < 0.0001 compare to scopolamine alone treated group (DW + Sco). <sup>a</sup>p < 0.05, <sup>b</sup>p < 0.001, <sup>c</sup>p < 0.0001 compare to control mice (DW + Sal). DW, distilled water; DI, D. integrifolia; Sal, saline; Sco, scopolamine; Tac, tacrine.

the increase of aging population and increase lifetime expectancy (Kalaria et al., 2008; Olayinka and Mbuyi, 2014).

In this study, we evaluated the effect of the decoction of D. integrifolia against scopolamine model of AD. Scopolamine is an alkaloid extracted from the Solanaceae Datura stramonium and which impairs short-term and long term-memory in animals and humans (Rabiei et al., 2015). Through the interference with acetylcholine in the brain, scopolamine can cause oxidative stress leading to cognitive impairment (Rahnama et al., 2015). Thus, scopolamine-induced memory impairment is a valid model for the evaluation of anti – amnesic effects of new drugs. Diverse behavioral animal models are usually used for the evaluation and validation of new drugs against dementia (Rajendran et al., 2014).

In the present study, 10 days pretreatment of animals with D. integrifolia significantly counteracted the reduction of the percentage of spontaneous alternation induced by scopolamine suggesting significant improvement of space related short – term memory by the decoction of D. integrifolia (Hritcu et al., 2012, 2015; Beppe et al., 2014). The effect of D. integrifolia and that of tacrine a cholinesterase inhibitor used in the treatment of AD were comparables.

The EPM test was used for the evaluation of learning and memory. The EPM is based on the apparent natural aversion of rodents to open and high spaces, and originally, it is used for measurement of anxiety (Pellow et al., 1985; Pellow and File, 1986; Lister, 1987). Some parameters of the EPM such as retention transfer latency (the time take by the animal to move from the open arms to the enclosed arms) is used for the evaluation of memory. And an animal that has previously (acquisition trial) experienced entering the open arms have the shortened transfer latency in the retention trial (Itoh et al., 1990). In the EPM test, D. integrifolia significantly reduced the initial transfer latency on day 10. These results suggest that D. integrifolia has a nootropic effect because it ameliorates the retention of informations in absence of any memory impairment inducer (Itoh et al., 1990; Rajendran et al., 2014). Furthermore, D. integrifolia significantly reduced the retention transfer latency on day 10 after scopolamine injection suggesting that D. integrifolia ameliorates learning and retention of information and plays a role in memory formation (Itoh et al., 1990; Sharma and Kulkarni, 1992; Joshi and Parle, 2007; Kulkarni et al., 2011; Rajendran et al., 2014).

The NOR test was used to evaluate the effect of D. integrifolia on recognition memory. We found that, 10 days pretreatment of mice with D. integrifolia significantly reversed the reduction of the discrimination index induced by scopolamine suggesting the effect of the plant on recognition memory (Ennaceur and Delacour, 1988). All these above results clearly show that D. integrifolia has a neuroprotective activity in vivo by counteracting memory impairment induced by scopolamine in a variety of behavioral paradigms.

To delineate the mechanism by which D. integrifolia exerts his neuroprotective activity, D. integrifolia was administered during 10 days consecutively 1 h prior scopolamine (1 mg/kg i.p.) injection. The MWM task was used as behavioral task. The results obtained show that like tacrine, D. integrifolia significantly reduced the learning and retention deficits caused by repeated does of scopolamine. D. integrifolia reduced the time to the invisible platform during acquisition and the latency time to the non-existing platform during retention phase. D. integrifolia also significantly increased the time spent in the target quadrant during this retention phase. Our results with the MWM suggest that D. integrifolia improves spatial long-term memory (Cheng et al., 2011). The results of MWM confirmed that pretreatment with D. integrifolia counteracted scopolamine induced learning and memory deficit thus D. integrifolia is neuroprotective (Konar et al., 2011; Hritcu et al., 2015).

The results of biochemical assays show that 10 days administration of scopolamine increased the activity of AChE and the level of MDA, a measure of brain lipid peroxidation (Hritcu et al., 2015) and reduced the level of GSH, the main antioxidant enzyme of the body (Agarwal et al., 2011). Our results are in accordance with literature that shows that administration of scopolamine in rodents can lead to increase of AChE activity and oxidative status in the brain (Cheng et al., 2011; Konar et al., 2011; Baradaran et al., 2012; Rajendran et al., 2014). Pretreatment

of mice with D. integrifolia reversed the increase of the activity of AChE and oxidative stress induced by scopolamine, thus protecting animals against learning and memory loss.

The results of histopathological studies demonstrated that 10 days administration of scopolamine resulted in neurodegenerative processes in the dentate gyrus when compared to naïve mice.

This cell death in the hippocampus dentate gyrus was significantly prevented by a pretreatment with D. integrifolia. The dentate gyrus is the part of the brain where adult neurogenesis takes place and it is also implicated in hippocampal neurogenesis and plasticity (Kempermann et al., 2015). Besides the facts that memory impairment induced by scopolamine is a result of an increase in AChE activity and brain oxidative status (Chen et al., 2008; Konar et al., 2011), it can be also assumed that scopolamine impairs neurogenesis in the brain which in turn leads to cognitive deficits as in AD (Demars et al., 2010; Lazarov and Marr, 2013). By antagonizing the cell death in the dentate gyrus induced by scopolamine, D. integrifolia can be a good treatment for cognitive deficits and AD. There are cumulative evidences in literature that scopolamine influences acquisition, consolidation and recall of informations and that scopolamine is a cholinergic blocker (Agrawal et al., 2009; Konar et al., 2011; Rajendran et al., 2014). By counteracting the effect of scopolamine, D. integrifolia can have the same mechanism of action, as tacrine wish is a cholinergic enhancer widely used in the treatment of AD. Furthermore, D. integrifolia had demonstrated many beneficial effects against others diseases such as hepatotoxicity probably due to its antioxidant properties (Ngueguim et al., 2016). This property of the plant may also be strongly involved in its neuroprotective effects observed in our study. This study shows that D. integrifolia has an ability to improve learning of information, ameliorates spatial short-term and long-term memory and recognition memory. The mechanism by which D. integrifolia exerts its effects may

## REFERENCES


be related to the reduction of AChE level associated with antioxidant properties and improvement of adult neurogenesis. The overall results of this study can explain the wide usage of D. integrifolia in the treatment of dementia in Central Africa.

## CONCLUSION

The results of this study shows that the decoction of D. integrifolia counteracted scopolamine-induced memory impairment and oxidative stress. Thus, it can be concluded that D. integrifolia can be a valuable plant resource for the management of dementia in general an age-related cognitive deficit of Alzheimer's type in particular. Nevertheless, more studies with D. integrifolia targeting other hypotheses of AD are needed to clarify the exact mechanism of action of the plant.

## AUTHOR CONTRIBUTIONS

NK, GT, SP, FM, GTN, and EB conceived and designed the work. NK, GT, SP, JN, GCNN, and DP collected and analyzed the data. NK, GT, and EB wrote and revised the manuscript. All authors read and approved the final manuscript.

## ACKNOWLEDGMENTS

This research was supported by the University of Buea, the University of Ngaoundere and the University of Maroua. The authors acknowledged Drs. Manfo T. F. Pascal, Fotio T. Agathe, Dongmo N. Mireille, and Dzeufiet D. P. Desiré for their technical assistance during this work. A special acknowledgment goes to Mr. Egbe B. Besong who edited the manuscript.



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Kouémou, Taiwe, Moto, Pale, Ngoupaye, Njapdounke, Nkantchoua, Pahaye and Bum. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Anti-inflammatory, Antinociceptive, and Antioxidant Activities of Methanol and Aqueous Extracts of Anacyclus pyrethrum Roots

Houria Manouze<sup>1</sup> , Otmane Bouchatta<sup>1</sup> , A. Chemseddoha Gadhi<sup>2</sup> , Mohammed Bennis<sup>1</sup> , Zahra Sokar<sup>1</sup> and Saadia Ba-M'hamed<sup>1</sup> \*

<sup>1</sup> Laboratory of Pharmacology, Neurobiology and Behavior (URAC-37), Faculty of Sciences Semlalia, Cadi Ayyad University, Marrakech, Morocco, <sup>2</sup> Unit of Phytochemistry and Pharmacology of Aromatic and Medicinal Plants, Laboratory of Biotechnology, Protection and Valorization of Plant Resources (URAC35), Faculty of Sciences Semlalia, Cadi Ayyad University, Marrakech, Morocco

#### Edited by:

Vivienne Ann Russell, University of Cape Town, South Africa

#### Reviewed by:

Willias Masocha, Kuwait University, Kuwait Regina A. Mangieri, University of Texas at Austin, United States

> \*Correspondence: Saadia Ba-M'hamed bamhamed@uca.ma

#### Specialty section:

This article was submitted to Ethnopharmacology, a section of the journal Frontiers in Pharmacology

Received: 14 May 2017 Accepted: 18 August 2017 Published: 05 September 2017

#### Citation:

Manouze H, Bouchatta O, Gadhi AC, Bennis M, Sokar Z and Ba-M'hamed S (2017) Anti-inflammatory, Antinociceptive, and Antioxidant Activities of Methanol and Aqueous Extracts of Anacyclus pyrethrum Roots. Front. Pharmacol. 8:598. doi: 10.3389/fphar.2017.00598 Anacyclus pyrethrum (L.) is a plant widely used in Moroccan traditional medicine to treat inflammatory and painful diseases. The objective of the present study was to evaluate the antinociceptive, anti-inflammatory and antioxidant activities of aqueous and methanol extracts of Anacyclus pyrethrum roots (AEAPR and MEAPR). The antiinflammatory effect of AEAPR and MEAPR was determined in xylene–induced ear edema and Complete Freund's Adjuvant (CFA)-induced paw edema. The antinociceptive activity of AEAPR and MEAPR (125, 250, and 500 mg/kg) administered by gavage was examined in mice by using acetic acid-induced writhing, hot plate, and formalin tests, and the mechanical allodynia were assessed in CFA-induced paw edema. In addition, the in vitro antioxidant activities of the extracts were determined by using 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging method, ferric reducing power and β-carotene-linoleic acid assay systems. AEAPR and MEAPR produced significant reductions in CFA-induced paw edema and xylene-induced ear edema. A single oral administration of these extracts at 250 and 500 mg/kg significantly reduced mechanical hypersensitivity induced by CFA, which had begun 1 h 30 after the treatment, and was maintained till 7 h. Chronic treatment with both extracts significantly reduced mechanical hypersensitivity in persistent pain conditions induced by CFA. Acute pretreatment with AEAPR or MEAPR at high dose caused a significant decrease in the number of abdominal writhes induced by acetic acid injection (52.2 and 56.7%, respectively), a marked increase of the paw withdrawal latency in the hot plate test, and also a significant inhibition of both phases of the formalin test. This antinociceptive effect was partially reversed by naloxone pretreatment in the hot plate and formalin tests. Additionally, a significant scavenging activity in DPPH, reducing power and protection capacity of β-carotene was observed in testing antioxidant assays. The present study suggests that AEAPR and MEAPR possess potent anti-inflammatory, antinociceptive and antioxidant effects which could be related to the presence of alkaloids and phenols in the plant. In addition, the antinociceptive effect of APR extracts seems to partly involve the opioid system. Taken together, these results suggest that Anacylcus pyrethrum may indeed be useful in the treatment of pain and inflammatory disorders in humans.

Keywords: Anacyclus pyrethrum, anti-inflammatory, antinociceptive, antioxidant, mice

## INTRODUCTION

fphar-08-00598 September 1, 2017 Time: 16:33 # 2

Chronic pain is a serious problem globally. Pain affects one in five adults, while an estimated one in ten suffers from chronic pain each year (Goldberg and McGee, 2011). Because of their involvement in most diseases, inflammation and pain have become the most important topic of several scientific researches. Although non-steroidal anti-inflammatory drugs (NSAIDs) and opiates have been classically used to manage pain, some adverse reactions may occur with these drugs, such as gastrointestinal disturbances, renal damage, respiratory depression, and possible dependence (Burke et al., 2006; Farshchi et al., 2009; Vittalrao et al., 2011). Therefore, new and more effective anti-inflammatory and analgesic drugs without side effects are targeted as an alternative to NSAIDs and opiates (Kumara, 2001; Dharmasiri et al., 2003). Recently, research has focused on herbal medicines used in traditional medicine due to their low cost, efficacy and safety. According to the World Health Organization (WHO), about 80% of the world population still relies mainly on herbal remedies (Kumara, 2001; Li et al., 2003).

Anacylcus pyrethrum (L.) Link (Asteraceae) is a native plant of North Africa (Boulos, 1983), locally known as "Aqar-qarha" or "Tigandizt" by Moroccan people. The root is widely used in Moroccan traditional medicine to treat rheumatism, sciatica, colds, neuralgia, and paralysis (Bellakhdar, 1997). It is also considered to be a sialagogue, sudorific and to relieve toothache (Doudach et al., 2012; Zaidi et al., 2013). Previous studies have indicated that the plant possesses antimicrobial activities (Doudach et al., 2012; Jalayer et al., 2012; Selles et al., 2013), and it has anti-diabetic (Tyagi et al., 2011; Selles et al., 2012b), aphrodisiac (Vikas et al., 2009) and hepatoprotective effects (Usmani et al., 2016). The plant is reported to have immunemodulatory and immune-stimulating properties (Bendjeddou et al., 2003; Ching et al., 2007; Dalila et al., 2010), in addition to anti-inflammatory and antioxidant potential (Müller et al., 1993; Sujith et al., 2011a; Selles et al., 2012a).

The phytochemical screening of Anacylcus pyrethrum has led to the identification of various secondary metabolites such as alkaloids, reducing compounds, tannins, flavonoids and coumarins (Hanane et al., 2014). This species also contains saponins, sesamin, inulin, gum and traces of essential oil (Sujith et al., 2011b; Selles et al., 2012b). The most important phytoconstituents present in its root are N-isobutyldienediynamide and polysaccharides (Crombie, 1954; Bendjeddou et al., 2003; De Spiegeleer et al., 2011; Boonen et al., 2012).

Although some Anacyclus species such as Anacyclus clavatus have been studied for their anti-inflammatory and antioxidant activities (Aliboudhar and Tigrine-Kordjani, 2014), no scientific reports on the antinociceptive or anti-inflammatory activities of Anacyclus pyrethrum in vivo have been conducted. Thus, the aim of the present study was to provide scientific evidence for the antinociceptive and anti-inflammatory activities of methanol and aqueous extracts of Anacyclus pyrethrum roots (APR) using appropriate models in mice. In addition, as inflammation is a process linked to oxidative stress and to the over-production of the reactive oxygen species (ROS), the antioxidant capacity of both extracts was also evaluated.

## MATERIALS AND METHODS

## Preparation of the Extracts and Phytochemical Screening Plant Material

Anacylcus pyrethrum (L.) Link (Asteraceae) was collected in June 2014, from Oukaïmeden (74 km from Marrakech) at 2,600 m of altitude in the High-Atlas Mountains (Morocco). The plant was identified by Professor A. Ouhammou, a taxonomist in the department of Biology, Faculty of Sciences Semlalia, Cadi Ayyad University. A voucher specimen was deposited at the Faculty's Herbarium (Mark 8258).

### Extracts Preparation

The roots were separated from the aerial parts of the plant and dried under shade. They were ground to a fine powder using a grinder apparatus. Root powder (1 g) was stirred with distilled water (20 ml) for 24 h at room temperature. The aqueous macerate was centrifuged (1200 rpm) for 15 min. The supernatant was lyophilized (yield = 20% w/w) then stored in a freezer at −20◦C until experimental use. For methanol extract preparation, the powder of APR (400 g) was exhaustively extracted with methanol in a Soxhlet apparatus. The methanol extract was concentrated to dryness under vacuum. The residue (21.8% w/w) was stored at −20◦C for several months, until the experimental use.

#### Phytochemical Screening of APR Extracts

Aqueous and methanol extracts of APR were screened for the presence of flavonoids, alkaloids, terpenoids, tannins and saponins. The qualitative determination of these phytochemicals

**Abbreviations:** AEAPR, Aqueous Extract of Anacyclus pyrethrum Roots; BCB, β-carotene bleaching; BHT, butylated hydroxy toluene; CFA, Complete Freund's adjuvant; COX, cyclooxygenase; DPPH, 2, 2-di-phenyl-1-Picryl-Hydrazyl; FRAP, ferric reducing antioxidant power; i.p., intraperitoneal administration; Indo, indomethacin; LOX, 5-lipoxygenase; MEAPR, Methanol Extract of Anacyclus pyrethrum Roots; NSAIDs, non-steroidal anti-inflammatory drugs; ROS, reactive oxygen species.

was conducted using previously reported methods (Harborne, 1998; Bargah, 2015).


## Animals

Adult Swiss male mice (25–35 g) were used for in vivo bioassays. The animals were provided by the Animal Care Facility of the Faculty of Sciences Semlalia, Cadi Ayyad University, Marrakech, Morocco. The mice were kept under constant conditions of ambient temperature (22 ± 2 ◦C) under a 12 h light/12 h dark cycle, with ad libitum access to food and water. All animal procedures were in strict accordance with the guidelines of the European Council Directive (EU2010/63). Care was taken to minimize the number of animals used for the experiments. All efforts were made to minimize any animal suffering, and the study met the ethical standards and approvals of the Council Committee of the research laboratories of the Faculty of Sciences, Cadi Ayyad University of Marrakech.

## Drugs and Reagents

Indomethacin was purchased from Laprophan (Morroco) and naloxone from Hospira (United States). Acetic acid, formalin, xylene, xylazine, ketamine and CFA (suspension of heat-killed Mycobacterium tuberculosis in oil, 1 mg/ml) were obtained from Sigma–Aldrich (France).

## Acute Toxicity

The acute toxicity study was conducted according to the Organization for Economic Cooperation and Development (OECD) guideline no. 423 (OECD, 2001), where the limit test dose of 5000 mg/kg was used. Mice were divided equally into nine groups (six animals per group). Eight groups were orally treated by gavage with different doses (500, 1000, 2000, and 5000 mg/kg) of AEAPR or MEAPR solution at 10 ml/kg. One group that received vehicle (distilled water) was included as a negative control. To detect signs of toxicity and death, mice were observed within the first 12 h after drug administration. The mice were daily weighed and observed for 14 days after treatment. At the end of the 14-day period, the animals were injected with a urethane lethal dose (1 g/kg, i.p.) and the vital organs were immediately removed and weighed.

## Assessment of APR Extracts Anti-inflammatory Activity Xylene-Induced Ear Edema

The xylene-induced ear edema test was performed as previously described (Tang et al., 1984). Briefly, each mouse was given by gavage one dose (125, 250, or 500 mg/kg) of AEAPR or MEAPR, indomethacin (10 mg/kg) or vehicle (10 ml/kg) 1 h before induction of ear edema by topical application of 0.02 ml xylene on the inner and outer surfaces of the right ear. The left ear was used as a control. One hour after xylene application, mice (six animals per group) were sacrificed by cervical dislocation. Circular sections of 5 mm were excised and weighed. To evaluate the extent of the ear edema, we calculated the weight difference between the right and the left ear sections of the same animal.

## Complete Freund's Adjuvant-Induced Paw Edema

Eight groups of animals (six mice per group) were assigned to this test. Mice were anesthetized with a mixture of ketamine (50 mg/kg) and xylazine (2 mg/kg) cocktail. All groups, except vehicle control group, received 20 µl of CFA by a subcutaneous injection in the plantar surface of the right hind paw (Bortalanza et al., 2002). Twenty four hours after CFA injection, mice were treated by gavage with AEAPR, MEAPR (125, 250, or 500 mg/kg), indomethacin (10 mg/kg), or vehicle (10 ml/kg). The effect of each treatment on edema development induced by the intraplantar injection of CFA was evaluated according to the previously reported method of Milano et al. (2008). The level of inflammation of the right hind paw was measured using a caliper at several time-points (0, 0.5, 1, 2, 4, 5, 6, and 7 h) and was expressed in millimeters.

## Assessment of APR Extracts Antinociceptive Activity Mechanical Allodynia

To measure the nociceptive reactivity to the application of mechanical stimuli to the hind pad, each mouse was placed in an individual observation cage (12 cm × 12 cm × 12 cm) with a mesh floor allowing access to the ventral surface of the hind pads. The animals were accustomed to the cage and the experiment room for at least 10–15 min or at the end of the exploratory behavior. Then mechanical hypersensitivity was assessed as described by Chaplan et al. (1994). Calibrated Von Frey filaments of increased tension were applied through the wire mesh floor of the cage, 10 times per paw with enough force to

cause buckling of the filament. The withdrawal threshold for Von Frey assay was determined as the filament at which the animal withdrew its paw at least five times in ten applications. Basal responsiveness to mechanical stimuli was assessed on the day before CFA injection. Twenty-four hours after CFA injection, mice were treated by gavage with AEAPR, MEAPR (125, 250, or 500 mg/kg), indomethacin (10 mg/kg), or vehicle (10 ml/kg) and then, the withdrawal threshold was evaluated at several time points, but always between 9 am and 6 pm.

To investigate the effect of chronic treatment on paw withdrawal, and the possible development of tolerance, mice (six animals per group) were treated with AEAPR, MEAPR (125, 250, or 500 mg/kg), indomethacin (10 mg/kg), or vehicle (10 ml/kg) once a day for 5 days successively. In order to investigate the possible development of tolerance, the treatment was interrupted for 3 days (from day 6 to day 8) and reinitiated for 2 days again (day 9 and day 10). The mechanical hypersensitivity was assessed 3 h after each daily treatment (the time where the maximal response was observed in the acute treatment). For the days 6– 8 (without treatment), the test was performed at exactly the same time as the previous days.

### Thermal Hyper-Nociception

In this test, animals were individually placed on a hot plate with an adjustable temperature (to 55 ± 1 ◦C) (Bhandare et al., 2010). The reaction time was defined as the latency for the animal to lick its paw(s) or jump from the plate. The cutoff time for the hot plate latencies was set at 30 s. Ten groups of six mice each were used for this test. They were treated orally with 125, 250, or 500 mg/kg of AEAPR or MEAPR, vehicle (10 ml/kg), indomethacin (10 mg/kg), naloxone (1 mg/kg) and AEAPR (500 mg/kg) or naloxone (1 mg/kg) and MEAPR (500 mg/kg). Naloxone, an opioid receptor antagonist, was administered alone or 30 min before the administration of MEAPR or AEAPR. The latency of nociceptive response was recorded before treatment and at 30, 60, 90, 120, 150, 180, 210, and 240 min after drug administration.

## Acetic Acid-Induced Abdominal Writhing

This test was performed in mice, according to the method described by Ferreira et al. (2000). Briefly, 0.6% acetic acid solution (10 ml/kg) was injected intraperitoneally. Animals (six mice per group) were pre-treated with AEAPR or MEAPR (125, 250, or 500 mg/kg), vehicle (10 ml/kg) or indomethacin (10 mg/kg), 30 min prior to peritoneal irritation. The resulting writhes and stretching were observed and counted over a period of 60 min after acetic acid injection.

## Formalin-Induced Pain

Formalin test was carried out as previously reported (De Miranda et al., 2001). Ten groups of six mice each were used for this test: the negative control group (vehicle, 10 ml/kg), the positive control group (indomethacin, 10 mg/kg), six groups receiving AEAPR or MEAPR (125, 250, or 500 mg/kg) and three groups treated with naloxone (1 mg/kg): alone, 30 min before the administration of MEAPR or AEPRR (500 mg/kg). After treatment administration, each animal received an intraplantar injection of 2% formalin (20 µl/ animal) into the right paw. Total time spent licking the injected paw was recorded during two phases: 5–10 min after formalin injection and 15–30 min after formalin injection (Hunskaar and Hole, 1987; Tjølsen et al., 1992).

## Motor Performance and Locomotor Activity

To evaluate any coordination disruption, non-specific musclerelaxant or sedative effects of APR extracts, mice were subjected to the rotarod task and open-field test. The motor coordination of the mice was evaluated on the rotarod apparatus at a constant speed of 12 rotations per minute. Twenty-four hours prior the drug testing, animals were tested and those who remained at the rotating bar for the full 300 s during three consecutive trials were used for the subsequent experiments. The selected animals were randomly distributed into groups of six mice and received AEAPR or MEAPR orally at a dose of 125, 250, or 500 mg/kg. The control group received the same volume of vehicle (10 ml/kg). An additional group received indomethacin (10 mg/kg) and served as a positive control. Rotarod tests were performed prior to drug administration (basal) and at 30, 60, and 120 min after administration. Latency to fall off was measured for each session (up to 300 s).

In order to evaluate eventual motor impairment induced by plant extract, the mice were placed individually in an observation chamber 60 min after oral treatment with vehicle, indomethacin or APR extracts. The open field apparatus used was a 50 × 50 cm square arena with 30 cm high black walls. The animal was placed in the center of the arena and distance traveled was scored during 10 min, using the videotracking EthoVision XT8.5 software (Noldus, Netherlands). The animal was returned to its home cage, and the apparatus cleaned with ethanol 70% to remove any odor.

## Antioxidant Activity

## DPPH Free Radical-Scavenging Activity

The hydrogen or electron donation ability of APR extracts (MEAPR, AEAPR) was measured using the stable radical 1, 1diphenyl-2-picryl hydrazyl (DPPH) assay (Burits and Bucar, 2000). Various concentrations of MEAPR and AEAPR were prepared in methanol and methanol-water, respectively (V/V). An aliquot (2 ml) of a 60 µM methanol solution of DPPH was mixed with 50 µl solution of each sample. After 20 min incubation at room temperature, the absorbance was read at 517 nm. DPPH free radical-scavenging activity of each sample was expressed as the percentage decrease of DPPH absorbance versus control. It was calculated using the formula: % Inhibition = [(A<sup>b</sup> – Aa)/Ab] × 100, where A<sup>b</sup> is the absorbance of DPPH alone in methanol (control) and A<sup>a</sup> is the absorbance of DPPH in the presence of the test substance. Quercetin and butylated-hydroxyl-toluene (BHT) were used as positive controls.

## β-Carotene/Linoleic Acid Bleaching Assay

The β-carotene/linoleic acid test evaluates the lipoperoxydation inhibitory effect of a compound or a mixture of compounds. The method described by Miraliakbari and Shahidi (2008), was used with slight modifications. A mixture of β-carotene

and linoleic acid was prepared by adding together 0.5 mg β-carotene in 1 ml chloroform (HPLC grade), 25 µl linoleic acid and 200 mg Tween 40. The chloroform was then completely evaporated under vacuum and 100 ml of oxygenated distilled water was subsequently added to the residue and mixed to form a clear yellowish emulsion. Three hundred fifty microliters of various concentrations of the sample (MEAPR, AEAPR, BHT or Quercetin) were added to 2.5 ml of the above emulsion in test tubes and mixed. The test tubes were incubated in a water bath at 50◦C for 2 h together with a negative control (blank) containing methanol instead of samples. The absorbance values were measured at 470 nm. The lipoperoxydation inhibitory effect of testing samples was evaluated by calculating the inhibition percentage using the following equation: I% = (Aβ−carotene after 2 hassay/Ainitial <sup>β</sup>−carotene) × 100, where Aβ−carotene after 2 hassay is the absorbance of β-carotene remaining in the sample after 2 h and Ainitial <sup>β</sup>−carotene is the absorbance of β-carotene at the beginning of the experiment.

#### Reducing Power Determination

The ability of APR extracts to reduce Fe+<sup>3</sup> to Fe+<sup>2</sup> was investigated using the method of Oyaizu (1986). The tested sample (MEAPR, AEAPR or control) was mixed with phosphate buffer (2.5 ml, 0.2 M, pH 6.6) and potassium ferricyanide (2.5 ml, 1%). The mixture was then incubated at 50◦C for 20 min. Subsequently, 2.5 ml of trichloroacetic acid (10%) was added to the mixture, which was then centrifuged for 10 min at 3000 rpm. Finally, the upper layer of the solution (2.5 ml) was mixed with distilled water (2.5 ml) and FeCl<sup>3</sup> (0.5 ml, 0.1%), and the absorbance was measured at 700 nm in a spectrophotometer. BHT and Quercetin were used as positive controls.

## Statistical Analysis

Statistical analysis was performed using SigmaPlot 11.0 software. All data were presented as mean ± SEM for six mice per group. A two-way analysis of variance (ANOVA) with repeated measures followed by Holm–Sidak post hoc analysis was used to examine the time-courses of mechanical, thermal and anti-edema effects after various treatments. Kruskal–Wallis ANOVA or one-way ANOVA followed by post hoc testing with the Student-Newman–Keuls test for multiple comparisons were used to measure variance of mouse behavior between groups. The significance of the differences between the means in the antioxidant activity was assessed by Student's t-test. The significance threshold was set at p < 0.05.

## RESULTS

## Phytochemical Screening of APR Extracts

The qualitative phytochemical screening of AEAPR and MEAPR showed the presence of alkaloids, flavonoids, tannins, saponins, and terpenoids in both extracts.

## Acute Toxicity

The oral administration of AEAPR or MEAPR at doses up to 5000 mg/kg did not produce any visible signs or symptoms of toxicity in mice. No mortality and no significant changes in body weights (p > 0.05) or in organ weights (p > 0.05) were observed at 14 days after AEAPR or MEAPR administration (**Table 1**).

## Anti-inflammatory Effect of Anacyclus pyrethrum root Extracts Xylene-Induced Ear Edema

Topical application of xylene caused a significant increase in the right ear section's weight (11.68 ± 0.49 mg) when compared to the vehicle-treated left ear (5.22 ± 0.29 mg) (p < 0.001). Indomethacin, as well as both APR extracts at tested doses, reduced xylene-induced ear edema compared to the vehicletreated control group (**Figure 1**). Indeed, Kruskal–Wallis oneway analysis of variance confirmed a high significant difference between all groups [H(7) = 30.63, p < 0.001]. The post hoc analysis showed that the pretreated groups with APR extracts or indomethacin decreased significantly (p < 0.001) the edema induced by xylene in comparison with vehicle-treated group. In the group treated with indomethacin, the inhibition of ear edema induced by xylene was significantly higher than that of the group treated with 125 mg/kg of MEAPR or AEAPR (q = 4.61 and q = 4.77, p < 0.001 respectively), whereas the oral administration of the two extracts at a dose of 500 mg/kg exhibited a very strong anti-edema effect compared to the indomethacin group (q = 5.05 and q = 6.46, p < 0.001; respectively) (**Figure 1**). However, we observed no significant difference in ear edema between AEAPR and MEAPR at a dose of 250 mg/kg vs. indomethacin (q = 2.73 and q = 1.95, p > 0.05) nor between AEAPR vs. MEAPR groups (q = 1.03, p > 0.05). The positive control, indomethacin, inhibited edema by 49%, whereas MEAPR and AEAPR at a dose of 500 mg/kg inhibited it respectively by 65% and 62% compared to the group treated with the vehicle.

#### CFA-Induced Inflammatory: Paw Thickness

Complete Freund's Adjuvant injection caused a significant increase (t = 10.59, p < 0.001) in paw volume after 24 h (from 4.50 ± 0.08 mm to 5.10 ± 0.05 mm). The paw thickness was reduced respectively by MEAPR, AEAPR and indomethacin after 30, 60, and 120 min of oral administration (5.04 ± 0.00 mm, 4.94 ± 0.04 mm and 4.90 ± 0.03 mm, respectively) (**Figure 2**). This reduction was statistically different between experimental groups [F(8,54) = 23.49, p < 0.001], and also between different post-treatments' times [F(8,54) = 399.38, p < 0.001]. Both APR extracts showed dose- and time-dependent anti-inflammatory effects. At higher doses (250 and 500 mg/kg), this effect became significant, starting from the third hour after AEAPR administration (**Figure 2A**) and starting from the second hour for MEAPR (**Figure 2B**). It does not become significant until the fourth and fifth hour after the administration of the low dose (125 mg/kg) of MEAPR or AEAPR (t = 2.16 and t = 2.39, p < 0.05, respectively). In addition, at a higher dose, AEAPR and MEAPR exhibited a greater reduction effect in paw thickness than indomethacin from time points 3 h (t = 2.50 and t = 2.69,

TABLE 1 | Effect of a single oral administration of AEAPR and MEAPR on body weight and relative organ weights of mice.


Data are expressed as mean ± SEM. There were no statistically significant differences in the body weight (b.wt) or relative organ weights.

p < 0.05; respectively), to 7 h (t = 3.10 and t = 3.59, p < 0.01; respectively).

## Antinociceptive Effect of Anacyclus pyrethrum Root Extracts

#### CFA-Induced Mechanical Hypersensitivity

The intraplantar injection of CFA produced noticeable mechanical hypersensitivity (The paw withdrawal threshold reduced from 1.57 ± 0.15 g at baseline to 0.42 ± 0.04 g (t = 12.21; p < 0.001), 24 h after CFA injection (**Figure 3**). The single oral administration of AEAPR or MEAPR, at 250 and 500 mg/kg, reduced the mechanical hypersensitivity induced by CFA. The paw withdrawal threshold increased significantly from point time 1 h 30 to 3 h and remained high up to 7 h (**Figure 3**). Two-way repeated measures ANOVA revealed that the withdrawal threshold was significantly different between groups [F(8,54) = 51.89, p < 0.001] and varied significantly with time [F(8,54) = 178.95, p < 0.001]. The post-hoc analysis showed a significant increase of withdrawal threshold at time point of 1h30 in mice treated with AEAPR or MEAPR (250 and 500 mg/kg) and indomethacin compared to the CFA group (AEAPR: t = 2.22, p < 0.05 and t = 3.50, p < 0.001; MEAPR: t = 2.70, p < 0.01 and t = 3.98, p < 0.001; indomethacin: t = 4.38, p < 0.001; respectively). The CFA group did no differ at 125 mg/kg of both extracts, except for MEAPR at the time point 3 h (t = 2.23, p < 0.05). Moreover, no significant difference was observed between indomethacin and MEAPR or AEAPR at 500 mg/kg (t = 0.40 and t = 0.88, p > 0.05).

To investigate the effects of repeated treatment, mice received daily MEAPR or AEAPR (125, 250, or 500 mg/kg), indomethacin, or vehicle for 5 days, interrupted for 3 days, then given daily for 2 more days (**Figure 4**). The results demonstrated that the threshold responses of animals after AEAPR or MEAPR treatments were increased during the observation period compared to CFA group. When the treatments were interrupted for 3 days, mechanical allodynia was re-established. On the 9th day, the treatments were restarted and it was observed that both extracts of APR once again reduced mechanical allodynia.

A two-way repeated measures ANOVA (treatment and day) revealed a significant effect of treatment [F(8,54) = 171.13, p < 0.001] and time [F(8,54) = 60.97, p < 0.001]. On the first day of treatment (day 1), except for the dose of 125 mg/kg of both extracts, the post hoc analysis showed significant differences between CFA group and MEAPR (250 and 500 mg/kg; t = 4.114 and t = 6.098, p < 0.001; respectively), AEAPR (250 and 500 mg/kg; t = 2.775, p < 0.01 and t = 5.433, p < 0.001; respectively) and indomethacin (t = 8.091, p < 0.001) treated groups. In addition, there was no significant difference, neither between MEAPR and AEAPR groups at 500 mg/kg nor between MEAPR and indomethacin groups (p > 0.05), whereas a significant difference between AEAPR and indomethacin groups (t = 2.66, p < 0.01) was shown. During the first treatment period (from day 1 to day 5) with MEAPR, AEAPR and indomethacin, the withdrawal thresholds were significantly increased (p < 0.001) compared to the day before the treatment (day 0). The interruption of the treatment induced a reduction in paw withdrawal threshold of animals treated with APR extracts at all doses. However, this weakening of the antinociception reaction after the treatment interruption

FIGURE 2 | Effect of acute administration of AEAPR (A) and AEAPR (B) on CFA-induced paw edema. Animals (six mice each group) were treated orally with vehicle (distilled water), indomethacin (Indo), MEAPR or AEAPR (125, 250, and 500 mg/kg), 24 h after intraplantar injection of Complete Freund's adjuvant (CFA: 20 µl). Each point represents the mean ± SEM of the paw thickness (in mm), before (basal level: BL), 24 h after the injection of CFA (time point 24 h) and at the time points indicated after drug administration. Holm–Sidak post hoc analysis; <sup>a</sup>p < 0.05, <sup>b</sup>p < 0.01 and <sup>c</sup>p < 0.001 in comparison with CFA + vehicle treated group; <sup>d</sup>p < 0.05; <sup>e</sup>p < 0.01 in comparison with CFA + indomethacin treated group, using Holm–Sidak post hoc analysis.

FIGURE 3 | Effect of acute administration of AEAPR (A) and MEAPR (B) on mechanical hypersensitivity in mice. Animals (six mice each group) were treated orally with vehicle (distilled water), indomethacin (Indo), MEAPR or AEAPR (125, 250, and 500 mg/kg), 24 h after intraplantar injection of Complete Freund's adjuvant (CFA; 20 µl). Each point represents the mean ± SEM of the paw withdrawal threshold (in grams) to hind paw stimulation, before (basal level: BL), 24 h after the injection of CFA (time 24 h) and at the times indicated after drug administration. <sup>a</sup>p < 0.05, <sup>b</sup>p < 0.01 and <sup>c</sup>p < 0.001 in comparison with CFA + vehicle treated group; <sup>d</sup>p < 0.05; <sup>e</sup>p < 0.01 and <sup>f</sup>p < 0.001 in comparison with CFA + indomethacin treated group, using Holm–Sidak post hoc analysis.

was not significant at the higher dose. The significant increase of threshold responses reappeared with the resumption of treatment (**Figure 4**).

#### Acetic Acid-Induced Abdominal Writhing

The oral administration of AEAPR or MEAPR at a dose of 250 and 500 mg/kg reduced the number of writhes compared to the vehicle group (**Figure 5**). The one-way ANOVA analysis showed a significant effect of treatment on the number of writhes [F(7,48) = 8.14, p < 0.001]. In fact, post hoc analysis showed that the extent of induced writhing at doses of 250 and 500 mg/kg of AEAPR or of MEAPR was significantly reduced compared to the vehicle group (AEAPR: q = 4.66, p < 0.01; q = 7.80, p < 0.001; and MEAPR: q = 4.76, p < 0.01; q = 8.55, p < 0.001); whereas both extracts of APR at 125 mg/kg, did not produce an antinociceptive effect (p > 0.05). In addition, the number of mouse abdominal constrictions induced by acetic acid in MEAPR treated group did not differ from that of the AEAPR treated group (p > 0.05) at high doses. Moreover, the test revealed that both extracts, at a dose of 125 mg/kg, reduced the number of writhes

FIGURE 4 | Effect of chronic administration of AEAPR (A) or MEAPR (B) on mechanical hypersensitivity in mice. Animals (six mice each group) were treated orally with vehicle (distilled water), indomethacin (Indo), MEAPR or AEAPR (125, 250, and 500 mg/kg), 24 h after intraplantar injection of Complete Freund's adjuvant (CFA; 20 µl). Each point represents the mean ± SEM of the paw withdrawal threshold (in grams) to hind paw stimulation before (basal level: BL), 24 h after CFA injection and at the times (days) indicated after drug administration. The treatment was given once a day for 5 days successively. On the 6th, 7th, and 8th day the drug treatment was avoided and reinitiated for 2 days again in order to evaluate resistance effect of the APR extracts. <sup>a</sup>p < 0.05, <sup>b</sup>p < 0.01 and <sup>c</sup>p < 0.001 in comparison with CFA + vehicle treated group; <sup>d</sup>p < 0.05; <sup>e</sup>p < 0.01 and <sup>f</sup>p < 0.001 in comparison with CFA + indomethacin treated group, using Holm–Sidak post hoc analysis.

but to a lesser degree than that observed with indomethacin (MEAPR: q = 4.09 and AEAPR: q = 3.80, p < 0.05), unlike the effect of the two higher doses (250 and 500 mg/kg), which did not significantly differ from indomethacin's one (p > 0.05). APR extracts treatments at 125, 250, and 500 mg / kg induced contraction inhibition of 19.2, 30.9, and 52.2% respectively for AEAPR and 22.3, 31.6, and 56.7%, respectively, for MEAPR, while the standard drug (indomethacin) had an inhibition rate of 46.4%.

#### Thermal Sensitivity

In the hot plate test, the oral administration of indomethacin, MEAPR or AEAPR significantly increased the paw withdrawal latency compared to the vehicle-treated group (**Figure 6**). This effect began 60 min after oral administration of high doses and then disappeared after 240 min. At a lower dose of APR extracts (125 mg/kg), the paw withdrawal latency started to increase significantly after 90 min of treatment with MEAPR and only after 120 min of treatment with AEAPR (**Figure 6**); two-way repeated measures ANOVA (group and time effect) confirmed these findings [F(7,48) = 17.13, p < 0.001 and F(7,48) = 69.52, p < 0.001, respectively]. Compared to baseline, the post hoc analysis showed that the treatment with indomethacin, MEAPR or AEAPR at higher doses (250 and 500 mg/kg) increased significantly the paw withdrawal latency, starting from 60 to 180 min (p < 0.001). In addition, naloxone (an opioid antagonist), which showed any significant effect on thermal sensitivity (in comparison with control: t = 0.746, p > 0.05), partially reversed the effect induced by AEAPR or MEAPR at 500 mg/kg in the hot plate test. In fact, we observed a decrease of the paw withdrawal latency of 33.68 or 38.77%, respectively, in comparison with both AEAPR and MEAPR at time point 120 min.

#### Formalin-Induced Pain

In the formalin test, pain responses such as licking of the right hind paw were expressed in the first and second phases. The licking time of the first phase was reduced when the mice were treated with 125, 250, and 500 mg/kg of MEAPR (**Figure 7A**), and one-way ANOVA confirmed these differences [F(7,48) = 4.95,

time points after the drug administration. <sup>a</sup>p < 0.05, <sup>b</sup>p < 0.01, and <sup>c</sup>p < 0.001 in comparison with the vehicle-treated group; <sup>d</sup>p < 0.05, <sup>e</sup>p < 0.01, and <sup>f</sup>p < 0.001

p < 0.001]. The post hoc analysis showed a significant difference between treated groups with 250 and 500 mg/kg of MEAPR or AEAPR and vehicle-treated group (p < 0.05). In addition, the analysis showed a significant difference between AEAPR at 125 mg/kg and indomethacin (p < 0.05), whereas there was no significant difference between indomethacin and MEAPR treated groups (p > 0.05). Concerning the second phase, all treatments markedly reduced the licking time of the injected paw in comparison with the group treated with vehicle (**Figure 7B**). The one-way ANOVA indicated that all treatments induced significant differences of the mean paw licking time in comparison to vehicle-treated group [F(7,48) = 18,25, p < 0.001]. The post hoc analysis showed that the licking times at all doses of each APR extract and indomethacin were significantly (p < 0.001) less than that of the vehicle group. In contrast to the first phase, the effect of MEAPR and AEAPR at 500 mg/kg scored higher (p < 0.05) than indomethacin in the second phase with percentages of inhibition of 66, 64, and 43%, respectively. The pre-treatment with naloxone (1 mg/kg) reversed the antinociceptive activity of AEAPR and MEAPR at a dose of 500 mg/kg, i.e., an increase in the licking time of the first phase (23.80 and 27.09%; respectively) and the second phase (66.43 and 65.21%; respectively) in the formalin test. While the effect of naloxone treatment alone, did not differ from that of the vehicle (p > 0.05).

in comparison with the indomethacin-treated group, using Holm–Sidak post hoc analysis.

## Measurement of Motor Performance and Locomotor Activity

Treatment with MEAPR or AEAPR, even at the highest dose tested (500 mg/kg), did not significantly reduce locomotion in the open field (vehicle: 2772.43 ± 144.72 cm; MEAPR: 2841.30 ± 338.26 cm; AEAPR: 2948.83 ± 320.80 cm; Indo: 2947.02 ± 189.08 cm, n.s), and produced no difference in latency to fall off the bar in the rotarod test (vehicle: 281.66 ± 14.30 s; MEAPR: 274.04 ± 25.91 s; AEAPR: 279.75 ± 20.21 s; Indo: 269.70 ± 19.11 s, n.s).

## Antioxidant Activity of Anacyclus pyrethrum Extracts

The antioxidant activity of Anacylcus pyrethrum root was assessed by three complementary in vitro antioxidant assays: the DPPH, the FRAP and the BCB assays. The concentrations that led to 50% inhibition (IC50) are given in **Table 2**. Note that low IC<sup>50</sup> values reflect better protective action. The results showed that both Anacylcus pyrethrum extracts (MEAPR and AEAPR) exhibited similar and interesting antioxidant activity, especially in DPPH test with IC<sup>50</sup> values of 12.38 ± 0.28 µg/ml and 13.41 ± 0.67 µg/ml respectively. The antioxidant potencies of both extracts were significantly less than those of the reference antioxidants butylated hydroxytoluene (BHT) and quercetin (**Table 2**). No significant difference was observed between AEAPR and MEAPR in any of the three antioxidant assays (p > 0.05). Recorded results showed that among the used bio-assays, the antioxidant activities of MEAPR and AEAPR were significantly higher, with DPPH than with other assays (p < 0.001).

## DISCUSSION

In the present work, APR extracts were first investigated for acute toxicity in male Swiss mice. AEAPR and MEAPR did not produce any sign of acute toxicity nor mortality up to the maximum dose of 5000 mg/kg during the 14 days following their single administration. Thus, both extracts have an LD<sup>50</sup> higher than 5000 mg/kg. According to the chemical labeling and classification of acute systemic toxicity recommended by OECD, AEAPR and

MEAPR were assigned to the lowest toxicity class (OECD, 2001), which is in accord with the conclusions of previous studies

pyrethrum root; Indo, indomethacin; Nal, naloxone.

(Doudach et al., 2012; Kishor Kumar and Lalitha, 2013). In the present study, the evaluation of the activity of MEAPR and AEAPR revealed that both extracts possessed potent anti-inflammatory effects in both acute and chronic models of inflammation. Indeed, we have shown that that AEAPR and MEAPR exhibited higher activity to counter the acute inflammation in a xylene-induced-ear-edema model (65 and 62%; respectively) than that of indomethacin (49%). This model is widely used to evaluate anti-inflammatory topical steroids and non-steroidal anti-inflammatory agents, especially those that inhibit phospholipase A2 (Zanini et al., 1992; Núñez Guillén et al., 1997). In addition, several studies have shown that the xylene increases vascular permeability, which causes edema that indicates inflammation, due to the release of inflammatory agents such as bradykinin, prostaglandin, histamine and serotonin, which in turn release neuropeptides that activate their receptors,

TABLE 2 | Antioxidant activities, expressed as IC50 values using DPPH, FRAP and BCB assays, for AEAPR, MEAPR, BHT and quercetin.


Values are the mean of three determinations ± SEM, ∗∗∗p < 0.001 significantly different from the BHT group and ###p < 0.001 significantly different from the quercetin group. DPPH, 2,2-diphenyl-1-picrylhydrazyl; FRAP, ferric reducing antioxidant power; BCB, β-carotene bleaching; BHT, butylated hydroxy toluene.

producing neurogenic inflammation (Carlson et al., 1985; Banki et al., 2014). Substance P is one of these neuropeptides, which allows the release of nitric oxide from endothelial cells, triggering vasodilation and plasma exudation, which is the origin of edema formation (Barry et al., 2011). On the basis of these data, the significant inhibition of xylene-induced ear swelling by AEAPR and MEAPR is certainly due to the action of the phytochemicals detected in those extracts and which may act, individually or in synergy, at different levels of the multifactorial process of inflammation. Among the potential active metabolites flavonoids have a membrane-stabilizing effect by reducing vasodilatation, which ameliorates the strength and integrity of blood vessel walls (Pathak et al., 1991); while alkaloids may act through the prevention of the neurogenic inflammation.

In addition, to investigate the effects of APR in a sustained inflammation model, we evaluated the effect of AEAPR and MEAPR on inflammation induced by intraplantar injection of CFA. Our results showed for the first time that acute or chronic oral treatments of animals with MEAPR or AEAPR were effective in preventing not only paw edema caused by CFA injection, but also mechanical hypersensitivity. Moreover, the anti-edematogenic and antinociceptive actions of APR extracts were evident from an early stage (1 h30) and maintained up to 7 h.

It has been reported that CFA induces persistent pain that results mainly from the involvement of macrophages and T lymphocytes in the injected rat paw, followed by a paw swelling and leukocyte infiltration of the synovium and surrounding tissue, which contribute to chronic inflammation and osteolytic lesions (Butler et al., 1992; Romas et al., 2002; Mo et al., 2013). In addition, activation of macrophages results in the production of pro-inflammatory cytokines (IL-1β and TNF-α),

growth factors and other inflammatory mediators, resulting in peripheral sensitization (Kumar et al., 2010; Mo et al., 2013). Indeed, inflammation causes the induction of COX-2 (Vane and Botting, 1998) leading to the release of nitric oxide, bradykinin and prostanoids, which induce phosphorylation of ion channels in nociceptor terminals, enhancing excitability and reducing the nociception threshold, and therefore development of peripheral sensitization. This mechanism of nociceptors sensitization is widely involved in all types of inflammatory pain and is associated with chronic pain (Basbaum, 1999; Basbaum et al., 2009).

Since the antinociceptive effect of AEAPR and MEAPR is associated with anti-inflammatory action, its continuing antinociceptive effect in the chronic pain model may be due to a reduction in the cytokine and prostanoid release which reduced sensitization of the nociceptors. It should be noted that Crombie (1954) has isolated dodeca-2E,4E-dienoic acid isobutylamide from APR (i.e., pellitorine) and that Cech et al. (2010) have reported that the same extract component from Echinacea purpurea roots exerted immuno-modulatory effects, especially with a decrease of plasma protein levels of certain pro-inflammatory cytokines (IL-8 and IL-6) and inversely an increased expression of anti-inflammatory molecules such as IL-10. Thus, although we didn't quantify the levels of the cytokines, we may predict that APR extract may act via the same product (dodeca-2E,4E-dienoic acid isobutylamide) to blunt the induced inflammation.

Assessment of AEAPR and MEAPR effects on the abdominal constrictions elicited by acetic acid showed a marked suppression of writhing response in the visceral pain model. This test is mainly used to screen antinociceptive activity (Utsunomiya et al., 1998). It has been reported that acetic acid induces the release of endogenous mediators that activate the nociceptive neurons (Collier et al., 1968). Indeed, acetic acid acts by releasing biogenic amines (e.g., bradykinin, and serotonin), cyclooxygenases and their metabolites (e.g., PGE2 and PGF2α) in the peritoneal fluid (Derart et al., 1980; Dhara et al., 2000). It also activates peritoneal receptors (Bentley et al., 1983; Medzhitov, 2008) and stimulates nociceptive nerve terminals (Duarte et al., 1988). The present work demonstrated that indomethacin causes a significant inhibition of acetic acidinduced pain, which is in agreement with previous reports indicating that this test is sensitive to non-steroidal antiinflammatory drugs (NSAIDs) (Gené et al., 1998; Vane and Botting, 2003). According to Rimbau et al. (1999), the alkamides from APR extracts act as a dual inhibitor of cyclooxygenase (COX) and 5-lipoxygenase (LOX) enzymes. COX catalyzes the conversion of arachidonic acid to prostaglandin (Williams et al., 1999), leading to the activation and sensitization of peripheral nociceptors. Since the vast majority of studies of alkamides shows their peripheral antinociception effects (de la Rosa-Lugo et al., 2017), on the basis of the Rimbau et al. (1999) study and our current findings, we suggest that the peripheral antinociception activity of AEAPR and MEAPR may be due to their alkamides.

The central antinociceptive activity of MEAPR and AEAPR was evaluated in the hot plate test. This test is a widely used model for acute thermal nociception to evaluate specifically central nociception (Eddy and Leimbach, 1953). Through this test of complex responses to inflammation and nociception (Bhandare et al., 2010), centrally acting antinociceptive drugs (i.e., opioid agents) elevate the nociception threshold of rodents toward heat via spinal and supraspinal receptors (Zakaria et al., 2008; Amabeoku and Kabatende, 2012). The present results showed that the oral treatment with both extracts of APR provided antinociceptive effects in a dose-dependent manner, indicating likely a central action. Furthermore, treatment with indomethacin (an NSAID drug) induced a significant increase in the latency time in the hot plate test, which is in accordance with a previous study indicating that the NSAID shows a central anti-nociception mediated by either serotoninergic (5-HT2/5-HT3) or adrenergic (α1/α2) receptors at the spinal/supraspinal level (Arslan and Bektas, 2015). Therefore, the efficacy of AEAPR and MEAPR in the hot plate test might be due to analgesic agent(s) acting primarily at the spinal, medullary, and/or higher levels of the CNS or by some indirect mechanisms as suggested for narcotic substances by Yaksh and Rudy (1977) and Tjølsen et al. (1992). Otherwise, it has been reported that the antinociceptive effect of affinin, alkamides isolated from Heliopsis longipes, could be a result of the activation of opioidergic, serotoninergic and GABAergic systems (Déciga-Campos et al., 2010). Indeed, the most important phytoconstituents present in the APR are alkamides (Crombie, 1954; Boonen et al., 2012). Thus, the central anti-nociception of MEAPR and AEAPR may due to its alkamides component.

To discriminate between the peripheral and central antinociceptive effects of APR extracts, the formalin test was used. Our results revealed that both AEAPR and MEAPR have acted effectively in both phases of the formalin test. It is known that the intraplantar injection of formalin induces two phases of pain sensitivity (Tjølsen et al., 1992). The first and the second phases are characterized as neurogenic pain and inflammatory pain, respectively. The mechanisms underlying these two phases are previously reported (Hunskaar and Hole, 1987; Corrêa and Calixto, 1993; do Amaral et al., 2007; McNamara et al., 2007). Based on previous literature, we believe that our results show MEAPR and AEAPR exert their analgesic action at both central and peripheral levels.

Moreover, in this study, we demonstrated that the antinociceptive action of AEAPR and MEAPR was partially antagonized by naloxone in the hot plate, and in the formalin test (first and the second phase). These results suggest that the APR extracts mechanism involves in part opioid receptors.

In addition to this, we did not detect any disturbances in the locomotor activity or motor performance in animals treated by AEAPR or MEAPR up to 500 mg/kg. This suggests that both extracts at the highest effective dose, have no muscle-relaxant or central depressant action in models of nociception used in our study. Our result is in opposition of the study of Zaidi et al. (2013), which demonstrated that ethanolic extract of APR impaired motor coordination at a dose of 1600 mg/kg in

Rotarod performance. However, this difference could be due to environmental conditions which could influence the expression of phytochemical compounds in the same plants grown in different areas (Harborne, 1993; Lee et al., 2007). In addition, several studies have reported variations in the biological activities of extracts prepared using different extraction techniques (Dhanani et al., 2017).

Inflammation is a process that involves a series of phenomena that may be due to several agents. It is usually associated with pain which is a secondary phase resulting from the release of analgesic mediators (Tsai et al., 2001). The inflammatory reaction could also be initiated by oxidative stress, which is defined as an overproduction of oxidizing molecules namely ROS. These agents induce cytokine release and pro-inflammatory enzyme activation; which are involved in the inflammatory process (Gupta et al., 2005). Indeed, inflammation, pain and oxidative stress are interrelated processes. In the present study, AEAPR and MEAPR showed an anti-inflammatory activity, therefore, both extracts of Anacyclus pyrethrum root were investigated for their antioxidant action against free radicals using diverse methods (DPPH, FRAP and BCB). The results showed that APR extracts, according to DPPH method, have a strong scavenging activity, but with the capacity to protect against lipid peroxidation (BCB test). Indeed, Anacyclus pyrethrum effectively inhibits oxidative stress, which is in accordance with previous studies (Sujith et al., 2011a; Selles et al., 2012b).

The antioxidant potential of APR extracts may be due to their phytochemical constituents. In fact, phytochemical screening has proved the occurrence of alkaloids, flavonoids, saponins, tannins, triterpenes and sterols. Detection of these phytochemicals in APR agrees with earlier findings (Sujith et al., 2011a; Selles et al., 2012a; Hanane et al., 2014). The phenolic compounds may contribute directly to the antioxidative action as has been shown by Nagendrappa (2005). Generally, flavonoids, which have powerful antioxidant activities in vitro, are able to scavenge a wide range of ROS like superoxide and nitric oxide radicals (Halliwell, 2008). In addition, as opioid receptors have been partially involved in APR extract's effects, detected alkaloids may be among the active compounds of APR extracts.

## REFERENCES


## CONCLUSION

Our study demonstrated that aqueous and methanol extracts of Anacyclus pyrethrum roots are non-toxic substances, with good central and peripheral antinociceptive effects, which is beneficial and researched in traditional medicine. The qualitative phytochemical analysis and the antioxidant activity in vitro have revealed the presence of several antioxidant phytoconstituants such as flavonoids, alkamides, saponins and tannins in Anacyclus pyrethrum root extracts.

Many mechanisms of specific action to those phytochemical compounds could be responsible for anti-inflammatory and antinociceptive activities observed in the present work. However, further studies are needed to isolate the pharmacologically active compounds and elucidate their exact molecular mechanism in the anti-inflammatory and antinociceptive process of APR.

## AUTHOR CONTRIBUTIONS

HM, MB, ZS, and SB designed the experiments; HM and OB performed the experiments, HM, SB, MB, and ACG performed the analysis of the data; HM And SB assembled the figures. HM, SB, MB, and ACG wrote and edited the manuscript. All authors validated it.

## ACKNOWLEDGMENTS

This research was supported by NEUREN Project (PIRSES-GA-2012-318997). All authors are thankful to Prof. A. OUHAMMOU (LEE, Cadi Ayyad University, Morocco) for identifying our experimental plant, to Prof. A. BENHARREF's LBCNSR laboratory and M. TAOURIRTE's laboratory of LBMC, Cadi Ayyad University, Morocco for providing all the support and equipment to achieving extracts preparation. We thank also A. REGRAGUI for his support by providing us experimental animals. Special thanks also to G. KUJOTH, Ph.D. (Senior Scientist Department of Pediatrics University of Wisconsin-Madison) for reading and editing the manuscript.




Sujith, K., Ronald Darwin, C., and Suba, V. (2011a). Antioxidant activity of ethanolic root extract of Anacyclus pyrethrum. Int. Res. J. Pharm. 2, 222–226.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Manouze, Bouchatta, Gadhi, Bennis, Sokar and Ba-M'hamed. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Antihyperalgesic Activities of Endocannabinoids in a Mouse Model of Antiretroviral-Induced Neuropathic Pain

Neha Munawar1,2, Mabayoje A. Oriowo<sup>1</sup> and Willias Masocha<sup>2</sup> \*

<sup>1</sup> Department of Pharmacology and Toxicology, Faculty of Medicine, Kuwait University, Safat, Kuwait, <sup>2</sup> Department of Pharmacology and Therapeutics, Faculty of Pharmacy, Kuwait University, Safat, Kuwait

Background: Nucleoside reverse transcriptase inhibitors (NRTIs) are the cornerstone of the antiretroviral therapy for human immunodeficiency virus/acquired immune deficiency syndrome (HIV/AIDS). However, their use is sometimes limited by the development of a painful sensory neuropathy, which does not respond well to drugs. Smoked cannabis has been reported in clinical trials to have efficacy in relieving painful HIV-associated sensory neuropathy.

#### Edited by:

Vivienne Ann Russell, University of Cape Town, South Africa

#### Reviewed by:

Regina A. Mangieri, University of Texas at Austin, USA Luigia Trabace, University of Foggia, Italy

> \*Correspondence: Willias Masocha masocha@hsc.edu.kw

#### Specialty section:

This article was submitted to Neuropharmacology, a section of the journal Frontiers in Pharmacology

Received: 26 January 2017 Accepted: 06 March 2017 Published: 20 March 2017

#### Citation:

Munawar N, Oriowo MA and Masocha W (2017) Antihyperalgesic Activities of Endocannabinoids in a Mouse Model of Antiretroviral-Induced Neuropathic Pain. Front. Pharmacol. 8:136. doi: 10.3389/fphar.2017.00136 Objectives: The aim of this study was to evaluate whether the expression of endocannabinoid system molecules is altered during NRTI-induced painful neuropathy, and also whether endocannabinoids can attenuate NRTI-induced painful neuropathy.

Methods: BALB/c mice were treated with 25 mg/kg of 2<sup>0</sup> ,30 -dideoxycytidine (ddC, zalcitabine), a NRTI, to induce thermal hyperalgesia. The expression of endocannabinoid system molecules was evaluated by real time polymerase chain reaction in the brain, spinal cord and paw skin at 6 days post ddC administration, a time point when mice had developed thermal hyperalgesia. The effects of the endocannabinoids, N-arachidonoyl ethanolamine (AEA) and 2-arachidonoyl glycerol (2-AG), the cannabinoid type 1 (CB1) receptor antagonist AM 251, CB2 receptor antagonist AM 630, and G protein-coupled receptor 55 (GPR55) antagonists ML193 and CID 16020046 on ddC-induced thermal hyperalgesia were evaluated using the hot plate test.

Results: ddC treatment resulted in thermal hyperalgesia and increased transcripts of the synthesizing enzyme Plcβ1 and decreased Daglβ in the paw skins, but not Napepld, and Daglα compared to vehicle treatment. Transcripts of the inactivating enzymes Faah and Mgll were downregulated in the brain and/or paw skin but not in the spinal cord of ddC-treated mice. Both AEA and 2-AG had antihyperalgesic effects in mice with ddCinduced thermal hyperalgesia, but had no effect in ddC-naïve mice. The antihyperalgesic activity of AEA was antagonized by AM251 and AM630, whereas the activity of 2-AG was antagonized by AM251, ML193 and CID 16020046, but not by AM630.

Conclusion: These data show that ddC induces thermal hyperalgesia, which is associated with dysregulation of the mRNA expression of some endocannabinoid

**132**

system molecules. The endocannabinoids AEA and 2-AG have antihyperalgesic activity, which is dependent on cannabinoid receptor and GPR55 activation. Thus, agonists of cannabinoid receptors and GPR55 could be useful therapeutic agents for the management of NRTI-induced painful sensory neuropathy.

Keywords: endocannabinoid, 2-arachidonoyl glycerol, anandamide, neuropathic pain, hyperalgesia, ddC, nucleoside reverse transcriptase inhibitor, antiretroviral

## INTRODUCTION

Pain is a major cause of poor quality of life among human immunodeficiency virus (HIV) infected patients (Larue et al., 1997). It can be due to the HIV infection, opportunistic infections, cancers and drug treatment (Nair et al., 2009). Pain is almost underestimated as it is subjective especially in HIV/acquired immune deficiency syndrome (AIDS) patients where the main focus is on other principal symptoms of the infection such as immunosuppression and opportunistic infections (Larue et al., 1997; Ferrari et al., 2006). About 25–50% of all pain clinic visits are due to neuropathic pain (Verma et al., 2004). Neuropathic pain is felt by around 20–40% of people with AIDS (Hitchcock et al., 2008; Ebirim and Otokwala, 2013). Some of the symptoms of neuropathic pain include hyperalgesia (an increased response to normally painful stimuli), allodynia (pain triggered by normally nonpainful stimuli, such as cloth rubbing against the skin) and spontaneous sensations such as burning, shooting, numbness, spasm and prickling (Dworkin et al., 2003; Cherry et al., 2012). It particularly affects the feet, hands and face, thus it can make the performance of day to day tasks such as cooking and other physical tasks very difficult (Hitchcock et al., 2008). This can have a serious negative impact on psychosocial well-being and overall quality of life of patients. These symptoms might also lead to discontinuation of antiretroviral therapy resulting in failure to suppress viral replication and worsening of HIV infection/AIDS.

Nucleoside reverse transcriptase inhibitors (NRTIs), in combination with other antiretroviral drugs, are effective in controlling the replication of HIV and form the backbone of most regimens used in the treatment of HIV/AIDS. However, their use is sometimes hampered by adverse effects including the development of dose-dependent painful peripheral neuropathy. Some NRTIs, such as didanosine (ddI), stavudine (d4T), and zalcitabine (ddC), have been removed/are being removed from regimens because of the development of peripheral neuropathy in about 15–30% of patients (Fichtenbaum et al., 1995; Moyle and Sadler, 1998; Dalakas, 2001). First-line medications recommended for managing neuropathic pain include amitriptyline, nortriptyline, duloxetine, venlafaxine, gabapentin, pregabalin, and 5% topical lidocaine (Dworkin et al., 2007). However, patients are not satisfied with current treatment options because of inadequate pain relief (Shlay et al., 1998; Simpson et al., 2000; Phillips et al., 2010) and the adverse side effect profiles which limit therapeutic efficacy and contribute to poor pain relief (Rahn and Hohmann, 2009). However, some HIV patients with painful neuropathy report relief after using cannabis (Woolridge et al., 2005; Phillips et al., 2010). Smoked cannabis has been reported in two randomized clinical trials to have efficacy in relieving painful HIV-associated sensory neuropathy (Abrams et al., 2007; Ellis et al., 2009). It has been observed that treatment with cannabinoid receptor agonists such as WIN 55,212-2 produced antinociception and antihyperalgesia in rodent models of HIV and NRTI-induced neuropathic pain (Wallace et al., 2007a,b). However, neither changes in the endocannabinoid system nor the effects of endocannabinoids against NRTI-induced painful neuropathy have been investigated.

Endocannabinoids and exogenous cannabinoid ligands produce their effects via two known cannabinoid receptors, CB1 and CB2 receptors (Guindon and Hohmann, 2009). Endogenous cannabinoid ligands are lipid molecules that are produced from phospholipid precursors in the cell membrane (Di Marzo et al., 1998; Piomelli, 2005) upon an "on demand" fashion (Giuffrida et al., 1999; Varma et al., 2001; Kim et al., 2002). N-arachidonoylethanolamine (anandamide, AEA) and 2-arachidonoyl-glycerol (2-AG) are the two most extensively studied endocannabinoids so far (Di Marzo and Piscitelli, 2015). AEA is a partial agonist at the CB1 and CB2 receptors (Hillard et al., 1999; Howlett et al., 2002; Chavez et al., 2010; Grueter et al., 2010). On the other hand, 2-AG behaves as a complete agonist at CB1 and 2 receptors (Stella et al., 1997; Gonsiorek et al., 2000; Savinainen et al., 2001; Howlett et al., 2002). Several pathways are involved in the synthesis of AEA and 2-AG (Di Marzo, 2008). N-acylphosphatidyl ethanolamine specific phospholipase D (NAPE-PLD) hydrolyses N-arachidonylphosphatidyl ethanolamine (NAPE) to produce AEA and phosphatidic acid (Natarajan et al., 1981; Schmid et al., 1983; Di Marzo et al., 1994; Okamoto et al., 2004). 2-AG is synthesized through the catalysis of the membrane bound phosphatidylinositol-4,5-biphosphate (PIP2) to diacylglycerol (DAG) by phospholipase C (PLC)-β1 (Farooqui et al., 1989; Ayakannu et al., 2013; Di Marzo and Piscitelli, 2015). The intermediate DAG is further catalyzed by the action of one of two diacylglycerol lipases (DAGLs), DAGL-α and DAGL-β, to 2-AG (Bisogno et al., 2003; Tanimura et al., 2010). In order to maintain endocannabinoid homeostasis, AEA and 2-AG after release in the synaptic space are either transported back into the cells by a transporter or degraded by enzymes. AEA is hydrolyzed to arachidonic acid and ethanolamine by fatty acid amide hydrolase (FAAH), an intracellular membrane-bound enzyme (Cravatt et al., 1996). 2-AG is hydrolyzed to arachidonic acid and glycerol by the enzyme monoacylglycerol lipase (MAGL) (Dinh et al., 2002; Saario et al., 2004).

The objective of this study was to evaluate whether the expression of molecules of the endocannabinoid system is

altered in the central nervous system (CNS) and the periphery of mice during NRTI-induced painful neuropathy, and also whether endocannabinoids can attenuate NRTI-induced painful neuropathy.

## MATERIALS AND METHODS

## Animals

The animals used in this study were female BALB/c mice (2–3 months old; 20–25 g) supplied by the Animal Resources Centre at the Health Sciences Centre (HSC), Kuwait University, Kuwait. The mice were kept in temperature controlled (24 ± 1 ◦C) rooms with food and water ad libitum. All experiments were performed during the same period of the day (8:00 AM to 4:00 PM) to exclude diurnal variations in pharmacological effects. The animals were handled in compliance with Directive 2010/63/EU of the European Parliament and of the Council on the protection of animals used for scientific purposes. All animal experiments were approved by the Ethical Committee for the use of Laboratory Animals in Teaching and in Research, HSC, Kuwait University.

#### Administration of 2<sup>0</sup> ,30 -Dideoxycytidine (ddC) to Induce Neuropathic Pain

2 0 ,30 -dideoxycytidine (ddC, zalcitabine) (Sigma-Aldrich, St. Louis, MO, USA) was prepared freshly in normal saline (0.9% NaCl) on the day of the experiment. ddC 25 mg/kg or its vehicle was administered to mice in a single intraperitoneal (i.p.) injection, in a volume of 10 ml/kg. This treatment regimen has been reported to produce painful neuropathy in mice (Sanna et al., 2014).

## Drug Administration

All the drugs were purchased from Tocris, Bristol, UK. N-arachidonoyl ethanolamine (AEA, anandamide), was dissolved in Tocrisolve; 2-arachidonoyl glycerol (2-AG) in normal saline containing 5% ethanol, 5% cremophor, and 5% DMSO (Guindon et al., 2011); AM 251 and AM 630 in normal saline containing 5% Tween 80 and 5% propylene glycol; ML193 and CID 16020046 in normal saline containing 5% ethanol, 5% cremophor, and 5% DMSO. The drugs and their vehicles were freshly prepared and further diluted with normal saline to lower concentrations before administration and administered i.p. to mice at a volume of 10 ml/kg body mass.

The doses of AEA and 2-AG, 1, 10, and 20 mg/kg, were chosen based on those previously shown to have antinociceptive and/or antihyperalgesic activity in mice (Mechoulam et al., 1995; Calignano et al., 2001). The drugs were administered to naïve mice and ddC-treated mice at 6 days after administration of ddC, when mice had developed thermal hyperalgesia.

To evaluate the receptors involved in the antihyperalgesic activities of the endocannabinoids CB1 (AM 251 3 mg/kg), CB2 (AM 630 3 mg/kg), and GPR55 (ML193 and CID 16020046 both at 10 mg/kg) antagonists were administered 15 min before the administration of AEA and 2-AG to mice with ddC-induced thermal hyperalgesia.

## Assessment of Thermal Nociception

Reaction latencies of mice to hot plate (Panlab SL, Barcelona, Spain) at 55 ± 1 ◦C in the form of the first sign of nociception, paw licking, flinching or jump response to avoid the heat were measured, as described before (Parvathy and Masocha, 2013), before (baseline latency), at day 6 after injection of ddC and at various times after drug treatment. A cut-off period of 20 s was maintained to avoid damage to the paws.

## Tissue Preparation and Real Time RT-PCR

Mice were anesthetized with halothane, sacrificed by decapitation on day 6 post-administration of ddC. Brains, spinal cords and paw skins were dissected, snap frozen in liquid nitrogen and kept at - 70◦C prior to mRNA extraction.

Gene transcripts of cannabinoid type 1 receptor (Cnr1), Cnr2, N-arachidonoyl ethanolamine-specific phospholipase D (Napepld), phospholipase C-beta 1 (Plcβ1), diacylglycerol lipasealpha (Daglα), Daglβ, fatty acid amide hydrolase (Faah), and monoacylglycerol lipase (Mgll) were quantified in the brains, spinal cords and paw skins of vehicle-treated and ddC-treated mice by real time polymerase chain reaction (PCR). Total RNA was extracted from the fresh frozen brains, spinal cords and paw skins using the RNeasy Kit (Qiagen GmbH), reverse-transcribed, and the mRNA levels were quantified on an ABI Prism <sup>R</sup> 7500 sequence detection system (Applied Biosystems) as previously described (Masocha, 2009). The primer sequences which were used, listed in **Table 1**, were ordered from Invitrogen (Life Technologies) and/or synthesized at the Research Core Facility (RCF), HSC, Kuwait University. Threshold cycle (Ct) values for all cDNA samples were obtained and the amount of mRNA of individual animal sample (n = 5–10 per group) was normalized to Ppia (cyclophilin A, housekeeping gene) (1Ct). The relative amount of target gene transcripts was calculated using the 2 <sup>−</sup>11Ct method as described previously (Livak and Schmittgen, 2001). These values were then used to calculate the mean and standard error of mean of the relative expression of the target gene mRNA in the brains, spinal cords and paw skins of vehicleand ddC-treated mice.

## Statistical Analyses

Statistical analyses were performed using unpaired Student's t-test (to compare the effects of ddC to those of the vehicle on mRNA expression on day 6 post ddC injection; and to compare the effects of ddC on reaction latency on day 6 post-injection and pretreatment reaction latency), one-way analysis of variance (ANOVA) followed by Bonferroni's multiple comparison posttests (to compare the effects of pretreatment with CB receptor and GPR55 antagonists before administration of either AEA or 2-AG to the antihyperalgesic effects of AEA and 2-AG in mice with ddC-induced thermal hyperalgesia at day 6 post-injection of ddC), or two-way repeated measures ANOVA followed by Bonferroni's multiple comparison post-tests (to compare the effects of ddC to vehicle on reaction latency over time; and to compare the effects of AEA and 2-AG to that of the vehicle in mice with ddC-induced thermal hyperalgesia) using GraphPad

#### TABLE 1 | Polymerase chain reaction (PCR) primer sequences of cyclophilin A and endocannabinoid system molecules.


Prism software (version 5.0). The differences were considered significant at p < 0.05. The results in the text and figures are expressed as the means ± SEM.

## RESULTS

## ddC-Induced Thermal Hyperalgesia in Female BALB/c Mice

Mice treated with ddC developed thermal hyperalgesia on day 6 after treatment, i.e., reduction in reaction latency compared to the baseline latency and vehicle-treated mice (7.3 ± 0.5 s compared to 12.7 ± 0.9 s and 12.8 ± 1.0 s, respectively; n = 18 for both vehicle- and ddC-treated mice; p < 0.001 for both comparisons; **Figure 1**). There was a significant interaction between treatment and time after treatment with ddC (F1,<sup>34</sup> = 60.72, p < 0.0001).

The mRNA expression of endocannabinoid molecules (in the brain, spinal cord, and paw skin) and the antihyperalgesic activity the endocannabinoids AEA and 2-AG were analyzed at day 6 in separate groups of mice, at this time point mice showed ddC-induced thermal hyperalgesia.

## Expression of the Endocannabinoid System Molecules mRNA during ddC-Induced Thermal Hyperalgesia in the Brain, Spinal Cord, and Paw Skin

Treatment with ddC did not affect the expression of endocannabinoid-synthesizing enzymes (Napepld, Plcβ1, Daglα,

vehicle-treated mice (Student's t-test).

and Daglβ) in the brain or the spinal cord compared to vehicle treatment. However, treatment with ddC significantly increased the transcripts of Plcβ1 (p = 0.0021), decreased the transcripts of Daglβ (p = 0.0406), but did not significantly affect the expression of Daglα and Napepld (p > 0.05) in the paw skin compared to vehicle treatment (**Figure 2A**).

Treatment with ddC did not affect the expression of endocannabinoid-inactivating enzymes (Faah and Mgll) in the spinal cord compared to vehicle treatment. However, treatment with ddC significantly decreased the transcripts of Faah (p = 0.0055), but did not significantly affect the expression of Mgll (p > 0.05) in the brain compared to vehicle treatment. In the paw skin, treatment with ddC significantly decreased the transcripts of Mgll (p = 0.0275), but did not significantly affect the expression of Faah (p > 0.05; **Figure 2B**).

The expression of the cannabinoid receptors Cnr1 and Cnr2 were not significantly modulated by treatment with ddC in all the three tissues analyzed, brain, spinal cord and paw skin, compared to vehicle treatment (p > 0.05; **Figure 2C**).

## Effects of Treatment with the Endocannabinoids AEA and 2-AG on Naïve Mice and Mice with ddC-Induced Thermal Hyperalgesia

Mice with ddC-induced thermal hyperalgesia and naïve mice were treated with 1, 10, and 20 mg/kg of the endocannabinoids AEA and 2-AG.

The intraperitoneal administration of vehicle did not change the reaction latency to thermal stimuli in mice with ddCinduced thermal hyperalgesia compared to before administration at day 6 (p > 0.05; **Figures 3A,B**). However, all the doses (1, 10, and 20 mg/kg) of AEA and 2-AG administered produced significant increase in reaction latency in mice with ddCinduced thermal hyperalgesia at all time points from 10 to 70 min post-drug administration, when the experiment was terminated, compared to mice treated with vehicle and before endocannabinoid administration at day 6 (p < 0.01; **Figures 3A,B**). There was a significant interaction between treatment and time after treatment for AEA doses of 1 mg/kg (F7,<sup>98</sup> = 50.22, p < 0.0001), 10 mg/kg (F7,<sup>98</sup> = 73.17, p < 0.0001), and 20 mg/kg (F7,<sup>98</sup> = 48.48, p < 0.0001) in mice with ddC-induced thermal hyperalgesia. There was also a significant interaction between treatment and time after treatment for 2- AG doses of 1 mg/kg (F7,<sup>98</sup> = 662.69, p < 0.0001), 10 mg/kg (F7,<sup>98</sup> = 218.53, p < 0.0001) and 20 mg/kg (F7,<sup>98</sup> = 1230.23, p < 0.0001) in mice with ddC-induced thermal hyperalgesia.

The intraperitoneal administration of AEA or 2-AG did not change the reaction latency to thermal stimuli in naïve mice (without ddC treatment) at any time point compared to mice treated with vehicle and before endocannabinoid administration (p > 0.05; **Figures 3C,D**). There was no significant interaction between treatment and time after treatment for AEA doses of 1 mg/kg (F7,<sup>105</sup> = 0.55, p = 0.7928) and 10 mg/kg (F7,<sup>105</sup> = 0.48, p = 0.8456), but at 20 mg/kg (F7,<sup>105</sup> = 2.61, p = 0.0155) in mice with ddC-induced thermal hyperalgesia. There was also no significant interaction between treatment and time after treatment for 2-AG doses of 1 mg/kg (F7,<sup>98</sup> = 0.79, p = 0.5996), and 20 mg/kg (F7,<sup>98</sup> = 1.07, p = 0.3871) in naïve mice.

## Effects of Cannabinoid Receptor and GPR55 Antagonists on the Antihyperalgesic Activities of the Endocannabinoids AEA and 2-AG in Mice with ddC-Induced Thermal Hyperalgesia

The administration of the CB1 receptor antagonist AM 251 (3 mg/kg) or the CB2 antagonist AM 630 (3 mg/kg) to mice with ddC-induced thermal hyperalgesia did not alter the reaction latency to hot-plate test compared to vehicle-treated mice at 30 min post-treatment (p > 0.05; **Figure 4**).

The CB1 receptor antagonist AM 251 significantly antagonized the antihyperalgesic effect of both AEA and 2- AG, i.e., a 38% reduction in reaction latency to AEA, 13.1 ± 0.1 s for AEA alone compared to 8.2 ± 0.2 s for AEA + AM 251 and 33% reduction in reaction latency to 2-AG, 11.7 ± 0.3 s for 2-AG compared to 7.9 ± 0.3 s for 2-AG + AM 251 (p < 0.01; **Figures 4A,B**).

hot-plate test. Reaction latency times of naïve mice at different times after treatment with (C) AEA (1, 10, and 20 mg/kg) or its vehicle, and (D) 2-AG (1 and 20 mg/kg) or its vehicle in a hot-plate test. Each bar represents the mean ± SEM of values obtained from 8 to 9 animals. ∗∗p < 0.01 compared to drug vehicle at the same time point after treatment (two-way repeated measures ANOVA followed by Bonferroni's Multiple Comparison Test).

The CB2 receptor antagonist AM 630 significantly antagonized the antihyperalgesic effect of AEA but not 2- AG, i.e., a 11% reduction in reaction latency to AEA, 13.1 ± 0.1 s for AEA alone compared to 11.7 ± 0.3 s for AEA + AM 630 (p < 0.01; **Figure 4A**) and no difference in reaction latency to 2-AG, 11.7 ± 0.3 s for 2-AG compared to 12.0 ± 0.4 s for 2-AG + AM 630 (p > 0.05; **Figure 4B**).

Since the antihyperalgesic activity of 2-AG was partially antagonized by the CB1 receptor antagonist but was not affected by the CB2 receptor antagonist, and 2-AG has been reported to activate GPR55 (Ryberg et al., 2007), the effects of two GPR55 antagonists CID 16020046 and ML193 on the activities of 2-AG were also evaluated. The administration of the GPR55 antagonists CID 16020046 (10 mg/kg) and ML193 (10 mg/kg) to mice with ddC-induced thermal hyperalgesia did not alter the reaction latency to hot-plate test compared to vehicle-treated mice at 30 min post-treatment (p > 0.05; **Figure 4C**). The GPR55 antagonists CID 16020046 and ML193 significantly antagonized the antihyperalgesic effect of 2-AG, i.e., a 48% reduction in reaction latency to 2-AG caused by pretreatment with CID 16020046, 11.2 ± 0.1 s for 2-AG alone compared to 5.8 ± 0.04 s for 2-AG + CID 16020046 and 49% reduction in reaction latency to 2-AG caused by pretreatment with ML193, 11.2 ± 0.1 s for 2- AG alone compared to 5.7 ± 0.1 s for 2-AG + ML193 (p < 0.01; **Figure 4C**).

## DISCUSSION

This study presents the first data on antihyperalgesic activities of endocannabinoids and the expression of endocannabinoid molecules mRNA in the CNS and periphery during NRTIinduced thermal hyperalgesia. Mice with ddC-induced thermal hyperalgesia had altered mRNA expression of endocannabinoidsynthesizing enzymes Plcβ1 and Daglβ in the paw skin, and endocannabinoid-inactivating enzymes Faah and Mgll in the brain and paw skin, respectively. The endocannabinoids AEA and 2-AG had antihyperalgesic activity against ddC-induced thermal hyperalgesia but had no activity in naïve mice. The antihyperalgesic activity of AEA was dependent on activation of both CB1 and CB2 receptors, whereas that of 2-AG was dependent on CB1 receptor and GPR55, but not CB2 receptor.

Changes in endocannabinoid expression have been found in various models of neuropathic pain (Jhaveri et al., 2007). In the

periphery, AEA has been found to be increased in the paw skin and dorsal root ganglia (DRG) of rats with spinal nerve ligation (SNL)-induced painful neuropathy (Mitrirattanakul et al., 2006; Jhaveri et al., 2007). 2-AG has been found to be increased in the DRG of rats with SNL-induced painful neuropathy (Mitrirattanakul et al., 2006). In the CNS, AEA, and/or 2-AG have been reported to be increased in different areas of the brain and spinal cord of rats with neuropathic pain induced by chronic constriction injury (CCI) of the sciatic nerve (Palazzo et al., 2006; Petrosino et al., 2007). In an animal model of chemotherapyinduced neuropathic pain (CINP) AEA and 2-AG were increased in the spinal cord, whilst 2-AG, but not AEA, was decreased in the paw skin (Guindon et al., 2013). Endocannabinoid molecules, such as AEA and 2-AG, are synthesized in an on demand fashion (Luchicchi and Pistis, 2012). Thus, changes in the enzymes that synthesize or degrade them would have a significant effect on the amount of endocannabinoids available when needed. There were no changes in the expression of mRNA of Napepld, the main AEA synthesizing enzyme, in DRGs or in the spinal cord of rats with CCI-induced neuropathic pain (Malek et al., 2014). However, NAPE-PLD immunoreactivity was decreased in the DRGs of rats with SNL (Sousa-Valente et al., 2016). In the current study, ddC-induced neuropathic pain did not significantly affect the level of the mRNA of Napepld in the CNS or periphery, similar to what has been found in the CCI model (Malek et al., 2014). However, there was an increase in the mRNA expression of Plcβ1, an enzyme involved in the synthesis of DAG, an intermediate in the synthesis of 2-AG, in mice with ddC-induced neuropathic pain in the periphery but not in the CNS. On the other hand, there were no changes in the expression of mRNA of Plcβ1 mRNA in DRGs or in the spinal cord of rats with CCI-induced neuropathic pain (Malek et al., 2014). The protein levels of DAGL-α, one of the main enzymes involved in the synthesis of 2-AG from DAG, were decreased in the spinal cord of mice with diabetic neuropathic pain (Ikeda et al., 2013). In the current study, the mRNA expression of Daglβ, but not Daglα, was decreased in the paw skin but not spinal cord or brain of mice with ddCinduced neuropathic pain. The mRNA expression of Faah, the main enzyme in the degradation of AEA, was increased in the spinal cord but not in the DRGs (Malek et al., 2014). Similarly, in a rat model of CINP Faah was increased in the spinal cord but not in the paw skin (Guindon et al., 2013). FAAH immunoreactivity was increased in the DRGs of rats with SNL (Sousa-Valente et al., 2016). In the rostroventromedial medulla (RVM) area of the brain of rats with diabetic neuropathy FAAH protein levels were increased (Silva et al., 2016). In contrast, mRNA expression of Faah was decreased in the brain but not the spinal cord, and showed a tendency toward decrease in the paw skin of mice with ddC-induced thermal hyperalgesia. There were no changes in the mRNA expression of Mgll, the main enzyme in the degradation of 2-AG, in the spinal cord or paw skins of rats with CINP (Guindon et al., 2013). In contrast mRNA expression of Mgll was decreased in the paw skin, but not spinal cord and brain, of mice with ddC-induced thermal hyperalgesia. The findings of the current study show a unique characteristic of a decrease in the mRNA expression of the enzymes that inactivate AEA and 2-AG, which might result in low levels of the enzymes and increased endocannabinoids.

N-arachidonoyl ethanolamine and 2-AG are important endogenous inhibitors of nociception (Rodriguez de Fonseca et al., 2005; Guindon and Hohmann, 2009). Exogenous administration locally in the paw skin of both AEA and 2-AG in rats with partial sciatic nerve ligation (PNL)- or diabetes-induced neuropathic pain and intrathecally into the spinal cord of rats with CCI-induced neuropathic pain have produced antihyperalgesic activities (Guindon and Beaulieu, 2006; Desroches et al., 2008; Schreiber et al., 2012; Starowicz et al., 2012; Desroches et al., 2014). In the current study, systemic administration (i.p.) of both AEA and 2-AG produced antihyperalgesic activities and abrogated ddC-induced thermal hyperalgesia, but had no activity in naïve mice (mice without neuropathic pain). Calignano et al. (1998) observed that

locally administered AEA had antinociceptive effects, whereas intraperitoneally administered AEA had no effect on nociception in naïve mice. However, in another study they observed that intraperitoneally administered AEA had antinociceptive effects that lasted up to 20 min (Calignano et al., 2001). In another study, AEA administered intraperitoneally did not produce antinociceptive effects in rats (Costa et al., 1999). Our findings of antihyperalgesic activity of AEA and 2-AG administered intraperitoneally, that lasted up to 70 min when the experiment was terminated, in mice with ddC-induced thermal hyperalgesia but not in naïve mice suggest that possibly the reduction in the inactivating enzymes FAAH and MAGL in mice with ddC-induced thermal hyperalgesia contributed to the prolonged antihyperalgesic activity of the endocannabinoids. This is in line with a study that showed that administration of AEA to mice that lack FAAH (FAAH−/<sup>−</sup> mice) produced antinociception but AEA did not have antinociceptive effects in wild-type (FAAH+/<sup>+</sup> mice) (Cravatt et al., 2001).

N-arachidonoyl ethanolamine and 2-AG produce their effects via two known cannabinoid receptors, CB1 and CB2 receptors (Guindon and Hohmann, 2009). AEA has been reported to produce its antinociceptive effects via activation of both CB1 and CB2 receptors (Schreiber et al., 2012). However, other studies have shown that AEA produces its antinociceptive effects via activation of CB1 receptors but not CB2 receptors (Calignano et al., 1998; Guindon and Beaulieu, 2006). Pre-treatment with either a CB1 or CB2 antagonist antagonized the antihyperalgesic activity of AEA in mice with ddC-induced thermal hyperalgesia. The antagonism was much more with the CB1 antagonist than the CB2 antagonist, thus suggesting that AEA had antihyperalgesic activity mainly via CB1 receptors although CB2 receptors also had a minor role. Pre-treatment with a CB1 antagonist, but not a CB2 antagonist, antagonized the antihyperalgesic activity of 2-AG in mice with ddCinduced thermal hyperalgesia. Thus, suggesting that 2-AG had antihyperalgesic activity via CB1 receptors but not CB2 receptors. This is in contrast with a study done in rats with PNL-induced neuropathic pain where both CB1 and CB2 antagonists equally inhibited the antihyperalgesic effects of 2-AG administered locally, subcutaneously in the dorsal surface of the hind paw (Desroches et al., 2008, 2014). The current study shows that CB1 receptors play a more important role than CB2 receptors in the antihyperalgesic effects of both endocannabinoids AEA and 2-AG against ddC-induced hyperalgesia.

Since the antihyperalgesic activity of 2-AG was partially antagonized by the CB1 receptor antagonist but was not affected by the CB2 receptor antagonist, and 2-AG has been reported to activate GPR55 (Ryberg et al., 2007), the effects of two GPR55 antagonists CID 16020046 and ML193 were also evaluated. We used two different antagonists of the same receptors because we did not find any reports of GPR55 receptor antagonists on the antinociceptive or antihyperalgesic activity of 2-AG. Pre-treatment with either GPR55 antagonist inhibited the antihyperalgesic activity of 2-AG in mice with ddC-induced thermal hyperalgesia much more than the CB1 receptor antagonist. Loss of GPR55 (GPR55−/<sup>−</sup> mice) resulted in increased sensitivity to thermal nociception, thermal hyperalgesia (Staton et al., 2008; Wu et al., 2013; Bjursell et al., 2016). However, GPR55−/<sup>−</sup> mice did not develop PNL-induced hyperalgesia, whereas the wild-type mice did, suggesting that GPR55 has a pro-nociceptive activity (Staton et al., 2008). Lysophosphatidylinositol (LPI), which is considered the main endogenous agonist of GPR55 (Oka et al., 2007; Henstridge et al., 2009), has pro-nociceptive activities (Deliu et al., 2015). Thus, our results show that 2-AG, in contrast to LPI, produces antinociceptive activity via activation of GPR55 receptors. Possibly 2-AG and LPI bind to and activate GPR55 at different sites or in different ways. This possibility warrants further research.

## CONCLUSION

Our results show that ddC induces thermal hyperalgesia that is associated with dysregulation of the mRNA expression of endocannabinoid molecules, more importantly downregulation of Mgll and Faah, which are involved in the inactivation of the endocannabinoids AEA and 2-AG. The endocannabinoids AEA and 2-AG had antihyperalgesic activity against ddC-induced thermal hyperalgesia, but had no activity in naïve mice, possibly due to the reduced mRNA expression of Mgll and Faah in mice with ddC-induced thermal hyperalgesia. The antihyperalgesic activity of AEA was dependent on activation of both CB1 and CB2 receptors, whereas that of 2-AG was dependent on CB1 receptors and GPR55, but not CB2 receptors. FAAH and MAGL inhibitors might not be very useful in the treatment of NRTI-induced neuropathic pain, since these two enzymes are reduced during ddC-induced thermal hyperalgesia. However, stable formulations of AEA and 2-AG or agonists of both CB1 receptors and GPR55, with activities similar to AEA and 2-AG, might be useful in the treatment of NRTI-induced neuropathic pain.

## AUTHOR CONTRIBUTIONS

NM: performed the experiments, analyzed the data, wrote the paper; MO: contributed reagents/materials/analysis tools, analyzed the data, wrote the paper; WM: conceived and designed the experiments, contributed reagents/materials/analysis tools, analyzed the data, wrote the paper; all authors read and approved the final manuscript.

## FUNDING

This work was supported by grants YM02/15 and SRUL02/13 from Kuwait University Research Sector.

## ACKNOWLEDGMENT

We are grateful to Dr. Subramanian S. Parvathy for her technical assistance and to the staff from the Animal Resources Centre, HSC, Kuwait University for their support.

## REFERENCES

fphar-08-00136 March 16, 2017 Time: 15:1 # 9




role in motor coordination. PLoS ONE 8:e60314. doi: 10.1371/journal.pone. 0060314

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Munawar, Oriowo and Masocha. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Neural Damage in Experimental Trypanosoma brucei gambiense Infection: The Suprachiasmatic Nucleus

#### Chiara Tesoriero<sup>1</sup> , Yuan-Zhong Xu1† , Dieudonné Mumba Ngoyi <sup>2</sup> and Marina Bentivoglio1,3 \*

<sup>1</sup>Department of Neuroscience, Biomedicine and Movement Sciences, University of Verona, Verona, Italy, <sup>2</sup> Institut National de Recherche Biomedicale (INRB), Kinshasa, Democratic Republic of Congo, <sup>3</sup>National Institute of Neuroscience (INN), Verona Unit, Verona, Italy

#### Edited by:

Nouria Lakhdar-Ghazal, Mohammed V University, Morocco

#### Reviewed by:

Arshad M. Khan, University of Texas at El Paso, United States Paul Manger, University of the Witwatersrand, South Africa

> \*Correspondence: Marina Bentivoglio marina.bentivoglio@univr.it

#### †Present address:

Yuan-Zhong Xu, Brown Foundation Institute of Molecular Medicine, University of Texas Health Science Center, Houston, TX, United States

Received: 09 November 2017 Accepted: 12 January 2018 Published: 13 February 2018

#### Citation:

Tesoriero C, Xu Y-Z, Mumba Ngoyi D and Bentivoglio M (2018) Neural Damage in Experimental Trypanosoma brucei gambiense Infection: The Suprachiasmatic Nucleus. Front. Neuroanat. 12:6. doi: 10.3389/fnana.2018.00006 Trypanosoma brucei (T. b.) gambiense is the parasite subspecies responsible for most reported cases of human African trypanosomiasis (HAT) or sleeping sickness. This severe infection leads to characteristic disruption of the sleep-wake cycle, recalling attention on the circadian timing system. Most animal models of the disease have been hitherto based on infection of laboratory rodents with the T. b. brucei subspecies, which is not infectious to humans. In these animal models, functional, rather than structural, alterations of the master circadian pacemaker, the hypothalamic suprachiasmatic nucleus (SCN), have been reported. Information on the SCN after infection with the human pathogenic T. b. gambiense is instead lacking. The present study was aimed at the examination of the SCN after T. b. gambiense infection of a susceptible rodent, the multimammate mouse, Mastomys natalensis, compared with T. b. brucei infection of the same host species. The animals were examined at 4 and 8 weeks post-infection, when parasites (T. b. gambiense or T. b. brucei) were detected in the brain parenchyma, indicating that the disease was in the encephalitic stage. Neuron and astrocyte changes were examined with Nissl staining, immunophenotyping and quantitative analyses. Interestingly, significant neuronal loss (about 30% reduction) was documented in the SCN during the progression of T. b. gambiense infection. No significant neuronal density changes were found in the SCN of T. b. brucei-infected animals. Neuronal cell counts in the hippocampal dentate gyrus of T. b. gambiense-infected M. natalensis did not point out significant changes, indicating that no widespread neuron loss had occurred in the brain. Marked activation of astrocytes was detected in the SCN after both T. b. gambiense and T. b. brucei infections. Altogether the findings reveal that neurons of the biological clock are highly susceptible to the infection caused by human pathogenic African trypanosomes, which have the capacity to cause permanent partial damage of this structure.

Keywords: human African trypanosomiasis, sleeping sickness, circadian rhythms, biological clock, astrocytes, neurodegeneration, arginine-vasopressin, vasoactive intestinal polypeptide

## INTRODUCTION

Human African trypanosomiasis (HAT), also known as sleeping sickness, is a parasitic disease still threatening an estimated population of 65 million people in 36 countries of sub-Saharan Africa (WHO, 2017). The disease is caused by the extracellular protozoan parasites Trypanosoma brucei (T. b.), transmitted to humans through bites of tsetse flies (genus Glossina). After an epidemic in the 1990s, the number of reported cases has considerably declined due to sustained control programs, and the disease is targeted by WHO for elimination. About 2800 new cases of HAT have been recorded in 2015 (WHO, 2017). Concerns, however, are raised by underreporting and the identification of asymptomatic carriers (Franco et al., 2014; Büscher et al., 2017).

The disease occurs in humans in two forms: an acute form caused by T. b. rhodesiense and a chronic form caused by T. b. gambiense, both considered fatal if left untreated (Büscher et al., 2017). T. b. gambiense infections are currently responsible for 97% of reported HAT cases (WHO, 2017). The disease evolves in two stages (Kennedy, 2013; Büscher et al., 2017). The first, hemolymphatic stage, in which the blood and peripheral tissues are infected, progresses to a meningoencephalitic stage when trypanosomes invade the central nervous system. The infection leads insidiously to a complex neuropsychiatric syndrome including characteristic disturbances of the sleep-wake cycle with daytime somnolence and nocturnal insomnia, and alterations of the structure of sleep (Buguet et al., 2001, 2014), documented also in rodent models (Darsaud et al., 2003; Seke Etet et al., 2012; Laperchia et al., 2016, 2017). Most experimental studies on the brain in African trypanosomiasis have been based up to now, for obvious safety reasons, on the use of the T. b. brucei subspecies, which is a livestock pathogen not infectious to humans.

In mammals, the sleep-wake cycle represents a main endogenous biological rhythm driven by the suprachiasmatic nucleus (SCN), the master circadian pacemaker, located in the anterior ventral hypothalamus (van Esseveldt et al., 2000; Antle and Silver, 2005; Morin and Allen, 2006). Functional changes in the absence of overt structural alterations have been reported in the SCN of T. b. brucei-infected laboratory rats (Peng et al., 1994; Lundkvist et al., 1998, 2002). No studies, however, have been hitherto performed on the SCN of T. b. gambiense-infected hosts.

The aim of the present study was to examine the SCN after infection with the human pathogen T. b. gambiense. For this purpose, the multimammate mouse, Mastomys natalensis, was used as an animal model. M. natalensis, the most widespread rodent in sub-Saharan Africa (Coetzee, 1975), is a sensitive recipient of T. b. gambiense infection (Mehlitz, 1978; Büscher et al., 2005). Laboratory rats and mouse strains show little or no susceptibility to most T. b. gambiense isolates (Giroud et al., 2009), as also shown by early histopathological studies on the brain infection in mice, in which very few extravascular parasites were detected in the brain parenchyma (Van Marck et al., 1981), or the infection had a very long duration and succeeded only in a proportion of animals (Poltera et al., 1982). The SCN was also here investigated in M. natalensis infected with T. b. brucei for comparison with T. b. gambiense infection.

A wealth of data indicates that the SCN is composed anatomically and functionally by neuronal subpopulations on the basis of chemoarchitectonic criteria and circuitry organization (Van den Pol, 1980; Abrahamson and Moore, 2001; Antle and Silver, 2005; Morin and Allen, 2006; Moore, 2013; Hastings et al., 2014). Two neuropeptides which characterize main neuronal populations in the SCN are represented by vasoactive intestinal polypeptide (VIP) and arginine-vasopressin (AVP). VIP is expressed by neurons densely aggregated in the ventrolateral (VL) portion of the rodent SCN, which is also named ''core'' in the classical partitioning of the nucleus (Moore, 2013) and is the main target of retinal fibers in the SCN. AVP is mainly expressed by neurons located in the dorsomedial (DM or ''shell'') portion of the nucleus, to which VIP neurons project in the intrinsic circuitry of the SCN and which gives origin to the SCN output. Communication between these two main subregions is effected by neural pathways and paracrine signaling (Hastings et al., 2014). Astrocytes are densely distributed in the SCN, where they represent a prominent and functionally important cell population (Morin et al., 1989; Becquet et al., 2008; Marpegan et al., 2011; Ng et al., 2011; Brancaccio et al., 2017), which can act as mediator of immune signals in the SCN (Leone et al., 2006). On this basis, the distribution of AVP and VIP neurons and astrocytes was here investigated in the SCN of M. natalensis, and neuronal density and astrocyte activation were evaluated after African trypanosome infection.

## MATERIALS AND METHODS

## Infection and Tissue Processing

Adult M. natalensis of 3–4 or 5–6 months of age (lifespan in laboratory conditions: 13–15 months; Coetzee, 1975), both males and females, of 15–35 g body weight (for T. b. brucei infection) or 45–65 g body weight (for T. b. gambiense infection) at the beginning of the experiments were obtained by the local breeding colony, established in the animal facility of the Institut National de Recherche Biomedicale (INRB), Kinshasa, Democratic Republic of Congo (DRC). The animals were maintained under a 12 h:12 h light/dark cycle, with food and water ad libitum. Under approval of the ethical committee of the Ministry of Health of DRC, experimental procedures were performed in strict adherence to the European Communities Council (86/609/EEC) directives and the ARRIVE guidelines. All efforts were made to minimize animal number and suffering.

Two subspecies of trypanosomes were obtained from the cryobank collections of INRB: T. b. gambiense (originally isolated from a patient in DRC in 2006 and identified as T. b. gambiense MHOM/INRB/2006/11A), and T. b. brucei (AnTat 1.1E, a pleiomorphic clone isolated in 1966 from the blood of Tralephagus scriptus in Uganda). The cryostabilates were re-thawed in a water bath at 37◦C and viability was assessed with the matching method (Herbert and Lumsden, 1976) before use. After dilution in 0.1 M phosphate buffer, pH 7.4, supplemented with glucose to achieve a concentration of 106.9–107.2 trypanosomes/ml, the inoculation was done by intraperitoneal injection of 0.25 ml per animal. The infected animals were monitored weekly. At each examination, blood samples were obtained from the tail tip to verify parasitaemia and body weight was recorded. Uninoculated animals, kept under the same conditions, were used as controls (n = 4 matched with either T. b. subspecies).

Previous disease monitoring at INRB has shown that T. b. gambiense infection of M. natalensis lasts about 4 months, and T. b. brucei infection about 2 months. On this basis, a survival time of 4 or 8 weeks was here adopted. At the time of sacrifice, no consistent body weight loss was found in the infected animals, which showed, however, high interindividual variability. The animals (n = 3 or 4 per time point and parasite subspecies) were sacrificed, during daytime (M. natalensis is a nocturnal rodent; Coetzee, 1975), under anesthesia, by transcardial perfusion with ice-cold saline followed by freshly prepared 4% paraformaldehyde in PB. The brains were immediately removed, postfixed for a few hours, and stored until sectioning at 4◦C in 0.01 M phosphate buffered-saline, pH 7.4 (PBS) containing 0.1% sodium azide. Following cryoprotection in 30% sucrose in PBS, the brains were sectioned coronally at a thickness of 30 µm with a freezing microtome. All sections through the SCN were collected in adjacent series of one every sixth section.

One series of sections from each animal was stained with cresyl violet for cytoarchitectural observation and cell counts. These sections were mounted onto gelatinized slides, air dried, and stained with 0.1% cresyl violet for 5–10 min. They were then dehydrated in a graded series of ascending ethanol concentrations, cleared with xylene, and coverslipped. The Nisslstained series of sections was immediately adjacent to the series processed for the visualization of the neuropeptides AVP and VIP in the SCN with immunofluorescence (see below), in order to complement the cytoarchitectonic observations with chemoarchitectonic criteria.

## Immunocytochemistry

Neurons in the SCN were labeled using NeuN immunoperoxidase, as well as AVP and VIP immunofluorescence; glial fibrillary acidic protein (GFAP) was used as marker of astrocytes in immunoperoxidase and immunofluorescence. In all procedures, sections from control and infected brains were processed in parallel and in the same solutions. In preliminary experiments, attempts to stain microglia in sections from M. natalensis brains turned out to be unsuccessful, and microglia could not, therefore, be investigated.

For the study of neurons and astrocytes with immunoperoxidase, two series of sections from each animal were first preincubated in 1% H2O<sup>2</sup> in PBS for 20 min to inactivate endogenous peroxidase activity, and then in a solution of 5% bovine serum albumin, 0.3% Triton X-100 in PBS for 1 h at room temperature. Subsequently, the sections were incubated overnight at 4◦C with mouse monoclonal anti-NeuN antibodies (1:500, Chemicon, Temecula, CA, USA) or rabbit polyclonal anti-GFAP (1:500; Dako, Glostrup, Denmark) diluted in 0.2% Triton X-100 and 1% bovine serum albumin in PBS. After thorough rinsing, the sections were incubated for 2 h at room temperature in biotinylated secondary goat anti-rabbit or goat anti-mouse IgGs (1:200, Vector, Burlingame, CA, USA) in the solution used for primary antibody dilution. The sections were finally reacted with avidin-peroxidase complex (1:100, Vector) for 1 h and visualized using 3<sup>0</sup> ,3-diaminobenzidine as chromogen. After rinsing in PBS, the sections were mounted onto gelatin-coated slides, air-dried, dehydrated, cleared and coverslipped. Specific immunostaining was absent in control sections in which the primary antibody was omitted.

Two adjacent series of sections from each animal were processed for double immunofluorescence to examine AVP or VIP together with GFAP. Briefly, following the blocking procedure in 5% normal donkey serum and 0.3% Triton X-100 in PBS, these sections were incubated overnight at 4◦C in a mixture of mouse monoclonal anti-GFAP (1:500; Chemicon) antibodies, and rabbit polyclonal anti-AVP (1:1000; Phoenix, Burlingame, CA, USA) or anti-VIP (1:500, Santa Cruz Biotechnology, Santa Cruz, CA, USA) antibodies. Immunopositivity was visualized with species-specific secondary antibodies raised in donkey and conjugated with Cy2 or Cy3 (1:100; Jackson ImmunoResearch, Suffolk, UK).

Brain sections were also used to ascertain the presence of parasites in the parenchyma. These sections were processed for double immunofluorescence using a mixture of rabbit polyclonal antibodies anti-variant surface glycoprotein of the AnTat 1:1E stabilate (1:200; kindly provided by Philippe Büscher, Institute of Tropical Medicine, Antwerp, Belgium) to label trypanosomes, and goat polyclonal anti-glucose transporter-1 antibodies (1:100, Santa Cruz Biotechnology) to label blood vessel walls (Pardridge et al., 1990). Species-specific secondary antibodies were used as above.

The material processed for immunofluorescence was observed with a confocal laser scanning microscope (Zeiss, LSM 510) equipped with an argon laser emitting at 488 nm (Cy2) and a helium/neon laser emitting at 543 nm (Cy3).

## Data Analysis

All quantitative analyses were performed blindly of the experimental group assignment. Three animals per experimental group were used for all analyses.

## Stereological Neuron Cell Counts

The borders of the SCN were delineated in each section using anatomical landmarks (optic chiasm and third ventricle) and cytoarchitectonic criteria. A previously validated stereological method (Abercrombie, 1946; Tsukahara et al., 2005) with modifications was used to estimate the number of neurons in either the whole SCN or its DM and VL subdivisions. Three equally spaced cresyl violet-stained or NeuN-immunostained sections through the rostrocaudal extent of the SCN were used. In cresyl-violet-stained sections, cell counts were performed in regions of interest (ROIs) randomly sampled using a 40× objective and a rectangular frame of 120 × 160 µm in the DM and VL portions (two ROIs per subregion per section, on each side of the SCN). The ROIs were consistently placed close to the ventral border of the SCN, near the optic chiasm and in the center of the dorsal portion of the SCN; the counting unit was a cell body containing a nucleus with a clearly visible nucleolus, discarding glial cells identified as smaller elements without nucleoli. In the NeuN-immunostained sections, the entire SCN area of dense immunoreactivity was used as ROI, and the counting unit was the immunoreactive cell nucleus, considering both faintly and intensely stained neurons. In each animal, the major axis of the neuronal nuclei was measured and averaged for the Abercrombie-based correction. According to this formula, the mean neuronal cell number per ROI was corrected by T/T+h, where T = section thickness (30 µm) and h = mean major axis of the neuronal nucleus (Liang et al., 2012).

The number of neurons was also counted in NeuN-immunostained sections through the hippocampal dentate gyrus (DG) of control and T. b. gambiense-infected M. natalensis at 8 weeks post-infection. This area was chosen on the basis of the key role of the hippocampus in the interaction between circadian rhythmicity and cognition (Kyriacou and Hastings, 2010), and the report of altered clock gene expression in the DG as long-term effect of acute sepsis (O'Callaghan et al., 2012). In each animal, three equally spaced sections through the middle portion of the hippocampus were sampled at corresponding rostrocaudal levels. Bilaterally, three ROIs (each of the same size used for cell counts in Nissl-stained sections of the SCN) were randomly sampled in the granule cell layer of the DG and the number of NeuN-immunostained neuronal nuclei was evaluated as above.

## Densitometric Analysis of GFAP Immunosignal Intensity

For quantitative evaluation the GFAP immunoreactivity in the SCN, three equally spaced sections per animal through the middle portion of the SCN were used. The analysis was pursued at 4 weeks after T. b. brucei infection, as well as at 4 and 8 weeks after T. b. gambiense infection, vs. matched controls. Under constant calibrated parameters, images were taken using a digital camera (QImaging, Surrey, BC, Canada) connected to the microscope (objective 40×, NA 0.75). Analysis of digitized images was aided by the software Image Pro Plus 5.0 for Windows (Media Cybermetics, Silvers Springs, MD, USA). In each image derived from the DM and VL subregions, the values of GFAP immunosignal intensity were measured in a square ROI (100 × 100 µm). The mean optical density (OD) value from each ROI was normalized against the background, defined as the signal measured in an area devoid of specific immunostaining. Two ROIs per SCN subregion were evaluated in each section. The value was expressed as grand mean per SCN computed from the three mean values obtained from each section.

#### Statistics

Data are given as means ± standard error of the mean (SEM) in the entire SCN, and in the DM or VL portion, respectively. Neuronal density (mean number of neurons/ROI), relative neuronal density (neuronal density in infected animals vs. controls, setting controls as 1), and OD values of GFAP immunosignal were evaluated. Statistical evaluation was performed with the Student t-test or two way analysis of variance (ANOVA) followed by the Bonferroni's post hoc test, as appropriate, in each series from T. b. brucei or T. b. gambiense experimental groups. Significance threshold was set at P < 0.05. The statistical analysis was performed using GraphPad.

## RESULTS

## Trypanosome Neuroinvasion

Numerous trypanosomes of either subspecies were observed in the brain of M. natalensis at 4 and 8 weeks post-infection (**Figure 1**). As demonstrated by the visualization of parasites and blood vessel walls in double immunofluorescence (**Figure 1**), many trypanosomes had invaded the parenchyma. T. b. gambiense parasites appeared more numerous than T. b. brucei in the host brain parenchyma, especially at 8 weeks post-infection, though no detailed analysis of parasite load was performed.

## The SCN of M. natalensis

According to taxonomic criteria, the multimammate mouse Mastomys ''holds an intermediate position between the house mouse and the ship (roof) rat'' (quoted from Coetzee, 1975). In Nissl-stained sections, the cytoarchitectural features of the SCN of M. natalensis appeared similar to previous descriptions in the laboratory mouse and rat (Abrahamson and Moore, 2001; Moore et al., 2002). Located dorsal to the optic chiasm on either side of the third ventricle in the anterior hypothalamus, the SCN appeared as a compact, bilateral cell aggregates, with an oval shape in the coronal plane, especially in its middle third (**Figures 2A–D**). The boundaries of the SCN, clearly delimited ventrally by the optic chiasm, were distinguished also medially, laterally and dorsally on the basis of the high cell packing density with respect to the surrounding hypothalamic regions (**Figure 2A**).

Concerning the chemoarchitectural organization of the two peptidergic cell populations here examined, a relatively extended portion of the SCN, from its rostral pole to the caudal end, was filled by AVP-immunoreactive neurons, whose packing density decreased along the dorsoventral axis leaving unstained the most ventral portion of the nucleus (**Figure 2E**). VIP-immunopositive neurons were densely aggregated in a more restricted VL portion of the SCN (**Figure 2F**). This organization is grossly similar to that described in the laboratory mouse and rat (Abrahamson and Moore, 2001; Moore et al., 2002; Morin and Allen, 2006), although the dorsal portion of the SCN containing AVP neurons appeared relatively more extended in M. natalensis than in laboratory rodents.

## The SCN of T. b.-infected M. natalensis

In Nissl-stained sections through the SCN of the infected animals, the most striking change was a decrease of neuronal density and gliosis after T. b. gambiense infection, especially at 8 weeks (**Figures 3C,F**) as compared to matched controls (**Figures 3A,D**) and to T. b. brucei-infected animals (**Figures 3B,E**).

The neuronal cell counts in Nissl-stained sections confirmed the qualitative observations, showing about 17% neuronal loss (83 ± 9% neurons per ROI as compared to controls) in the SCN at 4 weeks after T. b. gambiense infection and

FIGURE 2 | Suprachiasmatic nucleus (SCN) cytoarchitecture of Mastomys natalensis. (A–D) Cresyl violet-stained coronal sections through the rostral, middle and caudal levels of the SCN. The dotted lines in (C) outline the extent of the regions of interest (ROIs) used for the quantitative analyses; the asterisks in (B,D) are placed at the center of the same ROIs in the respective rostral and caudal sections. (E,F) Confocal microscopy images showing the distribution of the neuropeptide arginine-vasopressin (AVP) (E) and vasoactive intestinal polypeptide (VIP) (F), in the middle level of the SCN. Abbreviations: DM, dorsomedial portion; ox, optic chiasm; VL, ventrolateral portion; 3V, third ventricle. Scale bars: 350 µm in (A), 100 µm in (B–D); 150 µm in (E,F).

about 30% neuronal loss (70 ± 11% neurons as compared to controls) at 8 weeks. The decrease of neuron density in the SCN of the T. b. gambiense-infected animals vs. matched controls was significant at 8 weeks (P < 0.05). No significant decrease of neuronal density as compared to controls was instead found in the SCN of T. b. brucei-infected animals.

Neuronal cell counts in the two subregions of the SCN showed that the neuronal density decrease documented after T. b. gambiense infection affected both the DM portion (79 ± 9% neurons per ROI as compared to controls at 4 weeks; 67 ± 13% at 8 weeks) and the VL portion (84 ± 10% neurons at 4 weeks; 71 ± 9% at 8 weeks), with a significant decrease in each SCN subdivision as compared to matched controls at the latest time point (**Figures 3G–J**).

In the NeuN-immunostained sections, the density of immunopositive neurons in the SCN appeared lower than in the cresyl violet-stained sections at corresponding levels, especially in the medial portions of the nucleus, and the immunostaining appeared fainter in the T. b. gambiense-infected animals than in controls, though clearly visible and with the same distribution as in controls (**Figures 3K,L**). This is consistent with the findings that the expression of NeuN, a reliable and conserved neuronal marker, can be downregulated in pathological conditions (Duan et al., 2016), and that in other rodent species (hamster and mouse) NeuN does not label all SCN neurons, in particular medially (Morin et al., 2011).

Regarding our quantitative evaluation, performed in the SCN regions of dense NeuN immunoreactivity, the density of immunopositive neurons at 8 weeks after T. b. gambiense

infection showed a decrease (68 ± 5% NeuN-immunopositive neurons as compared to matched controls) similar to that documented in the cresyl violet-stained sections (see above), which was significantly lower than in the SCN of the control group (**Figures 3M,N**).

Regarding the observations of sections through the SCN processed for AVP or VIP immunofluorescence, the peptide immunosignal seemed to be downregulated and peptidergic cells appeared decreased in the T. b. gambiense-infected animals. This decrease, however, was not documented further by an objective quantitative evaluation.

In the hippocampus, NeuN immunostaining did not show differences in the distribution and labeling intensity between control and T. b. gambiense-infected animals (**Figures 4A,B**). At the quantitative evaluation, no significant changes were found at 8 weeks after T. b. gambiense infection animals vs. controls (**Figure 4C**).

As previously reported in laboratory rats (Tamada et al., 1998; Becquet et al., 2008), GFAP immunostaining in the SCN of M. natalensis was obviously more intense than in the immediately adjacent hypothalamic areas (**Figure 5A**). After trypanosome infection, GFAP immunostaining was markedly increased throughout the SCN, and especially in its VL portion (**Figures 5B,C,F–K**). The immunopositive astrocytes exhibited an activated phenotype, with hypertrophic and intensely immunostained cell bodies and processes, at both the sampled time points during the progression of

T. b. brucei (**Figures 5G,H**) and T. b. gambiense (**Figures 5J,K**) infections.

Densitometric analysis demonstrated that the GFAP immunosignal intensity in the SCN was increased in T. b. bruceiinfected animals at 4 weeks, although this trend did not reach statistical significance (**Figures 5D,E**). After T. b. gambiense infection, GFAP immunosignal intensity was significantly increased (a 2.5 fold increase) at both of the sampled time points.

## DISCUSSION

This is the first histopathological investigation on brain structures in an animal model of infection with the human pathogen T. b. gambiense and is focused on the SCN. The main finding of the present study is the occurrence of structural damage in the master circadian pacemaker, with about 30% neuronal loss during the progression of the encephalitic stage of T. b. gambiense infection in M. natalensis. Despite the technical limitations in the use of NeuN as marker of SCN neurons (Morin et al., 2011) and the limited sample size, the finding was here replicated using cresyl violet staining and NeuN immunophenotyping. No neuronal loss was found in the same animals in the sampled hippocampal region, indicating that no widespread neurodegenerative phenomena had occurred in the brain. In the same recipient species, no significant neuronal loss was found in the SCN after T. b. brucei infection. Although it is likely that the small number of analyzed cases affected the statistical power, the comparison between the two infection paradigms suggests a structural vulnerability of the biological clock to T. b. gambiense infection.

Several studies in T. b. brucei-infected rodents and neuropathological studies in the brain of victims of T. b. gambiense HAT have indicated that African trypanosome infection leads to a neuroinflammatory pathology without the hallmarks of a widespread neurodegeneration (Quan et al., 1999; Kristensson et al., 2010; Buguet et al., 2014). A neuroinflammatory response to the infection was here confirmed by the activation of astrocytes in M. natalensis infected with T. b. gambiense, which was marked also after T. b. brucei infection, as previously observed after T. b. brucei infection in the laboratory mouse (Hunter et al., 1992; Keita et al., 1997) and rat (Kennedy, 2008).

No overt structural alterations have been observed in previous studies on the SCN of T. b. brucei-infected rats, as supported by the present data in M. natalensis. Functional alterations in the SCN of T. b. brucei-infected rats have, however, been reported, with changes in spontaneous and light-induced Fos expression (Bentivoglio et al., 1994; Peng et al., 1994), melatonin receptor binding (Kristensson et al., 1998), as well as reduced neuronal firing rate and phase advance of its peak (Lundkvist et al., 1998), and reduced excitatory synaptic activity (Lundkvist et al., 2002) in SCN slice preparations. Similar alterations of spontaneous single cell unit activity in SCN slices have been elicited by exposure to the proinflammatory cytokines tumor necrosis factor-α and interferon-γ (Lundkvist et al., 2002). This indicates that inflammatory mediators which are protagonist of African trypanosomiasis (Kristensson et al., 2010; Kennedy, 2013) can disrupt the synaptic machinery of SCN neurons (Lundkvist et al., 2002), which are cell autonomous oscillators (Hastings et al., 2014). In T. b. brucei-infected rats, a decrease in the expression of glutamate receptor subunits was also found, despite normal density and distribution of the excitatory retinal fibers which target the SCN (Lundkvist et al., 1998). The present experimental data sets indicate that the SCN could be even more susceptible to T. b. gambiense than to T. b. brucei infection, leading in this nucleus to widespread neuron loss.

Alterations of the daily alternation of sleep and wakefulness are a characteristic feature of HAT (Buguet et al., 2001, 2014). Other endogenous circadian rhythms, in particular some hormone secretion cycles, are also altered in HAT (Radomski et al., 1994; Brandenberger et al., 1996). The infection, therefore, affects the regulation of circadian rhythmicity. Experimental data based on T. b. brucei infection have indicated that the hypothalamus, and especially the posterior hypothalamus, is targeted early in the encephalitic stage by the active process of trypanosome traversal of the blood-brain barrier (Laperchia et al., 2016). The SCN, which expresses receptors

to inflammatory mediators (Coogan and Wyse, 2008), could be especially sensitive to intrahypothalamic inflammatory signaling.

No information is available on the neuropathology of the SCN in the brain of HAT victims. Monitoring of the sleep-wake cycle in T. b. gambiense HAT patients with polysomnography during therapy has indicated that alterations of the sleep-wake cycle can slowly recover, which can be considered a clinical marker of therapy efficacy (Buguet et al., 1999; Mpandzou et al., 2011), as also shown by the use of non-invasive actigraphy (Njamnshi et al., 2012). Such clinical findings indicate, therefore, that SCN dysfunction in HAT can be largely compensated when the infection is cured.

Recovery after functional damage of the SCN has been reported in experimental paradigms of chronic (Palomba and Bentivoglio, 2008) and acute (O'Callaghan et al., 2012) neuroinflammation elicited by peripheral administration of the bacterial endotoxin, lipolysaccharide. Concerning structural damage, studies on partial SCN lesions with targeted methodological approaches, mostly based on genetic manipulations, have shown that lesions of cell populations with a specific chemical signature, e.g., those containing the calcium binding protein calbindin, which are located in the ''core'' of the rodent SCN (Antle and Silver, 2005), can lead to sustained loss of physiological and behavioral rhythmicity, while other microlesions can be compensated (Kriegsfeld et al., 2004). No detailed characterization of the SCN cell phenotypes spared by T. b. gambiense infection was possible in the present study. However, our findings indicate that neuronal loss was widely distributed in the SCN, therefore pointing to a neurodegenerative process which, based on experimental data (Kriegsfeld et al., 2004), could permit functional recovery. No clinical studies, however, are available on long-term, possibly subtle, sequels of permanent partial SCN damage in humans.

In conclusion, the present findings show that neurodegenerative events are caused in the SCN in a rodent model of infection with a causative agent of HAT. T. b. gambiense HAT is a chronic progressive disease with a mean duration of 3 years and, due to the nonspecific clinical signs and symptoms of the first stage, patients mostly present at the observation when the disease is already in the encephalitic stage (Büscher et al., 2017; Njamnshi et al., 2017), and the partial neuronal damage here documented in the SCN of an animal model might have already occurred. The outcome of an infection depends on the

## REFERENCES


delicate and complex interplay between the pathogen and the host, and the potential translational implications of the present findings should, therefore, be regarded with caution. Even taking this into account, our study further underlines the urgency of an effective control and elimination of this dreadful African disease.

## AUTHOR CONTRIBUTIONS

CT and Y-ZX: data analysis and manuscript preparation. Y-ZX tissue processing. DMN: infection design and animal experiments. MB: study design and article writing.

## FUNDING

This work was supported by intramural funds at INRB and at the University of Verona.


Franco, J. R., Simarro, P. P., Diarra, A., and Jannin, J. G. (2014). Epidemiology of human African trypanosomiasis. Clin. Epidemiol. 6, 257–275. doi: 10.2147/clep. s39728


**Conflict of Interest Statement**: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Tesoriero, Xu, Mumba Ngoyi and Bentivoglio. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Neural Damage in Experimental Trypanosoma brucei gambiense Infection: Hypothalamic Peptidergic Sleep and Wake-Regulatory Neurons

#### Edited by:

Nouria Lakhdar-Ghazal, Mohammed V University, Morocco

#### Reviewed by:

Jackson Cioni Bittencourt, University of São Paulo, Brazil Khalid El Allali, Institut Agronomique et Vétérinaire Hassan II, Morocco

> \*Correspondence: Marina Bentivoglio marina.bentivoglio@univr.it

#### †Present address:

Claudia Laperchia, Department of Fundamental Neurosciences, Faculty of Biology and Medicine, University of Lausanne, Lausanne, Switzerland Yuan-Zhong Xu, Brown Foundation Institute of Molecular Medicine, University of Texas Health Science Center, Houston, TX, United States Tiziana Cotrufo, Department of Cell Biology, Physiology and Immunology, Faculty of Biology, University of Barcelona, Barcelona, Spain

> Received: 07 December 2017 Accepted: 07 February 2018 Published: 27 February 2018

#### Citation:

Laperchia C, Xu Y-Z, Mumba Ngoyi D, Cotrufo T and Bentivoglio M (2018) Neural Damage in Experimental Trypanosoma brucei gambiense Infection: Hypothalamic Peptidergic Sleep and Wake-Regulatory Neurons. Front. Neuroanat. 12:13. doi: 10.3389/fnana.2018.00013 Claudia Laperchia1† , Yuan-Zhong Xu1† , Dieudonné Mumba Ngoyi <sup>2</sup> , Tiziana Cotrufo1† and Marina Bentivoglio1,3 \*

<sup>1</sup>Department of Neuroscience Biomedicine and Movement Sciences, University of Verona, Verona, Italy, <sup>2</sup> Institut National de Recherche Biomédicale (INRB), Kinshasa, Democratic Republic of Congo, <sup>3</sup>National Institute of Neuroscience (INN), Verona Unit, Verona, Italy

Neuron populations of the lateral hypothalamus which synthesize the orexin (OX)/hypocretin or melanin-concentrating hormone (MCH) peptides play crucial, reciprocal roles in regulating wake stability and sleep. The disease human African trypanosomiasis (HAT), also called sleeping sickness, caused by extracellular Trypanosoma brucei (T. b.) parasites, leads to characteristic sleep-wake cycle disruption and narcoleptic-like alterations of the sleep structure. Previous studies have revealed damage of OX and MCH neurons during systemic infection of laboratory rodents with the non-human pathogenic T. b. brucei subspecies. No information is available, however, on these peptidergic neurons after systemic infection with T. b. gambiense, the etiological agent of 97% of HAT cases. The present study was aimed at the investigation of immunohistochemically characterized OX and MCH neurons after T. b. gambiense or T. b. brucei infection of a susceptible rodent, the multimammate mouse, Mastomys natalensis. Cell counts and evaluation of OX fiber density were performed at 4 and 8 weeks post-infection, when parasites had entered the brain parenchyma from the periphery. A significant decrease of OX neurons (about 44% reduction) and MCH neurons (about 54% reduction) was found in the lateral hypothalamus and perifornical area at 8 weeks in T. b. gambiense-infected M. natalensis. A moderate decrease (21% and 24% reduction, respectively), which did not reach statistical significance, was found after T. b. brucei infection. In two key targets of diencephalic orexinergic innervation, the peri-suprachiasmatic nucleus (SCN) region and the thalamic paraventricular nucleus (PVT), densitometric analyses showed a significant progressive decrease in the density of orexinergic fibers in both infection paradigms, and especially during T. b. gambiense infection. Altogether the findings provide novel information showing that OX and MCH neurons are highly vulnerable to chronic neuroinflammatory signaling caused by the infection of human-pathogenic African trypanosomes.

Keywords: human African trypanosomiasis, orexin, hypocretin, melanin-concentrating hormone, neuroinflammation, sleep, wake

## INTRODUCTION

Many infections disturb sleep, and somnolence is part of the so-called sickness behavior in response to infections (Poon et al., 2015). The alterations of sleep in human African trypanosomiasis (HAT) are, however, so characteristic and severe that they gave to the disease the alternative name of sleeping sickness (Buguet et al., 2014; Büscher et al., 2017). HAT is caused by the extracellular protozoan parasites Trypanosoma brucei (T. b.), inoculated through bites of the hematophagous tsetse flies (genus Glossina). The disease is still endemic, with a focal distribution mostly in remote rural areas, in the vast sub-Saharan region of the vector's habitat. After an epidemic in the 1990s, this neglected tropical disease has attracted attention in the public health agenda. Due to sustained control activities, the reported cases have declined in recent years (dropping to less than 3000 in 2015) and the disease is currently targeted for elimination (WHO, 2017). There are, however, concerns on the fight against HAT and worries about disease re-emergence, especially due to inaccurate case reporting and to silent carriers (Welburn et al., 2016; Büscher et al., 2017).

The vast majority of HAT cases (97%; WHO, 2017) are due to the infection of T. b. gambiense, for which humans provide the reservoir, causing a chronic progressive form of the disease with a course from 1 to 3 years. T. b. gambiense HAT is endemic in western and central Africa, and especially in the Democratic Republic of Congo (DRC) where over 70% of cases in the last 10 years have been reported (WHO, 2017). A more acute form of HAT, lasting 6–8 months, is endemic in eastern and southern Africa and is caused by T. b. rhodesiense, representing a zoonosis with wildlife and livestock reservoirs and therefore difficult to control. Most experimental studies on nervous system infection caused by African trypanosomes have been performed using a third parasite subspecies, T. b. brucei, which is infectious to animals but not to humans.

During the progression of T. b. infection, the parasites, which initially invade peripheral organs through the blood and lymph, enter the central nervous system parenchyma. HAT is considered fatal if untreated (WHO, 2017) and the therapy of the meningoencephalitic stage is very toxic. Ongoing clinical trials based on oral therapy with fexinidazole are opening new perspectives for the cure of HAT (Mesu et al., 2018).

The disease causes severe chronic neuroinflammation, with high levels of inflammatory mediators (Kennedy, 2013; Büscher et al., 2017). Early autoptic studies of HAT victims have shown glial activation with myelin pallor but no features of neurodegeneration (Lejon et al., 2013). The clinical picture, initially nonspecific, leads to a constellation of neurological and psychiatric alterations during disease progression (Kennedy, 2013; Lejon et al., 2013; Buguet et al., 2014; Büscher et al., 2017). Since historical descriptions, striking clinical features of HAT in endemic regions are represented by diurnal somnolence and nocturnal insomnia, as well as episodes of irresistible sleep during wakefulness (sleep attacks; Lhermitte, 1910; Blum et al., 2006; Buguet et al., 2014). Polysomnographic recordings of the sleep-wake cycle in HAT patients have shown that the disease does not cause hypersomnia but rather a circadian disruption of sleep-wake alternation, as well as alterations of the sleep structure (Buguet et al., 2001, 2005, 2014).

Sleep-wake cycle changes during African trypanosomiasis implicate the master circadian pacemaker, the hypothalamic suprachiasmatic nucleus (SCN), which drives endogenous biological rhythms in the mammalian brain (van Esseveldt et al., 2000; Golombek and Rosenstein, 2010; Moore, 2013). Findings on neuronal cell loss in the SCN of an animal model of T. b. gambiense infection provided by the multimammate mouse, Mastomys natalensis (Mehlitz, 1975; Büscher et al., 2005) are presented in a companion article (Tesoriero et al., 2018). On the other hand, alterations of sleep architecture in African trypanosomiasis implicate damage to structures of the distributed neural network which regulates sleep and wakefulness (Saper et al., 2010; Scammell et al., 2017). In particular, the observation of sudden sleep episodes during wakefulness, as well as sleep fragmentation and disruption of the normal sleep sequence which also occur in T. b. brucei-infected rats (Darsaud et al., 2003; Seke Etet et al., 2012; Laperchia et al., 2016, 2017) recall the chronic sleep disorder narcolepsy (Sateia, 2014; Scammell, 2015), whose pathogenesis is due to impaired orexinergic signaling (Liblau et al., 2015). Such alterations point, therefore, to dysfunction of neurons which contain the orexin (OX)/hypocretin peptides, located in the posterior lateral hypothalamus. These neurons play a key role in wakefulness stability and sleep-wake transitions (de Lecea and Huerta, 2014), besides their implication in energy homeostasis and other physiological functions underlying motivated behaviors (Sakurai, 2014; Li et al., 2016).

Orexinergic neurons are intermingled with those expressing another peptide, melanin-concentrating hormone (MCH), which are sleep-promoting and also regulate food intake and other functions implicated in the motivational aspects of behavior (Torterolo et al., 2015; Ferreira et al., 2017). Structural and functional alterations of these two major hypothalamic neuronal populations has been previously reported in T. b. bruceiinfected laboratory rats and mice (Palomba et al., 2015), but no information on these cell groups during T. b. gambiense infection is available. To fill this gap of knowledge, the present study was aimed at the histopathological investigation of OX and MCH cell bodies and OX fibers in the animal model of T. b. gambiense infection provided by M. natalensis. This rodent is sensitive to T. b. gambiense infection (Mehlitz, 1978), at variance with laboratory rodents which are not susceptible to most isolates of this human-pathogenic parasite subspecies (Giroud et al., 2009).

OX neurons give origin to extensive projections (Peyron et al., 1998). In particular, in laboratory rodents orexinergic fibers innervate the area surrounding the SCN (Peyron et al., 1998; Marston et al., 2008), and show a discrete distribution in the thalamus along the midline with dense innervation of the thalamic paraventricular nucleus (PVT; Peyron et al., 1998; McGranaghan and Piggins, 2001; Mintz et al., 2001; Kirouac et al., 2005) documented also in African rodents (Bhagwandin et al., 2011a). The density of OX innervation was here evaluated in these two diencephalic targets of orexinergic projections, the peri-SCN region and PVT, given their high functional relevance for interactions between circadian and vigilance state regulation (Colavito et al., 2015).

## MATERIALS AND METHODS

## Animals and Infection

Experimental procedures were performed under approval of the ethical committee of the Ministry of Health of DRC, adhering to the European Communities Council (86/609/EEC) directives and the ARRIVE (''Animal Research Reporting of in vivo Experiments'') guidelines. All efforts were made to minimize animal number and suffering. The present investigation was based on the same brains of the animals used for the study of the SCN presented in a companion article (Tesoriero et al., 2018).

In brief, adult M. natalensis of both sexes (a total number of 45 animals, 30 destined to the two infection paradigms and 15 controls) were obtained from the breeding colony at the Institut National de Recherche Biomedicale (INRB, Kinshasa, DRC).

The rodent M. natalensis is very widespread in sub-Saharan Africa, and holds a taxonomic position between the mouse (house mouse) and the rat (ship, roof rat; Coetzee, 1975). The definition of multimammate derives from a uniquely large number of mammae in the female (Isaäcson, 1975). M. natalensis (also called Praomys natalensis) is nocturnal, with the peak of activity in the first 3 h of the period of darkness (Coetzee, 1975). The animal is omnivorous, uses preferentially burrows of other rodents for nesting and adapts easily to different environments (Isaäcson, 1975), representing the major rodent pest in sub-Saharan Africa (e.g., Mulungu et al., 2013).

The animals were maintained under a 12 h:12 h light/dark cycle, with free access to food and water. One group of animals was infected with T. b. gambiense (MHOM/INRB/2006/11A, originally isolated from a patient in DRC in 2006), and another group with T. b. brucei (AnTat 1.1E). The parasites derived from the cryostabilates collected at INRB, and each experimental group of infected M. natalensis included matched uninfected controls. The infection was done with intraperitoneal injection of 0.25 ml per animal of a solution (0.1 M phosphate buffer, pH 7.4, supplemented with glucose) containing 106.9–107.2 trypanosomes/ml. Parasitaemia was verified weekly from the tip of the tail vein. At 4 and 8 weeks post-infection, the animals (n = 3 or 4 infected animals per time point and control animals per experimental group) were sacrificed during daytime, under anesthesia, by transcardial perfusion with ice-cold 0.9% saline followed by formaldehyde solution (obtained dissolving paraformaldehyde in 0.1 M phosphate buffer, pH 7.2, at a 4% concentration). Previous monitoring at INRB of the natural course of the infections in M. natalensis indicated that at 8 weeks T. b. brucei infection is in a very advanced phase (and this was therefore selected as end-point of the study), whereas T. b. gambiense infection can last several weeks.

After perfusion, the brains were removed from the skull, posfixed for a few hours, and then stored until processing at 4◦C in 0.01 M phosphate buffered-saline, pH 7.4 (PBS), containing 0.1% sodium azide.

## Tissue Processing and Immunohistochemistry

Following cryoprotection in 30% sucrose in PBS, the brains were cut on the coronal plane with a freezing microtome into 30 µmthick sections, which were collected in six series. One series of sections was stained with cresyl violet for cytoarchitectonic control.

Two series of sections through the hypothalamus were processed free-floating for OX immunohistochemistry. The peptides OX-A and OX-B, also called hypocretin-1 and 2, cleaved from the common precursor prepro-OX, are largely co-localized in the same neurons (Nixon and Smale, 2007), and OX-A is more stable than OX-B. OX-A immunophenotyping was therefore used to investigate orexinergic neurons in the present study. The sections were first soaked in 1% H2O<sup>2</sup> in PBS for 20 min to inactivate endogenous peroxidase activity, and then pre-incubated in a solution of 5% bovine serum albumin (BSA) and 0.3% Triton X-100 in PBS for 1 h at room temperature. Subsequently, the sections were incubated overnight at room temperature in polyclonal goat anti-OX-A antibody (1:500; Santa Cruz, CA, USA) diluted in PBS containing 1% BSA, 0.2% Triton X-100. Following thorough washes, the sections were incubated for 2 h in biotinylated horse anti-goat IgGs (1:200, Vector, Burlingame, CA, USA), and finally reacted with avidinperoxidase complex (1:100, Vector, Burlingame, CA, USA) for 1 h and with 3,3<sup>0</sup> -diaminobenzidine (DAB) as chromogen.

The second series was processed further for MCH immunohistochemistry based on a two-color protocol (Peng et al., 1995). These sections were incubated overnight at room temperature in polyclonal rabbit anti-MCH antibody (1:1000; Phoenix Pharmaceuticals, Burlingame, CA, USA). After repeated washing, the sections were incubated for 2 h in biotinylated goat anti-rabbit IgGs (1:200, Vector) and then reacted with avidin-peroxidase complex (1:100; Vector) for 1 h. In the last step of the procedure, the sections were reacted in a freshly prepared and filtered solution containing 0.05% α-naphthol, 0.1% ammonium carbonate, and 0.003% H2O<sup>2</sup> in PBS. The dark blue reaction product was turned into pink by an additional incubation in 0.1% phosphate-buffered pyronin B. This procedure results in the simultaneous visualization in bright-field microscopy of brown DAB reaction products of the first incubation and pink reaction products of the second run (Peng et al., 1995).

After immunohistochemical processing, all sections were thoroughly washed in PBS, mounted on gelatin-coated slides, air-dried, dehydrated, cleared and coverslipped with Entellan. Specific immunostaining was absent in control sections in which the primary antibodies were omitted.

The presence of parasites in the brain parenchyma was investigated in additional sections through the telencephalon and diencephalon, which were processed for double

immunofluorescence. These sections were incubated with a mixture of primary antibodies: rabbit polyclonal antibodies which recognize the anti-variant surface glycoprotein of the AnTat 1:1E stabilate (1:200; kindly provided by Philippe Büscher, Institute of Tropical Medicine, Antwerp, Belgium) to visualize the parasites, and goat polyclonal anti-glucose transporter-1 antibodies (1:100, Santa Cruz Biotechnology) to visualize blood vessel endothelia (Pardridge et al., 1990). The sections were rinsed in PBS and incubated in a solution of species-specific secondary antibodies conjugated with Cy2 or Cy3 (1:100; Jackson ImmunoResearch, Suffolk, UK), rinsed in PBS, mounted on slides using a fluorescencecompatible medium (Dako, Hamburg, Germany) and stored at 4◦C.

## Data Analysis and Statistics

The sections processed for immunofluorescence for parasite detection were observed with a confocal laser scanning microscope (Zeiss LSM 510 Carl Zeiss, Jena, Germany), equipped with an argon laser emitting at 488 nm (Cy2) and a helium/neon laser emitting at 543 nm (Cy3). The other sections were investigated in bright-field microscopy and quantitative analyses

FIGURE 2 | Low power view of orexin (OX)-A-immunoreactive neurons in the perifornical area and lateral hypothalamus of Mastomys natalensis at 4 weeks (W) and 8 W after infection with Trypanosoma brucei brucei (Tbb) or Trypanosoma brucei gambiense (Tbg) and in uninfected animals. Note the rarefaction of immunostained cell bodies and the reduction of neuropil immunolabeling in the infected animals. Abbreviations: f, fornix; opt, optic tract. Scale bar: 250 µm.

were pursued in three animals per group, blindly of the animal's experimental group assignment.

The series of sections processed for OX-A single immunohistochemistry was used for OX cell counts, and that processed for double immunohistochemistry for MCH cell counts. The number of OX and MCH cells was determined stereologically using the optical fractionator method throughout the rostrocaudal extent of the perifornical area and lateral hypothalamus. Cells were visualized using an Olympus BX51 microscope (20× objective) with a motorized stage connected to a digital camera (JVC CCD KY-F58) and equipped with the image analysis digital system Stereo-Investigator software (MicroBrightfield Corp.). For a systematic random sampling, a grid centered on the fornix and adjacent medially to the wall of the third ventricle was used. The grid was divided into nine counting frames, allowing also partial cell counts in the medial, middle and lateral portions of the lateral hypothalamus. Only cells within the frame or touching one of the frame borders were counted.

For densitometric evaluation, three regularly spaced sections were sampled through the middle portion of the SCN and PVT, respectively, from each animal. In each section, under calibrated constant light parameters, four images from the PVT and two images from the peri-SCN region (within an area of 300 µm in length lateral to the SCN) were randomly sampled on each side using a 40× objective (NA 0.75; yielding a frame of 0.0256 mm<sup>2</sup> ). OX immunoreactivity signal was then measured by defining the zero value of optical density (OD) as background, measured in a portion of the section tissue devoid of specific immunostaining. For each region of interest (ROI), the values from each section were averaged. A grand mean for each ROI was then computed from the mean value derived from each section.

Data are reported as mean ± standard error of the mean (SEM). Kruskal-Wallis analysis of variance followed by the Dunn's test for pairwise comparison was used for the statistical evaluation of cell counts. One-way analysis of variance followed by the LSD post hoc test was used for OD values of immunoreactivity of OX fibers in the peri-SCN region and PVT. Significance threshold was set at P < 0.05.

## RESULTS

### Trypanosomes in the Brain Parenchyma

Double immunofluorescence showed in M. natalensis the occurrence of T. b. gambiense and T. b. brucei outside blood vessels within the brain parenchyma, in variable amounts, at both 4 weeks and 8 weeks (**Figure 1**) after infection, as also presented in the companion article (Tesoriero et al., 2018). This indicated that in both paradigms the infection was in the encephalitic stage at the sampled time points.

## OX and MCH Cell Bodies

In M. natalensis, OX-immunolabeled neurons showed a distribution similar to that described in laboratory rodents (Peyron et al., 1998; Mintz et al., 2001) and in a variety of African rodents (Bhagwandin et al., 2011a,b; Sweigers et al., 2017). The largest population was concentrated in the perifornical area, from which it extended to the lateral hypothalamus (**Figure 2**). MCH neurons also showed in the hypothalamus the distribution described in rodents, very similar, in particular, to that reported in the rat (Bittencourt, 2011). Thus, MCH neurons were distributed rostrally in the incerto-hypothalamic area, between the fornix and the internal capsule at the level of the tuberal lateral hypothalamus. Proceeding posteriorly, MCH neurons were densely aggregated around the fornix, and with sparser distribution medially to the internal capsule, as well as dorsally in the zona incerta, and with a medial, periventricular location. MCH neurons were intermingled with OX neurons in the perifornical area and lateral hypothalamus, and also surrounded OX neurons dorsally and ventrally. OX and MCH somata were morphologically similar, with a multipolar or fusiform shape (**Figures 3A,B**, **4**), giving origin to varicose fibers.

A decrease in the density of OX cell bodies and neuropil immunostaining after African trypanosome infection at both the sampled time points was evident even at low power observation along the rostrocaudal and mediolateral axes (**Figure 2**). At

FIGURE 3 | (A,B) Images of orexin-A-immunoreactive neurons in uninfected Mastomys natalensis (B represents at higher power the area boxed in A), and at 4 weeks (W) and 8 W after infection with Trypanosoma brucei brucei (Tbb) or Trypanosoma brucei gambiense (Tbg) (C–F). Abbreviations: f, fornix; opt, optic tract. Scale bars: 250 µm in (A), 150 µm in (E) (applies also to B–D,F).

FIGURE 4 | Images of OX-A-immunoreactive neurons (brown) and melanin-concentrating hormone (MCH)-immunoreactive neurons (purple) through the middle third (upper row) and posterior third (lower row) of the lateral hypothalamus of uninfected Mastomys natalensis and at 8 weeks (W) after infection with Trypanosoma brucei brucei (Tbb) or Trypanosoma brucei gambiense (Tbg). Note the marked shrinkage of many MCH-immunostained neurons. Scale bar: 150 µm.

higher power, the intensity of the soma immunostaining appeared relatively preserved in the infected animals, but with less extensive dendritic filling than in matched controls, and some cell bodies appeared shrunken (**Figures 3C–F**). Also the density of MCH neurons appeared decreased in both paradigms of infection, with shrinkage of many cell bodies (**Figure 4**).

Cell counts documented a progressive decrease of both peptidergic cell populations, which was more marked after T. b. gambiense infection than in the T. b. brucei-infected animals (**Figure 5**). In particular, after T. b. brucei infection the reduction of OX-immunostained cells with respect to controls was 11.12 ± 6.8% at 4 weeks and 21.34 ± 9.8% at 8 weeks, and that of the MCH-immunostained cells, evaluated in the lateral hypothalamus and perifornical area, was 17.53 ± 10.6% at 4 weeks and 24.45 ± 2.2% at 8 weeks. These decreases did not reach, however, statistical significance. After T. b. gambiense

FIGURE 6 | Top: images of the ventral part of the anterior hypothalamus showing OX-A immunoreactive fibers in the region lateral to the suprachiasmatic nucleus (SCN, delimited by a dotted line), in Mastomys natalensis at 4 weeks (W) and 8 W after infection with Trypanosoma brucei brucei (Tbb) or Trypanosoma brucei gambiense (Tbg). Note that the decrease in the immunoreactive fiber density, especially at 8 weeks after T. b. gambiense infection. Scale bar: 250 µm. Bottom: bar graphs of the quantitative densitometric evaluation of OX-A immunosignal in the peri-SCN region in control (ctrl) and infected animals. <sup>∗</sup>P < 0.05; ∗∗∗P < 0.001. Abbreviations: OD, optical density; OX, optic chiasm.

infection, 21.58 ± 7.6% reduction of OX-immunostained cells was documented at 4 weeks, and 43.78 ± 7.4% reduction at 8 weeks, when the cell number decrease vs. matched controls was significant. Even more marked was the decrease of MCH neurons (48.15 ± 4% reduction at 4 weeks and 54.15 ± 6.3% at 8 weeks), which was significant at both survival times.

Partial cell counts showed that loss of immunostained cell bodies occurred throughout the extent of their distribution in the lateral hypothalamus and perifornical area.

## OX Fibers

In uninfected M. natalensis, orexinergic fibers were found to be densely aggregated laterally to the SCN, as well as along the thalamic midline, innervating PVT throughout its extent. A significant density decrease was found in the orexinergic innervation of the peri-SCN region at 4 and 8 weeks after T. b. brucei infection (**Figure 6**). In PVT, T. b. brucei infection resulted in a reduction of orexinergic fiber density at 4 weeks, which was further reduced, and significant, at 8 weeks (**Figure 7**). In the peri-SCN region of T. b. gambiense-infected

Mastomys natalensis showing OX-A-immunoreactive fibers at 4 weeks (W) and 8 W after infection with Trypanosoma brucei brucei (Tbb) or Trypanosoma brucei gambiense (Tbg). Note the marked decrease of PVT orexinergic innervation, especially after T. b. gambiense infection (see also Figure 6). Scale bar: 250 µm. Bottom: bar graphs of the quantitative densitometric evaluation (OD, optical density) of OX-A immunosignal in PVT of control (ctrl) and infected animals. <sup>∗</sup>P < 0.05; ∗∗∗P < 0.001.

animals the decrease in the density of orexinergic fibers was very marked and highly significant at 8 weeks (**Figure 6**). Extreme reduction of orexinergic innervation was found in PVT at both 4 and 8 weeks after T. b. gambiense infection (**Figures 7**, **8**).

## DISCUSSION

The present findings show that T. b. gambiense infection leads to progressive quantitative decrease of OX and MCH neurons, and that this is more marked than that caused by T. b. brucei infection in the same host species, susceptible to both parasite subspecies. At the endpoint of the study (8 weeks postinfection), the reduction of OX neurons in M. natalensis was about 44% and that of MCH neurons accounted for the loss of about 54% of these neurons in the lateral hypothalamus and perifornical area. Of note, considerable damage of MCH neurons was documented in M. natalensis also at an earlier phase of the encephalitis (about 48% reduction of MCH neurons and about 21% reduction of OX neurons at 4 weeks postinfection).

Immunophenotyping could not here reveal whether the decrease was due to cell death phenomena or downregulation of peptide expression below the threshold of immunohistochemical visualization. However, in the infected animals immunostaining showed cell damage suggestive of an ongoing degenerative process, previously documented also in T. b. brucei-infected laboratory rodents in the absence of overall cell loss in the hypothalamus (Palomba et al., 2015).

The present analyses in infected M. natalensis also demonstrated significantly impoverished orexinergic innervation, which was especially marked after T. b. gambiense infection. This was here shown in the peri-SCN region, which is in turn involved in the circadian control of OX neuron activation (Marston et al., 2008; Belle et al., 2014; Belle and Piggins, 2017) that could be thus impaired during the infection. Decreased orexinergic input was also here documented in PVT, the thalamic midline structure in which OX release exerts powerful functional effects, and which plays a key role in funnelling state-dependent behavior information into the limbic system and prefrontal cortex (Colavito et al., 2015).

Loss of OX neurons (90% or greater) is the neuropathological hallmark of narcolepsy, and in particular of narcolepsy with cataplexy (sudden loss of postural muscle tone, triggered especially by emotion; Peyron et al., 2000; Thannickal et al., 2000). OX neurons degenerate also in narcolepsy without cataplexy, as reported in a post-mortem brain examination which showed loss of 33% of OX neurons (Thannickal et al., 2009). Besides narcoleptic-like changes of sleep architecture in HAT, the clinical phenotype of the sleep-wake cycle is, however, very different in this disease and in narcolepsy, since narcolepsy does not include circadian rhythm disturbances (Dantz et al., 1994).

The investigation of OX levels in the cerebrospinal fluid (CSF) of T. b. gambiense HAT patients (Dauvilliers et al., 2008) and T. b. brucei-infected rats (Palomba et al., 2015) have shown a decrease, which, however, was not significant and exhibited high interindividual variability. The concentration of OX in the CSF of narcoleptics with typical cataplexy is very low, whereas it is normal in most cases of narcolepsy without cataplexy, as well as in other neurodegenerative or neurotraumatic conditions which lead to partial loss of OX neurons (Bourgin et al., 2008). On the other hand, experimental evidence based on lesions of OX neurons in the rat has shown that a very large cell loss is needed to impair OX level in the CSF, which points to a considerable compensatory capacity of surviving OX neurons (Gerashchenko et al., 2003).

Importantly, polysomnographic recording of a limited cohort of T. b. gambiense HAT patients has shown that the narcoleptic-like sleep structure alterations can slowly recover after trypanocydal therapy, especially when patients are not severely ill (Buguet et al., 1999). This suggests that functional compensation can be effective when the disease is promptly cured.

The present finding of MCH neuron damage shows that T. b. gambiense infection does not affect OX neurons selectively. This also occurs in other conditions. For example, sleep disorders represent prominent non-motor symptoms in Parkinson's disease, and loss of both OX and MCH neurons, increasing with disease severity, has been reported in post-mortem studies of the hypothalamus of victims of this disease (Thannickal et al., 2007). It is relevant to recall, in this respect, that MCH neurons are instead spared in the brain of narcoleptic patients with or without cataplexy (Peyron et al., 2000; Thannickal et al., 2000, 2009). Altogether the findings indicate that T. b. gambiense infection disrupts the interplay between OX and MCH neurons, supporting data obtained in laboratory rodents after T. b. brucei infection (Palomba et al., 2015). These peptidergic neurons not only play reciprocal roles in the regulation of vigilance states (Konadhode et al., 2015), but are also interconnected. Notably, MCH neurons exert an inhibitory influence on OX neurons, tuning the overall output of these two systems (Rao et al., 2008).

Interestingly, the reduction of OX and MCH neurons here documented in T. b. gambiense infection is much more marked than that provoked by local (intra-hypothalamic) infusion of the endotoxin lipopolysaccharide for 1 month (Gerashchenko and Shiromani, 2004). Experimental studies based on T. b. brucei infection have indicated an escalating inflammatory response during disease progression (Kristensson et al., 2010)

## REFERENCES


in which the posterior hypothalamus is especially involved (Laperchia et al., 2016). Parasites reside in the median eminence, as in other circumventricular organs, since the first stage of hemolymphatic infection (Kristensson et al., 2010). The posterior hypothalamus is an early site of parasite traversal of bloodbrain barrier through gradients of permeability from the medianeminence-arcuate nucleus complex (Laperchia et al., 2016). This hypothalamic complex is known to represent a first order station for peripheral signals, transmitting information to the second order network provided by OX and MCH neurons (Guyon et al., 2009; Rostène et al., 2011). It may therefore not be surprising that these peptidergic cell populations are targeted by chronic inflammatory signaling during African trypanosomiasis.

Several data sets point to vulnerability of OX (Grossberg et al., 2011) and MCH (Sergeyev et al., 2001) neurons to inflammatory molecules. OX neurons are especially sensitive to nitric oxide toxicity (Obukuro et al., 2013). MCH neurons are very sensitive to signaling mediated by the CCL2 chemokine (Le Thuc et al., 2016). The expression of this chemokine in the brain during African trypanosomiasis remains to be explored, but CCL2 is part of the panel of inflammatory mediators increased in the CSF of T. b. gambiense HAT patients (Hainard et al., 2009).

In conclusion, the present findings show that sleep-wakeregulatory OX and MCH neurons are vulnerable to the infection caused by a human-pathogenic parasite responsible for the vast majority of HAT cases, and indicate that MCH neurons are especially susceptible to this infection. HAT causes a paradigmatic chronic progressive neuroinflammation, and the present data could therefore have pathogenetic implications for sleep disorders in other chronic neuroinflammatory conditions. Concerning HAT, the data recall the importance of an early diagnosis and therapy to favor compensation and recovery of vulnerable neuronal cell types damaged during the disease.

## AUTHOR CONTRIBUTIONS

CL and Y-ZX: data analysis and manuscript preparation. Y-ZX: tissue processing. DMN: infection design and animal experiments. TC: manuscript preparation: MB: study design and article writing.

## FUNDING

This work was supported by intramural funds at Institut National de Recherche Biomédicale (INRB) and at the University of Verona.

Bhagwandin, A., Fuxe, K., Bennett, N. C., and Manger, P. R. (2011a). Distribution of orexinergic neurons and their terminal networks in the brains of two species of African mole rats. J. Chem. Neuroanat. 41, 32–42. doi: 10.1016/j.jchemneu. 2010.11.002

Bhagwandin, A., Gravett, N., Hemingway, J., Oosthuizen, M. K., Bennett, N. C., Siegel, J. M., et al. (2011b). Orexinergic neuron numbers in three species of African mole rats with rhythmic and arrhythmic chronotypes. Neuroscience 199, 153–165. doi: 10.1016/j.neuroscience.2011. 10.023


by suppression of orexin neuron activity. J. Neurosci. 31, 11376–11386. doi: 10.1523/jneurosci.2311-11.2011


**Conflict of Interest Statement**: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Laperchia, Xu, Mumba Ngoyi, Cotrufo and Bentivoglio. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Cerebrospinal Fluid Amino Acid Profiling of Pediatric Cases with Tuberculous Meningitis

Shayne Mason<sup>1</sup> \*, Carolus J. Reinecke<sup>1</sup> and Regan Solomons <sup>2</sup>

*<sup>1</sup> Faculty of Natural Sciences, Centre for Human Metabolomics, North-West University, Potchefstroom, South Africa, <sup>2</sup> Department of Pediatrics and Child Health, Faculty of Medicine and Health Sciences, Stellenbosch University, Tygerberg, South Africa*

Background: In Africa, tuberculosis is generally regarded as persisting as one of the most devastating infectious diseases. The pediatric population is particularly vulnerable, with infection of the brain in the form of tuberculous meningitis (TBM) being the most severe manifestation. TBM is often difficult to diagnose in its early stages because of its non-specific clinical presentation. Of particular concern is that late diagnosis, and subsequent delayed treatment, leads to high risk of long-term neurological sequelae, and even death. Using advanced technology and scientific expertise, we are intent on further describing the biochemistry behind this devastating neuroinflammatory disease, with the goal of improving upon its early diagnosis.

#### Edited by:

*Vivienne Ann Russell, University of Cape Town, South Africa*

#### Reviewed by:

*Ravindra Kumar Garg, King George's Medical University, India Zongde Zhang, Beijing Chest Hospital, Capital Medical University, China*

\*Correspondence:

*Shayne Mason nmr.nwu@gmail.com*

#### Specialty section:

*This article was submitted to Neurodegeneration, a section of the journal Frontiers in Neuroscience*

Received: *28 July 2017* Accepted: *13 September 2017* Published: *26 September 2017*

#### Citation:

*Mason S, Reinecke CJ and Solomons R (2017) Cerebrospinal Fluid Amino Acid Profiling of Pediatric Cases with Tuberculous Meningitis. Front. Neurosci. 11:534. doi: 10.3389/fnins.2017.00534* Method: We used the highly sensitive analytical platform of gas chromatography-mass spectrometry (GC-MS) to analyze amino acid profiles of cerebrospinal fluid (CSF) collected from a cohort of 33 South African pediatric TBM cases, compared to 34 controls.

Results: Through the use of a stringent quality assurance procedure and various statistical techniques, we were able to confidently identify five amino acids as being significantly elevated in TBM cases, namely, alanine, asparagine, glycine, lysine, and proline. We found also in an earlier untargeted metabolomics investigation that alanine can be attributed to increased CSF lactate levels, and lysine as a marker of lipid peroxidation. Alanine, like glycine, is an inhibitory neurotransmitter in the brain. Asparagine, as with proline, is linked to the glutamate-glutamine cycle. Asparagine is associated with the removal of increased nitrites in the brain, whereas elevated proline coincides with the classic biochemical marker of increased CSF protein in TBM. All five discriminatory amino acids are linked to ammonia due to increased nitrites in TBM.

Conclusion: A large amount of untapped biochemical information is present in CSF of TBM cases, of which amino acid profiling through GC-MS has potential in aiding in earlier diagnosis, and hence crucial earlier treatment.

Keywords: gas chromatography-mass spectrometry (GC-MS), tuberculous meningitis (TBM), pediatric, cerebrospinal fluid (CSF), amino acid profiling

## INTRODUCTION

Tuberculosis (TB), caused by Mycobacterium tuberculosis (Mtb), is an ancient, persistent disease that remains a huge, deadly issue to this day. According to the World Health Organization (WHO, 2016), during 2015 there were an estimated 10.4 million new (incident) TB cases worldwide, of which 5.9 million were men, 3.5 million were women, and 1 million were of children. Six countries accounted for 60% of the new cases: India, Indonesia, China, Nigeria, Pakistan, and South Africa. An estimated 1.8 million people died from TB in 2015, with the mortality of cohorts of more than 100 individuals with extreme drug-resistant (XDR) TB being highest (>40%) in India and South Africa. The incidence of TB in South Africa (population 55 million) for 2015 was estimated to be 454,000 (294,000–649,000) individuals, of whom 33,000 (21,000–44,000) were children (<14 years of age). Hence, South Africa is one of the worst-stricken regions under the scourge of the M. tuberculosis bacillus, which depends on humans for its transmission.

TB is most commonly known in its pulmonary form; however, Mtb is not only localized in the lungs but, because of the systematic spread of the tubercule bacilli, can lead to extrapulmonary forms. A preferred site for Mtb—with high blood flow and oxygen content—is the brain; the pathogen is capable of crossing the blood-brain barrier and entering the meninges, which consist of three membranes between the skull and the brain. Infection within the meninges by Mtb leads to tuberculous meningitis (TBM)—the most severe manifestation of TB. The pathology that occurs in the central nervous system (CNS) is similar to that of its pulmonary type, in that lesions (turberculomas) form and can rupture releasing inflammatory markers. CNS-TB represents up to an estimated 10% of all forms of extra-pulmonary TB (and 1% of total TB) cases (Rock et al., 2005; Bhigjee et al., 2007; Cherian and Thomas, 2011), of which TBM is the principal manifestation.

The gold standard for diagnosis of TBM requires collection of cerebrospinal fluid (CSF), typically through a lumbar puncture. In 2010, Marais et al. proposed a uniform research case definition of TBM based on CSF: (1) TBM could be classified as "definite" when CSF demonstrated acid-fast bacilli on microscopy, a positive Mtb culture and/or passed a positive CSF Mtb commercial nucleic acid amplification test in an individual with symptoms or signs suggestive of the disease. (2) TBM could be classified as "probable" according to a scoring system based on clinical, CSF and neuroimaging criteria, as well as evidence of extraneural TB. In practice, however, TBM is difficult to diagnose in its early stages due to its non-specific clinical presentation. A particular concern is that late diagnosis, and subsequent delayed treatment, leads to high risk of long-term neurological sequelae, and even death.

The two primary biochemical markers currently considered for differential diagnosis of TBM are CSF protein and glucose levels (Solomons et al., 2015). Protein levels are defined as being either elevated—greater than 40 mg/dl (lower cut-off), or significantly elevated—greater than 100 mg/dl (higher cut-off) (Youssef et al., 2006; Hristea et al., 2012). Depressed glucose is defined as <2.2 mmol/l as an absolute value, or relatively as <0.5 CSF:blood glucose ratio as standardized cut-off values (Marais et al., 2010; Solomons et al., 2015).

Our group recently conducted an untargeted proton magnetic resonance (1H-NMR) metabolomics study on CSF from 17 pediatric cases with TBM (Mason et al., 2015), which revealed lowered glucose (as expected) and highly elevated lactate as the most defining biochemical markers of the disease, along with perturbed amino acids. The presence of highly elevated CSF lactate was further examined (Mason et al., 2016) and shown to be produced only by the host (with zero contribution from invading Mtb)—an interesting outcome in that the human brain mass produces lactate during a neuroinflammatory disease such as TBM. Indeed, CSF lactate has gained much attention recently in neuroenergetics (Mason, 2017), perhaps as a consequence of being an especially important biochemical marker of Mtb infection. Interestingly, the most prominent CSF metabolites identified by our untargeted <sup>1</sup>H-NMR metabolomics study as indicators of TBM were also derived in a completely independent study using exactly the same NMR data set and a novel nonparametric classification system (van Reenen et al., 2016)—a method of mathematical modelling for deductive verification of the AMLS (astrocyte–microglia lactate shuttle) hypothesis.

A secondary CSF biosignature of TBM revealed by our previous untargeted NMR metabolomics study consisted largely of perturbed amino acids—alanine, branched-chain amino acids (leucine, isoleucine, and valine), and lysine. These gluconeogenic amino acids, together with ketones, were postulated to be involved in directing (through shuttling mechanisms) CSF lactate, produced by glycolysis in astrocytes, from neurons preferentially into activated microglia in an attempt to aid the eradication of invading Mtb bacilli. Collectively, this biosignature led us to postulate the astrocyte–microglia lactate shuttle. Here, we now report on a more detailed targeted metabolomics study on the amino acid profiles in TBM from a larger collection of CSF samples.

Perturbations in amino acids in TBM cases are not a novel discovery as they have been observed previously in biochemical studies. In 1981, Corston et al. examined eleven viral meningitis and four TBM cases and reported that total amino acid concentrations in CSF were markedly higher in TBM than in viral meningitis. In 1998, Qureshi et al. compared several neurochemical markers in CSF between 11 cases of viral meningitis and 12 of TBM. These authors, using highpressure liquid chromatography, reported significantly increased aspartic acid, glutamic acid, GABA, glycine, and tryptophan in all cases, whereas only TBM cases exhibited significantly elevated phenylalanine, arginine, and homocysteine. From their results, Qureshi et al. postulated that inflammatory changes in meninges may interfere with amino acid transport across the blood–brain barrier, and proposed several therapeutic approaches. These earlier results on perturbed amino acids in TBM were, however, based on small sample sizes.

Here, we present the first report on a detailed amino acid profiling study of CSF obtained from a cohort of 33 TBM pediatric cases, compared to 34 controls. The method of analysis used was the sensitive platform of gas chromatographymass spectrometry (GC-MS). Based upon stringent statistical analyses and quality control measures, five amino acids—alanine, asparagine, glycine, lysine, and proline—were identified as being significantly increased in our group of TBM patients, of which we discuss the biological implications in the context of TBM. These observations justify the need for a comprehensive study on biochemical markers for TBM to be validated in a large-scale follow-up investigation.

## METHODS

## Sampling

This study was focused on a pediatric group with suspected TBM from the region surrounding Tygerberg Hospital in the Western Cape province of South Africa—an area endemic for TB. A specialized pediatric neurology unit within the hospital focuses on diagnosing and treating cases of TBM in the region. For the purposes of this study, a diagnosis of TBM was based on the uniform research case definition of Marais et al. (2010). Only children (>3 months and <13 years of age) with "definite" and "probable" TBM were included in the experimental patient group (n = 33). Our controls (n = 34) were age-matched pediatric patients suspected of meningitis, but later confirmed to be meningitis negative. Detailed clinical information on the study cohort is included in Supplementary Information (Table S1). Informed and written consent was obtained from each patient's caregiver and assent if the child was older than 7 years and competent to do so. The study was approved by the Human Research Ethics Committee of Stellenbosch University, South Africa (study no. N11/01/006).

## Gas Chromatography-Mass Spectrometry

Each CSF sample was filtered to remove bacteria—to ensure safety while handling—as well as for the removal of proteins and other macromolecules. Filtration was done using the Sartorius Centrisart <sup>R</sup> 1 10-kDa centrifugal unit by centrifugation of CSF samples for 15 min at 3,000 rpm. Samples were prepared using the commercially available EZ:faastTM analysis kit (Phenomenex) for the analysis of free (physiological) amino acids by GC-MS. See SI for full details on sample preparation, as well as instrument details.

In conjunction with the CSF samples, several external quality control (EQ) samples were included—commercial lyophilized human serum samples spiked with known concentrations of amino acids (Fowler et al., 2008). Calibration standard (CS) samples—commercial standard mixtures of amino acids as part of the EZ:faastTM analysis kit, were also included. The CS samples were used to quantify each block of samples (e.g., a calibration curve was created using CS2 and CS3 and this curve was used to quantify samples in block ➀, CS5 and CS6 were used to calibrate EQ3, and so on). Each block of samples consisted of either seven or eight randomized experimental samples, with a total of 10 blocks. The experimental run was as follows:

Blank||CS1|EQ1A|EQ1B|CS2|➀|CS3|EQ2|CS4|➁|CS5|EQ3| CS6|➂|CS7|EQ4|CS8|➃|CS9|EQ5|CS10|➄|CS11|EQ6|CS12|➅|

CS13|EQ7|CS14|➆|CS15|EQ8|CS16|➇|CS17|EQ9|CS18|➈|CS19| EQ10|CS20|➉|CS21|EQ11|EQ1C|EQ1D|CS22||end

## Data Analysis

The GC-MS data were deconvoluted, identified and annotated using NIST spectral libraries, and quantified using AMDIS (Automated Mass Spectral Deconvolution and Identification System). These data were amalgamated into one (n × m) data matrix composed of m amino acid identities and n sample identities. Each entry consisted of a concentration value calculated in micromoles per liter (µmol/l). Statistical analyses were performed using The Unscrambler <sup>R</sup> X (V10.4, CAMO software AS, Norway) and the online metabolomics suite, Metaboanalyst 3.0 (www.metaboanalyst.ca) (Xia et al., 2015). Unsupervised principal component analysis (PCA) and Hotelling's T 2 -test, with a confidence level of 95%, were used to remove case outliers—in two TBM and two control cases. The final data matrix consisted of 29 amino acids and 67 cases (TBM = 33, control = 34). Shifted log transformation and autoscaling was applied for multivariate analyses.

## RESULTS

## Data Quality Assessment

The entire experiment was run as one batch to ensure no batch effects. EQ samples were analyzed at equal intervals throughout the run to assess the quality of the GC-MS data produced. The first EQ (EQ1) was injected four times—twice at the start (EQ1A and EQ1B) and twice at the end (EQ1C and EQ1D) of the run—to assess the stability of each analyzed amino acid over time as the entire run time of the experiment was approximately 22 h. Each sample was pre-loaded onto an autosampler at room temperature. The means of the first two repeat EQ1 samples were compared with those of the last two repeat EQ1 samples. The results of the data quality assessment are given in the Table S2. Overall, the derivatized samples remained stable while within the autosampler. The only significantly notable amino acid concentration differences were those of cystathionine, which dropped by 42%, and of cystine and leucine, which increased by 34 and 40%, respectively. However, manual inspection of the data showed that the first value of cystine (EQ1A) was extremely low (an outlier); removal of this outlier revealed that cystine changed by only 3.42% over the entire run, indicating that the associated increase was a false assessment. By examining the EQ samples over the complete run and comparing them with ERNDIM consensus values (Fowler et al., 2008), we found that asparagine and cystine fell within the expected 95% confidence intervals—indicating very good reliability (indicated by the green section in Table S2). The amino acids with poor performance (see red section in Table S2) were glutamine, leucine, ornithine, tryptophan, and valine; glutamine demonstrated the greatest variability over the experimental run. The remaining 15 amino acids (alanine, alpha-aminobutyric acid, aspartic acid, cystathionine, glutamic acid, glycine, histidine, isoleucine, lysine, methionine, phenylalanine, proline, serine, threonine, and tyrosine) fell within the lab expected ranges of

the ERNDIM consensus value and therefore had good reliability (highlighted in yellow in Table S2).

## Identification of the Important Discriminatory Amino Acids

(circled) variables.

**Figure 1** illustrates the application of unsupervised PCA and supervised partial least squares discriminant analysis (PLS-DA). Based on these analyses, the two groups can be differentiated based purely on amino acids, but not completely separated. Within the PLS-DA correlation loadings plot there are 11 amino acids (circled on the figure)—identified as being the most important discriminatory metabolites based upon leave-one-out cross-validation (R <sup>2</sup> = 63.9%, Q <sup>2</sup> = 55.3%). These 11 amino acids are: alanine, alpha-aminobutyric acid, asparagine, glycine, hydroxylysine, lysine, ornithine, proline, serine, threonine, and valine. Soft independent modeling of class analogy (SIMCA) analysis was performed (**Figure 2**) based on the PLS-DA model, indicating three misclassifications for the TBM group. Hence, based on the prediction value of our PLS-DA model, three TBM cases were misclassified as control cases. The strength of this model is that there are no false positive classifications.

Univariate statistical analyses were also performed using Mann–Whitney p-values and fold changes. The results of these univariate analyses are illustrated in the volcano plot (**Figure 3**). Amino acids with significant values (Mann–Whitney p < 0.05 and fold change > 1.0) are indicated, namely, alanine, asparagine, glycine, histidine, lysine, proline, hydroxylysine, and valine.

Based on the combination of both multivariate and univariate analyses, therefore, the most important, common, discriminatory amino acids of the control and TBM cases are: alanine (TBM: 77.3 ± 52 µmol/l, control: 25 ± 9.5 µmol/l, ref.<sup>1</sup> : 11–53.5 µmol/l); asparagine (TBM: 19.9 ± 14 µmol/l, control: 5.3 ± 1.7 µmol/l,

<sup>1</sup>Reference ranges given for children (0–10 years) obtained from the Human Metabolome Database, based upon values reported values from the University of British Columbia, B.C.'s Children's Hospital Biochemical Genetics Lab.

ref.: 4–10 µmol/l); glycine (TBM: 58.2 ± 31.7 µmol/l, control: 24.4 ± 1.4 µmol/l, ref.: 1–10 µmol/l); lysine (TBM: 36.5 ± 25.2 µmol/l, control: 14.5 ±5.2 µmol/l, ref.: 6–37 µmol/l); and proline (TBM: 24.3 ± 25.8 µmol/l, control: 1.8 ± 0.5 µmol/l, ref.: 0–3 µmol/l) (shown as box plots in **Figure 4**). From the quality control assessment, these five amino acids indicate good reliability and hence are confidently identified as the important discriminatory variables for this study. It is important to note that hydroxylysine and valine were present as important amino acids but valine failed in the EQ reliability test and hydroxylysine was not part of the quality control assessment, hence neither of the last two was included in the discussion.

## DISCUSSION

The present targeted GC-MS data correspond with our previous untargeted NMR study (Mason et al., 2015), which highlighted alanine, lysine and branched-chain amino acids as important as well as other, non-amino acid metabolites. Similar to our earlier untargeted NMR study, the five discriminatory metabolites identified here are associated with perturbed energy metabolism. As previously underscored in a review of lactate shuttles (Mason, 2017): "Understanding brain energy metabolism—neuroenergetics—is becoming increasingly important as it can be identified repeatedly as the source of neurological perturbations." Here, alanine is closely linked

to lactate and the associated shuttling systems. The data of the study presented here further highlight the importance of understanding neuroenergetics when the brain is in crisis. Also, alanine, like glycine, is an inhibitory neurotransmitter in the brain. Furthermore, as discussed in Mason et al. (2015), elevated lysine seen in CSF from TBM patients has been associated

with mental retardation and in other motor neuron diseases (Shaw et al., 1995); it has also been shown to form adducts with other compounds, such as acrolein–lysine, a marker of lipid peroxidation in childhood meningitis (Tsukahara et al., 2002).

Asparagine, synthesized from central metabolic pathway intermediates, has the precursor oxaloacetic acid—an intermediate of the energy (Krebs) cycle. Asparagine is important in the metabolism of toxic ammonia in the body through the action of asparagine synthase, which attaches ammonia to aspartic acid in an amidation reaction. According to Qureshi et al. (1998), nitrites are significantly elevated in TBM cases. Hence, increased nitrogen excretion is expected in TBM patients in the form of elevated ammonia. This nitrogen excretion by-product is thus expected to be removed from the brain with asparagine as the carrier. Of special note, all five identified discriminatory amino acids are either directly or indirectly linked to ammonia. Asparagine is also involved in the glutamate-glutamine cycle, in which the enzyme asparagine synthetase produces asparagine, AMP, glutamate, and pyrophosphate from aspartate, glutamine and ATP. In the asparagine synthetase reaction, ATP is used to activate aspartate, forming beta-aspartyl-AMP. Glutamine donates an ammonium group, which reacts with beta-aspartyl-AMP to form asparagine and free AMP. Although glutamic acid and glutamine were not identified as significantly perturbed in this study, it can be postulated that a significant increase in TABLE 1 | Summary of concentrations of alanine, asparagine, glycine, lysine, and proline as reported for TBM in this study, compared to previously reported values for Alzheimer's and schizophrenia (concentrations given as µmol/l).


asparagine can still be linked to a slightly perturbed glutamateglutamine cycle in conjunction with increased ammonia—being particularly toxic in the brain as it alters neurotransmitter homeostasis. This postulate can be extended to proline, which is biosynthetically derived from glutamate and its immediate precursor, 1-pyrrole-5-carboxylate. Proline is a proteinogenic amino acid and is incorporated into proteins by prolyl-tRNA. Since elevated protein levels in CSF is a classic hallmark of TBM, it is not surprising to find proline at elevated levels. Proline is, however, used mainly for collagen formation and is important in maintaining good health in muscle, joints, and tendons. Perturbed levels of proline are normally associated with degenerative arthritis. Increased turnover of proline was, however, found in probable Alzheimer's disease cases (Fonteh et al., 2007) without arthritis, possibly reflecting increased brain degeneration.

In a study on the change in CSF amino acid profiles following an acute tonic-clonic seizure, Rainesalo et al. (2004) measured amino acid levels in CSF before and after a seizure. Rainesalo et al. reported an increase in alanine (21.5 ± 1.2 to 23.1 ± 1.4 µmol/l), asparagine (6.2 ± 0.3 to 7.7 ± 0.5 µmol/l), and glycine (5.8 ± 0.5 to 7.4 ± 0.7 µmol/l), and a slight decrease in lysine (17.7 ± 0.9 to 17.1 ± 0.9 µmol/l)—none statistically significant. Understandably, an acute tonic-clonic seizure (full body, grand mal, seizure) is a severe physiological response, yet Rainesalo et al. reported only slight changes in most amino acids (significant changes in taurine, ornithine, and phenylalanine). **Table 1** compares the five discriminatory amino acids identified in our study, as being most important in TBM, with those of studies of a chronic neuroinflammatory disease, Alzheimer's, and a psychiatric disorder, schizophrenia. This comparison indicates that these particular five amino acids occur at levels 2–10 times greater in TBM, further highlighting the severity of the pathophysiological condition of this disease.

## CONCLUDING REMARKS

In this study, we used the sensitive method of GC-MS to profile the amino acids of CSF collected from pediatric patients classified as being "definite" or "probable" cases of TBM. Through stringent quality assessment measures and statistical analyses

## REFERENCES


(univariate and multivariate) we identified five amino acids in our experimental TBM patients as being reliable and having strong discriminatory power. The concentrations of the amino acid markers of TBM here occur at levels many times greater than other neuropathological conditions, highlighting the severity of TBM. Our latest observations justify the need for more detailed studies on biochemical markers of TBM to be validated in largerscale, and preferably multi-institutional, follow-up investigations as amino acid profiling has potential in aiding in earlier diagnosis, and hence crucial earlier treatment of TBM.

## AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

## ACKNOWLEDGMENTS

Research funding was provided by the Technological Innovation Agency of the Department of Science and Technology of South Africa.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fnins. 2017.00534/full#supplementary-material


meningitis using simple clinical and laboratory parameters. Diagn. Microbiol. Infect. Dis. 55, 275–278. doi: 10.1016/j.diagmicrobio.2006.01.027

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Mason, Reinecke and Solomons. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# White Matter Abnormalities in Children with HIV Infection and Exposure

Marcin Jankiewicz <sup>1</sup> \*, Martha J. Holmes <sup>1</sup> , Paul A. Taylor <sup>2</sup> , Mark F. Cotton<sup>3</sup> , Barbara Laughton<sup>3</sup> , André J. W. van der Kouwe<sup>4</sup> and Ernesta M. Meintjes <sup>1</sup>

<sup>1</sup> Division of Biomedical Engineering, Department of Human Biology, University of Cape Town, Cape Town, South Africa, <sup>2</sup> Scientific and Statistical Computing Core, National Institutes of Health, Bethesda, MD, United States, <sup>3</sup> Stellenbosch University, Stellenbosch, South Africa, <sup>4</sup> Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, MA, United States

Background: Due to changes in guidelines and access to treatment, more children start combination antiretroviral therapy (ART) in infancy. With few studies examining the long-term effects of perinatal HIV infection and early ART on neurodevelopment, much is still unknown about brain maturation in the presence of HIV and ART. Follow-up studies of HIV infected (HIV+) children are important for monitoring brain development in the presence of HIV infection and ART.

Methods: We use diffusion tensor imaging (DTI) to examine white matter (WM) in 65 HIV+ and 46 control (HIV exposed uninfected (HEU) and HIV unexposed uninfected (HU)) 7-year-old children. This is a follow up of a cohort studied at 5 years, where we previously reported lower fractional anisotropy (FA) in corticospinal tract (CST) and mean diffusivity (MD) increases in inferior/superior longitudinal fasciculi (ILF/SLF), inferior fronto-occipital fasciculus (IFOF) and uncinate fasciculus (UF) in HIV+ children compared to uninfected controls. In addition, we also found a difference in FA related to age at which ART was initiated.

Results: At 7 years, we found two regions in the left IFOF and left ILF with lower FA in HIV+ children compared to controls. Higher MD was observed in a similar region in the IFOF, albeit bilaterally, as well as multiple clusters bilaterally in the superior corona radiata (SCR), the anterior thalamic radiation (ATR) and the right forceps minor. Unlike at 5 years, we found no impact on WM of ART initiation. In HEU children, we found a cluster in the right posterior corona radiata with higher FA compared to HU children, while bilateral regions in the CST demonstrated reduced MD.

Conclusions: At age 7, despite early ART and viral load (VL) suppression, we continue to observe differences in WM integrity. WM damage observed at age 5 years persists, and new damage is evident. The continued observation of regions with lower FA and higher MD in HIV+ children point to disruptions in ongoing white matter development regardless of early ART. Lastly, in HEU children we find higher FA and lower MD in clusters in the CST tract suggesting that perinatal HIV/ART exposure has a long-term impact on WM development.

Keywords: HIV, DTI, HIV exposure, HIV infection, children

#### Edited by:

Nilesh Bhailalbhai Patel, University of Nairobi, Kenya

#### Reviewed by:

Dong-Hoon Lee, University of Sydney, Australia Francesco Sammartino, The Ohio State University Columbus, United States Marzia Duse, Sapienza Università di Roma, Italy

> \*Correspondence: Marcin Jankiewicz m.jankiewicz@gmail.com

Received: 31 July 2017 Accepted: 20 September 2017 Published: 29 September 2017

#### Citation:

Jankiewicz M, Holmes MJ, Taylor PA, Cotton MF, Laughton B, van der Kouwe AJW and Meintjes EM (2017) White Matter Abnormalities in Children with HIV Infection and Exposure. Front. Neuroanat. 11:88. doi: 10.3389/fnana.2017.00088

## 1. INTRODUCTION

The treatment of childhood HIV infection has changed dramatically over the past 10 years. Changes in guidelines and access to treatment have increased the number of children beginning combination antiretroviral (ARV) therapy (ART) in infancy. The introduction of ART in infancy and early childhood may influence the effect of the virus, and may also have a direct effect on the child's maturing central nervous system. Since few studies have examined the long-term effects of perinatal HIV infection and early ART on neurodevelopment, little is known about brain maturation in the presence of HIV and ART. Neuroimaging studies have identified HIV associated differences in children, including regional and global white matter (WM) atrophy (van Arnhem et al., 2013; Sarma et al., 2014; Cohen et al., 2016; Yadav et al., 2017). Although WM continues developing well into adulthood (Reiss et al., 1996), the greatest increases in organization are typically within the first 10 years of growth (e.g., Bashat et al., 2005; Lebel et al., 2008; Lebel and Beaulieu, 2011), highlighting the importance of identifying damage or delayed development before adolescence.

Diffusion tensor imaging (DTI) is a non-invasive MRI technique that provides quantitative measures of WM microstructure. In childhood, fractional anisotropy (FA) typically increases with age and has been associated with increased myelination, greater axonal count and axonal density (Giedd et al., 1999; Filippi et al., 2002; Barnea-Goraly et al., 2005; Brouwer et al., 2012; Lentz et al., 2014). Additional DTI measures include mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD). MD is associated with structural organization, with lower values indicating well-organized structure and denser axonal packing (Feldman et al., 2010). In children RD values tend to decrease with age, which is interpreted as representing increased myelination and/or more densely packed axons; conversely, AD values tend to increase with age, which has been interpreted as improved fiber coherence as axons straighten and merge within the bundle (Lebel and Beaulieu, 2011).

DTI studies in HIV infected (HIV+) children and adolescents on ART have found lower whole brain (Uban et al., 2015) and regional FA, including the corpus callosum (CC), fornix, corona radiata, frontal and parietal WM, pre-/postcentral gyrus, and superior longitudinal fasciculus (Hoare et al., 2015; Li et al., 2015). Reductions in FA were attributed to RD increases, suggesting effects of HIV on myelination. However, most children in these studies initiated ART after 2 years of age, according to concurrent guidelines or delayed diagnosis. Although early ART is recommended to limit the effects of HIV infection in children, it is not yet known to what extent early ART can prevent or reverse HIV-associated WM damage.

Here we present DTI results in HIV+ children at age 7 years (7.01–7.84 years) from the CHER (Children with HIV Early Antiretroviral Therapy) clinical trial (Violari et al., 2008; Cotton et al., 2013) in follow-up at the Family Infectious Diseases Clinical Research Unit, Tygerberg Children's Hospital, Cape Town. These children initiated ART before 18 months of age and have been followed since enrollment under 12 weeks of age. We have previously reported lower FA in corticospinal tract and MD increases in inferior/superior longitudinal fasciculi (ILF/SLF), inferior fronto-occipital fasciculus (IFOF), and uncinate fasciculus (UF) in these children compared to uninfected controls at age 5 years (Ackermann et al., 2016), indicating that early ART may not fully protect WM development. Follow-up studies of infected children are particularly important for monitoring brain maturation in the presence of both HIV infection and ART, and the results can be used to identify critical time points for intervention.

Importantly, compared to Ackermann's study (Ackermann et al., 2016), the present study has an increased sample size and a group of HIV exposed, uninfected (HEU) children. While HEU children tend to be healthier than HIV+ children, studies find increased rates of infections and developmental delays compared to their HIV unexposed uninfected (HU) peers. Due to the increased success of preventing mother to child transmission (PMTCT) programs, there is a growing population of HEU children in regions with high HIV prevalence, such as South Africa. It is unclear if the reasons behind the increased risks are related to maternal HIV infection, perinatal ART exposure, environmental factors or a combination thereof. Two recent studies used DTI to study HIV exposure, with one finding no exposure differences (Jahanshad et al., 2015) while the other identified a region in which HEU infants had higher FA (Tran et al., 2016) then infants.

Here we examine group differences in WM integrity between HIV+ and uninfected children at age 7 years, and between HEU and HU children. In addition, within the HIV+ children we look at possible treatment differences between those who initiated ART before and after 12 weeks of age. Based on previous results from cohort at an earlier age and on the other studies in the literature, we hypothesized that HIV+ children would demonstrate several WM regions of reduced FA and increased MD compared to uninfected controls, while comparisons of HEU to HU children would not reveal any differences. Further, we also hypothesized that children who initiated ART before 12 weeks of age would show fewer WM alterations compared to controls than those who started ART later.

## 2. MATERIALS AND METHODS

## 2.1. Participants

Participants were 72 Xhosa and Cape Colored 7-year old HIV+ children from the CHER trial (Violari et al., 2008; Cotton et al., 2013) and 56 uninfected age-matched controls from the same community recruited as part of a parallel vaccine study (Madhi et al., 2010). As part of the CHER trial, infants with CD4 percentage (CD4%) of at least 25% were randomized to one of the following three treatment arms: ART-Def (ART deferred until CD4% < 25% in first year or CD4% < 20% thereafter, or if clinical disease progression criteria presented); ART-40W (ART initiated before 12 weeks of age and interrupted after 40 weeks); and ART-96W (ART initiated before 12 weeks of age and interrupted after 96 weeks). ART was restarted for any child in the in ART-40W and ART-96W groups if CD4% declined or clinical evidence of disease progression was observed. Since several children in the ART-Def arm met criteria for almost immediate initiation of ART, the children were grouped here based on age at treatment initiation, specifically those who received ART at or before 12 weeks (before-12wk) and those who received treatment after 12 weeks (after-12wk).

## 2.2. Image Acquisition

All children were scanned without sedation on a 3 Tesla Siemens Allegra MRI (Erlangen, Germany) at the Cape Universities Brain Imaging Centre (CUBIC) in South Africa with a single channel head coil according to protocols that had been approved by the Human Research Ethics Committees of the participating institutions. Parents/guardians provided written informed consent and children oral assent. Children were first familiarized with the scanning procedures on a mock scanner.

For each child we acquired a structural T1-weighted (T1w) volume using a volumetric navigated (Tisdall et al., 2012) multiecho magnetization prepared rapid gradient echo (MEMPRAGE) sequence (van der Kouwe et al., 2008) (voxel size = 1.3 × 1.0 × 1.0 mm<sup>3</sup> , FOV = 224 × 224 × 144 mm<sup>3</sup> , TR = 2,530 ms, TI = 1,160 ms, TEs = 1.53/3.19/4.86/6.53 ms, flip angle = 7 ◦ ) and a pair of diffusion weighted data sets with opposite phase encodings (here, anterior-posterior and posterior-anterior, AP-PA) for EPI distortion correction during processing (Irfanoglu et al., 2012) using a volumetric navigated (Alhamud et al., 2012) twice-refocused spin echo sequence (voxel size = 2.0 × 2.0 × 2.0 mm<sup>3</sup> , matrix size = 112 × 112 × 72, FOV= 224 × 224 × 144 mm<sup>3</sup> , TR/TE = 10,000/86 ms, 30 noncollinear gradient directions with DW factor b = 1, 000 s mm−<sup>2</sup> , and four non-DW b = 0 s mm−<sup>2</sup> (b0) acquisitions).

## 2.3. Image Preprocessing

Structural and DWI images were converted from the DICOM format to NiFTi format using the dcm2nii tool (http://www. cabiatl.com/mricro). The DWI data were first inspected visually for the presence of volumes with large motion artifacts or dropout slices, which were removed prior to processing. Motioncorrupted volumes were removed from both AP and PA acquisitions. Only subjects with more than 15 diffusion directions remaining were retained for subsequent analyses. DWI data were then corrected for motion, eddy current and EPI distortions using the DIFF\_PREP and DR-BUDDI tools within TORTOISE v.2.5.2 (Pierpaoli et al., 2010; Irfanoglu et al., 2015). For the purposes of having a b0-like contrast for default TORTOISE processing, the inversion of the relative contrast of tissues (IRCT) method was applied to each subject's T1w volume to generate a T2w-like contrast volume (see Appendix A of Taylor et al., 2016); the T2w-like volumes were only used for registration within TORTOISE. After preprocessing, the diffusion tensors (DTs) and associated parameters (FA, MD, etc.) were computed using 3dDWItoDT in AFNI (Cox, 1996).

A standard space and WM mask for this study were created as follows. Each subject's T1w structural image was non-linearly warped to the 2 mm isotropic Haskins pediatric template using AFNI's 3dQwarp. The warping transformation was then applied to each of the DTI parameter volumes of interest. To generate a group-level WM mask based on the subject data, the coregistered FA maps of all subjects were thresholded at FA > 0.2 and then combined together as an intersection across the subjects in order to restrict analysis to WM (Mori and van Zijl, 2002). Voxelwise comparisons of DTI measures were only performed within this WM mask.

## 2.4. Voxelwise Image Analyses

Voxelwise group comparisons were performed in FSL based on a general linear model with randomize (Winkler et al., 2014) that included gender and ethnicity as confounds. Clusters where DTI parameters differed between groups were identified from uncorrected p-value maps for 3 sets of comparisons: HIV+ vs. Controls; Before-12wk vs. After-12wk; and HEU vs. HU. To control for Type I error, Monte-Carlo simulations were performed using AFNI-3dClustsim (Forman et al., 1995). We set our cluster size threshold at pth = 0.005 and α = 0.05 (nearest neighbors NN = 3, two-sided voxelwise thresholding), which yielded a minimum cluster size of 112 mm<sup>3</sup> for Controls vs. HIV+, and 120 mm<sup>3</sup> for the HEU vs. HU and before- and after-12 wk group comparisons.

## 3. RESULTS

Data from 17 children (7 HIV+/10 controls) were excluded due to incomplete scans or the presence of motion artifacts in more than half the DTI volumes. After exclusions, we present data from 65 HIV+ children (51 before-12wk; 14 after-12wk) and 46 uninfected controls (19 HEU; 27 HU) (**Table 1**). Notably, in all except 9 children, VL was suppressed by age 2 years. At the time of scan, VL was suppressed in all but 3 HIV+ children (95%).

## 3.1. Controls vs. HIV+

Two regions in the left inferior fronto-occipital fasciculus (IFOF) and left inferior longitudinal fasciculus (ILF), respectively, showed lower FA in HIV+ children compared to controls. Higher MD was found in a similar region in the inferior frontooccipital fasciculus (IFOF), albeit bilaterally, as well as multiple clusters bilaterally in the superior corona radiata and the anterior thalamic radiation (ATR), and right forceps minor (**Table 2** and **Figure 1**). These FA decreases and MD increases were largely attributable to higher RD in HIV+ children.

## 3.2. Before-12wk vs. After-12wk

No regions showed FA or MD differences based on when treatment was initiated. To confirm that the potential benefit of earlier treatment in the before-12wk children group was not compromised by interruption, we also compared the after-12wk group separately to children in the before-12wk group on continuous and interrupted treatment, respectively, and examined associations of FA with age of ART initiation adjusting for sex, ethnicity and duration of interruption. No regions showed significant effects for any of these analyses.

## 3.3. HEU vs. HU

A cluster in the right posterior corona radiata had higher FA in HEU children than in HU children, while regions in the right- and left-corticospinal tract showed lower MD (**Table 3** and **Figure 2**).




Values are mean ± standard-deviation; VL, plasma viral load (RNA copies/ml). <sup>a</sup>Data were not available for 6 children.

<sup>b</sup>Age and duration of ART interruption mean and standard deviation based only on children in whom treatment was interrupted (N = 36).

<sup>c</sup>Median and interquartile range.

FIGURE 1 | Clusters showing lower FA (red) and higher MD (blue) in HIV infected children compared to controls. The clusters were overlayed on masks of the corresponding tracts (JHU White-Matter Tractography Atlas).

TABLE 2 | Peak (top) and center of gravity (COG) (bottom) MNI coordinates of clusters showing significant differences in FA or MD between Controls and HIV+ children.


Group means (stdev) of FA and MD are shown, as well as AD and RD, for each cluster. Units of MD, AD, and RD are 10−3mm<sup>2</sup> s −1 . Values are Mean (SD).

## 4. DISCUSSION

This study presents DTI findings at age 7 years in an expanded group of children who had been scanned at age 5 (Ackermann et al., 2016). Similar to earlier findings, our results point to alterations in WM microstructure in the presence of HIV infection despite early ART. Numerous regions were identified with lower FA or higher MD in HIV infected children compared to their uninfected peers. Although AD showed significant group differences in most regions, similar to the findings at 5 years, effects were largely attributable to RD increases in infected children, pointing to regional myelin damage, reduced myelination or myelin loss (Alexander et al., 2007). Contrary to the results at 5 years, we did not find differences based on timing of treatment initiation suggesting that by age 7 there is no protective effect in WM from starting treatment before or after 12 weeks. In addition, our analysis here revealed one cluster with higher FA and two with lower MD in HEU children compared to HU children demonstrating differences related to ART/HIV exposure in children.

## 4.1. Controls vs. HIV+

In contrast to other studies reporting FA reductions in the corpus callosum (CC) in children and youths on ART (Hoare et al., 2015; Li et al., 2015), we continue, as at 5 years (Ackermann et al., 2016), to find no evidence of microstructural CC damage at age 7 years in these children who all initiated ART by 18 months of age. The lack of HIV associated WM damage in the CC suggests that early treatment is neuroprotective to the CC. Children in the studies reporting CC damage initiated ART when clinically indicated.

However, despite early ART and VL suppression in 86% of the children by age 2 years, we continue to observe diffuse differences in WM integrity. Similar to our findings at 5 years (Ackermann et al., 2016), we again find HIV associated WM damage in the IFOF, ILF, and forceps minor, indicating that damage in these tracts may occur early during infection and persist during childhood. This concurs with spectroscopy findings from the same cohort studied here at age 5 years, where basal ganglia metabolite levels (choline, NAA) were associated with CD4/CD8 at enrollment (Mbugua et al., 2016).

A recent study of HIV+ adolescents on ART showed that WM integrity in these same three tracts were associated with measures of HIV disease severity (Uban et al., 2015). In particular peak VL was associated with reduced FA in the right IFOF, which partially mediated the effect of higher peak VL on poorer working memory performance (Uban et al., 2015). In addition, higher peak VL was related to higher streamline count (i.e., the number of fiber bundles) in the left ILF, which the authors interpret as a compensatory mechanism to deal with the impact of HIV and/or ART in this region. We found no evidence of compensation in this tract, but a loss of WM integrity. Disease severity (nadir CD4%) was also associated with AD and MD in the forceps minor (Uban et al., 2015), a fiber bundle that connects the lateral and medial surfaces of the frontal lobes and crosses the midline via the genu of the corpus callosum, and is responsible for interhemispheric sensory and auditory connectivity.

In addition to the above regions, we find MD increases at age 7 years in several clusters bilaterally along the superior corona radiata (SCR) and ATR that were not evident at 5 years. Two of the regions in the SCR are in similar locations, albeit contralateral, to the parietal corticospinal tract (CST) cluster where lower FA was found at age 5 years. Other regions, specifically the uncinate fasciculus (UF) and internal capsule and brain stem regions of the CST, showed WM damage at 5 years (Ackermann et al., 2016) but not at 7 years. These findings suggest that WM developmental delay in some regions may resolve, while other regions (viz. the SCR and ATR) may be sensitive to ongoing HIV infection and/or ART exposure.

While we found HIV-related RD increases in the SCR, Hoare et al. (2015) reported RD increases in the anterior corona radiate in children aged 6–16 years on ART. In infected children aged 13–17 years on ART, Li et al. (2015) found lower FA in the superior and posterior corona radiata, frontal and parietal WM, pre-/postcentral gyrus, and superior longitudinal fasciculus (SLF), all due to RD increases.

## 4.2. Before-12wk vs. After-12wk

Previously, at age 5 years, one cluster was found in which the before-12wk group demonstrated lower FA compared to the after-12wk group (Ackermann et al., 2016). The difference was attributed to the children in the before-12wk group whose treatment was interrupted, pointing to possible harmful effects of treatment interruption. Here we did not observe any differences based on timing of treatment initiation, indicating that HIV associated damage occurs either very early during infection (for example, in the IFOF, ILF and forceps minor) and as such affects children initiated before and after 12 weeks similarly, or later in development (for example, SCR and ATR) when all children are impacted equally.

In contrast to our findings, Li et al. (2015) found that longer ART duration and earlier age of treatment initiation was associated with lower frontal FA in infected youths. In their study, children initiated treatment when clinically indicated. As such, sicker children, in whom one might expect the most WM damage, would have initiated ART earlier and would have been on ART for longer. In our study, children were randomized to receive ART before or after 12 weeks, so that timing of ART initiation is not related to disease severity. Notably, we observe increasing damage in frontal WM (viz. ATR) in our children from 5 to 7 years, which overlaps with the time when children in (Li et al., 2015) start to initiate ART (age of ART initiation: 50–190 months). These findings suggest that frontal WM may be more vulnerable to ongoing HIV infection over this period of development, or that an early insult negatively impacts later development.

## 4.3. HEU vs. HU

We found one cluster in the right posterior corona radiata showing higher FA, accompanied by higher AD and lower RD, in HEU children compared to HU children. As the white matter sheet of the corona radiata is one of the first cells to form in embryos, it is possible that the observed increased FA in this region is related to in utero exposure to HIV and/or ART. TABLE 3 | Peak (top) and center of gravity (COG) (bottom) MNI coordinates of clusters showing significant differences in FA or MD between HU and HEU children.


Group means (stdev) of FA and MD are shown, as well as AD and RD, for each cluster. Units of MD, AD, and RD are 10−3mm<sup>2</sup> s −1 . Values are Mean (SD).

children. The clusters were overlayed on masks of the corresponding tracts (JHU White-Matter Tractography Atlas).

Regional increases in FA are typically interpreted as representing higher WM connectivity due to more densely packed axons, greater axon diameter or myelination. However, increased FA has been observed in pathology with differing explanations, such as accelerated maturation in autistic children (Bashat et al., 2005) and excessive, thick myelin in children with attentiondeficit/hyperactivity disorder (ADHD) (Li et al., 2010). In HEU infants, a recent study identified a region with higher mean FA in the middle cerebellar peduncles, which the authors interpret as potentially corresponding to microscopic deficits or reductions in axons (Tran et al., 2016).

In addition, we found clusters bilaterally in the corticospinal tract that demonstrated lower MD, with lower AD and RD. MD is associated with structural organization, with lower values indicating well organized structure. Denser axonal packing is thought to be related to lower MD values. In children, RD values decrease with age and is interpreted as representing increased myelination and/or more densely packed axons (Lebel and Beaulieu, 2011). The corona radiata is associated with the corticospinal tract, pointing to a possible relationship between the clusters showing exposure effects.

While a recent neuroimaging study of HEU children (Jahanshad et al., 2015) did not detect any group differences in DTI measures, the authors reported that higher FA and lower MD were each associated with higher IQ scores in both HEU and HU children. These results support the interpretation of increased connectivity with higher FA and lower MD, and suggest an absence of WM damage or delayed development in HEU children.

Further work exploring the relationship between DTI measures with other neuroimaging modalities as well as neuropsychological performance may help better understand these results.

## 4.4. Strengths and Limitations

The strength of our study is that the cohort has been followed from a young age and is well characterized. The relatively narrow age ranges over which imaging has been performed also facilitate a description of longitudinal changes. Unfortunately the cohort was not assessed for prenatal or perinatal HIV infection due to tests available at the time of birth. We also did not determine the effects of nutrition or other infections, e.g., cytomegalovirus.

## 5. CONCLUSIONS

This study presents a follow up of a cohort studied at 5 years, revealing ongoing WM alterations at age 7 years in HIV infected children compared to controls despite early ART and VL suppression. WM damage observed at age 5 years in the IFOF, ILF, forceps minor and CST persists. In addition, new WM damage is evident in multiple clusters along the SCR and ATR. The continued observation of clusters with lower FA and higher MD in HIV infected children point to disruptions in ongoing white matter development regardless of early ART. The fact that treatment initiation before or after 12 weeks does not influence WM integrity at this age further suggests that WM damage occurs either very early in infection or later in development when children initiating ART before and after 12 weeks are impacted similarly. In addition, in HEU children we find higher FA and lower MD in clusters in the CST suggesting that perinatal HIV/ART exposure has a long-term impact on WM development.

## ETHICS STATEMENT

This study was carried out in accordance with the recommendations of Human Research Ethics Committees of the participating institutions with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the Human Research Ethics Committees of the participating institutions.

## AUTHOR CONTRIBUTIONS

MJ and PT were involved in designing and performing data analyses. EM, AvK, and BL conceived, designed and obtained funding for the study. MJ, MH, EM, BL, and MC provided interpretation of data for the work. MJ and MJH drafted the

## REFERENCES


work and all other authors provided critical revision of the manuscript.

## ACKNOWLEDGMENTS

We thank the participants and their parents for being willing to take part in this study, research assistants Lungiswa Khethelo and Thandiwe Hamana for their expertise in supporting the children during neuroimaging, and the radiographers at CUBIC. This work was supported by NIH grants R01HD071664, R21MH096559, and R21MH108346; South African Medical Research Council (SAMRC); South African National Research Foundation (NRF) grants CPR20110614000019421 and CPRR150723129691; and the NRF/DST South African Research Chairs Initiative; PT: The research and writing of the paper were supported by the NIMH and NINDS Intramural Research Programs (ZICMH002888) of the NIH/HHS, USA. Support for the CHER study, which provided the infrastructure for the neurodevelopmental substudy, was provided by the US National Institute of Allergy and Infectious Diseases through the CIPRA network, Grant U19 AI53217; the Departments of Health of the Western Cape and Gauteng, South Africa; and GlaxoSmithKline/Viiv Healthcare. Additional support was provided with Federal funds from the National Institute of Allergy and Infectious Diseases, National Institutes of Health, United States Department of Health and Human Services, under Contract No. HHSN272200800014C. Permission to conduct the substudy on this cohort was granted by Doctors Avy Violari, Shabir Madhi and Mark Cotton and the CHER steering committee.

children compared to matched healthy controls. Neurology 86, 19–27. doi: 10.1212/WNL.0000000000002209


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Jankiewicz, Holmes, Taylor, Cotton, Laughton, van der Kouwe and Meintjes. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Functional Connectivity Alterations between Networks and Associations with Infant Immune Health within Networks in HIV Infected Children on Early Treatment: A Study at 7 Years

Jadrana T. F. Toich<sup>1</sup> , Paul A. Taylor 1, 2, 3, Martha J. Holmes <sup>1</sup> \*, Suril Gohel <sup>4</sup> , Mark F. Cotton<sup>5</sup> , Els Dobbels <sup>5</sup> , Barbara Laughton<sup>5</sup> , Francesca Little<sup>6</sup> , Andre J. W. van der Kouwe<sup>7</sup> , Bharat Biswal <sup>8</sup> and Ernesta M. Meintjes <sup>1</sup>

*<sup>1</sup> MRC/UCT Medical Imaging Research Unit, Division of Biomedical Engineering, Department of Human Biology, Faculty of Health Sciences, University of Cape Town, Cape Town, South Africa, <sup>2</sup> African Institute for Mathematical Sciences, Muizenberg, South Africa, <sup>3</sup> Scientific and Statistical Computing Core, National Institutes of Health, Bethesda, MD, United States, <sup>4</sup> Department of Health Informatics, School of Health Professions, Rutgers University, Newark, NJ, United States, <sup>5</sup> Family Clinical Research Unit, Department of Paediatrics and Child Health, Stellenbosch University, Stellenbosch, South Africa, <sup>6</sup> Department of Statistical Sciences, University of Cape Town, Cape Town, South Africa, <sup>7</sup> Department of Radiology, Massachusetts General Hospital, Boston, MA, United States, <sup>8</sup> Department of Biomedical Engineering, New Jersey Institute of Technology, Newark, NJ, United States*

#### Edited by:

*Nouria Lakhdar-Ghazal, Faculty of Science, Mohammed V University, Morocco*

#### Reviewed by:

*Kristina T. Legget, University of Colorado Denver School of Medicine, United States Xu Lei, Southwest University, China*

> \*Correspondence: *Martha J. Holmes martha.j.holmes@gmail.com*

Received: *31 August 2017* Accepted: *12 December 2017* Published: *11 January 2018*

#### Citation:

*Toich JTF, Taylor PA, Holmes MJ, Gohel S, Cotton MF, Dobbels E, Laughton B, Little F, van der Kouwe AJW, Biswal B and Meintjes EM (2018) Functional Connectivity Alterations between Networks and Associations with Infant Immune Health within Networks in HIV Infected Children on Early Treatment: A Study at 7 Years. Front. Hum. Neurosci. 11:635. doi: 10.3389/fnhum.2017.00635* Although HIV has been shown to impact brain connectivity in adults and youth, it is not yet known to what extent long-term early antiretroviral therapy (ART) may alter these effects, especially during rapid brain development in early childhood. Using both independent component analysis (ICA) and seed-based correlation analysis (SCA), we examine the effects of HIV infection in conjunction with early ART on resting state functional connectivity (FC) in 7 year old children. HIV infected (HIV+) children were from the Children with HIV Early Antiretroviral Therapy (CHER) trial and all initiated ART before 18 months; uninfected children were recruited from an interlinking vaccine trial. To better understand the effects of current and early immune health on the developing brain, we also investigated among HIV+ children the association of FC at 7 years with CD4 count and CD4%, both in infancy (6–8 weeks) and at scan. Although we found no differences within any ICA-generated resting state networks (RSNs) between HIV+ and uninfected children (27 HIV+, 18 uninfected), whole brain connectivity to seeds located at RSN connectivity peaks revealed several loci of FC differences, predominantly from seeds in midline regions (posterior cingulate cortex, paracentral lobule, cuneus, and anterior cingulate). Reduced long-range connectivity and increased short-range connectivity suggest developmental delay. Within the HIV+ children, clinical measures at age 7 years were not associated with FC values in any of the RSNs; however, poor immune health during infancy was associated with localized FC increases in the somatosensory, salience and basal ganglia networks. Together these findings suggest that HIV may affect brain development from its earliest stages and persist into childhood, despite early ART.

Keywords: HIV infection, fMRI, functional connectivity, resting state networks, seed-based correlation analysis, children, neurodevelopment, CD4

## INTRODUCTION

Increased access to antiretroviral therapy (ART) has transformed human immunodeficiency virus (HIV) infection from a fatal to a chronic illness. However, unlike HIV, many antiretrovirals (ARVs) do not effectively penetrate the blood-brain barrier of the central nervous system (CNS), so that the brain becomes a sanctuary site for HIV resulting in long-term damage and delayed neurodevelopment (see for example Martin et al., 2006; Smith et al., 2006; van Rie et al., 2009; Laughton et al., 2013; van Arnhem et al., 2013; Whitehead et al., 2014).

Even in the ART era, HIV infected (HIV+) children demonstrate cognitive delay and motor deficits compared to uninfected controls, along with impaired language abilities, failure to reach developmental milestones (Martin et al., 2006; van Rie et al., 2006; Koekkoek et al., 2008; Laughton et al., 2013; van Arnhem et al., 2013), and behavioral problems (Govender et al., 2011; Musielak and Fine, 2016), demonstrating the ongoing influence of the virus on the developing brain on ART.

Neuroimaging allows direct examination of how the pediatric brain is altered in the presence of both HIV and its treatment. Previous findings in HIV+ children on ART include ventricular enlargement, white matter (WM) abnormalities, cortical and subcortical volume alterations, and calcification of the basal ganglia and corpus callosum (Sarma et al., 2013; van Arnhem et al., 2013; Hoare et al., 2014; Uban et al., 2015; Cohen et al., 2016; Lewis-de los Angeles et al., 2016, 2017; Yadav et al., 2017). Within these studies, clinical, immunologic, and virologic measures were associated with volumetric measures, WM alterations, diffusivity markers, and shape deformation (van Arnhem et al., 2013; Uban et al., 2015; Cohen et al., 2016; Lewisde los Angeles et al., 2016). Since HIV penetrates the CNS during the first 3 weeks of life of perinatally HIV+ children (González-Scarano and Martín-García, 2005), which corresponds to a critical period in development, markers of early immune health, or virologic status may play an integral part in determining later neurological outcomes (Bilbo, 2013). Notably, children in these earlier studies initiated ART at different ages, mostly after 2 years of age, and often with limited viral load (VL) suppression. Earlier ART initiation and VL suppression could potentially prevent or reduce these HIV-related brain changes.

Following the landmark Children with HIV Early Antiretroviral Therapy (CHER) trial (Violari et al., 2008; Cotton et al., 2013) showing reduced infant mortality and HIV progression in infants initiating ART below 12 weeks of age compared to standard 2006 guidelines that advised initiating ART when CD4 lymphocyte percentage (CD4%) declined below 25% or for severe clinical disease (WHO, 2006; Violari et al., 2008; Cotton et al., 2013; Laughton et al., 2014), all guidelines now recommend initiating ART as soon as possible for all HIV+ infants regardless of CD4 measures, even if asymptomatic (WHO, 2013). Although early ART improves neurodevelopmental outcomes (Laughton et al., 2013; Brahmbhatt et al., 2014; Crowell et al., 2015), the long-term effects of early lifelong ART on brain development has not been established. We have found, for example, that alterations in brain WM and basal ganglia metabolism are evident at age 5 years in children from the CHER cohort despite starting ART before 75 weeks of age and VL suppression (Ackermann et al., 2016; Mbugua et al., 2016). In addition, there is concern about possible adverse effects of long-term ART including metabolic abnormalities (Vigano et al., 2010) and neurotoxicity (Robertson et al., 2012).

Resting state functional magnetic resonance imaging (RSfMRI) provides unique information regarding the functional connectivity (FC) of spatially distinct brain regions and the integrity of intrinsic resting state brain networks (RSNs). Since brain activity is measured when subjects are not performing a specific task, it greatly reduces the potentially confounding influences of attention, task performance, and language comprehension and is ideally suited to pediatric studies. It is a sensitive marker of alterations in brain development (Superkar et al., 2010; Thomason et al., 2011; de Bie et al., 2012) and disease (Greicius, 2008).

In HIV+ adults, RS-fMRI studies show reduced FC within various brain networks, including the visual (Wang et al., 2011), default mode (DM), executive control and salience networks (Thomas et al., 2013), as well as HIV-related changes in integration within the DM and executive control networks (Thomas et al., 2015), attenuated frontostriatal connectivity (Ipser et al., 2015), and both decreases (DM to dorsal attention, DM to salience, executive control to sensorimotor) and increases (executive control to salience) in internetwork correlations (Thomas et al., 2013). Conversely, Ortega et al. (2015) found similar FC within the DM network (DMN) in patients on ART and uninfected controls, and higher FC within the ventral attention network in patients on ART than those not receiving ART, suggesting that ART may mitigate HIVrelated FC alterations. Notably, partial correlations between subcortical seeds revealed no changes in subcortical connectivity in HIV+ adults on long-term ART with at least 1 year of undetectable plasma HIV ribonucleic acid (RNA) compared to uninfected controls (Janssen et al., 2017). The only RS-fMRI study performed to date in HIV+ youth, all of whom were on ART, showed associations of disease severity, characterized by higher peak HIV RNA and lower nadir CD4%, with poorer FC within the DMN, as well as decreases and increases in connectivity of seeds within the DMN to regions in the executive control, sensorimotor, salience, anterior cingulate/precuneus and visual networks (Herting et al., 2015). Peak plasma HIV RNA and nadir CD4% reflect the worst virologic status and immune health of subjects, respectively. The finding of lower within- and greater between-network connectivity, a pattern of connectivity that occurs earlier in development (Fair et al., 2007, 2009; Power et al., 2010), suggests developmental delay in youths with more advanced disease severity.

Here we use RS-fMRI to examine FC differences at age 7 years in HIV+ children from the CHER cohort compared to uninfected controls and, among infected children, associations of FC with measures of immune health. All HIV+ children initiated ART before 18 months of age and were virologically suppressed at the time of scanning. Major strengths of this study include close monitoring since birth, standardized ART regimens, recruitment from similar socio-demographic and economic backgrounds, and scanning within 6 months of their 7th birthdays. First, we hypothesized that, compared to uninfected children, HIV+ children would show reduced FC within and between the DM, executive control, somatosensory, salience and visual networks. Second, we postulated that improved immune health, measured by CD4 count and percentage in infancy and at scan, would be related to greater functional connectivity in these networks.

## METHODS

## Participants

Participants were 38 HIV+ Xhosa children (mean age ± standard deviation = 7.22 ± 0.16 years; 17 males) from the randomized CHER trial in follow-up at the Family Clinical Research Unit, Tygerberg Children's Hospital, in Cape Town, South Africa (Violari et al., 2008; Cotton et al., 2013) and 29 uninfected children (7.17 ± 0.10 years; 14 males) from an interlinking vaccine trial (Madhi et al., 2010). The two studies in parallel recruited infected (CHER) and uninfected (vaccine trial) infants from the same community in Cape Town. Inclusion criteria for both studies included birth weight >2,000 g and no CNS problems (other than due to HIV) or dysmorphic syndromes. A summary description of socioeconomic data from a subset of the cohort is published elsewhere (Holmes et al., 2017); although that study only included uninfected children, the data are representative of the community.

In the CHER trial, HIV+ infants 6–12 weeks of age with CD4% ≥25% were randomized to one of three treatment regimens: limited ART for either 40 or 96 weeks and restart when clinical and/or immunological criteria were met, or to start ART only if they became symptomatic or CD4% dropped below 20% (25% in the first year; Violari et al., 2008; Cotton et al., 2013), as per guidelines at the time (WHO, 2006). All HIV+ children had started ART before 18 months of age and received comprehensive immunological and clinical follow-up thereafter as described previously (Violari et al., 2008; Cotton et al., 2013). First line ART regimen consisted of Zidovudine (ZDV) + Lamivudine (3TC) + Lopinavir-Ritonavir (LPV/r, Kaletra) (Violari et al., 2008; Cotton et al., 2013). Children born to HIV+ mothers were exposed to treatment for prevention of mother-to-child transmission (PMTCT), mostly Zidovudine antenatally from 28 to 34 weeks and a single dose of Nevirapine (NVP) to the mother and Zidovudine for a week and a single dose of NVP to the infant. Of the 18 uninfected children, 8 were born to HIV+ mothers. Other than this single dose given to exposed infants as part of PMTCT, uninfected children never received ART.

## MRI Acquisition

Children were scanned on a 3T Allegra MRI (Siemens, Erlangen, Germany) at the Cape Universities Brain Imaging Centre (CUBIC) in Cape Town, South Africa, according to protocols approved by the Faculty of Health Sciences Human Research Ethics Committees of both the Universities of Cape Town and Stellenbosch. All parents and guardians provided written informed consent and all children provided oral assent.

T1-weighted structural images were acquired in the sagittal plane using a motion navigated (Tisdall et al., 2009) multi echo magnetization prepared rapid gradient echo (MEMPRAGE) sequence (van der Kouwe et al., 2008) with TR 2,530 ms, TEs 1.53/3.19/4.86/6.53 ms, inversion time (TI) 1,160 ms, flip angle 7 ◦ , resolution 1.3 × 1 × 1 mm<sup>3</sup> , and field of view (FOV) 224 × 224 × 144 mm<sup>3</sup> . RS-fMRI data were acquired using an interleaved multi-slice 2D gradient echo, echo planar imaging (EPI) sequence: 33 interleaved slices, slice thickness 4 mm, slice gap 1 mm, voxel size 3.44 × 3.44 × 5 mm<sup>3</sup> , FOV 220 × 220 × 164 mm<sup>3</sup> , TR/TE 2,000/30 ms, flip angle 77◦ , 180 volumes.

## RS-fMRI Processing

RS-fMRI data were preprocessed in AFNI (Cox, 1996) with a pipeline specified using the afni\_proc.py tool (see Appendix A for details). Briefly, preprocessing included: removal of the first 5 TRs; despiking; slice timing alignment; alignment to the skull-stripped structural image and nonlinear warping to 3 mm Talairach-Tournoux (TT) standard space; volume registration using 6 degrees of freedom (DOF); spatial smoothing with a Gaussian kernel of 6 mm full width at half maximum (FWHM); segmentation of the structural image into WM, gray matter (GM) and cerebrospinal fluid (CSF), and regression of the eroded WM and CSF average time series along with their derivatives; and bandpass filtering between 0.01–0.1 Hz as low frequency fluctuation (LFF) interval. Subjects were excluded if their structural or RS-fMRI data sets were of a poor image quality, contained signal dropout, or significant artifacts, or could not be aligned to the standard template. Time series were truncated to exclude suprathreshold subject motion, defined as >3 mm translation or >3 degrees rotation in any direction. Subjects with fewer than 130 time points after truncation were excluded and the time series of all remaining subjects were reduced to 130 time points to maintain equal weightings per subject.

A single, representative motion parameter was also estimated for each subject for inclusion as a control variable in the model design of the main analyses. First, the framewise displacement (FD<sup>i</sup> = q (x<sup>i</sup> − xi−1) <sup>2</sup> + y<sup>i</sup> − yi−<sup>1</sup> <sup>2</sup> <sup>+</sup> (z<sup>i</sup> <sup>−</sup> <sup>z</sup>i−1) 2 ) (Yan et al., 2013) was calculated with an in-house script for each volume relative to the previous volume using the translation parameters computed during motion correction. Then, FD<sup>i</sup> values were averaged for each participant to estimate the time series mean framewise displacement (FD). Two sample t-tests were used to compare FD values between the HIV+ and uninfected groups.

Group analyses were performed using tools within AFNI (Cox, 1996), FSL (Smith et al., 2004) and in-house scripts. Group independent component analysis (ICA) was performed to define RSNs and locations of peak FC within each network, which were subsequently used as seeds in our seed-based correlation analysis (SCA). SCA was performed to study whole brain (WB) connectivity to the areas of peak network connectivity. While ICA is useful for examining functional connectivity within networks, SCA generates seed-to-WB connectivity maps and permits an examination of FC differences that may occur between networks (as well as within networks, without the ICA-based condition of spatial independence of components).

## ICA-Generated RSNs

Standard RSNs were identified using group ICA with FSL's MELODIC function. Twenty independent components (ICs) were generated from the complete set of processed time series (i.e., from all subjects, after any exclusion criterion from quality control, etc. were applied), based on standard dimensionality reduction used in RS-fMRI studies of similar group size (Smith et al., 2009). Each IC was visually inspected and quantitatively compared to the standard set of Functional Connectome Project (FCP) template RSN maps (Biswal et al., 2010) using the 3dMatch function in FATCAT (Taylor and Saad, 2013). ICs containing known networks were thresholded at Z > 3 and binarized RSN masks created. The remaining ICs (representing non-GM tissue, subject motion, etc.) were discarded. The FSL function dual\_regression (Beckmann et al., 2009) was also used to generate FC maps(Z-scores) associated with each RSN for each individual.

## SCA-Generated WB FC Maps

For SCA, spherical seeds of 5 mm radius, constrained by the ICAgenerated RSN masks, were placed at the global peak of each ICA-generated RSN. In RSNs with large anteroposterior (AP) spread, a second seed was placed at a distant local maximum along the AP direction to explore potentially varied features of the network. Additional seeds were not selected in predominantly lateral networks, as left-right homotopy tends to be reflected in high temporal correlation along the left-right axis. The average time series of each seed was correlated with that of every voxel in the WB. The Pearson r-values from SCA were Fisher Ztransformed to generate a WB FC map for each seed for each subject.

## Statistical Analyses

FC in HIV+ and uninfected children were compared both within ICA-generated RSNs and SCA-generated WB FC maps using voxelwise two sample, unpaired t-tests with FSL randomize (Winkler et al., 2014). Among infected children, we also used FSL randomize to examine associations of FC (from dual\_regression) within the ICA-generated RSNs with measures of immune health (CD4 and CD4%) both at infancy and time of scan. Subject sex and FD were included in the model as confounding variables; subject age was not included due to the narrow age range of participants.

The significance of clusters was determined with AFNI's 3dClustSim using mixed autocorrelation function (ACF) modeling to account for the spatial smoothness of noise (Cox et al., 2017) at a voxelwise significance threshold of p = 0.005 and clusterwise significance of α < 0.05 (with 5,000 Monte Carlo simulations).

## RESULTS

Of 38 HIV+ and 29 controls, 9 (5 HIV+) were excluded due to significant ghosting artifacts or poor image quality, and 13 (6 HIV+) due to not meeting motion criteria. Therefore, our final sample included 27 HIV+ (18 female) and 18 uninfected (11 female) children (**Table 1**). Groups did not differ in age, sex, birth weight, or FD during scanning. Children initiated ART at TABLE 1 | Sample characteristics.


*Values are Mean* ± *Standard Deviation or Median (Interquartile range).*

*<sup>a</sup>Motion assessed using Framewise Displacement.*

*<sup>b</sup>9 children were interrupted around 40 weeks of age, and 3 children around 96 weeks. <sup>c</sup>CD8 missing for one child.*

a median age of 10 weeks (IQR: 8–23), and were all still on first line ART with plasma HIV RNA below detectable limits at time of scanning. Due to early ART, VLs were suppressed at a young age in all children—by 12 months in 81% of children, and by 2 years in 96% of children.

Twelve cortical and subcortical RSNs of interest were identified using group ICA (**Figure 1**). The infected and uninfected groups showed no significant FC differences within the ICA-defined RSNs. The 17 spherical seeds that were created at peak FC locations across the 12 RSNs are in **Table 2**.

Five regions in four WB FC maps showed reduced connectivity to their respective seeds in HIV+ children compared to uninfected controls. The clusters of reduced FC in HIV+ children are shown with their respective seeds in **Figure 2**, using the 3-dimensional viewer SUMA (Saad et al., 2004; Saad and Reynolds, 2012) within AFNI. Cluster sizes and peak locations are in **Table 3**, along with overlapping regions in the TT atlas determined using the whereami function in AFNI. From a seed in the posterior portion of the left (L) cingulate gyrus

in the DMN there were two significant clusters: one overlapping the L inferior frontal gyrus, and another mainly overlapping the L and R anterior cingulate and L medial frontal gyrus. A seed in the L paracentral lobule (somatosensory network) yielded a cluster overlapping the L and R cingulate gyrus and L medial frontal gyrus. A seed in the R cuneus of the posterior DMN exhibited a cluster mainly in the R inferior occipital gyrus, lingual gyrus, and middle occipital gyrus. Finally, a seed in the R middle frontal gyrus (executive control network) resulted in a cluster overlapping the R supramarginal gyrus and inferior parietal lobule.

In addition, two regions showed greater FC to their seeds in HIV+ children compared to uninfected controls. These are also shown in **Figure 2**, with accompanying information in **Table 3**. A cluster overlapping the L superior and middle temporal gyri showed greater FC in infected children to a seed in the R postcentral gyrus (motor network). A seed in the R anterior cingulate within the salience network resulted in a cluster in the L medial and superior frontal gyri and L anterior cingulate.

Among HIV+ children, no regions in any RSN showed association of FC with CD4 or CD4% at time of scan. In contrast, poorer immune health in infancy, as reflected by either lower CD4 or CD4% at enrollment (6–8 weeks), was associated with greater FC in three regions in three different RSNs, namely the basal ganglia network (R lentiform nucleus, putamen, and lateral globus pallidus), the somatosensory network (R precuneus, superior parietal lobule, paracentral lobule), and the salience network (R inferior frontal gyrus and insula). The clusters are shown within their respective networks in **Figure 3**, together with associations of average FC in these clusters with CD4 or CD4%; peak coordinates, location, and volume information are in **Table 4**.

## DISCUSSION

This study investigated HIV-associated FC changes in 7 year old children on two levels: firstly, comparing FC between HIV+ and uninfected cohorts, and secondly, examining relations of FC and HIV clinical measures within the infected group. Contrary to our first hypothesis, we found no group differences between infected and uninfected subjects within the ICA-generated RSNs. However, whole brain SCA from 17 seeds distributed across 12 RSNs revealed 5 connections with lower and 2 with



*The spheres (radius* = *5 mm) of each seed are shown in* Figure 2*. Here and below, coordinates are in RAI Dicom standard ("right," "anterior," and "inferior" have negative values).*

\**Based on seed's center in Talairach-Tournoux (TT) atlas.*

§*Based on components generated by ICA.*

*R, right; L, left; DMN, default mode network; pDMN, posterior DMN; vDMN ventral DMN. a vis1 and vis2 refer to two distinct components of the visual network (visual lingual and visual occipital) generated by ICA.*

higher connectivity in HIV+ children than controls. Most seeds were in networks previously implicated in HIV (DM, executive control, somatosensory, and salience networks). Of the connections found, all but one (L posterior cingulate to medial prefrontal cortex within DMN) were between networks. Among HIV+ children we observed no association in any of the ICA-generated RSNs with measures of immune health at time of scan. In contrast, poorer immune health in infancy was associated with localized FC increases at age 7 years in basal ganglia, somatosensory and salience networks. While we predicted association of FC in somatosensory and salience networks with immune health, the directionality of our findings is opposite to what we hypothesized.

## HIV+ vs. Uninfected FC Comparisons

The lack of observed HIV-related intra-network differences (i.e., within ICA-generated RSNs) may be due to the developing brain being characterized by less within-network but greater between-network connectivity (Fair et al., 2008; Power et al., 2010; Khundrakpam et al., 2016). It has been postulated that network regions in children are neither isolated fragments of an immature adult system nor unified into cohesive RSNs, but instead integrated into a different network structure organized by anatomical proximity (Fair et al., 2007). Focusing solely upon within-network changes without considering external relationships therefore risks missing critical details about the functional development of RSNs, as well as how specific networks interact with outside brain regions (Power et al., 2010; Khundrakpam et al., 2016) to create the large-scale brain networks essential for efficient functioning (Chen et al., 2008, 2011).

Studies in typically developing healthy children find that longdistance connections between functionally related regions tend to be relatively weak and strengthen with age, while short-distance relationships are stronger and weaken with development (Fair et al., 2009; Power et al., 2010). Synaptic pruning has been proposed as a possible mechanism for reductions in local FC, while myelination throughout childhood and adolescence could facilitate increased long-range correlations (Paus et al., 2001). Here, four of the five connections that demonstrated lower FC in infected children are between frontal and parietal regions, suggesting an HIV-related delay in long-range connection increases. Similarly, greater correlated brain activity in HIV+ children between the AC seed and L medial and superior frontal gyri may result from delay in the age-related decrease of short-range connections. Since decreased prefrontal-parietal connectivity is associated with poorer working memory capacity (Nagy et al., 2004) and performance (Olesen et al., 2004), these developmental delays may have functional consequences requiring further investigation.

While primary sensorimotor connectivity is well established by early childhood (5–8 years), paralimbic connectivity tends to mature in late childhood (8.5–11 years), and connectivity between higher order association regions only in late adolescence (15–18 years) (Khundrakpam et al., 2013). Using interregional correlations in cortical thickness as a measure of structural brain connectivity, Khundrakpam et al. (2013) found that connectivity decreased with age in primary sensorimotor regions but increased in association areas. Greater connectivity in the present study in HIV+ children at 7 years between the R postcentral gyrus in the motor network and L temporal regions in the somatosensory network could therefore reflect a delay in the age-related connectivity reductions between sensorimotor regions.

Using diffusion spectrum imaging, Hagmann et al. (2008) identified a structural core, a single integrated system from which processes in both cortical hemispheres appear to be coordinated, comprising the posterior cingulate cortex (PCC), precuneus, cuneus, paracentral lobule, isthmus of the cingulate, banks of the superior temporal sulcus, and inferior and superior parietal cortices. They further demonstrated that structural and functional connections were strongly correlated, indicating that these regions may similarly be hubs of functional connectivity. It is striking that in the present study all five connections demonstrated FC reductions in HIV+ children involve seeds or clusters located within key components of this core in the posterior medial and parietal cortex. These findings suggest that the structural core may be particularly vulnerable to the effects of HIV and/or ART. Further, since seeds were based on

#### TABLE 3 | Functional connections showing alterations in HIV infected children.


\**Based on seed center and cluster overlap within the Talairach-Tournoux (TT) atlas.*

§*Based on cluster overlap with the Functional Connectome Project networks (Biswal et al., 2010).*

*L, left; R, right; AC, anterior cingulate; DMN, default mode network; pDMN, posterior DMN.*

TABLE 4 | Regions within which immunocompromise in infancy (6–8 weeks), defined by low CD4 count and CD4%, is associated with functional connectivity increases at age 7 years.


\**Based on cluster overlap in Talairach-Tournoux (TT) atlas.*

*R, right; FC, functional connectivity.*

connectivity peaks in our ICA-generated RSNs, our results affirm the important role of these regions in functional integration.

In addition to this mainly posterior medial core, we also observed effects of HIV in medial frontal regions—rostral anterior cingulate (AC) and caudal AC clusters show lower FC to seeds in the PCC and paracentral lobule, respectively, and a seed in the R AC has greater FC to L frontal cortex. In total, four of the six distinct seeds with altered FC in the HIV+ children are medial, and two of these involve connections to medial frontal regions. Neurogenesis during prenatal development occurs in the ventricular zone in the center of the brain, from where neurons migrate radially out to the developing neocortex and connect with other neurons to establish rudimentary neural networks (Stiles and Jernigan, 2010). By the end of the prenatal period, major fiber pathways, including the thalamocortical pathway, are complete. The fact that midline brain regions appear disproportionately affected by HIV suggests that the changes causing the observed HIV-related developmental delays may be occurring early in development.

To our knowledge, only one study previously examined resting state FC in HIV+ youth, in a cohort aged 12–21 years (Herting et al., 2015). Here, WB FC was examined using SCA to 5 seeds within the DMN, but without controls. Functional connections were related to measures of disease severity, specifically peak HIV RNA and nadir CD4%. Greater HIV disease severity was related to both decreases and increases in BOLD signal correlations, and both withinand between networks. Notably, youths with more advanced HIV disease severity showed effects characteristic of a "less mature" DMN, providing additional evidence of HIV-related developmental delay. Internetwork correlations showing effects of disease severity occurred between the DMN seeds and clusters in the executive control, sensorimotor, salience, anterior cingulate/precuneus, and visual networks, with decreased functional connections between the DMN and executive and visual networks being related to worse processing speed scores (Herting et al., 2015). In the present study of 7 year olds, we similarly found HIV-related decreases and increases in FC between the DMN and salience, executive control, and visual networks, as well as lower within-DMN FC. It is noteworthy that many of the same regions are involved in these altered functional connections, specifically the medial prefrontal cortex, PCC, R lateral parietal and occipital cortices, R middle frontal gyrus, L superior frontal gyrus, as well as inferior frontal gyri albeit in different hemispheres. Our results, along with those of Herting et al. point to these regions as being at particular risk of alteration by HIV and/or ART in pediatric populations.

## Functional Connectivity Associations with Clinical Measures

In our study, we could not examine associations of FC with peak VL, as done in Herting et al. (2015). In our study peak VLs were truncated at a maximum value of 750,000 copies/mL at baseline. While one might expect timing of worst virological status (peak VL) and poorest immune health (nadir CD4%) to differentially affect FC, due to critical stages of development occurring at different times in childhood in different brain regions and networks, these timings are less meaningful in the context of our cohort where all infected children had either limited ART initiated between 6–12 weeks or deferred treatment when clinically indicated. Notably, Herting et al. (2015) controlled for age of peak VL and nadir CD4% in their analyses. In the CHER cohort where treatment was not based on disease severity but group assignment, nadir CD4% and peak VLs occurred immediately before treatment initiation for most children in whom treatment was deferred, and either at enrollment or after treatment interruption (if interrupted) in children initiating ART before 12 weeks. Therefore, we examined here the influence of immune health on brain development within the HIV+ children by observing the associations between FC and clinical measures at both study enrollment in infancy and time of scanning.

Similar to Thomas et al. (2013), who examined associations of VL and CD4 with FC measures in adults across 5 networks, we also found no regions within any of our ICA-generated RSNs showing a relationship of FC with current CD4 count or CD4%. It is possible that SCA, which assesses also between-network connectivity, could be more sensitive to detect connections affected by current immune health at this age. In contrast, poorer immune health in infancy was associated with increased FC in three right-lateralized regions in separate RSNs—basal ganglia, somatosensory, and salience networks. These findings imply that infant immune health has long-term consequences on brain development.

An MR spectroscopy study by Mbugua et al. (2016) of 5-yearold children from the same cohort similarly found that immune health measures at 6–8 weeks were related to N-acetyl aspartate (NAA) and choline levels in the basal ganglia, despite early ART, and VL suppression. The metabolite NAA is associated with neuronal density and integrity, and the result suggests that poor immune health in infancy relates to basal ganglia neuronal populations at age 5. If early HIV infection impacts basal ganglia neuronal density or integrity, neuronal activity and therefore FC in the region may be altered; however additional work is needed to directly examine possible relationships between altered FC and metabolite levels within this cohort. Notably, the basal ganglia are one of the mostly widely reported HIV-affected regions of the brain across modalities (e.g., Berger and Arendt, 2000; Moore et al., 2006; Ellis et al., 2007; Gongvatana et al., 2013).

Synaptogenesis and synaptic pruning start around 20 weeks gestational age (GA), and myelination around GA 30–32 weeks (Casey et al., 2005). These processes start in primary sensorimotor regions and sensory tracts, progressing to parietal and temporal association cortex, and finally prefrontal cortex (Khundrakpam et al., 2016). Correlated brain activity has been demonstrated in premature infants from 30 weeks GA, including in the somatosensory, visual, auditory, pDMN, and salience networks (Kiviniemi et al., 2000; Fransson et al., 2007; Redcay et al., 2007; Lin et al., 2008; Smyser et al., 2010). It is worth noting that the three networks where we found effects of immune health in infancy on RSFC are all involved in primary motor and sensory functions. The somatosensory network processes peripheral inputs and tactile sensations and is important for controlling action (Lin et al., 1996). The salience network, comprising the dorsal AC, the left and anterior right insula, and the adjacent inferior frontal gyri (Seeley et al., 2007), is important in coordinating behavioral responses (Medford and Critchley, 2010), initiating cognitive control (Menon and Uddin, 2010), and maintaining and implementing task sets (Dosenbach et al., 2006; Nelson et al., 2008). The basal ganglia network, which includes the putamen and caudate bilaterally as well as anterior parts of the thalamus (Szewczyk-Krolikowski et al., 2014), primarily regulates motor control, but also plays a role in human reasoning and adaptive function, the control of reward-based learning, sequencing, and cognitive function (Leisman et al., 2014). Since these networks support functional domains that are crucial when an infant starts to interact with his/her environment, they are amongst the first to mature and may be more sensitive to poor immune health during critical stages of development in infancy. However, it remains unclear why resulting FC would be increased at the observed stage of childhood. Connectivity increases with greater HIV disease severity were also observed by Herting et al. (2015) in youth between the R inferior temporal cortex within the DMN and the brainstem, R middle frontal gyrus (anterior cingulate/precuneus network), and R frontal pole (salience), and between the executive control and salience networks in HIV+ adults compared to uninfected controls (Thomas et al., 2013). In contrast to our finding of hyperconnectivity within networks in the children with the poorest immune health in infancy, the connectivity increases reported by the two other studies were between networks, indicating less independent brain networks in infected individuals, consistent with impairment. It is not clear whether the within-network FC increases observed here reflect an advantage or a deficit.

Since connectivity within local networks decreases with age from as young as 4–9 months (Damaraju et al., 2014), the FC increases observed at age 7 years in children with poorer immune health in infancy could be due to decreased synaptic pruning. The immune system plays a critical role in normal brain development and following injury (Merrill, 1992; Zhao and Schwartz, 1998; Hanamsagar and Bilbo, 2016), and elevated levels of cytokines and their receptors from perinatal infection have been linked with abnormal brain development and an increased risk of neurodevelopmental disorders (Urakubo et al., 2001; Pang et al., 2003; Meyer et al., 2006). The morphology and function of microglia, the primary immune cells in the brain, shift from an immature to a mature state throughout brain development in an age- and brain region-dependent manner (Bilbo, 2013). Animal models have shown that a single neonatal infection alters microglial functioning, leading to exaggerated cytokine production within the brain in response to subsequent immune challenges and an increased risk of cognitive deficits later in life (Bilbo, 2013). Since microglia are long-lived, functionally altered microglia may remain in the brain into adulthood. Among their many roles, microglia aid in synaptic pruning and regulate synaptic plasticity and function (Schafer et al., 2012; Hong et al., 2016). Following localization of C1q, the initiating protein within the classical complement cascade of the immune system, to synapses within the postnatal brain intended for elimination, microglia expressing the complement receptor for this protein are activated (Stevens et al., 2007; Schafer et al., 2012). We hypothesize that changes in the developmental trajectory of microglia arising from perinatal HIV infection and neuroinflammation in infancy alters later-life immune function, causing disruptions in synaptic pruning and connectivity increases within affected networks in childhood.

Given the overlapping functionality of the three affected networks, our findings provide impetus for further investigation of FC with motor and sensory performance measures. Such analysis may provide insight into whether the observed FC increases impact children positively, in the form of a possible compensatory mechanism, or negatively, such as delayed or impaired synaptic pruning, at this age.

## Limitations

Here, we used a voxelwise threshold of p = 0.005 during the clustering procedure. We note that more conservative voxelwise thresholding at p = 0.001 produced no significant results, likely due to the small sample size in this study. However, voxelwise thresholding with p = 0.005 showed adequate familywise error rate control when using the mixed ACF modeling (Cox et al., 2017) implemented here. In addition, because all HIV+ children were on ART it is impossible to disentangle the contributions of HIV infection and ART to our findings. Lastly, in these children we do not know whether HIV infection occurred prenatally or during birth. This knowledge would allow us to better understand the timing of damage during the fetal period and exposure to other viruses or bacteria which may have primed the immune system.

## CONCLUSION

HIV infection in conjunction with early ART alters betweennetwork connectivity in children (here, measured at age 7 years). The predominance of medial brain regions suggest that HIV affects brain development from its earliest stages. The networks implicated include DMN, somatosensory, salience, motor, basal ganglia, visual and auditory, as well as the higher-order executive control network. Weaker long-distance and stronger short-range connections in HIV+ children suggest developmental delay. Further, although no associations were found with current immune health, poor immune health during infancy is associated with localized FC increases in somatosensory, salience, and basal ganglia networks, indicating that effects of immunocompromise during critical stages of development in early infancy persist into childhood, despite early ART and viral suppression. These neurobiological alterations may contribute to cognitive problems among HIV infected children (e.g., Lewis-de los Angeles et al., 2017; Yadav et al., 2017) and require further investigation.

## AUTHOR CONTRIBUTIONS

EM, BL, and AvdK were involved in the study design and acquisition of data. JT, MH, PT, FL, SG, and BB were involved in data and statistical analyses. JT, EM, PT, and MH drafted the work

## REFERENCES


and all other authors provided critical revision of the manuscript. JT, PT, MH, EM, BL, ED, and MC provided interpretation of data.

## ACKNOWLEDGMENTS

We thank the participants and their parents for being willing to take part in this study, research assistants Lunges Khethelo and Thandiwe Hamana for their expertise in supporting the children during neuroimaging, the investigators on the CHER Plus Neuro study, and the data management teams at FAMCRU and PHRU.

This work was supported by NIH grants R01HD071664, R21MH096559, and R21MH108346; South African Medical Research Council (SAMRC); South African National Research Foundation (NRF) grants CPR20110614000019421 and CPRR150723129691; and the NRF/DST South African Research Chairs Initiative. Support for the CHER study, which provided the infrastructure for the neurodevelopmental substudy, was provided by the US National Institute of Allergy and Infectious Diseases through the CIPRA network, Grant U19 AI53217; the Departments of Health of the Western Cape and Gauteng, South Africa; and GlaxoSmithKline/Viiv Healthcare. Additional support was provided with Federal funds from the National Institute of Allergy and Infectious Diseases, National Institutes of Health, United States Department of Health and Human Services, under Contract No. HHSN272200800014C, and by the NIMH and NINDS Intramural Research Programs of the NIH.

Permission to conduct the substudy on this cohort was granted by Doctors Avy Violari, Shabir Madhi, and Mark Cotton and the CHER steering committee.


eight-year old children. Hum. Brain Mapp. 33, 1189–1201. doi: 10.1002/hbm. 21280


head micromovements on functional connectomics. Neuroimage 76, 183–201. doi: 10.1016/j.neuroimage.2013.03.004

Zhao, B., and Schwartz, J. P. (1998). Involvement of cytokines in normal CNS development and neurological diseases: recent progress and perspectives. J. Neurosci. Res. 52, 7–16. doi: 10.1002/(SICI)1097- 4547(19980401)52:1<7::AID-JNR2>3.0.CO;2-I

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Toich, Taylor, Holmes, Gohel, Cotton, Dobbels, Laughton, Little, van der Kouwe, Biswal and Meintjes. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

## APPENDIX A

The afni\_proc.py command within AFNI was used to specify all steps (or "blocks") and options for the full processing pipeline in this study. The command-based configuration provides a succinct form for generating a flexible analysis pipeline, and, unlike a GUI-generated analysis, it both ensures that identical steps are carried out across subjects and maintains an exact record of all steps for reproducibility. The afni\_proc.py command used in the present study is presented in **Table A1**, and the summary of steps is provided in the section Methods.

```
TABLE A1 | The afni_proc.py command using in AFNI (Cox, 1996).
```


*The first three variables (\$sub\_name, \$rest\_set and \$anat\_set) are respectively set for each subject's ID, resting state EPI data set and anatomical volume.*

# Perinatal HIV Infection or Exposure Is Associated With Low N-Acetylaspartate and Glutamate in Basal Ganglia at Age 9 but Not 7 Years

Frances C. Robertson<sup>1</sup> \*, Martha J. Holmes<sup>1</sup> , Mark F. Cotton<sup>2</sup> , Els Dobbels<sup>2</sup> , Francesca Little<sup>3</sup> , Barbara Laughton<sup>2</sup> , André J. W. van der Kouwe4,5 and Ernesta M. Meintjes<sup>1</sup>

<sup>1</sup> Medical Imaging Research Unit, Division of Biomedical Engineering, Department of Human Biology, Faculty of Health Sciences, University of Cape Town, Cape Town, South Africa, <sup>2</sup> Family Clinical Research Unit, Department of Paediatrics and Child Health, Tygerberg Children's Hospital and Faculty of Health Sciences, Stellenbosch University, Stellenbosch, South Africa, <sup>3</sup> Department of Statistical Sciences, Faculty of Sciences, University of Cape Town, Cape Town, South Africa, <sup>4</sup> A. A. Martinos Centre for Biomedical Imaging, Department of Radiology, Massachusetts General Hospital, Boston, MA, United States, <sup>5</sup> Department of Radiology, Harvard Medical School, Harvard University, Boston, MA, United States

Abnormalities of the basal ganglia are frequently seen in HIV-infected (HIV+) children despite antiretroviral treatment (ART) initiation during childhood. Assessment of metabolites associated with neuronal integrity or with glial proliferation can present a sensitive description of metabolic events underlying basal ganglia structural changes. We used magnetic resonance spectroscopy to examine differences in creatine, choline, N-acetylaspartate (NAA), glutamate, and myo-inositol between HIV+ children and HIVunexposed controls, as well as between HIV-exposed uninfected (HEU) children and HIV-unexposed controls at age 7 and at age 9. No differences in metabolites relative to the HIV-unexposed control group were found at age 7. However, at 9 years, both HIV+ and HEU had lower NAA and glutamate than unexposed control children. HEU children also had lower creatine and choline than control children. At age 7, lower CD4/CD8 ratio at enrollment was associated with lower choline levels. At age 9 lower CD4/CD8 at enrollment was associated with lower myo-inositol. Low NAA and glutamate at age 9, but not 7, suggest that basal ganglia neurons may be particularly affected by perinatal HIV/ART and that neuronal damage may be ongoing despite early ART and viral suppression. Reduced basal ganglia metabolite levels in HEU children suggest an effect of HIV exposure on childhood brain development that merits further investigation using neuroimaging and neurocognitive testing.

Keywords: HIV, early ART initiation, NAA, glutamate, magnetic resonance spectroscopy, basal ganglia, HIV exposure

## INTRODUCTION

Although infants with perinatal HIV infection can now expect to survive childhood, they face life-long treatment with antiretrovirals (ARVs) in order to maintain a healthy immune system. Starting antiretroviral therapy (ART) before 3 months of age yields good clinical, immunological, and developmental outcomes (Violari et al., 2008; Laughton et al., 2012; Cotton et al., 2013;

#### Edited by:

Nilesh Bhailalbhai Patel, University of Nairobi, Kenya

#### Reviewed by:

Gunther Helms, Lund University, Sweden Graham J. Galloway, Translational Research Institute, Australia

> \*Correspondence: Frances C. Robertson

frances.robertson@uct.ac.za

Received: 05 September 2017 Accepted: 03 April 2018 Published: 07 May 2018

#### Citation:

Robertson FC, Holmes MJ, Cotton MF, Dobbels E, Little F, Laughton B, van der Kouwe AJW and Meintjes EM (2018) Perinatal HIV Infection or Exposure Is Associated With Low N-Acetylaspartate and Glutamate in Basal Ganglia at Age 9 but Not 7 Years. Front. Hum. Neurosci. 12:145. doi: 10.3389/fnhum.2018.00145

**196**

Crowell et al., 2015). However, it is known that ART cannot completely reverse the neurological effects of HIV (Laughton et al., 2013; van Arnhem et al., 2013; Whitehead et al., 2014) and neurodevelopmental delay and neurocognitive deficits remain (Govender et al., 2011; Donald et al., 2014; Wilmshurst et al., 2014; Musielak and Fine, 2015). In addition, neurotoxic effects of ART itself (Robertson et al., 2012) may be detrimental to brain development.

Abnormalities of the basal ganglia are frequently seen in HIV-infected (HIV+) children and adults (George et al., 2009; Hoare et al., 2014), and subcortical structures may contain the highest levels of the virus (Tornatore et al., 1994). Neuroimaging shows that HIV+ children may have shape and volume alterations in subcortical gray matter (Lewis-de los Angeles et al., 2016; Yadav et al., 2017) and calcification of the basal ganglia (Govender et al., 2011), despite ART initiation during childhood.

Magnetic resonance spectroscopy (MRS) measures brain metabolites including creatine (Cr), involved in energy metabolism, N-acetyl-aspartate (NAA), reflecting neuronal integrity, and glutamate (Glu), the primary excitatory neurotransmitter, as well as choline (Cho) and myo-Inositol (Ins). Ins is considered a glial marker and Cho reflects cell membrane turnover – both present in larger concentrations in glial cells than in neurons. Assessment of metabolites associated with neuronal integrity — NAA and Glu, and those associated with glial proliferation — Ins and Cho, can present a sensitive description of metabolic events underlying, and even preceding, the structural changes in basal ganglia seen in HIV+ children.

Early single voxel spectroscopy (SVS) studies in children with HIV encephalopathy (HIVE) found lower NAA/Cr ratios in the basal ganglia compared to HIV+ children without encephalopathy and controls (Pavlakis et al., 1995; Lu et al., 1996). In earlier studies where most HIV+ children did not have encephalopathy, no alterations in basal ganglia NAA were found (Lu et al., 1996; Keller et al., 2004; Prado et al., 2011). More recently, Mbugua et al. (2016) reported higher absolute NAA levels in HIV+ children initiating ART before 12 weeks compared to a control group comprising mostly (80%) HIV-exposed uninfected (HEU) children.

Findings on glial metabolites in the basal ganglia are less consistent. Keller colleagues and Lu colleagues found lower Cho (Keller et al., 2004) and Cho/Cr (Lu et al., 1996) in HIV+ children than controls. However, other studies found higher basal ganglia Cho/Cr (Ashby et al., 2015) and Ins/Cr (Ashby et al., 2015) in HIV+ children, as well as higher absolute Cho (Mbugua et al., 2016) in HIV+ children who started ART before 12 weeks compared to children who initiated treatment later and a predominantly HEU control group.

Although earlier studies do not report ART use (Pavlakis et al., 1995; Lu et al., 1996), basal ganglia neurometabolite alterations reported subsequently were observed even in HIV+ children on ART (Keller et al., 2004; Gabis et al., 2006; Ashby et al., 2015; Mbugua et al., 2016). However, these studies used relatively small sample sizes (between 8 and 45 HIV+ subjects) spanning a wide age range and starting ART at varying stages of childhood. One study examined basal ganglia neurometabolites in young children who all initiated ART before 18 months of age (Mbugua et al., 2016). In that study, lower CD4/CD8 ratio at enrollment, aged 6–8 weeks, was associated with lower basal ganglia NAA and Cho at age 5 years, despite early ART initiation and viral load suppression.

The aim of this study was to investigate whether immune system impairment in infancy remains associated with lower NAA and Cho in the basal ganglia at age 7 and at age 9 in an expanded group from the same cohort studied by Mbugua et al. (2016). Because increased basal ganglia Cho and NAA in HIV+ children who initiated ART before 12 weeks was attributed to the use of primarily HEU controls (Mbugua et al., 2016), we aimed to compare HIV+ children to HIV-unexposed controls. In addition, although some studies report neurocognitive deficits in HEU children (Van Rie et al., 2008; Kerr et al., 2014), none have investigated the effects of perinatal HIV and ART exposure on neurometabolite levels in childhood. Our second aim therefore, was to investigate differences in basal ganglia neurometabolite levels between HEU and HIV-unexposed children (controls) at ages 7 and 9.

## MATERIALS AND METHODS

## Participants

Participants included 78 HIV+ children from the Children with HIV Early Antiretroviral therapy (CHER) trial (Violari et al., 2008; Cotton et al., 2013) and 53 HIV-uninfected (HIV-) children from the same community enrolled in a longitudinal neuroimaging study (Holmes et al., 2017) in Cape Town, South Africa.

On the CHER trial, infants with CD4<sup>+</sup> percentage (CD4%) of at least 25% were randomized to begin ART early (before 12 weeks of age) and have treatment interrupted after either 40 or 96 weeks, or to have ART deferred until CD4% < 20% (25% in the first year) or on presentation of clinical symptoms of disease. A small group with CD4% < 25% were randomized to the early treatment arms only.

The first line ART regimen consisted of Zidovudine (ZDV) + Lamivudine (3TC) + Lopinavir-Ritonavir (LPV/r, Kaletra <sup>R</sup> ) (Violari et al., 2008; Cotton et al., 2013). All HIV+ children started ART before 76 weeks of age and received regular clinical and immunological follow-up. All but nine children had plasma HIV RNA below detectable limits (<400 RNA copies/mL) by 2 years of age.

The HIV- children, comprising both unexposed (control) children born to HIV-seronegative mothers (N = 32) and HEU children (N = 21) born to HIV+ mothers, were recruited from a linked vaccine trial (Madhi et al., 2010). HEU children were exposed to treatment for prevention of mother-to-child transmission (PMTCT), mostly zidovudine antenatally from 28 to 34 weeks and a single dose of nevirapine (NVP) to the mother and zidovudine for a week and a single dose of NVP to the infant.

## Neuroimaging

Neuroimaging was performed without sedation according to protocols approved by the Human Research Ethics Committees of the Universities of Cape Town and Stellenbosch.

Parents/guardians provided written informed consent and children provided oral assent. A senior radiologist reviewed all structural scans, and children with abnormalities were excluded from analysis.

At 7 years, children were scanned on a 3T Allegra MRI scanner (Siemens, Erlangen, Germany) using a single channel head coil. At 9 years, a 3T Siemens Skyra MRI scanner with a 32-channel head coil was used.

On both scanners, the protocol included a high-resolution T1-weighted multiecho magnetisation prepared rapid gradient echo acquisition (MEMPRAGE; Van der Kouwe et al., 2008) FOV 224 mm × 224 mm, TR 2530 ms, TI 1160 ms, TE's = 1.53/3.19/4.86/6.53 ms, bandwidth 650 Hz/px, 144 slices, 1.3 mm × 1.0 mm × 1.0 mm) and single voxel 1H-MRS (PRESS: 1.5 cm × 1.5 cm × 1.5 cm voxel; TR 2000 ms, TE 30 ms, 64 averages) in the basal ganglia with Chemical Shift Selective (CHESS) water suppression. A water reference was acquired in the same voxel without water suppression. On the Allegra, an EPI volumetric navigated (vNav) PRESS (Hess et al., 2011) sequence was used that applies prospective motion- and shim correction throughout the acquisition and has been shown to provide high quality repeatable spectra in young children (Hess et al., 2013). The basal ganglia voxel comprised approximately 60% gray matter and anatomically represents the frontal limb of the internal capsule and part of the caudate nucleus, putamen, and globus pallidus (representative voxel placement shown in **Figure 1**). Shimming was performed over the voxel volume, first automatically using the scanner's "Advanced" adjustment, then manually if necessary to reduce the spectral linewidths reported by the scanner.

LCModel (Provencher, 2001) was used to perform eddy current correction and to calculate metabolite ratios to creatine, as well as absolute concentrations using the water scaling method (Ernst et al., 1993; Kreis et al., 1993). SPM12 software was used to segment the MEMPRAGE regions corresponding to the SVS voxel into gray matter, white matter, and cerebrospinal fluid (CSF) for partial volume correction and water concentration calculation. Spectra were eliminated if the quality was poor (signal-to-noise ratio < 7, line width at half the peak maximum > 0.07 parts per million as reported by LCModel).

## Statistical Analysis

Because metabolite levels are scanner-dependent and measured in institutional units, data from scans at different ages were treated independently with separate cross-sectional analyses at each age, to examine differences in Cr, Cho, NAA, Glu, and Ins between HIV+/HEU children and HIV-unexposed controls. Linear regression models in the R programming language (R Core Team, 2013) were used, with age at scan, gender, ethnicity, and voxel gray matter content as confounders. To ensure that results were not driven by influential outliers, all analyses were repeated excluding concentration estimates removed more than 1.5 times the interquartile range from the median value for that metabolite.

In the HIV+ group we also used linear regression analysis to examine the relationship of metabolite levels to clinical measures including CD4% at enrollment, CD4/CD8 ratio at enrollment, CD4% closest to scan and incidence of CDC stage C HIV disease.

## RESULTS

At age 7, spectra from 17 HIV+ (38%), 6 HEU (43%), and 7 control children (33%) did not meet the quality control criteria and were excluded. At age 9, data from 1 HIV+ child were excluded. Scans from 37 HIV+ children, 10 HEU children and 13 control children provided data at both 7 and 9 years. **Figure 1** shows the voxel placement and example basal ganglia spectra from the Allegra and Skyra scanners.

Demographic data for subjects included at age 7 and age 9 are presented in **Table 1**. At age 7 there was no difference in birth weight or age between groups. At the 9-year scan there was no difference in birth weight between groups, but the HIV+

children were younger than the control (p = 0.0003) and HEU (p = 0.00004) children.

Clinical data for the HIV+ children are presented in **Table 2**. Plasma HIV RNA was undetectable (<400 copies/mL) in 91% of the children (all except 4) at the 7-year and 97% (all except 2) at the 9-year scan. At the 7-year scan 3 children had not yet restarted ART after interruption, but by the 9-year scan all children were on ART.

No differences in metabolites relative to the control group were found at age 7 (**Table 3**). However, at 9 years, both HIV+ and HEU had lower NAA and Glu than control children (**Figure 2** and **Table 3**). HEU children also had lower Cr and Cho than control children.

There was no relationship between CD4% at enrollment and neurometabolites at age 7 or 9 years (**Table 4** and Supplementary Table 1). However, at age 7, lower CD4/CD8 ratio at enrollment was associated with lower Cho levels. Also, there was a weak association between lower CD4% at scan and NAA. The same relationships were not evident at age 9, but at this age lower CD4/CD8 at enrollment was associated with lower Ins, and a CDC stage C diagnosis showed a trend-level association with lower NAA and higher Ins. Relationships of CD4/CD8 ratio at enrollment to Cho at ages 7 and 9 are illustrated in **Figure 3**.

## DISCUSSION

This study presents SVS basal ganglia neurometabolite data in a larger sample of older HIV+ children than previously studied, 91% of whom had undetectable viral loads at 7 and 97% at 9 years, and all of whom had started ART before 76 weeks of age. Although we find no neurometabolite differences at age 7, lower NAA and Glu is apparent at age 9.

Contrary to findings in a subset of the same children at age 5 (Mbugua et al., 2016), we do not find an effect of immune health in infancy on basal ganglia NAA at either 7 or 9 years, suggesting that subsequent health, or other events during childhood, have a stronger influence on neuronal integrity at these older ages. However, as at age 5, higher CD4/CD8 ratio at enrollment remains associated with higher basal ganglia Cho at age 7, suggesting the effect of early immune health on glial cells may persist for some time.

TABLE 1 | Biographical data for HIV+, HIV-exposed uninfected (HEU), and HIV-unexposed uninfected controls scanned at 7 and 9 years.


Values are mean ± standard deviation. <sup>a</sup>Data missing for 1 subject.

TABLE 2 | Clinical data for HIV+ children scanned at 7 and 9 years.


Values are number (percentage of group), mean ± standard deviation, or median (interquartile range) and range when specified. Age and duration of interruption data only for children with treatment interruption. CDC classification data missing for 1 subject. <sup>a</sup>Median interval 63 days from 7 year scan; median interval 22 days from 9 year scan.

The p-values presented here should, however, be interpreted with caution, as due to the number of statistical tests the family-wise error rate is greater than 5%. Findings significant at p < 0.05 would not have survived correction for multiple comparisons. Although the observed associations may therefore represent type I errors, they nevertheless present a starting point for investigation in future studies.

## Decreased NAA and Glutamate in HIV+ Children at Age 9

Similar to previous findings in children without HIVE, at age 7 we find no differences in NAA or other metabolites between


TABLE 3 | Unstandardised regression coefficients (B), standard error and p-values for basal ganglia absolute metabolite levels relative to control children at age 7 and 9, controlling for sex, age at scan, ethnicity, and voxel gray matter content.

Coefficients significant at p < 0.05 marked in bold font.

HIV+ children and controls. Although NAA/Cr levels have been shown to be decreased in basal ganglia in children with HIVE (Pavlakis et al., 1995; Lu et al., 1996), in HIV+ children without encephalopathy NAA levels have been reported to remain unchanged (Keller et al., 2004). In our own previous study of the same cohort at age 5 years, children who initiated ART before 12 weeks had higher NAA and choline compared to the uninfected group of which 80% were HIV exposed (Mbugua et al., 2016). Notably, at age 9 we find lower NAA and Glu (but not NAA/Cr or Glu/Cr, Supplementary Table 2) in HIV+ children,


even though only six children in our 9-year old HIV+ group had a prior HIVE diagnosis.

The reason these differences in NAA and Glu are not observed at age 7 may be that the normal age-related increase in NAA (Holmes et al., 2017) is altered in HIV (Keller et al., 2004), such that NAA is no different from controls at age 7 but has dropped below control levels by age 9. At age 5 (Mbugua et al., 2016), only two of the uninfected children were HIV-unexposed, and the rest HEU, so that we cannot say with any certainty how NAA levels in HIV+ children compared to those of HIV-unexposed children at age 5. This makes it difficult to interpret these findings.

No previous study has reported reduced basal ganglia Glu in HIV+ children; together with reduced NAA this suggests neuronal cell loss. Neurons in the basal ganglia may be more vulnerable than other regions to damage via excitotoxicity because of their greater density of NDMA receptors (Lipton et al., 1991) or because of increased blood–brain barrier damage in these structures associated with high plasma viral load (Avison et al., 2004).

Although in basal ganglia both lower (Lu et al., 1996; Keller et al., 2004) and higher (Ashby et al., 2015) Cho/Cr and Cho have previously been found in HIV+ children, we found no difference in basal ganglia Cho between HIV+ children and controls at either age. In adults, elevated basal ganglia Cho/Cr normalizes after ART (Sailasuta et al., 2012). In these children, ART may have caused normalization of basal ganglia Cho levels, without preventing cell loss. However, the strongest neurometabolite effect observed in regressions with clinical data (B/SE, **Table 4**), is that of Cho at 7 years against CD4/CD8 at enrollment. This replicates the finding in the same cohort at age 5 (Mbugua et al., 2016), suggesting that better immune health in infancy may be associated with a greater cell density in childhood. Although the regression coefficient for Cho is 20 times smaller by 9 years of age, at 9 years the regression coefficient for Ins against baseline CD4/CD8 is similarly large.

An alternative reason for the lack of group differences at age 7 may be the lower SNR obtained from the single channel head coil on the Allegra scanner, resulting in higher standard deviations for metabolite concentrations estimated with LCModel at age 7, even though spectra still met strict quality control criteria. The standard errors for the regressions at each age were comparable, showing similar between-subject variability on each scanner. However, effect sizes were much larger in the 9 year old data (**Table 3**), suggesting greater sensitivity to detect differences on the Skyra. It is notable, however, that no group differences were found at age 7 even when regressions were weighted by the inverse of the metabolite standard deviations, to provide heavier weighting to metabolite concentrations estimated with greater precision by LCModel. Moreover, in our study on the same cohort at age 5 years, we were able to detect in data from the same scanner, with similar spectra of similar quality, and in a smaller sample, metabolite differences between HIV+ children initiating ART before and after 12 weeks (Mbugua et al., 2016).

Alterations in NAA and Glu observed at age 9, but not 5 and 7, suggest that basal ganglia neurons may be particularly affected by perinatal HIV/ART and that neuronal damage may be ongoing in this region despite early ART and viral suppression.

The results should be considered exploratory, and suggest that longitudinal investigation should be performed to clarify the timing and persistence of these effects during childhood.

## Reduced Neuronal and Glial Neurometabolites in HEU Children at Age 9

Surprisingly, at age 9 we detect reductions in several neurometabolites in HEU children relative to controls. Notably, regression coefficients for NAA, Glu, and Cr suggest larger reductions in HEU than HIV+ children, and a reduction in Cho is seen relative to controls in HEU but not HIV+ children. Although few neuroimaging studies have been done in HEU children, one diffusion tensor imaging (DTI) study in infants identified a region with high fractional anisotropy in the cerebellum (Tran et al., 2016) and another did not detect DTI or brain volume differences between HEU and control children (Jahanshad et al., 2015). Ours is the first to demonstrate neurometabolite alterations related to HIV/ART exposure. One previous study found elevated white matter Cho/Cr and NAA/Cr ratios in neonates exposed to HIV and ART in utero, however, the HIV status of these infants was not determined (Cortey et al., 1994).

Although at age 9 HIV+ children on ART show no difference in Cho level to control children, in HEU children Cho levels are lowered, suggesting that cell density may be reduced. In addition, we found reduced Cr, reflecting lowered energy metabolism, as well as reduced NAA and Glu, possibly reflecting loss of neurons or reduction in normal neurotransmission. Together, these neurometabolite reductions support the suggestion of loss of both neuronal and glial cells.

It is not clear why HIV exposure should be associated with greater basal ganglia neurometabolite reductions than HIV infection. It is likely that in the context of this study HEU children experience some of the same environmental and social stressors as HIV+ children, as well as nutritional and health challenges, but do not have access to the same level of clinical care as the HIV+ children under follow-up on the CHER trial. This may have resulted in poorer neurodevelopmental outcomes for HEU children.

Concern has also been raised about mitochondrial dysfunction in HEU children due to in utero and postpartum exposure to ARVs, particularly Zidovudine (Poirier et al., 2003), which may affect neurodevelopment. Neuroimaging in ART-exposed children without symptoms may show abnormalities similar to those in children with congenital mitochondrial disease (Tardieu et al., 2005). Reassuringly, however, the Surveillance Monitoring for ART Toxicities (SMARTT) cohort of the Pediatric HIV/AIDS Cohort Study, including more than 3500 children, recently found no association between in utero exposure to ART drugs and cognitive or academic scores in school-age children (Van Dyke et al., 2016).

Although most studies in HEU children under 3 years of age found no neurodevelopmental differences to controls when confounding factors were controlled (Alimenti et al., 2006; Gómez et al., 2009; Williams et al., 2010; Springer et al., 2012; Ngoma et al., 2014), studies from resource-limited settings suggest that HEU infants in Africa demonstrate subtle cognitive and motor impairment, and expressive language delay (Boivin et al., 1995; Le Doare et al., 2012).

Similarly, studies of school-aged HEU children demonstrate subtle deficits compared to control children, particularly in language-related cognitive performance (Van Rie et al., 2008; Kerr et al., 2014; Milligan and Cockcroft, 2017). A recent longitudinal study reported that neurodevelopment of HEU children is initially similar to their HIV-unexposed peers,

but neurocognitive performance starts to fall behind that of HIV-unexposed children during childhood (Smith et al., 2017). This might correspond to a lack of normal age-related increase in Cho, Glu, and NAA in the basal ganglia during childhood (Holmes et al., 2017), which could explain our observations of reduced levels relative to control children at age 9.

It is, however, notable that a recent longitudinal study in the same cohort showed no differences in linear metabolite change between HEU and controls between 5 and 10 years, using a smaller sample of 9-year-old children scanned on a different scanner a few months earlier (Holmes et al., 2017). A limitation of the current study is that data were acquired on different scanners at each age, which does not allow straightforward examination of change in metabolites with age. The cross-sectional analyses presented here suggest that the neurometabolite increase with age may not be linear in one or both of these groups, with a difference manifesting only at the later end of this age range in the basal ganglia.

Future work should investigate the association of basal ganglia metabolites with cognitive performance. Although the basal ganglia are critically involved in motor control, they are also implicated in language processing (Booth et al., 2007). It would be interesting to determine whether basal ganglia neurometabolite alterations are related to subtle cognitive and language deficits in HEU children.

## ETHICS STATEMENT

This study conforms to the ethical guidelines and principles of the international Declaration of Helsinki, and was approved by the Faculty of Health Sciences Human Research Ethics Committees of both the Universities of Cape Town and Stellenbosch. Parents/guardians provided written informed consent and children oral assent.

## REFERENCES


## AUTHOR CONTRIBUTIONS

EM, BL, and AvdK were involved in the study design and acquisition of data. FR, MH, and FL were involved in data and statistical analyses. FR drafted the work and all other authors provided critical revision of the manuscript. FR, MH, EM, BL, ED, and MC provided interpretation of data.

## FUNDING

Support for this study was provided by NRF/DST South African Research Chairs Initiative; US National Institute of Allergy and Infectious Diseases (NIAID) through the CIPRA network (Grant No. U19 AI53217); NIH (Grant Nos. R01HD071664, R21MH096559, and R21MH108346); NRF (Grant No. CPR20110614000019421); GlaxoSmithKline/Viiv Healthcare and the South African Medical Research Council (MRC).

## ACKNOWLEDGMENTS

We thank the CUBIC radiographers Marie-Louise de Villiers, Nailah Maroof, Petronella Samuels and Ingrid Op't Hof, our research staff Thandiwe Hamana and Rosy Khethelo, and Shabir A. Madhi for access to control participants on the CIPRA-SA04 trial.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fnhum. 2018.00145/full#supplementary-material

in neonates born to HIV- positive mothers. Am. J. Neuroradiol. 15, 1853–1859.



associated with neurometabolic reductions in the basal ganglia at age 5 years despite early antiretroviral therapy. AIDS 30, 1353–1362. doi: 10.1097/QAD. 0000000000001082



in pediatric HIV. Neuroimage Clin. 14, 316–322. doi: 10.1016/j.nicl.2017. 01.032

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Robertson, Holmes, Cotton, Dobbels, Little, Laughton, van der Kouwe and Meintjes. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Larger Subcortical Gray Matter Structures and Smaller Corpora Callosa at Age 5 Years in HIV Infected Children on Early ART

Steven R. Randall <sup>1</sup> \*, Christopher M. R. Warton<sup>1</sup> , Martha J. Holmes <sup>2</sup> , Mark F. Cotton<sup>3</sup> , Barbara Laughton<sup>3</sup> , Andre J. W. van der Kouwe<sup>4</sup> and Ernesta M. Meintjes <sup>2</sup>

*<sup>1</sup> Department of Human Biology, Faculty of Health Sciences, University of Cape Town, Cape Town, South Africa, <sup>2</sup> MRC/UCT Medical Imaging Research Unit, Division of Biomedical Engineering, Department of Human Biology, Faculty of Health Sciences, University of Cape Town, Cape Town, South Africa, <sup>3</sup> Children's Infectious Diseases Clinical Research Unit, Department of Paediatrics and Child Health, Tygerberg Children's Hospital & Faculty of Medicine and Health Sciences, Stellenbosch University, Cape Town, South Africa, <sup>4</sup> Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, MA, United States*

#### Edited by:

*Nouria Lakhdar-Ghazal, Faculty of Science, Mohammed V University, Morocco*

#### Reviewed by:

*Owen Carmichael, Pennington Biomedical Research Center, United States Francesco Sammartino, Ohio State University Columbus, United States*

> \*Correspondence: *Steven R. Randall srrandall88@gmail.com*

Received: *30 July 2017* Accepted: *16 October 2017* Published: *02 November 2017*

#### Citation:

*Randall SR, Warton CMR, Holmes MJ, Cotton MF, Laughton B, van der Kouwe AJW and Meintjes EM (2017) Larger Subcortical Gray Matter Structures and Smaller Corpora Callosa at Age 5 Years in HIV Infected Children on Early ART. Front. Neuroanat. 11:95. doi: 10.3389/fnana.2017.00095* Sub-Saharan Africa is home to 90% of HIV infected (HIV+) children. Since the advent of antiretroviral therapy (ART), HIV/AIDS has transitioned to a chronic condition where central nervous system (CNS) damage may be ongoing. Although, most guidelines recommend early ART to reduce CNS viral reservoirs, the brain may be more vulnerable to potential neurotoxic effects of ART during the rapid development phase in the first years of life. Here we investigate differences in subcortical volumes between 5-year-old HIV+ children who received early ART (before age 18 months) and uninfected children using manual tracing of Magnetic Resonance Images. Participants included 61 Xhosa children (43 HIV+/18 uninfected, mean age = 5.4 ± 0.3 years, 25 male) from the children with HIV early antiretroviral (CHER) trial; 27 children initiated ART before 12 weeks of age (ART-Before12Wks) and 16 after 12 weeks (ART-After12Wks). Structural images were acquired on a 3T Allegra MRI in Cape Town and manually traced using MultiTracer. Volumetric group differences (HIV+ vs. uninfected; ART-Before12Wks vs. ART-After12Wks) were examined for the caudate, nucleus accumbens (NA), putamen (Pu), globus pallidus (GP), and corpus callosum (CC), as well as associations within infected children of structure volumes with age at ART initiation and CD4/CD8 as a proxy for immune health. HIV+ children had significantly larger NA and Pu volumes bilaterally and left GP volumes than controls, whilst CC was smaller. Bilateral Pu was larger in both treatment groups compared to controls, while left GP and bilateral NA were enlarged only in ART-After12Wks children. CC was smaller in both treatment groups compared to controls, and smaller in ART-After12Wks compared to ART-Before12Wks. Within infected children, delayed ART initiation was associated with larger Pu volumes, effects that remained significant when controlling for sex and duration of treatment interruption (left β = 0.447, *p* = 0.005; right β = 0.325, *p* = 0.051), and lower CD4/CD8 with larger caudates controlling for sex (left β = −0.471, *p* = 0.002; right β = −0.440, *p* = 0.003). Volumetric differences were greater in children who initiated ART after 12 weeks. Results suggest damage is ongoing despite early ART and viral load suppression; however, earlier treatment is neuroprotective.

Keywords: HIV/AIDS, antiretroviral, MRI, volumetric segmentation, WM, GM, pediatric

## INTRODUCTION

Since the advent of combination antiretroviral (ARV) therapy (ART), human immunodeficiency virus (HIV) infection has become a chronic condition with ongoing damage to the body and central nervous system (CNS), especially in the developing brain of the fetus, infant and young child (van Rie et al., 2007). As CNS penetration by ART is limited, the brain may become a reservoir for the virus, with few drugs available to impact these reservoirs (van Rie et al., 2007). Substantial brain development in the first few years of life puts infected children at greater risk of neurological impairment compared to HIV infected (HIV+) adults (Tardieu et al., 2000; Mitchell, 2001). For example, language functions are more impaired in HIV+ children than in adults (van Rie et al., 2007).

To minimize effects of HIV in children, new guidelines recommend that life-long ART be initiated as soon as possible (WHO, 2013). However, some ARVs can cause neurodevelopmental impairment despite virological suppression in the CNS (Tardieu et al., 2005; van Rie et al., 2007). For example, prenatal exposure to zidovudine has been linked to mitochondrial dysfunction within the CNS (Tardieu et al., 2005), as well as chronic ARV treatment in the form of combination therapy results in abnormal BOLD response within frontal brain regions (Chang et al., 2008). Few studies have examined the long-term neurological effects of HIV infection in children in whom plasma viral loads (VL) are suppressed following early ART (before 18 months of age; Le Doaré et al., 2012; Laughton et al., 2013; Phillips et al., 2016), and very little is known about the complex nature of brain recovery following ART initiation and subsequent brain development (Tamula et al., 2003; Shanbhag et al., 2005; van Rie et al., 2007; Govender et al., 2011; Laughton et al., 2012; Whitehead et al., 2014).

Although neuroimaging can provide insights into the mechanisms underpinning neurobehavioral outcomes in HIV+ children (Hoare et al., 2014; Musielak and Fine, 2016), such studies are rare, even in developed countries, and have typically included children across wide age ranges who either started ART late (Hoare et al., 2015) or some on early ART but small numbers with VL suppression (van Arnhem et al., 2013; Sarma et al., 2014). Studies in children and youths where most (>85%) have suppressed viral loads, and some received early ART, have demonstrated lower global and local cortical and total gray matter (GM) volumes (Cohen et al., 2016; Lewis-de Los Angeles et al., 2017), white matter (WM) alterations (Ackermann et al., 2014, 2016; Andronikou et al., 2014; Uban et al., 2015), and volume reductions (Cohen et al., 2016), both subcortical volume increases and decreases and shape deformations (Lewis-de Los Angeles et al., 2016; Yadav et al., 2017), altered cortical thickness (Yadav et al., 2017), and effects of immunocompromise in infancy on basal ganglia metabolism at 5 years (Mbugua et al., 2016). These findings suggest that HIV- and/or ART-related damage may be irreversible or ongoing.

In this study we compare over a narrow age range (4.9– 6.5 years of age) manually traced volumes of select subcortical structures and the corpus callosum in a unique cohort of HIVinfected children, all of whom initiated ART before 76 weeks of age and were followed since birth, to HIV-uninfected children from the same community. This is the first study using goldstandard manual tracing (Morey et al., 2009) to assess structural brain changes in perinatally acquired HIV on early ART (Hoare et al., 2014; Musielak and Fine, 2016). Amongst HIV+ children from this cohort, initiating treatment before 12 weeks showed improved cognitive performance at 11 months and improved overall health compared to starting treatment after 12 weeks (Violari et al., 2008; Laughton et al., 2012; Cotton et al., 2013). We also examined the effect of timing of treatment initiation on volume alterations. We hypothesized that HIV+ children on early treatment would experience similar perturbations observed previously in both children and adults, with neuronal thinning, resulting overall in global atrophy and loss of gray matter and white matter volume. In addition, we hypothesized that earlier treatment would mitigate the effects of HIV and accompanying brain damage leading to smaller reductions in volume in children initiating ART earlier.

## MATERIALS AND METHODS

This study included 43 HIV+ Xhosa children from the randomized CHER (Children with HIV early antiretroviral) trial in follow-up at the Children's Infectious Diseases Clinical Research Unit, Tygerberg Children's Hospital, Cape Town (Laughton et al., 2012). As part of the CHER trial, infants with CD4 percentage (CD4%) of at least 25% were randomized to one of the following three treatment arms: ART-Def (ART Deferred until CD4% < 25% in first year or CD4% < 20% thereafter, or if clinical disease progression criteria presented); ART-40W (ART initiated before 12 weeks of age and interrupted after 40 weeks); and ART-96W (ART initiated before 12 weeks of age and interrupted after 96 weeks). ART was restarted in ART-40W and ART-96W for CD4% decline or clinical evidence of disease progression. (Violari et al., 2008; Cotton et al., 2013). One child not adhering to ART was excluded. Since some children in the ART-Def arm met criteria for almost immediate initiation of ART, the children were grouped here based on age at treatment initiation, specifically those who received ART at or before 12 weeks (ART-Before12wks) and those who received treatment after 12 weeks (ART-After12wks). Of the 27 who received ART before 12 weeks, 9 remained on continuous ART in line with clinical criteria governing interruption. Treatment was interrupted in 18, two of whom had not met ART restart criteria at the time of scan as they remained clinically well with CD4% of at least 20%.

Eighteen HIV uninfected Xhosa children were recruited from an interlinked vaccine trial as controls (Madhi et al., 2010). They comprised 12 HIV exposed uninfected (HEU) children born to HIV+ mothers but who tested HIV negative (PCR) at baseline and 30 days after the third dose of vaccine, and 6 HIV unexposed and uninfected children (HU) born to HIV seronegative mothers (tested after 24 weeks gestation) who remained seronegative at enrolment.

All children received MRI scanning on a 3T Allegra MRI (Siemens, Erlangen, Germany) at the Cape Universities Brain Imaging Centre (CUBIC) according to protocols approved by the Human Research Ethics Committees of the Universities of Cape Town and Stellenbosch. All parents provided written informed consent and all children provided oral assent. Highresolution T1 weighted images were acquired using a volumetric echo-planar imaging (EPI) navigated (Tisdall et al., 2012) multi echo magnetization prepared rapid gradient echo (MEMPRAGE) sequence (van der Kouwe et al., 2008). Imaging parameters were: FOV: 224 × 224 mm<sup>2</sup> ; 144 sagittal slices, TR: 2,530 ms; TE: 1.53/3.19/4.86/6.53 ms; TI: 1,160 ms; Flip angle: 7◦ ; voxel size: 1.3 × 1.0 × 1.0 mm<sup>3</sup> , scan time 5:20 min. The 3D EPI navigator provided real-time motion tracking and correction. BrainVoyager QX software (Brain Innovation, Maastricht) was used to transform DICOM files into the AC-PC plane and rotate images for hemispherical symmetry.

Due to its frequency in the HIV literature, the basal ganglia (**Figure 1A**) was the primary subcortical region of interest (ROI), with the corpus callosum (**Figure 1B**) functioning as the white matter analog. All gray matter structures within the basal ganglia were manually segmented, with the exception of the substantia nigra and subthalamic nuclei. The structures included the caudate head, nucleus accumbens (NA), globus pallidus (GP), and putamen (Pu). All structures were manually traced using MultiTracer software (http://bishopw.loni.ucla.edu/ MultiTracer/) on a Lenovo ThinkPad X200 tablet and integrated active digitizer stylus.

All tracings were performed at 4X magnification as further magnification resulted in visible pixelation of the images. Image files were scaled according to the prescribed voxel intensity range (Woods, 2003); the contrast used in this study ranged from Pixel Intensity Units of 13,000 to 18,000 dependent on the structure. Screen brightness for the tracing tablet was set to 100% and brightness of images was set to 1.34 (49% of a maximum unadjusted default scale value of 2.72) for most structures (Woods, 2003).

The contours of selected brain structures were completed by manually outlining the structures on MR images, slice by slice. For standardization purposes, tracing of all subcortical gray matter structures were performed in the coronal plane. Based on this tracing, we present the frust volumes output by MultiTracer. The expert neuroanatomist (SR) who performed all tracings was blinded to all participant data. Inter-rater and intra-rater reliabilities were assessed for each structure using independent measurements for 10 participants who were randomly selected by another expert neuroanatomist (CW) and the primary tracer (SR), respectively, and assessed via Pearson and intra-class correlation.

Data and statistical analyses were performed in IBM SPSS Statistics 23. ANOVA was used to compare structure volumes of uninfected children to infected children, and to children who initiated ART before and after 12 weeks. Additionally, amongst infected children, we used linear regression to examine associations of structure volumes with timing of ART initiation. Since treatment was interrupted in a subset of the children who received ART before 12 weeks and interruption is characterized by a subsequent drop in CD4 cell count and increased VL,

which may negate the potential benefits of earlier ART, multiple regression was used to control for duration of interruption. Duration of interruption was set to zero for children in whom treatment was not interrupted. Sex was controlled for in all analyses as it affects brain development (Gur et al., 1999; Gilmore et al., 2007). Finally, to explore whether volume alterations may be due to an inflammatory response, we examined relationships between structure volumes and CD4/CD8 ratio as a proxy for immune health, both at enrolment and at scan. Due to the fact that statistical analyses were repeated for 9 anatomical structures, we indicate for each test which results survive Bonferroni correction (p < 0.0056).

Since lower total GM and WM volumes have been reported in HIV+ youths (Cohen et al., 2016; Lewis-de Los Angeles et al., 2017), total intracranial volumes (ICVs) were derived for all children using FreeSurfer (Fischl, 2012) and considered a potential confounder. Additional control variables considered included age at scan and birthweight. Control variables showing group differences were controlled for using analysis of covariance (ANCOVA) and multiple regression.

## RESULTS

We present data for 61 Xhosa children (mean age ± s.e. = 5.4 ± 0.04; 25 male). Of 18 uninfected controls, 12 were HEU and 6 HU. Of the 43 infected children, 27 received ART before 12 weeks, and 16 after. Sample characteristics are summarized in **Table 1**. Since uninfected children were slightly older than infected children, and children who received ART before 12 weeks had smaller ICVs than uninfected controls, we controlled for ICV and age at scan in all analyses involving uninfected and infected children. While no children had suppressed VLs at enrolment, VL was suppressed in 93% of infected children at time of scanning. Children receiving ART before and after 12 weeks did not differ on any of the clinical variables. No significant effect of interruption on structure volume was observed, nor did duration of interruption show any relation with structure volumes (all p's > 0.18).

All inter-rater Pearson correlations for manually traced volumes were significant (all p's < 0.05) and ranged from r = 0.71 for the left caudate to r = 0.93 for the corpus callosum. Cronbach's α's were above 0.8 in all regions. Intra-rater Pearson correlations were greater than r = 0.82 (all p's < 0.001), and Cronbach's α's were all above 0.83.

In **Table 2** we compare structure volumes between HIV+ and uninfected children, and in **Figure 2** between uninfected children and children who received ART before and after 12 weeks, controlling for sex, ICV, and age at scan. Infected children showed gray matter (GM) volume increases in left (L) NA, albeit below conventional levels of significance, right (R) NA, bilaterally in Pu, and L GP, effects that appear to be largely attributable to volume increases in children who initiated ART after 12 weeks. In contrast to GM volume increases, CC was smaller in infected children with reductions evident both in children initiating ART before and after 12 weeks. Only in CC did volumes differ significantly between children initiating ART before and after 12 weeks, with the latter showing greater volume reductions compared to uninfected controls than children who initiated ART before 12 weeks. HIV-related volume increases in the Pu and reductions in the CC remain significant after Bonferroni correction for multiple comparisons.

Amongst infected children, increasing age of ART initiation was associated with larger Pu volumes bilaterally (**Figure 3**, **Table 3**), with the association on the left surviving Bonferroni correction for multiple comparisons. Although caudate volumes did not show any group differences or relation with timing of ART initiation, amongst infected children, lower CD4/CD8 ratio at the time of scanning was associated with caudate volume increases bilaterally (left r = −0.487, p = 0.007; right r = −0.469, p = 0.001; **Figure 4**), effects that remained significant when controlling for sex (left β = −0.471, p = 0.002; right β = −0.440, p = 0.003) and adjusting for multiple comparisons. Lower enrolment CD4/CD8 was also weakly associated with greater caudal volumes at age 5 years (left r = −0.252 p = 0.107; right r = −0.298 p = 0.056), although these relations did not survive when controlling for sex (left β = −0.217, p = 0.201; right β = −0.226, p = 0.172). No other regions showed association between volumes and immune health at enrolment (all p's > 0.2) or at scan (all p's > 0.5).

## DISCUSSION

This is the first study to examine over a narrow age range the impact of HIV on subcortical gray matter and corpus callosum volumes in a cohort of children receiving ART before 18 months of age and VL suppression by 109 weeks in all but three children. In contrast to most previous studies in HIV+ adults and children that reported gray matter atrophy (Tardieu et al., 2000; Crain et al., 2010; Becker et al., 2011; Chang et al., 2011; Cohen et al., 2016; Musielak and Fine, 2016; Lewis-de Los Angeles et al., 2017), we found HIV-related subcortical GM volume increases at 5 years in the left GP and bilaterally in the NA and Pu, but CC reductions. Association of increasing age of ART initiation with greater volumes in the Pu suggest that treatment timing plays a role in these observed volume increases. The fact that both GM and WM volume differences were largest in children who initiated ART after 12 weeks point to greater protection from earlier treatment, which is consistent with the findings from other studies in this cohort (Laughton et al., 2012; Mbugua et al., 2016). The fact that associations of Pu and NA volumes with treatment timing strengthened after adjustment for interruption supports findings from a previous DTI study in the same cohort that found the lowest fractional anisotropy in a region in the corticospinal tract in treatment-interrupted children suggesting that ART interruption may negate the benefit of earlier ART (Ackermann et al., 2016).

Even though most studies in children find cortical and total gray matter atrophy in HIV infection (Cohen et al., 2016; Lewis-de Los Angeles et al., 2017), regional GM volume increases have been reported (Sarma et al., 2014). Although, few studies have examined subcortical volumes specifically, two recent studies reported HIV-related regional subcortical GM increases. Blokhuis et al. (2017) found trend-level Pu increases


TABLE 1 | Sample

characteristics.

*aART-Before12*

*bART-Before12*

*cOnly including 18 children from the ART-Before12*

α*CD8 count at enrolment data missing for 1 ART-After12*

§*Plasma Viral Load Chi Tests had fewer than 5 cases.*

 *wks, ART-After12*

 *wks* <

 *wks* <

*Uninfected Controls (p* < *0.01).*

*Uninfected Controls (both p's* < *0.05).*

 *wks group who were interrupted;*

 *wks girl.*

 *2 had not yet restarted ART at time of scanning.*

TABLE 2 | Comparison of structure volumes between HIV infected and uninfected children.


*F-value from Analysis of Covariance adjusted for sex, intracranial volume and age at scan.*

TABLE 3 | Associations in infected children of structure volume with age at ART initiation.


*r is the Pearson correlation coefficient;* β*1 is the standardized regression coefficient adjusted for sex;* β*2 is the standardized regression coefficient adjusted for sex and duration of interruption.*

in HIV+ children, and Yadav and colleagues larger nucleus accumbens and smaller hippocampi (Yadav et al., 2017). The authors postulated inflammatory processes and chronic stress as possible explanations for the volume increases. Putamen hypertrophy in HIV+ adults has also been attributed to possible inflammation and dopaminergic system dysfunction (Castelo et al., 2007). The absence of association of CD4/CD8 ratio, as a marker of immune health, with volumes in regions showing HIV-related increases in our study, suggest that observed volume increases are not necessarily due to inflammatory processes in these specific regions.

Compared to the relative abundance of literature attributing structural atrophy to gliosis and inflammation (Bates et al., 2002; Miller et al., 2002; Mehta et al., 2003; Pekny and Pekna, 2016; Pérez-Cerdá et al., 2016), the mechanisms for pathology-induced hypertrophy of brain structures are not wellunderstood. Although enlargement of the striatum has been associated with dopaminergic system dysfunction in recreational and antipsychotic drug use (Selemon et al., 1999; Jacobsen et al., 2001; Dazzan et al., 2005; Jernigan et al., 2005), and linked specifically to elevated glial densities in rhesus monkeys administered antipsychotic drugs (Selemon et al., 1999), smaller volume differences in children who initiated ART earlier in our study suggests that the gray matter hypertrophy observed here is not due to gliosis resulting from ART drug interactions. Notably, gray matter hypertrophy has been observed in patients with obstructive sleep apnea (Rosenzweig et al., 2013). Since shortterm hypoxic events have been shown to induce reactive gliosis and neuronal death in rats (Aviles-Reyes et al., 2010), ischaemic preconditioning may cause long-lasting neuronal changes within the brain. This may be relevant here as cerebrovascular alterations have been reported in HIV+ children (Shah et al., 1996; Patsalides et al., 2002). Increased metabolic demand in the presence of reactive gliosis (Epstein and Gelbard, 1999; Walsh et al., 2004) and mitochondrial dysfunction in HIV+ children (Tardieu et al., 2005; Crain et al., 2010; Takemoto et al., 2017) may create localized subclinical ischaemic/hypoxic stress within the CNS. Angiogenesis from ischaemic preconditioning may be exacerbated by the HIV-Tat gene, a heparin-binding angiogenic growth factor (Das et al., 2016), expressed by infected cells. The exact mechanisms of gray matter hypertrophy in HIV+ children requires further investigation.

Volumetric MRI studies in HIV have mostly relied on automated/semi-automated techniques (Hoare et al., 2014; Sarma et al., 2014; Musielak and Fine, 2016), which typically involve some form of co-registration to a template based on an adult brain, using contrast or other feature discriminators. These techniques may be suboptimal for diagnoses where regional or global brain volumes may be affected, or when examining children. It has been shown that using age-matched pediatric brain templates in pediatric studies lead to considerably different tissue distribution from that obtained with an adult-based template (Yoon et al., 2009). This may explain why previous studies of subcortical volumes using FreeSurfer either failed to detect group differences (Lewis-de Los Angeles et al., 2016), or detected fewer and less significant differences (Blokhuis et al., 2017; Yadav et al., 2017) than in our study. Using FreeSurfer automated segmentation in our cohort, within the gray and white matter structures investigated, there were no significant volumetric differences between groups, apart from the left globus pallidus (GP) which was smaller in infected children (L GP: Mean (SD): HIV 1,775 ± 192 mm<sup>3</sup> , controls 1,949 ± 255 mm<sup>3</sup> , F = 7.848, p = 0.007), and the CC that tended to be larger (HIV 469 ± 89 mm<sup>3</sup> , controls 431 ± 68 mm<sup>3</sup> , F = 2.871, p = 0.096). In both structures, the findings were opposite to those generated using manual segmentation, indicating that automated methods may be inappropriate for pediatric populations, especially in the presence of pathology, and that manual segmentation may be more sensitive to detect subtle changes.

The highest concentration of HIV is observed in the caudate nuclei and CC (Thompson et al., 2006; Ances et al., 2012), possibly due to their proximity to the ventricles. The proximity of these structures to the ventricles, and thus CSF, allows for easier penetration of HIV-infected mononuclear cells permitting higher concentrations of HIV toxins in these sites (Oster et al., 1993; McClernon et al., 2001; Thompson et al., 2006; Kumar et al., 2007; Ances et al., 2012; Andronikou et al., 2014). This would

in turn also apply to certain ARVs that more readily permeate through the blood-brain-barrier (BBB) and pass within the CSF to these sites. As such, location may explain the association seen in the left and right caudate with CD4/CD8 ratio at the time of scanning as a measure of immune health in which the more immunocompromised children at the time of scan had larger caudate volumes bilaterally. This relationship, which survives correction for multiple comparisons, suggests that at this age the caudate is particularly susceptible to the concurrent immune state.

Due to their greater distance from the ventricles, the Pu and NA may be less readily penetrated by the HIV-infected mononuclear cells. However, higher cerebral blood flow of these areas have been observed in HIV+ children (Blokhuis et al., 2017). Another hypothesis may be that once HIV invades the area and viral infection spreads, removal of metabolic waste

and virions may be slower in these structures compared to the caudate. This may explain the larger Pu and NA observed in HIV+ children. Thus, presence of inflammation in the form of reactive gliosis and leukocyte infiltration, resulting in slightly increased structure size, may be because of prolonged HIV exposure in the Pu and NA rather than a present acute, transient, and compromised immunological state, as in the left and right caudate.

Our finding of reduced CC volume in HIV infection is consistent with that of most previous studies (Thompson et al., 2006; Chiang et al., 2007; Hasan et al., 2009; Dewey et al., 2010; Heaps et al., 2012; Hoare et al., 2012; Sarma et al., 2014; Yadav et al., 2017), although CC volume and thickness were similar to those in controls in a study by Andronikou et al. (2015) including infected children from the same cohort studied here. In our study, the largest volume difference between infected and uninfected children was observed in the CC, with the CC of HIV+ children being on average 24% smaller at 5 years of age than in their uninfected counterparts. Thompson et al. (2006), who similarly found a 25% decrease in the thickness of the CC in HIV+ adults compared to uninfected controls (Thompson et al., 2006), postulated that the overall thickness of the CC is impacted by the loss of peripheral WM. It has been suggested that callosal thickness and volume may be used as biomarkers of overall global WM integrity (Thompson et al., 2006; Andronikou et al., 2014). However, DTI in children from the same cohort studied here revealed no differences in WM integrity between infected and uninfected children in the CC at 5 years of age (Ackermann et al., 2016), despite volume reductions. Since DTI analyses involve co-registration to a template before performing voxel wise statistical comparisons, DTI outcomes would not be impacted by volume differences. As such, making inferences regarding microstructural integrity from macrostructural data may be inappropriate.

The clinical manifestations of HIV infection in these children do not paint a clear picture of causation. VL was suppressed (≤399 RNA/ml) in 93% of children at time of scanning. A previous study in 128 perinatally infected children, who were born prior to the adoption of preventative treatment or ART guidelines, found that 21% showed characteristic evidence of HIV infection induced encephalopathy, despite at least 74% of these children having VL suppression at the time of diagnosis (Cooper et al., 1998). Findings of VL suppression in plasma may, however, not be representative of the compartmentalized and unique viral reservoirs in the CNS (Strain et al., 2005; Pillai et al., 2006). Further, there could be ongoing effects from damage or

delayed development arising during the initial phases of HIV invasion, prior to ART initiation, which may have occurred in utero in some participants. In our study, all HIV+ treatment groups had elevated VL at study entry (7 weeks of age) thus possibly facilitating HIV entry into the CNS early on (Ivey et al., 2009). Neuroinvasion by HIV can occur as early as the initial 10 days post-infection (Lackner et al., 1994). Early ART administered at around 8.4 ± 1.6 weeks of age is perhaps too late for individuals already experiencing high viral loads. Damage to the basal ganglia, and thus corpus striatum is dictated by the rate of initial insult (Brouwers et al., 2000; Becker et al., 2011). This may be evidenced by our observation of CD4/CD8 ratio at enrolment showing weak trend-like association with the size of the left and right caudate. This ties in with spectroscopy findings involving the same cohort at the same age, where basal ganglia metabolite levels (choline, NAA) were associated with CD4/CD8 at enrolment, suggesting that early infiltration and damage caused by HIV persists into early childhood (Mbugua et al., 2016). A confounding issue here may be the duration of ART itself. A trend of duration of ART with prefrontal CC thickness was observed in this cohort previously (Ackermann et al., 2014; Andronikou et al., 2014). It has also been shown that ART decreases motor and working memory performance in both children and adults (von Giesen et al., 2003; Chang et al., 2008; Laughton et al., 2012) which suggests that there may be some

complicated interplay between the benefits and risks associated with early ART.

A limitation to the study is the inclusion of total ICV calculated using FreeSurfer automated methods. Intrinsically, using an automated measure to standardize manually derived volumes may add to inaccuracies (Morey et al., 2009; Yoon et al., 2009; Dewey et al., 2010; Narayanan et al., 2016). However, our results do not change significantly without the inclusion of ICV. In addition, a limitation in sampling bias may exist. As the HIV+ Xhosa children formed part of a longitudinal cohort followed since birth with regular follow-up visits to the clinic, it is possible that they are at an added advantage as they could be receiving better health care than the normal standard community care, despite standardized treatment as per the current WHO guidelines. These children also received neuropsychological and behavioral testing giving them additional opportunities for individual case specific and HIV-related support. Children who maintained participation may also have nuanced home environments which allowed them to continue longitudinally.

Overall findings of this study suggest that perinatal HIV infection targets select structures of the basal ganglia, and that these effects are observable at 5 years of age despite early ART and VL suppression. In HIV+ children, both the NA and Pu are enlarged bilaterally, as well as the left GP. In contrast, the CC is smaller compared to uninfected children. Our results suggest that earlier treatment is neuroprotective. Early therapy before severe viral replication would be advantageous, as evidenced by increased structure size of the bilateral Pu with increased delay to initiate ART.

## AUTHOR CONTRIBUTIONS

SR, CW, and EM designed this study and developed the methodology. SR, MH, and EM drafted the manuscript. EM, AvdK, and BL are Principal Investigators. AvdK provided Technical expertise and data processing. SR conducted manual segmentation and performed statistical analyses. CW trained SR in manual segmentation protocol and conducted initial inter-rater reliabilities. MC and BL offered clinical expertise, and monitored and collected clinical data of participants longitudinally. All authors contributed to the manuscript and revised it critically.

## ACKNOWLEDGMENTS

We thank the participants and their parents for being willing to take part in this study and research assistants Lunges Khethelo and Thandiwe Hamana for their expertise in supporting the children during neuroimaging, and Stevie Biffen for Inter-rater segmentation.

This work was supported by NIH grants R01HD071664, R21MH096559, and R21MH108346; South African Medical Research Council (SAMRC); South African National Research Foundation (NRF) grants CPR20110614000019421 and CPRR150723129691; and the NRF/DST South African Research Chairs Initiative. Support for the CHER study, which provided the infrastructure for the neurodevelopmental substudy, was provided by the US National Institute of Allergy and Infectious Diseases through the CIPRA network, Grant U19 AI53217; the Departments of Health of the Western Cape and Gauteng, South Africa; and GlaxoSmithKline/Viiv Healthcare. Additional support was provided with Federal funds from the National Institute of Allergy and Infectious Diseases, National Institutes of Health, United States Department of Health and Human Services, under Contract No. HHSN272200800014C.

Permission to conduct the substudy on this cohort was granted by Doctors Avy Violari, Shabir Madhi, and Mark Cotton and the CHER steering committee.

## REFERENCES


apnea: a role for the ischemic preconditioning? PLoS ONE 8:e83173. doi: 10.1371/journal.pone.0083173


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Randall, Warton, Holmes, Cotton, Laughton, van der Kouwe and Meintjes. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Working Memory Profiles in HIV-Exposed, Uninfected and HIV-Infected Children: A Comparison with Neurotypical Controls

Robyn Milligan and Kate Cockcroft\*

Department of Psychology, School of Human and Community Development, University of the Witwatersrand, Johannesburg, South Africa

This study compared the working memory profiles of three groups of children, namely HIV-infected (HIV-I; n = 95), HIV-exposed, uninfected (HIV-EU; n = 86) and an HIV-unexposed, uninfected, (HIV-UU; n = 92) neurotypical control group. Working memory, an executive function, plays an important role in frontal lobe-controlled behaviors, such as motivation, planning, decision making, and social interaction, and is a strong predictor of academic success in school children. Memory impairments have been identified in HIV-I children, particularly in visuospatial processing. Verbal working memory has not been commonly investigated in this population, while it is unknown how the working memory profiles of HIV-EU children compare to their HIV-I and HIV-UU peers. Of interest was whether the working memory profiles of the HIV-EU children would be more similar to the HIV-I group or to the uninfected control group. The results revealed no significant differences in working memory performance between the HIV-I and HIV-EU groups. However, this does not mean that the etiology of the working memory deficits is the same in the two groups, as these groups showed important differences when compared to the control group. In comparison to the controls, the HIV-I group experienced difficulties with processing tasks irrespective of whether they drew on a verbal or visuospatial modality. This appears to stem from a generalized executive function deficit that also interferes with working memory. In the HIV-EU group, difficulties occurred with verbally based tasks, irrespective of whether they required storage or processing. For this group, the dual demands of complex processing and using a second language seem to result in demand exceeding capacity on verbal tasks. Both groups experienced the greatest difficulties with verbal processing tasks for these different reasons. Thus, disruption of different cognitive abilities could result in similar working memory profiles, as evidenced in this study. This has implications for the underlying developmental neurobiology of HIV-I and HIV-EU children, as well the choice of appropriate measures to assist affected children.

#### Edited by:

Vivienne Ann Russell, University of Cape Town, South Africa

#### Reviewed by:

Michael J. Boivin, Michigan State University, United States Eileen Martin, Rush University, United States

> \*Correspondence: Kate Cockcroft kate.cockcroft@wits.ac.za

Received: 09 May 2017 Accepted: 19 June 2017 Published: 06 July 2017

#### Citation:

Milligan R and Cockcroft K (2017) Working Memory Profiles in HIV-Exposed, Uninfected and HIV-Infected Children: A Comparison with Neurotypical Controls. Front. Hum. Neurosci. 11:348. doi: 10.3389/fnhum.2017.00348

Keywords: HIV-exposure, HIV-infection, working memory

Working memory is a limited capacity system responsible for briefly holding and manipulating visuospatial and verbal material in a readily accessible form (Baddeley, 2000). It is regarded as a key executive function, together with cognitive flexibility and inhibitory control of automatized behaviors, and as such plays an important role in frontal lobe-controlled behaviors such as motivation, planning, decision making, and social interaction (Miyake et al., 2000; Miyake and Friedman, 2012). Measures of working memory are significant predictors of academic potential and success, even better than IQ scores (Berninger and Swanson, 1994; Swanson and Sachse-Lee, 2001; DeStefano and LeFevre, 2004; Swanson et al., 2004; Alloway and Alloway, 2013; Alloway and Copello, 2013; Alloway and Gregory, 2013). Furthermore, working memory assessments have reduced cultural and socioeconomic bias in comparison to conventional intelligence and scholastic measures, making them suitable for assessing learning in non-Western, low resource and developing communities (Engel et al., 2008; Rinderman et al., 2010; Cockcroft et al., 2016).

There is a growing body of research on the working memory functioning of neurotypical children. What is known within this population is that working memory is fractionated into several inter-related components, generally supporting the Baddeley (2000, 2012) model. Although several alternative working memory models exist (e.g., Ericsson and Kintsch, 1995; Cowan, 1999; Oberauer, 2010), Baddeley's (2000) was selected as the theoretical basis for this study, as there appears to be the most theoretical consensus around this model, it is the most widely researched with pediatric populations, and it can be readily operationalized by psychometric measurement (Miyake and Shah, 1999). According to this model, working memory comprises three storage components, namely a visuospatial sketchpad (specialized for briefly holding and refreshing visual and spatial material), a phonological loop (responsible for auditory-verbal material), and an episodic buffer (responsible for integrating different types of material into meaningful episodes and communicating with long-term memory). These storage components are supervised by a flexible attentional controller, the central executive, responsible for controlled processing such as co-ordination of multiple tasks, temporary activation of longterm memory, maintaining task goals, and resolving interference during complex cognition (Baddeley et al., 1998a,b; Baddeley, 2000). The processing functions of the central executive may overlap with other executive control functions (Miyake and Friedman, 2012).

In neurotypical children, scores on working memory measures predict reading achievement, math ability, as well as learning and skill acquisition (Bull and Scerif, 2001; Swanson and Sachse-Lee, 2001; Cowan and Alloway, 2008; Alloway et al., 2009). Research into working memory functioning in neurodivergent populations includes children with Attention Deficit/Hyperactivity Disorder (AD/HD), Developmental Co-ordination Disorder (DCD), Specific Language Impairment (SLI), Autistic Spectrum Disorder (ASD), Down Syndrome, Williams Syndrome, Dyslexia, Dyscalculia, and general intellectual disabilities (Hughes et al., 1994; Bull and Johnston, 1997; McLean and Hitch, 1999; Passolunghi and Siegel, 2001, 2004; Swanson and Sachse-Lee, 2001; Laws and Bishop, 2003; Marton and Schwartz, 2003; Geurts et al., 2004; Pickering and Gathercole, 2004; Alloway and Gathercole, 2006; Pickering, 2006; Whitehouse et al., 2006; Williams et al., 2008; Alloway, 2011; Henry, 2012; Henry et al., 2012). There is general support for the fractionation of working memory following the Baddeley (2000, 2012) model, as well as evidence of distinct patterns of working memory deficits in these populations. For example, children with AD/HD show difficulties in central executive functioning, while those with DCD have deficits primarily in visuospatial working memory (Alloway and Gathercole, 2006; Alloway, 2011). Studies of children with SLI show that they have particular weaknesses in the phonological loop and central executive (Laws and Bishop, 2003; Marton and Schwartz, 2003; Pickering and Gathercole, 2004; Henry, 2012; Henry et al., 2012), while children with ASD experience greatest difficulty on central executive and visuospatial memory tasks (Hughes et al., 1994; Geurts et al., 2004; Whitehouse et al., 2006; Williams et al., 2008). Children with Dyslexia tend to be characterized by deficits in the phonological loop and central executive, while those with Dyscalculia also show central executive difficulties, as well as poor visuospatial working memory (Bull and Johnston, 1997; McLean and Hitch, 1999, Passolunghi and Siegel, 2001, 2004; Swanson and Sachse-Lee, 2001; Passolunghi et al., 2005; Passolunghi, 2006). These findings are useful as they enable theorizing about neurodivergent cognitive development in specific disorders, facilitate tracking the developmental progression of these disorders, and allow for appropriate remediation strategies to be implemented for affected children.

A population that is markedly absent from these studies is children with Human Immunodeficiency Virus (HIV) infection and exposure. Given the high incidence of pediatric HIV infection in Africa (World Health Organization [WHO], 2017), together with evidence that HIV-infected (HIV-I) children show an increased prevalence of learning difficulties (Sherr et al., 2009), as well as the importance of working memory for learning (Alloway et al., 2009), an investigation of working memory profiles in HIV-I and HIV-exposed but uninfected (HIV-EU) children is clearly warranted. An estimated 1,752,300 South African children between the ages of 5 and 14 years have reportedly been exposed to the virus in utero (Statistics South Africa, 2016), indicating a large group of children who are likely to be in need of specialized medical, neurocognitive, and educational assistance.

Most studies of neurocognition in HIV-I children have been generalist in nature [i.e., focusing on a general developmental or IQ score, obtained from the Griffiths Mental Development Scales (GMDS), the Bayley Scales of Infant Development, the Kaufmann Assessment Battery for Children (K-ABC), or the Wechsler Individual Scales for Children (WISC; Eley et al., 2008; Shead et al., 2010; Kandawasvika et al., 2011; Laughton et al., 2012, 2013; Lowick et al., 2012)]. Such studies may miss more subtle deficits within specific cognitive domains. A focused investigation of working memory profiles can provide detail on how this important executive function may be impacted by HIV infection

and exposure. Of the few studies which investigated working memory, none appear to have measured all of its components (i.e., verbal and visuospatial storage, verbal and visuospatial processing), and most employed only a single measure, typically visuospatial in nature, thereby limiting construct validity. In most cases, the investigation of working memory was secondary to the investigation of another cognitive construct, such as executive functioning or general intelligence (Bagenda et al., 2006; Koekkoek et al., 2008; Allison et al., 2009). Our study addressed these shortcomings by using a detailed assessment comprising three measures of each component of working memory in order to compile comprehensive profiles of working memory functioning in three groups of children between 6 and 8 years matched for age, English-language ability, and socioeconomic status (SES). The three groups included HIV-I, HIV-EU, and HIV-unexposed, uninfected (HIV-UU) neurotypical children. Such profiles could assist with distinguishing the neurobiological mechanisms underlying HIV infection, and HIV exposure with no infection.

Working memory deficits have been identified in individuals infected with HIV, but these studies have generally focused on adults (Stout et al., 1995; Farinpour et al., 2000; York et al., 2001; Hinkin et al., 2002; Reger et al., 2002; Sacktor et al., 2002; Heaton et al., 2004; Dawes et al., 2008). It is difficult to reconcile the results of the few investigations of neurocognitive functioning in HIV-I children which include working memory measures due to sample differences in sociocultural and economic backgrounds (Lowick et al., 2012), different assessment measures, and very wide sample age ranges (Wachsler-Felder and Golden, 2002). Taking these difficulties into account, there is some evidence for the preservation of verbal storage (Blanchette et al., 2002; Bagenda et al., 2006; Martin et al., 2006; Klaas et al., 2009), and for impairments in visuospatial processing in HIV-I children (Boivin et al., 1995; Koekkoek et al., 2008). However, these studies had relatively small sample sizes (N = 14–41), and other than Bagenda et al. (2006), who included 6- to 12-year olds, all considered an older cohort of children (9+ years) than in the present study. None had a specific focus on working memory.

An important issue when investigating neurocognition in HIV-I children from Africa is the availability and adherence to antiretroviral treatment (ART), since only 28% of children in low and middle income countries who require ART actually receive it (Grantham-McGregor et al., 2007; Walker et al., 2011). The early initiation of ART appears to be key in preserving neurocognitive functions and possibly even reversing some of the damage caused by HIV infection, particularly in infants (Laughton et al., 2012). Similarly, there is evidence that HIV-I children who are not on ART show significant neurocognitive deficits relative to their uninfected counterparts (Shead et al., 2010). While ART plays a significant role in protecting neurocognition in HIV-I children, these children nonetheless struggle with neurocognitive deficits and delay relative to their uninfected peers (Smith et al., 2008; Cotton et al., 2009). This may be due to the neurotoxic effects of HIV infection which may cause permanent structural damage to the central nervous system before ART is initiated, and/or due to incomplete penetration of the blood–brain barrier by antiretroviral agents (Wolters et al., 1997; Van Rie et al., 2007; Cotton et al., 2009; Lowick et al., 2012). In our sample, all of the HIV-I children were on combination ART and had been virologically and immunologically stable for at least 6 months.

Due to improvements in preventative mother-to-child antiretroviral treatment (PMTCT) and its administration, the HIV-EU child population is rapidly overtaking the number of children born with HIV infection (Filteau, 2009; Shapiro and Lockman, 2010; Morden et al., 2016). The HIV-EU child is believed to have a unique neurocognitive profile due to exposure to the immunological side effects of HIV in utero (as a result of immune activation in the mother), as well as exposure to the prophylactic effects of PMTCT (Kuhn et al., 2001; Le Chenadec et al., 2003; Bunders et al., 2005; Nyoka, 2008; Garay and McAllister, 2010; Claudio et al., 2013). There has been limited research into the neurocognitive functioning of this population, and none that has explored how their working memory profiles may differ from those of children with HIV infection, or from uninfected controls. Investigations into the neurocognitive functioning of HIV-EU children have produced ambiguous findings. For example, there is some evidence that their neurocognition does not differ from that of neurotypical children (Kandawasvika et al., 2015), while other studies have found significant impairments in verbal functioning, sequencing, memory, and quantitative reasoning (Levenson et al., 1992; Brackis-Cott et al., 2009; Kerr et al., 2014). Kerr et al. (2014) compared the neurocognitive functioning of HIV-EU children from Thailand (n = 160) and Cambodia (n = 202) to an unexposed control group (n = 167). The groups were compared on the Child Behavior Checklist, the Beery Visual Motor Integration Test, the Stanford Binet-II and the Wechsler Preschool and Primary Scales of Intelligence, third edition (WPPSI-III). No significant group differences were found in executive function ability, but a significantly greater proportion of the HIV-EU sample had attentional difficulties. However, these results should be interpreted cautiously given the very wide age range of the sample (2–15 years). In another study, Kandawasvika et al. (2011) investigated the risk of HIV-associated neurocognitive impairment in 65 HIV-I infants who were part of a Zimbabwean PMTCT program in primary healthcare clinics. They were compared to 188 HIV-EU and 287 HIV-UU infants. A translated version of the Bayley Infant Neurodevelopmental Screener (BINS) was administered when the infants were 3, 6, 9, and 12 months old. This is a screening measure which identifies risk for developmental delay and neurological impairment in four areas, namely neurological functions, expressive functions, receptive functions, and cognitive processing. Infants were then classified according to their risk (low, moderate, or high), with 17% of those at high risk from the HIV-I group, 9% from the HIV-EU group, and 9% HIV-UU group (the remainder were of unknown status). The results showed that for this high risk group, the threat of neurocognitive impairment between the ages of 3 and 9 months increased from 3 to 6%, and this risk was highest among HIV-I infants (10% versus 2%: p < 0.001). The mothers of the HIV-I children tended to be older, more likely to be single, have no financial subsistence and to be co-infected with syphilis than the mothers of the other groups, factors which may have contributed to their infants' underperformance.

By 9 months, one-third of the uninfected infants were in the moderate risk group. These findings suggest that, even with PMTCT, there is risk of progressive encephalopathy for HIV-EU infants, and neurocognitive impairment may become increasingly more evident as these infants develop (Pollack et al., 1996).

Much of the research with HIV-I and HIV-EU children emerges from English-first language, Western, well-resourced contexts, and so has limited generalizability to children from sub-Saharan Africa, who experience a different, more virulent clade of HIV, and who are growing up in very different cultural, linguistic, and contextual circumstances (Chase et al., 2000; Wachsler-Felder and Golden, 2002; Laughton et al., 2013). This concern is also applicable to the studies that have profiled working memory functioning in other neurodevelopmental disorders, as well as in neurotypical children (Passolunghi et al., 2005; Alloway, 2011; Henry, 2012). The application of these findings to under-resourced, developing, African contexts where ART tends to be initiated later, is likely to be limited (Blanchette et al., 2002; Feinstein, 2003; Bagenda et al., 2006; Paxson and Schady, 2007; Smith et al., 2008). Importantly, samples from developing countries may be erroneously identified as showing significant neurocognitive delays if compared to Western norms in the absence of any matched neurotypical control group from the same socioeconomic, linguistic, and contextual backgrounds. For example, Lowick et al. (2012) compared the neurodevelopmental functioning of 30 HIV-I South African children on ART to that of 30 neurotypical controls (age range: 55–75 months). The HIV-I children had received combination ART for at least a year, and were virologically and immunologically stable. The standard scores from the GMDS-ER were used to categorize participants into developmentally delayed or not delayed groups. The HIV-I group had consistently higher proportions of children with developmental delay on all subscales of the GMDS-ER compared to the controls. Nearly half (46.7%) of the HIV-I group demonstrated severe developmental delay, compared to 10% of the neurotypical control group (p < 0.05); this reflects a sevenfold increase in the likelihood of severe neurodevelopmental delay in the HIV-I group (OR = 7.88; CI 1.96–31.68). Worryingly, 87% of the control group was classified as functioning in the below average to borderline range on the GMDS-ER, which may be a result of the context of extreme poverty and deprivation in which these children were growing up. The GMDS-ER is standardized on neurotypical British infants and children (Laher and Cockcroft, 2013). Similarly, Shead et al. (2010) found that the scores of their HIV uninfected group of infants (16– 42 months) were more than one standard deviation below the age appropriate norms on the Bayley Scales. While these Scales have norms for South Africans (Richter and Griesel, 1988), they are outdated. This highlights the importance of using appropriate neurotypical comparison groups when sampling from non-Western, developing countries, and when using developmental measures that were normed in Western, developed and wellresourced contexts that tend to be culturally, socioeconomically, linguistically, and educationally different. Consequently, it was necessary in our profiling of the working memory functioning of HIV-I and HIV-EU children, to include a control group of

neurotypical children from similar socioeconomic, linguistic, and cultural backgrounds, for comparison purposes.

The present study compared the working memory profiles of children who were HIV-I, HIV-EU, and an HIV-UU control group. As demonstrated in the review, working memory impairments have been identified in HIV-I children, particularly in visuospatial processing (Boivin et al., 1995; Koekkoek et al., 2008). It is unknown how the working memory profiles of HIV-EU children compare to their HIV-I and HIV-UU peers. Of interest was whether the working memory profiles of the HIV-EU children would be more similar to the HIV-I group or to the uninfected control group. We hypothesized that there would be significant differences in the working memory profiles of these groups, with the HIV-I group performing worst, and the HIV-UU group best.

## MATERIALS AND METHODS

## Participants

All participants were African or Colored (mixed race), with an African language as their mother tongue and English as their second language. The HIV-I group comprised 95 children (49 girls; mean age = 7.42 years, SD = 0.85). All participants were in Grade 1 for the first time, attending English medium schools. We chose this age group as it marks the commencement of formal education in South Africa and is a valuable point at which to identify children who could benefit from additional educational assistance. Participants were drawn from public hospitals in Johannesburg, Gauteng. Inclusion criteria were an HIV positive status, on combination antiretroviral treatment (cART), viral suppression for at least 6 months (which includes adherence and a positive response to the cART, with no debilitating side effects from the medication, stable CD4 counts and viral loads). The HIV-EU group consisted of 86 children (47 girls, mean age = 7.36 years, SD = 0.88). These children were HIV negative, but their biological mother was HIV positive at the time of their birth. The participants received PMTCT at birth and subsequently seroconverted. The HIV-UU group comprised 92 children (55 girls; mean age = 7.05 years, SD = 0.86). Both they and their biological mother were HIV negative. Exclusion criteria for all groups were attendance at specialized schooling, institutionalization in an orphanage, or neurological compromise, such as epilepsy, traumatic brain injury, and/or previous diagnoses of Meningitis or Encephalitis.

## Measures

Each participant was individually assessed in a quiet room for one session of approximately 1 h. Measures were administered in a fixed order, designed to vary task demands and minimize participant fatigue. All tests were administered in English.

#### Working Memory

The 12 tests from the Automated Working Memory Assessment (AWMA; Alloway, 2007), an individual, computer based battery were administered. The AWMA comprises three tests of each component of Baddeley's (2000) working memory model,

namely Verbal (phonological loop) and Visuospatial (visuospatial sketchpad) Storage, Verbal and Visuospatial Processing (tapping central executive resources). It has a pre-determined sequence, is automatically scored and converted to a standard score based on the participant's age (M = 100; SD = 15).

#### **Verbal storage**

These span tasks (Digit Recall, Word Recall, and Non-word Recall) measure the storage capacity of the phonological loop using different types of verbal material. In these tasks, either digits, words or non-words are stated in sequences that increase in number over each trial, starting with two items. The participant must recall them in the same order in which they were heard.

#### **Verbal processing**

Phonological loop and central executive functioning were measured in the complex span tasks of Listening Recall, Counting Recall, and Backward Digit Recall. In Listening Recall, participants judge the legitimacy of a spoken sentence by noting it as 'true' or 'false,' and must recall the final word of each sentence in sequence, after hearing a minimum of one and a maximum of six sentences. For Counting Recall, an array of shapes are presented, and participants must count and report the red circles, and then attempt to recall the total red circles for each array, in the original sequence. In Backward Digit Recall, participants must reverse the order of a sequence of heard digits, starting with two digits.

#### **Visuospatial storage**

Visuospatial sketchpad functioning (storage only) was measured with the Dot Matrix, Mazes Memory, and Block Recall tasks. In Dot Matrix, participants view four-by-four matrices and must identify the location of a previously shown red dot, by tapping on the correct square on the computer screen. For Mazes Memory, participants view a maze with a red pathway drawn through its course, and after a three second delay, must trace the path on a blank maze. In Block Recall, participants view a series of tapped blocks, and must reproduce the same sequence by tapping on each block on the screen.

#### **Visuospatial processing**

Visuospatial sketchpad and central executive functioning were measured with the Odd One Out, Mister X, and Spatial Recall tests. The Odd One Out test comprises three shapes, each presented in a row, and participants must detect the shape that is odd. At the end of each presentation (starting with one and reaching a maximum of six rows), participants must tap on the screen to recall the location of each odd-one-out shape in the correct order presented. In Mister X, pictures of two Mister X characters are shown, each wearing different colored hats, each holding a red ball, and each positioned in different orientations. Participants must identify whether the Mister X with the blue hat is holding the ball in the same hand as the Mister X with the yellow hat. At the end of six presentations, participants must recall, in the correct order, the position of each red ball by pointing to its location on the screen. The Spatial Recall test presents two objects (the target image has a red dot above it), and participants must identify whether the target object is identical or opposite of another presented object. The position of the red dot must be recalled at the end of each set of six presentations by pointing to its location on the screen.

Each test starts with a series of three practice trials, immediately followed by test trials, which progressively increase in difficulty. On practice trials, the correct response is given following the participant's response, while no feedback is given on test trials. Each level offers six attempts, four of which have to be correct to proceed to the next level. Each level increases in difficulty with an added length to the item. Reliability and validity of the AWMA with British children are reported in Alloway et al. (2006, 2008). This test has not been standardized for South African children, but has been used in studies of South African children (Cockcroft and Alloway, 2012; Cockcroft, 2016; Cockcroft et al., 2016).

#### Non-verbal Intelligence

The Ravens Colored Progressive Matrices (RCPM; Raven et al., 1998) measured general intellectual ability to control for potential differences in this regard between the groups. The RCPM is a culturally reduced, non-verbal test consisting of 36 items in three sets, with 12 items per set. A single raw score is produced that can be converted to a percentile. The RCPM has good retest and split-half reliability, with no gender or ethnicity differences (Raven et al., 1990). Validity studies comparing the RCPM and WISC found strong correlations (0.91, 0.84, and 0.83) between the Full Scale, Verbal and Performance IQs, respectively (Martin and Wiechers, 1954).

#### English Language Proficiency

Since assessment was undertaken in English, which was not the mother tongue of the participants, the Sentence Repetition Test (SRT; Redmond, 2005) was used as a brief measure of English language proficiency in order to determine whether participants were proficient enough to complete the tests, and also to control for potential language differences between the groups. The SRT includes 16 ten-word sentences, each between 10 and 14 syllables long, with an even number of active and passive sentences. The participant must recall and repeat the sentences exactly as they are read and are scored either a 0, 1, or 2 based on their performance. The SRT was originally developed as a screen for children with Specific Language Impairment, however, it was subsequently found to be particularly sensitive to tapping the English proficiency of children who are not first language English speakers (Komeili et al., 2012; Komeili and Marshall, 2013). Inter-rater reliability was calculated by independent comparisons of marked responses (number of agreements/number of agreements + number of disagreements), and a value of 95% (sentence recall probe) and 98% (past tense elicitation probe) were found (Redmond, 2005).

#### Socioeconomic Status

The inclusion of a measure of SES was important due to its relationship with neurocognitive development and HIV infection and exposure (Coscia et al., 2001; Dobrova-Krol et al., 2010). The Living Standard Measure (LSM; South African Advertising Research Foundation [SAARF], 2012) is the industry standard

when considering consumer patterns in South Africa. It is not dependent on reports of income or personal demographics (employment status, race, age, gender), and instead collects information about access to basic facilities, ownership of appliances and other assets, and factors related to residential location and type of dwelling in order to measure living standards as a proxy for SES. The measure is a 30 item binary questionnaire which marks the presence or absence of an appliance or facility in the family home (i.e., dishwasher, TV, mobile phone) as an indicator of wealth.

The study was approved by the Medical Research Ethics Committee of the University of the Witwatersrand, Johannesburg. Parents/guardians of participants provided informed, written consent, while participants granted assent to participate. There were appropriate opportunities for withdrawal at any point without prejudice.

## RESULTS

## Descriptive Statistics

Skew and kurtosis for all of the variables met the criteria for univariate normality (Kline, 2005). The groups were balanced in terms of gender [HIV-I(female) = 51.6%; HIV-EU(female) = 53.41%; HIV-UU(female) = 59.78], while there were significant between group differences on age [F(2,272) = 4.93; p = 0.008], SES [F(2,272) = 11.03; p < 0.001], English proficiency [F(2,272) = 31.29; p < 0.001] and intelligence [F(2,272) = 24.764; p < 0.001]. Descriptive statistics for the working memory, intelligence, English proficiency and SES measures are shown in **Table 1**, as well as analyses of variance between the three groups on these variables. Working memory composites reflect the average of the three tests that measure each of the following components: verbal storage, verbal processing, visuospatial storage and visuospatial processing.

## Between Group Comparisons

A multivariate analysis of covariance (MANCOVA) between the working memory tests, with age, SES, language proficiency and intelligence as covariates, was significant for group [Wilk's λ = 0.80, F(8,526) = 7.98, p < 0.0001, η 2 <sup>p</sup> = 0.11, power = 1, Hotelling's Trace = 0.244, F(8,524) = 7.85, p < 0.0001, η 2 <sup>p</sup> = 0.11, power = 1]. Subsequent univariate analyses of covariance (ANCOVAs) indicated that significant between group differences were present in all composite scores except Visuospatial Storage [Verbal Storage F(2,270) = 11.28, p < 0.001, η 2 <sup>p</sup> = 0.078; Verbal Processing F(2,270) = 18.16, p < 0.0001, η 2 <sup>p</sup> = 0.12; Visuospatial Processing F(2,270) = 7.36, p < 0.001, η 2 <sup>p</sup> = 0.053; refer to **Table 1**]. The Bonferroni correction method was used to protect against an inflated familywise error rate, due to multiple comparisons, and therefore α was set at 0.0125.

Following from the ANCOVAs, pairwise comparisons (incorporating the same covariates mentioned above) highlight the differences between groups on the working memory composite scores (See **Table 2**). The Bonferroni correction for multiple comparisons (α = 0.0125) was used. The HIV-I group fared significantly poorer than the HIV-UU group on processing tasks (p < 0.001) irrespective of modality (verbal or visuospatial). The former group performed on a par with the neurotypical controls on the storage tasks across both modalities. In contrast, the HIV-EU group performed significantly worse than the HIV-UU group on tasks tapping the verbal modality, irrespective of whether their focus was on storage or processing (p < 0.001). The HIV-EU group's performance was not significantly different from the neurotypical controls on measures drawing on the visuospatial domain, irrespective of whether these required storage or processing. The HIV-I and HIV-EU groups did not differ significantly from one another on any of the working memory composites.

## Within Group Comparisons

In order to determine the individual working memory profiles (relative strengths and weaknesses) for each group, repeated measures analyses of variance (rANOVA) were conducted between the four working memory composites within each group (see **Table 3**). For the HIV-I group, Verbal Processing was significantly weaker than the other three composites (Verbal Storage: p < 0.001, d = −0.65; Visuospatial Storage: p = 0.001, d = −0.41; Visuospatial Processing: p < 0.001, d = −0.72) with medium effect sizes. Visuospatial Processing was significantly stronger than Visuospatial Storage (p = 0.01, d = −0.28), but the effect size was small. For the HIV-EU group, post hoc analyses showed that Visuospatial Processing was significantly better than the other three composites (Verbal Storage: p < 0.001, d = 0.83; Verbal Processing: p < 0.001, d = 0.86; Visuospatial Storage: p < 0.001, d = 0.58), with medium to large effect sizes. The two verbal composites appear to be weaknesses as they have relatively lower scores, with no significant difference between the processing and storage components. For the HIV-UU group, Visuospatial Processing was significantly stronger than both Verbal Processing (p < 0.001, Cohen's d = −0.44) and Visuospatial Storage (p < 0.001, Cohen's d = 0.56), with medium effect sizes. Visuospatial Processing appears to be the strongest composite for this group, followed by Verbal Storage, Verbal Processing, and Visuospatial Storage in that order.

In summary, while the HIV-EU and HIV-I groups' performance did not differ significantly from one another on any of the working memory measures, when compared with the HIV-UU group, they revealed different areas of deficit. In particular, the HIV-I group showed difficulties with processing tasks irrespective of modality, while the HIV-EU group showed difficulties with verbal tasks, irrespective of whether they drew on storage of processing. When all the tasks were considered, both the HIV-I and HIV-EU groups showed the greatest relative difficulty with Verbal Processing.

## DISCUSSION

The objective of this study was to investigate whether working memory performance differed between HIV-I children, HIV-EU children, and a group of neurotypical controls. While the overall results might suggest equivalent memory functioning between

TABLE 1 | Descriptive statistics, ANCOVAs for working memory tests, ANOVAs for age, intelligence, English proficiency and SES by group.


<sup>∗</sup>0.0125 correction for ANCOVAs between working memory tests after covariation of age, intelligence, English proficiency, and SES scores; ∗∗0.05 for all other ANOVAs.

TABLE 2 | Pairwise comparisons between groups on the four working memory composites.


<sup>∗</sup>p < 0.0125 (Bonferroni correction).

the HIV-I and HIV-EU groups, there were some important differences in memory functioning between them that emerged in the comparisons with the HIV-UU group. In comparison to the control group, the HIV-I group showed difficulties with processing tasks, regardless of whether they drew on a verbal or visuospatial modality. On the other hand, the HIV-EU group experienced difficulty with tasks that drew on the verbal modality, irrespective of whether they were storage or processing based. It is likely that these different working memory profiles stem from different etiologies.

Understanding these findings requires some detail about the measures used in this study. Tasks designed to tap working memory processing are complex, requiring simultaneous storage and processing of information, and thus draw on multiple cognitive functions including attention and long-term memory (Baddeley and Logie, 1999; Cowan, 1999; Duff and Logie, 2001). The complex memory span tasks used in the current study impose significant burdens on concurrent processing associated with the central executive (general working memory system), while the phonological and visuospatial storage/short-term


memory tasks impose minimal processing loads, tapping instead the storage capacity of the respective working memory storage systems (Baddeley and Logie, 1999). Alternative theoretical accounts of complex working memory processing, where performance is supported by a unitary limited capacity resource that can support processing and storage in isolation or jointly, can also be explained by these tasks (Just and Carpenter, 1992; Cowan, 1999). Cognitive processing problems would therefore be expected to manifest themselves to a greater extent on the processing tests than the storage only tasks. This was evident when these tasks were compared within the HIV-I group; performance was significantly poorer on processing tasks relative to storage tasks.

The HIV-I children's working memory profile (poor central executive processing) is similar to that of children with general learning difficulties and AD/HD (Bull and Scerif, 2001; Pickering and Gathercole, 2004; Alloway, 2011). Our finding of a processing impairment in the HIV-I sample is supported by evidence of impaired visuospatial working memory processing in HIV-I adult samples (York et al., 2001; Hinkin et al., 2002; Heaton et al., 2004; Dawes et al., 2008), and in two pediatric samples (Koekkoek et al., 2008; Boivin et al., 2010b). In our sample, we found that this processing impairment also impacted verbal working memory. Verbal working memory has not been generally investigated in other studies. This processing impairment may be linked to the diminished capacity of frontostriatal white matter networks, which are implicated in working memory processing and broader executive control (Van Rie et al., 2007). Since working memory shares integral links with other executive functions, particularly inhibitory control (Miyake and Friedman, 2012), it is most likely that the working memory impairment is secondary to a more general processing deficit, and is not in itself the cause of the difficulty. Due to these shared underlying processes, remediating working memory often has positive effects on broader executive functioning as well (Miyake and Friedman, 2012; Blakey and Carroll, 2015). In keeping with this line of reasoning, Hinkin et al. (2002) propose that working memory deficits across both verbal and visuospatial domains are a result of executive functioning impairment secondary to HIV infection, and not a result of localized damage of the more isolated cortical regions housing the verbal or spatial stores (left and right dorsal and ventral pathways; Van Rie et al., 2007; Makuuchi and Friederici, 2013). This argument is also supported by evidence that stimulant medication administered to treat comorbid AD/HD in HIV-I children improves both inhibitory control and working memory (Mehta et al., 2004). While such pharmacological treatments may benefit HIV-I children, behavioral interventions, such as working memory interventions, may also be of value, as these have demonstrated success with children with AD/HD (Klingberg, 2010). Impaired executive processes in the HIV-I group probably account for their considerably poorer performance with processing tasks.

In contrast, the HIV-EU group performed poorly on both storage and processing tasks that drew on the verbal modality. Their visuospatial processing and storage were on a par with that of the neurotypical controls. This does not suggest a general

TABLE 3 | Repeated measures ANOVAs comparing

 working memory performance

 within each group.

fnhum-11-00348 July 4, 2017 Time: 16:3 # 8

learning deficit, but rather a language-related difficulty (Pickering and Gathercole, 2004; Archibald and Gathercole, 2006; Pickering, 2006). Given that these children were assessed in their second language (English), and that they were in their first year of formal schooling in this language, it is most likely that the increased load of complex processing together with use of a second language meant that demand exceeded capacity in the verbal processing tasks for the HIV-EU group. There is evidence that the executive and attentional components of working memory are linked to long-term memory (e.g., Cowan, 1999; Oberauer et al., 2005). As such, the availability of language knowledge stored in long-term memory is likely to influence the efficiency of verbal working memory (Majerus, 2013). The capacity to process and store information, as captured in complex verbal working memory tasks, may be crucial to support learning generally (Pickering and Gathercole, 2004; Gathercole et al., 2006). This finding suggests that mother tongue instruction in the Foundation phase years would benefit these children and that too early a transition to second language instruction in English could be detrimental. It is important to note that the capacity of verbal working memory is likely to change as literacy development progresses (Conant et al., 1999; Conant et al., 2003; Boivin et al., 2010a). We did not measure early literacy skills, such as phonological awareness and alphabetic knowledge, in this study and so cannot comment on the extent to which the development (or lack thereof) of these skills may underpin this group's weaker verbal working memory.

It is not possible to rule out general neurocognitive compromise in the context of HIV-exposure as an explanatory framework for the HIV-EU group. This may stem from the effects of ART toxicity and/or chronic maternal viral infection which disrupts neurogenesis in the fetus, and may result in permanent structural changes in the developing brain (Garay and McAllister, 2010). The fact that the HIV-EU group's working memory performance generally fell in between that of the HIV-I and controls, suggests some depression of functioning in this group. Consequently, both the HIV-I and HIV-EU groups are likely to benefit from an integrated and targeted working memory intervention that focuses on strategies to promote working memory functioning, especially with regard to verbal processing.

A finding worth comment was the absence of impairment in visuospatial storage in both HIV-affected groups. Impairment in short-term visuospatial storage in HIV-I adults is usually characteristic of late stage infection, and is unlikely to show significant compromise during periods of good health (Reger et al., 2002). In addition, there is evidence for the dominance of visuospatial encoding in preschool children in keeping with the earlier maturation of the visual areas of the brain. This dominance begins to shift to a gradually increasing reliance on verbal encoding from approximately 6 years, the stage at which most children are exposed to literacy instruction (Hitch et al., 1988, 1993; Conant et al., 2003; Alloway et al., 2006; Boivin et al., 2010a). The finding that the visuospatial shortterm stores in the two HIV-affected groups were no poorer than that of the control group could reflect this relative proficiency, as the participants were all school beginners who were just starting to negotiate this transition. This strength could be developed to compensate for poor verbal storage and processing capacities. Children with weaker verbal storage capacity may be able to capitalize on imagery or other kinds of visuospatial mediation to overcome some of their verbal learning difficulties. Strategic visuospatial mediation may be particularly valuable in mathematics (McLean and Hitch, 1999) and literacy (Johnston and Anderson, 1998), but less useful in the context of general language learning, as phonological forms are the basic representational medium (Pickering and Gathercole, 2004).

There are some limitations of the current study that warrant acknowledgment. Although children with formal diagnoses of AD/HD and learning disabilities were excluded from the study, it is possible that these difficulties may have been undiagnosed, particularly since the samples came from low SES circumstances with limited access to specialized educational or health care. Further, the effects of HIV infection and exposure without infection are accentuated by a host of socio-economic and psychosocial factors including additional illness, poor nutritional status, caregiver stress, and adverse living conditions (Walker et al., 2011). Studies where these risks were covaried (Floyd et al., 2007; Walker et al., 2011) found that the cognitive and motor deficits in HIV-EU children became non-significant. A socio-demographic analysis of the three groups in the current study indicated that the HIV-EU group were particularly disadvantaged in comparison to the other three groups. Their SES was the poorest and significantly lower than that of the HIV-UU group (p = 0.015); they had the highest familial burden of care (looking after an immediate family member with special needs) (21.95%), the highest proportion of recipients of a child support grant (91.9%), the lowest levels of maternal education (only 21.8% finishing high school), and the highest levels of paternal absenteeism (53.4%). In contrast, HIV-I children are frequently followed up on by specialized ART clinics which offer them support and access to social and allied therapeutic services, while the HIV-EU children seldom receive any of these auxiliary services, a finding replicated elsewhere (Kerr et al., 2014).

On the positive side, a strength is that this appears to be the first study to give a detailed comparative analysis of the working memory functioning of HIV-I and HIV-EU children. The participants were from poor socioeconomic backgrounds, and are representative of young children who access the public health system in South Africa. Thus, the results from this study should be generalizable to such a population where the incidence of HIV infection and exposure are highest, and which is most in need of support and intervention.

## CONCLUSION

There are widespread concerns about the early developmental wellbeing and loss of intellectual potential in South African children due to poverty, poor health and nutrition, and deprived environments (Grantham-McGregor et al., 2007). Of the reasons for this, lack of stimulation and HIV infection

feature prominently (Walker et al., 2007). A key reason for the failure of HIV-I and HIV-EU children to attain their developmental potential could be because of poor working memory functioning since this would adversely affect their ability to mentally hold and manipulate information, skills vital for learning. Such failure would have farreaching repercussions on long-term development and functioning. Early intervention at the level of working memory is clearly needed for HIV-affected children, and could have positive consequences for these children's academic functioning, as well as their social and emotional efficacy.

## AUTHOR CONTRIBUTIONS

KC helped to conceptualize this study, supervised the data collection and analyses, and wrote up the paper. RM helped to

## REFERENCES


conceptualize this study, collected the data and analyzed it, and assisted with writing up the paper.

## FUNDING

The financial assistance of the National Research Foundation (NRF) toward this research is hereby acknowledged. Opinions expressed and conclusions arrived at, are those of the authors and are not necessarily to be attributed to the NRF.

## ACKNOWLEDGMENTS

This study formed part of the following Doctoral thesis: Milligan, R. (2016). A comparison of working memory profiles in HIV positive and HIV-exposed, uninfected children. Unpublished Doctoral thesis, University of the Witwatersrand, Johannesburg.


with human immunodeficiency virus. Pediatrics 106, 1–10. doi: 10.1542/peds. 106.2.e25



of life in south africa: the IeDEA-SA collaboraion. PLoS ONE 11:e0151762. doi: 10.1371/journal.pone.0151762



in developing countries. Lancet 369, 145–157. doi: 10.1016/S0140-6736(07) 60076-2


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Milligan and Cockcroft. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Reductions in Corpus Callosum Volume Partially Mediate Effects of Prenatal Alcohol Exposure on IQ

Stevie C. Biffen<sup>1</sup> \*, Christopher M. R. Warton<sup>1</sup> , Nadine M. Lindinger <sup>1</sup> , Steven R. Randall <sup>1</sup> , Catherine E. Lewis <sup>1</sup> , Christopher D. Molteno<sup>2</sup> , Joseph L. Jacobson1, 2, 3 , Sandra W. Jacobson1, 2, 3 and Ernesta M. Meintjes 1, 4 \*

<sup>1</sup> Department of Human Biology, Faculty of Health Sciences, University of Cape Town, Cape Town, South Africa, <sup>2</sup> Department of Psychiatry and Mental Health, Faculty of Health Sciences, University of Cape Town, Cape Town, South Africa, <sup>3</sup> Department of Psychiatry and Behavioral Neurosciences, Wayne State University School of Medicine, Detroit, MI, United States, <sup>4</sup> MRC/UCT Medical Imaging Research Unit, Division of Biomedical Engineering, Department of Human Biology, Faculty of Health Sciences, University of Cape Town, Cape Town, South Africa

#### Edited by:

Nilesh Bhailalbhai Patel, University of Nairobi, Kenya

#### Reviewed by:

Corrado Romano, Unit of Pediatrics and Medical Genetics, IRCCS Associazione Oasi Maria Santissima, Italy Lauren Jantzie, University of New Mexico, United States Alberto Granato, Università Cattolica del Sacro Cuore, Italy

#### \*Correspondence:

Stevie C. Biffen steviebiffen@gmail.com Ernesta M. Meintjes ernesta.meintjes@gmail.com

Received: 28 July 2017 Accepted: 18 December 2017 Published: 12 January 2018

#### Citation:

Biffen SC, Warton CMR, Lindinger NM, Randall SR, Lewis CE, Molteno CD, Jacobson JL, Jacobson SW and Meintjes EM (2018) Reductions in Corpus Callosum Volume Partially Mediate Effects of Prenatal Alcohol Exposure on IQ. Front. Neuroanat. 11:132. doi: 10.3389/fnana.2017.00132 Disproportionate volume reductions in the basal ganglia, corpus callosum (CC) and hippocampus have been reported in children with prenatal alcohol exposure (PAE). However, few studies have investigated these reductions in high prevalence communities, such as the Western Cape Province of South Africa, and only one study made use of manual tracing, the gold standard of volumetric analysis. The present study examined the effects of PAE on subcortical neuroanatomy using manual tracing and the relation of volumetric reductions in these regions to IQ and performance on the California Verbal Learning Test-Children's Version (CVLT-C), a list learning task sensitive to PAE. High-resolution T1-weighted images were acquired, using a sequence optimized for morphometric neuroanatomical analysis, on a Siemens 3T Allegra MRI scanner from 71 right-handed, 9- to 11-year-old children [9 fetal alcohol syndrome (FAS), 19 partial FAS (PFAS), 24 non-syndromal heavily exposed (HE) and 19 non-exposed controls]. Frequency of maternal drinking was ascertained prospectively during pregnancy using timeline follow-back interviews. PAE was examined in relation to volumes of the CC and left and right caudate nuclei, nucleus accumbens and hippocampi. All structures were manually traced using Multitracer. Higher levels of PAE were associated with reductions in CC volume after adjustment for TIV. Although the effect of PAE on CC was confounded with smoking and lead exposure, additional analyses showed that it was not accounted for by these exposures. Amongst dysmorphic children, smaller CC was associated with poorer IQ and CVLT-C scores and statistically mediated the effect of PAE on IQ. In addition, higher levels of PAE were associated with bilateral volume reductions in caudate nuclei and hippocampi, effects that remained significant after control for TIV, child sex and age, socioeconomic status, maternal smoking during pregnancy, and childhood lead exposure. These data confirm previous findings showing that PAE is associated with decreases in subcortical volumes and is the first study to show that decreases in callosal volume may play a role in fetal alcohol-related impairment in cognitive function seen in childhood.

Keywords: fetal alcohol spectrum disorders, MRI, subcortical volumes, corpus callosum, IQ

## INTRODUCTION

Prenatal alcohol exposure (PAE) is associated with a range of neurocognitive and behavioral problems. Fetal alcohol spectrum disorders (FASD) is an umbrella term under which the entire spectrum of outcomes related to PAE falls. Fetal alcohol syndrome (FAS), the most severe of the FASD, is characterized by growth deficits, small head circumference, distinctive craniofacial dysmorphology (small palpebral fissures, flattened philtrum, and thin vermillion) and cognitive problems that relate to specific brain abnormalities (Jones and Smith, 1973; Hoyme et al., 2005). Partial FAS (PFAS) is diagnosed in individuals exhibiting the facial dysmorphology characteristic of FAS, whose mothers are known to have drunk heavily during pregnancy and who exhibit growth deficits, small head circumference or neurobehavioral impairment (Hoyme et al., 2005). Adverse effects on brain development range from microstructural, neurochemical and cellular dysfunction (such as widespread apoptotic neurodegeneration that can result in the death of millions of cells in the forebrain during gestation) to gross structural abnormalities (Ikonomidou, 2000; Olney, 2004; Spadoni et al., 2007; Lebel et al., 2008).

In the U.S. and in Western countries, FAS occurs at a rate estimated to be between 0.05 and 7 per 1,000 live births (May and Gossage, 2001; May et al., 2009). In contrast, certain highrisk communities in South Africa have been found to have an FAS prevalence rating that is 18–148 times greater than in the U.S. (May et al., 2000; Viljoen et al., 2005; Urban et al., 2008). Within the Western Cape province of South Africa, rates are highest amongst the Colored population of mixed Asian, African and European ancestry (May et al., 2000, 2005, 2013; Viljoen et al., 2005).

Magnetic resonance imaging (MRI) allows for non-invasive and quantitative investigation of the structural and functional brain alterations that are observed in FASD. Brain regions that have been consistently implicated in FASD include the corpus callosum (CC), cerebellum, parietal lobe, frontal lobe, temporal lobe, caudate nucleus, hippocampus and increased volume of lateral ventricles (see review papers, Lebel et al., 2011; Donald et al., 2015). While most regions show alcohol-related volume reductions, findings in certain subcortical brain regions have not always been consistent. For example, some studies report bilateral decreases in hippocampal volume (Willoughby et al., 2008; Astley et al., 2009; Coles et al., 2011; Nardelli et al., 2011; Treit et al., 2013) with PAE, while others find relative sparing (Archibald et al., 2001), or even volume increases (Riikonen et al., 2007). Methodological differences, such as automated vs. manual segmentation, failure or different strategies to control for total intracranial volume (TIV), small sample sizes, or large age ranges may explain some of these differences. Notably, one study found increased hippocampal volume relative to brain volume with age (9–15 years) in controls, but not in children with FASD (Willoughby et al., 2008), highlighting the potential contribution of age-related changes in brain development when comparing diagnostic groups and the need for narrow age ranges. It is not yet known to what extent developmental trajectories in children prenatally exposed to alcohol are altered.

Although manual tracing is considered the gold standard for volumetric analysis, yielding the greatest consistency and sensitivity (Morey et al., 2009), few studies employ this method due to high time demands (Archibald et al., 2001; Riikonen et al., 2007; Willoughby et al., 2008; Astley et al., 2009). Here we use manual tracing of study participants within a narrow age range (10.4 ± 0.4 years) to examine the effects of PAE on volumes of select subcortical structures and the CC in 81 children from the Cape Town Longitudinal Cohort (Jacobson et al., 2008), for whom diagnostic and detailed prospective prenatal alcohol and drug exposure data are available. Most previous studies have only been able to examine volumetric differences between diagnostic groups and not dose-dependent effects, since the amount of alcohol used during pregnancy is difficult to recall reliably in retrospective case-controlled studies (Jacobson et al., 2002).

Due to the labor intensive nature of manual tracing, only subcortical structures previously implicated in FASD were investigated (Archibald et al., 2001; Morey et al., 2009; Nardelli et al., 2011). These include the caudate nucleus and hippocampus, which are involved in cognition, memory and emotional networks (Devinsky and D'Esposito, 2004) that are amongst the most affected domains in FASD (Uecker and Nadel, 1996; Mattson et al., 1998; Willoughby et al., 2008; Green et al., 2009; Jacobson et al., 2011; Lewis et al., 2016). Although findings in the nucleus accumbens have rarely been reported in FASD (Archibald et al., 2001), we report it here because the manual tracing protocol we used for the caudate initially involved tracing both structures together. The CC was included as it is clearly visible on MRIs, is easy to delineate, and has been consistently shown to be reduced in FASD (Mattson et al., 1992; Sowell et al., 2001, 2008; Bookstein et al., 2002; Autti-Rämö et al., 2007; Lebel et al., 2008; Astley et al., 2009), even in newborns (Jacobson et al., 2017). PAE has been linked to a larger angle of the splenium (Bookstein et al., 2007), reductions in CC thickness and area in the splenium and anterior third (Yang et al., 2012), and CC volume reductions in the splenium, genu and area just anterior to the splenium (Riley et al., 1995).

In addition, we examined whether alcohol-related regional volumetric changes mediate the effect of PAE on cognitive performance on the Wechsler Intelligence Scale for Children-Fourth Edition (WISC-IV) and the California Verbal Learning Test-Children's Version (CVLT-C). Poorer IQ is frequently seen in FASD (Streissguth et al., 1990, 1991; Jacobson et al., 2004) as is impaired performance on the CVLT-C (Delis et al., 1994), a list learning task (Kerns et al., 1997; Mattson et al., 1998; Crocker et al., 2011; Vaurio et al., 2011; Suttie et al., 2013; Lewis et al., 2015).

We hypothesized that increasing PAE would be associated with volumetric reductions in all regions examined and that these effects would remain significant after controlling for reductions in total brain volume. We also hypothesized that volumetric reductions in hippocampus, caudate nucleus and CC would be associated with poorer WISC IQ and CVLT-C performance and that the effects of PAE on these outcomes would be partially mediated by volumetric changes in the regions that were traced.

## METHODS

## Participants

From 1998 to 2002 pregnant women were recruited into the Cape Town Longitudinal Cohort (Jacobson et al., 2008) from an antenatal clinic in a Cape Colored community in Cape Town known to have a high prevalence of alcohol consumption (Croxford and Viljoen, 1999). Detailed alcohol exposure data collected prospectively during pregnancy were available for all of these children. Mothers were interviewed at enrollment and at two subsequent visits using the timeline follow-back (TLFB) approach to record alcohol consumption during pregnancy (Jacobson et al., 2002). We adapted the interview to include information about the type of beverage consumed, container size (including pictures of different containers, bottles, cans, glass size) and sharing (size of container divided by number of women drinking together) to reflect how pregnant women in this community tend to drink. The mother was asked about her drinking on a day-by-day basis during a typical 2-week period around the time of conception, with recall linked to specific times of day and activities. If her drinking had changed since conception, she was also asked about her drinking during the past 2 weeks and when her drinking had changed. At the two follow-up visits, the mother was again asked about her drinking during the previous 2 weeks. Volume was recorded for each type of alcohol beverage consumed each day and converted to oz of absolute alcohol (AA) using multipliers proposed by Bowman et al. (1975) (liquor—0.4, beer—0.05, wine—0.12, cider—0.06). Six summary measures were constructed—average oz AA per day at conception, AA per day averaged across pregnancy, AA per drinking day (quantity per occasion) at conception and across pregnancy, and number of drinking days per week (frequency) at conception and across pregnancy.

Any woman who reported drinking at least 1.0 oz AA per day (≈2 standard drinks per day) or at least 2 instances of binge drinking (≥5 drinks per occasion) during the first trimester was invited to participate. Women initiating antenatal care who reported no binge drinking and minimal alcohol use (<0.5 oz AA/ day) were invited to participate as controls. In this sample all but one of the mothers in the control group reported abstaining from drinking during pregnancy; the single control mother who drank consumed only 2 drinks/occasion on 2–3 days during the pregnancy. Smoking was recorded as the number of cigarettes smoked per day. Participants were excluded if they were <18 years of age, presented with chronic medical problems including epilepsy, diabetes, HIV, or cardiac problems. Infant exclusionary criteria were major chromosomal anomalies, neural tube defects, multiple births, and seizures.

In 2005, we organized a clinic in which the children in the study were each independently examined by two expert dysmorphologists (E.H. Hoyme, M.D., and L.K. Robinson, M.D.) for growth and FAS related dysmorphic features using a standard protocol (Hoyme et al., 2005) based on the Revised Institute of Medicine criteria (Jacobson et al., 2008). Case conferences were held in which the dysmorphologists, SWJ, JLJ, and CDM determined which of the children met criteria for FAS or PFAS. Children who did not meet criteria for FAS of PFAS were designated as either non-syndromal heavily exposed (HE) or controls, depending on the maternal alcohol history. Five children who could not attend the 2005 clinic were examined by another expert FASD dysmorphologist (N. Khaole, M.D.), whose diagnoses were all subsequently confirmed by examinations conducted in follow-up clinics we held with the same dysmorphologists in 2009 and with HEH in 2013 and 2016.

## Magnetic Resonance Image (MRI) Acquisition

Magnetic resonance image (MRI) scans were acquired at the Cape Universities Brain Imaging Centre (CUBIC) using a 3T Siemens Allegra MRI scanner (Siemens Medical Systems, Erlangen, Germany) for 81 right-handed children from this cohort at age 9–11 years. High-resolution T1 weighted images were obtained using a volumetric navigated (Tisdall et al., 2012) multi-echo magnetization prepared rapid gradient echo (MEMPRAGE) sequence optimized for morphometric neuroanatomical analysis using FreeSurfer software (van der Kouwe et al., 2008). Imaging parameters were: FOV 256 × 256 mm, 128 sagittal slices, TR 2,530 mm, TE 1.53/3.21/4.89/6.57 mm, TI 1,100 ms, flip angle 7◦ , voxel size 1.3 × 1.0 × 1.3 mm<sup>3</sup> , acquisition time 8:07 min. The volumetric navigator provides real-time motion tracking and correction to reduce artifacts resulting from motion.

Human subject participation was approved by the ethics committees of the University of Cape Town Faculty of Health Sciences and Wayne State University. Mothers provided written informed consent and children, oral assent. Examiners were blind with respect to PAE history and FASD diagnosis, except in the most severe cases of FAS. Total intracranial volumes (TIV) were obtained from FreeSurfer (Dale et al., 1999; Fischl et al., 2002). Called eTIV or ICV, within FreeSurfer, the programme calculates TIV by using an atlas scaling factor that it obtains from registering the images from the sample to an average template and using that scaling factor to estimate each participant's TIV (Buckner et al., 2004, see also http://surfer.nmr.mgh.harvard. edu/fswiki/eTIV).

## Manual Tracing Protocol

The CC and left and right caudate nuclei, nucleus accumbens and hippocampi were manually traced on the MR images using Multitracer (Woods, 2003) software on a tablet laptop (Lenovo ThinkPad X200, Wacom digitizer pen enabled). A single blinded investigator performed all manual tracings (SCB). Tracing was supervised and reviewed by CW, the senior neuroanatomist at the University of Cape Town. All structures were traced in the coronal plane, with the exception of the CC that was traced in the sagittal plane. Orthogonal planes were used to inform border delineation. "Frust" volumes were recorded for each structure, which assumes that a contour applies to the center of the slice on which it is drawn and that the square root of the cross-sectional area changes linearly across the slice thickness (Woods, 2003).

The CC (**Figure 1A**) is clearly visible as a curved WM structure near the midline slice. The superior border is the cingulate gyrus and the CSF (cerebrospinal fluid) of the lateral

and posterior fissure form the inferior border. The midline slices (about slices 127–130) were determined by locating the smallest area of the thalamus. Bookstein et al. (2002) used the term "midline" to denote the slices located most medially in the CC (i.e., along the midline of the brain). The CC was traced on two contiguous midline slices. These were averaged and volume calculated.

The caudate nucleus (**Figure 1B**) is a gray matter (GM) structure that lies on the lateral wall of the lateral ventricle and is traced until it disappears. The medial border is the lateral ventricle and the lateral border is formed by WM. The inferior border is the nucleus accumbens (**Figure 1C**), which can only be differentiated histologically by cytoarchitecture, however, it can be removed from the caudate nucleus reliably by drawing a straight line from the inferior border of the lateral ventricle to the bottom-most border of the internal capsule. Laterally, it is separated from the putamen by a line descending from the internal capsule. It was traced until the anterior commissure bridges the midline.

The anterior border of the hippocampus (**Figure 1D**) is the inner surface of the alveus. Tracing proceeds past the crura of the fornices until hippocampal GM disappears. The medial border includes the subiculum to the supramedial border of WM in the temporal lobe. The lateral border is the inner surface of the alveus within the temporal horn of the lateral ventricle. Inferiorly, the boundary is the interface between the WM and GM.

Three participants (2 HE; 1 control) were excluded due to pathology (unrelated to the research question) and 7 (1 FAS; 2 HE; 4 control) due to compromised image quality. Pathology included one scan where there were hypointense inclusions on the superior surface of the right lateral ventricle (indicative of a potentially conflicting diagnosis), and two scans where the lateral ventricles were occluded, making delineation of caudate nuclei and nucleus accumbens problematic.

## Statistical Analyses

Analysis of variance (ANOVA) was performed for each region of interest (ROI) to examine whether structural volumes differed between diagnostic groups. Pearson correlation was used to examine the relations of ROI volumes to each of the potential confounders. Potential control variables included TIV, child's sex and age (years) at scan, socioeconomic status (SES; Hollingshead, 2011), maternal smoking during pregnancy (cigarettes/day), and postnatal lead exposure (ug/dl) obtained from a blood sample at 5 years. ANCOVA was used to examine whether volumetric group differences remained significant after controlling for confounders even weakly (p < 0.10) related to the outcome. Although marijuana (days/week) and cocaine (days/week) were considered, only 5 mothers in this sample reported marijuana use and 1 cocaine. Any observed betweengroup differences were, therefore, rerun excluding children born to these mothers to see whether the effects persisted without them.

The association between continuous measures of PAE (both at conception and across pregnancy) and structural volumes were examined using Pearson correlation. Hierarchical multiple regression analyses were used to control for confounders. The alcohol measure was entered in the first step of each analysis for each outcome. Total intracranial volume (TIV) was then added to the regressions for the regional volumes to determine if the exposure affected the regional volume over and above its effect on total brain volume. Control variables related to the outcome at less than p < 0.10 were entered in the third step to determine if the effect of the prenatal alcohol measure on structure volume continued to be significant after statistical adjustment for these potential confounders. Observed associations with extent of alcohol exposure were rerun excluding children born to mothers who used marijuana or cocaine during pregnancy to see whether effects remained without them.

Pearson correlations were used to examine whether alcoholrelated volumetric changes were associated with performance on the WISC-IV IQ and CVLT-C, two outcomes known to be related to PAE. Mediation of effects of PAE on these cognitive outcomes by structural volumes associated with PAE was examined using multiple regression. The effect of PAE on the cognitive outcome was entered in Step 1 of the regression; the structure volume, in Step 2. Mediation was inferred if the effect of PAE was significantly reduced when the structure volume was entered in Step 2, based on the Clogg test (Clogg et al., 1992). **Table 1** shows a list of abbreviations used.

## RESULTS

Following the exclusions described above, we report data for 71 children. Demographic characteristics are presented in **Table 2**. Although the FASD diagnostic groups did not differ by sex, prenatal exposure to cigarettes, marijuana, or cocaine, or childhood lead exposure, children in the HE group were slightly older than those in the FAS and PFAS groups. As expected,


children in both the FAS and PFAS groups had lower IQs than the HE children; those in the PFAS group also had lower scores than the controls. Total brain volumes were smaller in the children with FAS compared to all other groups. SES was lower for mothers of children with FAS or PFAS than mothers of children from the HE and control groups.

As expected, all of the alcohol measures for the control group were statistically lower compared to the alcohol-using groups (**Table 2**). In addition, children with FAS were more exposed than HE children on all measures. They were also exposed to more alcohol/drinking day at conception than those in the PFAS group. At conception and across pregnancy, AA/day and drinking days/week for the PFAS group were also significantly higher than for the HE group. Although number of days of alcohol use was reduced from 1.5 (at conception) to 1.1 (across pregnancy) day/week on average [t(68) = 3.88, p < 0.001), the number of drinks/occasion remained virtually unchanged, [t(69) = −0.33, p = 0.741]. Moreover, alcohol consumption at conception was highly correlated with alcohol use throughout pregnancy (r = 0.923). Due to the small sample size of the FAS group, the two syndromal groups, FAS and PFAS, were combined for subsequent volumetric analyses.

Volumes of traced regions are reported for each diagnostic group in **Table 3** and associations of control variables with structural volumes in **Table 4**. Children in the FAS/PFAS group had smaller right hippocampi compared to both the HE and control children, but this effect was no longer significant after adjustment for TIV. Volumes of four regions were smaller in girls than in boys. Among the remaining potential confounders, smoking during pregnancy and postnatal lead exposure were related to size in one region—the CC.

**Table 5** presents the results of the multiple regression analyses examining the relation of continuous measures of PAE to the structural volumes. PAE was associated with volume reductions bilaterally in the caudate and hippocampus and smaller CC, but the effect on the CC was no longer significant after controlling for smoking during pregnancy and postnatal lead exposure. When the regression for CC was rerun controlling only for lead, the effect of absolute alcohol/day remained significant (β = −0.281, p = 0.022). In a regression of the effect of alcohol and smoking on CC volume, the effects of both of these exposures fell just short of significance (alcohol β = −0.236, p = 0.065; smoking β = −0.217, p = 0.089), suggesting that neither effect was attributable to the effect of the other exposure.

Prenatal alcohol exposure (PAE) was associated with lower IQ (r = −0.402; p = 0.001) and CVLT-C (r = −0.248; p = 0.045) scores. Larger CC was associated with higher IQ for the sample as a whole and with better CVLT-C performance for the FAS/PFAS group (**Table 6**). Unexpectedly, smaller hippocampi were associated with better CVLT-C scores for the FAS/PFAS group.

The hypothesis that CC size would mediate the effect of PAE on IQ was tested using multiple regression. PAE was entered in Step 1. When CC size was added in Step 2 of the regression analysis, the effect of AA/day on IQ was reduced from −0.40 to −0.33, a reduction that was statistically significant, Clogg (t = −2.86, p < 0.01). These data thus indicate that the effect

#### TABLE 2 | Sample characteristics (N = 71).


For skewed data medians and interquartile ranges (IQR) were used. HE, heavily exposed non-syndromal; FAS, fetal alcohol syndrome; PFAS, partial FAS; WISC-IV, Wechsler Intelligence Scales for Children-Fourth Edition.

<sup>a</sup>HE > FAS, PFAS (both p < 0.05).

<sup>b</sup>FAS < HE (p = 0.011); PFAS < HE, Control (both p ≤ 0.01).

<sup>c</sup>FAS < PFAS (p = 0.018), HE (p = 0.003), Control (p = 0.038).

<sup>d</sup>Hollingshead, 2011; FAS < HE, Control (both p ≤ 0.05); PFAS < HE, Control (both p < 0.01).

<sup>e</sup>Control < FAS, PFAS, HE (all p < 0.001); HE < FAS, PFAS (both p < 0.001).

<sup>f</sup> Control < FAS, PFAS, HE (all p < 0.001); HE < FAS (p < 0.001); PFAS < FAS (p = 0.027).

<sup>g</sup>Control < FAS, PFAS, HE (all p ≤ 0.001); HE < FAS, PFAS (both p < 0.01).

<sup>h</sup>Control < FAS, PFAS, HE (all p < 0.001); HE < FAS, PFAS (both p ≤ 0.001).

<sup>i</sup>Control < FAS, PFAS, HE (all p < 0.001); HE < FAS (p = 0.036).

<sup>j</sup>Control < FAS, PFAS, HE (all p ≤ 0.001); HE < FAS, PFAS (both p < 0.01).

\*p ≤ 0.05, \*\*p ≤ 0.01, \*\*\*p ≤ 0.001.

TABLE 3 | Comparison of structure volumes by diagnostic group.


Values are mean(SD). Volumes are in mm<sup>3</sup> . NA, nucleus accumbens; CC, corpus callosum; HE, heavily exposed non-syndromal; FAS, fetal alcohol syndrome; PFAS partial (FAS). Bold print denotes significance at p ≤ 0.05. #FAS/PFAS<HE non-syndromal (p = 0.012) and Control (p = 0.052). p<sup>1</sup> after controlling for TIV. p<sup>2</sup> after controlling for TIV, as well as (<sup>a</sup> ) sex of child, (<sup>b</sup> ) cigarettes/day during pregnancy, (<sup>c</sup> ) lead concentration (ug/dl).


Bold print denotes significance at p ≤ 0.1. Values are Pearson r (p). NA, nucleus accumbens; CC, corpus callosum. Sex (male = 1, female = 2).

TABLE 5 | Association of alcohol consumption measures with structure volumes.


Bold print denotes significance at p ≤ 0.05.

β<sup>1</sup> is the standard regression coefficient adjusted for TIV.

β<sup>2</sup> is the standard regression coefficient controlling for TIV, as well as (<sup>a</sup> ) sex of child, (<sup>b</sup> ) cigarettes/day during pregnancy, (<sup>c</sup> ) lead concentration (ug/dl).

of PAE on IQ was partially mediated by the alcohol-related reduction in CC size (**Figure 2**).

## DISCUSSION

This study shows that increasing PAE, both in terms of quantity and frequency, is associated with disproportionate volume reductions bilaterally in the caudate nuclei and hippocampi and smaller CC. By contrast, volume differences between FASD diagnostic groups were evident only in the right hippocampus. Smaller CC was associated with poorer IQ. Within the FAS/PFAS group, smaller CC was also associated with lower CVLT-C scores, but hippocampal volume reductions were associated with better CVLT-C performance. The results from our mediation analysis suggest that the adverse effect of PAE on IQ is partially attributable to the reductions in CC size.


NA, nucleus accumbens; CC, corpus callosum; HE, heavily exposed non-syndromal; FAS, fetal alcohol syndrome; PFAS, partial (FAS). Bold print denotes significance at p ≤ 0.05. Values are Pearson r (p).

When the regional volumetric measures were compared across the three diagnostic groups, no significant differences were seen after adjustment for TIV. By contrast, the continuous measures of PAE were associated with volumetric differences in all of the regions examined after adjustment for brain size, except the bilateral NA. These findings are consistent with other neuroimaging studies performed with children from the same Cape Town cohort, which found that continuous measures of PAE were more sensitive than diagnosis to fetal alcohol-related alterations in brain structure and function (De Guio et al., 2014; Meintjes et al., 2014; du Plessis et al., 2015).

Our finding of alcohol-related volumetric reductions bilaterally in the caudate nuclei are consistent with studies reporting decreased basal ganglia volume in FAS (Mattson et al., 1994, 1996, 1999, 2001; Archibald et al., 2001; Roussotte et al., 2012). Animal studies suggest that the physiological mechanisms underlying smaller striatal volumes may be explained by a reduction in dendritic tree formation rather than a reduction in number of parvalbumin interneurons (De Giorgio et al., 2012). The caudate nucleus is involved in many of the neurobehavioural deficits exhibited by children prenatally exposed to alcohol, including executive function domains, such as cognitive flexibility, concept formation and reasoning, planning and response inhibition (Mattson et al., 1999, 2001). This region has also been shown to be activated less by children with FASD during behavioral inhibition tasks (Fryer et al., 2007) and to mediate cognitive control and verbal learning in children with PAE (Fryer et al., 2012). Although the latter study reported that larger caudate nuclei were associated with better performance in verbal learning in alcohol-exposed children, we did not confirm that finding in our study.

The caudate nuclei may be more susceptible to the effects of PAE than the NA. It has been suggested that some midline brain structures are more affected by alcohol exposure during embryological development than others (Sulik et al., 1981; Zhou et al., 2003; Meintjes et al., 2014). Our results are consistent with those of Archibald et al. (2001), who also failed to find alcoholrelated volume reductions in the NA using a similar tracing methodology. Because the NA can only truly be differentiated from the caudate nucleus histologically, the lack of association in both these studies may be due to poor sensitivity by these tracing methods to differentiate the NA from the caudate nuclei. Thus, although our results suggest that the NA may be relatively spared in PAE, further investigation using more sensitive methods, such as diffusion tensor imaging or higher resolution structural imaging, is needed.

Prenatal alcohol exposure (PAE) was also associated with volumetric reductions in both the left and right hippocampi– effects that remained significant after control for TIV. Reductions in pyramidal and granule cell number and density were observed in rats exposed prenatally to alcohol (Livy et al., 2003), which suggests that this may be the case in humans. The hippocampal formation has been shown to be particularly sensitive to PAE, showing a diverse range of negative effects and is one of the few areas of the brain which continues to exhibit these deficits as it develops into adulthood (see review, Gil-Mohapel et al., 2010). Associations of smaller hippocampi with better memory performance in typically-developing young adults (Van Petten et al., 2004) suggest a neural pruning process with age that may alter the relation of hippocampal volume in healthy controls as they grow older. Although our findings for the caudate nuclei, NA and CC are consistent with those of Archibald et al. (2001), who used a similar tracing protocol to ours, fetal alcohol-related differences in hippocampal volumes in their study did not survive after controlling for TIV. This disparity in results may be due to the wide range of ages included in that study (7–24 years) compared to the more limited range in ours (9–11 years). The observed association of smaller hippocampal volume with better learning and memory performance in the FAS/PFAS group in the current study was not expected at this age and is difficult to interpret.

AA/day and IQ by corpus callosum (CC) volume. The figure shows that the effect of prenatal alcohol exposure on IQ is partially mediated by the fetal alcohol-related corpus callosum volume reduction. When CC size was added in Step 2 of the regression analysis, the effect of AA/day on IQ was reduced from −0.40 to −0.33, a reduction that was statistically significant, Clogg (t = −2.86, p < 0.01).

Prenatal alcohol exposure (PAE) was associated with smaller CC after adjustment for TIV and postnatal lead exposure. However, regression analyses showed that the effects of smoking and alcohol on CC volume were confounded in this sample, and both fell slightly short of conventional levels of statistical significance when included in the same analysis. The effects of maternal smoking on the child's brain are not well understood, but decreases in CC size and changes in microstructure of other frontal WM structures have been linked with maternal smoking during pregnancy (Jacobsen et al., 2007; Jacobson et al., 2008; Paus et al., 2008). Given that the PAE and maternal smoking effects were similar in magnitude when adjusted for each other in the regression analysis, neither effect appears to be attributable to the effect of the other exposure, and a larger sample would likely have shown their effects to be independent.

The hypothesis that volumetric reductions in the caudate nuclei and hippocampi would mediate effects of PAE on cognitive performance was not supported. Volumetric reductions in the caudate nuclei did not relate to the cognitive outcomes; and, contrary to expectation, the hippocampal volumes in the FAS/PFAS group were inversely related to learning and memory performance.

In this study, larger CC area was associated with higher IQ scores and, amongst children in the FAS/PFAS group, with better learning and memory performance. Mediation analysis indicated that the effect of PAE on IQ was partially mediated by reductions in CC size. Our findings are consistent with several studies that have suggested that WM is more susceptible to the teratogenic effects of PAE than cortical GM (Archibald et al., 2001; Lebel et al., 2008; Spottiswoode et al., 2011). It is of interest that microstructural changes in two regions of the CC have also been found to mediate effects of PAE on IQ (Fan et al., 2016).

Morphology of the CC has been shown to have functional effects, with a thinner midline being associated with poorer motor activity and thicker than average midline being associated with poorer executive function (Bookstein et al., 2002). The CC is particularly important for functional outcomes requiring interhemispheric transfer of information, and it has been shown that even moderate exposure to alcohol may influence this process (Roebuck et al., 2002; Dodge et al., 2009). Since cognitive function assessed on IQ tests depends on a broad range of cognitive domains (Wechsler, 2003; Willoughby et al., 2008) utilizing a range of GM areas that communicate via axons in the CC, poorer interhemispheric transfer could contribute to lower IQ (Dodge et al., 2009). Smaller CC size may be due to either poorer myelination or fewer axons in the CC, resulting in slower signal processing or communication with greater numbers of GM cell bodies, respectively. CC volume reductions may also be due to changes to CC projection neurons, as reductions in size, number and location of these cells were observed in rats prenatally exposed to alcohol (Livy and Elberger, 2008). However, as no one animal model exhibits all the features of FASD (Cudd, 2005), further investigation into the underlying physiological mechanisms in humans is needed to better understand the links between CC and behavioral outcomes. As such, it is not surprising that the most heavily exposed children, those in the FAS/PFAS group, had the smallest CC areas and the lowest IQ scores.

One potential limitation of this study that is common to other longitudinal PAE studies is that it relies on the mother's report to assess alcohol consumption. However, we have previously validated the TLFB interview used in this study in relation to levels of fatty acid ethyl ester metabolites in meconium samples obtained from newborns in this cohort (Bearer et al., 2003). In addition, this TLFB interview has been shown to predict a broad range of behavioral (Jacobson et al., 2002, 2008; Lewis et al., 2015; Lindinger et al., 2016; Fortin et al., 2017) and neuroimaging outcomes (De Guio et al., 2014; Meintjes et al., 2014; du Plessis et al., 2015; Jacobson et al., 2017). A second limitation relates to the difficulty assessing effects of timing of exposure in a human study. Animal models have demonstrated that timing of exposure is a major determinant of specific longterm neurobehavioral outcomes, since different brain regions are susceptible to PAE during unique sensitive periods in their development (Livy et al., 2003; Valenzuela et al., 2012; Sadrian et al., 2014). Unfortunately, multicollinearity between maternal alcohol use at conception and later in pregnancy in this cohort, as in most human samples, was too high to discriminate effects attributable to timing of exposure. Polysubstance abuse is another common limitation in PAE studies. The current study assessed potential confounding by prenatal cocaine, marijuana and smoking exposure in order to address this problem; none of these exposures were related to the outcomes examined. Strengths of this study include the use of a sample size that was large relative to other manual tracing studies, limited age range, a population that is geographically and culturally homogenous albeit diverse in terms of genetic ancestry, and the availability of continuous measures of prospectively collected alcohol consumption.

In conclusion, this study, which used manual tracing, confirms previous reports linking PAE to decreases in subcortical volumes in the caudate nucleus and hippocampus after adjustment for TIV and also showed that PAE is associated with a disproportionate reduction in CC size. This is the first study to provide evidence that fetal alcohol-related reductions in CC partially mediate effects of PAE on IQ. This finding adds to the growing body of evidence suggesting that WM may be particularly vulnerable to the teratogenic effects of PAE on development of the fetal brain.

## ETHICS STATEMENT

Ethical approval for the study was acquired from the University of Cape Town Faculty of Health Sciences Human Research Ethics Committee and the Wayne State University Institutional Review Board. Informed written consent was obtained from the mothers and oral assent from the children in accordance with the Declaration of Helsinki.

## AUTHOR CONTRIBUTIONS

SB performed the tracing of the volumes under the supervision of CW and with the assistance of SR. She also reviewed the literature, performed the data analyses, contributed to the interpretation and wrote up the findings. EM provided

## REFERENCES


overall project supervision, collaborated on the design of the neuroimaging study, the data analysis, interpretation and write-up of the findings. SJ and JJ designed the study, supervised recruitment and maternal interviews and child assessments, collaborated on the data analysis, the interpretation of the findings and write-up of the paper. CM administered the maternal interviews, which included sociodemographic information and alcohol, smoking and drug ascertainment. NL and CL performed neuropsychological assessments.

## ACKNOWLEDGMENTS

This study was supported by NIH/NIAAA grants R01AA016781 and U01AA014790; South African Research Chairs Initiative of the Department of Science and Technology and National Research Foundation of South Africa; Medical Research Council of South Africa; and the Lycaki-Young Fund, State of Michigan. We thank H. Eugene Hoyme, M.D., Luther K. Robinson, M.D., and Nathaniel Khaole, M.D., who conducted the Cape Town dysmorphology examinations in 2005 in conjunction with the NIAAA Collaborative Initiative on FASD and Dr. Hoyme and the other dysmorphologists who participated in the 2009, 2013, and 2016 clinics. We thank R. Colin Carter, M.D./M.M.Sc. for his consultation, the CUBIC radiographers Marie-Louise de Villiers and Nailah Maroof, and our University of Cape Town and Wayne State University research staff Nicolette Hamman, Mariska Pienaar, Maggie September, Emma Makin, Renee Sun, and Neil Dodge. We also thank the parents and children for their long-term participation in and contribution to the study.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Biffen, Warton, Lindinger, Randall, Lewis, Molteno, Jacobson, Jacobson and Meintjes. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Altered Parietal Activation during Non-symbolic Number Comparison in Children with Prenatal Alcohol Exposure

Keri J. Woods1,2 \*, Sandra W. Jacobson2,3,4, Christopher D. Molteno<sup>4</sup> , Joseph L. Jacobson2,3,4 and Ernesta M. Meintjes1,2 \*

<sup>1</sup> Division of Biomedical Engineering, Department of Human Biology, University of Cape Town, Cape Town, South Africa, <sup>2</sup> Department of Human Biology, Faculty of Health Sciences, University of Cape Town, Cape Town, South Africa, <sup>3</sup> Department of Psychiatry and Behavioral Neurosciences, Wayne State University School of Medicine, Detroit, MI, United States, <sup>4</sup> Department of Psychiatry and Mental Health, Faculty of Health Sciences, University of Cape Town, Cape Town, South Africa

Number processing is a cognitive domain particularly sensitive to prenatal alcohol exposure, which relies on intact parietal functioning. Alcohol-related alterations in brain activation have been found in the parietal lobe during symbolic number processing. However, the effects of prenatal alcohol exposure on the neural correlates of nonsymbolic number comparison and the numerical distance effect have not been investigated. Using functional magnetic resonance imaging (fMRI), we examined differences in brain activation associated with prenatal alcohol exposure in five parietal regions involved in number processing during a non-symbolic number comparison task with varying degrees of difficulty. fMRI results are presented for 27 Cape Colored children (6 fetal alcohol syndome (FAS)/partial FAS, 5 heavily exposed (HE) non-sydromal, 16 controls; mean age ± SD = 11.7 ± 1.1 years). Fetal alcohol exposure was assessed by interviewing mothers using a timeline follow-back approach. Separate subject analyses were performed in each of five regions of interest, bilateral horizontal intraparietal sulci (IPS), bilateral posterior superior parietal lobules (PSPL), and left angular gyrus (left AG), using the general linear model with predictors for number comparison and difficulty level. Mean percent signal change for each predictor was extracted for each subject for each region to examine group differences and associations with continuous measures of alcohol exposure. Although groups did not differ in performance, controls activated the right PSPL more during non-symbolic number comparison than exposed children, but this was not significant after controlling for maternal smoking, and the right IPS more than children with fetal alcohol syndrome (FAS) or partial FAS. More heavily exposed children recruited the left AG to a greater extent as task difficulty increased, possibly to compensate, in part, for impairments in function in the PSPL and IPS. Notably, in non-syndromal heavily exposed children activation was impaired in the right PSPL, but spared in the right IPS. These results extend previous findings of poor right IPS recruitment during symbolic number processing in FAS/PFAS, indicating that mental representation of relative quantity is affected by prenatal alcohol exposure for both symbolic and non-symbolic representations of quantity.

Keywords: arithmetic deficits, fetal alcohol spectrum disorders, fetal alcohol syndrome, fMRI, non-symbolic number processing, parietal lobe, intraparietal sulci, numerical distance effect

#### Edited by:

Vivienne Ann Russell, University of Cape Town, South Africa

#### Reviewed by:

Sarah C. Treit, University of Alberta, Canada Jiaojian Wang, University of Electronic Science and Technology of China, China

> \*Correspondence: Keri J. Woods

keri.woods@gmail.com Ernesta M. Meintjes ernesta.meintjes@gmail.com

Received: 31 July 2017 Accepted: 08 December 2017 Published: 08 January 2018

#### Citation:

Woods KJ, Jacobson SW, Molteno CD, Jacobson JL and Meintjes EM (2018) Altered Parietal Activation during Non-symbolic Number Comparison in Children with Prenatal Alcohol Exposure. Front. Hum. Neurosci. 11:627. doi: 10.3389/fnhum.2017.00627

Prenatal alcohol exposure causes impairment in brain structure and function, leading to cognitive and behavioral deficits (Archibald et al., 2001; Sowell et al., 2001; Riley and McGee, 2005; Astley et al., 2009). The alcohol-related cognitive deficits include lower IQ (Streissguth et al., 1990; Jacobson et al., 2004), poor attention, and executive function (Kodituwakku et al., 1995; Coles et al., 1997; Mattson et al., 1999; Burden et al., 2005b), and slower cognitive processing speed (Streissguth et al., 1990; Jacobson et al., 1993, 1994; Coles et al., 2002). Among the cognitive deficits seen in relation to prenatal alcohol exposure, arithmetic has been found to be especially sensitive, and mathematical deficits persist, even after controlling for IQ (Streissguth et al., 1990, 1994a; Coles et al., 1991; Chiodo et al., 2004; Jacobson et al., 2004; Burden et al., 2005a). When academic achievement tests are administered to alcoholexposed individuals, arithmetic is consistently more impaired than reading or spelling (Streissguth et al., 1991; Goldschmidt et al., 1996; Kerns et al., 1997; Howell et al., 2006). Moreover, alcohol-related deficits in numerosity can already be detected in infancy (Jacobson et al., 2011b).

Fetal alcohol syndrome (FAS), the most severe of the fetal alcohol spectrum disorders (FASD), is characterized by distinctive craniofacial dysmorphology [short palpebral fissures, thin upper lip (vermillion), flat philtrum], small head circumference and pre- and/or postnatal growth retardation (Hoyme et al., 2005). In partial FAS (PFAS), some of the craniofacial dysmorphology is seen, as well as small head circumference, retarded growth, or neurobehavioral deficits. Heavily exposed (HE) individuals lacking the distinctive dysmorphology are diagnosed with alcohol related neurodevelopmental disorder (ARND) if they exhibit cognitive and/or behavioral impairment (Stratton et al., 1996; Hoyme et al., 2005).

Since the beginning of the 20th century, studies have shown that the parietal lobe is involved in number processing (Henschen, 1919), but more recently functional magnetic resonance imaging (fMRI) has provided a more extensive understanding of the neuroanatomy of this domain of processing. Based on brain lesion and neuroimaging findings, Dehaene (1992) and Dehaene and Cohen (1996) have proposed a triplecode model of number processing that incorporates three different systems of representation: the quantity system, the verbal system and the visual system. They have posited that the core quantity system—a non-verbal abstract representation of numerical quantity—is localized bilaterally in the anterior portion of the horizontal segment of the intraparietal sulci (IPS). This area is hypothesized to support number processing irrespective of the notation used, that is, whether represented symbolically as Arabic numbers or sequences of words or nonsymbolically by, for example, numbers of dots. Verbal processing of numbers is posited to be based in the left angular gyrus (left AG; close to the language areas of the brain), while the bilateral posterior superior parietal lobules (PSPL) are hypothesized to be involved in spatial and non-spatial attentional processes contributing to the visual processing of quantity.

Magnitude comparison can be studied by asking the subject to determine which of two numbers is larger or in a proximity judgment (PJ) paradigm ("Which of two numbers is closer to a third?"). Behavioral studies of magnitude comparison have shown that reaction time is slower for comparison of numbers that are closer together (e.g., 2 and 3) than numbers that are farther apart (e.g., 2 and 9), a phenomenon referred to as the "distance effect" (Moyer and Landauer, 1967). Consistent with this behavioral pattern, studies have shown that the IPS is more active when comparing numbers that are closer together (Pinel et al., 2001; Kaufmann et al., 2005; Ansari and Dhital, 2006; Mussolin et al., 2010a).

In the first fMRI study of number processing in FASD, adults with and without prenatal alcohol exposure were administered a subtraction task (Santhanam et al., 2009). Exposed individuals with alcohol-related dysmorphology exhibited poorer task performance and lower activation in parietal and frontal regions known to be associated with arithmetic processing, including the right inferior and left superior parietal regions and medial frontal gyrus, compared with controls. In the first fMRI study of children with FASD, the participants were administered tasks involving PJ and single-digit addition problems, which they performed as well as controls because the tasks had been simplified for administration in the scanner (Meintjes et al., 2010). Children with FAS and PFAS activated a markedly more diffuse parietal region extending into the precuneus and posterior cingulate, and for exact addition (EA), also into the postcentral gyrus. The FAS/PFAS group also exhibited significantly greater activation of the left AG than the control children in the PJ task. However, no significant differences were detected in the anterior portion of the IPS and other regions which have been linked by Dehaene et al. (2003) and associates to number processing, possibly due to lack of statistical power in the small sample on which this whole brain voxel-wise analysis was conducted. Using a region of interest (ROI) analysis, we recently examined brain activation patterns in the five parietal regions identified by Dehaene et al. (2003) as most critical for number processing in a sample consisting of the FAS/PFAS and control children from the Meintjes et al. (2010) study and a group of nonsyndromal HE children, which was assessed on both the PJ and EA tasks (Woods et al., 2015). During both tasks, higher levels of prenatal alcohol exposure were associated with weaker activation of the right IPS. Additionally, during PJ, children in the FAS/PFAS group activated the left AG more than control or HE children.

These and other studies examining the effects of prenatal alcohol exposure on neural correlates of number processing have, however, focused on symbolic representation of quantity (e.g., using Arabic numbers or written number words) (Streissguth et al., 1994a,b; Kopera-Frye et al., 1996; Burden et al., 2005a; Santhanam et al., 2009; Meintjes et al., 2010; Jacobson et al., 2011a; Woods et al., 2015). To date, no neuroimaging studies have examined non-symbolic number processing (e.g., dot patterns or collections of objects) in alcohol exposed children. Though symbolic representation of number appears to be localized in the IPS in adults (Dehaene et al., 2003), a meta-analysis of studies comparing number processing in children vs. adults suggests that

the location of parietal activations is more notation specific in children (Kaufmann et al., 2011). One aim of the current study is to examine the degree to which the effects of prenatal alcohol exposure on symbolic number processing are also seen in nonsymbolic number processing, which also depends heavily on the core quantity system.

Studies of other pediatric conditions associated with mathematical difficulties, such as developmental dyscalculia (DD; characterized by impairment in the processing of numerical and arithmetical information in individuals with normal intelligence), have shown a more pronounced behavioral distance effect in affected children than typically developing controls (Price et al., 2007; Mussolin et al., 2010b), and poorer math achievement has also been associated with a greater distance effect (Gullick et al., 2011). Similarly, higher levels of prenatal alcohol exposure were found to be associated with a more pronounced distance effect in a behavioral study using a Sternberg reaction time paradigm (Burden et al., 2005a). In DD, activations of the bilateral IPS fail to exhibit the increased response to differences in numerical distance seen in normal control children (Mussolin et al., 2010a).

In this study, we used fMRI to investigate the effect of prenatal alcohol exposure on the neural correlates of the numerical distance effect during a non-symbolic number comparison task. To our knowledge, the effect of prenatal alcohol exposure on the neural correlates of the distance effect and non-symbolic number comparison has not to date been investigated. Based on studies of DD (Price et al., 2007; Mussolin et al., 2010a) and our previous math studies (Meintjes et al., 2010; Woods et al., 2015), we hypothesized that prenatal alcohol exposure would be associated with weaker activation in the right IPS during non-symbolic number comparison, as well as weaker increases in activation in the right IPS arising from the distance effect. Based on our previous studies, we also expected to find increased compensatory activation in the left AG by children with FASD.

## MATERIALS AND METHODS

## Participants

Participants were right-handed children from the Cape Colored (mixed ancestry) community in Cape Town, South Africa. The Cape Colored community is composed primarily of descendants of white European settlers, Malaysian slaves, Khoi-San aboriginals, and black African ancestors. The incidence of FASD in this population is exceptionally high due to poor socioeconomic circumstances and historical practices of compensating farm labor with wine, which have contributed to a tradition of heavy recreational weekend binge drinking (May et al., 2007, 2013).

Data from 7 of 34 children recruited into the study were excluded from the analyses: technical error (1 control child), incomplete imaging data (2 controls), functional data due to failure to meet the performance criteria (1 FAS, 1 control), and excessive motion (1 FAS, 1 control). Mean and maximum displacements did not differ between exposure groups [t(25) = 1.49, p = 0.162 and t(25) = 1.55, p = 0.148, respectively].

After exclusions, there were 27 children with usable data (11 alcohol exposed and 16 controls). The children with usable scanner data did not differ from those excluded in terms of alcohol exposure or FASD diagnostic group (all ps > 0.20) or other demographic variables, including child age, sex, IQ, and blood lead concentration, and maternal education, age at delivery, parity, and smoking during pregnancy (all ps > 0.10).

13 of the 27 children were the older siblings of participants in our Cape Town Longitudinal Cohort (Jacobson et al., 2008). The others were identified by screening all of the 8- to 12-yearold children from an elementary school in a rural section of Cape Town, where there is a very high incidence of alcohol abuse among local farm workers (Meintjes et al., 2010; Jacobson et al., 2011c).

## Procedure

A research nurse and staff driver transported the mother and child from their home to the Cape Universities Brain Imaging Centre (CUBIC). All examiners were blind with regard to maternal alcohol history and the child's FASD diagnostic status, except in the most severe cases. Written informed consent was obtained from each mother and assent from each child. Approval for human research was obtained from the ethics committees at Wayne State University and the UCT Faculty of Health Sciences.

Each mother was interviewed in her primary language, Afrikaans or English, regarding her alcohol consumption and smoking during pregnancy. The alcohol interviews used a timeline follow-back approach (Sokol et al., 1985; Jacobson et al., 2002) to determine incidence and amount of drinking on a day-by-day basis during pregnancy. By contrast to most of our Cape Town studies, in which the mothers were interviewed prospectively during pregnancy, the interviews for this sample were conducted retrospectively. Any child whose mother reported consuming at least 14 standard drinks/week (1.0 oz absolute alcohol (AA)/day) on average or engaged in binge drinking during pregnancy (4 or more drinks/occasion) was considered heavily exposed. Controls were children whose mothers reported abstaining or drinking less than seven drinks/week and no binge drinking during pregnancy. Volumes recorded for each type of beverage consumed each day were converted to oz AA using multipliers proposed by Bowman et al. (1975) to provide three continuous measures of drinking during pregnancy: average oz AA/day, AA/drinking day (dose/occasion) and frequency (days/wk). Number of cigarettes smoked on a daily basis was also recorded; use of illicit drugs was obtained as days/month. Mothers were also interviewed regarding their education and occupational status and that of their spouse/partner to determine socioeconomic status (SES) on the Hollingshead (2011) scale. Mothers and children were given breakfast, lunch, and a snack at each laboratory visit. The mother received a small monetary compensation for each visit, as well as a photograph of her child, and the child was given a small gift.

Each child was examined for growth and FAS dysmorphology by two expert United States-based dysmorphologists during a clinic conducted in 2005 (Jacobson et al., 2011c) using a standard protocol based on the Revised Institute of Medicine criteria (Hoyme et al., 2005). The criteria for full FAS were

at least two of the principal dysmorphic features, small head circumference (bottom 10th percentile), and low weight or short stature (bottom 10th percentile); for partial FAS (PFAS), two features and small head circumference, low weight, or short stature. Alcohol-exposed children who did not meet criteria for either FAS or PFAS were designated non-syndromal HE.

There was substantial agreement among the examiners on the assessment of all dysmorphic features, including the threeprincipal fetal alcohol related features—philtrum and vermilion measured on the Astley and Clarren (2001) rating scales and palpebral fissure length (median r = 0.78). Two children who could not attend the 2005 clinic were examined by a Cape Town based FASD expert dysmorphologist (N. Khaole, MD, United States), whose diagnoses were subsequently confirmed by examinations conducted in follow-up clinics we held with the same dysmorphologists in 2009 and with HEH in 2013 and 2016.

## Neuropsychological Assessment

Each child was assessed on 7 of the 10 subtests from the Wechsler Intelligence Scale for Children, Third Edition (WISC-III)—Similarities, Arithmetic, Digit Span, Symbol Search, Coding, Block Design, and Picture Completion—and Matrix Reasoning from the WISC-IV. The IQ subtests were selected to represent the four dimensions of the WISC-III: Verbal Comprehension (Similarities), Perceptual Organization (Block Design, Picture Completion, Matrix Reasoning), Freedom from Distractibility (Digit Span, Arithmetic), and Processing Speed (Coding, Symbol Search). Similarities was the only subtest administered in the verbal domain to limit use of tests that may be more dependent on the cross-cultural context. IQ was estimated from these subtests using Sattler's (1992) formula for computing Short Form IQ; validity coefficients for Short Form IQ based on 5 or more subtests consistently exceed r = 0.90. Handedness was assessed on the Annett (1970a,b) Behavioral Handedness Inventory.

# Neuroimaging Assessment

#### Magnetic Resonance Imaging Protocol

All scans were acquired using a 3T Allegra MRI scanner (Siemens Medical Systems, Erlangen, Germany). High-resolution anatomical images were acquired in sagittal orientation using a magnetization prepared rapid gradient echo (MPRAGE) sequence (TR 2300 ms, TE 3.93 ms, TI 1100 ms, 160 slices, flip angle 12◦ , 1.3 mm × 1.0 mm × 1.0 mm, 6.03 min). During the fMRI protocol, 126 functional volumes sensitive to blood oxygen level dependent contrast were acquired with a T2<sup>∗</sup> -weighted gradient echo, echo planar imaging sequence (TR = 2000 ms, TE = 30 ms, 34 interleaved slices, 3 mm thick, gap 0.9 mm, 200 mm field of view, resolution 3.125 mm × 3.125 mm × 3 mm, 4.2 min). Total scan time was less than 1 hour since these tasks formed part of a longer protocol.

## Functional MRI Experimental Task

For this study, we designed an fMRI non-symbolic number comparison task, which we refer to here as the "smarties" test. For this test, the child is presented with a screen split vertically in half, each half containing different numbers of smiley faces, and is asked to press the button on the side with the most smiley faces. This non-symbolic number comparison task was administered using a fixed-paced block design and had nine 16-s task blocks, each with eight problems, interleaved with 10-s rest blocks. Each task block comprised problems at 1 of 3 levels of difficulty, defined in terms of the ratio of number of faces on one side of the screen to the other (1:2, 2:3, 3:4) (see **Figure 1**). Stimuli were shown for only 1 s to prevent counting. During rest blocks, the child looked at a fixation square.

Each child practiced this task initially in a mock scanner to reduce anxiety, thereby facilitating completion of the MRI scans. The experimental task was programmed using E-Prime software (Psychology Software Tools, Inc., Pittsburgh, PA, United States) and was presented on a rear projection screen using a data projector located in a room behind the scanner through a waveguide in-line with the bore of the magnet. The child held a Lumitouch response system (Photon Control Inc., Burnaby, CO, Canada) in his/her right hand and responded using the right index and middle finger. The child was able to talk to the examiner using an intercom built into the scanner and could stop the scan at any time by squeezing a ball held in his/her left hand.

## Behavioral Performance

Behavioral responses were recorded on a computer; number of problems correct and reaction time for correct responses were tabulated. Blocks with fewer than 5 correct responses were considered bad blocks. If 5 or more of a subject's nine active blocks were bad, their data were excluded (1 exposed, 1 control). All children met the performance criterion of at least one usable block at each difficulty level.

## fMRI Analysis

All fMRI analyses were performed using Brain Voyager QX (Brain Innovation, Maastricht, Netherlands). Four dummy images were acquired in each run that were excluded from all analyses. Images were motion corrected relative to its first volume with trilinear/sinc interpolation. Images were corrected for different slice acquisition times and linear trends and temporally smoothed with a high pass filter of 2 cycles/point. For each child, data from the largest section with no movement greater than 3 mm displacement or 3.0◦ rotation were analyzed. Children were excluded from further analyses if the section of usable data did not include at least one block from each condition, resulting in two additional children being excluded (1 exposed, 1 control). Each child's functional data were co-registered to his/her high-resolution anatomical MRI, rotated into the AC-PC plane and normalized to Talairach space using a linear transform calculated on the anatomical images. The 3.125 mm × 3.125 mm × 3 mm fMRI voxels were interpolated during Talairach normalization to 3 mm × 3 mm × 3 mm.

A priori regions of interest (ROIs) were defined for each of the five parietal regions identified in Dehaene et al.'s (2003) metaanalysis; namely, bilateral anterior horizontal IPS, bilateral PSPL

and left AG. Each ROI consisted of a sphere, radius 6 mm, centered on the coordinates derived from the meta-analysis. These regions are illustrated in **Figure 2**.

To create a parametric model, difficulty levels were weighted by the ratio of the number of faces on the two sides of the stimuli within each block. We used two predictors of interest. The first predictor ("main") gives the signal variation explained by non-symbolic number processing across all task difficulties. The other predictor ("parametric") gives the parametric increase in activation as the difficulty level increases and is a measure of the strength of the distance effect. Separate subject analyses were performed on the average signal in each ROI using the general linear model with predictors of interest convolved by the standard hemodynamic function. The six motion correction parameters were z-transformed and then added as predictors of no interest. For each predictor of interest, beta maps estimating mean percent signal change were created for each subject for each ROI.

#### Statistical Analyses

All variables were examined for normality of distribution. AA/day was skewed (skewness > 3) and was log transformed (log X + 1). The following variables with outliers greater than 3 SDs beyond the mean were transformed by recoding all outlying values to one point beyond the next most extreme observed value: maternal education (n = 1), AA/occasion (n = 1), and proportional drinking days (n = 1).

Seven control variables were assessed for consideration as potential confounders of the relation of prenatal alcohol exposure to number processing: four demographic characteristics (parity, SES, mother's age at delivery, and years of education), two child characteristics (child sex and age at assessment), and two exposure variables (number of cigarettes smoked per day during pregnancy and level of postnatal lead exposure). Lead exposure, which was based on a venous blood sample obtained from the child, was included because lead levels in this population are within the range in which subtle but meaningful effects on cognitive function have consistently been reported (e.g., Lanphear et al., 2000; Chiodo et al., 2004). Each control variable that was even weakly related to a given outcome measure (at p < 0.10) was considered a potential confounder of the effect of alcohol exposure on the outcome in question.

T-tests were used to examine differences between diagnostic groups (exposed; control) on the behavioral and neuroimaging outcome measures. Analysis of covariance (ANCOVA) was used to test whether differences remained significant after controlling for potential confounders. Differences in performance between difficulty levels were examined using a repeated measures ANOVA with Greenhouse–Geisser correction. The relation of the three continuous measures of prenatal alcohol exposure— AA/day, AA/drinking day and proportional drinking days—to each of the outcome measures was examined using Pearson correlation. Multiple regression analyses were then run relating each of the continuous exposure measures and potential confounders to each of the outcomes.

Association of percent signal change in each of the ROIs to behavioral performance was examined using Pearson correlation.

## RESULTS

## Sample Characteristics

fnhum-11-00627 January 4, 2018 Time: 18:16 # 6

The sample characteristics are summarized in **Table 1**. The mothers of the alcohol exposed children reported an average of 13.4 standard drinks of alcohol per drinking occasion during pregnancy on 2–3 days per week. By contrast, all but two women in the control group abstained from drinking during pregnancy; the one control mother drank two drinks once during pregnancy and the other control mother drank two drinks about once a month. The groups were generally similar in terms of the other background characteristics, except that mothers of the exposed children smoked more during pregnancy than mothers of controls. None of the mothers in this sample reported using drugs during pregnancy. Of the 11 alcohol-exposed children, 6 met criteria for either FAS or PFAS, while 5 were designated as non-syndromal HE. As expected, the IQ scores of the exposed children were lower than those of the controls. The low IQ scores of both groups reflect the highly disadvantaged circumstances and poor quality of education available in this community.

## Neuropsychological Assessments

Behavioral data are summarized in **Table 2**. Overall accuracy and reaction time did not differ between exposed and control children (all ps > 0.30), nor did accuracy and reaction time at each difficulty level (all ps > 0.25). None of the behavioral measures were related to the continuous measures of prenatal alcohol exposure, although the relation between reaction time in the easy level of the task and AA/occasion fell just short of significance (r = 0.34, p < 0.010), with greater alcohol exposure associated with slower reaction time, as expected.

TABLE 1 | Sample characteristics.


Means (SD); †p < 0.10, <sup>∗</sup>p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; <sup>a</sup>Data only available for 25 subjects.

Performance accuracy was high in both groups, with scores over 80% for all but the most difficult level. A repeated measures ANOVA with Greenhouse–Geisser correction showed that mean accuracy differed significantly between levels [F(1.9,48.7) = 7.60, p = 0.002], with greater accuracy in the easy and intermediate levels than in the difficult level (p = 0.020 and p = 0.005, respectively). Mean reaction times also differed between difficulty levels [F(1.9,49.6) = 9.89, p < 0.0005], with reaction times in the easy condition being faster than in the intermediate (p < 0.013) and difficult (p < 0.001) conditions.

## Neuroimaging Assessments

**Table 3** shows mean % signal change during non-symbolic number comparison in the Dehaene parietal ROIs for children in each of the two groups. The only significant group difference was in the right PSPL, where the control group showed greater activation than the exposed group, in which activations were actually reduced lower than baseline, on average, however, this effect was no longer significant after adjustment for smoking during pregnancy. By contrast, the left AG showed a stronger distance effect in the exposed children than in the controls (see **Table 4**).

Since we did not observe a difference in activation during number comparison between exposed and control children in the right IPS as hypothesized, we compared activation levels in this region in the FAS/PFAS group only to the controls. As predicted, the control children activated the right IPS more than the FAS/PFAS group [t(20) = 2.14, p = 0.045, means ± SD = –0.15 ± 0.12 and 0.11 ± 0.28 for the FAS/PFAS and control groups, respectively]. When the FAS/PFAS, non-syndromal HE, and control children were compared, activation patterns in the right PSPL were similar for the FAS/PFAS and HE groups [t(9) = 1.21, p = 0.257], whereas activation patterns in the right IPS of HE children were more similar to those of controls [t(19) = 0.021, p = 0.986].

Higher levels of prenatal alcohol exposure on all three continuous measures were related to greater distance effects in the left AG (rs = 0.46, 0.41, and 0.45 for AA/day, AA/occasion


Values are Means (SD).



F<sup>1</sup> = before adjustment for potential confounders; F<sup>2</sup> = after adjustment for potential confounders; <sup>∗</sup>p < 0.05; d.f. = (1,25); <sup>a</sup>adjusted for smoking during pregnancy; <sup>b</sup>adjusted for maternal education; <sup>c</sup>adjusted for child's sex; <sup>d</sup>adjusted for child's age.



F<sup>1</sup> = before adjustment for potential confounders; F<sup>2</sup> = after adjustment for potential confounders; <sup>∗</sup>p < 0.05; d.f. = (1, 25); <sup>a</sup>controlled for maternal education; <sup>b</sup>controlled for smoking during pregnancy.

and frequency of drinking, respectively, all ps < 0.05; **Figure 3**). More drinking days/week was related to reduced activation during number comparison in the right PSPL (r = –0.41, p = 0.036; **Figure 4**).

Children who showed a greater distance effect in the left PSPL (i.e., activated the left PSPL more with increasing task difficulty) had shorter reaction times (r = –0.43, p = 0.03) and those who showed greater activation of this region had better accuracy (r = 0.46, p = 0.02).

## DISCUSSION

This study examined the relation of FASD diagnosis and continuous measures of prenatal alcohol exposure to activation of five parietal regions, which have been identified by Dehaene et al. (2003) as most critical for number processing, during a non-symbolic number comparison task with varying degrees of difficulty. Despite similar behavioral performance, prenatal alcohol exposure was associated with altered patterns of brain activation. Control children showed greater activation of the right PSPL during non-symbolic number comparison than exposed children; however, this effect was no longer significant after adjusting for smoking. Control children also showed greater activation of the right IPS than children with a diagnosis of FAS or PFAS. Both greater activation of the left PSPL and a greater distance effect on the activation of this region were associated with better task performance. In the right PSPL, the activation patterns of HE children were similar to those of the FAS/PFAS group, while in the right IPS their activation patterns were similar to that of the controls. With respect to the distance effect, the exposed children showed greater activation with increasing levels of task difficulty in the left AG, compared to the controls.

## Effects on the IPS

As predicted, prenatal alcohol exposure was associated with weaker activation of the right IPS during non-symbolic number comparison, although the effect was seen only in the syndromal children in this sample. This finding of an adverse effect of prenatal alcohol exposure on activation of the right IPS during non-symbolic number processing is consistent with previous studies of symbolic number processing (Santhanam et al., 2009; Meintjes et al., 2010; Woods et al., 2015). During a subtraction task (Santhanam et al., 2009), adults with alcohol related dysmorphology showed weaker activation of the right inferior

parietal lobe (just slightly superior and posterior to our right IPS ROI) than unexposed controls. This effect was not seen in the non-dysmorphic prenatal alcohol exposure group. In a whole brain analysis (Meintjes et al., 2010), we found that control children activated the right IPS more than children with FAS/PFAS during a PJ task, though this region was centered more posteriorly than the right IPS ROI used in this study. There were no non-dysmorphic HE children included in that study. In our previous study of symbolic number processing using the same a priori ROIs as in the present study, we found that increasing alcohol exposure was related to weaker activation in the right IPS during both PJ and simple addition (Woods et al., 2015). These results, taken together with the findings from the current study, indicate that prenatal alcohol exposure impairs the activation of the right IPS during both symbolic and non-symbolic number processing.

The bilateral IPS have frequently been linked to non-verbal representation of quantity (Dehaene et al., 2003). This region is activated when numbers are attended to, even without any explicit number processing task requirements (Eger et al., 2003), during number comparison (Chochon et al., 1999), and during mental arithmetic (Burbaud et al., 1999; Chochon et al., 1999; Pesenti et al., 2000). The region is independent of modality of number and is activated whether number is presented as Arabic numbers, written number words, spoken numbers or sets of dots (Le Clec'H et al., 2000; Pinel et al., 2001; Piazza et al., 2002). It is more active when manipulating large numbers (Kiefer and Dehaene, 1997; Stanescu-Cosson et al., 2000), performing arithmetic with three rather than two operands (Rivera et al., 2002), approximating addition rather than computing the exact sum (Dehaene et al., 1999), and performing subtraction rather than multiplication (Chochon et al., 1999; Lee, 2000). This region has also been shown to exhibit a distance effect, meaning that it is activated more when comparing numbers that are closer together (Dehaene and Cohen, 1996; Pinel et al., 2001). Although we did not observe a behavioral difference in this fMRI study, evidence from our Detroit and Cape Town behavioral studies suggests a specific effect of prenatal alcohol on magnitude comparison (Burden et al., 2005a; Jacobson et al., 2011a,b). The poorer recruitment of the IPS in children with FAS or PFAS observed here provides further evidence of a fetal alcohol exposure related change in mental representation and manipulation of quantity.

The non-dysmorphic HE children did not show poorer recruitment on the IPS than control children. Given that some of the non-dysmorphic children were very heavily exposed, perhaps the effect of prenatal alcohol on the right IPS relates more to the timing of drinking during pregnancy than on the quantity the mother drank.

White matter microstructure directly underlying the left IPS has been shown to be related to mathematical abilities in children with prenatal alcohol exposure. Higher math scores were associated with greater fractional anisotropy in this region (Lebel et al., 2010). These findings, combined with our results and the results of previous fMRI studies, suggest that both gray and white matter of the IPS are involved in mathematical processing and that both are affected by prenatal alcohol exposure.

Based on the findings from developmental dyscalculia (Price et al., 2007; Mussolin et al., 2010a), we expected to find a weaker distance effect in the right IPS in children with prenatal alcohol exposure; however, this was not the case. Although the control children activated the right IPS more than the FAS/PFAS group during non-symbolic number comparison averaged across the three difficulty levels, there was no difference in the distance effect between the groups in this region. The lack of a group difference in the distance effect in this region may be related to the fact that the distance effect was not clearly evident in the right IPS in the control children, possibly because they were too young and their brains still too immature. In a symbolic number comparison task, adults showed a distance effect in parietal regions, including the intraparietal sulcus (Kaufmann et al., 2005), while in an identical paradigm, 8- to 12-year-old children showed no significant distance effect in any parietal region, even though they showed a behavioral distance effect (Kaufmann et al., 2006). An alternative reason may relate to task design. Our "smiley face" stimuli remained the same size throughout trials, so it is possible that children used non-numerical cues, such as density of the faces, or the total area covered by the faces to select the correct answer. Other studies have varied the sizes of their stimuli, so that the participants could not use these non-numerical cues to select the correct answer (Ansari et al., 2006; Holloway and Ansari, 2009; Gullick et al., 2011; Kucian et al., 2011; Lonnemann et al., 2011). A third possibility is that the task did not increase enough in difficulty for the additional intraparietal neuronal resources to be required. However, this is unlikely, as the behavioral results did show distance effects.

## Effects on the Left AG

Consistent with our hypothesis, exposed children showed compensatory activation of the left AG, demonstrating a greater

distance effect in that region than typically developing controls. This finding was also evidenced by an association between level of prenatal alcohol exposure and the distance effect in the left AG. These results suggest that alcohol exposed children need to recruit the left AG to a greater extent as the task difficulty increases, possibly to compensate for deficits in quantity representation in the IPS. In contrast, controls showed an inverse distance effect in this region, that is, reduced left AG activation during the more difficult conditions, presumably due to better functioning of regions specialized for quantity representation. This area has also been implicated in our previous studies of number processing in FASD. In a whole brain voxelwise analysis, we found a significant group difference in a nearly identical region of the left AG [–42,–65,36] with greater activation in the FAS/PFAS group than the controls (Meintjes et al., 2010), and in an ROI study, we found that during PJ children with FAS or PFAS demonstrated greater activation of the left AG ROI than HE or control children (Woods et al., 2015).

The AG is adjacent to the perisylvian language processing network and is associated with the verbal processing of numbers (Dehaene et al., 2003). It is more highly activated during addition and multiplication than during subtraction, presumably because addition and multiplication facts are more likely to be retrieved from long-term memory (Stanescu-Cosson et al., 2000; Simon et al., 2002; Delazer et al., 2005). It is also more active during symbolic number processing than non-symbolic number processing (Holloway et al., 2010; Price and Ansari, 2011). Although the left AG is usually associated with verbal strategies for solving number processing problems, that is not likely to be the case here, as these non-symbolic number comparison problems do not involve the recall of arithmetic facts (Menon et al., 2000), verbal manipulations of number (Dehaene et al., 2003), or the mapping from symbols to numerical magnitudes (Grabner et al., 2007; Ansari, 2008). However, the left AG is also involved in visuospatial attention (Nobre et al., 1997; Corbetta and Shulman, 1998; Gitelman et al., 1999; Corbetta et al., 2000; Rushworth M.F.S. et al., 2001; Rushworth M.F. et al., 2001) and has been shown to be involved in mentally maintaining a spatial representation of numbers similar to a mental number line (Göbel et al., 2001). Based on our data, it appears that the exposed children rely on this spatial representation increasingly as the difficulty of the problems increases, instead of relying on quantity processing mediated by the right IPS.

## Effects on the PSPL

Exposed children activated the right PSPL less than control children, and more frequent drinking was associated with reduced activation. Although this effect was no longer significant after adjustment for maternal smoking during pregnancy, smoking was confounded with prenatal alcohol exposure in this sample (r = 0.44, p = 0.010), making it difficult to tease out the degree to which each of these exposures was implicated in the observed effect. Prenatal alcohol exposure has not been shown previously to affect the activation of the right PSPL during number processing, but in our previous study we found that greater prenatal alcohol exposure was related to less activation of the left PSPL during EA (Woods et al., 2015). Similarly, a study of subtraction (Santhanam et al., 2009) found that a similar region in the left hemisphere was activated more by controls than by exposed young adults with alcohol related dysmorphology. The PSPL, which is activated during counting (Piazza et al., 2002) and a variety of visual–spatial tasks, is believed to support the engagement of attention during visual processing of numbers (Pinel et al., 2001; Dehaene et al., 2003). These findings suggest that at this age alcohol exposed children seem to be less able to recruit the attentional systems associated with number processing.

Interestingly, activation patterns in the right PSPL for the nonsyndromal HE children were similar to those in the FAS/PFAS group, while those in the right IPS were similar to controls. These data suggest that, although the activation of right PSPL appears to be impaired in the non-syndromal HE children, the functioning of the right IPS is apparently spared. Other studies have also found more extensive neural impairment in the FAS/PFAS group than in the HE. For example, both functional and structural connectivity have been found to be lower in heavily alcohol exposed children in only a subset of the regions affected in children with FAS/PFAS (Fan et al., 2016, 2017).

Greater activation of the left PSPL, as well as a greater distance effect on the left PSPL, were both associated with quicker and more accurate behavioral performance. Dehaene et al. (2003) have emphasized that the PSPL can be engaged when attending to specific quantities on the number line. It is possible that the children who are better able to recruit the left PSPL (and recruit it more with increasing task difficulty) are better able to position each array of stimuli on the mental number line and, therefore, to make magnitude comparisons more quickly and accurately.

## Comparisons of Effects on Symbolic and Non-symbolic Number Processing

Because our previous study (Woods et al., 2015) investigating the effect of prenatal alcohol exposure on symbolic number processing in children used the same a priori ROIs as in this study, the results of these two studies can be examined to compare the effect of prenatal alcohol exposure on symbolic and non-symbolic number processing. The tasks investigated in the previous study of 49 children (8–12 years of age) were simple EA and a magnitude comparison task, PJ.

In both studies, although the activation of the right IPS was impaired by prenatal alcohol exposure, we found no relation of alcohol or FASD diagnostic group to activation of the left IPS. It is not clear whether the right IPS is more affected by prenatal alcohol exposure than the left IPS or whether they are equally affected, but only the effect on the right IPS is apparent because the left IPS is recruited to a lesser extent by these tasks. In the Dehaene meta-analysis (Dehaene et al., 2003) used to identify the critical parietal number processing regions used in our studies, the IPS activation was bilateral in all but two studies. To examine whether the absence of an alcohol effect on the left may be due to differences in degree of activation on the left and right, we compared the magnitudes of the activation in the left and right IPS in control children during symbolic and nonsymbolic number processing. In control children, neither the left

nor the right IPS showed significant activation increases during non-symbolic number processing. In the PJ task, activation increases were seen bilaterally in the IPS (right IPS mean % signal change ± SD = 0.09 ± 0.12, p = 0.007; and left IPS mean % signal change ± SD = 0.07 ± 0.13, p = 0.045), but during EA activation increases were seen only on the right (right IPS mean % signal change ± SD = 0.07 ± 0.08, p = 0.003). It is, thus, not clear whether the left IPS is more sensitive to prenatal alcohol exposure or merely less active in the tasks examined to date.

While during the Smarties and PJ tasks, exposed children showed increased activation of the left AG, no compensatory activation was seen during EA (see also Meintjes et al., 2010). These data suggest that exposed children show compensatory activation of the left AG during magnitude comparison tasks, both symbolic and non-symbolic, but not during arithmetic tasks relying on verbal recall. Whereas for magnitude comparison tasks, the control children appear to use the optimal strategy, relying on the core quantity system in the IPS, the alcohol exposed children recruit the left AG to a greater extent, possibly due to alcohol-related damage to the IPS. In contrast, for arithmetic tasks, verbal recall of number facts (relying on the left AG) is the most efficient strategy and appears to be used by both the exposed and control children.

While prenatal alcohol exposure was associated with lower activation of the right PSPL during the Smarties task, no association with alcohol exposure was found in this region during either PJ or EA (see also Santhanam et al., 2009; Meintjes et al., 2010). In contrast, alcohol-related reductions were seen in activation in the left PSPL during EA (Woods et al., 2015) and in a comparable region during subtraction (Santhanam et al., 2009). Thus, it appears that both the left and right PSPL are impaired by prenatal alcohol exposure, but the impairment of the right PSPL is evident only during non-symbolic magnitude comparison, whereas impairment of the left PSPL has been seen during symbolic arithmetic operations. Neither of the studies using the PJ task to assess symbolic magnitude comparison found alcohol-related impairment in the left nor the right PSPL.

Although all but one of the heavily alcohol exposed children showed relatively low levels of activation in the right PSPL during the Smarties task, there was a notably wide range of signal change among the controls (see **Figure 4**). We examined numerous sample characteristics (child IQ, age at scan, grade in school, ADHD status, maternal education and SES, smoking and other prenatal drug exposures, and postnatal lead exposure) in relation to these activation levels in the control group. One of the two controls with the lowest % signal change had a somewhat elevated prenatal exposure to smoking (seven cigarettes/day). There were no other differences that would help explain the wide range in responses, suggesting that these activations represent a "normal" range for the controls. What is striking in our findings is that the activation levels in this brain region are low for virtually all of the alcohol-exposed children.

## Limitations

One limitation of this study was that, by contrast to most of the studies our group has conducted in Cape Town, the maternal report of drinking during pregnancy for this cohort was obtained retrospectively several years after the child's birth. Nevertheless, the validity of these reports is supported by the fact that they were predictive of neuroimaging and neurobehavioral outcomes (Meintjes et al., 2010, 2014; Jacobson et al., 2011b,c; De Guio et al., 2014; du Plessis et al., 2014; Lewis et al., 2015; Lindinger et al., 2016). Predictive validity of the maternal pregnancy drinking reports for childhood IQ in this sample was substantial (rs = –0.54 and –0.53, for AA/day and AA/occasion, respectively, both ps < 0.001). However, drinking during pregnancy is more difficult to recall reliably in a retrospective interview than smoking, which is a more addictive, more consistent, and less of an episodic practice, as well as packaged so that quantity is more readily recalled (Jacobson and Jacobson, 1992). These differences in ability to recall quantity between alcohol exposure and smoking make it particularly difficult to tease out the degree to which each of these exposures is responsible for the outcomes, such as activation of the right PSPL, that are related to both exposures in this sample. A second limitation was the small size of the FAS/PFAS (n = 6) and HE (n = 5) groups. In addition, because all of the children come from socioeconomically and educationally disadvantaged environments, we cannot determine the degree to which the results would hold for children from an educationally less deprived background. We did not control for multiple comparisons, due to the fact that we examined only five regions rather than the whole brain; ROI analyses increase SNR by averaging across the voxels in a region. Although the use of an ROI approach was appropriate for this study, a whole brain voxelwise analysis could have revealed additional activation differences between groups.

## CONCLUSIONS

This study found poorer recruitment of the right IPS during non-symbolic number comparison in syndromal children with FAS/PFAS compared with controls, extending our previous finding of poorer right IPS recruitment during symbolic number processing (Woods et al., 2015). To our knowledge, this is the first to show that heavy prenatal alcohol exposure impairs mental representation and manipulation of quantity for nonsymbolic, as well as symbolic, representations. As hypothesized, this impairment appeared to be compensated for by increased activation in the left AG, with only the exposed children recruiting the left AG to a greater extent as task difficulty increased. The inverse relation between prenatal alcohol exposure and activation of the right PSPL in children with prenatal alcohol exposure suggests that alcohol impairs the ability of exposed children to employ the attentional systems required for optimal number processing. Notably, the non-syndromal HE children's activation was impaired in the right PSPL, which mediates attention during number processing, but spared in the right IPS, which mediates quantity comparison.

## ETHICS STATEMENT

Written informed consent was obtained from each mother and assent from each child. This study was conducted according to the ethical guidelines and principles of the international Declaration of Helsinki, as well as South African research ethics guidelines, as documented by the Department of Health in "Ethics in Health Research: Principles, Processes and Structures", Second edition, 2015. Approval for human research was obtained from the Wayne State University Institutional Review Board and the UCT Faculty of Health Sciences Ethics Committee.

## AUTHOR CONTRIBUTIONS

fnhum-11-00627 January 4, 2018 Time: 18:16 # 11

KW performed the neurobehavioral and neuroimaging data analyses, reviewed the literature, interpreted the findings, and wrote up the paper. EM provided overall project supervision, collaborated on the design of the neuroimaging study, oversaw the neuroimaging assessments, and reviewed the paper. SJ and JJ designed the original study and the neuroimaging task, supervised recruitment and maternal and child assessments, provided suggestions regarding data analysis and interpretation, and reviewed the paper. CM administered the maternal interviews, which included sociodemographic information and alcohol, smoking and drug ascertainment.

## FUNDING

This research was funded by a Fogarty International Research Collaboration Award from the National Institutes of Health (NIH) (R03 TW007030), a Focus Area grant (FA2005040800024)

## REFERENCES


from the National Research Foundation of South Africa, the South African Research Chairs Initiative of the Department of Science and Technology and National Research Foundation of South Africa, the Medical Research Council of South Africa, a Children's Bridge grant from the Office of the President of Wayne State University (WSU), and a grant from the Lycaki-Young Fund from the State of Michigan. Data analysis was funded, in part, by a grant from the NIH/National Institute on Alcohol Abuse and Alcoholism (NIAAA) R01AA016781. The dysmorphology clinic assessments were also funded in part by NIAAA grants U01AA014790, U24AA014815, and U01AA014809 in conjunction with the Collaborative Initiative on Fetal Alcohol Spectrum Disorders.

## ACKNOWLEDGMENTS

We thank H. E. Hoyme and L. K. Robinson, who performed the dysmorphology examinations of the children in the 2005 and 2009 clinics, and H. E. Hoyme and the team of dysmorphologists, G. De Jong, M.D., P. Shah, M.D., H. Bezuidenhout, M.D., E. Krzesinski, M.D., as well as R. Colin Carter, M.D., who examined the children in 2013 and 2016. We thank M. September, N. Dodge, and our UCT and WSU research staff, for their work on this project, and the staff at the Cape Universities Brain Imaging Centre for their contributions to the acquisition of the neuroimaging data. We also thank the mothers and children who have participated in our Cape Town research study.

adjustment in a sample of problem drinkers. J. Stud. Alcohol Drugs 36, 1154–1172. doi: 10.15288/jsa.1975.36.1154



prenatally. Alcohol. Clin. Exp. Res. 19, 1558–1564. doi: 10.1111/j.1530-0277. 1995.tb01024.x


word attack and arithmetic scores on standardized tests: Adolescent data from a population-based prospective study. Alcohol. Clin. Exp. Res. 18, 248–254.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The handling editor VAR declared a shared affiliation, though no other collaboration, with the authors.

Copyright © 2018 Woods, Jacobson, Molteno, Jacobson and Meintjes. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Changes in the Cholinergic, Catecholaminergic, Orexinergic and Serotonergic Structures Forming Part of the Sleep Systems of Adult Mice Exposed to Intrauterine Alcohol

Oladiran I. Olateju , Adhil Bhagwandin , Amadi O. Ihunwo and Paul R. Manger\*

School of Anatomical Sciences, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa

We examined the effect of chronic prenatal alcohol exposure on certain neuronal systems involved with the sleep-wake cycle of C57BL/6J mice exposed to prenatal alcohol once they had reached 56 days post-natal. Pregnant mice were exposed to alcohol, through oral gavage, on gestational days 7–16, with recorded blood alcohol concentration (BAC)s averaging 1.84 mg/ml (chronic alcohol group, CA). Two control groups, an oral gavage sucrose control group (chronic alcohol control group, CAc) and a non-treated control group (NTc), were also examined. At 56 days post-natal, the pups from each group were sacrificed and the whole brain sectioned in a coronal plane and immunolabeled for cholineacetyltransferase (ChAT), tyrosine hydroxylase (TH), serotonin (5HT) and orexin-A (OxA) which labels cholinergic, catecholaminergic, serotonergic and orexinergic structures respectively. The overall nuclear organization and neuronal morphology were identical in all three groups studied, and resemble that previously reported for laboratory rodents. Quantification of the estimated numbers of ChAT immunopositive (+) neurons of the pons, the TH+ neurons of the pons and the OxA+ neurons of the hypothalamus showed no statistically significant difference between the three experimental groups. The stereologically estimated areas and volumes of OxA+ neurons in the CA group were statistically significantly larger than the groups not exposed to prenatal alcohol, but the ChAT+ neurons in the CA group were statistically significantly smaller. The density of orexinergic boutons in the anterior cingulate cortex was lower in the CA group than the other groups. No statistically significant difference was found in the area and volume of TH+ neurons between the three experimental groups. These differences are discussed in relation to the sleep disorders recorded in children with fetal alcohol spectrum disorder (FASD).

Keywords: fetal alcohol spectrum disorder, brain, sleep, sleep disorders, immunohistochemistry, quantitative studies

## INTRODUCTION

Adequate quality sleep, especially in children, is important for the development of the brain (Chen et al., 2012; Ipsiroglu et al., 2013). Certain neural functions such as learning, cognition, motor activities, attention, alertness, as well as brain growth and metabolic activities, are only performed effectively if there has been adequate sleep (Chen et al., 2012; Ipsiroglu et al., 2013).

#### Edited by:

Nilesh Bhailalbhai Patel, University of Nairobi, Kenya

#### Reviewed by:

Jean-Pierre Hornung, University of Lausanne, Switzerland Susana Pilar Gaytan, Universidad de Sevilla, Spain

> \*Correspondence: Paul R. Manger paul.manger@wits.ac.za

Received: 10 July 2017 Accepted: 13 November 2017 Published: 27 November 2017

#### Citation:

Olateju OI, Bhagwandin A, Ihunwo AO and Manger PR (2017) Changes in the Cholinergic, Catecholaminergic, Orexinergic and Serotonergic Structures Forming Part of the Sleep Systems of Adult Mice Exposed to Intrauterine Alcohol. Front. Neuroanat. 11:110. doi: 10.3389/fnana.2017.00110 Sleep problems are often observed in concert with neurodevelopmental conditions such as autism (Malow et al., 2006), attention deficit hyperactivity disorders (Cortese et al., 2009) and fetal alcohol syndrome (FAS; Streissguth et al., 2004; May et al., 2009; Olson et al., 2009; Chen et al., 2012; Ipsiroglu et al., 2013). Children with FAS show an unwillingness to go to bed (Meltzer and Mindell, 2004; Wengel et al., 2011), experience short sleep duration (Jan et al., 2010; Wengel et al., 2011), and experience sleep anxiety with frequent sleep disruptions (Haydon et al., 2009; Wengel et al., 2011). Moreover, FAS children report night terrors (Durmer and Dinges, 2005), exhibit sleep walking (Randazzo et al., 1998), daytime tiredness (Lancioni et al., 1999) and suppressed sensory information processing (Wengel et al., 2011) when compared to neurotypical children. Similarly, in laboratory rodents exposed to intrauterine alcohol, polysomnographic recording of the sleep-wake cycle demonstrated that total sleep duration was significantly reduced, with a concomitant increase in total wake time, and increases in the numbers of both sleep and wake episodes (Hilakivi, 1986; Stone et al., 1996).

The neural systems that control and regulate sleep are comprised of neurons in specific nuclear clusters in the basal forebrain, hypothalamus and pons, which produce a variety of neurotransmitters and project throughout the brain. These neurons depolarize in specific patterns during wake, slow wave sleep (SWS) and rapid eye movement sleep (REM; Datta and MacLean, 2007; Lyamin et al., 2008; Takahashi et al., 2010; Dell et al., 2012; Bhagwandin et al., 2013; Petrovic et al., 2013). Due to the oscillatory relationship of these neuronal groups in the sleep-wake cycle, it may be that the anatomy of these neuronal systems differ from neurotypical in FAS children with sleep disorders; however, to the author's knowledge, there appear to be no reports that directly correlate morphological changes in the sleep centers in FAS subjects that are known to experience sleep disorders. In order to determine whether morphological changes in the neural systems involved in the production and regulation of sleep may be correlated to the sleep disorders observed in FAS children, we targeted for investigation the organization, morphology and numbers of pontine cholinergic (laterodorsal tegmental, LDT and pedunculopontine nuclei), pontine catecholaminergic (locus coeruelus complex, LC), hypothalamic orexinergic and midbrain serotonergic (dorsal raphe) neurons of 56 day old C57BL/6J mice following exposure to prenatal alcohol. In addition to qualitative examination of these neurons involved in sleep, stereological analysis of neuronal numbers and neuronal size was undertaken for the cholinergic LDT and pedunculopontine tegmental nucleus (PPT), the noradrenergic LC, the orexinergic neurons of the hypothalamus, and orexinergic bouton density in the anterior cingulate cortex.

## MATERIALS AND METHODS

## Breeding and Prenatal Ethanol Exposure

All animals were treated and used according to the guidelines of the University of the Witwatersrand Animal Ethics Committee (Clearance No. 2012/15/2B), which parallel those of the NIH for the care and use of animals in scientific experimentation. Female C57BL/6J mice (Mus musculus), 12 weeks of age, were allocated into three experimental groups: Chronic Alcohol exposure (CA), control for Chronic Alcohol exposure (CAc) and a Non-Treatment control group (NTc). For effective mating, 1–2 female mice were introduced into the cage of a C57BL/6J male mouse for 12 h, which was considered gestational day 0 (GD 0). In all, a total of 14 female mice (4–5 mice assigned to each experimental group) and 8 male mice were used to generate the required numbers of pups used in the present study.

For the CA group, each pregnant mouse received a dose of 7.5 µl/g of 50% alcohol in distilled water (2.9 g/kg) per day (Haycock and Ramsay, 2009; Knezovich and Ramsay, 2012) for 10 consecutive days by oral gavage, starting from GD 7 (Webster et al., 1980, 1983; Sulik et al., 1981; Choi et al., 2005; Redila et al., 2006; Parnell et al., 2009), while each pregnant mouse in the CAc group received an equivalent dose of isocaloric sucrose (704 g/L) by oral gavage over the same period (Haycock and Ramsay, 2009; Knezovich and Ramsay, 2012). To control for the possible influence of post-traumatic stress in the pregnant mice, pregnant mice in the NTc group did not undergo any oral gavage. Food and water was provided ad libitum to the mice, except in the control groups (CAc and NTc), where it was withheld for 2 h post-gavage in order to partially control for the reduction in feeding during the period of peak intoxication of the alcohol-treated dams (Haycock and Ramsay, 2009). The pups were weaned 21 days after birth and then the male and female pups separated. Three pups of the same sex from each experimental group were kept in separate cages (cage dimensions: 200 × 200 × 300 mm) with adequate food and water supplies until post-natal day (PND) 56.

## Blood Alcohol Concentration Assay in the Pregnant Mice

On the last day (10th day) of oral gavage (GD 16), a small lesion was made at the site of the saphenous vein on the left hind legs of all the pregnant mice in the CA and CAc experimental groups. The saphenous bleeding procedure was performed on the pregnant mice in the sucrose group in order to mimic the effects of the bleeding on the alcohol exposed pregnant mice. The non-treatment pregnant mice served as controls for the possible effects of the bleeding and/or the oral gavage procedures. Fifty microliter of blood was drawn into heparinized capillary tubes at 30 min post-gavage (Bielawski and Abel, 1997) to determine the blood alcohol concentration (BAC). The blood samples from the FAS model and the sucrose control were stored at 4◦C overnight after which they were centrifuged with Vivaspin500 100 µm membrane tubes (Biotech, South Africa) for 30 min before the serum was extracted and the BAC analyzed using an EnzyChromTM Ethanol Assay Kit (BioVision, South Africa). The pregnant mice belonging to the CA group that were administered alcohol had an average BAC of 1.84 mg/ml (s.e. = 0.39), which is above the pharmacologically significant level of 1 mg/ml reported by Rhodes et al. (2005) and Sulik (2005).

## Sacrifice and Tissue Processing

At PND 56, when the mice reached adulthood, a total number six mice (n = 6) from each experimental group (1–2 mice randomly selected from each litter to control for potential genetic influences) were weighed and then euthanized (Euthanaze 1 ml/kg, contains sodium pentobarbitone 100 mg/ml, intra-peritoneally) before being perfused trans-cardially with 0.9% cold (4◦C) saline followed immediately by cold 4% paraformaldehyde in 0.1 M phosphate buffer (PB). The brain was removed from the skull, weighed and post-fixed for 24 h in 4% paraformaldehyde in 0.1 M PB at 4◦C. The brains were then cryoprotected by immersion in 30% buffered sucrose solution in 0.1 M PB at 4◦C until they equilibrated. The brains for all 18 mice was then frozen in crushed dry ice, and sectioned in a coronal plane at 50 µm thickness using a sliding microtome. A one in four series of sections was taken. The first series of sections, stained for Nissl substance to reveal cytoarchitectural features, was mounted on 0.5% gelatine-coated slides, dried overnight, cleared overnight in a 1:1 mixture of 100% ethanol and 100% chloroform and stained with 1% cresyl violet in H2O. The second series of sections was immunostained with an antibody to cholineacetyltransferase (ChAT, AB144P, Chemicon, raised in goat) to reveal cholinergic neurons, and the third series was immunostained with an antibody to tyrosine hydroxylase (TH; AB151, Chemicon, raised in rabbit) to reveal catecholaminergic neurons. The fourth series of sections was divided at the level of the posterior commissure. All sections rostral to the posterior commissure were immunostained with an antibody to orexin-A (OxA, AB3704, Merck-Millipore, raised in rabbit) to reveal the hypothalamic orexinergic neurons and orexinergic boutons in the cerebral cortex, while all section caudal to the posterior commissure were immunostained with an antibody to serotonin (5HT, AB938, Chemicon, raised in rabbit) to reveal the serotonergic neurons.

## Immunostaining Protocol

All sections used for immunohistochemical staining were initially incubated in a solution containing 1.6% of 30% H2O2, 49.2% methanol and 49.2% 0.1 M PB solution, for 30 min to reduce endogenous peroxidase activity, which was followed by three 10 min rinses in 0.1 M PB. To block non-specific binding sites the sections were then preincubated at room temperature for 2 h in a blocking buffer solution containing 3% normal serum (normal rabbit serum, NRS, Chemicon, for ChAT sections and normal goat serum, NGS, Chemicon, for the remaining sections), 2% bovine serum albumin (BSA, Sigma) and 0.25% Triton X-100 (Merck) in 0.1 M PB. The sections were then placed in a primary antibody solution (blocking buffer with correctly diluted primary antibody) and incubated at 4◦C for 48 h under gentle shaking. To reveal cholinergic neurons, anti-ChAT at a dilution of 1:3000 was used. To reveal putative catecholaminergic neurons, anti-TH was used at a dilution of 1:3000. To reveal serotonergic neurons, anti-5-HT at a dilution of 1:5000 was used. To reveal hypothalamic orexinergic neurons and cortical orexinergic boutons, anti-OxA at a dilution of 1:3000 was used.

This was followed by three 10 min rinses in 0.1 M PB, after which the sections were incubated in a secondary antibody solution for 2 h at room temperature. The secondary antibody solution contained a 1:1000 dilution of biotinylated anti-goat IgG (BA-5000, Vector labs, for ChAT sections) or biotinylated anti-rabbit IgG (BA-1000, Vector Labs, for the remaining sections) in a solution containing 3% NGS/NRS and 2% BSA in 0.1 M PB. This was followed by three 10 min rinses in 0.1 M PB after which the sections were incubated in an avidin-biotin solution (Vector Labs) for 1 h. After three further 10 min rinses in 0.1 M PB, the sections were placed in a solution of 0.05% diaminobenzidine in 0.1 M PB for 5 min (1 ml/section), followed by the addition of 3 µl of 30% H2O<sup>2</sup> to the solution in which each section was immersed. The precipitation process was stopped by immersing the sections in 0.1 M PB and then rinsing them twice more in 0.1 M PB. To check for non-specific staining from the immunohistochemistry protocol, we omitted the primary antibody and the secondary antibody in selected sections, which produced no evident staining. The sections were then mounted on 0.5% gelatine coated glass slides, dried overnight, dehydrated in a graded series of alcohols, cleared in xylene and coverslipped with DPX.

## Qualitative and Quantitative Determination of Cell Numbers and Statistical Analysis

The distribution of immunopositive cells was compared qualitatively between the experimental groups using both low and higher power light microscopy. Digital photomicrographs of these cells were captured using Zeiss Axioshop and Axiovision software. No pixelation adjustments or manipulation of the captured images was undertaken, except for the adjustment of contrast, brightness and levels using Adobe Photoshop 7.

For the quantification of Chat+, TH+, OxA+ cells and OxA+ boutons, an unbiased systematic random sampling stereological design protocol was employed. We used a MicroBrightfield (MBF; Colchester, VT, USA) system with three plane motorized stage, Zeiss.Z2 vario axioimager and StereoInvestigator software (MBF, version 11.08.1; 64-bit). Separate pilot studies for each immunohistochemical stain for each group, were conducted to optimize sampling parameters, such as the counting frame and sampling grid sizes, and achieve a coefficient of error (CE) below 0.1 (Gundersen et al., 1988; West et al., 1991; Dell et al., 2016). At this point we would like to acknowledge that while every effort was made to obtain a CE below 0.1, this was not always achieved mostly due to a limited number of countable sections in those instances. In addition we measured the tissue section thickness at every 10th sampling site and the vertical guard zone was determined according to tissue thickness to avoid errors/biases due to sectioning artifacts (West et al., 1991; Dell et al., 2016). We decided to maintain consistency amongst sampling parameters between the groups studied for each neuroanatomical region and employed a single-blind procedure, to reduce unfavorable


stereological estimation biases. **Table 1** provides a detailed summary of the parameters used for each neuroanatomical region and between the groups in the current study. Finally, with regard to the OxA+ bouton densities, a region of interest (ROI) measuring 1200 × 800 µm encompassing all six layers in similar locations of the anterior cingulate cortex in all three groups was used to calculate the densities reported.

To estimate the total number of pontine Chat+ neurons (LDT and PPT), pontine locus coeruleus TH+ neurons, hypothalamic OxA+ neurons and cortical OxA+ boutons, we used the ''Optical Fractionator'' probe and the following equation (West et al., 1991; Dell et al., 2016):

$$N = Q/(\text{SSF} \times \text{ASF} \times \text{TSF}) \tag{1}$$

where N was the total estimated neuronal number, Q was the number of neurons counted, SSF was the fraction of the sections sampled, ASF was the area sub fraction (which is calculated by the ratio of the size of the counting frame to the size of the sampling grid), and TSF was the thickness sub fraction (which is calculated by the ratio of the disector height relative to the section thickness).

To determine neuronal sizes, we used the ''Nucleator'' probe. For all individuals counted these probes were used concurrently while maintaining strict criteria, e.g., only neurons with complete cell bodies were counted, and obeying all commonly known stereological rules.

An analysis of variance (ANOVA) was performed to determine if there was a significant variation in the mean ChAT+, TH+, or OxA+ neuronal counts, areas and volumes of all the experimental groups (CA, CAc and NTc). In addition, ANOVA was used to determine sex differences in the mean ChAT+, TH+, or OxA+ neuronal counts, areas and volumes within and/or between the sexes within and between groups. Where the ANOVA was significant, a post hoc analysis using Tukey's pairwise comparison revealed where significant differences existed. In this report, only statistically significant differences in the mean neuronal counts, areas and volumes are reported in the results. All statistical analyses were performed using SPSS Inc programme (version 22.0). A significance level of 5% was used as an indicator of significant differences for all statistical analyses.

## RESULTS

## General Observations on the Body and Brain

The pups that experienced chronic prenatal alcohol exposure (CA group) showed no overt signs of FAS, in that no craniofacial abnormalities were readily apparent and there was no evident reduction in overall body mass. At the time of sacrifice, the average body masses of the mice were: CA male—19.75 g (s.e. 0.75 g), CA female—15.13 g (s.e. 0.55 g); CAc male— 19.88 g (s.e. 0.4 g), CAc female—16.00 g (s.e.

mouse in the coronal plane immunostained for cholineacetyltransferase (ChAT) in the three different groups analyzed in the present study, the group exposed to chronic prenatal alcohol (CA; A,B), the prenatal gavage control group (CAc; C,D) and the non-treated control group (NTc; E,F). The nuclear organization and number of ChAT immunostained cells was similar between groups, but the soma of the neurons in the CA group, exposed to chronic prenatal alcohol, were statistically significantly smaller than the two control groups (CAc and NTc). In all images dorsal is to the top and medial to the left. Scale bar in (E) = 250 µm and applies to (A,C,E) scale bar in (F) = 100 µm and applies to (B,D,F).

0.29 g); NTc male—20.25 g (s.e. 0.75 g), NTc female— 15.63 g (s.e. 0.24 g). In addition, there were no observable differences in the general morphology of the brains of mice treated with alcohol (CA group), sucrose (CAc group) or the non-treated control group (NTc). The average brain masses recorded for each group were: CA male—0.420 g (s.e. 0.02 g), CA female—0.390 g (s.e. 0.007 g); CAc male—0.396 g (s.e. 0.005 g), CAc female—0.376 g (s.e. 0.003 g); NTc male—0.400 g (s.e. 0.006 g), NTc female—0.390 g (s.e. 0.007 g). No statistically significant differences were observed between experimental groups in terms of body or brain mass when mice of the same sex were compared.

## Cholinergic Neurons of the Laterodorsal Tegmental and Pedunculopontine Nuclei

For all mice in all three experimental groups, there were no marked differences in the location and morphology of the ChAT+ neurons within the PPT and the LDT nuclei. The ChAT+ neurons forming the PPT were located within the dorsal aspect of the pontine tegmentum extending from the ventrolateral border of the periaqueductal/periventricular gray matter to the superior cerebellar peduncle (**Figure 1**). A moderate to high density of ChAT+ neurons were observed in this region. The PPT and LDT ChAT+ neurons exhibited a variety of somal shapes due to being multipolar (**Figure 1**).

Our quantitative estimation of the numbers of ChAT+ neurons in the LDT and PPT in the brain of mice from the three different experimental groups revealed a distinct homogeneity in numbers between individuals in the same group, and between groups. For the CA group, the average number of ChAT+ neurons was 1741.4 (s.e. 276.8) (male: 1933.2 (s.e. 427.8), female: 1549.6 (s.e. 450.8)), for the CAc group it was 1844.3 (s.e. 312.7) (male: 1686.9 (s.e. 593.5), female: 1646.1 (s.e. 311.9)) and for the NTc group it was 2091.9 (s.e. 405.2) (male: 1256.0 (s.e. 355.0), female: 2961.4 (s.e. 62.6); **Figure 2A**). While there is a trend for the CA and CAc groups to have less ChAT+ neurons than the NTc group, these were not statistically significant differences. ANOVA and post hoc pairwise comparisons revealed that the lower number of ChAT+ neurons in the CA group was not statistically significantly different from the CAc group (p = 0.977) or the NTc group (p = 0.763). A comparison between CAc and NTc groups was also not statistically significantly different (p = 0.872).

In terms of average somal area, the CA group had an average somal area of 89.7 µm<sup>2</sup> (s.e. 1.0) (male: 78.8 (s.e. 1.3), female: 100.5 (s.e. 1.6)), for the CAc group it was 95.0 µm<sup>2</sup> (s.e. 0.7) (male: 100.1 (s.e. 0.9), female: 86.4 (s.e. 1.1)) and for the NTc group it was 92.9 µm<sup>2</sup> (s.e. 0.8) (male: 98.9 (s.e. 1.4), female: 90.0 (s.e. 0.9); **Figure 3A**). Statistical analyses using ANOVA and post hoc pairwise comparisons revealed that the smaller somal area of the ChAT+ neurons in the CA group was statistically significantly smaller than the CAc group (p = 3.8 × 10−<sup>5</sup> ) and the NTc group (p = 0.017). A comparison of somal areas between the CAc and NTc groups was not statistically significantly different (p = 0.170). In regards to average somal volume, the CA group exhibited an average somal volume of 707.5 µm<sup>3</sup> (s.e. 12.5) (male: 572.1 (s.e. 14.9), female: 827.9 (s.e. 20.0)), for the CAc group it was 763.9 µm<sup>3</sup> (s.e. 8.9) (male: 823.0 (s.e.

16.1), female: 663.9 (s.e. 13.1)) and for the NTc group it was 739.4 µm<sup>3</sup> (s.e. 9.4) (male: 816.4 (s.e. 18.2), female: 703.3 (s.e. 10.8); **Figure 3B**). Statistical analyses revealed that the smaller somal volumes of the ChAT+ neurons in the CA group was statistically significantly smaller than the somal volumes of the CAc group (p = 3 × 10−<sup>4</sup> ), but not significantly different from the NTc group (p = 0.070). A comparison of the somal volumes between the CAc and NTc groups was not statistically significantly different (p = 0.208). Thus, for the pontine cholinergic neurons, the only substantive difference

immunostained cells in the LC was similar between groups. In all images dorsal is to the top and medial to the left. Scale bar in (E) = 250 µm and applies to (A,C,E) scale bar in (F) = 100 µm and applies to (B,D,F).

observed between the groups was the reduction in size, by around 5%–10%, of the soma in the group exposed to prenatal alcohol.

When comparing the two sexes, the somal areas and volumes in the CA group was statistically significantly different from the CAc (p = 2.2 × 10−<sup>5</sup> ) and NTc (p = 2.2 × 10−<sup>5</sup> ) groups. There was similarly sex differences between the two sexes in each experimental group for the somal areas (CAmale vs. CAfemale : p = 2 × 10−<sup>5</sup> ; CAcmale vs. CAcfemale : p = 2 × 10−<sup>5</sup> ; NTcmale vs. NTcfemale : p = 3.5 × 10−<sup>5</sup> ) and somal volumes (CAmale vs. CAfemale : p = 2 × 10−<sup>5</sup> ; CAcmale vs. CAcfemale : p = 2 × 10−<sup>5</sup> ; NTcmale vs. NTcfemale : p = 2.6 × 10−<sup>5</sup> ).

## Catecholaminergic Neurons of the Locus Coeruleus Complex

For all mice in all three experimental groups, there were no marked differences in the location and morphology of the TH+ neurons within the LC of the pontine region, where the TH+ neurons forming the locus coeruleus were readily observed (**Figure 4**). This complex contained five nuclei: the subcoeruleus compact portion (A7sc), subcoeruleus diffuse portion (A7d), locus coeruleus compact portion (A6c), fifth arcuate nucleus (A5) and the dorsolateral division of locus coeruleus (A4). The distribution of the neurons forming these five nuclear subdivisions of the LC was the same as what has been previously described in other laboratory rodents (Dahlström and Fuxe, 1964; Olson and Fuxe, 1972), thus an extensive description is not provided here. All neurons throughout the LC showed a similar variety of somal shapes and all were multipolar.

Our quantitative estimation of the numbers of TH+ neurons in the A6 and A7 nuclei of the LC from the three different experimental groups revealed a distinct homogeneity in numbers between individuals in the same group, and between groups (**Figure 2B**). For the CA group, the average number of TH+ neurons was 549.9 (s.e. 113.6) (male: 424.8 (s.e. 180.5, female: 674.9 (s.e. 116.4)), for the CAc group it was 510.2 (s.e. 60.6) (male: 576.9 (s.e. 38.0), female: 465.7 (s.e. 96.4)) and for the NTc group it was 512.8 (s.e. 104.4) (male: 409.2 (s.e. 148.5), female: 668.2 (s.e. 51.8); **Figure 2B**). While the number of TH+ neurons is highest in the CA group, statistical analyses using ANOVA and post hoc pairwise comparisons revealed that the number of TH+ neurons in the CA group was not statistically significantly different from the CAc group (p = 0.952) or the NTc group (p = 0.958). A comparison between CAc and NTc groups was also not statistically significantly different (p = 0.999).

In terms of average somal area, the CA group had an average somal area of 111.1 µm<sup>2</sup> (s.e. 1.9) (male: 109.8 (s.e. 2.2), female: 113.8 (s.e. 3.6)), for the CAc group it was 115.5 µm<sup>2</sup> (s.e. 1.9) (male: 113.9 (s.e. 2.4), female: 118.2 (s.e. 3.0)) and for the NTc group it was 114.1 µm<sup>2</sup> (s.e. 2.5) (male: 139.1 (s.e. 5.1), female: 101.7 (s.e. 2.3); **Figure 3C**). Statistical analyses using ANOVA and post hoc pairwise comparisons revealed that the average somal areas of the TH+ neurons in the CA group was not statistically significantly different from the CAc group (p = 0.305) or the NTc group (p = 0.578). A comparison between CAc and NTc groups was also not statistically significantly different (p = 0.884). In regards to average somal volume, the CA group exhibited an average somal volume of 955.5 µm<sup>3</sup> (s.e. 25.2) (male: 937.5 (s.e. 29.5), female: 991.9 (s.e. 47.1)), for the CAc group it was 1011.3 µm<sup>3</sup> (s.e. 25.6) (male: 993.8 (s.e. 32.7), female: 1041.3 (s.e. 40.8)) and for the NTc group it was 1007.1 µm<sup>3</sup> (s.e. 33.7) (male: 1347.6 (s.e. 72.4), female: 838.7 (s.e. 28.7); **Figure 3D**). Statistical analyses revealed that the average somal volume of the TH+ neurons in the CA group was not statistically significantly different from the CAc group (p = 0.340) or the NTc group (p = 0.397). A comparison between CAc and NTc groups was similarly not statistically significantly different (p = 0.994). Thus, for the neurons of the LC, no differences in nuclear organization, neuronal morphology, neuronal numbers, or somal size was detected between the three experimental groups investigated.

When analyzing the two sexes independently, the somal areas for the NTc group were significantly different from the CA group (male: p = 2 × 10−<sup>5</sup> ; female: p = 0.010) and the CAc group (male: p = 2 × 10−<sup>5</sup> ; female: p = 2 × 10−<sup>4</sup> ). Likewise, somal volume in the NTc group was significantly different from the CA group (male: p = 2 × 10−<sup>5</sup> ; female: p = 0.010) and the CAc group (male: p = 2 × 10−<sup>5</sup> ; female: p = 5 × 10−<sup>4</sup> ). When comparing between the sexes within the same group, the somal areas and volumes in the NTc group (i.e., NTcmale vs. NTcfemale) were significantly different (somal area: p = 2 × 10−<sup>5</sup> ; somal volume: p = 2 × 10−<sup>5</sup> ).

## Serotonergic Neurons of the Dorsal Raphe Nuclear Complex

For all mice in all three experimental groups, there were no marked differences in the location and morphology of the 5-HT+ neurons within the dorsal raphe nuclear complex (**Figure 5**). Within the dorsal raphe nuclear complex there were six distinct nuclei: the dorsal raphe interfascicular (DRif) nucleus, dorsal raphe ventral (DRv) nucleus, dorsal raphe dorsal (DRd) nucleus, dorsal raphe lateral (DRl) nucleus, dorsal raphe peripheral (DRp) nucleus and the dorsal raphe caudal (DRc) nucleus (**Figure 5**). The distribution of the serotonergic neurons forming these nuclei and their relationship to architectonic borders is similar to that previously described for laboratory rodents (e.g., Steinbusch, 1981), and thus a full description is not provided herein. The neurons of the DRp, DRl and DRc nuclei were readily distinguishable from the DRif, DRv and DRd nuclei since they were qualitatively substantially larger and multipolar (**Figures 5B,D,F**). Thus, for the neurons of the dorsal raphe complex, no differences in nuclear organization or neuronal morphology was detected between the three experimental groups investigated.

## Orexinergic Neurons of the Hypothalamus

For all mice in all three experimental groups, there were no marked differences in the location and morphology of the OxA+ neurons within the hypothalamus (**Figure 6**). Within this aggregation of OxA+ hypothalamic neurons we could identify three clusters—a main cluster (Mc), a zona incerta cluster (Zic) and an optic tract cluster (Otc; **Figure 6**). The distributions of the neurons forming these clusters are similar to that reported in several rodent species previously and are thus not described in detail herein. All the OxA+ neurons were multipolar and exhibited a variety of somal shapes.

Our quantitative estimation of the numbers of OxA+ neurons in the hypothalamus of the three different experimental groups revealed a distinct homogeneity in numbers between individuals in the same group, and between groups. For the CA group, the average number of OxA+ neurons was 840.1 (s.e. 115.6) (male: 669.9 (s.e. 71.1), female: 1095.6 (s.e. 98.7)), for the CAc group it was 762.6 (s.e. 107.4) (male: 773.8 (s.e. 239.0), female: 751.5 (s.e. 21.7)) and for the NTc group it was 684.8 (s.e. 131.4) (male: 631.7 (s.e. 178.5), female: 738.0 (s.e. 227.1); **Figure 2C**). The number of OxA+ neurons was highest in the CA group, but statistical analyses using ANOVA and post hoc pairwise comparisons revealed that the number of OxA+ neurons in the CA group was not statistically significantly different from the CAc group (p = 0.892) or the NTc group (p = 0.639). A comparison between CAc and NTc groups was also not statistically significantly different (p = 0.891).

In terms of average somal area, the OxA+ neurons in the CA group had an average somal area of 63.3 µm<sup>2</sup> (s.e. 1.3) (male: 66.7 (s.e. 2.0), female: 60.3 (s.e. 1.5)), for the CAc group it was 58.9 µm<sup>2</sup> (s.e. 1.3) (male: 62.2 (s.e. 2.1), female: 56.2 (s.e. 1.6)) and for the NTc group it was 55.8 µm<sup>2</sup> (s.e. 1.1)

(male: 53.5 (s.e. 1.8), female: 57.5 (s.e. 1.4); **Figure 3E**). Statistical analyses using ANOVA and post hoc pairwise comparisons revealed that the OxA+ somal areas in the CA group were statistically significantly larger than the OxA+ neurons of the CAc group (p = 0.026) and the NTc group (p = 5 × 10−<sup>5</sup> ). A comparison of somal area between the CAc and NTc groups was not statistically significantly different (p = 0.165; **Figure 3E**). In regards to average somal volume, the OxA+ neurons in the CA group exhibited an average somal volume of 426.3 µm<sup>3</sup> (s.e. 13.1) (male: 457.0 (s.e. 20.9), female: 391.0 (s.e. 14.2)), for the CAc group it was 379.3 µm<sup>3</sup> (s.e. 12.5) (male: 411.7 (s.e. 21.1), female: 352.9 (s.e. 14.6)) and for the NTc group it was 350.9 µm<sup>3</sup> (s.e. 10.9) (male: 334.9 (s.e. 18.6), female: 363.2 (s.e. 13.0); **Figure 3F**). Similar to the somal area, statistical analyses revealed that the somal volume in the OxA+ neurons in the CA group was statistically significantly larger than the OxA+ neurons of the CAc group (p = 0.030) and the NTc group (p = 9 × 10−<sup>5</sup> ). A comparison of somal volume between CAc and NTc groups was not statistically significantly different (p = 0.217; **Figure 3F**). Thus, for the hypothalamic orexinergic neurons, the only substantive difference observed between the groups was the increase in size, by around 7%–20%, of the soma in the group exposed to prenatal alcohol.

control group (CAc) (C,D) and the NTc (E,F). The nuclear organization and number of OxA immunostained cells was similar between groups, but the soma of the neurons in the CA group, exposed to chronic prenatal alcohol, were statistically significantly larger than the two control groups (CAc and NTc). In all images dorsal is to the top and medial to the left. Scale bar in (E) = 250 µm and applies to (A,C,E), scale bar in (F) = 100 µm and applies to (B,D,F). Mc, main orexinergic cluster; OTc, optic tract cluster; Zic, zona incerta cluster.

When comparing males between experimental groups, the somal areas of the NTc group were significantly different from the CA group (p = 3 × 10−<sup>5</sup> ) and the CAc group (p = 6 × 10−<sup>3</sup> ). In addition, the somal volumes of the NTc group were significantly different from the CA (p = 9 × 10−<sup>5</sup> ) and the CAc (p = 0.021) groups.

## Orexinergic Boutons of the Cerebral Cortex

For all mice in all three experimental groups, there were no marked differences in the location and morphology of the OxA+ boutons along the orexinergic axonal fibers in the anterior cingulate cortex (**Figure 7**). The ramifications of terminal axonal fibers were broadly distributed across all the cortical cell layers in the cerebral cortex and they appeared to have no specific spatial organization, although many traversed the cortex in a vertical manner. The OxA+ en passant boutons were distinct along the terminal axonal fibers and exhibited readily distinguishable small and large boutons as reported previously for this system (Dell et al., 2015). Qualitatively, in all three experimental groups, there appeared to be more small boutons than large boutons (**Figure 7**).

Our quantitative analysis of the density of OxA+ boutons in the anterior cingulate cortex of the three different experimental

FIGURE 7 | High magnification photomicrographs of the orexinergic axons and boutons in the anterior cingulate cortex of the mouse in the three different groups analyzed in the present study, the group exposed to chronic prenatal alcohol (CA; A), the prenatal gavage control group (CAc; B) and the NTc (C). Note the lower density of orexinergic axons and boutons in the CA group compared to the two control groups (CAc and NTc). Scale bar in (C) = 100 µm and applies to all. In all images the pial surface is towards the top of the image.

groups revealed a distinct homogeneity in numbers between individuals in the same group. For the CA group, the average density of OxA+ boutons was 60 boutons/mm<sup>3</sup> (s.e. 4), for the CAc group it was 80 boutons/mm<sup>3</sup> (s.e. 6) and for the NTc group it was 100 boutons/mm<sup>3</sup> (s.e. 8; **Figure 2D**). The density of OxA+ boutons was lowest in the CA group, but highest in the NTc group. Statistical analyses using ANOVA and post hoc pairwise comparisons revealed that the number of OxA+ boutons in the CA group was statistically significantly different from the NTc group (p = 0.002) but not the CAc group (p = 0.081). A comparison between CAc and NTc groups was also not statistically significantly different (p = 0.136). Thus, it appears that the prenatal exposure to alcohol lowers the density of orexinergic boutons in the cerebral cortex.

## DISCUSSION

In the present study we examined the nuclear organization, neuronal morphology, stereologically estimated total cell number and average cell size of four specific clusters of nuclei known to be involved in the control and regulation of the sleep-wake cycle. In addition, we examined the axonal bouton density of the orexinergic system in the anterior cingulate cortex. As well as differences between groups, we examined differences between the sexes within and between experimental groups; however, the differences between the sexes within and between groups were not systematic and are not discussed further. As outlined earlier, sleep in children suffering from fetal alcohol spectrum disorder (FASD) is perturbed in many ways, therefore a targeted analysis of some of the critical neuronal regulators of sleep was undertaken in mice exposed to a chronic prenatal alcohol regime to determine whether any clues to the basis for sleep disorders in FASD children could be found. For the most part, we did not record any major differences between mice exposed to prenatal alcohol and those not. The organization of the nuclei that form the pontine cholinergic, pontine catecholaminergic, midbrain serotonergic and hypothalamic orexinergic was similar across all mice studied. Furthermore, the specific neuronal morphology within the various nuclei was also unaltered by prenatal exposure to alcohol. The somal volumes and areas of the pontine catecholaminergic neurons of the LC were not affected by prenatal alcohol exposure; however, the somal volumes and areas of the pontine cholinergic and hypothalamic orexinergic did show significant variation in the group exposed to prenatal alcohol compared to both control groups. In addition, the density of orexinergic boutons in the anterior cingulate cortex was lower in the group exposed to prenatal alcohol than the control groups. These three specific differences are discussed in terms of the effect they may have on the control and regulation of the sleep wake cycle and how this relates to the observed sleep disorders in children with FASD.

## The Soma of the Pontine Cholinergic Neurons Are Smaller in Mice Exposed to Prenatal Alcohol

The first specific difference observed in the mice that underwent a chronic prenatal alcohol regime compared to the control mice was the smaller size (around 95%) of the pontine cholinergic neuronal soma of the LDT and PPT nuclei. The cholinergic neurons of the LDT and PPT are known to be involved in the generation of REM sleep, and the phasic events of REM sleep such as ponto-geniculo-occipital (PGO) spikes, through their projections to the thalamus and the basal forebrain (Webster and Jones, 1988; Semba et al., 1990). Additionally, these neurons are involved in the activation of thalamocortical systems by blocking synchronized oscillations such as seen in SWS (Paré et al., 1988). In the FASD mouse model generated in this study, these cholinergic neurons were reduced in size, which would hypothetically indicate that in the FASD model mice they would have a higher threshold potential, thus being less excitable, and being smaller, they may only be able to support a smaller axon and axon terminal field (Shepherd, 1979), making their action upon the recipient neurons less intense. Indeed, in experimental models where these neurons have undergone chemical ablation it was found that the amount of time spent in wake was significantly increased (Webster and Jones, 1988). Thus, it appears that the difficulty experienced in going to sleep, short sleep durations and increased sleep disruptions observed in FASD children (Meltzer and Mindell, 2004; Jan et al., 2010; Wengel et al., 2011) may in part be related to a reduction in size of the cholinergic neurons of the LDT and PPT.

## The Soma of the Hypothalamic Orexinergic Neurons Are Larger in Mice Exposed to Prenatal Alcohol, but Bouton Density in the Anterior Cingulate Cortex Is Lower

The second specific difference observed in the mice treated with a chronic prenatal alcohol regime when compared to the control groups was the significantly larger size of the soma of the hypothalamic orexinergic neurons. Interestingly, the third specific difference observed was that the orexin bouton density in the cerebral cortex was lower in the mice treated with a chronic prenatal alcohol regime. The hypothalamic orexinergic neurons are known to project throughout the entire central nervous system, but exhibit specifically intense projections to other regions of the brain involved in arousal (e.g., Peyron et al., 1998). Thus, orexin (or hypocretin) has been implicated in arousal as one of the main functions of this system (e.g., Saper et al., 2001; Siegel, 2004a,b). Many of the disorders associated with sleep in FASD children appear to be related to problems associated with arousal, and thus the arousal systems of the brain (Meltzer and Mindell, 2004; Haydon et al., 2009; Jan et al., 2010; Wengel et al., 2011). In this sense, the changes observed in the orexinergic system in the currently used mouse model of FASD is of interest—perhaps the changes observed in this mouse model can help explain some of the sleep disturbances observed in FASD children.

The orexinergic neurons in the FASD mouse model were approximately 1.1–1.2 times larger than those measured in the control mice. One of the key cellular level phenomena associated with an increase in neuron size, and thus an increase in neuronal surface area, is a lower threshold potential (Shepherd, 1979). Thus, the larger orexinergic neurons found in the FASD mouse model would, theoretically, be more readily excitable when compared to the control groups. In this sense, the brain globally, and in particular the arousal systems receiving intense projections from the orexinergic neurons of the hypothalamus, would potentially be in receipt of a greater number of excitatory action potentials from the hypothalamus in the FASD mouse model than the controls. This may augment the arousal function of the orexinergic neurons. The second key feature associated with increased neuron size would be the theoretical ability of the larger cell to support a larger and more extensively branched axon and axonal bouton terminal field (Shepherd, 1979), but this was not observed. Indeed, the contrary was observed, at least in the anterior cingulate cortex, where the orexin bouton density was lower in the group exposed to a chronic prenatal alcohol regime than the controls. Interestingly, in previous studies of orexinergic cortical projections in Cetartidoactyls, the cetaceans had a lower density of orexinergic boutons in the cerebral cortex despite having greater numbers of smaller neurons than artiodactyls (Dell et al., 2012, 2015). The cetaceans are under pressure to maintain arousal in their aquatic environment and through the manner in which they sleep. Thus, while the differences in the orexinergic system of the mice exposed to chronic prenatal alcohol cannot directly explain the disorders of arousal observed in FASD children, the difference observed do indicate that there may be problems with the arousal system in FASD children.

## Further Studies

In the current study we have observed three specific differences in two distinct neuronal systems involved in the regulation and control of the sleep-wake cycle. These three differences, when viewed in light of previous experimental studies of the sleep-wake cycle, appear to be related to the disorders of sleep experienced in children suffering from FASD. It would appear that prenatal alcohol exposure causes deficits within the arousal systems of both the hypothalamus (orexinergic) and pons (cholinergic), and that these deficits may cause the disorders observed in FASD children. It would be useful in future to undertake polysomnographic recording of sleep in the FASD mouse model and FASD children to determine to what degree the sleep disorders in the FASD children match any potential sleep disorders in the FASD mouse model, as well as examine in detail potential changes to the terminal projection fields of these systems in various regions of the mouse brain. Given the findings of the current study and the information to date for sleep in FASD children, REM sleep would appear to be an interesting target for study. Given the success in treatment of sleep disorders, such as narcolepsy (Didato and Nobili, 2009), further research into the precise mechanism causing sleep disorders in FASD models and children may open avenues for the use of neurotherapeutics to ameliorate and potentially alleviate sleep disorders associated with FASD. In future, such symptoms as night terrors (Durmer and Dinges, 2005), sleep walking (Randazzo et al., 1998) and daytime tiredness (Lancioni et al., 1999) may be treatable in FASD children and thus increase their quality of life significantly by keeping these symptoms under control.

## AUTHOR CONTRIBUTIONS

OIO, AOI and PRM conceptualized the study. OIO, AB and PRM undertook the practical work and analysis. OIO and PRM wrote the first draft of the article and subsequently edited it with the suggestions of AB and AOI.

## REFERENCES


## FUNDING

The research reported herein was funded by the National Research Foundation of South Africa (PRM).

## ACKNOWLEDGMENTS

The authors are grateful for the assistance of the Central Animal Services staff at the University of the Witwatersrand for their assistance in the first phase of this project.


and disturbs sleep-wake state transitions structure. Exp. Neurol. 247, 562–571. doi: 10.1016/j.expneurol.2013.02.007


**Conflict of Interest Statement**: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Olateju, Bhagwandin, Ihunwo and Manger. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Mutation Analysis of Consanguineous Moroccan Patients with Parkinson's Disease Combining Microarray and Gene Panel

*Ahmed Bouhouche1 \*, Christelle Tesson2 , Wafaa Regragui1 , Mounia Rahmani1 , Valérie Drouet2 , Houyam Tibar <sup>1</sup> , Zouhayr Souirti3 , Rafiqua Ben El Haj1 , Naima Bouslam1 , Mohamed Yahyaoui1 , Alexis Brice2 , Ali Benomar1 and Suzanne Lesage2*

*1Research Team in Neurology and Neurogenetics, Faculty of Medicine and Pharmacy, Genomics Center of Human Pathologies, University Mohammed V, Rabat, Morocco, 2Sorbonne Universités, UPMC Université Paris 6 UMR\_S 1127, INSERM U 1127, CNRS UMR 7225, Institut du Cerveau et de la Moelle épinière, ICM, Paris, France, 3Clinical Neurosciences Laboratory, Faculty of Medicine and Pharmacy, Sidi Mohamed Ben Abdellah University, Fez, Morocco*

#### *Edited by:*

*Nilesh Bhailalbhai Patel, University of Nairobi, Kenya*

### *Reviewed by:*

*Thomas M. Durcan, Mcgill University, Canada Gunnar P. H. Dietz, Schwabe Pharma Deutschland, Germany*

#### *\*Correspondence:*

*Ahmed Bouhouche a.bouhouche@um5s.net.ma*

#### *Specialty section:*

*This article was submitted to Neurodegeneration, a section of the journal Frontiers in Neurology*

*Received: 26 July 2017 Accepted: 10 October 2017 Published: 31 October 2017*

#### *Citation:*

*Bouhouche A, Tesson C, Regragui W, Rahmani M, Drouet V, Tibar H, Souirti Z, Ben El Haj R, Bouslam N, Yahyaoui M, Brice A, Benomar A and Lesage S (2017) Mutation Analysis of Consanguineous Moroccan Patients with Parkinson's Disease Combining Microarray and Gene Panel. Front. Neurol. 8:567. doi: 10.3389/fneur.2017.00567*

During the last two decades, 15 different genes have been reported to be responsible for the monogenic form of Parkinson's disease (PD), representing a worldwide frequency of 5–10%. Among them, 10 genes have been associated with autosomal recessive PD, with *PRKN* and *PINK1* being the most frequent. In a cohort of 145 unrelated Moroccan PD patients enrolled since 2013, 19 patients were born from a consanguineous marriage, of which 15 were isolated cases and 4 familial. One patient was homozygous for the common *LRRK2* G2019S mutation and the 18 others who did not carry this mutation were screened for exon rearrangements in the *PRKN* gene using Affymetrix Cytoscan HD microarray. Two patients were determined homozygous for *PRKN* exon-deletions, while another patient presented with compound heterozygous inheritance (3/18, 17%). Two other patients showed a region of homozygosity covering the 1p36.12 locus and were sequenced for the candidate *PINK1* gene, which revealed two homozygous point mutations: the known Q456X mutation in exon 7 and a novel L539F variation in exon 8. The 13 remaining patients were subjected to next-generation sequencing (NGS) that targeted a panel of 22 PD-causing genes and overlapping phenotypes. NGS data showed that two unrelated consanguineous patients with juvenile-onset PD (12 and 13 years) carried the same homozygous stop mutation W258X in the *ATP13A2* gene, possibly resulting from a founder effect; and one patient with late onset (76 years) carried a novel heterozygous frameshift mutation in *SYNJ1*. Clinical analysis showed that patients with the *ATP13A2* mutation developed juvenile-onset PD with a severe phenotype, whereas patients having either *PRKN* or *PINK1* mutations displayed early-onset PD with a relatively mild phenotype. By identifying pathogenic mutations in 45% (8/18) of our consanguineous Moroccan PD series, we demonstrate that the combination of chromosomal microarray analysis and NGS is a powerful approach to pinpoint the genetic bases of autosomal recessive PD, particularly in countries with a high rate of consanguinity.

Keywords: Parkinson's disease, Moroccan patients, consanguinity, chromosomal microarray analysis, nextgeneration sequencing gene panel

#### Bouhouche et al. Genetics of Consanguineous PD Patients

## INTRODUCTION

Parkinson's disease (PD; MIM #168601) is a common neurodegenerative disorder with a prevalence of >1% in populations over 60 years of age (1). PD is clinically characterized by rigidity, bradykinesia, tremor, and postural instability, and may be accompanied with dementia and depression (2, 3). The disease etiology is likely to be multifactorial, involving complex interactions between genetic and environmental factors. In the past 20 years, genetic studies of PD families have provided strong support for the hypothesis that PD has a significant genetic component. To date, 13 genes have been described for hereditary PD (4, 5), and at least 10 of these genes are associated with autosomal recessive (AR) forms of PD. Although mutations in PARK2 (*PRKN*; MIM #602544), PARK6 (*PINK1*; MIM #605909), and PARK7 (*DJ1*, MIM #606324) are infrequent in the PD population, they are responsible for a majority of early-onset PD and are usually known to cause typical PD with indistinguishable clinical signs (1, 4, 6). Moreover, the other, more rare mutations, including PARK9 (*ATP13A2*; MIM #610513), PARK14 (*PLA2G6*; MIM #612953), PARK15 (*FBXO7*; MIM #260300), PARK19 (*DNAJC6*; MIM #615528), PARK20 (*SYNJ1*; MIM #615530), PARK23 (*VPS13C*; MIM #616840), and *PODXL* (MIM #602632) have been implicated in parkinsonism characterized by juvenile onset and atypical clinical signs (5, 7–10).

*PRKN*, located on chromosome 6q26, is one of the largest human genes spanning 1.38 Mb with 12 exons and large intronic regions (11). Mutations in this gene explain up to 50% of AR PD and about 15% of sporadic cases with early onset (12, 13). A large number of mutations have been identified in all populations studied, regardless of ethnic origin, including exon rearrangements and point mutations. *PINK1* contains eight exons that span 1.8 kb and encode a 581-amino-acid serine/threonine kinase protein. This protein exhibits an N-terminal mitochondrial targeting sequence, a putative transmembrane anchor as well as a C-terminal kinase domain (14, 15). It was noticed that the mutational hotspot is found in exon 7. More than 70 mutations have been reported in *PINK1*; two-thirds of which are loss-of-function mutations affecting its kinase domain (16). *DJ1* is composed of eight exons spanning 24 kb and encodes a 189-amino-acid protein (17, 18). The causative PD mutations identified in *DJ1* are mainly point and structural mutations resulting in loss of protein function (17, 19). All *PRKN*, *PINK1*, and *DJ1* mutation carriers present with similar clinical features that are practically indistinguishable from idiopathic forms of PD, including a good response to levodopa with a tendency to develop levodopa-induced dyskinesia and a slow progression.

In addition, mutations in *ATP13A2* cause a juvenile parkinsonism characterized by a rapid progression, supranuclear gaze palsy, pyramidal signs, and dementia (4, 20). Mutations in *PLA2G6* display both early-onset and juvenile forms of recessively inherited atypical parkinsonism. The early-onset form is associated with levodopa-responsive dystonia-parkinsonism, pyramidal signs, cognitive dysfunction, and Lewy body disease accumulation in the brain, whereas the juvenile one is distinguished by dystonia, cerebellar ataxia, spasticity in all limbs, and cognitive decline (21). Mutations in *FBXO7* are also responsible for a juvenile form that presents with early dystonia and pyramidal signs (22). Furthermore, the gene *DNAJC6* encodes for the HSP40 Auxilin protein. Mutations in *DNAJC6* induce an earlyonset parkinsonian-pyramidal phenotype with rapid progression and a poor response to levodopa. Patients with these mutations can also show juvenile parkinsonism with a relatively slow disease progression (7, 23, 24). Also responsible for early-onset PD, *SYNJ1* mutations cause atypical parkinsonism with a number of features, such as dystonia, oculomotor apraxia, dementia, and seizures (25–27). *VPS13C* is a multi-exonic gene (86 exons) responsible for early-onset parkinsonism, and it is associated with a severe phenotype characterized by rapid progression, cognitive deterioration, and a widely distributed presence of Lewy bodies (9, 23). Recently, a homozygous frameshift mutation in the *PODXL* gene was described as a causal factor for juvenile parkinsonism. Although these data are currently insufficient to support this assertion, the clinical features appear similar to those of classic PD (10).

In the present study, we analyzed the genetic bases of a series of consanguineous PD patients from Morocco by combining chromosomal microarray analysis (CMA) and gene panel nextgeneration sequencing (NGS) to target 22 genes associated with PD and overlapping phenotypes.

## PATIENTS AND METHODS

## Patients

From 2013 to 2016, a total of 145 Moroccan PD patients were enrolled at the Movement Disorder Unit of the Department of Neurology (Specialties Hospital, Rabat, Morocco). Clinical Diagnosis of PD was made using the United Kingdom Parkinson's Disease Society Brain Bank criteria (28). Patients were submitted to a structured clinical interview as described previously (29). Nineteen of 145 patients (13%) were born from a consanguineous marriage, of which 15 were isolated cases and 4 were familial cases. Genomic DNA was extracted from peripheral blood leukocytes using Isolate II Genomic DNA kit from Bioline. This study was approved by the Biomedical Research Ethics Committee of the Medical School of Rabat (CERB) and written informed consent was obtained from all subjects in accordance with the Declaration of Helsinki*.*

## Genetic Analysis

To assess the 19 consanguineous PD patients selected for this study, we first sequenced exon 41 of Leucine-rich repeat kinase 2 gene (*LRRK2*; MIM #609007) to screen for the G2019S mutation that was reported to represent 41% of Moroccan PD patients (29).

## Chromosomal Microarray Analysis

DNA samples of patients negative for the G2019S mutation were then screened for exon rearrangements using Affymetrix Cytoscan HD microarray according to the manufacturer's protocol. With a median inter-marker distance of 500–600 bp, CytoScan HD offers the highest physical coverage of the genome for detecting human chromosomal abnormalities. Indeed, these chips include 750,000 single-nucleotide polymorphism (SNP) and 2.6 million copy number variation (CNV) markers that enable high-resolution (25–50 kb resolution) detection of CNVs, region of homozygosity (ROH), uniparental disomy, and low-level mosaicism. Briefly, 250 ng of DNA samples were digested with *Nsp1*, amplified with TITANIUM Taq DNA polymerase (Clontech, Mountain View, CA, USA), fragmented with Affymetrix fragmentation reagent, and labeled with biotin endlabeled nucleotides. DNAs were hybridized to the microarrays for 16 h, washed and stained on the GeneChip Fluidics Station 450, and scanned on the GeneChip Scanner 3000 7G (Affymetrix). Data analysis was performed using Chromosome Analysis Suite software version 1.2.2 (Affymetrix). Data were considered significant only when they met the quality control criteria set by the manufacturer [the Median Absolute Pairwise Difference scores (MAPD) < 0.25, the Waviness Standard Deviation (WSD) < 0.12, and the SNP Quality Control (SNP-QC) > 0.15].

## DNA Sanger Sequencing

DNAs of patients with ROH covering the 1p36.12 locus were sequenced for *PINK1*. All eight of the coding exons and intron– exon boundaries of *PRKN* and *PINK1* were polymerase chain reaction (PCR) amplified and the PCR products were sequenced using Big Dye Terminator Cycle Ready Reaction 3.1 Kits and an ABI 3130xl automated sequencer for the patients and the 96 controls. The collected sequence data were analyzed using SeqScape2.1 software (Applied Biosystems, Foster City, CA, USA).

## Next-Generation Sequencing (NGS) Target Gene Panel and Validation

We designed an NGS-based screening of the 22 currently most prevalent parkinsonism-associated genes (Table S1 in Supplementary Material). The custom Design KAPPA Library Preparation Kit (Roche) was used to capture all exons, intron– exon boundaries, 5′- and 3′-UTR sequences and 10-bp flanking sequences of target genes (RefSeq database, hg19 assembly). Specific probes for NGS target enrichment were designed using NimbleDesign1 software and amplicon length varied between 250 and 500 bp. Runs were performed on Illumina MiSeq sequencer. The assay was performed according to the manufacturer's recommended protocol. Variants were prioritized based on the following criteria: frequencies <0.01% in public databases (ExAC/ GnomAD) and our in-house database of 500 exomes, nucleotide and amino-acid conservation (based on alignments), relation of the gene to disease (per family), and inheritance pattern. All reported variants were confirmed by Sanger sequencing.

## Bioinformatics Analysis of Gene Panel Data

Human reference genome UCSC hg19 was used for sequence alignment and variant calling with the Burrows-Wheeler Aligner (BWA)2 (30) and the Genome Analysis Toolkit (GATK)3 (31). PCR duplicates were removed prior to variant calling using Picard4 software. Variants were annotated with ANNOVAR software (32). The mean coverage was 993× (range 594–1241×), and the mean percent coverage at 30× was 98.7% (range 96.5–99.6%) for all individuals tested. Targeted exons with a coverage less than 30 reads were screened subsequently by Sanger sequencing. Analyses were performed using custom Polyweb software (Paris Descartes, France).

## RESULTS

Of the 145 PD patients recruited during the 4-year period, 19 individuals were born from a consanguineous marriage, which represents an inbreeding rate of 13%. Of them, seven patients were males (37%) and four had a positive family history of PD (21%). The mean age at examination was 56.2 years (range 19–84) and the mean age at onset was 47.4 (range 12–77) years. Sanger screening for the *LRRK2* exon 41 showed the G2019S mutation in the homozygous state for one patient without family history of the disease (**Table 1**, patient ID 3332). The clinical phenotype of this patient, whose disease onset was recorded at 48 years and who had a disease duration of 9 years, is no different from the heterozygous G2019S carriers described previously (29). Because the G2019S mutation is very common in Morocco, this patient will be included in a large series of G2019S carriers to be analyzed for the dopamine metabolism genes in order to determine their effects on age of onset, clinical phenotype and response to treatment.

The remaining 18 patients negative for the *LRRK2* G2019S mutation were subjected to high-resolution CMA using an Affymterix platform and CytoscanHD microarrays, which revealed microdeletion chromosomal region 6q26 of the *PRKN* in 3 of the 18 consanguineous PD patients (16.7%). The deletions were determined to be homozygous in two patients and compound heterozygous in one patient. Data collected from Patient 3020, who carries a compound heterozygous deletion, are shown in **Figure 1**. This male patient (III.3) is a sibling to four other, unaffected children born from a common consanguineous marriage of the first degree (**Figure 1A**). Despite the patient's consanguineous heritage, the allele-difference plot did not show ROH at chromosome 6q (**Figure 1B**). The weighted log2 ratio and copy-number-state plots of chromosomal region 6q26 show two heterozygous *PRKN* deletions of 166 and 587 kb encompassing exons 3–4 and exons 3–7, respectively (**Figure 1C**). The centromeric and telomeric break points of the smallest deletion were located at positions 162.600.881 and 162.767.227, and the break points of the largest deletion were located at positions 162.195.309 and 162.783.147, respectively. Absence of ROH at region 6q26 suggests that the two deletions were inherited from two different ancestors, and thus the patient's father (II.1) and mother (II.2) must carry one of the two mutations in the heterozygous state.

In addition, patient 3158 (III.6), a 49-year-old male, was also born from a consanguineous marriage of the first degree

<sup>1</sup>https://design.nimblegen.com/nimbledesign/app/.

<sup>2</sup>http://bio-bwa.sourceforge.net/.

<sup>3</sup>http://www.acronymfinder.com/Genome-Analysis-Toolkit-(software)-(GATK). html.

<sup>4</sup>http://picard.sourceforge.net/.


Table 1 | Clinical features of the nine consanguineous Moroccan PD patients with gene mutation.

*Hmz, homozygous; Htz, heterozygous; W, woman; M, male.*

(**Figure 2A**). The allele-difference plot shows a 6.5 Mb ROH at chromosome 6q (**Figure 2B**). The weighted log2 ratio and copy-number-state plots of the 6q26 region show a homozygous deletion of 154 kb that encompasses exon 9 of the *PRKN* gene (**Figure 2C**). The centromeric break point of the deletion was located within intron 9 (161.835.769), and the telomeric break point was located in intron 7 (161.990.516).

Data collected from patient 3468, who is also homozygous for a *PRKN* gene deletion, are shown in **Figure 3**. This patient (III.1) was born from a first-degree consanguineous marriage and has a brother known to have PD (**Figure 3A**). The allele-difference plot shows three long, contiguous stretches of homozygosity at chromosome 6q: (1) 12.49 Mb from 6q16.1-q21, (2) 15.79 Mb from 6q22.31-q23.3, and (3) 17.2 Mb from 6q25.2-qter (**Figure 3B**). The weighted log2 ratio and copy-number-state plots of the 6q26 region show a homozygous deletion of 339kb encompassing exons 6 and 7 of the *PRKN* gene (**Figure 3C**). While the centromeric break point of the deletion was located within intron 7 (162.101.291), the telomeric break point was located in intron 5 (162.441.087).

While Patients 3158 and 3468 presented with early-onset PD with mixed phenotypes, patient 3020 presented with juvenileonset disease with a phenotype of tremors and dystonia. The three patients were improved by very low doses of antiparkinsonian drugs without motor fluctuations or dyskinesia, even after up to 35 years of disease duration in Patient 3020 (**Table 1**).

For patients without *PRKN* exon deletions, the high-resolution karyotype was normal and no CNV was identified within the known PD loci. However, because Patients 3528 and 3223 showed ROH covering the 1p36.12 locus, they were also sequenced for the candidate *PINK1* gene. Results revealed two homozygous point mutations: Q456X (c.1366C > T) (**Figures 4A,B**) mutation in exon 7 and novel L539F (c.1617G > C) variation in exon 8 (**Figures 4C,D**). The patient with the Q456X mutation showed an early-onset disease, mixed phenotype without balance impairment after 20 years of disease duration, dyskinesia with very low dose levodopa therapy (400 mg/day), and a mild degree of cognitive impairment. Interestingly, the patient with the L539F mutation manifested parkinsonism at 50 years of age with high doses of levodopa (1000mg/day) after 4 years of disease progression (**Table 1**).

## NGS Panel and Validation Results

From the 13 patients analyzed by NGS, we identified causative mutations affecting three individuals. In patients 3022 and 3868, we identified the same p.W258X (c.774 G > A) stop mutation in *ATP13A2* (**Figure 5**). For individual 3022, a homozygous mutation in *ATP13A2* was suspected, given that a loss of

heterozygosity was detected at this locus on microarray analysis. Patients with the W258X mutation in *ATP13A2* showed a juvenile-onset disease (12 and 13 years), an akinetic-rigid phenotype with shuffling gait and postural instability. Patient 3022 had dystonia and swallowing difficulties in the very onset of the disease. He also displayed motor fluctuations and dyskinesia at very low doses of dopaminergic drugs (400 mg of levodopa equivalent daily dose) after 7 years of disease duration. Patient 3868 presented with some cognitive decline 12 years after disease onset. No other non-motor symptoms are present in both patients (**Table 1**). Finally, we identified in Patient 3467 two heterozygous mutations in *SYNJ1*: a p.S552Ffs\*5 (c.1655delG) frameshift mutation and a p.T1236M (c.3707 C > T) missense variant (**Figure 6**). The missense variant is most likely benign

region of homozygosity.

as it is reported at relatively high frequency in the ExAC database (rs145937537, minor allele frequency in all populations: 0.0015 and in African population: 0.0017) but was not reported in the homozygous state. It is predicted benign by 6 of 7 *in silico* pathogenicity prediction tools (SIFT,5 PolyPhen-2,6

Mutation Taster,7 LRT,8 and FATHMM Radial SVM, LR score).9 This patient was a 76-year-old male born from consanguineous parents (**Figure 6**) with a slowly progressive form of typical PD, and a phenotype of akinetic-rigidity,

8http://www.genetics.wustl.edu/jflab/lrt\_query.html.

5http://sift.jcvi.org/www/SIFT\_enst\_submit.html.

<sup>7</sup>http://www.mutationtaster.org/.

<sup>6</sup>http://genetics.bwh.harvard.edu/pph2/index.shtml.

<sup>9</sup>http://fathmm.biocompute.org.uk/.

tremors, postural instability, sleep disorders, and constipation. He was treated by very low doses of levodopa and displayed no apparent cognitive impairment. His father was diagnosed with parkinsonism at the age of 63 years, and he died 10 years later while still walking independently.

The remaining 10 patients did not have any identifiable mutations in the targeted PD-associated genes. These patients displayed a mean age of onset of 55.6 years [42–77]. Their clinical phenotype ranged between akinetic-rigid to mixed form without dystonia but with postural instability in up to 70% of cases after 5.7 years [1–19] of disease duration, on average. Motor fluctuations are present in 5 of 10 patients, even those with short disease duration of 3 or 4 years, whereas the dyskinesia phenotype is only observed in two patients that each have had PD for more than 10 years. All patients displayed at least two non-motor signs; urinary dysfunction being the most frequent (Table S2 in Supplementary Material).

(B) Allele-difference plot. Arrow indicates CNV at position 6q26. (C) Weighted log2 ratio and copy number state plots. CNV, copy number variation.

## DISCUSSION

Of the *PRKN* mutations reported so far, large deletions are the most frequent and can be present in either homozygous or compound heterozygous states (16, 33, 34). However, due to limitations of the multiplex ligation-dependent probe amplification (MLPA) and quantitative PCR methods used commonly, only few studies have determined the size and break point locations of the rearrangements. We overcame this issue by employing highresolution CMA, which allows for rapid and effective detection of CNV and their break point locations. Through use of Cytoscan HD arrays, the screening of 18 consanguineous, Moroccan PD patients revealed four deletions in three probands. Our results show a rearrangement frequency of 16.7% in consanguineous Moroccan PD patients. The deletions were determined to be homozygous in two patients and compound heterozygous in one patient. All the rearrangements observed in our sample were deletions located between exons 2 and 9, and their break point locations appear to be unique. In patient 3020, although the two deletions were inherited from two individuals of the same, highly consanguineous family, the break point locations of their intron were different. These findings suggest that the deletions were independent and recurrent events, which confirms previous studies reporting that most *PRKN* gene deletions occurred between exons 2 and 8. This region of *PRKN* comprises the FRA6E center and is considered a deletion hotspot because it contains certain microhomology sequences that have been frequently implicated in the main rearrangement process (33, 35, 36).

Moreover, in addition to the detection of CNV, the CMA method enables detection of ROH, which is effective for identifying candidate genes of recessive diseases. Indeed, two patients without *PRKN* CNV showed ROH at the 1p36 locus and had mutations in the *PINK1* gene. These mutations consisted of the

already known Q456X mutation in exon 7 and a new L539F variant in exon 8; the latter of which is located in the C-terminal domain of the *PINK1* protein that we have already shown to be probably pathogenic (37).

Furthermore, we demonstrate the feasibility of NGS as a research and potential diagnostic tool for patients with parkinsonism. For instance, we identified the same homozygous nonsense mutation of p.W258X in *ATP13A2* in two unrelated consanguineous patients, probably resulting from a common founder. This mutation falls into an ROH at chromosome 1p, detected in one patient. We also found two different heterozygous variants in *SYNJ1* in one patient: a novel frameshift mutation and a missense change that is likely benign.

Clinically, our patients with mutations in *PRKN* and *PINK1* presented with a classical PD phenotype with early onset, good response to levodopa, and benign course as previously reported in the literature (12). It should be noted that the patient sharing two compound heterozygous deletions in *PRKN* with a juvenile onset, tremoric phenotype, and dystonia, did not exhibit nonmotor symptoms after 35 years of disease duration. Also, for the patient with the *PINK1* stop mutation, postural instability was not seen after 20 years of disease duration. Otherwise, mutations in *ATP13A2* was reported to cause Kufor–Rakeb syndrome with juvenile onset and atypical clinical features including pyramidal signs, dystonia, cognitive decline, supranuclear gaze palsy (38) and may become unresponsive to levodopa as the disease progresses (39). Our patients who had the same novel R258X mutation had few atypical signs, namely dystonia and cognitive impairment, but one patient also displayed bulbar symptoms at the onset of the disease. These bulbar symptoms can be explained by either bilateral pyramidal signs (pseudo-bulbar syndrome). Furthermore, mutations in *SYNJ1* have been reported previously to cause AR PD with symptom onset occurring after about 30 years of age and with disease progression ranging from severe to stable. Also, it has been observed that *SYNJ1* mutations induce severe dyskinesia with generalized seizures in childhood, dystonia, developmental delay, cognitive impairment, and oculomotor disturbances even when coupled with low-dose levodopa therapy (25, 40). However, we report here a male patient heterozygous for a novel frameshift mutation in *SYNJ1* who displays a late-onset, slowly progressive form of typical PD and responds well to low-dose levodopa therapy without apparent cognitive impairment. Though unavailable for examination, his father also had parkinsonism, who by the patient's accounts followed a similarly benign course. This case is thus consistent with a pseudo-dominant pattern of inheritance or recessive inheritance in compound association with another heterozygous mutation in a gene that not exists in the panel used.

In conclusion, this study provides evidence that supports the power and efficiency of the high-resolution CMA method for uncovering micro-rearrangements in the *PRKN* gene for patients with AR PD. By focusing on patient populations with a high prevalence of consanguinity in Morocco, we also demonstrate the potential of this approach to be used a first-line diagnostic test. Furthermore, the combination of CMA with NGS provides an important understanding of the genetic bases of AR PD.

## ETHICS STATEMENT

This study was approved by the biomedical research ethics committee of the Medical School of Rabat (CERB) and written informed consent was obtained from all subjects in accordance with the Declaration of Helsinki.

## AUTHOR CONTRIBUTIONS

WR, HT, MR, and ZS phenotyped patients. RB and NB performed DNA extraction and banking. RB performed Sanger Sequencing. ABo performed microarray analysis in the Medical School of Rabat. CT and VD performed NGS analyses at the ICM platform. SL and AB supervised the work at the ICM institute. ABo, ABe, and MY supervised the work and obtained funding

## REFERENCES


support. ABo and RB wrote the manuscript. All authors edited the final version of the manuscript.

## ACKNOWLEDGMENTS

We are grateful to the patients for participation in this study, and Dr. Khoussine J. for improving the English language of the manuscript.

## FUNDING

This work was supported by the "Centre National de Recherche Scientifique et Technique" (CNRST) of "Ministère de l'Enseignement Supérieur, de la Recherche Scientifique et de la Formation des Cadres" (MESRSFC) and the Mohammed V University in Rabat (UM5R), Morocco.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at http://www.frontiersin.org/article/10.3389/fneur.2017.00567/ full#supplementary-material.


disease. *Am J Med Genet B Neuropsychiatr Genet* (2005) 133B:120–3. doi:10.1002/ajmg.b.30119


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2017 Bouhouche, Tesson, Regragui, Rahmani, Drouet, Tibar, Souirti, Ben El Haj, Bouslam, Yahyaoui, Brice, Benomar and Lesage. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Deep Brain Stimulation in Moroccan Patients With Parkinson's Disease: The Experience of Neurology Department of Rabat

Mounia Rahmani <sup>1</sup> \*, Maria Benabdeljlil <sup>1</sup> , Fouad Bellakhdar <sup>2</sup> , Mustapha El Alaoui Faris <sup>1</sup> , Mohamed Jiddane<sup>3</sup> , Khalil El Bayad<sup>4</sup> , Fatima Boutbib<sup>1</sup> , Rachid Razine<sup>5</sup> , Rachid Gana<sup>2</sup> , Moulay R. El Hassani <sup>3</sup> , Nizar El Fatemi <sup>2</sup> , Meryem Fikri <sup>3</sup> , Siham Sanhaji <sup>1</sup> , Hennou Tassine<sup>4</sup> , Imane El Alaoui Balrhiti <sup>1</sup> , Souad El Hadri <sup>1</sup> , Najwa Ech-Cherif Kettani <sup>3</sup> , Najia El Abbadi <sup>2</sup> , Mourad Amor <sup>6</sup> , Abdelmjid Moussaoui <sup>6</sup> , Afifa Semlali <sup>7</sup> , Saadia Aidi <sup>1</sup> , El Hachmia Ait Benhaddou<sup>4</sup> , Ali Benomar <sup>4</sup> , Ahmed Bouhouche<sup>4</sup> , Mohamed Yahyaoui <sup>4</sup> , Abdeslam El Khamlichi <sup>8</sup> , Abdessamad El Ouahabi <sup>8</sup> , Rachid El Maaqili <sup>2</sup> , Houyam Tibar <sup>4</sup> , Yasser Arkha<sup>8</sup> , Adyl Melhaoui <sup>8</sup> , Abdelhamid Benazzouz <sup>9</sup> and Wafa Regragui <sup>4</sup>

<sup>1</sup> Research Team in Neurology and Neurogenetics, Department of Neurology A and Neuropsychology, Faculty of Medicine and Pharmacy, Hôpital des Spécialités ONO, University Mohammed V, Rabat, Morocco, <sup>2</sup> Department of Neurosurgery, Faculty of Medicine and Pharmacy, Hôpital Ibn Sina, University Mohammed V, Rabat, Morocco, <sup>3</sup> Department of Neuroradiology, Faculty of Medicine and Pharmacy, Hôpital des Spécialités ONO, University Mohammed V, Rabat, Morocco, <sup>4</sup> Research Team in Neurology and Neurogenetics, Department of Neurology B and Neurogenetics, Faculty of Medicine and Pharmacy, Hôpital des Spécialités ONO, University Mohammed V, Rabat, Morocco, <sup>5</sup> Laboratory of Biostatistics, Clinical Research and Epidemiology, Faculty of Medicine and Pharmacy, University Mohammed V, Rabat, Morocco, <sup>6</sup> Department of Anesthesia and Intensive Care, Faculty of Medicine and Pharmacy, Hôpital des Spécialités ONO, University Mohammed V, Rabat, Morocco, <sup>7</sup> Department of Surgical Intensive Care, Faculty of Medicine and Pharmacy, Hôpital Ibn Sina, University Mohammed V, Rabat, Morocco, <sup>8</sup> Department of Neurosurgery, Faculty of Medicine and Pharmacy, Centre de Rehabilitation et de Neurosciences, Hôpital des Spécialités ONO, University Mohammed V, Rabat, Morocco, <sup>9</sup> Centre National de la Recherche Scientifique, Institut des Maladies Neurodégénératives, Univ. de Bordeaux UMR 5293, Bordeaux, France

Introduction: Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is known as a therapy of choice of advanced Parkinson's disease. The present study aimed to assess the beneficial and side effects of STN DBS in Moroccan Parkinsonian patients.

Material and Methods: Thirty five patients underwent bilateral STN DBS from 2008 to 2016 in the Rabat University Hospital. Patients were assessed preoperatively and followed up for 6 to 12 months using the Unified Parkinson's Disease Rating Scale in four conditions (stimulation OFF and ON and medication OFF and ON), the levodopa-equivalent daily dose (LEDD), dyskinesia and fluctuation scores and PDQ39 scale for quality of life (QOL). Postoperative side effects were also recorded.

Results: The mean age at disease onset was 42.31 ± 7.29 years [28–58] and the mean age at surgery was 54.66 ± 8.51 years [34–70]. The median disease duration was 11.95 ± 4.28 years [5–22]. Sixty-three percentage of patients were male. 11.4% of patients were tremor dominant while 45.71 showed akinetic-rigid form and 42.90 were classified as mixed phenotype. The LEDD before surgery was 1200 mg/day [800-1500]. All patients had motor fluctuations whereas non-motor fluctuations were present in 61.80% of cases. STN DBS decreased the LEDD by 51.72%, as the mean LEDD post-surgery was 450 [188-800]. The UPDRS-III was improved by 52.27%, dyskinesia score by 66.70% and motor fluctuations by 50%, whereas QOL improved by 27.12%. Post-operative side effects were hypophonia (2 cases), infection (3 cases), and pneumocephalus (2 cases).

Edited by:

Nilesh Bhailalbhai Patel, University of Nairobi, Kenya

#### Reviewed by:

Adam Olding Hebb, Colorado Neurological Institute (CNI), United States Mark Janssen, Maastricht University Medical Centre, Netherlands

> \*Correspondence: Mounia Rahmani

mouniarahmani4@gmail.com

#### Specialty section:

This article was submitted to Neurodegeneration, a section of the journal Frontiers in Neurology

Received: 30 September 2017 Accepted: 14 June 2018 Published: 31 July 2018

#### Citation:

Rahmani M, Benabdeljlil M, Bellakhdar F, El Alaoui Faris M, Jiddane M, El Bayad K, Boutbib F, Razine R, Gana R, El Hassani MR, El Fatemi N, Fikri M, Sanhaji S, Tassine H, El Alaoui Balrhiti I, El Hadri S, Ech-cherif Kettani N, El Abbadi N, Amor M, Moussaoui A, Semlali A, Aidi S, Ait Benhaddou EH, Benomar A, Bouhouche A, Yahyaoui M, El Khamlichi A, El Ouahabi A, El Maaqili R, Tibar H, Arkha Y, Melhaoui A, Benazzouz A and Regragui W (2018) Deep Brain Stimulation in Moroccan Patients With Parkinson's Disease: The Experience of Neurology Department of Rabat. Front. Neurol. 9:532. doi: 10.3389/fneur.2018.00532

**283**

Conclusion: Our results showed that STN DBS is an effective treatment in Moroccan Parkinsonian patients leading to a major improvement of the most disabling symptoms (dyskinesia, motor fluctuation) and a better QOL.

Keywords: Parkinson disease, deep brain stimulation, subthalamic nucleus, quality of life, surgical benefit, clinical outcome

## INTRODUCTION

Stereotactic surgery represents a highly effective therapy for the treatment of Parkinson's disease (PD) and other movement disorders refractory to medical treatment. The use of deep brain stimulation (DBS) for PD was driven by advances in the understanding of the pathophysiology and availability of animal models of the disease. In 1993, Benazzouz et al. (1) successfully performed high frequency stimulation of the subthalamic nucleus (STN) in Macaca mulatta monkeys rendered parkinsonian by MPTP (1-methyl-4phenyl-1,2,3,6-tetrahydropyridine).The authors have reported dramatic improvements of the motor symptoms without the development of abnormal involuntary movements. In 1994, Benabid et al. (2) and Siegfried and Lippitz (3) reported successful treatment of patients with PD who underwent DBS of the subthalamic nucleus (STN) and of the globus pallidus internus (GPi), respectively. The procedure is now commonly used in patients with intractable tremor, or with disabling drug-induced complications, especially motor fluctuations and/or dyskinesia. DBS in its current form is a symptomatic treatment that does not interfere with the progression of the disease, and does not affect the non-levodopa responsive motor and non-motor aspects of the disorder such as levodopa-refractory freezing of gait and balance problems nor non-motor aspects of the disease (4).

Nowadays, the STN is widely considered the target of choice (5–8). The mechanism of the stimulation effect on PD is not fully understood but thought to likely be related to the modulation of neuronal activity and the reinstatement of balance within basal ganglia connections (4, 9, 10).The postoperative clinical outcome depends on the quality of the inclusion clinical criteria and the precision of targeting for electrode implantation, which is based on neuroimaging techniques, intraoperative electrophysiology and test of the stimulation effects (11, 12).

Multiple series have reported on the long-term efficacy of DBS for PD (13–23). Here we report our experience of STN DBS performed in a cohort of Moroccan PD patients over a period of 9 years. We describe our results of this first Moroccan series, with a particular emphasis on evaluating the effectiveness and safety of this neurosurgical treatment on the first year of follow up.

## MATERIALS AND METHODS

We conducted a retrospective study of 35 patients with advanced PD who underwent bilateral STN DBS surgery from January 2008 to December 2016 in the University Hospital of Rabat. Surgery was performed in two different departments of neurosurgery, using the same technical procedure. The study was approved by ethics committee of medical school of Rabat and all patients provided their written informed consent.

## Patient's Selection

Evaluations were performed by neurologists specialized in movement disorders. Inclusion criteria for pre-operative assessment were diagnosis of PD according to the UK Brain Bank Criteria (24), age under 70 years old, severe parkinsonian motor symptoms or dyskinesia that limit activities of daily living despite optimal medical therapy for at least 6 months, no dementia or major general illness. A systematic psychiatric appraisal was made to assess and treat any severe depression.

All patients underwent a pre-operative testing. Levodopa challenge was analyzed, and an improvement of at least 30% was necessary to confirm levodopa responsiveness. A morphologic MRI was performed to exclude patients with severe cerebral atrophy, ischemic lesions or other brain injuries that may contra-indicate the surgical procedure. All subjects had also a neuropsychological testing. Global cognitive functioning was evaluated by the Mattis Dementia Rating Scales (MDRS) (25, 26) and the Montreal Cognitive Assessment (MoCA) (26, 27), intellectual capacities by the Progressive Matrices de Raven (PM47) (26, 28), executive functions by the Trail Making Test (TMT), Stroop test and Frontal Assessment Battery (FAB), memory by the verbal fluency and the Memory Impairment Screen (MIS-D), visual and constructive abilities by the Rey figure and the Benton Visual Retention Test (26).We considered a score of MDRS lower than 130/144 a cut off for DBS eligibility as well as severe executive troubles. Patients were eligible for surgery if they responded to levodopa, challenge with normal brain MRI, normal cognitive tests, and no major drug-resistant depression.

#### Pre-operative Evaluation

All patients were assessed using a form that specifies the demographic characteristics (age, gender, age of onset, duration of the disease), clinical features [laterality of symptoms, predominant features: tremulous, akinetic-rigid, or mixed subtypes according to criteria used in 2008 by Rajput et al. (29)], disease severity assessed by the motor section of the UPDRS score (Unified Parkinson's Disease Rating Scale) and Hoehn and Yahr scale (24), motor complications [fluctuations, dyskinesia, freezing of gait (FOG)] and non-motor fluctuations (dysautonomic troubles, sleep disorders, depression, cognitive disturbances, hallucinations, delirium), Levodopa Equivalent Daily Doses (LEDD) calculated based on a previously published algorithm combining dopamine agonist daily dose with levodopa daily dose (30),UPDRS I for mental cognitive assessment, UPDRS II and Schwab and England Scale for activity of daily living (ADL), Giovannoni criteria (31) for dopamine dysregulation syndrome (DDS), Montgomery-Asberg Depression Scale (MADRS) and Hamilton anxiety rating scale (HAM-A) to evaluate mood disorder and PDQ 39 for quality of life (32).

## Surgical Targeting and Procedure

A Leksell G stereotactic frame (Elekta AB) was placed under local anesthesia. We performed a preoperative 1.5 Tesla cerebral MRI with the following sequences: ventriculographic CISS (constructive interference in steady state), MPR (Multi-plan Reconstruction) with gadolinium, and coronal T2 DESS (double echo steady state) followed by a CT scan to check for any MRI-generated distortion. All the images were transferred to the surgical planning station (Elekta Surgiplan<sup>∗</sup> ). The STN coordinates were calculated using direct (based on MRI T2 DESS) and indirect (using statistical coordinates) methods (33, 34).

The first operated side was the one contralateral to the most impaired body-side. The electrodes were implanted in a single operative session under local anesthesia and the target was identified by a combination of neuroimaging, microelectrode recording, and stimulation tests. For each patient, trajectories were determined on the basis of individual anatomical variations and stereotactic MRI based software was used to plan an optimal trajectory from the defined entry point to the sensorimotor STN stimulation target (anatomically referred to as dorso-lateral STN) by avoiding critical brain structures.

The STN stimulation target was defined using a combination of statistical coordinates of STN (4 mm inferior, 3.9 posterior, and 12 mm lateral from midcomissural point), and direct visualization on MRI where the STN was chosen at the anterior margin of red nucleus and 2 to 3 mm lateral from its external border.

Stereotactic gadolinium enhanced T1-weighted images were used to visualize vessels to avoid injury of any vascular structure during surgery.

All trajectories were anterior to the motor strip close to the coronal suture on sagittal plane, and about 2 to 3 cm from midline in coronal plane. The final trajectory was defined as to get a maximum of definitive electrode plots within the visualized hypointensity of STN. To reach this, the trajectory direction should superimpose to the vertical axis of the STN.

Multi-track (3–5) microelectrodes were inserted for electrophysiological mapping of the STN. Subsequent macrostimulations were used to assess the efficacy and side effect profile of the tested electrodes. The optimal track (best micro-recording and widen therapeutic window on macro-stimulation) was chosen for each side and the permanent quadripolar leads were implanted (model 3389; Medtronic, Minneapolis, Minn., USA). Continuous fluoroscopy was used to monitor a potential electrode displacement and to confirm the definitive electrode positioning. Non-absorbable silk stich was used to anchor the definitive electrode and the burr hole was sealed by acrylic cement. The mean surgery duration was 5–7 h from the scalp incision.

Post-operative CT scan was performed immediately after surgery to rule out surgical complications such as hemorrhage and to confirm the final location of the implanted electrodes based on fusion of the preoperative MRI and postoperative CT scan images. The internal pulse generators (model 7428 Kinetra, or 37601 Activa PC; Medtronic) were implanted the same day or 3–5 days later in a subcutaneous pocket in the infra-clavicular region under general anesthesia.

## Stimulation Programming

The 8 lead contacts were assessed 1–3 months after surgery when the lesion-like effect responsible of spontaneous postoperative improvement of PD symptoms, has disappeared. The best contact that improved the symptoms without side effects was chosen for each side.

## Post-operative Evaluation

Patients were evaluated 6–12 months after surgery using section III of the UPDRS in four conditions (stimulation OFF and ON and medication OFF and ON). Subscales of UPDRS III were assessed as follow: speech score expresses item 18, tremor is the sum of items 20, and 21, rigidity is item 22, bradykinesia is the sum of items 23 to 27, 30, and 31, posture is item 28 and postural instability item 29. Patients were also assessed for the LEDD, dyskinesia score (the sum of items 32–35 of the UPDRS IV), motor fluctuations score (the sum of items 36-39 of the UPDRS IV), MADRS, HAM-A and PDQ39 scale expressed as summary index (SI) that ranges between 0 and 100% (100% is equivalent to bad quality of life) and as its eight dimensions (35). Postoperative side effects were also recorded.

## Statistical Analysis

SPSS 13.0 software was used for the statistical processing of our data. Quantitative data were expressed in mean ± standard deviation (SD) or median and interquartile range. Categorical variables were expressed as numbers and percentages. Data were tested for normal distribution by graphical methods. Pre and post intervention quantitative variables of normal distribution were compared using paired-t test (UPDRS III, Dyskinesia scores, Motor Fluctuations scores, PDQ39 SI, Bradykinesia subscale of UPDRS III) and pre and post intervention quantitative variables of non normal distribution (UPDRS I, UPDRS II, LEDD, PDQ 39 subscales, speech, tremor, rigidity, posture, and postural instability subscales of UPDRS III) and ordinal variables (UPDRS V and VI) were analyzed using Wilcoxon test. Pre and post intervention categorical data were compared using McNemar test. Statistical significance was assumed for tests yielding pvalues of less than 0.05.

## RESULTS

## Demographic Data

The mean age at disease onset was 42.31 ± 7.29 years and the mean age at surgery was 54.66 ± 8.51 years. The median disease duration was 11.95 ± 4.28 years. Sixty-three percentage of patients were male. Four patients were tremor-dominant, 16 patients showed an akinetic rigid form and 15 patients were classified as mixed subtype. The LEDD before surgery was 1200 mg/day [800-1500]. Mean time to appearance of motor fluctuations was 7.81±4.25 years [1–16], non-motor fluctuations 8.05±3.34 years [2–15] and dyskinesia 7.82±3.76 years [2– 14]. The most disabling symptom was akinetic state (40%)



\*Percentage (number), \*\*Means and standard deviation [minimum, maximum], \*\*\*Median and interquartile range, M, males; DDS, dopamine dysregulation syndrome; LEDD, levodopa equivalent daily doses.

followed by akinesia and dyskinesia in the same time (34.21%) then dyskinesia alone (7%). Only one patient suffered from disabling tremor associated with dyskinesia. More than one third of patients presented a dopamine dysregulation syndrome (**Table 1**).

## Surgery Related Complications

Some side effects were recorded during or immediately after surgery. They were mild and transient: confusional episode (2 cases), hallucination (1 case), hypophonia (1 case), aphonia (1 case), anxiety (2 cases), hypomania (1 case), and pneumocephalus (2 cases).

#### **DBS Related Complications**

Long term complications were as follow: two patients suffered from dysarthria that needed changes of stimulation parameters, three patients presented direct hardware-related complications; two patients exhibited a battery site infection 3 and 5 months after surgery that was resolved after antibiotic treatment for one patient and after removal of the battery and the lead for the other one. Another subject experienced, 6 months after surgery, an infection of the lead/wire connection site. The whole DBS system was removed then repositioned 8 months later. One patient had unilateral misplacement of the lead with a medial deviation of 3.2 mm. Following a reimplantation of the lead, the patient presented clinical improvement and the CT scan showed correct positioning of the lead.

## Changes in Motor Outcome

Comparison between pre-and postoperative clinical state is summarized in **Table 2**. There is no change in the part II of the UPDRS following surgery except for a tendency to a small improvement in activity of daily living during OFF medication state up to 10% but without reaching significance (p = 0.06).

There is a significant improvement of the UPDRS III especially when patients were in OFF state with a rate of 52.25%. Motor fluctuation scores were improved by 50% and dyskinesia score by more than 66%. There was no more OFF FOG after surgery. The England and Schwab score in OFF medication was also improved by 36.66%. LEDD was decreased up to 51.72%.

UPDRS III subscales analysis in the four conditions, medication OFF and ON with and without stimulation, showed that tremor, rigidity, and bradykinesia scores were significantly improved by STN DBS in both medication OFF and ON states. Posture was improved only in medication OFF state and there was no modification in speech and postural instability by stimulation. The whole UPDRS III score was improved by 50% or more in medication OFF and ON states (**Table 3**).

## Changes in Quality of Life and Non-motor Outcome

There was no modification in UPDRS part I, depression, anxiety scores or in the prevalence of the DDS. The quality of life (PDQ-39 SI) was improved by 27%. We also assessed the subscales of PDQ-39 score. The only improved dimensions were mobility (p = 0.004), ADL (p = 0.003), and Sigma (p = 0.038). The others, mainly emotional well-being (p = 0.166), social support (p = 0.806), cognition (p = 0.954), communication (p = 0.747), and bodily discomfort (p = 0.281) were not improved.

## Post-operative DBS Setting

As shown in **Table 4**, contact 3 (37.35%) was chosen most often for permanent stimulation followed by contact 2 (34.94%) and contact 1 (18.07%). Contact 0 was rarely selected (9.63%). Contact 0 refers to the most ventral contact and contact 3 to the most dorsal one. The mean proportion of patients who needed bipolar stimulation was 11.42% whereas 12.85% needed two active contacts. The stimulation parameters (mean ± SD) were 2.82 ± 0.57 V, 163.29 ± 35.79 Hz and 65.67 ± 11.91 µs (2 patients had 60 µs at one side and 90 µs at the other).

## DISCUSSION

STN-DBS efficacy on PD motor symptoms is well documented in the short and medium terms, up to 5 years [(13)–(23, 36, 37)], while a few publications with a small number of examined patients addressed the long-term efficacy of this procedure (38– 41). Here, we report the one-year outcome of a cohort of 35 consecutive PD patients at advanced stage of the disease, who underwent bilateral STN-DBS.

The age at surgery of 54 years was in line with a Canadian series (38) but much lower than larger series where patients TABLE 2 | Comparison of pre and post-operative clinical state (stimulation ON).


Data are expressed as means ± standard deviation or median and interquartile range [25–75%], or percentage (number), MF, motor fluctuation; FOG, freezing of gate; LEDD, levodopa equivalent daily doses; DDS, dopamine dysregulation syndrome; PDQ 39SI, PDQuestionnaire-39 summary index; MADRS, Montgomery and Asberg Depression Rating Scale; HAM-A, Hamilton anxiety scale. Bold values refer to significant p value.

were operated between 58-61 years of age (36, 37, 42). This can be explained by the young age of onsetalready reported in our population with an age of onset < 55 years in 45% and < 45 years in 15% (43). The disease duration of 11.97 ± 4.28 years was similar to other series [15 16, 22, 23, 36–39, 42, 44]. The usefulness of DBS in early stage of the disease is currently a subject of debate. Results of the EarlyStim Trial (45) demonstrated that DBS was superior to medical therapy with respect to motor disability, activities of daily living, levodopainduced motor complications and time with good mobility and


TABLE 3 | Effects of STN DBS on UPDRS III subscales.

Data are expressed as median and interquartile range [25–75%] and means ± standard deviation. STIM, stimulation; Paired t-test was used to compare the pre and post intervention means of UPDRS III, Bradykinesia subscale of UPDRS III. Wilcoxon test was used for the other subscores. Bold values refer to significant p value. Bold values refer to significant p value.


Data are given as numbers (%). thirteen electrodes were employed for double monopolar contact stimulation and eight electrodes for bipolar stimulation. Contact 0 refers to the most ventral contact and contact 3 the most dorsal one. STN, subthalamic nucleus.

no dyskinesia. However, the poor access of our population to DBS surgery makes the question quite obsolete, as we are in the obligation to offer this therapeutic option to really disabled patients.

When compared to baseline, STN DBS in our patients allowed a major benefit in different components of motor function as widely noted in different series (46). It improved the UPDRS III score and the cardinal symptoms both in OFF and ON medication conditions. These results attest two points: (i) the superiority of DBS in relieving these symptoms as patients had better scores in ON medication/ON stimulation than in ON medication/OFF stimulation and (ii) the best motor state in the morning before taking their first dose of dopaminergic drugs (OFF medication/ON stimulation). Moreover, effective contacts were the most dorsal ones (contacts 2 and 3). This result may be explained by the phenotype of our patients (11.40% tremor dominant and 42.71 mixed forms) requiring current diffusion to zona incerta to relieve severe tremor (47).

In addition, STN DBS is equivalent to dopamine effect on the posture by means of rigidity relief. As expected, surgery did not ameliorate both postural instability and dysarthria. Indeed, the effect of stimulation on axial symptoms is known to be poor, their pathophysiology being different (46, 48–51). Moreover, patients can even exhibit a slight deterioration in axial symptoms, which is associated with the progression and the natural evolution of the disease (38–41, 46). An older age, intensity of axial symptoms and UDPRS II off-medication score (items 5–17) before surgery were predictive factors of dysarthria/hypophonia and postural instability after surgery (44). In our patients, the young age at surgery may explain in part the absence of such side effects. Otherwise, there was no more medication OFF FOG postoperatively confirming the benefit of DBS on dopa-sensitive symptoms.

Subsequently, we observed a considerable reduction in the daily doses of antiparkinsonian medication, up to 50% of the preoperative doses, which participated in the antidyskinetic effect of DBS. Indeed, we recorded a major reduction in dyskinesia by 66.70% and in the frequency and severity of motor fluctuations by 50%, both are known to significantly contribute to preoperative functional limitations (46) and considered by our patients as the most disabling symptoms. These results are in line with most series that reported a reduction of medication doses of 19 to 80.7%, motor fluctuation scores of 16 to 95% and dyskinesia scores of 53–92% (16, 22, 37, 42, 52–60).

There was a limited advantage of surgery on ADL assessed by UPDRS II in OFF medication state with just a trend to improvement and no effect in ON state for our patients. In several series, ADL in On medication condition had not improved or even declined at 1 year and remained stable at 5 years despite reductions in dyskinesia duration and severity, whereas in OFF medication, ADL improved by 49 to 54.2% 5 years after surgery (16, 36, 37, 42, 61, 62). However, in our series, when ADL was measured using the Schwab and England scale, we observed a significant improvement of ADL in OFF medication by 36.66%, which was in line with most series (16, 36, 37).

The good effect of STN DBS was also attested by an overall improvement of quality of life by 27% in our patients assessed by the PDQ-39 SI. This rate is in contrast with the dramatic improvement of the motor function especially the most disabling symptoms (dyskinesia and motor fluctuations) but is in agreement with previous studies. Indeed, the improvement of QOL reported varied from 30.2 to 50.6% (63, 64). Dimensions affected by DBS are subject of conflict. Some authors report that DBS ameliorates all dimensions of QOL, whereas others emphasize that the dimensions improved are those that surgery is expected to affect (ADL and mobility) but not the others (social support, cognition, and communication) (23, 64–66). Sobstyl et al. (66) demonstrated a correlation between dyskinesia and the improvement of "Mobility" and "ADL" dimensions and PDQ39 SI. This correlation can explain our results, as the significantly improved dimensions were "Mobility," "ADL," and "Stigma." The improvement of "Stigma" dimension may result from the impact of dyskinesia on social life especially in our country where hyperkinetic movements are culturally not appreciated. The disappearance of dyskinesia allowed patients to be involved in social life. Over all, further studies on a large number of patients and long follow up are needed to determine the impact of DBS on QOL taking into account both motor and non-motor symptoms. The moderate improvement of QOL by STN DBS reported up to now highlights the major influence of nonmotor symptoms on quality of life (67, 68).

Various studies found no effect of STN DBS on cognitive functions while others noted worsening of verbal fluency or transient cognitive impairments (56, 69–72). In our series, we did not observe any change in cognitive and mood scales. Patients with preexisting cognitive impairment were not selected for DBS. On the other hand, all subjects were screened for depression and psychiatric disorders in order to avoid the reported potential exacerbation of mood disorders after surgery (34, 71–74).

DDS is one of the clinical aspects of Impulse Control Behaviors (ICB). It has a prevalence of 13.6% in PD and may be considered as the neuropsychiatric equivalent of levodopainduced dyskinesia (75–77). Contrasting results of DBS on ICB are reported in the literature (78–83). Merola et al. (84) in their study of 150 consecutive PD STN-DBS-treated patients, reported only an overall trend for reduction of ICB but with significant improvement in hypersexuality, gambling and DDS after a follow up of 4.3± 2.1 years. In our series, there was no modification in the prevalence of DDS after surgery. Patients were assessed 6 to 12 months after surgery, which could be considered insufficient to appreciate the modification of their ICB. Nevertheless, new ICB may occur in some subjects with risk factors such as: younger age, female, lower dyskinesia improvement and schizoid traits of personality disorders (84). A longer follow up is needed to assess our patients for new ICB.

STN DBS can be regarded as a safe procedure in properly selected patients. Mortality and permanent morbidity are very low and surgical complications are relatively rare. However, numerous surgical, hardware-related, or infective complications may be developed after surgery or during the follow-up period, sometimes even years after the intervention for lead positioning (85). The rates of these complications are quite variable in the literature and include intracranial hemorrhage (0–10%), stroke (0–2%), infection (0–15%), lead erosion without infection (1– 2.5%), lead fracture (0–15%), lead migration (0–19%), and death (0–4.4%) (13, 85–89). In our series, we did not record any cerebral hemorrhage. Major complications included infections and hardware-related ones, occurring in 8.6% of cases.

Psychiatric disorders were seen during surgery in 6 patients. This difficulty to complete the surgical procedure has been recently reported by other groups as a factor that can be time wasting and frustrating for both the patient and the surgeon, observed mainly in early series. The most frequent cause is a psychiatric disturbance of the patient, with hallucinations and impossibility to cooperate during surgery (85, 90–94).

## CONCLUSION

Our results showed that STN DBS is an effective treatment in Moroccan Parkinsonian patients leading to a major improvement of the most disabling symptoms (dyskinesia, motor fluctuation) and a better QOL. These findings, which are in line with those previously reported in other caucasian and asian population, showed that in carefully selected Moroccan patients with a multidisciplinary management, STN DBS is a powerful treatment that alleviates the burden of advanced PD.

## AUTHOR CONTRIBUTIONS

MR and WR participated equally to the design of the study, edited the manuscript, participated in patients' selection, peroperative microrecording, and macrostimulation testing and programming. FoB, YA, and AdM performed lead targeting and surgery. RG, NEF, and NEA: participated in surgery. MJ, MoE, MF, and NEK performed acquisition, analysis, and interpretation of radiological data. FaB and SS performed neuropsychological assessment. AbM, MA, and AS performed anesthetic monitoring. MB participated in patient's selection, peroperative neurophysiology, and macrostimulation testing. MEF participated in patients' selection and gave agreement for clinical data. SE, HeT, IEB, and HoT participated in collecting data. KE performed and edited statistical analysis. RR supervised statistical analysis. AbB participated in per-operative neurophysiology and revising the manuscript. AhB and EAB participated in critical reading of the manuscript. SA, AlB, and MY gave agreement for clinical data. AEK, AEO, FoB, and RE gave agreement for surgical data.

## ACKNOWLEDGMENTS

We are very grateful to Janardan Vaidynathan, PhD, for his support in the surgical procedure and electrophysiology recording. We are also thankful to Dr. Kaswati Janane for his help on statistics.

## REFERENCES


dyskinesias to impulse control disorders. Lancet Neurol. (2009) 8:1140–49. doi: 10.1016/S1474-4422(09)70287-X


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer MJ declared a past co-authorship with one of the authors AbB to the handling Editor.

Copyright © 2018 Rahmani, Benabdeljlil, Bellakhdar, El Alaoui Faris, Jiddane, El Bayad, Boutbib, Razine, Gana, El Hassani, El Fatemi, Fikri, Sanhaji, Tassine, El Alaoui Balrhiti, El Hadri, Ech-cherif Kettani, El Abbadi, Amor, Moussaoui, Semlali, Aidi, Ait Benhaddou, Benomar, Bouhouche, Yahyaoui, El Khamlichi, El Ouahabi, El Maaqili, Tibar, Arkha, Melhaoui, Benazzouz and Regragui. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Both Reaction Time and Accuracy Measures of Intraindividual Variability Predict Cognitive Performance in Alzheimer's Disease

Björn U. Christ <sup>1</sup> \*, Marc I. Combrinck <sup>2</sup> and Kevin G. F. Thomas <sup>1</sup>

<sup>1</sup> Applied Cognitive Science and Experimental Neuropsychology Team Laboratory, Department of Psychology, University of Cape Town, Cape Town, South Africa, <sup>2</sup> Division of Geriatric Medicine, Groote Schuur Hospital, Department of Medicine, University of Cape Town, Cape Town, South Africa

Dementia researchers around the world prioritize the urgent need for sensitive measurement tools that can detect cognitive and functional change at the earliest stages of Alzheimer's disease (AD). Sensitive indicators of underlying neural pathology assist in the early detection of cognitive change and are thus important for the evaluation of early-intervention clinical trials. One method that may be particularly well-suited to help achieve this goal involves the quantification of intraindividual variability (IIV) in cognitive performance. The current study aimed to directly compare two methods of estimating IIV (fluctuations in accuracy-based scores vs. those in latency-based scores) to predict cognitive performance in AD. Specifically, we directly compared the relative sensitivity of reaction time (RT)—and accuracy-based estimates of IIV to cognitive compromise. The novelty of the present study, however, centered on the patients we tested [a group of patients with Alzheimer's disease (AD)] and the outcome measures we used (a measure of general cognitive function and a measure of episodic memory function). Hence, we compared intraindividual standard deviations (iSDs) from two RT tasks and three accuracy-based memory tasks in patients with possible or probable Alzheimer's dementia (n = 23) and matched healthy controls (n = 25). The main analyses modeled the relative contributions of RT vs. accuracy-based measures of IIV toward the prediction of performance on measures of (a) overall cognitive functioning, and (b) episodic memory functioning. Results indicated that RT-based IIV measures are superior predictors of neurocognitive impairment (as indexed by overall cognitive and memory performance) than accuracy-based IIV measures, even after adjusting for the timescale of measurement. However, one accuracy-based IIV measure (derived from a recognition memory test) also differentiated patients with AD from controls, and significantly predicted episodic memory performance. The findings suggest that both RT- and accuracy-based IIV measures may be useful indicators of underlying neuropathology. The present study therefore contributes toward an understanding of the relative utility of RT- and accuracy-based IIV measures in detecting neurocognitive impairment in older adults, and also advances the empirical evaluation of sensitive markers of cognitive change in patients with AD.

#### Edited by:

Nouria Lakhdar-Ghazal, Faculty of Science, Mohammed V University, Morocco

#### Reviewed by:

Rahul Goel, University of Houston, United States Rufus Olusola Akinyemi, University of Ibadan, Nigeria

> \*Correspondence: Björn U. Christ budoch@gmail.com

Received: 30 September 2017 Accepted: 13 March 2018 Published: 09 April 2018

#### Citation:

Christ BU, Combrinck MI and Thomas KGF (2018) Both Reaction Time and Accuracy Measures of Intraindividual Variability Predict Cognitive Performance in Alzheimer's Disease. Front. Hum. Neurosci. 12:124. doi: 10.3389/fnhum.2018.00124

Keywords: Alzheimer's disease, accuracy, cognition, episodic memory, intraindividual variability, reaction time

## INTRODUCTION

Twenty-three percent of the worldwide burden of disease occurs in individuals age 60 years and older, and up to 63% of individuals with age-related diseases such as dementia currently reside in low- and middle-income countries (LAMICs; Prince et al., 2015; World Health Organisation, 2015). In one such country, South Africa, the most recent census statistics indicate that 8% of the population (∼4.1 million individuals) are aged 60 years or older, and that that number will increase by as much as 40% over the next two decades (Statistics South Africa, 2014). Furthermore, community-based estimates suggest there is a higher prevalence of dementia in South Africa, and in LAMICs generally, compared to global estimates (Prince et al., 2013; de Jager et al., 2015). These epidemiological data underscore the urgency of conducting LAMIC-based dementia research.

Recently, the Alzheimer's Association's Research Roundtable (AARR), an interdisciplinary group of leading dementia researchers, prioritized the urgent need for sensitive measurement tools that can detect cognitive and functional change at the earliest (even prodromal) stages of Alzheimer's disease (AD; Snyder et al., 2014). Sensitive indicators of underlying neural pathology are important for the evaluation of early-intervention clinical trials, and may play a central role in alleviating the burden of age-related disease (Food and Drug Administration, 2013). One method that may be particularly well-suited to help achieve this goal involves the quantification of intraindividual variability (IIV; also known as inconsistency) in cognitive performance. Whereas conventional indicators of cognitive performance are based on measures of central tendency and involve assessment of an individual on a single measure administered on a single occasion, IIV indicators are based on measures of variability and involve assessing fluctuations in performance of an individual on a single measure administered on multiple occasions (Li et al., 2001; Hultsch et al., 2002; MacDonald et al., 2006).

Contemporary IIV research focuses primarily on inconsistency in performance on reaction time (RT) measures (see, e.g., Bielak et al., 2010; Saville et al., 2011; Bunce et al., 2013; Yao et al., 2016). Such latency-based measures are particularly well-suited to IIV research because they have larger ranges than traditional cognitive test scores, thus making them more sensitive than traditional cognitive tests to individual performance differences. RT tasks also (a) typically involve multiple trials, which allows for many samples of performance, and (b) are less sensitive to re-test effects (Allaire and Marsiske, 2005; Salthouse, 2012). Over the past two decades, a sizeable literature has established IIV in RT as an effective marker of general cognitive function in older adults: High levels predict impending cognitive decline, and are associated with a range of age-related neurological disturbances, with neurodegenerative disease (e.g., AD), and with mortality risk (Collins and Long, 1996; Hultsch et al., 2002; MacDonald et al., 2003; Burton et al., 2006; Shipley et al., 2006; Duchek et al., 2009; Bielak et al., 2010).

An alternative method for capturing IIV involves using accuracy-based measures. These measures are derived from tasks featuring stimuli to which the test taker makes either a correct or an incorrect response (e.g., Murphy et al., 2007; Tractenberg and Pietrzak, 2011). Because such tasks are used frequently in clinical practice, deriving an IIV score from them, and showing the predictive value of that score, is a useful undertaking. Although some studies report that accuracy-based IIV measures can, for instance, differentiate between patients with AD, those with Parkinson's disease, and healthy controls, and can aid in detecting prodromal AD (Darby et al., 2002; Burton et al., 2006; Murphy et al., 2007; Tractenberg and Pietrzak, 2011; Kälin et al., 2014), many researchers prefer latency-based measures. One reason for this preference is that statistically significant positive associations between accuracy-based IIV and age do not survive after controlling for mean performance. Associations of outcomes (e.g., age, clinical group status) with RT-based IIV measures are not affected by controlling for mean performance, and hence those measures are perceived to be superior in detecting underlying pathology (see, e.g., Li et al., 2001; Salthouse et al., 2006).

However, only one previous study in the aging literature provides a direct comparison of the relative sensitivity of RT- and accuracy-based IIV measures to cognitive compromise. Hultsch et al. (2000) measured trial-to-trial and session-to-session IIV in RT- and accuracy-based measures in three groups: healthy older adults, patients with arthritis, and patients with dementia (either mild AD or mild vascular dementia). They reported that, whereas there were no significant between-group differences in terms of accuracy-based IIV, RT-based measures differentiated the groups successfully, independent of mean-level predictors.

The current study seeks to systematically replicate and extend the findings of Hultsch et al. (2000). Specifically, we also compare directly the relative sensitivity of RT- and accuracybased estimates of IIV to cognitive status. The novelty of the present study, however, centers on the patients we tested and the measures we used. Where Hultsch et al. (2000) used a mixed-dementia group, we use a group of patients with AD. Inclusion of this more homogenous clinical group allows for improved sensitivity of accuracy-based tasks, which are typically designed to target specific domains of cognitive function (e.g., episodic memory). Furthermore, where Hultsch and colleagues' analyses were targeted toward categorical prediction of group membership (i.e., they asked whether RT- and accuracy-based IIV measures could distinguish healthy older adults from patients with arthritis and from patients with dementia), our analyses use more variable outcome measures (i.e., we ask not only whether RT- and accuracy-based IIV are significantly different in healthy older adults compared to patients with AD, but also whether those IIV measures are predictive of performance on a measure of general cognitive function and on a measure of performance in a cognitive domain that, typically, is sensitive to AD dysfunction). In summary, the specific aims of our analyses were to (a) use RT- and accuracy-based measures of IIV to differentiate between a clinical group of AD patients and a control group of demographically matched healthy individuals, (b) determine the relative contribution of RT- and accuracybased measures of IIV to the prediction of overall cognitive functioning and episodic memory functioning, and (c) evaluate the effect of the timescale of measurement on that relative

contribution of RT- and accuracy-based measures of IIV to the prediction of overall cognitive functioning and episodic memory functioning.

Hence, the present study contributes toward an understanding of the relative utility of RT- and accuracy-based IIV measures in detecting neurocognitive impairment in older adults, and also responds to the AARR call for empirical evaluation of sensitive markers of cognitive change in patients with AD.

## MATERIALS AND METHODS

## Design and Setting

The current study is the first report of data collected within an ongoing longitudinal investigation of AD progression taking place in Cape Town, South Africa. The parent study utilizes a measurement burst design (Nesselroade, 1991), in which each participant experiences three intervals of serial testing (or bursts; T1, T2, and T3) over the course of 12 months. Within each interval, each participant is tested three times (e.g., T1.1, T1.2, T1.3) over a 2-week period. The data we report here are from the first test interval (i.e., T1).

## Participants

All participants (N = 48; 34 women) were over the age of 55 years (M = 71.25, SD = 7.12). The sample consisted of a cognitively healthy control group (n = 25; 19 women) and a mildto-moderate stage possible or probable AD clinical group (n = 23; 15 women), with diagnosis following NINCDS-ADRDA criteria (McKhann et al., 1984).

Clinical participants were recruited from a state hospital's Memory Clinic. Recruitment was monitored by health professionals, including a neurologist (MIC) and a neuropsychologist (KGFT), who provide clinical service delivery at the Clinic. Control participants were community-dwelling volunteers from the greater Cape Town area. They received notice of the study via word-of-mouth or flyers distributed to seniors' clubs, old age homes, and retirement villages.

Inclusion criteria were (a) availability of medical health history; (b) age 55 years or above; (c) English literacy (i.e., basic ability to speak, read, and write in that language); and (d) availability of a close relative or similar who could provide information about recent changes in cognitive function. Exclusion criteria included (a) a diagnosis of HIV/AIDS, uncontrolled hypertension, uncontrolled diabetes mellitus, or any other medical condition that, in the opinion of the research team, might have a long-lasting effect on cognitive function; (b) current or present psychiatric illness; (c) a Geriatric Depression Scale (GDS; Yesavage et al., 1982) score > 9/30; (d) the presence of any major neurological disorder (e.g., Parkinson's disease, Huntington's disease) or past stroke; (e) any history of alcohol or drug abuse, or heavy smoking (> 20 cigarettes per day); and (f) Mini-Mental State Examination (MMSE; Folstein et al., 1975) score < 12.

The Research Ethics Committees of the University of Cape Town's Department of Psychology and Faculty of Health Sciences approved all study procedures. These procedures adhered to the guidelines published in the Declaration of Helsinki (World Medical Association, 2013).

## Measures and Procedures

The data we report on here were gathered across four sessions (one screening and three test sessions). All study participants signed consent forms before screening. Moreover, all the clinical participants were informed about the study and consent was signed in the presence of a guardian/caregiver/relative (who also signed the consent form). The screening session occurred no more than 30 days before the first test session (in most cases, there was a week or less of separation). For all participants, the three test sessions took place over a 2-week period.

Sessions were held in a private research room at Groote Schuur Hospital or at the participant's home, depending on his/her preference and travel capabilities. All tests were administered by BUC, or by a graduate student trained and supervised by him.

### Screening Session

This session included administration of (a) a detailed clinical interview that gathered information about biographical, medical, and psychiatric history, (b) the GDS, (c) the MMSE, and (d) the Cambridge Examination for Mental Disorders of the Elderly-Revised edition (CAMCOG-R; Huppert et al., 1995). The latter was developed as a cognitive screening measure for the early diagnosis of dementia in the elderly (Leeds et al., 2001). It consists of 67 items and measures cognitive performance within eight domains (orientation, language, memory, attention, praxis, calculation, abstract thinking, and perception). We used a version adapted for use with South African samples (James et al., 2014).

At the conclusion of the session, participants were invited back for repeated administration of a 10-test cognitive battery (see Supplemental Material for the full list of tests). Below, we describe the five tests for which data are reported.

#### Test Sessions

The order of test administration was varied for each session to prevent order effects (see Supplemental Material for the different test orders). All three test sessions were otherwise identical to one another. Each lasted ∼2 h.

#### **Reaction Time Tasks**

These tasks are part of the Cambridge Neuropsychological Test Automated Battery (CANTAB; Fray et al., 1996). All CANTAB tests are administered on a touch-screen computer. On the simple reaction time (SRT) task, a yellow dot appears inside a circle placed at the center of the computer screen. Participants are required to release a press pad and touch the dot as quickly as possible after its onset. On the choice reaction time (CRT) task, the yellow dot appears inside one of five circles located on the screen. Both tests include a 10-trial practice phase that precedes the test phase. Participants are required to obtain 90% accuracy on the practice trials before proceeding to the test phase. Those who fail to achieve this criterion are presented with a second practice phase. Thereafter, they proceed to the test phase regardless. The test phase for both the SRT and the CRT tasks consists of 30 trials. A single block of trials (e.g., 30 SRT trials) takes ∼5 min to complete.

We administered two SRT blocks and two CRT blocks in each session. Hence, after three test sessions and six blocks of administration we had collected data from 180 trials of SRT performance and 180 trials of CRT performance for each participant.

### **Accuracy-based Tasks**

We used two subtests from the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS), a short screening battery for identifying and characterizing dementia in the elderly (Randolph et al., 1998). These subtests measure immediate and delayed episodic memory, a prominent domain of dysfunction in the cognitive profile of AD (Traykov et al., 2007). There are four parallel forms for each of the RBANS memory subtests, making them appropriate for repeat assessments and allowing for the tracking of cognitive decline within neurodegenerative processes (Randolph et al., 1998).

On the RBANS List Learning subtest, the participant is read a list of 10 words, and is instructed immediately thereafter to recall as many as possible. This process is repeated four times. After a 25–35 min delay, the participant is asked to recall the list, and immediately thereafter is administered a recognition task (i.e., to identify which words from a group of 20 (10 targets and 10 foils) were present on the original list). On the RBANS Story Memory subtest, the participant is read a brief story, and is instructed immediately thereafter to recall as many elements of the story as possible. This process is repeated twice. After a 25–35 min delay, the participant is asked to recall the story.

Although we administered a different form of the List Learning and Story Memory subtests at each test session, the order of administration was the same for each participant (i.e., all participants received List A and Story A at the first test session, List B and Story B at the second session, and so on).

## Statistical Analyses Data Preparation

#### **RT Tasks: Filtering data**

We examined the RT data for outliers because unusually fast or slow responses may reflect spurious performance (e.g., temporary distraction, interruption, or fast guesses). Following convention (see, e.g., Hultsch et al., 2000; Bielak et al., 2010; Garrett et al., 2012), we removed scores that were either (a) below a lower limit for authentic responses at 150 ms, or (b) above an upper limit of 3 SD above the group RT mean for each block of testing. Missing data were then imputed for the outlier trials using a regression-based multiple imputations method (Lachaud and Renaud, 2011). This method of filtering the data is thought to offer conservative estimates of performance variability (e.g., Hultsch et al., 2002).

#### **RBANS and CAMCOG-R tasks: deriving variables**

We derived three scores from the RBANS subtests. The List Learning score is the sum of the number of words recalled correctly across the four learning trials (range 0–40). The List Recognition score is the total number of correctly identified words on the recognition trial (range 0–20). The Story Memory score is the sum of the number of items recalled correctly across the two learning trials (range 0–24).

We derived two scores derived from the CAMCOG-R: Total Score (assessing general cognitive function; range = 0–105), and the recent memory and learning subscale composite score (assessing episodic memory function; range = 0–21). We chose to use the latter because (a) episodic memory dysfunction is a key feature of AD (Peña-Casanova et al., 2012), and (b) the composite score is relatively resistant to the influence of education (James et al., 2014). This is an important consideration given that almost half of the variance in CAMCOG-R scores is accounted for by the effects of age and education (Pereiro et al., 2015).

#### **Extracting intraindividual variability**

Computing IIV scores requires an initial purification of systematic effects in the data that are explained by mean performance scores. Specifically, although one might calculate the intraindividual standard deviation (iSD), calculating raw SDs may introduce systematic effects associated with mean RT because slower mean RTs are strongly associated with higher SDs, and vice-versa (Hale et al., 1988; Hultsch et al., 2008). Therefore, before computing iSDs it is important to partial out any factors (e.g., group and time-on-task effects such as practice and fatigue) that may influence mean RT performance.

To determine which factors significantly influenced the means of RT- and accuracy-based variables, and to thus extract iSDs, we ran a random intercept model on the sample data for each of the SRT, CRT, List Learning, List Recognition, and Story Memory variables, and then added two sets of main effects, the first [featuring test order, blocks, trials (or sessions for the accuracy-based tasks)] to evaluate the impact of time-on-task effects, and the second (featuring group status, sex, monthly household income, age, and level of education) to evaluate the impact of group effects.

#### Inferential Statistical Analyses

We conducted all inferential analyses using SPSS (version 24), with α set at 0.05.

The first part of the analysis involved analyzing betweengroup differences in demographic, cognitive, and affective variables. We used independent-samples t-tests for parametric data, chi-squared tests of contingency for categorical data, Mann-Whitney U tests for non-parametric data, and when the assumption of homogeneity of variance for the Mann-Whitney U tests was not upheld we used independent-samples t-tests and bootstrapped 1,000 replicates using bias corrected (BCa) confidence intervals. To estimate effect sizes, we used Cohen's d, phi (φ), and r for t-tests, chi-squared tests, and Mann-Whitney U tests, respectively. We interpreted these effect sizes following Cohen's (1988) guidelines: for Cohen's d, 0.2 = small, 0.5 = moderate, 0.8 = large; for φ and r, 0.1 = small, 0.3 = moderate, 0.5 = large.

The second part of the analysis set the stage for subsequent regression modeling by examining bivariate associations (using Pearson's r correlation coefficient) between each candidate predictor (i.e., the iSD for each of the SRT, CRT, List Learning, List Recognition, Story Memory variables, and the mean for each of those variables) and each cognitive outcome variable (i.e., CAMCOG-R Total Score and CAMCOG-R Memory Composite).

The final part of the analysis involved creation of a series of sequential multiple regression models that sought to determine the relative contribution of RT- and accuracy-based IIV measures to the prediction of (a) overall cognitive functioning, and (b) episodic memory functioning.

## RESULTS

## Sample Characteristics

The groups were well matched in terms of age, sex distribution, monthly household income, and current depressive symptomatology, but there were significant between-group differences in terms of education, with participants in the control group having completed more years of formal schooling (see **Table 1**). As expected, the analyses also detected significant between-group differences (associated with large effect sizes) on the two CAMCOG-R outcome measures, with the control group scoring better in each case.

## Primary Analyses

#### Extraction of iSDs

We followed the extraction approach described by Hultsch et al. (2008). Random intercept models identified the following fixed effects that contributed significantly to mean performance on each of the candidate predictors: for SRT, blocks, group status, and sex significantly predicted trial-to-trial performance; for CRT, test order, blocks, group status, and sex significantly predicted trial-to-trial performance; for List Learning, group status significantly predicted session-to-session performance; for List Recognition, group status, and session significantly predicted session-to-session recognition performance; and for Story Memory, task session, group status, sex, and education level significantly predicted session-to-session scores. (See Supplemental Material for the full set of results.)

Next, we entered, for each candidate predictor, the significant fixed effects and all their higher-order interactions into a random coefficient model with random slopes on trials (or sessions for the accuracy-based variables) in order to partial out time-on-task and group effects. Finally, we captured the residuals, converted them to T-scores, and calculated the SD across the T-score values to compute the iSDs.

### Between-Group Differences: Predictor Variables

On both RT-based measures, iSD scores for controls were, on average, significantly lower than those for patients. In contrast, the same significant between-group difference was only present for one of the accuracy-based measures (List Recognition; see **Table 2**).

Regarding mean-level performance variables, control participants achieved significantly faster reaction times on the CANTAB tasks, and performed significantly more accurately on the RBANS subtests, than patients (see **Table 2**).

#### Regression Modeling

#### **Bivariate associations between predictor and outcome variables**

Among iSD scores, those for SRT, CRT, and List Recognition showed significant moderate-to-large negative associations with both outcome variables. Among mean scores, each predictor variable was significantly associated, with moderate-to-large magnitude and in the expected direction, with each outcome variable (see **Table 3**). Based on this set of findings, we excluded List Learning and Story Memory iSD scores from subsequent analyses.

#### **Sequential regression models**

We regressed a set of demographic variables (age, sex, education, income), after controlling for group status, on each

TABLE 1 | Descriptive statistics and between-group differences: sample demographic, affective, and cognitive characteristics (N = 48).


For the variables Age, Education, Income, GDS, MMSE, CAMCOG-R Total Score, and CAMCOG-R Memory Composite the second and third columns present means with standard deviations in parentheses. AD, Alzheimer's disease; ESE, effect size estimates (for t, Cohen's d, for chi square, ϕ, and for Mann-Whitney, Cohen's r); GDS, Geriatric Depression Scale; MMSE, Mini-Mental State Examination; CAMCOG-R, Cambridge Cognitive Examination for Mental Disorders of the Elderly-Revised. <sup>a</sup>Highest level of education attained. <sup>b</sup>Monthly household income, in South African Rands (ZAR). At the time of the study, the US\$:ZAR exchange rate was 1:13.53. \*p < 0.05, \*\*\*p < 0.001. All p-values are two-tailed.


TABLE 2 | Descriptive statistics and between-group differences: predictor variables (N = 48).

Data presented are means, with standard deviations in parentheses. AD, Alzheimer's disease; ESE, effect size estimates (for t, Cohen's d and for Mann-Whitney, Cohen's r), iSD, intraindividual standard deviation; SRT, simple reaction time; CRT, choice reaction time. <sup>a</sup>BCa 95% CI [1.26 to 4.04], p = 0.002; <sup>b</sup>BCa 95% CI [−5.73 to −3.50], p = 0.001. \*p < 0.05. \*\*p < 0.01. \*\*\* p <0.001. All p-values are two-tailed.

TABLE 3 | Bivariate correlations: predictor and outcome variables (N = 48).


CAMCOG–R, Cambridge Cognitive Examination for Mental Disorders of the Elderly-Revised. iSD, intraindividual standard deviation; SRT, simple reaction time; CRT, choice reaction time. \*\*p < 0.01. All p-values are one-tailed.

of the two CAMCOG-R outcome variables to determine which demographic factors were significant predictors of, respectively, overall cognitive functioning and episodic memory functioning. For CAMCOG-R Total Score, significant predictors were group (β = −0.72, t = −8.38, p < 0.001), sex (β = −0.27, t = −3.13, p < 0.01), and education (β = 0.27, t = 2.76, p < 0.01). For CAMCOG-R Memory composite, significant predictors were group (β = −0.80, t = −9.10, p < 0.001) and sex (β = −0.24, t = −2.77, p < 0.01).

Then, we created a set of models that described how trialto-trial variability on RT tasks, relative to session-to-session variability on accuracy-based tasks, predicted (a) CAMCOG-R Total Score, and (b) CAMCOG-R Memory Composite score. For each model, we entered the significant demographic factors identified above at the first step, iSD RT- and accuracy-based predictors at the second, and mean-based predictors at the third (see **Table 4**). The purpose of taking this third modeling step was to determine if the significant iSD predictors identified at Step 2 would continue to make a unique contribution toward prediction of the outcome variable after controlling for (a) means of iSD predictors entered at Step 2, and (b) means of the List Learning and Story Memory scores (entered because they are widely-used mean-level predictors of episodic memory performance in the clinical setting).

The most notable results at the second step were these: After controlling for demographic variables, iSDs for List Recognition and SRT contributed significantly to the prediction of CAMCOG-R Total Score [Model 1: 1R <sup>2</sup> = 0.04, F(2, 42) = 4.23, p = 0.02, Cohen's f = 0.20], and CAMCOG-R Memory composite score [Model 3: 1R <sup>2</sup> = 0.08, F(2, 43) = 8.08, p < 0.01, Cohen's f = 0.37]. iSDs for List Recognition and CRT contributed significantly to the prediction of Memory composite score [Model 4: 1R <sup>2</sup> = 0.07, F(2, 43) = 7.31, p <0.01, Cohen's f = 0.34]. List Recognition iSD score only predicted one outcome variable (viz., Memory Composite) significantly when it was entered together with SRT (Model 3). Regarding this latter result, we compared the slopes of the SRT and List Recognition iSD scores by computing a z-score of the residual difference between their unstandardized slopes. The relative contribution of the SRT iSD (b = −0.61, SE = 0.17) to the prediction of Memory Composite was significantly larger than that of the List Recognition iSD (b = −0.15, SE = 0.07), z = −2.50, p = 0.01.

The most notable results at the third step, given the aims of the model, were these: SRT iSD scores continued to contribute significantly to the prediction of CAMCOG-R Total Score (Model 1) and CAMCOG-R Memory Composite


TABLE 4 | Regression models: trial-to-trial reaction time IIV compared to accuracy-based IIV in the prediction of CAMCOG scores (N = 48).

Data presented are β (standard regression coefficient) values. CAMCOG-R, Cambridge Cognitive Examination for Mental Disorders of the Elderly-Revised. SRT, simple reaction time; CRT, choice reaction time. \*p < 0.05, \*\*p < 0.01, \*\*\*p < 0.001. All p-values are one-tailed.

score (Model 3), and CRT iSD score continued to contributed significantly to the prediction of CAMCOG-R Memory Composite score (Model 4).

Finally, to examine the influence of measurement timescale on the relative contribution of the different IIV measures to the prediction of cognitive performance, we created a set of models that described how session-to-session variability on RTand accuracy-based tasks predicted (a) CAMCOG-R Total Score, and (b) CAMCOG-R Memory Composite score. For each model, we entered the same demographic factors as in the previous models at the first step, iSD RT- and accuracy-based predictors at the second, and mean-based predictors at the third (see **Table 5**).

The most notable result at the second step was that, for the RT data, the magnitude of variability decreased markedly from that observed in the trial-to-trial models. For instance, although at Step 2 of the modeling procedure SRT iSD was a significant predictor of CAMCOG-R Total Score [Model 1: 1R <sup>2</sup> = 0.04, F(2, 42) = 4.27, p = 0.02, Cohen's f = 0.20] and of CAMCOG-R Memory Composite score [Model 3: 1R 2 change = 0.04, F(2, 43) = 3.99, p = 0.03, Cohen's f = 0.18], CRT iSD was not a significant predictor of either outcome. Again, List Recognition iSD score only predicted Memory Composite score significantly when it was entered together with SRT (Model 3). This time, the relative contributions of SRT iSD (b = −0.16, SE = 0.07) and List Recognition iSD (b = −0.18, SE = 0.08) to the prediction of Memory composite were not significantly different, z = −0.22, p = 0.59.

The most notable results at the third step were, again, that the predictive power of the RT iSD scores decreased markedly from that observed in the trial-to-trial models. Here, the only significant finding, given the aims of the model, was that SRT iSD scores continued to contribute significantly to the prediction of CAMCOG-R Total Score (Model 1).

## DISCUSSION

This study provided a direct comparison of the relative sensitivity of reaction time- and accuracy-based estimates of intraindividual variability to cognitive compromise. We systematically replicated findings presented by Hultsch et al. (2000), showing that (a) RT-based measures of IIV differentiated a dementia group from a group of healthy older adults, (b) increasing the timescale of measurement (i.e., measuring on a session-to-session rather than a trial-to-trial basis) reduced the sensitivity of RT-based TABLE 5 | Regression Models: session-to-session reaction time IIV compared to accuracy-based IIV in the prediction of CAMCOG scores (N = 48).


Data presented are β (standard regression coefficient) values. CAMCOG-R, Cambridge Cognitive Examination for Mental Disorders of the Elderly-Revised; SRT, simple reaction time; CRT, choice reaction time. \*p < 0.05, \*\*p < 0.01, \*\*\*p < 0.001. All p-values are one-tailed.

IIV, and (c) generally, RT-based IIV was a better predictor of cognitive status than accuracy-based IIV, even after adjusting for timescale of measurement. We extended upon previous findings by showing that accuracy-based IIV (a) could also differentiate patients with AD from healthy older adults, (b) correlated significantly with overall cognitive function and episodic memory performance in both patients and controls, and (c) was a significant predictor of episodic memory performance, even after controlling for sex and group status (AD patient vs. control).

Of the accuracy-based IIV measures that formed part of our investigation, only RBANS List Recognition was sensitive to between-group differences, correlated with CAMCOG-R Total Score and Memory Composite score, and predicted performance on the CAMCOG-R Memory Composite variable after controlling for sex and group status. Although Hultsch et al. (2000) also used measures of recognition memory to derive accuracy-based IIV, they found them to have no significant value in distinguishing dementia patients from controls. We argue that this cross-study difference is attributable to sample characteristics: Whereas we used a homogeneous group of patients with AD, Hultsch and colleagues used a heterogeneous clinical group [i.e., some of their patients had been diagnosed with vascular dementia (VaD) and others with AD]. Patients with VaD perform significantly better than those with AD on recognition memory tasks (Tierney et al., 2001; Román et al., 2002). Hence, including both VaD and AD patients in a single clinical group is likely to diminish the sensitivity of a recognitionbased measure to neurological compromise.

We suggest, therefore, that accuracy-based IIV measures are useful in detecting neurocognitive impairment, but that there must be a careful match between the type of task from which the IIV measure is derived and the purportedly compromised cognitive domain. In other words, accuracy-based IIV measures have less utility when they are considered as indicators of diffuse cognitive or neurological dysfunction: They are best used as indicators of a specific type of cognitive impairment linked to a specifically damaged neuroanatomical site or system. Murphy et al. (2007) demonstrated this point empirically. They administered parallel forms of a list-learning task eight times over 4 days to young (M = 23.4 years) and older (M = 73.3 years) adults. The groups were differentiated by accuracy-based IIV scores derived from tasks assessing frontal lobe function (e.g., false memory tests), but not by those derived from tasks assessing medial temporal lobe (MTL) function (e.g., learning, delayed recall). The authors proposed that age-related changes in the integrity of the frontal lobes (changes not typically present in the MTL) explained this finding. These specific structural changes made it much more likely that there would be increased variability in the performance of the older adults relative to the younger counterparts on the frontal tasks, but not the MTL tasks. Another minor empirical demonstration of this regional specificity consideration is that, among the patient group in the present study, the largest magnitude of variability we observed was on the List Recognition task (see **Table 2**).

Whereas List Recognition iSD scores differentiated between patients and controls, and were significantly associated with scores on the outcome measures, no such relationships were observed for the List Learning and Story Memory iSDs. Given that performance on all three tests requires participation from neural networks that are centered on the MTL and that are compromised by AD pathology (Traykov et al., 2007; Peña-Casanova et al., 2012), this result is unexpected: In this context, IIV on the three tasks should have been similar.

One possible reason for this unexpected result relates to the differing nature of the processing demands made by the List Learning, Story Memory, and List Recognition tasks. Although all three tasks require the participant to retrieve previouslyencoded information, the former two make heavier demands on cognitive resources because they are free recall, and not aided-recall, tasks. In other words, they present no cues to assist retrieval of the learned information, and therefore require more self-generated strategic processing (Moscovitch and Winocur, 1992; Dickerson et al., 2007). Tasks with greater strategic processing demands typically produce higher degrees of score variability, particularly when performance is measured across several learning trials (Allaire and Marsiske, 2005), as was the case with both List Learning and Story Memory. Hence, performance on those subtests may be more vulnerable than List Recognition performance to adaptive variability (Li et al., 2004). Because the presence of adaptive variability tends to increase IIV, a confound within the current design is that one of the three memory tasks we used to measure IIV featured lower strategic processing demands than the other two.

Nonetheless, there is clinical value in the finding that an accuracy-based measure of inconsistency can significantly predict episodic memory performance. In the clinic, accuracybased assessment is far more prevalent than latency-based assessment, and the List Recognition task we used here is a standard element of many clinical neuropsychological test batteries. Of note here, however, is that a current trend in IIV studies that use accuracy-based measures is to move away from operationalizing variability as inconsistency across time and toward dispersion across tasks (within and across cognitive domains) or across items within a test of global cognition (e.g., CAMCOG-R). Findings from IIV studies using this latter operationalization indicate successful prediction of cognitive decline and clinical dementia status above and beyond mean-level performance (Tractenberg and Pietrzak, 2011; Kälin et al., 2014). Accuracy-based measures of dispersion may be more practical than RT measures of IIV for clinicians as the test from which they are derived are already used frequently within standard neuropsychological test batteries, and they avoid the need for multiple trials of administration (Kälin et al., 2014).

## LIMITATIONS

Two limitations of the study's RT-based measures might have reduced their sensitivity to impairment on the cognitive outcomes we sought to measure. The first involves how engaging the RT tasks were for participants. We observed that, for healthy controls, bivariate associations between (a) RT means and CAMCOG-R Total Scores, and (b) variability scores and CAMCOG-R Total Scores were in the opposite direction from what might have been expected (see Supplemental Material). That is, participants with higher CAMCOG-R Total Scores showed slower and more variable performance on both the SRT and the CRT tasks. One explanation is that the repetitive nature of the serial assessments, combined with the relative ease of the RT tasks, may have resulted in a lack of engagement among higher-functioning individuals. This speculation is consistent with research indicating that a lack of task engagement (e.g., due to boredom) during prolonged repetitive tasks may reduce mean RT performance and increase RT variability (Pan et al., 1994; Langner et al., 2010; Garrett et al., 2012; Wang et al., 2014).

A second, and related, limitation involves the regional specificity of the RT measures. Performance on the kinds of SRT and CRT tasks used here activates a complex combination of cognitive control processes (including visual encoding, motor preparation, response selection, and execution), with common neural substrates located largely in the frontal lobes (MacDonald et al., 2000; Lo and Andrews, 2015). As noted above, accuracybased IIV measures are most useful when the task from which they are derived taps into functioning of the area purportedly compromised in the samples under scrutiny. Such regional specificity considerations may also apply to RT-based IIV measures (MacDonald et al., 2008). Following this line of argument, an RT measure better suited to the purposes of the current study may have been one derived from tasks sensitive to episodic memory function [e.g., the recognition latencies from the list and story tasks used by Hultsch et al. (2000)].

Hence, future research in the field might consider adapting RT tests to make them more engaging, and to ensure that they meet considerations related to regional specificity. Using latency scores from tasks that are typically used to produce accuracy-based outcomes may, in fact, also improve task engagement because participants typically find such tasks more challenging than basic RT tasks (Allaire and Marsiske, 1999).

## SUMMARY AND CONCLUSION

We set out to systematically replicate and extend important previous findings regarding the use of intraindividual variability measures in the detection of neurodegenerative disease (Hultsch et al., 2000). Our replication was successful: Results indicated that RT-based IIV measures are superior predictors of cognitive compromise than accuracy-based IIV measures, even after adjusting for timescale of measurement. Our extension was also successful: Results indicated that, by using a homogeneous clinical sample (i.e., early-to-mid-stage Alzheimer's disease patients) and measuring overall cognitive function as well as a performance within a targeted cognitive domain, accuracy-based IIV measures may be useful indicators of underlying pathology.

The present study therefore contributes toward understanding the relative utility of RT- and accuracy-based IIV measures in detecting neurocognitive impairment in older adults, and also responds to the AARR call for empirical evaluation of sensitive markers of cognitive change in patients with AD.

## ETHICS STATEMENT

This study was carried out in accordance with the recommendations of the University of Cape Town Research Ethics Code for Research Involving Human Participants with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the Research Ethics Committees of the University of Cape Town's Department of Psychology and the Faculty of Health Sciences.

## AUTHOR CONTRIBUTIONS

BC: contributed to the conception and design of the study, participant recruitment and acquisition of data, data analysis and

## REFERENCES


interpretation, drafting of the manuscript, and critical revisions of the manuscript for important intellectual content; KT: contributed to the conception and design of the study and critical revisions of the manuscript for important intellectual content; MC: contributed to participant recruitment and acquisition of data; All three authors (BC, KT, and MC) approved of the final version of the manuscript to be submitted for publication.

## FUNDING

Albertina and Walter Sisulu Institute of Ageing Grant Oppenheimer Memorial Trust Award (OMT Ref. 20865/01) Groote Schuur Hospital Neurology Postgraduate Scholarship. National Research Foundation Innovation Doctoral Scholarship (Grant UID: 83307). University of Cape Town Doctoral Research Scholarship.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fnhum. 2018.00124/full#supplementary-material


marker for prodromal Alzheimer's disease. Front. Aging Neurosci. 6:147. doi: 10.3389/fnagi.2014.00147


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Christ, Combrinck and Thomas. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Houyam Tibar1,2,3,4\*, Khalil El Bayad2 , Ahmed Bouhouche1,2, El Hachmia Ait Ben Haddou1,2, Ali Benomar 1,2, Mohamed Yahyaoui1,2, Abdelhamid Benazzouz 3,4† and Wafa Regragui1,2†*

*1Research Team in Neurology and Neurogenetics, Faculty of Medicine and Pharmacy, Genomics Center of Human Pathologies, University Mohammed V, Rabat, Morocco, 2Department of Neurology and Neurogenetics, Hôpital des spécialités de Rabat, Rabat, Morocco, 3University de Bordeaux, Institut des maladies neurodégénératives, UMR 5293, Bordeaux, France, 4CNRS, Institut des maladies neurodégénératives, UMR 5293, Bordeaux, France*

#### *Edited by:*

*Nilesh Bhailalbhai Patel, University of Nairobi, Kenya*

#### *Reviewed by:*

*Gianfranco Spalletta, Fondazione Santa Lucia (IRCCS), Italy Rodolfo Gabriel Gatto, University of Illinois at Chicago, United States*

#### *\*Correspondence: Houyam Tibar*

*tibarhouyam@gmail.com*

*† These authors have contributed equally to this work.*

#### *Specialty section:*

*This article was submitted to Neurodegeneration, a section of the journal Frontiers in Neurology*

*Received: 23 October 2017 Accepted: 06 March 2018 Published: 04 April 2018*

#### *Citation:*

*Tibar H, El Bayad K, Bouhouche A, Ait Ben Haddou EH, Benomar A, Yahyaoui M, Benazzouz A and Regragui W (2018) Non-Motor Symptoms of Parkinson's Disease and Their Impact on Quality of Life in a Cohort of Moroccan Patients. Front. Neurol. 9:170. doi: 10.3389/fneur.2018.00170*

Background: Non-motor symptoms (NMSs) are a real burden in Parkinson's disease (PD). They may appear in early pre-symptomatic stage as well as throughout the disease course. However, their relationship with the deterioration of the patient's quality of life (QoL) is still under debate. This study aimed to investigate the prevalence of NMSs and their impact on the QoL in a cohort of Moroccan patients.

Methods: We carried out a cross-transactional study, where a total of 117 patients were submitted to a structured clinical interview and examination investigating motor and NMSs based on common and conventional scales. Motor symptoms were assessed by the UPDRS I–VI during ON condition. The NMSs were evaluated with common scales and their relationship with the QoL was investigated.

results: The mean patient's age was 60.77 ± 11.36 years old, and the median disease duration was 6 years [2.5–9.5]. Motor's phenotype subtypes were the mixed form in 40.2% of patients, akinetic-rigid in 20.5% and a tremor-dominant form in 39.3%. The median Hoehn and Yahr staging was 2 [1–2.5]. Regarding NMSs, the most common were urinary dysfunctions (82.6%), sleep (80.6%), and gastrointestinal (80%) disorders. Other autonomic dysfunctions were also frequent: thermoregulatory dysfunctions 58.6%, cardiovascular troubles 50.9%, and sexual dysfunctions 47.9%. Depression was present in 47.9% and fatigue symptoms in 23.1%. The median score of SCOPA-AUT was 14 [7.75–21.80]. The median PD questionnaire 39-score index (PDQ39-SI) was 23.22% and the most affected dimension was "mobility." Univariate and multivariate analyses showed that the SCOPA-AUT score impacted the QoL (*p* = 0.001), especially the gastrointestinal (*p* = 0.007), and cardiovascular (*p* = 0.049) dimensions.

conclusion: Our data show that all patients have presented at least one NMS. Autonomic and sleep disorders were the most frequent, and in contrast to other studies, digestive and cardiovascular disorders were rather the factors influencing negatively the QoL of patients. Understanding the pathophysiology of these NMSs should be placed at the forefront in order to develop new therapeutic approaches by improving the QoL of PD patients.

Keywords: Parkinson's disease, motor symptoms, non-motor symptoms, quality of life, Moroccan patients

#### Tibar et al. Non-Motor Symptoms in Parkinson's Disease

## INTRODUCTION

Parkinson's disease (PD) is one of the most prevalent neurodegenerative diseases and the number of patients is increasing. It is reported to be the second most common neurodegenerative pathology of the central nervous system (1). Its prevalence is estimated between 1 and 2 per 1,000 in unselected populations (2). The prevalence of the disease can be variable with age, as it's affecting 1% of the general population above 60 years (3) and about 4% in highest age (4), and it is still a rare disease before the age of 50 (1, 5). PD is a complex neurological pathology characterized essentially by motor signs, including rest tremor, muscle rigidity, akinesia, and postural instability (6, 7). However, diverse non-motor symptoms (NMSs), such as sleep disorders, psychiatric disorders, autonomic disabilities, and sensory disturbances are also present in PD (8). These NMSs may contribute to the impairment of patient's quality of life (QoL).

From a pathophysiological point of view, the PD motor symptoms are attributed to the degeneration of the dopaminergic nigrostriatal system (7). Nevertheless, increasing evidences have shown that PD is a multisystem disorder characterized also by the degeneration of the meso-cortical dopaminergic system, the noradrenergic system of the locus coeruleus, the serotonergic system of the dorsal raphe nuclei, and the cholinergic system of the nucleus basalis of Meynert, in addition to the histaminergic, peptidergic, and olfactory-related systems (8, 9).

During the past decades, understanding the pathophysiology of the motor symptoms was at the origin of the development of pharmacological and surgical therapies allowing their successful management (10–13). However, the knowledge of the NMSs pathophysiology is still limited. Recently, the importance given to the NMSs is increasing and it is now recognized that they may precede motor symptoms and may be useful in identifying potential patients to develop a neurodegenerative disease, such as PD (14, 15). Interestingly, clinical studies suggested that olfactive deficits, rapid eye movement sleep behavior disorder (RBD), fatigue, and depression may constitute important markers of preclinical stages of PD (16, 17).

From several reports, it appears that NMSs are estimated to occur at least in one-third of PD patients (18–20). However, their prevalence and relationship with the QoL in Moroccan patients are still unknown. This study aimed first to evaluate the prevalence of the NMSs in a cohort of 117 Moroccan parkinsonian patients and second to determine the impact of NMSs on their QoL.

## MATERIALS AND METHODS

## Subjects

A cohort of 117 patients was gathered from different regions of Morocco in the department of Neurology and Neurogenetics in Ibn Sina University Hospital of Rabat (Hôpital des Spécialités de Rabat). All patients included in the study had a confirmed diagnosis of PD as per the UK Brain Bank Clinical Diagnostic criteria (21) and provided a written informed consent to take part in our study. The study was approved by the ethics committee of medical school of Rabat (CERB). Collected data included demographic characteristics, the medical history, the disease duration, presence of similar cases in the family, patients' profession, and the socio-economic levels. Regarding the cognitive profile of our patients, individuals with a mini mental state examination (MMSE) score less than 21 were excluded to avoid interferences of cognitive impairment in NMS evaluation. The Moroccan version of the MMSE we use, considers the cut-off for dementia at 21 for illiterate patients or with low level of education. We divided the studied population into two groups: a group with low level of education [level 1–2 of international standard classification of education (ISCED)] and middle to higher level of education (22). We used some specific exclusion criteria, such as: co-morbidity which might limit walking ability (e.g., inflammatory arthropathy); serious hearing deficit, other neurological problems (stroke, inflammatory diseases), and acute medical problems (e.g., cardiopathy) that can also impact QoL.

## Clinical Evaluation of Motor Symptoms

The motor stage of PD was evaluated according to the UPDRS scale (I–VI) during ON condition. UPDRS I assess mentation, behavior, and mood (range 0–16); UPDRS II evaluates Activities of daily living (ADL), including speech, swallowing, handwriting, dressing, hygiene, falling, salivating, turning in bed, walking, and cutting food (range 0–52). UPDRS III is a score for motor examination (range 0–108). Each item of those scales scored on a scale from 0 to 4. UPDRS IV assesses the treatment's complications in the week preceding the examination (items 32–35 for dyskinsia, items 36–39 for motor fluctuations, and items 40–42 for digestive, sleep, and autonomic complications). UPDRS V is the Hoehn and Yahr's staging of severity of PD (staging 1–5), and UPDRS VI is Schwab and England' to assess independency on activity of daily living on OFF and on ON conditions (we only assessed this scale during ON condition) (range 0–100% and higher scores refer to better functional independency).

The clinical type at onset of the disease was classified as tremor dominant, akinetic-rigid, or mixed form according to criteria used by Rajput et al. (23). Tremor dominant subtype referred to patients in whom the tremor was the dominant feature compared to bradykinesia and rigidity. Patients with prominent bradykinesia and rigidity with no visible tremor were classified as akinetic-rigid subtype and those who had comparable severity of bradykinesia, rigidity, and tremor were classified as mixed subtype. The first side and limb affected at the onset of the disease were also recorded. Levodopa equivalent daily dose (LEDD) was calculated based on Tomlinson et al. recommendations (24).

## Clinical Evaluation of Non-Motor Symptoms

The following symptoms were assessed: hyposmia, neuropsychiatric disorders (depression, anxiety, suicidal thoughts), autonomic dysfunctions (constipation and urinary dysfunctions), and sleep disturbances [excessive daytime sleepiness (EDS) and the quality of sleep]. A clinical interview was conducted to determine the presence of each NMS at the time of the examination. Informations on the current use of medications, such as laxatives, hypnotics, or antidepressants to treat some of these NMS, were also collected.

Smell loss was assessed by the validated Argentina Hyposmia Rating Scale (AHRS) (25) starting by asking the subjects whether they noted a change in their ability to smell. Patients with factors that could impair odor identification, such as: current smokers, medical history of nasal surgery (to correct a deviated septum or other plastic surgeries), allergic rhinitis, and traumatic nasal injuries were eliminated. According to these criteria, our sample for this test was restricted to 51 patients (23 males and 28 females). Hyposmia was considered to be present if the AHRS score was lower than 22.

To evaluate the presence of depression and its severity, we used the Montgomery–Asberg Depression Rating Scale (MADRS) (26). The cut-off scores we used were: <7 absent signs, mild signs = 7–18, moderate = 18–34, and ≥35 reflects severe depression. This scale consistently has the highest Cronbach's alpha levels reaching 0.92 (27). Patients with severe depression (MADRS > 35) were excluded from the study. Patients with scores ≥7 were considered having depression.

The anxiety at the time of this study was evaluated using the Hamilton Anxiety Rating Scale (HAM-A). The cut-off scores we used were: mild anxiety = 8–14; moderate anxiety = 15–23; severe signs of anxiety ≥24; and scores ≤7 were considered to reflect no or minimal signs of anxiety (28). Patients with score ≥8 were considered having anxiety.

Dysautonomic symptoms were assessed using the self-reported autonomic symptoms in PD the SCOPA-AUT questionnaire (29). The 25 items of the SCOPA-AUT are grouped into 6 domains: gastrointestinal functioning (7 items), urinary functioning (6 items), cardiovascular functioning (3 items), thermoregulatory functioning (4 items), pupillomotor functioning (1 item), and sexual function (2 items for men and 2 for women). The score for each item ranges from 0 (where the patient had never experienced the symptom) to 3 (where the patient describes that the symptom is often experienced). The SCOPA AUT total score ranges from 0 to 75 (29).

The quality of sleep was appreciated using the Pittsburgh sleep quality index (PSQI) questionnaire (Dimension 1–7) (30). A total PSQI-score of 5 or greater was considered indicative of poor quality of sleep. EDS was evaluated using the Epworth sleepiness scale (31). A total Epworth score greater than 8 is indicative of excessive diurnal sleepiness and a score of 15 or greater represents a severe EDS.

The fatigue was assessed using the Fatigue severity scale (FSS), the subject is asked to read each statement and circle a number from 1 to 7, depending on how appropriate they felt the statement applied to them over the preceding week. A low value indicates that the statement is not very appropriate whereas a high value indicates agreement. The scoring is done by determining the average response to the questions (adding up all the answers and dividing by nine) (32).

The pain was evaluated by the Douleur neuropathique-4 (DN4) questionnaire, which consists on interviewing questions (DN4-interview) and physical tests; the test was considered positive for a score >4, and the Visual analog scale (VAS) was used to quantify the intensity of that pain. The scale is presented to the patient who is asked to place a line perpendicular to the VAS line at the point that represents his pain intensity. It ranges from 0 (no pain at all) to 10 (my pain is as bad as it could possibly be) (33).

In order to clarify whether the form of the disease could possibly influence the NMSs, we studied the correlation between both of them. For statistical constraints, we divided the patients into two groups: the tremoric dominant form group and the akineticrigid and mixed form as a second group.

## Quality of Life

We chose the 39-item PD questionnaire (PDQ39) to evaluate the QoL. It is a PD-specific health status questionnaire comprising 39 items. It contains eight domains: mobility, activities of daily living (ADL), emotional well-being, stigma, social support, cognition, communication, and bodily discomfort. The PDQ-39SI is obtained by the sum of the eight PDQ-39 scale scores divided by eight, which gives a score between 0 and 100 ("0% = no difficulties" to "100% = maximum level of difficulty"). Higher scores indicated poorer QoL (34).

We analyzed the correlation between NMSs and the QoL of Moroccan patients with PD. After specifying the most affecting NMS on the QoL scores, we analyzed its impact on each of the eight dimensions of the PDQ39 questionnaire.

We also compared the differences between men and women regarding their QoL scores and identified the most affected dimension for each gender.

## Statistical Analysis

Demographic and clinical variables were analyzed using parametric and nonparametric tests as appropriate using SPSS 13.0 software. Quantitative data were expressed as mean ± SD or median and interquartile range. Categorical variables were expressed as numbers and percentages.

The comparison of NMS's prevalence between clinical phenotypes (tremoric dominant form versus akinetic-rigid and mixed form) was done using logistic regression.

To eliminate the possible confounding factor we assessed, in the univariate analysis, the relationships between clinical form and age, gender, disease duration, LEDD, UPDRS I, UPDRS II, UPDRS III, UPDRS IV, UPDRS V, UPDRS VI, PDQI, FSS, Epworth, DN4, VAS, PDQ39, SCOPA-AUT, MADRS, HAM-A, MMSE, AHRS. In the multiple regression, we adjusted for the variables that were statistically significant in the multivariate analysis (*p* < 0.05) (LEDD, UPDRS IV, UPDRS I, UPDRS II, UPDRS III, UPDRS IV, UPDRS VI, FSS, VAS, PDQ39, SCOPA-AUT, MADRS, HAM-A).

In order to identify factors that can impact the QoL we assessed the relationship between PDQ39 and the other variables (age, level of education, gender, disease duration, LEDD, UPDRS I, UPDRS II, UPDRS III, UPDRS IV, UPDRS V, UPDRS VI, SCOPA-AUT, Epworth, PSQI, MADRS, HAM-A, MMS, DN4, VAS, FSS). The correlation was determined, in the univariate analysis, using the Spearman test for quantitative variables and Mann–Whitney test for gender and level of education. Linear regression was used in the multivariate analysis. We adjusted for disease duration (*p* = 0.065) and variables that were statistically significant in the univariate linear regression (*p* < 0.05) (disease duration, UPDRS I, UPDRS VI, PSQI, FSS, VAS, SCOPA-AUT, HAM-A, MADRS). We did not introduce the VAS score (*p* < 0.001 in the univariate analysis) in the multivariate model because it has strong collinearity with the DN4.

The relationship between the dimensions of PDQ39 score and SCOPA-cardiovascular, SCOPA-gastrointestinal was analyzed using Spearman test.

Categorical variables were expressed as numbers and percentages and were compared using Chi-square test. For multiple testing, we corrected the p value by the Bonferroni method.

## RESULTS

## Demographic Data and Motor Aspects

We enrolled into this study a total of 117 patients who fulfilled the inclusion and exclusion criteria of PD. The demographic features are reported in **Table 1**.

The overall mean age of this study population was 60.77 ± 11.36 years. The mean age at onset of the disease was 54.28 ± 12 years and the median of the disease duration was 6 [2.5–9.5] years. The study cohort included 65 (55.6%) males and 52 (44.4%) females. 81 of our patients (69.2%) had low level of education.

Our patients presented different clinical forms with 40.2% (*n* = 47) of the mixed akinetic-rigid-tremoric form, 39.3% (*n* = 46) of the tremor-dominant form, and 20.5% (*n* = 24) of the akinetic-rigid form. The onset of the motor symptoms affected mainly the right side [54.7% (*n* = 64)] and in 2.5% (*n* = 3) of cases the onset was bilateral with asymmetry. The median LEDD was 325 mg per day [200–500].

The median UPDRS-III score was 13 [7–27.75]. The median score of Hoehn and Yahr was 2 [1–2.5] and 27.3% (*n*= 32) of cases had a score above 2. The median score of Schwab and England during ON condition was 90% [80–90%] (**Table 1**).

## Non-Motors Aspects

All our patients (100%) presented at least one NMS. The prevalence of NMS and the medians of the scores used to assess them are presented in **Table 2**. Regarding the clinical phenotype, it did not impact the prevalence of any of the NMSs (**Table 3**). The univariate analysis showed significant results for some NMSs (depression, anxiety, fatigue, and dysautonomic symptoms) and the PDQ39. However, none of these parameters was significant in the multivariate analysis.

#### Autonomic Dysfunctions

The median score of SCOPA-AUT test was 14 [7.75–21.80]. A large number of our patients suffered from urinary [*n* = 95 (82.6%)] and gastrointestinal dysfunctions [*n*= 92 (80%)]. Sexual impairments were observed in 49.3% (*n* = 34), cardiovascular dysfunctions in 50.9% (*n* = 59), thermoregulatory troubles in 58.6% (*n* = 68), and pupillomotor dysfunctions in 31% (*n* = 36) of cases. Patients presented at least two autonomic disorders at the same time in 86.3% (*n* = 101) of cases and only 2.5% (*n* = 3) had no autonomic dysfunctions (**Table 2**).

#### Sleep Disorders

Sleep disorders were the second more frequent symptoms in our cohort after urinary disturbances with 80.6% (*n* = 87) of cases experiencing poor quality of sleep. The median PSQI score was 8 Table 1 | Characteristics of Moroccan Parkinson's disease (PD) patients.


*a Median and interquartile ranges.*

*bMean* ± *SD.*

*c Indicative means for parameters with non-normal distribution.*

*LEDD, levodopa equivalent daily dose; UPDRS, unified Parkinson's disease rating scale; MMSE, mini-mental state examination.*

[6–12.75]. All those patients used hypnotics. The median score of Epworth score for excessive daily sleepiness was 5 [2.28–9.8] with 23.4% (*n* = 25) of our patients having EDS, among which 6.4% had severe EDS (Epworth score ≥16).

#### Neuropsychiatric Symptoms

Depressive symptoms were present in 47.9% (*n* = 56) of patients, 12.5% (*n* = 7) had moderate depression, and 87.5% (*n* = 49) mild depression. The median MADRS score was 6 [2–12]. Anxiety disorders were present in 50.9% (*n* = 58) of our patients with a median score of 8 [3–13.25]. In these patients, 55.1% (*n* = 32) had mild signs, 32.7% (*n* = 10) had moderate, and 12% (*n* = 7) had severe sings of anxiety. About 11% of our patients were treated for depression (*n* = 6) and 12% (*n* = 7) were treated for anxiety. About 7% (*n* = 8) of all the 117 patients had suicidal thoughts, 6% (*n* = 7) reported transient suicidal thoughts, and less than 2% (*n* = 2) only believed that suicide was considered as a possible solution.

#### Fatigue and Pain

About the quarter of our patients (23.1%, *n* = 27) suffered from fatigue and the median FSS was 3.22 [1.5–5.05] and only 11.1% (*n* = 13) suffered from neuropathic pain with a VAS >5.



*% (number of patients).*

*a Median and interquartile ranges.*

*bMean* ± *SD.*

*c Means for parameters with non-normal distribution.*

*HAM-A, Hamilton anxiety scale for anxiety; MADRS, Montgomery–Asberg Depression Rating Scale for depression; HAM-A, Hamilton anxiety rating scale; SCOPA-AUT, self-reported autonomic assessment for Parkinson's disease; VAS, visual analog scale; DN4, "Douleur neuropathique 4" questionnaire for pain; Epworth, scale for excessive daytime sleepiness; FSS, fatigue severity scale; PSQI, Pittsburg sleep quality index; MMSE, mini-mental scale examination; PDQ39-SI, Parkinson's disease questionnaire 39 score index; AHRS, Argentina hyposmia rating score.*

#### Olfaction

Only 50/117 (42.7%) patients were submitted to the AHRS test and their median score was 24 [20–24]. In these patients, 86% (*n* = 43) were aware of their smell loss, but only 28% (*n* = 14) of them had a score below 22 and 14% (*n* = 7) had severe scores (<10). The patients with a positive score estimated that the olfactory changes appeared before the manifestation of motor symptoms. The estimated duration was different from a patient to another with a median less than 1 year and ranges from 10 months to 20 years. Their mean age was 67.7 ± 8 years old vs 58.63 ± 1 for those without olfactory changes (*p* = 0.009). The side of onset of PD in those patients was equal between right and left sides (46.7% each) while 6.7% started the disease bilaterally with asymmetry.

#### QoL and Its Correlation With NMSs

The median score of the PDQ39-SI was 23.22 [13.36–36.69]. The scores of the PDQ39-SI dimensions are presented in **Table 4**. Mobility was the highest score with a median of 30 [11.25–57.5]. Regarding the non-motor dimensions, emotional well-being had a median score of 25 [8.3–54.16], cognition 25 [12.5–41.18], bodily discomfort 25 [0–50] ranking the top. Social support was the last score with a median of 0 [0–16.6].

We compared the PDQ39-SI scores between men and women. The median score of PDQ39-SI for women was 26.61 [16.34–39.27] vs 19.42 [11.09–35.72] for men, with no significant difference (*p* = 0.070). The only significant differences between the two groups when comparing the eight dimensions of this score were in the well-being and mobility, where women had a worst score (*p* = 0.005 and *p* = 0.002, respectively) (**Table 4**).

We also investigated the relationship between level of education and PDQ 39 SI and there was no difference in the PDQ39 SI score between the two groups (*p* = 0.398).

The univariate analysis of the correlation between the PDQ39 and other assessments are listed in the **Table 5**. Of interest, is that age, MMSE score, the daytime sleepiness, and the AHRS showed no correlations with the QoL while motor signs were well correlated as predicted. Moderate positive correlation was found between the PDQ39 and LEDD and SCOPA-AUT sexual dysfunction score (0.01 < *p* < 0.05). Strong positive correlation (*p* < 0.001) was found with the disease duration and the following NMSs-scores: UPDRSI, HAM-A, MADRS, DN4, PSQI, FSS, SCOPA-AUT gastrointestinal, urinary, cardiovascular, thermoregulatory, and pupillomotor dysfunctions.

In the multivariate analysis, the correlation between NMSs and PDQ39-SI (**Table 6**) aimed to see which NMS did affect the QoL of our patients. It showed that the most correlated ones were the autonomic dysfunctions (*p* < 0.001), especially the gastrointestinal *(B* = 1.067 [0.391–1.743], *p* = 0.007) and cardiovascular symptoms (*B* = 1.067 [0.391–1.743], *p* = 0.49). Constipation was the most frequent gastrointestinal sign (62.9%, *n* = 73) and swallowing difficulties was the least frequent complaint (23.3%, *n* = 27). Orthostatic hypotension is the cardiovascular sign sought by the SCOPA-AUT questionnaire, signs of light-headed after passing from sitting to standing position were present in 42.2% of cases (*n* = 49) and only 15.5% (*n* = 18) had fainted during the 6 months prior to the questionnaire. **Table 7** shows the results of the correlation between the PDQ39 dimensions and the most disabling autonomic dysfunctions (gastrointestinal and cardiovascular symptoms). Cardiovascular symptoms did not correlate with all the questionnaire's dimensions. Gastrointestinal had positive correlation with all dimensions except emotional well-being.

## DISCUSSION

This is the first study on NMSs in PD patients enrolled in Morocco. Our results show a high prevalence of NMSs in our patients, in


Table 3 | Comparison of non-motor signs between trembling form group and akinetic-rigid form and mixed form group.

*CI, confidence interval; LEDD, levodopa equivalent daily dose, UPDRS, Unified Parkinson's disease rating scale; PSQI, Pittsburg sleep quality index; FSS, fatigue severity scale; Epworth, scale for excessive daytime sleepiness; DN4, "Douleur neuropathique 4" questionnaire for pain; VAS, visual analog scale; SCOPA-AUT, self-reported autonomic assessment for Parkinson's disease; MADRS, Montgomery–Asberg Depression Rating Scale for depression; HAM-A, Hamilton anxiety scale for anxiety; MMSE, mini-mental scale examination; PDQ39-SI, Parkinson's disease questionnaire 39 score index; AHRS, Argentina hyposmia rating scale.*

Table 4 | Dimensions of the PDQ39 in the study population.


*Values are expressed as median and interquartile ranges as distribution was non-Gaussian and corresponding means and SD are presented below each median. PDQ39, Parkinson's disease questionnaire 39; ADL, activity of daily living.*

*Bold font means significant p values.*

whom at least one NMS has been found. Our results are in line with previous studies (35, 36), in which the authors have found a NMS prevalence rate of 100% in a cohort of 82 and 134 patients, respectively.

In our study, the incidence of PD was slightly higher in male than in female and the average age of our patients was 60.77 years, which is in line with the international reports. By focusing on the clinical manifestations, we found that the most common type of Table 5 | Correlation between motor and non-motor features and the PDQ39- SI (Spearman test).

Table 6 | Correlation between non-motor signs and PDQ39-SI (linear regression and multiple linear regression).


*\*p* < *0.05.*

*\*\*p* < *0.01.*

*PDQ39-SI, Parkinson's disease questionnaire 39 score index; LEDD, levodopa equivalent daily dose; UPDRS: unified Parkinson's disease rating scale; PSQI, Pittsburg sleep quality index; FSS, fatigue severity scale; Epworth, scale for excessive daytime sleepiness; DN4, "Douleur neuropathique 4" questionnaire for pain; VAS, visual analog scale; SCOPA-AUT, self-reported autonomic assessment for Parkinson's disease; MADRS, Montgomery–Asberg Depression Rating Scale for depression; HAM-A, Hamilton anxiety scale for anxiety; MMSE, mini-mental scale examination; AHRS, Argentina hyposmia rating scale.*

onset was the mixed type and the tremor predominant type with a very slight difference, while the akinetic-rigid form was less common. However, the clinical form did not impact the NMSs. Screening of NMSs showed that the autonomic symptoms are the most frequent, followed by sleep and psychiatric (depression) disorders. Fatigue and neuropathic pain have been the less common NMSs present in our patients.

The most frequent NMSs were the urinary disturbances, which is in line with several previous studies (20, 37–39). Urinary disturbances reported in PD patients are diverse and their prevalence through studies is different. The most reported symptoms are: urge incontinence (39, 40), neurogenic detrusor over-activity (40, 41), detrusor hyperreflexia (42), and nocturia (43).

The gastrointestinal symptoms were present in 80% of our patients. Using the SCOPA-AUT scale we were investigating different gastrointestinal symptoms, such as swallowing difficulties, excessive salivation, constipation, and difficult defecation. Constipation was the most common symptom in our patients and swallowing difficulties were around 20% which is in line with previous reports (44, 45). More other symptoms like gastroparesis, gastroesophageal reflux, or weight loss are also


*CI, confidence interval; LEDD, levodopa equivalent daily dose; UPDRS, unified Parkinson's disease rating scale; PSQI, Pittsburg sleep quality index; FSS, fatigue severity scale; Epworth, scale for excessive daytime sleepiness; DN4, "Douleur neuropathique 4" questionnaire for pain; VAS, visual analog scale; SCOPA-AUT, self-reported autonomic assessment for Parkinson's disease; MADRS, Montgomery– Asberg Depression Rating Scale for depression; HAM-A, Hamilton anxiety scale for anxiety; MMSE, mini-mental scale examination; PDQ39-SI, Parkinson's disease questionnaire 39 score index.*

Table 7 | Correlation between the PDQ39 dimensions and the most disabling autonomic items gastrointestinal, cardiovascular symptoms.


*\*p* < *0.05.*

*\*\*p* < *0.01.*

*PDQ39, Parkinson's disease questionnaire 39; SCOPA-AUT, self-reported autonomic assessment for Parkinson's disease, ADL, activity of daily living.*

frequent in all the stages of the disease course, as suggested by the Braak's theory for the pathodynamics of Lewy pathology in PD, and considered as premotor symptoms (46–48). All these symptoms affect at least one-third of the patients (49). Being one of the first organs where the alpha-synuclein is deposited, these symptoms reflect the dysfunction of the enteric nervous system and the stomach, or are considered as side effects of the antiparkinsonian drugs (49). In their study, Cersosimo et al. (49) showed that the gastrointestinal symptoms reported to occur before the onset of motor symptoms were constipation in 87% of patients, defecatory dysfunction in 58.9%, and dry mouth in 20.5%. The fact that their prevalence was similar to that of controls makes this outcome debatable.

While gastrointestinal and urinary dysfunctions usually occur early in PD, cardiovascular symptoms tend to appear and to become prominent with the progression of the disease (50). Some authors associate cardiovascular symptoms to sympathetic dysfunctions and suggest that they may appear in earlier stages even in drug-naïve patients or precede the appearance of motor symptoms (50–52). Cardiovascular symptoms (orthostatic hypotension) were present in more than half of our patients, which is in line with some previous reports (39).

Sexual dysfunctions were present 49.3% of our patients, but we believe that this percentage does not represent the reality of genital troubles that our patients face and it is rather a cultural finding, and consequently it is improper to compare it to other studies. Assessing these symptoms was mostly difficult. Our patients tended to be reserved regarding their sexual life, and those who answered tended to trivialize the existing troubles and relate them to their advanced age. In a previous study, Chaudhuri et al. (45) reported that the questions relating to sex were frequently left unanswered.

Sleep dysfunctions are prevalent in PD and represent an integral component of the disease, which may emerge over different phases of PD. It could be the result of the disease itself, secondary to other NMSs, or a side effect of medicines. It is affecting 42–98% of PD patients (53, 54). RBD is the most typical sleep disorder associated with PD, but fragmented sleep in PD, which is multifactorial is commonly reported (55). In addition, patients with PD are commonly affected by primary sleep disorders also existing in the general population in particularly restless-legs syndrome (RLS, also known as Willis–Ekbom disease) or periodic leg movements in sleep and sleep disordered breathing, notably obstructive sleep apnea (56). The regulation of the normal sleepwake cycle has been shown to involve the pedunculopontine nucleus (PPN) whose role is paramount. In PD, the PPN cholinergic neurons showed special selective vulnerability (57). In our cohort, a large number of patients (80.6%) suffered from poor sleep quality while only 23.4% had EDS. Previous studies showed that 50% of PD patients had prolonged daytime sleepiness or unconscious drowsy (58). This prevalence is much higher than in our study, which can be due to the inadequacy of the Epworth scale in our population as two items are not applicable for illiterate patients. Another explanation can be the early stage of the disease of our patients (72.6% had a Hoehn and Yahr stage <2) and the exclusion of patients with dementia from our study. The frequency of EDS in PD seems to correlate with disease severity, treatment duration, and additional symptoms like dementia and depression (59–61).

The incidence of mental and behavioral disorders in PD patients is generally low except for depression (41.5%), apathy (40.9%), and anxiety (39.4%) (58). Depression may occur at any time during the course of PD, or precede the manifestation of motor symptoms by 4–6 years (62). It may appear early in the Braak stage II, when Lewy bodies deposit in dorsal raphe nucleus and locus coeruleus, and thus may precede the motor symptoms. It is generally attributed to noradrenaline and serotonin deficiency (63, 64). However, depressive symptoms are definitely increasing with the disease progression. Compared to the general population, depression in PD patients is two- to threefold more prevalent (62). Our data confirm those of previous studies showing a prevalence of depression up to 47.9%. The dominant character of depressive signs in our cohort was rather mild depressive symptoms. Suicidal thoughts were present only in 8% of patients. This can be explained by the impact of religion in our society, helping patients to better accept their chronic pathologies.

Concerning anxiety, our results are in line with previous reports. Its estimated prevalence in PD is 25–40% (65) and it is considered as a wearing off phenomenon. As for depression, anxiety can appear at any stage in the course of the disease and it may also precede the motor signs. In regards to the semiological presentation of anxiety in PD, it is usually seen as a generalized anxiety disorder, phobic disorder, or panic disorder (65).

Regarding olfactory disorders, they exist in a large majority of PD patients (up to 90%) and most often are present at the time of diagnosis. Though, in more than 70% of cases, patients are unaware of their smell changes (66). Smell impairment is considered a marked warning of PD before motor symptoms, which is due to the early involvement of olfactory related brain regions by alphasynuclein according to Braak staging and has a good clinical predictive value for PD (46, 57). While some studies concluded that olfactory disturbances are independent of disease duration and stage (66, 67), others thought that they are associated with greater disease severity and progression and that severe hyposmia in individuals with PD can predict the course toward PD dementia (68). In our cohort, olfactory disorders were associated with a later age of diagnosis of the disease (67.5 years old), which is considered to be a predictor for functional dependency in PD (69). Some of our patients reported that the olfactory changes appeared years before the onset of motor symptoms. We used the AHRS (25), being simple and time conserver. It revealed 28% of patients with smell disorders, while 86% assumed that they were suffering from smell changes, which is making us questioning the sensitivity of this scale.

The fatigue, also described as a feeling of tiredness or exhaustion, has been increasingly known in the context of PD. Fatigue is mostly reported by PD patients as one of their most invalidating symptoms with the utmost negative effect on their QoL (70). 23.1% of our patients suffered from fatigue with a median score of 3.22. The pathophysiology of fatigue in PD is unknown, and no effective treatment exists. Goldman et al. reported that fatigue is associated with worse cognitive impairment (71) which can explain the poor frequency of fatigue recorded in our study as patients with dementia were excluded.

Pain was at the bottom of the NMSs frequency. Only 11.1% of our patients reported neuropathic pain. The prevalence of different pain types in PD varies from one study to other; it is commonly reported in the PD-NMSs studies in many various types: musculoskeletal, dystonic, or radicular central neuropathic pain. Their origin seems to be multifactorial, but responding well to Levodopa treatment (72–74). Santos-Garcia et al. (75) found that depression was an independent predictor of pain in PD patients, this may explain our results, since our patients presented mainly mild depression.

Despite the important prevalence of NMSs, the QoL of our patients was not strongly affected (median score at 23.22). Mobility dimension ranked the top of scores with a median of 30, while the median score of ADL was 20.88, which is expected, since PD constitutes primarily a motor handicap. The most affected non-motor dimensions of the PDQ39-SI in our patients were emotional well-being, bodily discomfort, and cognitive impairment. This is in line with the numerous studies that showed the adverse impact of depression and dementia on the QoL of PD patients (20, 76).

The least affected dimension was the social support. Indeed, the majority of our patients were well surrounded by their family members who provided support for their appointments. Stigma median score reached only 18.75%, we believe that our patients seemed to accept their disease and in the majority of cases do not feel embarrassed to reveal it.

In the correlation analysis, all the NMSs were correlated to the QoL, except for the EDS and olfactory changes. Although urinary symptoms were the most common in our patients, the multivariate analysis showed no correlation between urinary symptoms and the QoL. Surprisingly, gastrointestinal and cardiovascular symptoms had the most negative impact on the QoL of our patients. In the literature, the symptoms influencing the QoL of PD patients are diversely reported. A large number of studies reported fatigue, sleep disturbances, depression, apathy, and mood to be determinants of the QoL (38, 62, 77). Generally, depression is considered as the most important neuropsychiatric symptom impacting the QoL in PD patients (20, 37, 78–80), which is not found in our cohort. In line with our results, Svetlana Tomic et al. (39) found gastrointestinal symptoms to be impacting negatively the QoL of PD patients in a study focusing only on autonomic dysfunctions. Also, Gallagher et al. (37) reported QoL to be influenced by thermoregulatory, gastrointestinal, cardiovascular dysfunctions (in particular orthostatic hypotension), and urinary symptoms.

In this study, we analyzed the impact of those symptoms on each dimension of the PDQ39-SI and found that the gastrointestinal symptoms affected all the aspects of QoL (except for the emotional well-being). For cardiovascular symptoms, they did not influence cognition and communication, nor the stigma dimension, they affected negatively all the other aspects. These findings address the fact that before increasing dopaminergic drugs for patients' complaints of difficulty in ADL or reduced mobility, one has to exclude the presence of NMSs that do impact mobility and treating them will in some extent, improve motor state of the patients.

Regarding the gender differences, our female patients had a tendency of higher PDQ39-SI than men but the difference was not significant. There was no significant difference between male and female in our study regarding the assessment of their QoL, except for the emotional well-being and mobility dimensions. Tomic et al. (39) reported the same findings especially that their study was only regarding the autonomic dysfunctions, in addition to this Hristova et al. (81) reported also women to present worst scores in social support and bodily discomfort. In the majority of the few studies regarding the QoL in PD, women seem to experience worst QoL than men (82). On top of that, Moore et al. (83) concluded in their study done on 100 PD patients that gender identity might have a significant effect on their QoL. The androgynous group (men and women) had the best QoL and furthermore, androgynous women had the best scores of all groups (83).

## STUDY LIMITATIONS

The study has some methodological limitations that we took into consideration when interpreting the findings. In other hand, our patients had low disease severity (27.3% had a Hoen and Yahr score >2) and the cognitive impact on QoL was not assessed as we excluded patients with dementia. Also, our study is crosssectional, so the analysis was based on clinical data collected at a single point in time; therefore, any pattern of progression of the disease could not be estimated.

## CONCLUSION

The prevalence of NMSs in our study was 100%, as at least one NMS has been found in all our patients. Urinary symptoms were the most prevalent, although, gastrointestinal and cardiovascular symptoms were the factors that impacted the QoL. Our patients may underestimate the gastrointestinal symptoms without relating them to PD, especially if they exist long time before the onset of the motor symptoms. Also, cardiovascular symptoms constitute a real burden for our patients. These symptoms have to be investigated systematically and treated once present to improve the QoL of PD patients. More studies are needed to understand the pathophysiology of NMSs in order to develop new and effective therapeutic approaches.

## ETHICS STATEMENT

This study was carried out in accordance with the recommendations of the ethics committee of medical school of Rabat (CERB) with written informed consent from all subjects. The protocol was approved by the ethics committee of medical school of Rabat (CERB).

## AUTHOR CONTRIBUTIONS

HT designed the study, collected and analyzed the data, wrote and edited the manuscript. KB assisted with collection and analysis of the data. AB, EH, AB, and MY agreement for clinical data. WR and AB contributed to the conception and supervision of the work and participated in writing the paper, edited and approved the final version to be published. All authors read and approved the final manuscript.

## ACKNOWLEDGMENTS

We are thankful to Dr. Kaswati Janane for his help on statistics.

## REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Tibar, El Bayad, Bouhouche, Ait Ben Haddou, Benomar, Yahyaoui, Benazzouz and Regragui. This is an open-access article distributed*  *under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Hippocampal Neurodegenerative Pathology in Post-stroke Dementia Compared to Other Dementias and Aging Controls

#### Rufus O. Akinyemi 1, 2 \*, Louise M. Allan<sup>2</sup> , Arthur Oakley <sup>2</sup> and Rajesh N. Kalaria<sup>2</sup> \*

<sup>1</sup> Neuroscience and Ageing Research Unit, Institute for Advanced Medical Research and Training, College of Medicine, University of Ibadan, Oyo, Nigeria, <sup>2</sup> Neurovascular Research Group, Institute of Neuroscience, Newcastle University, Newcastle upon Tyne, United Kingdom

Neuroimaging evidence from older stroke survivors in Nigeria and Northeast England

#### Edited by:

Vivienne Ann Russell, University of Cape Town, South Africa

#### Reviewed by:

Wolfgang Härtig, Leipzig University, Germany Scott Edward Counts, Michigan State University, United States

#### \*Correspondence:

Rufus O. Akinyemi roakinyemi@com.ui.edu.ng; rufusakinyemi@yahoo.com Rajesh N. Kalaria raj.kalaria@ncl.ac.uk

#### Specialty section:

This article was submitted to Neurodegeneration, a section of the journal Frontiers in Neuroscience

Received: 30 September 2017 Accepted: 08 December 2017 Published: 19 December 2017

#### Citation:

Akinyemi RO, Allan LM, Oakley A and Kalaria RN (2017) Hippocampal Neurodegenerative Pathology in Post-stroke Dementia Compared to Other Dementias and Aging Controls. Front. Neurosci. 11:717. doi: 10.3389/fnins.2017.00717 showed medial temporal lobe atrophy (MTLA) to be independently associated with post-stroke cognitive impairment and dementia. Given the hypothesis ascribing MTLA to neurodegenerative processes, we assessed Alzheimer pathology in the hippocampal formation and entorhinal cortex of autopsied brains from of post-stroke demented and non-demented subjects in comparison with controls and other dementias. We quantified markers of amyloid β (total Aβ, Aβ-40, Aβ-42, and soluble Aβ) and hyperphosphorylated tau in the hippocampal formation and entorhinal cortex of 94 subjects consisting of normal controls (n = 12), vascular dementia, VaD (17), post-stroke demented, PSD (n = 15), and post-stroke non-demented, PSND (n = 23), Alzheimer's disease, AD (n = 14), and mixed AD and vascular dementia, AD\_VAD (n = 13) using immunohistochemical techniques. We found differential expression of amyloid and tau across the disease groups, and across hippocampal sub-regions. Among amyloid markers, the pattern of Aβ-42 immunoreactivity was similar to that of total Aβ. Tau immunoreactivity showed highest expression in the AD and mixed AD and vascular dementia, AD\_VaD, which was higher than in control, post - stroke and VaD groups (p < 0.05). APOE ε4 allele positivity was associated with higher expression of amyloid and tau pathology in the subiculum and entorhinal cortex of post-stroke cases (p < 0.05). Comparison between PSND and PSD revealed higher total Aβ immunoreactivity in PSND compared to PSD in the CA1, subiculum and entorhinal cortex (p < 0.05) but no differences between PSND and PSD in Aβ-42, Aβ-40, soluble Aβ or tau immunoreactivities (p > 0.05). Correlation of MMSE and CAMCOG scores with AD pathological measures showed lack of correlation with amyloid species although tau immunoreactivity demonstrated correlation with memory scores (p < 0.05). Our findings suggest hippocampal AD pathology does not necessarily differ between demented and non-demented post-stroke subjects. The dissociation of cognitive performance with hippocampal AD pathological burden suggests more dominant roles for non-Alzheimer neurodegenerative and / or other non-neurodegenerative substrates for dementia following stroke.

Keywords: Alzheimer's disease, cerebrovascular disease, mixed dementia, post-stroke dementia, stroke, vascular dementia

## INTRODUCTION

Stroke accounts for half of the global burden of neurological disorders while remaining the most common cause of acquired disability, and a common cause of cognitive impairment and dementia (Kalaria et al., 2016; Writing Group Members et al., 2016; GBD 2015 Neurological Disorders Collaborator Group, 2017). Magnetic resonance imaging studies in a cohort of older African stroke survivors participating in the Cognitive Function After STroke (CogFAST)—Nigeria Study showed that medial temporal lobe atrophy (MTLA) was an independent predictor of post-stroke cognitive impairment and dementia (Akinyemi et al., 2014, 2015). These findings were in concordance with previous findings of MTLA significantly predicting progression to cognitive impairment, dementia and death in the longitudinal CogFAST Newcastle Study (Firbank et al., 2007, 2012) while pathological studies unmasked hippocampal neuronal atrophy as an important neuropathological substrate of post-stroke dementia (PSD) in the Newcastle cohort (Gemmell et al., 2012).

The link between cerebrovascular disease, neurodegeneration and cognition has long been debated (de la Torre and Mussivand, 1993; Kalaria et al., 1993b). Evidence for this link has been provided by experimental models (Kalaria et al., 1993a; Kitaguchi et al., 2009) and from epidemiological studies (Schneider et al., 2004; Petrovitch et al., 2005). Accumulation of Alzheimer pathology in primary vascular brain disorders occurs largely as a result of shared mechanisms of neurovascular unit dysfunction (Iadecola, 2004; Kalaria et al., 2012). Experimental evidence from animal studies has shown that amyloid production may be exacerbated by cerebral hypoxia/ischaemia (Kalaria et al., 1993a; Whitehead et al., 2005).

In a large post-mortem study, Aho and colleagues using immunohistochemistry found no significant increase in amyloid load in subjects with cerebrovascular disease (CVD) compared to controls although there was a trend of increased deposition of Aβ-42 over Aβ-40 (Aho et al., 2006). In this context a post-mortem study of aging controls and subjects with vascular dementia (VaD), reported increased accumulation of total guanidine HCl extractable Aβ-42 peptides (over Aβ-40) by Enzyme Linked Immunosorbent Assay (ELISA) in the temporal cortex in the oldest VaD subjects as well as in older aged controls (Lewis et al., 2006) although these increases were not evident by immunohistochemistry.

Cross-sectional studies examining the relationship of Aβ with cognitive function have also yielded conflicting results. Whereas, some studies have reported significant correlation between metrics of AD pathology and cognitive performance (Bennett et al., 2006; Mormino et al., 2009) others have reported dissociation between Alzheimer pathological load and cognitive status especially in subjects with presumed high cognitive reserve, mixed pathologies or non-AD subjects (Mufson et al., 1999; Aizenstein et al., 2008; Stern, 2009). Similarly, the morphological variants and anatomical localization of the AD pathology may be important. Neuritic plaques and neurofibrillary tangles have stronger impact on cognition than diffuse plaques, and pathologies in the neocortical region exert more influence than those in the allocortical regions (Nelson et al., 2009, 2012).

The hippocampus plays a very strategic role in the neurobiology of memory, being involved in the hierarchical spread of amyloid and tau (neurofibrillary tangles) (Braak and Braak, 1991; Thal et al., 2002b). Whereas, amyloid is deposited in the hippocampus in late stages, tau deposition occurs quite early within the hippocampal formation during their natural histories, and both deposits appear to relate to the connections of the hippocampal circuitry (Lace et al., 2009).

Given the hypothesis ascribing MTLA to neurodegenerative processes (Henon et al., 1998; Cordoliani-Mackowiak et al., 2003; Firbank et al., 2007), we investigated Alzheimer pathology in the hippocampal formation and entorhinal cortex of brain tissues obtained from the CogFAST (Cognitive Function after Stroke) Newcastle post-stroke cohort. Our objective was to quantify the presence of markers of amyloid pathology (total Aβ, Aβ-40, Aβ-42, and soluble Aβ) and a marker of hyperphosphorylated tau pathology in post-stroke subjects with and without dementia compared to aging controls, Alzheimer's disease (AD) and mixed AD and vascular dementia (AD-VaD). We hypothesized that in tandem with MTLA, hippocampal Alzheimer pathology would be differentially expressed in demented and non-demented stroke survivors in comparison with other dementias and aging controls.

## METHODS

## Subjects

Ninety-four human post-mortem brains were retrieved from the Newcastle Brain Tissue Resource (NBTR) at the Campus for Ageing and Vitality, Newcastle University, UK. **Table 1** provides details of the demographic, cognitive and pathologic characteristics of the subjects. Post-stroke subjects in the longitudinal CogFAST-Newcastle Study were classified based on the performance at the last cognitive assessment before death. They were classified as post-stroke non-demented (PSND) if CAMCOG score was >80 and Clinical Dementia Rating (CDR) was less than 1, but as post-stroke demented (PSD) if CAMCOG score was <80 and CDR score was >1 (Allan et al., 2011). Control subjects were historical subjects that had no significant evidence of cognitive impairment upon scrutiny of their medical records and whose post-mortem brain tissue was considered devoid of sufficient vascular or degenerative pathologies beyond the threshold for assigning a specific pathologic diagnosis (Kalaria et al., 2004; Gemmell et al., 2012).

Autopsies were performed between 24 and 92 h after death and brains were fixed for between 6 and 34 weeks. Cognitive scores on the Mini-Mental State Examination (MMSE) and Cambridge Cognitive Examination (CAMCOG) proximate to death as well as APOE genotypes were retrieved from clinical and research records of the subjects in the CogFAST-Newcastle cohort. The CogFAST-Newcastle Study and ancillary studies had ethical approval from the local Newcastle Ethical committees and participants gave written consent for brain tissue donation. Use of brain tissue was also approved by the local Ethical committees (Newcastle upon Tyne Hospitals National Health Service Trust, UK) and the committee of the NBTR.


#### TABLE 1 | Characteristics of study subjects.

Values show mean ± SD unless otherwise indicated. Post-mortem delay between autopsy and fixation of tissue (hours) and Fixation length (weeks) ranged 24–92 h and 6–34 weeks, respectively. There was no correlation between these variables and presence of pathology. \*p < 0.001. AD, Alzheimer's disease; AD\_VaD, Mixed AD\_VaD; PSND, MCA; Middle cerebral artery; N/A, Not Applicable NCD, No Cognitive Data; NPD, No Pathological Data PCA, Posterior cerebral artery; PMD, Postmortem delay; PSD, Post-stroke demented; Post-stroke non-demented; VaD, Vascular dementia.

Post-mortem reports were retrieved for all the cases used in this study. Primary neuropathological diagnoses were made from brain tissue sampled at several coronal levels (Perry and Oakley, 1993) to check for pathological changes consistent with AD, VaD and mixed AD\_VaD in accordance with established pathologic diagnostic criteria (Hyman and Trojanowski, 1997; Kalaria et al., 2004), and following macroscopic and microscopic post-mortem examination of the brain tissue. Haematoxylin and Eosin was utilized as a standard stain for a general neuropathologic structural evaluation of the brain, and for the detection of infarcts and rarefactions. Gallyas and Bielschowsky's silver impregnation stains and AT8 immunohistochemistry were used to evaluate "CERAD" neuritic plaques and neurofibrillary tangles according to the methods of Braak (Braak and Braak, 1991; Mirra et al., 1991). In addition, vascular lesions (cortical and sub-cortical infarcts, border-zone infarcts, strategic infarcts, lacunar infarcts (<15 mm), microinfarcts (<5 mm) and mild, moderate and severe cerebral amyloid angiopathy were recorded (Kalaria et al., 2004). Final diagnoses were assigned during monthly clinicopathologic consensus meetings. A final diagnosis of VaD was made if there was clinical evidence of dementia (DSM IV) and pathologic evidence of multiple or cystic infarcts, lacunes, micro-infarcts, small vessel disease in the absence of severe degenerative pathology (Braak Stage <III) (Kalaria et al., 2004). Subjects were assigned mixed AD\_VaD if there was pathologic evidence of cerebrovascular disease in the presence of significant AD pathology (Braak Stage V or VI) and moderate to severe CERAD scores. A diagnosis of AD was assigned when there was significant Alzheimer pathology-Braak V–VI, moderate to severe CERAD score and absence of significant vascular pathology.

## APOE Genotyping

APOE genotyping was undertaken in the CogFAST cohort only (PSND and PSD) as shown in **Table 1** using restriction enzyme isoform genotyping as previously described (Hixson and Vernier, 1990; Rowan et al., 2005). In brief, ApoE restriction isotyping used oligonucleotides to amplify apolipoprotein E gene sequences containing amino acid positions 112 and 158. The amplification products were digested with HhaI and subjected to electrophoresis on polyacrylamide gels. Each of the isoforms was distinguished by a unique combination of HhaI fragment sizes that enabled unambiguous typing of all homozygotic and heterozygotic combinations. HhaI cleaves at GCGC encoding 112arg (E4) and 158arg (E3, E4), but does not cut at GTGC encoding 112cys (E2, E3) and 158cys (E2).

## Immunohistochemistry

Paraffin embedded brain tissue blocks taken from relevant coronal levels of the Newcastle Brain Map (Perry and Oakley, 1993) and containing the hippocampal formation and entorhinal cortex were cut into 10µm serial sections using a rotary microtome. Sections were mounted on slides coated with 2% APES (3-aminopropyltrethoxysilane) solution in acetone, and dried in a pre-heated oven at 600C for 30 min. The cut sections were then serially immunostained in duplicates with primary antibodies to different amyloid-β species and tau (**Table 2**). The sections were first de-paraffinized in two sequential solutions of Xylene for 15 min and then rehydrated using decreasing concentrations of ethanol (100, 95, 70, and 50%) to deionized water. Antigen retrieval was performed for Aß-immunolabeling using formic acid-pretreatment for 4 h and for tau immunolabeling using heat in the form of microwaving sections for 10 min in a solution of 0.01M citrate buffer (PH 6.0). The buffer was brought to boil in the microwave, slides were added, buffer was microwaved at 450W for 10 min, and the solution was then allowed to cool for 20 min following which slides were transferred to deionized water. Non-specific reaction was quenched with 0.9% hydrogen peroxide (unless otherwise stated.) in 5 Mm Tris buffered Saline (TBS) solution (pH 7.6) for 15 min in order to remove endogenous peroxidise. Non-specific antigens were blocked using normal horse serum (anti-mouse antibody: 4G8, NU-1 and AT8) and normal goat serum (antirabbit antibody: T-40 and T-42 for 60 min. The slides were then incubated for 2 h at room temperature (AT8) or overnight at 40C (4G8, T-40, T-42, and NU-1) with the primary antibody diluted to specific concentration with buffer: total amyloid β (4G8, 1: 1000, Mouse monoclonal, Signet 9220-10), amyloid β-42 (T-42, 1: 5000, Rabbit polyclonal, gift; H. Mori, Japan), amyloid β-40 (T-40, 1: 5000, Rabbit polyclonal, gift; H. Mori, Japan), soluble amyloid oligomer (NU-1, Mouse monoclonal, gift: C. Klein, US) and antibody to hyperphosphorylated tau (AT8, 1: 1000, Mouse monoclonal Autogen Bioclear). After washes in buffer, biotinylated secondary antibody was applied to the sections with the blocking serum for 30 min, followed by the addition of the streptavidin/biotinyl-horseradish peroxidase (HRP) complex (SABC) for 30 min to remove excess secondary antibody. Finally, the slides were immersed in a 0.025% diaminobenzidine (DAB) solution for a variable short period of time to visualize the positive antibody reaction. Sections were then rinsed in water and counterstained in haematoxylin as indicated. Sections were then dehydrated back through graded alcohols, cleared in xylene and mounted with glass coverslips using DPX mounting medium (Sigma, UK). After each step, with the exception of the blocking stage, sections were rinsed in buffer (TBS) three times for 5 min each. All immuno-histochemical protocols included a positive control and a negative control for which the primary antibody is omitted. All the antibodies used in the analysis have been previously well characterized. We have published on use of these antibodies as described in our previous publications (Chang et al., 2003; Lewis et al., 2006; Ndung'u et al., 2012; Mukaetova-Ladinska et al., 2015).

## Microscopy and Image Analysis

Stained sections were examined and digitized using a Zeiss Axioplan 2 research grade microscope coupled to an Infinity 2 camera. Magnification was set at X10 for the hippocampal subregions CA1, CA2 and CA3, and X5 for the subiculum and entorhinal cortex (EC). Five images were taken at random from the CA1, CA3 and subiculum, 3 images from CA2 and 4 × 3 from the EC from the pial surface to the white matter. Approximately 2,820 images were taken.

Images were analyzed using Image Pro Plus 6.3 (Media Cybernetics, Silver Spring, MD, USA). The area of interest (AOI) was delineated on-screen using the wand tool and the size of the area selected was recorded. Histogram based analysis was used to measure the number of pixels stained ("per area," PA) and intensity of stain (integrated optical density, IOD), as determined by the operator. Background was excluded by highlighting only the positive staining. To standardize the measurements, the range of stain intensity for each AOI was determined between 0 and 255 (where 0 = black and 255 = white, as set up for the Image Pro Plus Histogram based analysis). Using the sum of each of the three measures, further calculations were undertaken to generate the metrics of immunostaining: Percent of the area of interest positively stained, "Percent area (%PA)" = Per Area × 100; Mean intensity of stain per pixel, "Integrated Optical Density (IOD)" = 255 – (sum IOD/area). The mean % PA and IOD were then calculated for each subject by computing averages from the images taken from each hippocampal sub-region.

## Statistics

Statistical analysis was carried out using the IBM SPSS software (version 19.0). The Kolmogorov-Smirnov Test was used to establish normality of data. Comparisons across



NoHoS, Normal horse serum; NoGoS, Normal goat serum; O/N, Overnight; SABC, avidin biotin complex; DAB, diaminobenzidine.

groups of cases and across sub-regions were performed using parametric tests (ANOVA for group means and Tukey post-hoc analysis for between-group differences) and non-parametric tests Kruskal–Wallis and Mann–Whitney U-tests) for nonnormally distributed dataset. The relationship among markers, and with demographic, cognitive and pathological variables were assessed using Pearson's correlation coefficient (r) or Spearman's correlation (rho) as necessary depending on the normality of the dataset. Appropriate power calculation was performed using the G∗Power software (Faul et al., 2007) at significance level, α-level = 0.05 and assuming a moderate effect size Cohen's d = 0.4.

## RESULTS

## Characteristics of Study Subjects

The demographic, cognitive and pathological characteristics of the study participants [non-demented (PSND), demented (PSD) subjects from the CogFAST-Newcastle study; Control, VaD, AD and mixed AD-VaD groups] are shown in **Table 1**. There were no significant differences in the age (p = 0.786), gender distribution (p = 0.493), post-mortem delay (p = 0.902) and length of fixation of tissues (p = 0.589) across the groups. However, the PSND group had significantly higher scores on the cognitive batteries MMSE and CAMCOG compared to the PSD group (p < 0.05). Similarly, Braak stage, CERAD score and Thal stage were significantly higher in the AD, AD-VaD groups compared to the VaD and post-stroke groups (PSND and PSD) (p < 0.05). Hippocampal vascular scores were similar among the PSND, PSD, VaD and AD-VaD groups. The distribution of the ε3 and ε4 alleles of APOE were not significantly different between the PSND and PSD groups (Fisher's exact test = 2.13; p = 0.249). Given a total sample size of 94, a significance level, α = 0.05 and assuming a moderate effect size Cohen's d = 0.4, 6 sub-groups and 5 degrees of freedom, the computed Power (1-β) = 0.8424 using the G∗Power software (Faul et al., 2007).

## Quantification of Amyloid (Aβ) Burden

In quantifying amyloid burden both parenchymal as well as vascular amyloid immunoreactivities were incorporated in order to capture the total quantity of the different species of amyloid detected within the defined area of interest as previously performed (Lewis et al., 2006). **Figure 1** illustrates the immunostaining pattern with antibodies to various Aβ species in serial sections across the CA1 sub-region of the hippocampus.

## Immunolabeling of Total Aβ with 4G8 Antibody

Distribution of 4G8 antibody immunostaining dataset assessed by Kolmogrov-Smirnov test showed non-normal distribution**.** Spearman's rank correlation analysis revealed no significant associations between post-mortem delay, length of fixation period and 4G8 immunoreactivity (IR) measures in the hippocampal sub-regions and entorhinal cortex. Furthermore, there was significant positive correlation between scores on the semi-quantitative amyloid rating scales of CERAD, Thal, Braak and tau stages with the metrics of 4G8 total IR in the entorhinal

FIGURE 1 | Illustrative images of amyloid pathology in the CA1 sub-region across different markers and across disease groups and aging controls. There is higher expression of amyloid in the AD and AD\_VaD groups compared with the VaD, PSD, and PSND and control groups. The level of amyloid immunoreactivity is similar between 4G8 and T-42 but much lower in T-40 and NU-1.

cortex, subiculum and CA1 sub-regions (**Table 3**) suggesting good agreement between semi-quantitative and quantitative measures of amyloid and tau quantification.

In the CA1 region, total Aβ IR varied significantly across groups (p = 0.008, Kruskal-Wallis Test) (**Figure 2**). Between group differences assessed with Mann–Whitney U-Test showed that compared to the control group, total AβIR was significantly higher in AD (p = 0.002), showed a trend with AD-VaD (p = 0.064) and PSND (p = 0.076) but not significantly different between PSD and VaD. In the CA2 and CA3 regions, there was no significant variation across disease groups. In the subiculum, there was significant variation across disease groups and controls (p < 0.001, Kruskal–Wallis Test) with total IR being significantly higher in the AD (p = 0.001), AD-VaD (p = 0.008) and PSND (p = 0.015) groups but not significantly different between PSD and VaD. In the entorhinal cortex, total Aβ IR similarly showed significant variation across the sub-region (p < 0.001, Kruskal– Wallis Test) with values significantly higher in AD (p = 0.022), AD-VaD (p = 0.016) than in controls while PSD was significantly lower than PSND (p = 0.019), AD-VaD (p = 0.002) and AD (p = 0.002) (**Figure 2**).

### Aβ-42 Immunohistochemistry with T-42 Antibody

In the CA1 region, Aβ-42 immunoreactivity varied significantly across groups (p < 0.001, Kruskal-Wallis Test) (**Figure 3**). Between group differences assessed with Mann–Whitney U-Test showed that compared to the control group, T-42 immunoreactivity was significantly higher in AD (p = 0.002) and AD-VaD (p = 0.005), but not significantly different from VaD, PSND and PSD (**Figure 3**). In the CA2 region, immunoreactivity varied across disease groups (p = 0.001, Kruskal–Wallis Test) with the IR being significantly higher in AD-VaD (p = 0.046) and VaD (p = 0.014) groups compared to the control and PSD groups respectively (**Figures 4B**, **5**). In the CA3 region, values revealed normal distribution by Kolmogorov–Smirnov test of normality and ANOVA showed significant variation of Aβ-42 immunoreactivity across the regions (p < 0.001). Compared to the PSND group, immunoreactivity was significantly higher in AD (p < 0.001) and AD-VaD (p = 0.002) groups but not significantly different from control PSD and VaD groups (**Figure 3**). In the subiculum, Aβ-42 immunoreactivity varied significantly across groups (p < 0.001) and was significantly higher in the AD (p = 0.001) and AD-VaD (p = 0.008) groups compared to the control and VaD groups respectively. However, the VaD groups did not differ significantly from the PSND and PSD groups respectively (**Figures 4**, **5**). Similarly, Aβ-42 immunoreactivity varied significantly across the entorhinal cortex (p < 0.001) and was significantly higher in the AD (p = 0.001) and AD-VaD (p = 0.001) groups but not different from the VaD, PSND and PSD groups.

### Aβ-40 Immunoreactivity with T-40 Antibody

Immunostaining and distribution of T-40 antibody across disease groups varied significantly across the sub-regions (p = 0.001, Kruskal–Wallis Test). In CA1, the IR in the control group was not significantly different from that of PSND, PSD and, VaD but was significantly lower than AD (p = 0.004) and AD-VaD (p = 0.010). There was no difference between the PSND and PSD groups. In


α designates total Aβ (4G8) immunoreactivity in the CA1, CA2, CA3, subiculum (SB) and entorhinal cortex (EC) sub-regions of the hippocampus while β depicts CERAD score, Thal stage, Braak stage and tau stage. Statistical significance designated by the following p-values: \*\*p < 0.01; \*p < 0.05; † p < 0.1. Abbreviations: CERAD, Consortium to Establish a Registry for Alzheimer's Disease.

CA2, the control group IR was lower than AD, VaD and AD-VaD although this did not attain statistical significance (p > 0.05). However, VaD (p = 0.005) and AD-VaD (p = 0.006) groups were significantly higher than the PSD group. The PSND group did not differ from the PSD group (**Figure 4**). In the entorhinal cortex, Aβ-40 IR was significantly higher in the AD and AD-VaD groups compared to the control and PSD groups (p < 0.05, Kruska– Wallis Test). There was no significant variation in the CA3 region and subiculum.

We further explored the relationship of Aβ-42 and Aβ-40 based on the relative immuno-reactivities of T-42 and T-40 in the CA1 sub-region. The ratio varied between 1.17 and 748. 39 and the mean value was least in the control group and progressively increased in the PSND, PSD, VaD and AD groups to attain a maximum value in the AD-VaD group (Data not shown).

## Soluble Aβ-Immunoreactivity with NU-1 Antibody

There were no significant differences in the percentage area and total immunoreactivity of NU-1 across the disease groups in the CA1, CA2, and CA3 regions (p > 0.05, Kruskal Wallis Test). However, in the subiculum and entorhinal cortex,

the immunoreactivity varied significantly (p < 0.05 Kruskal– Wallis Test). Between group analysis showed significantly higher immunoreactivity in the AD-VaD (p = 0.007 Mann–Whitney U) group compared to the control group, and in the AD-VaD (p = 0.008) and AD (p = 0.040) compared to the PSD group (**Figure 5**). The PSND and PSD groups showed no significant differences.

#### Hyperphosphorylated tau Immunoreactivity with AT8 Antibody

**Figure 6A** shows tau (AT8) immunoreactivity across disease groups in the hippocampal subregion CA1 while **Figure 6B** shows the quantification across different subregions. Across disease groups, AT8 immunoreactivity was highest in AD and AD\_ VaD groups in comparison to each of the other groups of PSND, PSD, VaD and Controls (p < 0.05). There was slightly higher AT8 immunoreactivity in PSD compared to the PSND group in the CA2 and CA3 sub-regions (**Figure 6B**) although the difference did not attain statistical significance.

## Influence of APOE ε4 Genotype on Amyloid and tau Accumulation in Post-Stroke Sub-Cohort

Eight out of 16 subjects with PSND (50%) were APO ε4 positive compared to 3 out of 13 PSD cases (23.1%). Overall, 11 out of 29 (37.9%) post-stroke subjects were APOE ε4 positive.

Evaluation of the influence of APOE ε4 status on amyloid and tau deposition in the post-stroke sub-cohort revealed a statistically significant higher amyloid load in APOE ε4 positive post-stroke subjects compared to APOE ε4 negative subjects in the subiculum and entorhinal cortex respectively (p = 0.01) (**Figure 7A**). Presence of APOE ε4 also influenced tau deposition (AT8 immunoreactivity) in the post-stroke cohort, the immunoreactivity being significantly higher in APOE ε4 positive subjects in the subiculum (Mann–Whitney U-Test, p = 0.042) (**Figure 7B**).

#### Clinico-Pathological Correlations

We explored relationships between the various markers of Alzheimer pathology used in this study and cognitive scores of the post-stroke group utilizing Spearman correlation analysis. Measures of general cognitive functioning (MMSE and CAMCOG total) and functioning in the memory domain (CAMCOG memory) were correlated with two measures of Alzheimer pathological burden: Aβ-42 immunoreactivity (being the predominant β-amyloid species deposited in the hippocampus) and tau immunoreactivity measures in the CA1, subiculum and entorhinal cortex: sub-regions which demonstrated the most consistent patterns of variation of immunoreactivity across the hippocampal formation. **Table 4** demonstrates significant correlation of only AT8 immunoreactivity in the subiculum with CAMCOG memory (rho = −0.425, p = 0.024), whereas there was no significant correlation with Aβ-42 immunoreactivity.

## DISCUSSION

We hypothesized that Alzheimer pathology would be differentially expressed in demented and non-demented post-stroke subjects in comparison to normal aging controls and other dementias. We found (1) amyloid beta deposition was not remarkably different between PSD and PSND groups despite the finding that MTLA was a significant feature in PSD subjects. (2) As expected, concentrations of total amyloid beta and amyloid β-42 were significantly greater in AD and mixed subjects but low in post-stroke subjects consistent with insufficient concentrations for a diagnosis of AD (3) an association between APOE ε4 allele positivity and higher load of amyloid and tau pathology in the subiculum and entorhinal cortex of post-stroke cases (4) hyperphosphorylated tau immunoreactivity did not

differ significantly between PSND and PSD groups and (5) poor correlation of cognitive measures with the burden of amyloid and tau pathology in post-stroke subjects.

## Amyloid Accumulation, Aging and Cerebrovascular Disease

Consistently across all the markers of amyloid pathology, we found evidence of increasing accumulation of amyloid in controls, post-stroke groups, AD and AD\_VaD in that order. Our finding of amyloid accumulation in normal aging controls concurs with the biological phenomenon of aging-associated accumulation of amyloid that has been reported across species: in drosophilia (Rogers et al., 2012), mice (Yamada et al., 2011), non-human primates (Ndung'u et al., 2012) and man (Tomlinson et al., 1968; Katzman et al., 1988; Bennett et al., 2006; Lewis et al., 2006; Boyle et al., 2013b). This occurs as a result of aging-related compromise of the neurovascular unit resulting in increased production of amyloid and reduced clearance through the perivascular space (Iadecola, 2004; Kalaria, 2009; Kalaria et al., 2012). Amyloid accumulation in the post-stroke groups mirrored that in the VaD group and was less than in the AD and AD-VaD groups (**Figures 2**, **3**) in consonance with the findings of Lewis et al. (2006) showing enhancement of amyloid accumulation in VaD possibly triggered by cerebral hypoxia consequent to cerebral vascular disease (Lewis et al., 2006; Kalaria et al., 2016). This is in tandem with previous findings of enhanced accumulation of amyloid in animal models of chronic cerebral hypoperfusion (Kalaria et al., 1993a; Yamada et al., 2011) as well as increased Pittsburgh Compound B (PIB) uptake in PSD subjects in a pilot study (Mok et al., 2010). In addition, this suggests that beyond age-associated accumulation of amyloid, cerebral vascular disorders including stroke do exacerbate brain amyloid deposition as previously demonstrated in brain tissue from hypertensive (Petrovitch et al., 2000) and diabetic subjects (Luchsinger, 2010). Although Marchant et al. failed to establish a direct relationship between CVD and Aβ using PIB-PET approach in a cohort of non-demented elderly subjects (Marchant et al., 2012), and further suggested that the PIB-PET amyloid measures did not influence cognition, the authors admitted that the limited statistical power of the study may have failed to detect any existent interaction. Utilizing a semi-quantitative approach, a neuropathological study of 484 post-mortem brains did not find a relationship between amyloid deposition and cerebrovascular lesions (Aho et al., 2006). The previous study by Lewis et al. (2006) and this current work have utilized sensitive quantitative approaches to detect amyloid

load. Besides, findings from studies in non-demented elderly subjects may not necessarily simulate those in demented subjects with significant CVD as the mechanisms that produce cognitive decline and dementia may differ in different clinicopathological scenarios (Kalaria, 2012a,b).

Although the main Aβ variants detected in the human brain are Aβ1–40 and Aβ1–42, a significant proportion consists also of N-terminal truncated species (Aβn-40/42 where n = 2 to 11) of which pyroglutamate-modified Aβ peptides are predominant components. Most N-truncated Aβ are considered to be the degradation products of full-length Aβ although Aβ11–40/42 may be generated intracellularly directly from APP by BACE proteolysis occurring in trans-Golgi network (Perez-Garmendia et al., 2014). AβN3(pE), Aβ peptide bearing amino-terminal pyroglutamate at position 3, has been shown to be a major Ntruncated/modified constituent of intracellular, extracellular and vascular Aβ deposits in AD brain tissue (Saido et al., 1995) which progressively accumulates in the brain and could predate the development of AD symptoms. These truncated species are believed to constitute a potential seeding specie in the formation of pathological amyloid aggregates (Saido et al., 1995; Perez-Garmendia et al., 2014) Although we did not specifically assay for these pyroglutamate-modified Aβ peptides, they may have contributed to the intraneuronal and vascular amyloid which was

detected by the immunolabeling of total Aß (4G8) in the current study.

## Sub-Regional Variation in Hippocampal Amyloid Accumulation

Across the sub-regions of the hippocampal formation and entorhinal cortex, amyloid deposition was significantly higher in the CA1, subiculum and entorhinal cortex compared to the CA2 and CA3 regions respectively. This differential pattern may be related to the spatial localization and role of these regions in the hippocampal circuitry (Lavenex and Banta Lavenex, 2013), differential susceptibility of these sub-regions to different pathologies (Small et al., 2011) or the temporal evolution and hierarchical progression of cerebral amyloidosis (Thal et al., 2006).

The entorhinal cortex has been described as the gateway into the hippocampal formation whereas the subiculum and CA1 regions constitute the outflow stations (Goldman-Rakic et al., 1984; Suzuki and Amaral, 2004). Alzheimer pathology tends to spread along the hippocampal circuitry (Thal et al., 2002a; Lace et al., 2009) and this may explain the differential susceptibility and high β-amyloid load in these sub-regions. Besides, in the hierarchical evolution and natural history of amyloid and tau pathologies, these sub-regions are affected earlier in the disease course, compared to other regions such as CA2 and CA3 (Thal et al., 2002b; Lace et al., 2009). The later effect in the CA2 region, in particular may reflect the natural course of disease or the existence of some underlying protective mechanisms operating in the early stages of disease and only giving way in the advanced stage of the disease (Caruana et al., 2012).

The finding of a relatively higher ratio of Aβ-42 compared to Aβ-40 which further increases with the degree of accumulation of AD pathology is in concordance with previous findings (Aho et al., 2006; Selkoe, 2008) demonstrating the predominance of Aβ-42 over Aβ-40 in brain parenchymal amyloid deposits. We have also previously demonstrated a preponderance of Aβ (42) over Aβ-40 in parenchymal and vascular amyloid deposits in non-human primates including squirrel, rhesus monkeys and aging baboons (Ndung'u et al., 2012).

## Differential Amyloid and tau Deposition between PSND and PSD

Largely, there were no significant differences in the quantity of amyloid and tau deposited across hippocampal regions and markers in PSND compared to PSD groups. This suggests agerelated deposition of amyloid and tau in post-stroke survivors. However, total Aβ immunoreactivity was unexpectedly higher in the entorhinal cortex of PSND than PSD. Though not statistically significant, a similar pattern was observed with Aβ-42 in the subiculum, Aβ-40 in the entorhinal cortex, and


TABLE 4 | Correlation matrix showing association of neuropathologic measures of amyloid and tau pathology with cognitive scores in selected hippocampal sub-regions and entorhinal cortex.

soluble Aβ in the CA1 sub-region. This suggests that diffuse early amyloid accumulation in post-stroke subjects alone does not explain why some stroke survivors become demented while others remain cognitively intact. Cognitively normal elderly subjects with huge quantities of amyloid pathology but preserved cognitive functioning have been described in the literature (Bennett et al., 2006; Chetelat et al., 2010). It may also imply that amyloid needs the synergy of other pathologies including tau pathology, vascular lesions, brain atrophy, white matter pathology, MTLA in order to produce significant cognitive decline and dementia (Mormino et al., 2009). In a PIB-PET study of elderly subjects-normal, mild cognitive impairment (MCI) and AD, the investigators found that whereas amyloid load (PIB index) and hippocampal atrophy both predicted loss of episodic memory, amyloid deposition alone in the absence of hippocampal atrophy failed to predict episodic memory loss (Mormino et al., 2009). In addition, a recent study by Wong et al. (2016), found that amyloid retention measured by 11C-PiB PET showed no association with cognitive impairment and clusters of neuropsychiatric symptoms suggesting that other plausible biological pathways could be advanced to explain the development and progression of cognitive impairment and neuropsychiatric symptoms following stroke (Wong et al., 2016). A complementary study of our current cohort (Gemmell et al., 2012) found that whereas pyramidal neuronal volume was preserved in the CA regions and entorhinal cortex of the control and PSND groups, subjects in all the demented groups (including PSD) had significant atrophy of these neurons. This would, therefore, suggest that high amyloid load in the PSND group was insufficient to produce dementia because of preserved neuronal volume. This, indeed, may be a signature of brain/cognitive reserve, preserved synaptic integrity or some other compensatory mechanisms (Stern, 2009; Boyle et al., 2013a).

The observation of similarity between amyloid load in the PSD compared to the control group could imply that given similar quantities of amyloid pathology with respect to the controls, the PSD group was demented possibly because of the presence of additional cerebrovascular lesions which lowered the pathological threshold (Snowdon et al., 1997; Esiri et al., 1999); the presence of neuronal atrophy (Gemmell et al., 2012) or the slightly higher and more advanced hyperphosphorylated tau pathology in the CA2 and CA3 sub-regions of the PSD group. A recent study in which total and phosphorylated tau proteins were quantified in the frontal and temporal cortices of subjects with vascular dementia found a selective loss of total tau protein in VaD compared with controls and AD, whereas phosphorylated tau levels were similar to controls in VaD in both regions, but they were increased in the temporal lobes of patients with AD (Mukaetova-Ladinska et al., 2015). These results demonstrated that breaches of microvascular or microstructural tissue integrity subsequent to ischemic injury in older age may modify tau protein metabolism or phosphorylation and have effects on the burden of neurofibrillary pathology (Mukaetova-Ladinska et al., 2015).

## APOE ε4 Genotype and Accumulation of Amyloid and tau Pathologies

Further analysis of the post-stroke cohort suggested that the presence of the APOE ε4 allele was responsible for driving amyloid and tau accumulation in those who possessed the allele (which was present in 50%) of the PSND subjects. Despite the limited size of the sub-cohort, the amyloid load was significantly higher in the PSND group than the PSD group. APOE ε4 has been associated with accumulation of amyloid and/or tau pathology (Nagy et al., 1995; Saito et al., 2002) in AD but the relationship with post-stroke dementia has been conflicting and less well defined. Previous studies in the Newcastle cohort failed to establish a relationship between APOE ε4 with post-stroke cognitive impairment at 3 months after the stroke (Rowan et al., 2005) but predicted decline at 1-year follow up (Ballard et al., 2004). Furthermore, these results are in concordance with a previous study examining the genetic associations of vascular dementia subtypes in which an association was found between APOE-ε4 allele and mixed dementia, stroke-related dementia and subcortical ischemic vascular dementia (SIVD) as well as higher Aβ-(42) levels (Jones et al., 2011). Other studies had also reported both positive (Packard et al., 2007; Liu et al., 2012) and negative associations (Gdovinova et al., 2006) although these were largely clinical studies. It is plausible that the APOE ε4 allele might have contributed to the higher quantity of amyloid in subregions of the PSND group compared to the PSD group, but further research is required to explore this relationship further.

## Correlation of Cognitive Scores with AD Pathology

In the post-stroke cohort with available cognitive scores, very limited correlation of tau pathology with CAMCOG memory score was established. In this cohort, there was dissociation of cognitive performance and hippocampal Alzheimer pathologic burden. The implication of this may be that AD pathology probably does not contribute very strongly to the substrates of dementia after a stroke event as previously hypothesized (Henon et al., 1997). And if there was any contribution at all, tau pathology probably contributed more than amyloid pathology. Recent reports also suggest that MTLA which had hitherto been widely ascribed to AD pathology, may have a vascular basis (Bastos-Leite et al., 2007; O'Sullivan et al., 2008). The lack of association may also be due to the presence of other pathologies such as synucleinopathies, (the determination of which is beyond the scope of the current study), presence of robust cognitive reserve and other lifestyle and psychosocial factors might also offer reasons for the dissociation between clinical and cognitive measures in the current study. Our study had limitations. We did not quantify specific isoforms of phosphorylated tau protein (R3 or R4) in our cohort as this could yield different amounts in the different subgroups. Alternative techniques including neurochemical approaches such as ELISA could have also be employed to further substantiate the findings from this current study and the potential existence of familywise Type

## REFERENCES


1 error associated with multiple pairwise comparisons is also acknowledged.

In conclusion, we did not find quantitative AD pathological markers sufficient to separate the non-demented post-stroke from demented post-stroke subjects. Thus, hippocampal ADpathological mechanisms do not separate non-demented from demented stroke subjects. Neocortical Alzheimer pathology, other non-Alzheimer neurodegenerative pathologies as well as other non-neurodegenerative mechanisms such as vascular and inflammatory/immune mechanisms require further research in order to fully determine the precise pathological substrates of dementia following stroke.

## AUTHOR CONTRIBUTIONS

ROA performed experiments in the study and drafted the first draft of the manuscript. AO provided technical support. LA and AO contributed to critically revising the manuscript for important intellectual content. RK corrected drafts and obtained the funding. All co-authors approved the final version of the manuscript for submission.

## FUNDING

Our work is supported by grants from the Dunhill Medical Trust (UK), Medical Research Council (MRC, G0500247), Newcastle Centre for Brain Ageing and Vitality (BBSRC, EPSRC, ESRC and MRC, LLHW), and Alzheimer's Research (ARUK). Our work is also supported by the Newcastle Brain Tissue Resource, which is funded in part by a grant from the UK MRC (G0400074), by the Newcastle NIHR Biomedical Research Centre in Ageing and Age Related Diseases award to the Newcastle upon Tyne Hospitals NHS Foundation Trust, and by a grant from the Alzheimer's Society and ARUK as part of the Brains for Dementia Research Project. ROA is also supported by a grant from the University of Ibadan College of Medicine (CTR16A012) for the IBADAN Brain Bank Project.

## ACKNOWLEDGMENTS

ROA was recipient of a research fellowship and a return home fellowship from the International Brain Research Organization (IBRO), and an Overseas Research Studentship from Newcastle University, UK. We are also grateful to the patients, families, and clinical house staff for supporting our longitudinal studies. We are grateful for the technical assistance of Janet Slade in performing experiments.


Perry, R. H., and Oakley, A. (1993). Newcastle Brain Map. London: Wolfe.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Akinyemi, Allan, Oakley and Kalaria. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# A Putative Mechanism of Age-Related Synaptic Dysfunction Based on the Impact of IGF-1 Receptor Signaling on Synaptic CaMKIIα Phosphorylation

Olalekan M. Ogundele<sup>1</sup> \*, Joaquin Pardo<sup>2</sup> , Joseph Francis <sup>1</sup> , Rodolfo G. Goya<sup>2</sup> and Charles C. Lee<sup>1</sup> \*

<sup>1</sup>Department of Comparative Biomedical Sciences, School of Veterinary Medicine, Louisiana State University, Baton Rouge, LA, United States, <sup>2</sup> Institute for Biochemical Research of La Plata, School of Medicine, National University of La Plata, La Plata, Argentina

#### Edited by:

Nilesh Bhailalbhai Patel, University of Nairobi, Kenya

#### Reviewed by:

Jinfei Ni, Harvard Medical School, United States Jatin Tulsulkar, The Ohio State University, United States Christine Gall, University of California, Irvine, United States

> \*Correspondence: Olalekan M. Ogundele ogundele@lsu.edu Charles C. Lee cclee@lsu.edu

Received: 26 August 2017 Accepted: 18 April 2018 Published: 14 May 2018

#### Citation:

Ogundele OM, Pardo J, Francis J, Goya RG and Lee CC (2018) A Putative Mechanism of Age-Related Synaptic Dysfunction Based on the Impact of IGF-1 Receptor Signaling on Synaptic CaMKIIα Phosphorylation. Front. Neuroanat. 12:35. doi: 10.3389/fnana.2018.00035 Insulin-like growth factor 1 receptor (IGF-1R) signaling regulates the activity and phosphorylation of downstream kinases linked to inflammation, neurodevelopment, aging and synaptic function. In addition to the control of Ca<sup>2</sup><sup>+</sup> currents, IGF-1R signaling modulates the activity of calcium-calmodulin-dependent kinase 2 alpha (CaMKIIα) and mitogen activated protein kinase (MAPK/ErK) through multiple signaling pathways. These proteins (CaMKIIα and MAPK) regulate Ca<sup>2</sup><sup>+</sup> movement and long-term potentiation (LTP). Since IGF-1R controls the synaptic activity of Ca<sup>2</sup><sup>+</sup>, CaMKIIα and MAPK signaling, the possible mechanism through which an age-dependent change in IGF-1R can alter the synaptic expression and phosphorylation of these proteins in aging needs to be investigated. In this study, we evaluated the relationship between an age-dependent change in brain IGF-1R and phosphorylation of CaMKIIα/MAPK. Furthermore, we elucidated possible mechanisms through which dysregulated CaMKIIα/MAPK interaction may be linked to a change in neurotransmitter processing and synaptic function. Male C57BL/6 VGAT-Venus mice at postnatal days 80 (P80), 365 and 730 were used to study age-related neural changes in two brain regions associated with cognitive function: hippocampus and prefrontal cortex (PFC). By means of high throughput confocal imaging and quantitative immunoblotting, we evaluated the distribution and expression of IGF-1, IGF-1R, CaMKIIα, p-CaMKIIα, MAPK and p-MAPK in whole brain lysate, hippocampus and cortex. Furthermore, we compared protein expression patterns and regional changes at P80, P365 and P730. Ultimately, we determined the relative phosphorylation pattern of CaMKIIα and MAPK through quantification of neural p-CaMKIIα and p-MAPK/ErK, and IGF-1R expression for P80, P365 and P730 brain samples. In addition to a change in synaptic function, our results show a decrease in neural IGF-1/IGF-1R expression in whole brain, hippocampus and cortex of aged mice. This was associated with a significant upregulation of phosphorylated neural MAPK (p-MAPK) and decrease in total brain CaMKIIα (i.e., CaMKIIα and p-CaMKIIα) in the aged brain. Taken together, we showed that brain aging is associated with a change in neural IGF-1/IGF-1R expression and may be linked to a change in phosphorylation of synaptic kinases (CaMKIIα and MAPK) that are involved in the modulation of LTP.

Keywords: IGF-1/IGF-1R, aging, CaMKIIα, MAPK/ErK, KCa2.2

## INTRODUCTION

Brain aging has been implicated in the cause and progression of disease conditions, characterized in part by memory deficits (Reagh and Yassa, 2017; Seeley, 2017), notably in disorders like Alzheimer's disease (Heckman et al., 2017; Vemuri et al., 2017; Caballero et al., 2018). One underlying cause of the behavioral changes associated with normal or disease-related aging is modifications to synaptic morphology and molecular composition (Bertoni-Freddari et al., 1988, 1992, 1993), which often lead to neuronal cell death, oxidative stress and cytoskeletal defects, which are implicated in age-linked disorders (Wilson et al., 2016, 2017; Pellegrini et al., 2017; Salvadores et al., 2017).

Neurotrophic factors and receptors have important roles during development and in the adult brain. During development, neural progenitor cells develop cytoskeletal structures (neurites) required for cell migration in the developing brain (Chou and Wang, 2016; Hanamura, 2017). In the absence of neurotrophic factors, neuronal development is impaired (Sanford et al., 2008; Park and Poo, 2013). Depending on the stage of development, depletion of neurotrophins may halt neural cell migration and formation of synapses in several brain circuits (Park and Poo, 2013; Chou and Wang, 2016). Therefore, the process of neural circuit formation is a set of molecular events governed by neurotrophic activation of neurotropin receptors (Mousa and Bakhiet, 2013; Bertrand, 2017). Similarly, in the adult nervous system, neurotrophic factors and receptors are required for the maintenance of active synapses (Gómez-Palacio-Schjetnan and Escobar, 2008; Ito-Ishida et al., 2008; Garcia et al., 2012; Ivanov, 2014). There are several neurotropic factors in the brain; insulin-like growth factor-1 (IGF-1), nerve growth factor (NGF) and, brain derived neurotropic factor (BDNF) among others (Yuen et al., 1996; Park and Poo, 2013; Song et al., 2017; Zegarra-Valdivia, 2017), which have associated receptors, such as Insulin-like growth factor 1 receptor (IGF-1R), IGF-1 receptor 2 (IGF-2R), and Tyrosine kinase receptors (RTkA and RTkB; Mousa and Bakhiet, 2013; Dyer et al., 2016; Bertrand, 2017).

IGF-1 and IGF-1R are particularly important because of their role in neurodevelopment, synaptic function and aging (Bartke et al., 2003; Sonntag et al., 2005; Chiu and Cline, 2010; Dyer et al., 2016). As such, changes in their expression pattern have been implicated in the pathophysiology of developmental and age-related neuropsychiatric disorders (Deak and Sonntag, 2012; Green et al., 2014; Dyer et al., 2016). In addition to their involvement in formation of synapses, IGF-1 and IGF-1R act to maintain synapses in the adult brain (Chiu and Cline, 2010; Gazit et al., 2016; Nieto-Estévez et al., 2016; Decourtye et al., 2017; Reim and Schmeisser, 2017). Notably, IGF-1/IGF-1R signaling may alter the activity of proteins directly involved in synaptic plasticity, cognitive and memory function (Bartke et al., 2003; Sonntag et al., 2005; Deak and Sonntag, 2012).

The role of IGF-1—and other neurotropic factors such as BDNF—in neuronal development and synaptic plasticity has been described extensively (Nieto-Estévez et al., 2016; Reim and Schmeisser, 2017). However, a recent study demonstrated that IGF-1R is directly involved in the regulation of presynaptic Ca2<sup>+</sup> release during long-term potentiation (LTP) in the hippocampus (Gazit et al., 2016). Therefore, both IGF-1 and IGF-1R can directly modulate specific aspects of cognition and memory function in the hippocampus (Sonntag et al., 2005; Deak and Sonntag, 2012). IGF-1-mediated activation of neurotropin receptors, and IGF-1R activation (by IGF-1 or insulin) involves signaling of downstream proteins (Hiney et al., 2009; Liu et al., 2015; Law et al., 2017). These kinases are involved in several pathways associated with synaptic function, growth, inflammation and metabolism (Schumacher et al., 1991; Mynarcik et al., 1997; Siddle, 2011; Fernandez and Torres-Alemán, 2012).

IGF-1 activation of insulin receptor or IGF-1R can initiate Ras/ErK signaling (Lopaczynski, 1999; Moelling et al., 2002; Dyer et al., 2016). Furthermore, Ras/Raf signaling can modulate the phosphorylation of synaptic regulatory calcium-calmodulindependent kinase 2 alpha (CaMKIIα; Villalonga et al., 2001; Illario et al., 2003; Wu et al., 2011; DiBattista et al., 2015). Both mitogen activated protein kinase (MAPK/ErK) and CaMKIIα are likely colocalized at synaptic densities (Giovannini et al., 2001; Tsui et al., 2005). As such, the phosphorylation status of these proteins may alter hippocampal LTP and depression (LTD; Giovannini et al., 2001; Derkach et al., 2007). Therefore, a change in IGF-1/IGF-1R may affect synaptic function by altering the balance between synaptic MAPK/ErK and CaMKIIα activity.

CaMKIIα and MAPK/ErK act downstream of IGF-1/IGF-1R in various signaling pathways already described in neurons (Chiu and Cline, 2010; Song et al., 2010; Zuloaga et al., 2013). CaMKIIα controls LTP by regulating ionotropic receptors and ion movement at post-synaptic densities (PSDs; Wang and Kelly, 2001; Hinds et al., 2003; Mao et al., 2014). Since MAPK/ErK is co-localized with CaMKIIα at PSDs, it can alter the synaptic activity of CaMKIIα by increased phosphorylation (Giovannini et al., 2001; Tsui et al., 2005; Derkach et al., 2007). Physiologically, the LTP process is associated with a synchronous oscillation of Ca2<sup>+</sup> and K<sup>+</sup> ions (Bacci et al., 1999; Power et al., 2002; Allen et al., 2011). During LTP, CaMKIIα increase Ca2<sup>+</sup> currents from inotropic glutamate receptor activation (Sanz-Clemente et al., 2013; Mao et al., 2014; DiBattista et al., 2015) and inhibits small ion conductance channels, such KCa2.2 (Hammond et al., 2006; Lin et al., 2010; Griffith et al., 2016). Conversely, MAPK/ErK can inhibit CaMKIIα (Giovannini et al., 2001), while activating the pore forming sub-units of calcium-activated potassium (KCa2.2) channels (Schrader et al., 2006; Turner and Shieh, 2006).

KCa2.2 channels generates prolonged low-tone K<sup>+</sup> currents during LTP (Kim and Hoffman, 2008) and the hyperpolarization phase of the action potential (Power et al., 2002; Hammond et al., 2006; Lin et al., 2010). An increase in KCa2.2 activity reduces the threshold of the action potential due to a sustained after-hyperpolarization effect (Power et al., 2002; Stocker, 2004; Stocker et al., 2004; Lin et al., 2010). Thus, age-linked neural changes, which promotes loss of CaMKIIα function, can upregulate KCa2.2 activity through disinhibition of this channel. Additionally, age-dependent increase in Ras/ErK activation can promote KCa2.2 activity by attenuating (phosphorylating) CaMKIIα-linked inhibition of KCa2.2. Furthermore, Ras-ErK signaling can activate (phosphorylate) pore forming subunits of KCa2.2.

Therefore, we asked whether an age-related change in IGF-1/IGF-1R axis is related to changes in synaptic function through age-related alterations of MAPK/ErK/CaMKIIα and KCa2.2 in the hippocampus and PFC (**Figure 1**)? A decline in IGF-1 and IGF-1R expression has been described in age-related neuropsychiatric and degenerative diseases (Carro et al., 2002; Yaghmaie et al., 2006; Piriz et al., 2011; Puche and Castilla-Cortázar, 2012; Green et al., 2014; Werner and LeRoith, 2014). Therefore, in this study we assessed the differential expression of these synaptic kinases (i.e MAPK/ErK and CaMKIIα) with age in the hippocampus and medial prefrontal cortex (mPFC). Additionally, we examined the relationship between age-related change in neural MAPK/ErK/CaMKIIα activity and expression of synaptic markers—neurotransmitter transporters—in the hippocampus (CA1) and mPFC.

## MATERIALS AND METHODS

## Animal Strain

Male C57BL/6 VGAT-Venus mice of the following age groups were used for this study; postnatal days 80 (P80; young adult; n = 11), P365 (middle aged; n = 10), P730 (elderly; n = 8). The vesicular GABA transporter (VGAT) Venus mice have been previously developed and characterized by Wang et al. (2009). The animals (i.e., VGAT-Venus) used to establish the colony for this study were obtained from Dr. Janice Nagle at Wesleyan University and bred at the vivarium of the Louisiana State University School of Veterinary Medicine. VGAT-Venus mice are transgenic mice bred on a C57BL/6J background and carry no mutations or abnormalities. These mice express a fluorescence protein called Venus (a modified yellow fluorescence protein developed by Atsushi Miyawaki at RIKEN, Wako, Japan) in inhibitory GABAergic and Glycinergic neurons (Wang et al., 2009). VGAT-Venus mice were used because it enables a simple method for assaying changes in inhibitory neuronal composition through immunofluorescence imaging (Lee et al., 2015).

All animals used for this experiment weighed between 22–27 grams. Animals were kept under standard laboratory conditions and handled in accordance to NIH guidelines for animal care and use in research. All protocols used were reviewed and approved by the Institutional Animal Care and Use Committee of the Louisiana State University School of Veterinary Medicine.

## Sample Preparation

After the brains were collected, the right and left hemisphere were used for immunofluorescence or immunoblotting preparations, respectively.

## Immunofluorescence

Animals were deeply anesthetized via inhalation of isoflurane in an enclosed chamber, then perfused transcardially through the left ventricle using 10 mM phosphate buffered saline (PBS). The right half of the brain was collected and rapidly fixed in 4% PB paraformaldehyde (PFA) overnight at 4◦C. Subsequently, the fixed brain samples were transferred into 4% PB-PFA containing 30% sucrose for cryopreservation. Cryopreservation was performed at 4◦C for 72 h. Free-floating cryostat sections (20 µm thick) were obtained using a Leica Cryostat and collected in 10 mM PBS at 4◦C. The sections were washed three times (5 min each) in 10 mM PBS (pH 7.4) on a tissue rocker. Blocking was done in normal goat serum (Vector Labs), prepared in 10 mM PBS+0.03% Triton-X 100, for 2 h at room temperature. The sections were incubated in primary antibody solution overnight at 4◦C [Rabbit anti-IGF-1R (1:100; ThermoScientific-MA5- 15148), Mouse anti SK2.2 (EMD Millipore Q2650573; 1:250), Rabbit anti-MAPK/ErK1/ErK2 (1:100; Cell Signaling-#9102) and Mouse anti- CaMKIIα (1:100; Cell Signaling-#50049)]. The primary antibodies were diluted appropriately in 10 mM PBS, 0.03% Triton-X 100 and normal goat serum. It is important to note that VGAT-Venus mice express Venus in inhibitory GABAergic and Glycinergic neurons: therefore, no staining was necessary for Venus fluorescence observation. Subsequently, the sections were washed as previously described and incubated in secondary antibody solution [Goat anti Rabbit 568, Goat anti Rabbit 594 and Goat anti Mouse 568 (diluted at 1:1000) prepared in 10 mM PBS, 0.03% Triton X-100 and Normal Goat Serum] at room temperature (1 h). Immunolabeled sections were washed and mounted on gelatin-coated slides using a plain or DAPI containing anti-fade mounting medium (Vector Labs).

## Confocal Microscopy

Imaging of immunolabeled proteins in the hippocampus and cortex was performed by confocal microscopy (Olympus FluoView 10i). Fluorescence intensity was estimated for CaMKIIα, IGF-1R, MAPK/ErK and KCa2.2 using ImageJ (Burgess et al., 2010; McCloy et al., 2014). In addition, cell counting was conducted to determine the distribution of Venusexpressing neurons per unit area in the hippocampus (CA1- DG field) and mPFC (Layer V) using ImageJ (Grishagin, 2015). Fluorescence quantification and cell counting was conducted in n = 10 fields for n = 6 consecutive brain (serial) sections per animal. The average fluorescence intensity and cell count was determined and compared for all groups in One-Way analysis of

variance (ANOVA) with Tukey post hoc test. Statistical analysis was performed in GraphPad Prism Version 7.0.

## Immunoblotting

The left whole brain was rapidly frozen and homogenized using a low speed hand-held homogenizer. The brain homogenate was centrifuged at 12,500 g for 15 min (4◦C) to isolate whole brain lysate. Fifteen microgram of protein (3 biological replicates for all samples), obtained from brain tissue homogenate, was processed per well. After western blotting, protein was detected using the following primary antibodies; Rabbit anti IGF-1R (Cell Signaling-#3027s), Mouse anti-IGF-1 (abcam- # ab176523), Rabbit anti-MAPK/ErK1/ErK2 (1:100; Cell Signaling-#9102), Rabbit anti-phospho-MAPK/ErK1/ErK2 (Cell Signaling-#4370s; sites: Thr202/Tyr204), Mouse anti-CaMKIIα (Cell Signaling-#50049), Rabbit anti-phospho-CaMKIIα (Cell Signaling-#12716s; site: Thr286), Mouse anti SK2.2 (EMD Millipore-#Q2650573), Rabbit anti PSD-95 (Cell Signaling- #3450s), Rabbit anti Homer-1 (Proteintech-#124-33-1-AP), Rabbit anti Synaptophysin (Cell Signaling-#5461s), Rabbit anti vesicular glutamate transporter 2 (VGLUT2; abcam-#ab84103), Rabbit anti GAPDH (Cell Signaling-#5174s). Subsequently, the primary antibodies were detected with HRP-conjugated Goat anti Rabbit (Invitrogen #65-6120) and Goat anti Mouse (Invitrogen #65-6520) secondary antibodies following which the reaction was developed using a chemiluminescence substrate (Thermofisher-#34579). Protein expression was quantified and normalized with the housekeeping protein (GAPDH) and synaptic proteins (PSD-95, Homer1 and Synaptophysin) expression using Image Lab version 5.2.1 (BioRad, Hercules, CA, USA). Multiple controls were used for normalizing each protein of interest because of a general decline in neural proteins with age (Carney et al., 1991; Schimanski and Barnes, 2010). Subsequently, normalized protein expression data was analyzed through One-Way ANOVA (with Tukey Post hoc test) in GraphPad Prism Version 7.0. The outcome was presented as bar chart with error bars representing the mean ± SEM respectively.

## STATISTICS

Analysis was conducted with the GraphPad Prism Version 7.0. For the immunofluorescence results, the average fluorescence intensity and cell count was determined and compared for all groups in One-Way ANOVA with Tukey post hoc test. For immunoblotting protein expression data, One-Way ANOVA (with Tukey Post hoc test) was performed. The outcomes are presented as bar chart with error bars representing the mean ± SEM respectively.

## RESULTS

## Changes in Control Protein Expression With Age

We observed a significant decrease in expression level of control proteins—GAPDH, PSD-95 and Homer1—in brain lysate prepared from aged mice (P730) when compared with P80 and P365 brain samples. This may have resulted from age-related decrease in neural protein synthesis, or loss of protein due to oxidation (Carney et al., 1991; Schimanski and Barnes, 2010). To ascertain an age-linked protein depletion, equal volume (20 µl) and proteins concentration (15 µg/well) were examined via Western blot for each sample (animal) across all groups. As such, P730 mice exhibited a significant decrease in MAPK, CaMKIIα, p-CaMKIIα, IGF-1 and IGF-1R. The results were normalized by the corresponding expression of control proteins (GAPDH, Homer1, PSD-95 and synaptophysin) from the same sample in multiple trials. Interestingly, control proteins—GAPDH, Homer1 and PSD-95–were reduced significantly in P730 brain samples. Therefore, band intensity for a protein of interest was divided by band intensity for control proteins for the same sample. Ultimately, the average expression after normalizing with various control proteins was adopted as the normalized expression for the protein. However, not all proteins were reduced in the aged brain. Synaptophysin (a control) and p-MAPK/ErK were significantly upregulated in P730 brain lysates when compared with P80 and P365 samples. Therefore, a decrease in a control protein does not connote a decrease in protein sample loaded for the P730 group.

## Age-Dependent Change in Neural IGF-1 and IGF-1R Expression

First, we evaluated the distribution of IGF-1R in immunolabeled brain sections containing the cortex and hippocampus. IGF-1R expression was estimated through quantification of fluorescence (see ''Materials and Methods'' section). For this procedure, a constant exposure time and contrast was adopted for all sections, relative to background staining. We observed an age-dependent decrease in hippocampal IGF-1R expression when P365 (p < 0.001) and P730 (p < 0.001) mice were compared with their P80 counterparts (**Figures 2A,C**). In addition, there was a significant reduction of IGF-1R staining in the mPFC at P365 (p < 0.001) and P730 (p < 0.001) when compared with the P80 mice (**Figures 2B,D**).

In quantitative immunoblotting of whole brain lysate, there was an age-dependent decrease in neural IGF-1 level when P730 mice were compared with P80 and P365 mice (**Figures 2E,F**; p < 0.001). Interestingly, there was no significant change in neural IGF-1 level at P365 when compared with P80 IGF-1 expression. Similar to the observations from confocal quantification, there was a significant decrease (p < 0.001) in IGF-1R expression with age in whole brain lysate of P365 and P730 mice (**Figures 2E,F**). Furthermore, a significant decline (p < 0.001) was observed when comparing P730 with P365 neural IGF-1R expression (**Figures 2G,H**). Based on these outcomes, we deduced that a decline in both IGF-1 and IGF-1R are associated with aging in the hippocampus and cortex. While IGF-1 decline occurred later (P730) than IGF-1R depletion in the hippocampus and cortex (P365). This outcome is based on the age ranges adopted for this study. P80 expression was used as a baseline for immunofluorescence and immunoblotting analysis.

## Phosphorylation of Neural MAPK/ErK Increased With Age

Using confocal imaging and quantification techniques, we estimated and normalized the expression of MAPK/ErK (fluorescence) in whole brain sagittal sections (**Figure 3A**). In P365 vs. P80 mice, we recorded a significant increase in prefrontal cortical and hippocampal MAPK/ErK expression (**Figure 3B**; p < 0.001). At P730, MAPK/ErK expression varied between the hippocampus and mPFC (**Figure 3C**). Hippocampal MAPK/ErK expression increased at P730 when compared with P365 (p < 0.001) and P80 (p < 0.001), but reduced significantly in the mPFC when P730 mice were compared with P365 (**Figure 3C**; p < 0.001). However, mPFC MAPK/ErK expression at P730 was higher than the baseline (vs. P80; p < 0.05). Therefore, based on our hypothesis, a change in neural IGF-1/IGF-1R expression may be associated with the region-specific change in MAPK/ErK expression. As such, a reduction in hippocampal and mPFC IGF-1R expression was accompanied by an increase in MAPK/ErK.

Since MAPK/ErK is active in its phosphorylated form (Ferrer et al., 2001; Hoofnagle et al., 2004), we compared the distribution of MAPK/ErK and phosphorylated MAPK/ErK in whole brain lysates from P80, P365 and P730 mice. Subsequently, we determined the percentage phosphorylation of neural MAPK/ErK by comparing GAPDH-normalized expression of MAPK/ErK and p-MAPK/ErK in whole brain lysates [p-MAPK/(p-MAPK+MAPK) × 100]. At P365, there was no significant change in neural MAPK/ErK expression when compared with P80 (baseline) MAPK/ErK expression (**Figures 3D,E**). However, total brain MAPK/ErK expression reduced significantly at P730 (**Figures 3D,E**; p < 0.001). In subsequent analysis, we found age-related differences in non-phosphorylated to phosphorylated forms of MAPK/ErK (**Figures 3F,G**). As such, in P730 mice there was a significant increase in p-MAPK/ErK in total brain lysate when compared with P80 (p < 0.05) and P365 (p < 0.001). Based on these outcomes, the percentage of normalized phosphorylated MAPK/ErK was 85% for P730 mice when compared with P80 (47%) and P365 (25%; **Figure 3H**).

From these outcomes, we deduce that a change in IGF-1/IGF-1R signaling may be linked with an increased conversion of MAPK/ErK to p-MAPK/ErK in the brain of aged mice. It is important to note that immunoblot outcomes for protein expression gives brain-specific expression, while confocal imaging depicts region specific (mPFC: Layer V and CA1) expression. Although total brain MAPK/ErK did not change at P365 (**Figure 3E**), mPFC and CA1 MAPK/ErK expressions increased significantly when compared with P80 (**Figures 3B,C**). Likewise, in spite of a decrease in brain MAPK/ErK at P730 (**Figure 3E**), hippocampal and cortical distribution of the proteins were significantly higher than what was recorded at P80 (**Figures 3B,C**). Furthermore, mPFC expression reduced (p < 0.001; **Figure 3C**), while CA1 expression increased, significantly at P730 (p < 0.001) when compared with P365 levels.

## Age-Linked Depletion of Brain CaMKIIα

Based on our hypothesis (**Figure 1**), an increase in the expression of brain p-MAPK/ErK may alter synaptic CaMKIIα function through phosphorylation (inactivation). In support of this proposition, an increase in the percentage of brain p-MAPK/ErK

hippocampal IGF-1R expression when we compared P365 and P730 hippocampus. (E,F) Quantitative western blots showing a significant decrease in insulin-like growth factor-1 (IGF-1) expression in total brain lysate at P730 when compared with P80 (p < 0.001) and P365 (p < 0.001). No significant change in IGF-1 was observed at P365 when compared with P80. Protein expression per lane was normalized with GAPDH in 15 µg total protein for P80, P365 and P730 groups. (G,H) IGF-1R expression reduced in total brain lysate at P365 (p < 0.001) when compared with P80. There was a further decrease in neural IGF-1R expression at P730 (p < 0.001) when compared with P80 and P365. ∗∗p < 0.01, ∗∗∗p < 0.001.

was associated with a significant decrease in CA1 and mPFC CaMKIIα expression with age (**Figures 4A,B**). Normalized fluorescence intensity for immunolabeled CaMKIIα reduced significantly in the hippocampus at P365 (p < 0.001) and P730 (p < 0.001) when compared with the control (**Figure 4C**). Likewise, there was a significant reduction in prefrontal cortical expression of CaMKIIα for P365 and P730 mice when compared with P80 (p < 0.001; **Figure 4D**). The outcome for confocal fluorescence quantification was further confirmed through immunoblot quantification of CaMKIIα in whole brain lysate. As such CaMKIIα expression reduced significantly in the lysate prepared from P365 and P730 mice brains when compared with P80 brain lysate in immunoblotting (**Figures 4E,F**; p < 0.001). A further decline in brain CaMKIIα was observed at P730; when compared with P365 CaMKIIα expression (p < 0.05; **Figure 4F**).

Since p-MAPK/ErK can phosphorylate synaptic CaMKIIα (Giovannini et al., 2001; Tsui et al., 2005), we evaluated the significance of increased p-MAPK/ErK (P730) on the normalized

Equally, MAPK/ErK expression increased in the mPFC of P730 mice when compared with P80 (p < 0.05). (D) Immunoblots showing the expression of MAPK/ErK in total brain lysate for P80, P365 and P730 groups. (E) Bar chart showing a decrease in normalized expression of MAPK/ErK in the brain of P730 mice. (F–H) Quantitative immunoblots showing a change in the expression of phosphorylated MAPK/ErK at P365 and P730. Percentage p-MAPK/ErK increased significantly in the P730 (85%; p < 0.001) group when compared with P80 (47%) and P365 (25%). <sup>∗</sup>p < 0.05, ∗∗∗p < 0.001.

expression of p-CaMKIIα in whole brain lysate. In addition to a decrease in CaMKIIα, P365 and P730 mice exhibited a significant depletion of neural p-CaMKIIα when compared with P80 expression (p < 0.001 and p < 0.05 respectively; **Figures 4G,H**). CaMKIIα expression decreased significantly in P730 brain lysate when compared with P365 expression (p < 0.05; **Figures 4E,F**). Conversely, p-CaMKIIα expression increased significantly in P730 brain when compared with P365 expression (p < 0.01; **Figures 4G,H**). In subsequent analysis, we determined percentage phosphorylation of CaMKIIα by comparing normalized expression of CaMKIIα and p-CaMKIIα for P80, P365 and P730 brain lysates [p-CaMKIIα/(p-CaMKIIα+CaMKIIα) × 100] (**Figure 4I**). Interestingly, there was a significant increase in percentage phosphorylated CaMKIIα at P730 (p < 0.05) when compared with P80 and P365 groups. Taken together, our results show a significant decrease in total CaMKIIα (p < 0.05) at P365. However, for P730 brain, percentage phosphorylated CaMKIIα was upregulated in addition to a decrease in neural CaMKIIα expression (**Figures 4F–I**).

## Differential Expression of Small-Ion Conductance Channels KCa2.2 in the Hippocampus-PFC Axis

We hypothesized that CaMKIIα regulation of KCa2.2 may be altered because of a decrease in neural CaMKIIα expression (P356 and P730) and increased percentage of

phosphorylated CaMKIIα in the P730 brain (**Figures 4E–I**). Moreover, in addition to increased CaMKIIα phosphorylation by p-MAPK/ErK, the protein (i.e., p-MAPK/ErK) may directly activate the pore forming subunit of KCa2.2 (**Figure 1**). Owing to a decrease in CaMKIIα expression or an increased CaMKIIα phosphorylation, the activity of KCa2.2 may become upregulated. Consequently, either downregulation of CaMKIIα or an increased p-MAPK/ErKmediated KCa2.2 phosphorylation would lead to attenuation of synaptic potentials. This may be related to prolonged after-hyperpolarization currents that are linked to increased KCa2.2 activity at synapses (K<sup>+</sup> ion movement; **Figure 1**).

In KCa2.2 immunolabeled sections, we found that KCa2.2 expression varied in the hippocampus and cortex (**Figures 5A,B**). In the hippocampus, normalized fluorescence intensity of immunolabeled KCa2.2 (expression) increased significantly at P365 (p < 0.05) and P730 (p < 0.001) when compared with P80 (**Figure 5C**). Additionally, we observed an age-linked increase in hippocampal KCa2.2 expression when comparing P365 with P80 (p < 0.05), and P730 vs. P365 (p < 0.01; **Figure 5C**). The outcome for the hippocampal expression of IGF-1R, MAPK, CaMKIIα and KCa2.2 support our hypothesis. As such a change in hippocampal IGF-1R was associated with increased MAPK phosphorylation (**Figure 3B**), depleted CaMKIIα (**Figure 4C**) and upregulated KCa2.2 expression (**Figure 5C**).

Surprisingly, decreased cortical CaMKIIα (**Figure 4D**) and increased MAPK/ErK (**Figure 3C**) was associated with reduced KCa2.2 expression for the P365 mice (p < 0.001); when compared with P80 mice (**Figure 5D**). However, there was a significant increase in cortical KCa2.2 expression at P730 (p < 0.01) when compared with P365 expression levels. This outcome agrees partially with our hypothesis, since the mPFC exhibits a distinct pattern of KCa2.2 expression when compared with the hippocampus in aged mice. In the mPFC, CaMKIIα expression was significantly sustained at P365 when compared with P730 (p < 0.05; **Figures 4E,F**). Conversely, the expression of phosphorylated CaMKIIα was significantly lower in P365 brain lysate when compared with P730 expression (p < 0.01; **Figures 4G,H**). Pertaining to CaMKIIα regulation of KCa2.2 function, we deduced that a decrease in cortical CaMKIIα expression at P365 was not sufficient to cause upregulation of KCa2.2 in the mPFC. Rather, a decreased expression together with an increased percentage phosphorylation (inactivation) of CaMKIIα might have contributed to upregulation of cortical KCa2.2 at P730 (**Figure 5D**). Taken together, we infer that decreased expression and increased phosphorylation of CaMKIIα may be associated with dysregulation of KCa2.2 in the CA1 and mPFC.

In subsequent analysis, we found that neural KCa2.2 did not significantly change at P365 in whole brain lysate (**Figures 5E,F**). However, it is important to note that regional variations may occur, as observed for the hippocampus and cortex through immunohistochemical methods. As such, KCa2.2 expression increased in the hippocampus but decreased in the mPFC at P365 (**Figures 5C,D**) when assessed through immunohistochemistry. At P730, there was a significant decrease in whole brain KCa2.2 expression (**Figure 5F**; p < 0.001) although hippocampal KCa2.2 (confocal) expression remained significantly higher when compared with P80 (p < 0.001) and P365 (p < 0.01). In order to ascertain a change in post-synaptic profile, we evaluated

(E,F) Quantitative western blots showing a decreased KCa2.2 expression in total brain lysate at P730 when compared with P80 or P365 (p < 0.001). (G–I) Quantitative western blots showing a significant decrease in post-synaptic markers in P730 total brain lysate. PSD-95 (p < 0.001) and Homer-1 expression (p < 0.001) reduced significantly for the P730 group when compared with P80 or P365 mice. <sup>∗</sup>p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

the expression of post-synaptic structural proteins that are closely related to synaptic function, KCa2.2 and CaMKIIα expression at PSDs densities. In support of our results, there were significant changes in the expression of PSD-95 and Homer1 in whole brain lysate of aged (P730) animals when compared with P80 and P365 expression (p < 0.001; **Figures 5G–I**).

## Synaptic Excitatory and Inhibitory Transport

Since IGF-1R is involved in the modulation of presynaptic function (Gazit et al., 2016), we compared age-linked changes in the expression of IGF-1R and presynaptic proteins associated with vesicle and neurotransmitter transport. Furthermore, we highlighted possible links between dysregulated IGF-1R- CaMKIIα-KCa2.2 function and excitatory/inhibitory neurotransmitter transporter protein expression in the CA1 and mPFC regions. In addition to age-linked decreases in IGF-1R, there was a significant decrease in the count of inhibitory GABAergic and Glycinergic neurons expressing VGAT (**Figures 6A–D**). In the hippocampus (CA1-DG field), the count of VGAT-Venus neurons decreased at P365 (p < 0.001) and P730 (p < 0.001) when compared with P80 count (**Figures 6A,C**). Likewise, VGAT-Venus neuron count decreased in the mPFC of P365 and P730 mice when compared with P80 scores (**Figures 6B,D**). Equally, there was a significant loss of VGLUT2 in whole brain lysate of aged mice (**Figures 6E,F**). This suggests a significant change in presynaptic morphology; similar to changes in post-synaptic protein expression described previously (PSD-95 and Homer1; **Figures 5G–I**). Interestingly, synaptophysin, a presynaptic protein, increased in total brain lysate at P730 when compared with P80 (**Figures 6E,F**; p < 0.001) and P365 expression (p < 0.01). Although IGF-1R is known to mediate the synaptic activity of synaptophysin (Gazit et al., 2016), our results suggest an inverse relationship for this interaction. As such, a decrease in neural IGF-1R was accompanied by an increase in synaptophysin expression. This may represent a compensatory mechanism for the loss of synaptic function in the aging brain. Based on these outcomes, we deduced that loss of IGF-1R signaling in the aged brain may be linked to depletion of post-synaptic proteins, dysregulation of IGF-R-linked presynaptic neurotransmitter transport, and synaptophysin activity.

## DISCUSSION

The role of IGF-1 and IGF-1R have been extensively described in the pathophysiology of age-related brain disorders and developmental synaptic dysfunction (van Dam and Aleman, 2004; Chiu and Cline, 2010; Fernandez and Torres-Alemán, 2012; Dyer et al., 2016; Reim and Schmeisser, 2017; Wrigley et al., 2017). The outcome of this study demonstrates that age-related changes in IGF-1/IGF-1R activity may be associated with dysregulated synaptic MAPK/ErK and CaMKIIα function. We found that an age-related decrease in IGF-1/IGF-1R expression was associated with reduction of neural CaMKIIα expression and increased MAPK/ErK phosphorylation in the brain. Based on our hypothesis (**Figure 1**), the physiological implication of these outcome may involve increased activity

of small ion conductance channels (KCa2.2) at hippocampal PSDs, and decreased expression in the mPFC of aged mice. Increased KCa2.2 expression in the hippocampus may occur due to reduced CaMKIIα-dependent KCa2.2 inhibition, increased CaMKIIα phosphorylation (inactivation) by p-MAPK/ErK, and upregulated phosphorylation (activation) of KCa2.2 by p-MAPK/ErK. Taken together, our results suggest that age-linked changes in IGF-1R signaling may alter synaptic KCa2.2 regulation by disrupting the balance of regulatory synaptic proteins, CaMKIIα and MAPK/ErK.

and P730 (p < 0.001) when compared with P80 expression. <sup>∗</sup>p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

## IGF-1/IGF-1R-Linked Alteration in Synaptic Kinases

IGF-1R is activated by endogenous IGF-1 (Dyer et al., 2016; Gazit et al., 2016). This interaction contributes to the regulation of presynaptic Ca2<sup>+</sup> signaling and Ca2<sup>+</sup> release from the mitochondria (Gazit et al., 2016), N-type, and L-type calcium channels (Blair and Marshall, 1997). Although, synaptic function (LTP) involves a synchronous oscillation of Ca2<sup>+</sup> and K<sup>+</sup> ions (Bacci et al., 1999; Power et al., 2002), the effect of IGF-1/IGF-1R signaling on the activity of calcium-dependent potassium channels (KCa2.2) is poorly understood. Furthermore, how an age-dependent change in IGF-1/IGF-1R signaling contributes to dysregulation of synaptic KCa2.2 function in the aged brain has yet to be investigated. In this study, we described some of the possible pathways through which a change in IGF-1R signaling can alter synaptic KCa2.2 activity in the aging cortex and hippocampus.

## Alterations in Synaptic Kinases and KCa2.2 Expression

Our results indicate that a decrease in IGF-1 and IGF-1R with age was associated with a significant change in the expression and phosphorylation of synaptic kinases involved in synaptic function. Hippocampal and prefrontal cortical IGF-1R expression decreased by middle age as seen in P365 mice (**Figures 2C,D**). However, depletion of brain IGF-1 occurred much later in P730 brain lysate (**Figures 2G,H**). To test our hypothesis, we evaluated the significance of age-linked IGF-1/IGF-1R alteration on the relative expression of MAPK/ErK and CaMKIIα in the hippocampus and mPFC of mice. In addition to acting downstream of IGF-1R (Chiu and Cline, 2010; Deak and Sonntag, 2012; Dyer et al., 2016), both proteins (i.e., MAPK/ErK and CaMKIIα) are involved in the regulation of neurotransmitter receptors and ion channels at synapses (Giovannini et al., 2001; Tsui et al., 2005). Moreover, previous studies have described co-localization of MAPK/ErK and CaMKIIα at post-synaptic sites (Giovannini et al., 2001; Tsui et al., 2005; Hammond et al., 2006). Since IGF-1R regulates MAPK/ErK and CaMKIIα through the Ras/Raf/ErK pathway, a change in IGF-1R signaling may alter the synaptic activity of MAPK/ErK and CaMKIIα. In aging, a decline in IGF-1R may cause an increase in MAPK phosphorylation. Thus, an increase in activated p-MAPK/ErK, can facilitate phosphorylation (inactivation) of CaMKIIα thereby disrupting synaptic function (Ferrer et al., 2001; Giovannini et al., 2001; Hoofnagle et al., 2004).

Age-dependent changes in IGF-1/IGF-1R are associated with dysregulation of MAPK/ErK and CaMKIIα expression (**Figures 3**, **4**). MAPK/ErK and CaMKIIα modulates NMDARlinked calcium currents (Hinds et al., 2003; Mao et al., 2014) and calcium-activated potassium channels (KCa2.2) during LTP (Giovannini et al., 2001; Hammond et al., 2006; Lin et al., 2010). Therefore, we examined the differential expression of KCa2.2 in the hippocampus and cortex of aged mice, characterized by a decrease in IGF-1/IGF-1R and altered MAPK/CaMKIIα expression (**Figure 5**). Our results show that an increase in p-MAPK/ErK, and decreased CaMKIIα was associated with upregulated KCa2.2 expression in the hippocampus of aged mice. Although previous studies show a change in Ca2<sup>+</sup> currents relative to IGF-1R function (Blair and Marshall, 1997; Gazit et al., 2016), here we propose a possible mechanism through which IGF-1R might alter K<sup>+</sup> current through the regulation of substrates that modulate synaptic KCa2.2 function in aging.

The physiological implication of increased neural KCa2.2 activity has been described previously by Hammond et al. (2006). They demonstrate that an increase in KCa2.2 activity abolishes synaptic potentials and reduced memory function in mice (Hammond et al., 2006; Maingret et al., 2008; Lin et al., 2010). In separate studies, changes in the expression of KCa2.2 in young mice precipitated a decline in memory formation and retrieval (Stackman et al., 2002; Hammond et al., 2006). Evidently, the expression and activity of KCa2.2 represents a crucial part of synaptic regulation and LTP (Disterhoft and Oh, 2006, 2007). Expression of KCa2.2 and other post-synaptic proteins—PSD-95 and Homer1–show significant declines with age in the brain. Thus, differential hippocampal and cortical KCa2.2 expression suggest adaptive physiological changes in normal synaptic aging. Our results suggest that decreased CaMKIIα, increased CaMKIIα phosphorylation, and increased MAPK/ErK phosphorylation are possible causes of synaptic KCa2.2 dysregulation that may be linked to decreased IGF-1R signaling in the normal aging brain.

## Neurotransmission

In addition to a change in KCa2.2 expression and post-synaptic proteins (Homer1 and PSD-95), we found a significant change in presynaptic protein expression in the aged brain. While previous studies have shown the role of IGF-1R in the presynaptic activity of synaptophysin (Gazit et al., 2016), here we showed an inverse relationship between IGF-1R and synaptophysin expression in the aged (P730) brain. Our results indicate that a decrease in IGF-1R was associated with a significant increase in synaptophysin expression with age (**Figures 6G,H**). Conversely, there was a decrease in the expression of Venus in VGAT-positive puncta (confocal) and VGLUT2 (immunoblots) in the brain of aged mice (P730). From these outcomes, we deduced that an increased p-MAPK/ErK can increase the activity of synaptophysin independent of neurotransmitter transport. Yokomaku et al. (2003) support this proposition by showing that MAPK signaling inhibitors suppressed synaptophysin function in cultured neurons. Synaptophysin expression and activity were rescued by estradiol-mediated MAPK/ErK increase in vitro (Yokomaku et al., 2003). Since VGAT and VGLUT2 exist in the presynaptic area (Zander et al., 2010), it is logical to speculate that a change in CaMKIIα control of KCa2.2 function may alter the synaptic expression and activity of VGAT and VGLUT2 (Trimmer, 2015), although the mechanism remains to be resolved. As such, selective activation of IGF-1R and inhibition of KCa2.2 function are possible intervention methods for attenuating CaMKIIα loss and synaptic dysfunction in aging.

## CONCLUSION

Taken together, the outcome of this study showed that neural IGF-1/IGF-1R expression is reduced with age in the hippocampus and cortex. IGF-1/IGF-1R depletion is linked to increased neural MAPK/ErK phosphorylation and CaMKIIα depletion in the aged brain. We showed that loss of IGF-1/IGF-1R was also associated with a change in the synaptic expression of KCa2.2; especially in the hippocampus of aged mice. Ultimately, this might lead to a decline in presynaptic neurotransmitter function and loss of post-synaptic proteins.

## AUTHOR CONTRIBUTIONS

OMO, CCL and JF conducted specific aspects of the research. CCL, JF and RGG supervised the manuscript write up and analysis of data. JP and OMO completed manuscript write up and presentation of data and conducted additional experiments for protein analysis.

## REFERENCES


## FUNDING

This study was supported by the IBRO-ISN 2015 Fellowship and LSU SVM Faculty Start up awarded to OMO. NIH Grant R03 MH 104851, NIH R03 AG 05212 and Louisiana Board of Regents RCS Grant RD-A-09 awarded to CCL.


interaction with calmodulin. PLoS One 6:e21929. doi: 10.1371/journal.pone. 0021929


**Conflict of Interest Statement**: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Ogundele, Pardo, Francis, Goya and Lee. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Academic and Behavioral Outcomes in School-Age South African Children Following Severe Traumatic Brain Injury

Aimee K. Dollman<sup>1</sup> , Anthony A. Figaji <sup>2</sup> and Leigh E. Schrieff-Elson<sup>1</sup> \*

<sup>1</sup> Applied Cognitive Science and Experimental Neuropsychology Team, Department of Psychology, University of Cape Town, Cape Town, South Africa, <sup>2</sup> Division of Neurosurgery, Department of Surgery, School of Child and Adolescent Health, University of Cape Town, Red Cross War Memorial Children's Hospital, Cape Town, South Africa

Background: Children who have sustained severe traumatic brain injuries (TBIs) demonstrate a range of post-injury neurocognitive and behavioral sequelae, which may have adverse effects on their academic and behavioral outcomes and interfere with school re-entry, educational progress, and quality of life. These post-TBI sequelae are exacerbated within the context of a resource-poor country like South Africa (SA) where the education system is in a somewhat precarious state especially for those from disadvantaged backgrounds.

### Edited by:

Vivienne Ann Russell, University of Cape Town, South Africa

#### Reviewed by:

Miguel Angel García-Cabezas, Boston University, United States Megan Narad, Cincinnati Children's Hospital Medical Center, United States Kate Cockcroft, University of the Witwatersrand, South Africa Angelina Kakooza-Mwesige, Makerere University, Uganda

#### \*Correspondence:

Leigh E. Schrieff-Elson leigh.schrieff-elson@uct.ac.za

Received: 28 July 2017 Accepted: 28 November 2017 Published: 13 December 2017

#### Citation:

Dollman AK, Figaji AA and Schrieff-Elson LE (2017) Academic and Behavioral Outcomes in School-Age South African Children Following Severe Traumatic Brain Injury. Front. Neuroanat. 11:121. doi: 10.3389/fnana.2017.00121 Objectives: To describe behavioral and academic outcomes of a group of school-aged SA children following severe TBI.

Methods: The sample included 27 school-age children who were admitted to the Red Cross War Memorial Children's Hospital (RXH), SA, between 2006 and 2011 for closed severe TBI and who received intracranial monitoring. We collected behavioral data using the Child Behavior Checklist (CBCL) and the Behavior Rating Inventory of Executive Function (BRIEF) and academic information sourced from the BRIEF, CBCL, medical folders, and caregivers. Analyses include descriptive statistics and bivariate correlation matrices.

Results: The descriptive results show that (1) more than half of the participants experienced clinically-significant behavioral problems across the CBCL scales, (2) the working memory BRIEF subscale appeared to be the most problematic subdomain, (3) two thirds of the sample were receiving some form of, or were in the process of being placed in, special needs education, (4) there was a three-fold increase in the use of special education services from pre- to post-injury, and (5) more than half (n = 16) of the sample repeated at least one grade after returning to school post-injury. Correlation analyses results suggest that children with increased externalizing behavioral problems and executive dysfunction are more likely to repeat a grade post-injury; and that children with executive dysfunction post-TBI are more likely to require some form of special educational services.

Conclusion: While there is a vast amount of literature on pediatric TBI (pTBI) academic and behavioral outcomes, little literature exists on the pTBI population from the developing world and SA specifically. This is important to address given unique challenges that face the country and its educational system, and its implications for the management and care of children post-TBI.

Keywords: pediatric, traumatic brain injury, academic, behavior, outcome, developing countries

# INTRODUCTION

## TBIs in Developing World Countries Like South Africa (SA)

A disproportionate number of individuals who sustain traumatic brain injuries (TBIs) in the state sector are children and adolescents (Thurman, 2016). Although TBI is a global problem, and reported as a leading cause of mortality and morbidity among youth in high-income countries (HICs) the burden thereof is reportedly more potent in resource-poor countries (Hyder et al., 2007; Harris et al., 2008; Alexander et al., 2009; Kumar and Mahapatra, 2009; Bener et al., 2010; Figaji, 2017). This disproportionate effect on poorer economies is attributed to the quality of the environment and lack of resources. There are several factors often associated with resource-poor countries, such as poverty, lack of access to education, differences in infrastructure, and social problems, like alcoholism and higher road traffic accidents rates, that create greater risk for TBI (Levin, 2004; Hyder et al., 2007; Alexander et al., 2009). Further, home and work environments in indigent areas are often less secure, with residents more exposed to potential hazards. Prevention endeavors and access to rehabilitation may also be less available in such environments (World Health Organization, 2011). Other adverse factors that can impact on the incidence and/or consequences of TBI include the inconsistent scope of care for TBI survivors and health care facilities that are illprepared to cope with the degree of injury and care required for a public health problem of this magnitude (Hyder et al., 2007; Jerome et al., 2017). Hence, the burden of trauma, and in particular TBI, as the leading cause of death and neurological disability in trauma patients, is far greater in resource-poor countries than in the developed world (De Silva et al., 2009; Figaji, 2017).

The formidable economic sequelae associated with TBI results not only from the expenses associated with direct healthcare, but also from indirect costs linked to loss of the potential future productivity of that individual (Jaffe et al., 1993; Ragnarsson, 2002; Flanagan et al., 2008). The loss of potential future productivity is especially important in the case of children, clearly because most of their life will be spent in the shadow of the TBI. Also, there are associated costs, such as a loss of productivity for extended periods of time for caregivers. TBI can therefore be economically challenging and exhausting at the societal, individual and familial levels (Jaffe et al., 1993; Tilford et al., 2005; Gontkovsky et al., 2006). These factors are compounded in resource-poor environments.

Published literature on the epidemiology of pediatric TBIs (pTBIs) in developing world countries is generally limited. Researchers highlight the dearth of much-needed research of this nature generally and in SA specifically (Bruns and Hauser, 2003; De Silva et al., 2009; Haaring et al., 2011; Naidoo, 2013). Although it is suggested that the incidence of pTBI in SA must be high, exact rates are not available because systematic research on the topic is lacking (Levin, 2004; Penn et al., 2009; Corrigan et al., 2010; Naidoo, 2013; Tabish and Syed, 2015). There is little quantification of the neurological disability of survivors and the impact this has on families and state sector services.

## Education and Inequality in SA

Although SA is considered a developing country with upper to middle income levels (The World Bank, 2016), its Gini index, which represents inequality in the spread of income, is among the highest in the world. This unequal socio-economic climate, which stems from the country's apartheid history, is also clear in the school system where inequality (in terms of financial and resource provision) is rife (Du Plessis, 2001; Engelbrecht, 2006; Soudien and Baxen, 2006; Donohue and Bornman, 2014). The apartheid-based Bantu Education Act of 1953 engendered such unequal education (Donohue and Bornman, 2014; Letgotlo, 2014). Even after SA's movement to a democracy in 1994, the scars of fragmentation from past segregation and discrimination practices remain, which has long resulted in the deprivation of adequate education for large numbers of SA people (Du Plessis, 2001; Engelbrecht, 2006).

Even though some progress has been made with the SA Schools Act (SASA) of 1996 to democratize and make uniform the school system post-democracy, and despite the knowledge that all learners (as per the Bill of Rights) have the right to equity and quality education, SA's schooling system is in a precarious state (Letgotlo, 2014, South African Council for Educators SACE, 2016) . A recent news publication (The Economist, 2017) described some of the inequality (schools "with cricket pitched as smooth as croquet lawns" vs. others built from mud) and consequent dire outcomes (worrying percentages of learners who do not finish<sup>1</sup> school and not being able to read or work out basic division sums after 5–6 years of schooling) associated with some of the country's schools.

There are also reported problems with schools for learners with special education needs (LSEN<sup>2</sup> ) in the country. Donohue

<sup>1</sup>Legotlo (2014) reports that 10% of learners complete school within a reasonable period of time.

<sup>2</sup>From Du Plessis (2001) based on the 1995 White paper: "The term 'learners with special education needs' refer to all learners in need of educational support, it is learners whose special needs arise from intrinsic factors as well as learners whose needs arise from extrinsic (social, structural and systematic) factors" (p. 62).

and Bornman (2014) report that as many as 70% of LSEN who should be in school given their age, are not; most of those who are in school attend LSEN, rather than mainstream schools. This finding contrasts with the inclusive<sup>3</sup> education policy that the Department of Education (DOE) in SA have been aiming to implement, in line with the global trend of inclusive education and the Education for All initiative [United Nations Educational, Scientific, and Cultural Organization (UNESCO, 1990)]. SA's unique complex sociopolitical and economic background distinguishes it from other countries following this trend, however.

Amollo (2008) reports on a briefing by the SA DOE and their aims to reserve LSEN schools for those with severe disabilities and that as far as possible, have mainstream schools accommodate those with less severe disabilities, in a move from an exclusive toward a more inclusive educational model (Du Plessis, 2001). While a DOE policy document "White Paper 6" (Department of Education, 2001) was meant to frame and direct this transformation (Pillay and Di Terlizzi, 2009) toward inclusive education, the implementation thereof has been challenging and consequently poor (Engelbrecht, 2006; Donohue and Bornman, 2014).

In trying to understand the barriers to the implementation of inclusive education in the country, both societal and contextual factors need to be considered (Engelbrecht, 2006). Donohue and Bornman (2014) describe school-related and culture- and psychosocial-related barriers. In terms of the former, the authors conjecture that because most of the country's teaching complement are of the "older generation", that some might not have embraced or repositioned their thinking (perhaps as a function of previous training) to align with the new inclusive education strategy. They do however note that low resources remain the main obstruction, even among those who are aligned with the new inclusive policy. Regarding culture- and psychosocial-related barriers, in some communities, individuals with disabilities may be devalued and prejudiced such that schooling may not be viewed as major priority for them, as their potential is questioned. On the other hand, parents may want to protect their children with special needs from ill treatment and stigma, and may therefore choose to keep them home.

Other authors question the preparedness of the country to promote inclusive education. Pillay and Di Terlizzi (2009) note that more resources and infrastructure (facilities) are needed in mainstream schools first. From their case study, they report that despite the move toward inclusive education, LSEN schools, though few, may still be better equipped to accommodate learners with special needs than mainstream schools. According to principals of mainstream schools, some of the main challenges to inclusive quality education, is educators coping with learners with special needs who are included in mainstream schooling, managing behaviorally challenged learners and behaviors and emotions of children who fail to progress, perceived lack of support (including parental support) and training for teachers and management. Large class sizes and considerable workloads for educators were some additional barriers reported for educators (Materechera, 2014).

SA DOE audits reported on by Amollo (2008) also revealed a weak infrastructure and a host of similar and other problems in terms of some LSEN schools locally including classroom overcrowding, ill-prepared educators, and lack of necessary assistive devices. Issues with learner progression and skill recognition, nutrition, transport, and discrimination were also listed. There were also challenges in terms of the lack of LSEN schools in rural areas and closure of several LSEN schools (Amollo, 2008; Materechera, 2014). Further, principals of LSEN schools perceive lack of funding, adequately trained staff and specialists, parental support, resources, and support services, as well as stigmatization of special needs school learners by mainstream school learners and acceptance by peers, diversity in disabilities and closure of special needs schools to be the main barriers to inclusive education (Materechera, 2014).

When one considers the experience of pTBI survivors in SA, the commonly reported academic and behavioral post-TBI challenges may be exacerbated by these contextual limitations of the burden of disease and barriers to accessing appropriate education.

## TBI Outcomes

### Academic Outcomes

It is well-established globally that pTBI survivors experience impairments in a range of neurocognitive and behavioral domains, which may have adverse effects on academic outcomes (Van't Hooft, 2010; Li and Liu, 2012; Babikian et al., 2015). Changes in both academic performance and behavior can interfere with school re-entry, educational progress, and ultimately, quality of life of the injured child. These postpTBI effects extend beyond the child to their familial and social environment (Donders, 1994; Anderson and Yeates, 2010; Treble-Barna et al., 2016).

Post-pTBI cognitive sequelae include deficits in general intellectual functioning, attention, executive function, memory and learning, and language skills (Rao and Lyketsos, 2000; Mayfield and Homack, 2005; Anderson and Yeates, 2010; Yeates, 2010; Babikian et al., 2015). Academic performance depends on the integrity of these cognitive skills. For example, a child's ability to sustain their attention, learn the material presented to them and then remember it, is pertinent to successful academic progression (Arroyos-Jurado et al., 2000; Hawley, 2004; Prasad et al., 2017). Executive functions (such as working memory and inhibition)—involved in the coordination of goal-directed behavior—also play an important role in academic achievement, as well as behavioral and adaptive functioning (Anderson et al., 2002; St Clair-Thompson and Gathercole, 2006). TBI can also affect core academic skills such as reading, writing, mathematics, and spelling; however, some academic skills may be more affected than others (Taylor, 2010). Further, the often-extended absence from school during the post-injury convalescent period, which can result in less opportunity for learning, can also contribute to poor academic outcome (Ewing-Cobbs et al., 1998; Babikian and Asarnow, 2009).

<sup>3</sup> ". . . a means of education according to which the learner is schooled in the least restrictive environment possible, to overcome his or her challenges to learning and development" (Pillay and Di Terlizzi, 2009, p. 491).

Knowledge of pTBI survivors and any cognitive deficits they may experience within a classroom is important because teachers may assume that pTBI survivors are fully recovered from their injuries when no obvious physical deficits are seen and some teachers may be unaware that learners in their classroom may have sustained a TBI, particularly when the injury occurred prior to the child entering that class (Hawley, 2004; Jantz and Coulter, 2007). This may result in the lack of, or delayed implementation of, academic assistance and modifications in the classroom required by pTBI survivors (Mayfield and Homack, 2005; Jantz and Coulter, 2007).

### Behavioral Outcomes

While a child's outcome may be significantly impacted by post-TBI cognitive impairment, it is often the behavioral changes that are considered more debilitating, particularly following severe TBI, given the dose-response relationship between severity and outcome (Donders, 1994; Babikian and Asarnow, 2009; Taylor, 2010; Catroppa et al., 2012). Behavioral problems not only interfere with the functioning and educational progress of the injured child, but can also be disruptive to others in the home, community, or classroom, particularly when these problem behaviors persist over time (Savage et al., 2005; Yeates and Taylor, 2006). Behavioral impairments negatively impact school performance, by hindering both the continued development of current skills and acquisition of new skills (Keenan and Bratton, 2006; Jonsson et al., 2013; Babikian et al., 2015).

Patterns of behavioral problems following TBI may vary from one child to another and may include internalizing impairments such as anxiety, withdrawal, depression, and other emotional problems and externalizing impairments such as aggression, irritability, disinhibition, impulsivity, agitation, and distractibility (Fletcher et al., 1990; Mayfield and Homack, 2005; Yeates and Taylor, 2006; Li and Liu, 2012). These behavioral impairments may occur as a direct result of damage to the brain and resulting cognitive deficits. For example, damage to the vulnerable frontal and associated areas may lead to impaired executive functioning, including the inability to initiate tasks, self-monitor behavior, and inhibit responses (Mayfield and Homack, 2005; Yeates, 2009). Problems with behavioral inhibition may lead to a hostile environment in the classroom, especially when these problems are expressed through agitation or inappropriate (e.g., insulting) comments or actions (e.g., getting out of one's chair), which can be disruptive to other learners in the classroom (Mayfield and Homack, 2005). Behavioral dysfunction can also be an indirect consequence of the injury (Mayfield and Homack, 2005). For example, children may react negatively and act out in trying to resume daily activities while adjusting to post-injury deficits, leading to frustration when previously managed tasks become more challenging. This may emerge as children become more aware of their deficits. Negative behaviors post-injury may also occur in response to the family's reaction to the injury. Parental stress and unrealistic expectations (especially in the absence of physical injuries) may also indirectly lead to behavioral impairments (Max et al., 1999; Donders and Strom, 2000; Taylor et al., 2001; Bamdad et al., 2003; Mayfield and Homack, 2005). The demands placed on the child may then lead to increased feelings of frustration and inadequacy; such undesirable feelings consequentially reinforce poor behavior (Kinsella et al., 1999; Savage et al., 2005; Li and Liu, 2012).

Changes in academic performance and impairments in behavior can interfere with educational progress and quality of life of the injured child. The consequences of TBI can also be burdensome to the child's family and others in their social and classroom environments (Anderson and Yeates, 2010; Li and Liu, 2012). These post-TBI sequelae are exacerbated within the context of a resource-poor country like SA where the education system is in a somewhat precarious state especially for those from lower socioeconomic backgrounds. While there is a vast amount of literature on pTBI academic and behavioral outcomes, little literature exists on the pTBI population from the developing world and SA specifically. This is important to address given unique challenges that face the country and its educational system, and its implications for the management and care of children post-TBI. The purpose of this crosssectional, descriptive study was therefore to investigate and describe behavioral and academic outcomes following severe TBI in a group of South African children of school-going age.

## METHODS

## Sample

The sample included 27 children with severe TBI who were of school-going age at the time of their injury. These children were identified from a database of 137 children who had been admitted to Red Cross War Memorial Children's Hospital (RXH) for severe TBI (post-resuscitation Glasgow Coma Scale (GCS)<sup>4</sup> score of ≤8) over a 5-year period (2006–2011) and who underwent intracranial monitoring (Schrieff et al., 2013).

Participants had to be at least 1 year post-injury, because the recovery trajectory reportedly then tends to stabilize (Jaffe et al., 1995; Taylor, 2010). We excluded children who had sustained open TBIs, given the differing pathophysiology and outcomes (compared to closed TBI; Anderson et al., 2001), and those who were not attending school at the time of injury and assessment, as this limited the academic data available for those children. **Figure 1** shows the reasons for exclusion of participants, resulting in a final sample of 27 children.

## Procedure

We contacted caregivers of potential participants telephonically or approached them at their follow up Neurosurgery outpatient clinic visit if contact details were unavailable in the medical folders<sup>5</sup> . We collected the data between 2012 and 2013. Three caregivers completed the measures at their homes due to time or transport constraints; the remainder at RXH. First language

<sup>4</sup>The GCS has traditionally been used to classify severity of a TBI as mild, moderate or severe. Severe injury corresponds to a score of 8 or below on the GCS, moderate injury from 9 to 12, and mild injury a score of 13 or higher (Zillmer et al., 2008). <sup>5</sup>Hereafter referred to as "case file;" while the hospital medical folder is primarily used to store medical information and documents, non-medical information such as demographics, correspondence and school results can also be found in it.

isiXhosa-speaking research assistants assisted with interpreting when necessary.

## Measures

We had the original English versions of our measures translated into two other common local languages, Afrikaans and isiXhosa, to facilitate administration. These were linguistically validated through forward and back translations and authentication by Stellenbosch University's Language Services (Cape Town, SA).

We used a parent information questionnaire and asset index (Myer et al., 2008) to obtain demographic and socioeconomic background information of the participants. It captures demographic information such as parental/guardian employment and education, home language, and annual household income. The asset index groups asset ownership into three categories: 0–5 (low), 6–12 (medium), and 13–17 (high) and reflects the material and financial resources of the household, for example, appliances (e.g., microwave oven, refrigerator, television), a flushing toilet, running water and car, as well as whether the responder makes use of bank accounts and credit cards.

We used a questionnaire from the RXH pediatric neuropsychology clinic to obtain information on the developmental history (including pregnancy and birth, development, and family composition) of the participants.

We used the informant (parent) measures of the Child Behavior Checklist (CBCL; Achenbach and Rescorla, 2001) and the Behavior Rating Inventory of Executive Function (BRIEF; Gioia et al., 2000) to assess behavioral and emotional functioning, and executive functioning, respectively. These are recognized ecologically valid tools for assessing these outcomes (Schwartz et al., 2003; Gioia and Isquith, 2004).

The CBCL is a 112-item questionnaire used to assess behavioral and emotional functioning of children aged 6–18 years (Achenbach and Rescorla, 2001). Besides demographic questions, there are items that assess competency of activities, social functioning and school. Scores are also produced for eight syndrome scales: Anxious/Depressed, Withdrawn/Depressed, Somatic Complaints, Social Problems, Attention Problems, Rule-breaking Behavior and Aggressive Behavior, and six Diagnostic and Statistical Manual (DSM)-oriented scales: Affective Problems, Anxiety Problems, Somatic Problems, Attention Deficit/Hyperactivity Problems, Oppositional Defiant Problems, and Conduct Problems. On the syndrome and DSM-Oriented scales, age-standardized T-scores above 70 suggest clinically significant behavioral problems, and those from 67 to 70 are borderline clinical. Scores are also produced for two broad syndrome groups: Internalizing Problems and Externalizing Problems that together with the remaining syndrome scales produce a Total Problems score. Age-standardized T-scores above 63 suggest clinically significant behavioral problems and those from 60 to 63, are borderline clinical. The CBCL has both reported validity and reliability (Achenbach and Rescorla, 2001). There are some published research studies using the CBCL with South African samples (e.g., Shields et al., 2008; Palin et al., 2009; Schrieff-Elson et al., 2015).

The BRIEF is an 86-item standardized rating scale assessing everyday executive function behaviors within the home environment for children aged 5–18 years (Gioia et al., 2000). Two index scores [Behavior Regulation Index and Metacognition Index] and an overall composite score of executive function, the Global Executive Composite (GEC), are produced. Scores for eight clinical subscales, that assess interrelated executive function domains, are also produced. Inhibit, Shift, and Emotional Control are subscales of the BRI, while Initiate, Working Memory, Plan/Organize, Organization of Materials, and Monitor are subscales of the MI. T-scores of ≥63 are clinically relevant. The BRIEF shows high levels of internal consistency and stability, as well as test-retest reliability (Malloy and Grace, 2005; Chapman et al., 2010). It has been used cross-culturally in published studies, for example, in the Han Chinese and Dutch populations (Qian et al., 2010; Huizinga and Smidts, 2011) and in a published SA study (Schrieff-Elson et al., 2015).

We sourced academic information regarding education type and grade repetition for the entire sample from relevant questions on the CBCL. For four participants, specific details of the academic data (i.e., whether the child was making use of remedial services or not) were unclear; we therefore consulted the caregiver directly and/or consulted the participants' case files for these details.

## Statistical Analysis

We used SPSS 22.0 to compute the analyses.

#### Descriptive Statistics

We used descriptive statistics to present the demographic and clinical characteristics of the participants.

### Dependent Variables

#### **Academic outcomes**

Academic outcome is represented by two variables: (1) the child's education type at the time of assessment, and (2) whether the child had repeated any grades since returning to school after their injury (see **Table 1** for subcategories). The latter encompasses whether the child had to repeat the grade that they were in at the time of their injury (due to prolonged absence or for academic reasons), or whether they had to repeat subsequent grades postinjury. Both the education type and grade repetition variables were assigned values and converted to z-scores for purposes of statistical analyses. Greater scores represent poorer academic outcome on each of the variables.

### **Behavioral outcomes**

For correlational analyses, we used the scores from the CBCL Total, Internalizing, and Externalizing Problems scales, and the BRIEF's BRI, the MI, and the GEC scores. Subscale scores from the CBCL and index scores from the BRIEF formed part of the descriptive statistics. Greater scores represent more problem behaviors on both the CBCL and BRIEF.

## Correlation Matrices

We used bivariate correlation matrices (Pearson and Spearman's correlation coefficients) to explore the relationships between the academic and behavioral outcome variables. We used one-tailed correlational analyses based on established literature, which informed the direction of the expected relationships. The rstatistic provided a measure of effect size which are described as small, medium or large and represented by r-values of 0.10, 0.30, and 0.50, respectively (Field, 2009).

## ETHICAL CONSIDERATIONS

This protocol was approved by the University of Cape Town (UCT)'s Department of Psychology's Research Ethics Committee and Faculty of Health Sciences' Human Research Ethics Committee (Ref: 345/2011). We obtained written informed consent, as well as permission to obtain school data, from the caregivers of all children in the sample. All data were obtained from parents or legal guardians; hence we did not obtain assent as the children themselves did not complete any measures. This study adhered to the World Medical Association Declaration

TABLE 1 | Variables that represent academic outcome as assessed in this sample.


of Helsinki's ethical principles for medical research involving human subjects (World Medical Association, 2001). We obtained permission from the Western Cape Education Department, SA, to access school data.

## RESULTS

**Tables 2**, **3** present a description of the demographic and SES, and the injury characteristics of the sample, respectively. As per **Table 2**, more than half of the participants were isiXhosaspeaking and male. All participants had access to at least six or more material and financial resources (medium or high asset index bracket). These do, however, include basic amenities as previously described. Half of the caregivers of the children in the sample earned up to ZAR25 000 per annum, with 37% reporting earnings from ZAR25 000 to ZAR100 000 per year. This latter

TABLE 2 | Demographic and SES characteristics of the sample (N = 27).


<sup>a</sup>The home language recorded as being "other" was Swahili. However, the caregiver was fluent in English and therefore the participant was not excluded.

<sup>b</sup>Household income presented in South African Rands (ZAR).

<sup>c</sup>For one of the cases, information on education and employment was only provided for the participant's guardian and not for the parents.

TABLE 3 | Injury characteristics of sample (N = 27).


Means are presented with standard deviations in parentheses. GCS, Glasgow Coma Scale.

result, however, includes quite a wide range and one cannot be sure how many families are earning closer to the lower end of that range. Most parents had 8–11 years education. The youngest participant was 6 years 5 months, while the oldest was 12 years 7 months at the time of injury. The range of time since injury was 15 to 70 months (**Table 3**).

## Academic Outcome

**Table 4** compares the academic information obtained for the sample and presents the ratios of grade repetition, and type of education pre- and post-injury. When considering the enrollment in remedial and special needs education services from pre- to post-injury, there was a three-fold increase from before (n = 6; 22.22%) to after (n = 18; 66.67%) the TBI. This increase reveals that for 44.44% of the sample (n = 12), the use of remedial services or special needs education was required only after the TBI. Remedial help was usually in the form of extra lessons or placement in a remedial-oriented class at the mainstream school. For six of these 12 participants, school personnel were in the process of applying to a school that catered for their specific learning or physical needs and were awaiting the outcome of this application, or applications were already approved and they were awaiting transfer. In these cases, children continued to attend mainstream schools in the interim, where they may or may not have received remedial help during that time. Four (14.81%) of children in special needs schools at the time of the study had been placed in those schools when they resumed schooling post-injury, or were placed a later stage before commencement of the study. More than half of the sample (n = 16; 59.26%) had repeated a grade following their injury, which represents a 275% increase from pre- to post-TBI.

**Figure 2** presents a cross-tabulation of pre- and post-injury academic information. In the sample, 15 (55.55%) children who had not repeated a grade pre-injury, went on to repeat one or more grades post-injury. Of the four participants (14.80%) that repeated a grade pre-injury, only one (3.70%) also repeated a grade post-injury. This participant attended a special needs school pre- and post-injury and had premorbid cerebral palsy. The three participants (11.10%) who repeated a grade pre-, but not post-injury, attended mainstream schools both pre- and TABLE 4 | Academic information obtained for the sample (N = 27).


Frequencies are presented with percentages in parentheses.

post-injury. Only one of these participants required the use of remedial services post-injury. Of those that attended mainstream school pre-injury (n = 26; 96.30%), four (14.81%) were then placed in special needs school post-injury.

## Behavioral Outcomes CBCL

**Table 5** presents the descriptive statistics for the CBCL scores. Caregivers reported clinically significant Total Problems (65.38%), Internalizing Problems (65.38%), and Externalizing Problems (53.85%) for more than half of the participants. Although fewer participants scored within the clinical range on all other individual CBCL scales and subscales, some of these frequencies were still above 40% [e.g., Aggressive Behavior (46.15%) and Affective Problems (46.15%) scales].

## BRIEF

**Table 6** shows that on the BRIEF, the mean reported scores for most of the indices fell within the clinical range, with Working Memory Index (73.08%) as the highest, followed by Emotional Control (65.38%), and Plan/Organization (65.38%). Overall on the GEC, 69.23% of the sample scored in the clinical range. Looking specifically at the major index scales, more than half the sample scored within the clinical range on the BRI (61.54%) and the MI (61.54%).

## Correlation Matrix

**Table 7** shows the correlations between academic and behavioral outcomes. There were strong significant positive correlations between repeated grade post-injury and the CBCL Total Problems and CBCL Externalizing Problem scales and the three BRIEF indices. These relationships suggest that grade repetition post-injury was associated with poorer reported scores on these behavioral measures.

There were medium to strong positive correlations between post-injury school type and the BRIEF indices, suggesting that poorer executive function behaviors were associated with increased need for remedial and special needs education.

As expected, were the significant positive correlations between outcome scores on the CBCL and the BRIEF, which were all in the expected direction.

## DISCUSSION

This study aimed to contribute to the pTBI outcome literature in a developing world setting, by investigating and describing behavioral and academic outcomes in a group of school-going South African children who had sustained a severe TBI.

## Academic Outcomes

Two thirds (18/27) of the sample were, at the time of assessment, receiving some form of remedial or special needs education or were in the process of being placed in remedial or special needs education post TBI. Thus, there were more children requiring some form of specialized education following their TBI than those who were integrated back into mainstream schooling. This result is consistent with literature on pTBI survivors being reintegrated into the schooling system and the associated increased need for specialized educational services, particularly with more severe TBI (Donders, 1994; Kinsella et al., 1997; Ewing-Cobbs et al., 1998; Savage et al., 2005; Jantz and Coulter, 2007; Arnett et al., 2013; Prasad et al., 2017). Taylor et al. (2003) found that 62% of the children with severe TBI in their sample were in programs that catered for special education needs, even several years after their injuries. Placement in these programs occurred soon after injury. Kinsella et al. (1997) previously reported similar high rates, where 70% of children with severe TBI required special needs intervention or attended school part-time after their injury; while Donders (1994)reported a 40% increase in the number of children requiring special needs education and Hawley (2004) a two-fold increase in the number children in her TBI (mixed severity) sample. In the current study, there was a three-fold increase (from n = 6 to n = 18) in the number of children using some form of special education services, from pre- to post-TBI.

Post-injury, 22% (6/27) of learners had applied for placement in a LSEN school, or had already been offered placement, but were awaiting transfer to that school. There are several reasons that could account for the delay in placement at an LSEN school. It may reflect a delayed process of identification of needs, for example, when deficits are not immediately evident and but are only evident at a later stage when cognitive demands on the child increase. The delay in placement may also be due to the unavailability of a place at a suitable school, especially when resources are limited (Taylor et al., 2003; Mayfield and Homack, 2005; Ciccia and Threats, 2015). In SA, the delay in placement in LSEN schools is often due to it being a lengthy bureaucratic process. Furthermore, the number of special needs schools as well as their capacity for learners is limited with more than 10,000 learners on a waiting list for placement (Amollo, 2008; Donohue and Bornman, 2014). This lengthy waiting list is particularly problematic when one considers that once learners reach 16 years of age, they are considered too old to be placed in LSEN schools.

More than half (59%) of the sample had to repeat at least one grade after returning to school post-injury. This result is consistent with the outcomes for the "children group" (5–10 years) in the Ewing-Cobbs et al. (1998) study with 55% having repeated a grade within two years following their injury. There was a four-fold increase in the number of children who had to repeat one or more grades from before to after the TBI.

The reported reasons for these adverse academic outcomes in the literature are varied. The cognitive deficits associated with TBI, including difficulties in attention, memory, executive function, and essential skills such reading and writing, can impact on academic performance (Donders, 1994; Ewing-Cobbs et al., 1998; Arroyos-Jurado et al., 2000; Hawley, 2004; Taylor, 2010; Max, 2014). Other factors such as behavioral impairments and absence from school may also play a role (Ewing-Cobbs et al.,

#### TABLE 5 | Outcome scores obtained on the CBCL parent version (N = 26).


Means are presented with standard deviations in parentheses. Frequencies are presented with percentages in parentheses. One participant's scores are missing from the CBCL. The caregiver did not correctly complete the form and it could not be scored. The caregiver subsequently moved away and was lost to follow-up. CBCL, Child Behavior Checklist; ADH, Attention Deficit/Hyperactivity.

<sup>a</sup>Clinical range.

<sup>b</sup>Borderline clinical range.

TABLE 6 | Outcome scores obtained on the BRIEF parent version (N = 26).


Means are presented with standard deviations in parentheses. Frequencies are presented with percentages in parentheses. Data was missing for one participant. The caregiver of this participant had taken the measures home with him to complete; all measures were returned apart from the BRIEF. He was subsequently lost to follow-up due to a change in contact details after changing employment. BRIEF, Behavior Rating Inventory of Executive Function; BRI, Behavior Regulation Index; MI, Metacognition Index; GEC, Global Executive Composite.

<sup>a</sup>Clinical range.

1998; Babikian and Asarnow, 2009). In some cases, the child missed out on weeks or months of school during their post-TBI recovery period, especially when the TBI occurred during the school term. Frequently, the child also missed periods of school to attend doctors' appointments even after the initial hospital stay.

Given the already fragile state of the education system in SA and worrying statistics like 70% of children with special educational needs who should be in school, are not attending school, and given that many pTBI survivors (especially after severe TBI) have post-TBI special needs, one cannot help but be concerned about the future educational retention of these children. The tentative nature and lack of efficacy of the inclusive educational model also creates uncertainty around where SA pTBI survivors may be and are best placed.

## Behavioral Outcome CBCL

The sample's scores on the Total Problems, Internalizing Problems, and Externalizing Problems scales were on average in the clinically significant range, with relatively homogenous means. Elevated syndrome and DSM-Oriented scale scores, such as Aggressive Behavior, Affective Problems, Attention problems, Social Problems, and Conduct Problems, fell in the borderline clinical range. Overall, caregivers reported a greater frequency of clinically significant internalizing than externalizing problems. Such problem behaviors may persist over time and patterns thereof may vary. Within the internalizing and externalizing behavior scales, anxious/depressed behaviors and aggressive behaviors had the greatest frequency of clinically significant scores, respectively. The generally high occurrence of problem behaviors is consistent with numerous studies documenting the commonly reported sequelae following severe TBI and may reflect difficulties associated with self-regulation of behavior and emotions (see e.g., Fletcher et al., 1990; Kinsella et al.,



CBCL, Child Behavior Checklist; BRIEF, Behavior Rating Inventory of Executive Function; BRI, Behavior Regulation Index; MI, Metacognition Index; GEC, Global Executive Composite; Statistics presented are Pearson correlation coefficients (r) unless otherwise stated. All tests are 1-tailed.

<sup>a</sup>Statistics presented are Spearman correlation coefficients for continuous variables that are not normally distributed (rs).

<sup>b</sup>n = 26.

\*p < 0.05. \*\*p < 0.01.

1999; Mayfield and Homack, 2005; Yeates and Taylor, 2006; Dooley et al., 2008; Anderson and Yeates, 2010; Taylor, 2010; Catroppa et al., 2012; Babikian et al., 2015). Within both the internalizing and externalizing behavioral categories, although internalizing behaviors were generally more pronounced, on average, caregivers rated a greater degree (as indicated by the highest score) of aggressive behaviors than any other specific problem behavior assessed on the CBCL. While aggression can manifest as a direct result of damage to the brain and the associated cognitive deficits, Dooley et al. (2008) reported that aggressive behaviors are likely a result of anger and distress in response to one's injury and deficits, but may also be associated with emotional lability and decreased frustration tolerance.

#### BRIEF

In more than half of the sample, caregivers reported clinically significant problems across a range of executive function subdomains. Working memory appeared to be the most problematic. The results are consistent with literature documenting executive dysfunction in children who have sustained a TBI (Levin and Hanten, 2000; Anderson et al., 2002; Bamdad et al., 2003; Yeates et al., 2005; Yeates, 2010; Arnett et al., 2013). Working memory is crucial to optimal functioning of cognition and the assessment thereof generally (e.g., holding instructions in mind while executing a task) and is particularly susceptible to the effects of a TBI (Hillary et al., 2006). The high incidence of executive dysfunction, in addition to other problems with behavioral and emotional regulation (such as assessed on the CBCL and the BRIEF) in the sample likely reflect damage to the vulnerable frontal areas and associated neural circuitry mediating executive functioning and the regulation and self-monitoring of responses (Bamdad et al., 2003).

To promote inclusive education, the intention of the SA DOE is that only learners deemed to have severe disabilities be placed in LSEN schools, with learners with less severe disabilities being accommodated in mainstream schools (Amollo, 2008). However, it seems that educators may not all be adequately trained in this regard. Coping with learners with special needs in mainstream schools, managing behaviorally challenged learners (who could include pTBI survivors), and supporting those who fail to progress (who may struggle emotionally) are reported as concerns for educators and a barrier to the inclusive education model.

## Correlation Analyses

The significant correlations between repeated grade post-injury and the Externalizing Problems scale (and Total Problems, which is likely a function of this) on the CBCL suggest that pTBI survivors who repeated a grade post-injury may display a greater the degree of externalizing behavior problems as assessed on the CBCL, or conversely, that those with externalizing problems, specifically showing more rule-breaking or aggressive behavior, are more likely to repeat grades after their injury. These findings are consistent with literature describing the impact of behavioral impairments on school performance and educational outcome (Keenan and Bratton, 2006; Babikian and Asarnow, 2009; Arnett et al., 2013). Nelson et al. (2004) found that among students who had emotional or behavioral disorders, those who exhibited externalizing as compared to internalizing behavioral problems were more likely to have deficits in academic achievement. These authors conjecture that externalizing behaviors (such as inattentiveness and disruptive behaviors) may have more of a pervasive influence than internalizing behavior problems with regards to interfering with the learning process and consequently coping and progressing academically. This is especially the case when the behaviors involve poor behavior regulation and selfmonitoring secondary to executive dysfunction (Mayfield and Homack, 2005).

The positive significant associations between post-injury repeated grade and school type, and the outcome scores (GEC, BRI, MI) on the BRIEF, suggest that pTBI survivors with increased executive dysfunction, including problems with regulation of behavior and abilities related to problem solving, are more likely to be in need of special educational services or to have failed grades. The relationships between the academic outcome variables and the BRIEF outcomes are consistent with literature documenting the role of executive functions in the learning process and in academic achievement, as well as post-TBI executive function outcomes generally (Anderson et al., 2002; St Clair-Thompson and Gathercole, 2006; Babikian et al., 2015). In their sample of children and adolescents with Attention-Deficit Hyperactivity Disorder (ADHD), St Clair-Thompson and Gathercole (2006) found that there was increased risk of repeating grades, learning disabilities and poor academic achievement amongst those with deficits in executive functioning.

The significant positive relationships between the CBCL and BRIEF were expected due to the construct validity of these measures, and because the Internalizing Problems and Externalizing Problems scales, and the BRI and MI form part of the overall outcome scores (Total Problems and GEC, respectively) obtained on these measures.

In summary, the results show elevated problems with behavioral and executive functioning, and academic concerns in the sample. More than half of the participants experienced clinically significant behavioral problems and working memory appeared to be the most problematic subdomain of executive function. Two thirds of the children were receiving some form of, or in the process of being placed in, LSEN schools; and the increase in the number of children using some form of special education services from pre- to post-injury was three-fold. Furthermore, more than half of the sample repeated at least one grade after returning to school following their injury. The results in this study suggest that children with increased externalizing behavioral are more likely to repeat a grade post-injury; and that children with executive dysfunction post-TBI may be more likely to require some form of special educational services and more likely to repeat a grade post-injury (although the converse may also be true in both cases given the correlational nature of the study).

## Limitations and Directions for Future Research

Ideally, the study design should include a model of change in academic grade performance and would incorporate a change in performance over three time periods: the period before injury, the initial period upon return to school, and the time of assessment. A change in performance could then be determined as a percentage recovery from the initial decline in academic performance and the total decline in academic performance. This model of change was however not achievable due to challenges in obtaining complete academic history (pre-injury scholastic and behavioral assessments, and term and year-end results) for most the sample. These limitations are not uncommon for a developing world setting and are important to address because it limits good data collection.

Due to the small sample size, the results of this study should be viewed with caution. The exclusion of children who were not attending school at the time of the study became a limiting factor in the size of the sample. Furthermore, the small sample had a wide age range of participants (6–12 years).

Further, this study made use of parent self-report measures, which may be limited by response sets, social desirability bias, and sometimes unreliable recall from memory of past behaviors. The measures used have not been validated in the SA context—as is the case generally with neuropsychological measures (Schrieff-Elson et al., 2017). Nevertheless, informant measures were used in this study due to their psychometric properties and that they are commonly used assessment tools. Moreover, they are more ecologically valid that standard pencil and paper measures, despite the lack of norms (Anderson et al., 2002; Gioia and Isquith, 2004). Further studies should look to supplement informant measures, for example, by including structured interviews with caregivers and teachers, as well as direct observations of behavior.

## SUMMARY AND CONCLUSION

We investigated and described behavioral and academic outcomes for a group of school-aged severe pTBI survivors in SA. The results show that problems (e.g., increased need for special education services, behavioral problems and executive dysfunction) experienced by this sample are consistent with those reported for children with severe TBI in the literature (Babikian and Asarnow, 2009; Van't Hooft, 2010; Babikian et al., 2015; Prasad et al., 2017). The current study's results therefore advocate for increased awareness in identifying children, and indeed families, that are at greater risk for dysfunction and poorer academic outcomes following pTBI. This is particularly important in the developing world context like SA, where there are a limited number of LSEN schools (reserved for children with severe disabilities) and none that specifically cater for the unique needs of children with TBI (Levin, 2004). Interim remedial support should be provided until children who require special needs schooling are placed appropriately. For those who recover sufficiently to be accommodated in mainstream schools, ideally, this would include an increased availability of educational resources and learner support that focuses on the cognitive, behavioral and emotional sequelae associated with TBI. Advances in technology provide opportunities through which to view post-pTBI educational opportunities and support in developed compared to the developing world contexts and the obvious disparities in post-pTBI education reintegration. Clearly, changes in policy and a greater funding focus on this issue in the developing world context are needed (Chomba et al., 2014).

## AUTHOR CONTRIBUTIONS

AD carried out this research as her Masters research project. She was the lead author on for the write-up of this manuscript. AF was a co-supervisor on AD's MA research. He oversaw drafts of this article and provided edits and feedback. LS-E was the main supervisor for AD's MA research. She has overseen multiple drafts of the thesis and manuscript and provided edits and feedback.

## ACKNOWLEDGMENTS

This research was supported through funds obtained from the following sources: National Research Foundation, Victor Nell-SACNA Endowment for the Study of Neuropsychology in South Africa, KW Johnston Bequest, UCT Research Scholarship, and the Ernst and Ethel Ericksen Trust.

## REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The handling Editor declared a shared affiliation, though no other collaboration, with the authors.

Copyright © 2017 Dollman, Figaji and Schrieff-Elson. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Elevated Cortisol Leaves Working Memory Unaffected in Both Men and Women

#### Robyn Human<sup>1</sup> \*, Michelle Henry 1,2 , W. Jake Jacobs <sup>3</sup> and Kevin G. F. Thomas <sup>1</sup>

<sup>1</sup>ACSENT Laboratory, Department of Psychology, University of Cape Town, Rondebosch, South Africa, <sup>2</sup>Academic Development Programme, Centre for Higher Education Development, University of Cape Town, Rondebosch, South Africa, <sup>3</sup>Anxiety Research Group, Department of Psychology, University of Arizona, Tucson, AZ, United States

Activation of the hypothalamic-pituitary-adrenal (HPA) axis (as might occur, for example, when the organism encounters a threat to allostatic balance) leads to the release of cortisol into the bloodstream and, ultimately, to altered neural functioning in particular brain regions (e.g., the prefrontal cortex (PFC)). Although previous studies suggest that exposure to acute psychosocial stress (and hence, presumably, elevation of circulating cortisol levels) enhances male performance on PFC-based working memory (WM) tasks, few studies have adequately investigated female performance on WM tasks under conditions of elevated cortisol. Hence, we compared associations between elevated (relative to baseline) levels of circulating cortisol and n-back performance in a South African sample (38 women in the late luteal phase of their menstrual cycle, 38 men). On Day 1, participants completed practice n-back tasks. On Day 2, some completed the Trier Social Stress Test (TSST), whereas others experienced a relaxation period, before completing 1-back and 3-back tasks. We measured self-reported anxiety and salivary cortisol at baseline, post-manipulation and end of session. We reconstituted group assignment so that all women with elevated cortisol were in one group (EC-Women; n = 17), all men with elevated cortisol were in another (EC-Men; n = 19), all women without elevated cortisol were in a third (NoEC-Women; n = 21), and all men without elevated cortisol were in a fourth (NoEC-Men; n = 19) group. Analyses suggested this reconstitution was effective: in EC, but not NoEC, groups cortisol levels rose significantly from baseline to post-manipulation. Analyses of n-back data detected significant relations to task load (i.e., better performance on 1-back than on 3-back tasks), but no significant relations to sex, performance accuracy/speed, or cortisol variation. The data patterns are inconsistent with reports describing sex differences in effects of stress on WM performance. We speculate that cross-study methodological differences account for these inconsistencies, and, particularly, that between-study variation in the magnitude of baseline cortisol levels might affect outcomes. For instance, diurnal cortisol rhythms of South African samples might have flatter curves, and lower baseline values, than predominantly Caucasian samples from the United States and western Europe due to greater prenatal and lifetime stress, more socioeconomic disadvantage and faster ancestral life history (LH) strategies. We describe ways to disconfirm this hypothesis, and urge further cross-national research exploring these possibilities.

Keywords: cortisol, psychological stress, sex differences, Trier Social Stress Test (TSST), working memory

#### Edited by:

Nouria Lakhdar-Ghazal, Mohammed V University, Morocco

#### Reviewed by:

Mar Sanchez, Emory University, United States Annerine Roos, Stellenbosch University, South Africa Dustin Van Gerven, University of Victoria, Canada

> \*Correspondence: Robyn Human robynjvanvuuren@gmail.com

> Received: 28 August 2017 Accepted: 10 July 2018 Published: 24 July 2018

#### Citation:

Human R, Henry M, Jacobs WJ and Thomas KGF (2018) Elevated Cortisol Leaves Working Memory Unaffected in Both Men and Women. Front. Hum. Neurosci. 12:299. doi: 10.3389/fnhum.2018.00299

## INTRODUCTION

The construct of working memory (WM) describes a class of memory produced by a set of neural processes that underpin a variety of higher-order cognitive functions. These processes are integral in coordinating the temporary storage and subsequent manipulation of information that might assist the organism in achieving a variety of goal-directed behaviors, especially when that information must be applied outside of the immediate context (Baddeley and Hitch, 1974; Baddeley, 1986; D'Esposito and Postle, 2015). The purpose of the present investigation was to determine if elevated cortisol levels immediately and differentially affect WM in men and women. This proposal is based on evidence that a chain of predictable neurobiological events follow any threat to allostatic balance (McEwen, 2005). In humans, activation of the hypothalamic-pituitary-adrenal (HPA) axis leads to the release of cortisol into the bloodstream. After crossing the blood-brain barrier, cortisol and related hormones bind to glucocorticoid receptors (GRs), which, although spread throughout the brain, occur in particular abundance in the prefrontal cortex (PFC), hypothalamus and hippocampus (Reul and De Kloet, 1985; Alderson and Novack, 2002; Kemeny, 2003). Once bound, these hormones alter neural functioning in those regions. Of particular interest here is the alteration of PFC functioning when cortisol levels are elevated above baseline, and consequent effects on WM performance (Owen, 1997; Baddeley, 2001; Wolf, 2003; Arnsten, 2009).

Previous studies, built on this chain of evidence, demonstrate that acute psychosocial stress (and consequent elevations in cortisol levels) negatively affect speed and accuracy in the performance of WM tasks, especially at greater cognitive loads (e.g., Lupien et al., 1999; Wolf et al., 2001; Mizoguchi et al., 2004; Elzinga and Roelofs, 2005; Oei et al., 2006; Schoofs et al., 2008, 2009; Luethi et al., 2009; Barsegyan et al., 2010). A smaller group of studies suggest, however, that such cortisol elevations improve, or may have no effect, on WM task performance (Kuhlmann et al., 2005; Oei et al., 2009; Weerda et al., 2010; Stauble et al., 2013). Both groups of studies, however, enrolled only male participants. Although this is a justifiable position given their logic, it does mean these studies leave unexplored the nature of relations between stressor-induced increases in cortisol levels and female performance on WM tasks.

Although under normal circumstances there are negligible sex differences in terms of HPA-axis functioning, men generally exhibit greater stressor-induced increases in cortisol levels than women (Kudielka and Kirschbaum, 2005; Uhart et al., 2006; Kudielka et al., 2009). However, women who are in the late luteal phase of the menstrual cycle and free of oral contraceptives exhibit post-stressor cortisol responses comparable to those of men (Kirschbaum et al., 1999). This fact permits us to compare directly the performance of men and women on tasks representing WM processing in the presence of elevated cortisol.

Under normal circumstances, men and women perform differently on WM-related tasks (Speck et al., 2000; Lynn and Irwing, 2008; but see Evans and Hampson, 2015). These functional differences extend to performance on cognitive tasks related to WM under conditions of elevated cortisol. Although not testing WM directly, several well-designed studies demonstrate that acute exposure to a psychosocial stressor differentially affects performance by men and women on tasks requiring activity in neural regions that support WM processing (Jackson et al., 2006; Preston et al., 2007; Porcelli and Delgado, 2009; van den Bos et al., 2009; Thomas et al., 2010; but see Starcke and Brand, 2016).

Studies featuring exposure to a stressor and consequent elevated cortisol, and focusing on WM, report enhanced male performance but either impaired or unaffected female performance on an n-back task (Cornelisse et al., 2011; Schoofs et al., 2013). Cautious interpretation of these data is warranted because Cornelisse et al. (2011) did not control for female use of oral contraceptives or menstrual cycle phase, and, although Schoofs et al. (2013) did use such controls, their findings have yet to be replicated.

In the only other study to specifically investigate cortisolrelated sex differences in WM performance, Zandara et al. (2016) found that women whose cortisol levels decreased from pre- to post-manipulation measurement (n = 6) showed significantly improved performance, across that period, on a forward digit span test. On a backward digit span task, however, there were neither detectable sex differences nor significant changes in preto post-manipulation performance. Although the design did not control for female oral contraceptive use or for menstrual cycle phase, the authors did control for these factors statistically. After doing so, they found no detectable effects on digit span performance. Again, the results should be interpreted cautiously given that: (a) they have yet to be replicated, (b) group sizes with regard to oral contraception/menstrual cycle phase were small (n ≤ 12), and (c) there is some debate about the efficacy of forward digit span tasks as a measure of WM, primarily because these tasks do not require manipulation of the presented information and do not include a reaction time component (Jarrold and Towse, 2006; Lynn and Irwing, 2008; Schoofs et al., 2008; Egeland, 2015).

In summary, although previous studies suggest that stressinduced cortisol elevations enhance male performance on PFC-based WM tasks, few studies have adequately investigated female performance on WM tasks under such conditions. Therefore, in the present study, we exposed naturally cycling women in the late luteal phase of the menstrual cycle, and men, to the Trier Social Stress Test (TSST; Kirschbaum et al., 1993). We then compared the performance of women and men with elevated cortisol against that of women and men without elevated cortisol on an n-back task, a commonly accepted measure of the WM construct that reliably activates the PFC (Owen et al., 2005; Jaeggi et al., 2010), with some sex-specific differentiation (Speck et al., 2000; Li et al., 2010; although see Kane et al., 2007 and Schmidt et al., 2009 for contrasting data).

An important methodological consideration here is that performance characteristics on n-back tasks vary as a function of the level of cortisol and the specific version of the n-back being administered. On 0-back or 1-back tasks, for example, elevated cortisol has little or no effect on performance; in contrast, on 2- or 3-back tasks, the presence of elevated cortisol has medium- to large-sized effects on performance, especially in men (Schoofs et al., 2008, 2013; Cornelisse et al., 2011). Hence, we used performance on a 1-back task to contrast with that on a 3-back task.

Based on the methodological considerations regarding relations between level of cortisol and performance on n-back tasks, we predicted that: (a) overall WM performance, as measured by the n-back, is faster and more accurate on a 1-back than on a 3-back task regardless of biological sex or of exposure to a cortisol-elevating psychosocial stressor. We also sought to replicate the findings that, on a 3-back task, (b) men with cortisol elevations show enhanced performance, whereas (c) women with cortisol elevations show impaired performance (see Cornelisse et al., 2011; Schoofs et al., 2013).

## MATERIALS AND METHODS

## Design and Setting

This quasi-experimental study took place over two consecutive days, permitting us to ensure that, by the end of Day 1, all participants understood the requirements and nature of the n-back before undertaking the Day 2 tasks. On each day, the participant entered the laboratory at 16:00 h or 18:00 h and completed all procedures within 2 h, ensuring control of cortisol's circadian cycle, maximizing potential for elicitation of a strong HPA-axis response to the stressor and permitting us to investigate the effects of cortisol elevations occurring outside the normal diurnal cycle (Dickerson and Kemeny, 2004; Kudielka et al., 2004, 2009; Maheu et al., 2005).

## Participants

We recruited undergraduate students (45 men and 57 women), between the ages of 18 and 25 years (M = 19.33, SD = 1.51). Of these, 24 met at least one of the exclusion criteria listed below. Also, one set of cortisol data showed unusual patterns at baseline (>18 SD above the mean for participants exposed to the stressor), and another was lost due to experimenter error (see **Table 1**). Hence, our final sample consisted of data obtained from 38 men and 38 women. Each received course credit in exchange for participation.

TABLE 1 | Reasons recruited participants were excluded from final data analysis (n = 26).


Note. BDI-II, Beck Depression Inventory—Second Edition.

## Exclusion Criteria

We asked those people using any form of steroid medication, and women using any form of oral contraceptive, not to apply to the study. We used this exclusion criterion because these medications affect the magnitude of cortisol response to psychosocial stressors. We also asked those women experiencing an irregular menstrual cycle not to apply. We permitted women reporting a regular menstrual cycle to enrol because the study design specified testing women during the late luteal phase of the menstrual cycle (i.e., the 6-day window preceding the start of menses; Ferin et al., 1993; Symonds et al., 2004). Each woman indicated the date she expected to begin her next period and received an appointment within the 6 days preceding that date. Due to within- and between-woman variability in overall menstrual cycle length, this method appears to be an accurate way to predict phase of the menstrual cycle (Sherman and Korenman, 1975; Cole et al., 2009). Each enrolled woman contacted the experimenter on the first day of her next period to confirm, post-experimentally, the phase of the menstrual cycle during which she had been tested.

We excluded enrolled participants who: (a) scored 29 or above on the Beck Depression Inventory—Second Edition (BDI-II; Beck et al., 1996); (b) self-reported beginning menstruation on either Day 1 or Day 2 of the experimental protocol; (c) did not arrive for the Day 2 session; or (d) withdrew from the study at any point during the Day 1 or Day 2 protocols. We also excluded data sets from individuals who did not meet the n-back performance criterion during the Day 1 practice session (see below), and from women who reported, after completing the experimental protocols, they were more than 1 day outside of the late luteal phase of the menstrual cycle during Day 2.

This study was carried out in accordance with the recommendations of the University of Cape Town's Guidelines for Human Research, as interpreted by the Research Ethics Committees of that institution's Faculty of Health Sciences and Department of Psychology, with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki (World Medical Association, 2013). The protocol was approved by the Research Ethics Committees of the University of Cape Town's Faculty of Health Sciences and Department of Psychology.

## Materials and Procedures

#### Day 1

A female research assistant (RA) met participants at the laboratory. She assigned each participant pseudo-randomly to one of four groups: TSST-Women, TSST-Men, Relax-Women, and Relax-Men. Each participant then read and signed a consent form and completed the BDI-II and Form Y-2 of the State-Trait Anxiety Inventory (STAI; Spielberger et al., 1983). The four men and four women who met the BDI-II exclusion criterion did not self-report levels of depression requiring immediate intervention. Hence, the RA provided counselling referrals and dismissed them from the study. Results from the STAI-Trait form served as a measure of general anxiety levels, permitting us to ensure there were no detectable between-group differences in everyday experiences of anxiety.

The remaining participants then completed a set of practice n-back tasks. They completed one block of 20 0-back trials, followed by one block of 20 1-back trials, followed by one block of 20 3-back trials.

A standard desktop computer presented the n-back tasks using E-prime software version 1.1 (Psychology Software Tools, Pittsburgh, PA, USA). The E-prime n-back script was modified from one similar to that at https://step.talkbank.org/scriptsplus/. Participants saw a random series of letters, presented one at a time, on a computer screen. They were instructed to determine if the presented letter was a ''target'' or a ''non-target.'' If the former, they pressed the F key on the keyboard; if the latter, they pressed the J key. Each letter was displayed on-screen for 500 ms with an inter-stimulus interval of 3518 ms. Participants were required to achieve an accuracy score of at least 70% on the 0-back before proceeding to the 1-back, 70% on the 1-back before proceeding to the 3-back, and 70% on the 3-back to end the practice tasks.

At the end of the Day 1 session, the RA reminded participants of the appointment scheduled for the next day, asked them not to smoke, consume any food or drink, chew gum, or engage in physical exercise for 2 h before the start of the Day 2 session. This reminder paralleled protocols described by others (e.g., Kirschbaum et al., 1993; Schoofs et al., 2008; Verdejo-Garcia et al., 2015).

#### Day 2

The same RA met returning participants at the laboratory, and reminded them of their ethical right to withdraw from the study at any time without penalty. **Figure 1** illustrates the timeline of Day 2 experimental events.

Participants rated their current level of anxiety thrice using the STAI-State form: the first, at baseline, shortly after entering the laboratory (STAIB), the second at 5 min following the end of the stress or control manipulation (STAI1) and the third at 45 min after the end of the manipulation (STAI2).

The RA collected three saliva samples using SARSTEDT Salivette<sup>r</sup> Cortisol swabs (Sarstedt, Nümbrecht, Germany): the first, at baseline, shortly after the participant entered the laboratory (CORTB), the second at 5 min after the stress or control manipulation ended (CORT1) and the third at 45 min after the manipulation ended (CORT2). Immediately after each collection, the RA stored the saliva samples in individual, labeled tubes and placed them in a freezer where they were stored at −20◦C. Upon completion of data collection, we transported the tubes to a laboratory for analysis. Samples were analyzed using a competitive electrochemiluminescent immunoassay on the Roche Cobas 6000 (Roche Diagnostics GmbH, Mannheim, Germany) with a coefficient of variation of 4%.

At the end of the session, the RA debriefed participants completely and reminded the women to contact the experimenter on the first day of their next period. The study was then concluded.

## **Experimental Manipulation**

Participants in the TSST-Women and TSST-Men groups completed a modified form of the TSST to induce cortisol release. The TSST, which involves public speaking and mental arithmetic tasks, on the average induces large increases in cortisol levels (Kirschbaum et al., 1993; Dickerson and Kemeny, 2004; Foley and Kirschbaum, 2010). In the current study, participants were told that an interviewing panel (one man and one woman) in a separate room would analyze their speech and behavior with


TABLE 2 | Between-group comparisons: participant age and self-reported depression and anxiety (N = 76).

Note. Data presented are means with standard deviations in parentheses. ESE, effect size estimate (in this case, β 2 ); BDI-II, Beck Depression Inventory-Second Edition; STAI, State-Trait Anxiety Inventory.

the help of a video camera. Participants were then instructed to write and present a speech detailing their suitability for a job of their choosing. After 10 min of preparation, the RA escorted the participant to a room illuminated by a harsh, bright light. There, the participant spoke extemporaneously for 5 min while being observed by the interviewing panel. From this point, the protocol (including the actual speech and the arithmetic task) followed that described by Kirschbaum et al. (1993) closely.

Administration of the control procedures occurred in the same room as those of the stress induction. In this case, however, the room had normal lighting, no video camera and no judges. Participants in the Relax-Women and Relax-Men groups sat in comfortable chairs, read neutral-content magazines and listened to relaxing music for 20 min.

### **The n-Back Tasks**

Following the experimental manipulation, participants completed a 0-back task (1 block, 24 trials). They then completed four alternating 1-back and 3-back blocks (24 trials each). On eight trials within each block, the correct response was F (''target''); on the remaining 16, the correct response was J (''non-target''). Following Schoofs et al. (2008, 2013), the first three trials of each block were non-targets, and we did not include their data in the final analysis. This preparation permitted us to examine performance on three n-back versions (0-, 1- and 3-back) while the participant was under the influence of elevated cortisol (see above and Sliwinski et al., 2006, for arguments detailing the value of this approach).

## Statistical Analysis

After applying the exclusion criteria described above, the groups were constituted as follows: TSST-Women (n = 20), TSST-Men (n = 17), Relax-Women (n = 18), and Relax-Men (n = 21). Based on the moderate effect sizes reported in studies describing relations between stress and working memory performance (e.g., Schoofs et al., 2013), a power analysis indicated that an N of 76 is sufficient to detect the effects under consideration. Furthermore, our sample size compares favorably to those used in similar studies (Schoofs et al., 2008, 2013; Cornelisse et al., 2011).

We analyzed the data using SPSS (version 23.0), and set the Type I error rate at 0.05 unless otherwise specified. In cases where assumptions of parametric statistical tests were violated, we made suitable adjustments (e.g., log transformations of data).

## RESULTS

Common findings in this area of research are that: (a) there is considerable variability in response to laboratory-based stress induction methods (e.g., Buchanan and Tranel, 2008; van den Bos et al., 2009; Schlotz et al., 2011), and (b) more men than women show increased cortisol after TSST exposure (e.g., Stroud et al., 2002; Elzinga and Roelofs, 2005). These patterns were present in our data: post-manipulation measurement detected elevated cortisol levels in 12 of 20 women (60%) and 14 of 17 men (82%), in the TSST condition. In the Relax condition, 5 of 18 women (28%) and 5 of 21 men (24%), showed elevated post-manipulation cortisol levels.

Because our questions focused on relations between elevated cortisol and WM performance, we reconstituted the group assignment so that women with elevated cortisol (EC) were in one group (EC-Women; n = 17), men with elevated cortisol were in another (EC-Men; n = 19), women without elevated cortisol were in a third (NoEC-Women; n = 21) and men without elevated cortisol were in a fourth (NoEC-Men; n = 19). Here, we defined ''elevated cortisol'' as any increase in cortisol level from CORT<sup>B</sup> to CORT<sup>1</sup> (i.e., any elevation from baseline to the immediate post-manipulation measurement point). Subsequent analyses compared self-report and physiological measures of stress, and working memory performance, across the EC-Women, EC-Men, NoEC-Women and NoEC-Men groups.

## Sample Characteristics

A series of one-way ANOVAs detected no statistically significant between-group differences in mean age, BDI-II scores, or STAI-Trait scores (see **Table 2**).

## Group Assignment Check

We conducted 4 × 3 (Group [EC-Women, EC-Men, NoEC-Women, NoEC-Men] × Time [Baseline, Time 1, Time 2]) repeated-measures ANOVAs on data for the STAI-State scores and salivary cortisol levels (see **Table 3** for descriptive statistics). Planned comparisons tested a priori hypotheses regarding between- and within-group differences.

## Subjective Anxiety

The analysis detected a significant main effect of Time, F(2,144) = 16.18, p < 0.001, η 2 <sup>p</sup> = 0.18, but not of Group, F(3,72) = 2.11, p = 0.11, η 2 <sup>p</sup> = 0.08. The analysis also detected a


Note. Data presented are means with standard deviations in parentheses. STAI, State-Trait Anxiety Inventory. Cortisol levels are measured in nanomoles per liter (nmol/l). Where cortisol levels for a participant were measured at <0.50 nmol/l, 0.45 nmol/l was used as an estimate.

significant Group × Time interaction, F(6,144) = 4.12, p = 0.001, η 2 <sup>p</sup> = 0.15.

We performed planned contrasts on the Time and Group × Time effects. Regarding the main effect of Time, the STAI<sup>B</sup> vs. STAI<sup>1</sup> contrast was not significant, t(133.39) = −1.01, p = 0.31, Cohen's d = 0.17, but the STAI<sup>B</sup> vs. STAI<sup>2</sup> and STAI<sup>1</sup> vs. STAI<sup>2</sup> contrasts were significant at the Bonferroni-corrected p of 0.017, t(144.49) = 3.27, p = 0.001, d = 0.53 and t(118.77) = 3.62, p < 0.001, d = 0.59, respectively.

Regarding the Group × Time interaction, two of the five contrasts were significant at the Bonferroni-corrected p of 0.01. The first compared change from STAI<sup>B</sup> to STAI<sup>1</sup> in the EC-Women and EC-Men groups, taken together, t(216) = −2.89, p = 0.004, d = 0.64. The second compared, at STAI1, the average STAI-State scores of the EC-Women and EC-Men groups, taken together, vs. those of the NoEC-Women and NoEC-Men groups, taken together, t(216) = 3.96, p < 0.001, d = 0.70. The other three contrasts (first, comparing change from STAI<sup>B</sup> to STAI1in the NoEC-Women and NoEC-Men groups, taken together; second, comparing, at STAIB, the scores of the EC-Women and EC-Men groups, taken together, vs. those of the NoEC-Women and NoEC-Men groups, taken together; and third, comparing, at STAI2, the scores of the EC-Women and EC-Men groups, taken together, vs. those of the NoEC-Women and NoEC-Men groups, taken together) were not significant, t<sup>S</sup> < 1.13, p<sup>S</sup> > 0.26, d<sup>S</sup> < 0.27. This set of analyses confirms that, while undergoing cognitive testing, participants in the EC and NoEC groups were experiencing different levels of self-reported anxiety.

#### Cortisol Levels

**Figure 2** shows the fluctuations in group cortisol levels across the Day 2 experimental procedures.

Mauchly's test detected a violation of the assumption of sphericity, χ 2 (2) = 8.43, p = 0.02. Hence, we corrected degrees of freedom using Huynh-Feldt estimates of sphericity, ε = 0.96. The analysis detected main effects of Time, F(1.92, 138.16) = 38.20, p < 0.001, η 2 <sup>p</sup> = 0.35, and of Group, F(3,72) = 4.75, p = 0.004, η 2 <sup>p</sup> = 0.17, and a Group × Time interaction, F(5.76, 138.16) = 17.64, p < 0.001, η 2 <sup>p</sup> = 0.43.

We performed planned contrasts on all three effects. Regarding the main effect of Time, the CORT<sup>B</sup> vs. CORT<sup>1</sup> and CORT<sup>1</sup> vs. CORT<sup>2</sup> contrasts were significant at the Bonferronicorrected p of 0.017, t(121.03) = −3.28, p = 0.001, d = 0.54, and t(107.80) = 3.97, p < 0.001, d = 0.65, respectively, but

FIGURE 2 | Changes in cortisol levels across the Day 2 experimental procedures. Error bars represent standard deviations. Data for elevated cortisol (EC) groups are presented separately from those for NoEC groups to allow easier viewing of the magnitude of error bars. CORTB, cortisol at baseline; CORT1, cortisol at the first post-manipulation measurement point (i.e., 5 min post-manipulation); CORT2, cortisol at the second post-manipulation measurement point (i.e., 45 min post-manipulation).

the CORT<sup>B</sup> vs. CORT<sup>2</sup> contrast was not, t(144.41) = 0.79, p = 0.43, d = 0.13. This set of analyses confirms an impression given by **Figure 2**: Cortisol levels at baseline were equivalent to those at 45-min post-manipulation, with levels at 5-min post-manipulation greater than those at both baseline and 45-min post-manipulation.

Regarding the main effect of Group, two contrasts (EC-Women vs. EC-Men, and the EC-Women and EC-Men groups taken together vs. the NoEC-Women and NoEC-Men groups taken together) were significant at the Bonferronicorrected p of 0.017, t(91.86) = −2.83, p = 0.006, d = 0.56, and t(161.22) = 3.45, p < 0.001, d = 0.48, respectively, while the NoEC-Women vs. NoEC-Men contrast was not, t(106.63) = −1.07, p = 0.29, d = 0.20. This set of analyses confirms that our reconstituted group assignment effectively separated those with elevated cortisol from those without.

Regarding the Group × Time interaction effect, two of the five contrasts were significant at the Bonferroni-corrected p of 0.01. The first of these compared change from CORT<sup>B</sup> to CORT<sup>1</sup> in the EC-Women and EC-Men groups, taken together, t(40.70) = −5.86, p < 0.001, d = 1.32. The second compared, at CORT1, the average cortisol levels of the EC-Women and EC-Men groups, taken together, vs. those of the NoEC-Women and NoEC-Men groups, taken together, t(46.09) = 1.57, p < 0.001, d = 1.25. The other three contrasts (first, comparing change from CORT<sup>B</sup> to CORT1in the NoEC-Women and NoEC-Men groups, taken together; second, comparing, at CORTB, the scores of the EC-Women and EC-Men groups, taken together, vs. those of the NoEC-Women and NoEC-Men groups, taken together; and third, comparing, at CORT2, the scores of the EC-Women and EC-Men groups, taken together, vs. those of the NoEC-Women and NoEC-Men groups, taken together) were not significant, t<sup>S</sup> < 1.57, p<sup>S</sup> > 0.12, d<sup>S</sup> < 0.38. This set of analyses confirms that, while undergoing cognitive testing, participants in the EC and NoEC groups were experiencing different levels of circulating cortisol.

#### Interim Summary

Analyses of the STAI-State and salivary cortisol data detected: (a) no differences between the EC and the NoEC groups at baseline or at the end-of-session measurement point, (b) differences between those groups at the immediate post-manipulation measurement point, and (c) an increase in self-reported anxiety and physiological stress markers from baseline to the immediate post-manipulation measurement in the EC, but not in the NoEC, groups.

## Working Memory Tasks: Primary Analyses Day 1

Because performance on the Day 1 n-back tasks ensured that participants understood the requirements of those tasks, we examined descriptive statistics to give an indication of participants' performance, and then conducted a repeatedmeasures ANOVA to investigate whether: (a) participants required more blocks to reach criterion performance as the WM task load increased, and (b) there were between-group


Note. Data presented indicate number of people who reached criterion on the indicated trial.

differences in number of 0-back, 1-back and 3-back blocks required to reach criterion performance.

For the 0-back task, 73 participants (96%) achieved the required 70% accuracy at their first attempt, and the rest achieved it at the second attempt. For the 1-back task, 66 participants (87%) achieved the required accuracy at their first attempt, and the rest (bar one, who achieved it at the third attempt) achieved it at the second attempt. For the 3-back task, 58 participants (76%) achieved the required accuracy at their first attempt, and most others (14 of 18) achieved it at the second attempt (see **Table 4**).

Mauchly's test detected a violation of the assumption of sphericity, χ 2 (2) = 13.70, p = 0.001. Hence, we corrected degrees of freedom using Huynh-Feldt estimates of sphericity, ε = 0.91. The analysis detected a main effect of Task Load, F(1.81,130.39) = 7.54, p < 0.01, η 2 <sup>p</sup> = 0.10, but no main effect of Group, F(3,72) = 1.42, p = 0.24, η 2 <sup>p</sup> = 0.06, and no Group × Task Load interaction, F(5.43, 130.39) = 0.58, p = 0.730, η 2 <sup>p</sup> = 0.02.

Regarding the main effect of Task Load, the 0-back vs. 1-back and 1-back vs. 3-back contrasts were not significant at the Bonferroni-corrected p of 0.017, t(110.62) = −2.10, p = 0.038, and t(127.36) = −1.90, p = 0.06, respectively. However, the 0-back vs. 3-back contrast was significant, t(90.25) = −3.57, p = 0.001.

#### Day 2

We derived two outcome variables for each block of the 1-back and 3-back tasks: Correctly Identified Stimuli (CIS; the percent of target and non-target letters each participant correctly identified) and Mean Reaction Time (MRT) to CIS. Hence, the data set comprised eight CIS variables and eight MRT variables for each participant (see **Table 5**).

#### **CIS**

We conducted a 4 × 2 × 4 (Group [EC-Women, EC-Men, NoEC-Women, NoEC-Men] × Task Load [1-back, 3-back] × Block [1, 2, 3, 4]) repeated-measures ANOVA on this set of data. Because the data were not normally distributed, we log-transformed them and analyzed the transformed set.

Mauchly's test detected a violation of the assumption of sphericity for the Block factor, χ 2 (5) = 18.58, p = 0.002, and the


#### TABLE 5 | Descriptive statistics: Day 2 n-back performance (N = 76).

Note. Data presented are means, with standard deviations in parentheses. CIS, correctly identified stimuli; MRT, mean reaction time (measured in milliseconds). <sup>a</sup>Data based on n = 16 participants. One participant in this group used an incorrect key to provide responses for all non-target trials in this block.

Task Load × Block interaction, χ 2 (5) = 17.26, p = 0.004 data. Hence, in both cases we corrected degrees of freedom using Huynh-Feldt estimates of sphericity, ε = 0.94. Subsequently, the analysis detected a main effect of Task Load, F(1,71) = 85.61, p < 0.001, η 2 <sup>p</sup> = 0.55, and of Block, F(2.83,200.64) = 3.72, p = 0.014, η 2 <sup>p</sup> = 0.05. Regarding the main effect of Task Load, participants obtained a greater overall percentage of correct responses on the four 1-back blocks (M ± SE: 94.30 ± 0.68) than on the four 3-back blocks (M ± SE: 83.39 ± 1.12). In fact, participants in all four groups appeared to perform equivalently, and almost at ceiling, on the 1-back trials. Regarding the main effect of Block, participants performed best on the first block of trials, and worst on the second and fourth blocks (M ± SE: block 1 = 90.53 ± 0.77; block 2 = 87.86 ± 1.06; block 3 = 89.72 ± 0.92; block 4 = 87.93 ± 1.10). Post-hoc pairwise comparisons, using Tukey's LSD procedure, detected differences between performance on the first and second blocks, p = 0.014, d = 0.22, first and fourth blocks, p = 0.011, d = 0.22, and second and third blocks, p = 0.045, d = 0.02.

The analysis detected no main effect of Group, F(3,71) = 0.48, p = 0.695, η 2 <sup>p</sup> = 0.02, and no interactions, F<sup>S</sup> < 1.83, p<sup>S</sup> > 0.12, η 2 pS < 0.06. Hence, neither sex nor variation in cortisol detectably affected the accuracy of n-back performance. A series of bivariate correlational analyses, documenting associations between magnitude of cortisol change from Cort<sup>B</sup> to Cort<sup>1</sup> and the CIS variables, confirmed the veracity of this fact, r<sup>S</sup> < 0.14, p<sup>S</sup> > 0.25.

#### **MRT**

As an initial analytic step, we conducted a 4 × 2 × 4 (Group (EC-Women, EC Men, NoEC-Women, NoEC-Men) × Task Load [1-back, 3-back] × Block [1, 2, 3, 4]) repeated-measures ANOVA on this set of data. Because the data were not normally distributed, we analyzed a log-transformed set.

Mauchly's test detected a violation of the assumption of sphericity for data related to the Block factor, χ 2 (5) = 11.80, p = 0.038. Hence, for that factor we corrected degrees of freedom using Huynh-Feldt estimates of sphericity, ε = 0.97. The analysis detected a main effect of Task Load, F(1,71) = 115.60, p < 0.001, η 2 <sup>p</sup> = 0.62, and Block, F(2.91, 206.83) = 3.31, p = 0.022, η 2 <sup>p</sup> = 0.04, but not Group, F(3,71) = 0.51, p = 0.679, η 2 <sup>p</sup> = 0.02.

Regarding the main effect of Task Load, participants' reaction time on the four 3-back blocks (M ± SE: 824.69 ± 31.70) was significantly slower than that on the four 1-back blocks (M ± SE: 619.89 ± 18.93). Regarding the main effect of Block, participants performed best on the third block of trials and worst on the second block (M ± SE: block 1 = 723.43 ± 26.26; block 2 = 745.21 ± 25.84; block 3 = 708.45 ± 25.36; block 4 = 712.86 ± 23.70). Post-hoc pairwise comparisons, using Tukey's LSD procedure, detected differences between performance on the first and second blocks, p = 0.045, d = 0.08, second and third blocks, p = 0.002, d = 0.14, and second and fourth blocks, p = 0.008, d = 0.13.

The analysis did not detect interactions between Group and Task Load, F(3,71) = 1.39, p = 0.254, η 2 <sup>p</sup> = 0.06, between Group and Block, F(8.74, 206.83) = 1.76, p = 0.080, η 2 <sup>p</sup> = 0.07, or between Group, Task Load, and Block, F(9,213) = 0.92, p = 0.512, η 2 <sup>p</sup> = 0.04. It did, however, detect a Task Load × Block interaction, F(3,213) = 0.091, p < 0.001, η 2 <sup>p</sup> = 0.10.

Regarding the Task Load × Block interaction, the data depicted in **Figure 3** suggest this is a product of sampling error: for the set of 3-back blocks, MRT was at its lowest on block 3, whereas for the set of 1-back blocks, MRT was at its highest on block 3. There is no apparent experimental precedent or

theoretically justifiable reason for this data pattern, and hence we make no further comment about it.

In summary, we conclude from this set of analyses that neither sex nor variation in cortisol detectably affected the speed of nback performance. A series of bivariate correlational analyses, documenting associations between magnitude of cortisol change from Cort<sup>B</sup> to Cort<sup>1</sup> and the MRT variables, confirmed the veracity of this fact, r<sup>S</sup> < 0.40, p<sup>S</sup> > 0.073.

## Working Memory Tasks: Secondary Analyses

The fact that our analyses detected no significant relations among cortisol, sex and WM performance did not confirm our hypotheses that cortisol elevations enhance WM performance in men, but impair WM performance in women. The results are also inconsistent with data patterns described elsewhere (e.g., Schoofs et al., 2008, 2013; Cornelisse et al., 2011). Hence, we conducted two additional sets of analyses, taking different approaches to the data and thereby assessing the fidelity of our initial findings.

The first set of analyses retained the participants' original group assignments, and so permitted us to investigate the effects of TSST exposure (rather than elevated cortisol) on WM performance. These analyses, then, are more similar than those above to those conducted by, for instance, Schoofs et al. (2008, 2013) and Cornelisse et al. (2011). So, using the original group assignments and the log-transformed WM data, we conducted two (one on the CIS data and one on the MRT data) 4 × 2 × 4 (Experimental Condition [TSST-Women, TSST-Men, Relax-Women, Relax-Men] × Task Load [1-back, 3-back] × Block [1, 2, 3, 4]) repeated-measures ANOVAs. The only difference between these analyses and those reported above, sections ''CIS'' and ''MRT'', is that presence of the Experimental Condition factor in place of the Group factor (i.e., the Task Load and Block data were the same as analyzed above).

These secondary analyses detected the same main effects of Task Load and of Block on CIS and MRT scores, and the same Task Load × Block interaction effect on MRT scores as did the original analysis. Most pertinent, of course is that neither secondary analysis detected main effects of Experimental Condition, F<sup>S</sup> < 0.64, p<sup>S</sup> > 0.59, η 2 pS < 0.04, and that neither detected interactions involving Experimental Condition, F<sup>S</sup> < 5.24, p<sup>S</sup> > 0.85, η 2 pS < 0.10. Hence, both secondary analyses detected no relations among the participant's sex, the experimental condition to which s/he was exposed, the interaction between the two, or accuracy/speed of n-back performance.

Second, we created two separate general linear models to test if sex, magnitude of cortisol change from baseline to CORT1, and/or the interactions among those variables, account for a significant proportion of the variance in (a) CIS across the set of 3-back trials, or (b) MRT across the set of four 3-back blocks. Neither model detected main or interaction effects: for CIS, F<sup>S</sup> < 0.78, p<sup>S</sup> > 0.380, η 2 pS < 0.02, R <sup>2</sup> = 0.01, and for MRT, F<sup>S</sup> < 1.94, p<sup>S</sup> > 0.165, η 2 pS < 0.03, R <sup>2</sup> = 0.04. Again, these analyses detected no relations among sex, magnitude of cortisol change from baseline to CORT1, their interaction and accuracy/speed of 3-back performance.

## DISCUSSION

A highly controlled design, featuring a modified version of the TSST and using 76 healthy adults (38 women in the late luteal phase of their menstrual cycle, and 38 men), tested the following predictions: (a) overall WM performance, as measured by the n-back, is faster and more accurate on a 1-back than on a 3-back task, regardless of biological sex or elevated circulating cortisol; (b) cortisol elevations enhance performance on a 3-back task in men; but (c) impair performance on a 3-back task in women.

The literature commonly reports considerable variability in response to laboratory-based stress induction methods (e.g., Buchanan and Tranel, 2008; van den Bos et al., 2009; Zandara et al., 2016). Our findings are similar: 11 participants who completed the study procedures did not respond to the TSST with measurable increases in cortisol levels. Moreover, and again as is common in this area, more women (n = 8) than men (n = 3) did not respond to the TSST with such increases (e.g., Elzinga and Roelofs, 2005; Stephens et al., 2016).

Confirming the first hypothesis, the present analyses demonstrated that, overall, participants completed the 1-back trials (presumably, representing a lighter cognitive load) more quickly and accurately than the 3-back trials (presumably, representing a heavier cognitive load). This data pattern characterized mean performance in both men and women, regardless of whether exposure to the experimental manipulation elevated cortisol. This pattern replicates data showing that people are faster and more accurate on lighterthan heavier-load WM tasks, regardless of biological sex or cortisol status (Callicott et al., 1999; Speck et al., 2000; Pelegrina et al., 2015; Rac-Lubashevsky and Kessler, 2016). This performance difference likely arises because 1-back tasks, unlike 2-back and higher tasks, require no content manipulation, are not vulnerable to interference and do not require accessing information outside of immediate attentional focus (Verhaeghen and Basak, 2005; Oberauer, 2006; Schleepen and Jonkman, 2010).

The present analyses did not confirm the second and third hypotheses, detecting no between-group differences in 3-back task performance. Hence, elevated cortisol levels did not enhance performance in men, or impair performance in women, on a heavier-load WM task.

This data pattern does not systematically replicate those described elsewhere. For example, Cornelisse et al. (2011) reported that TSST exposure facilitated faster reaction time on a 2-back (but not a 3-back) task in men. Schoofs et al. (2013), using only a 2-back task, replicated that result and found that stress exposure was associated with slower reaction times in women. Our findings are, however, consistent with those described in other studies reporting that stress-induced cortisol elevations do not affect WM performance in men (Kuhlmann et al., 2005; Oei et al., 2009; Weerda et al., 2010; Stauble et al., 2013) or women (Cornelisse et al., 2011; Zandara et al., 2016).

## On Failures to Systematically Replicate

These failures to systematically replicate data patterns under apparently similar conditions may be due to a number of between-study methodological differences including, but not limited to, the task used to represent working memory (e.g., n-back vs. digit span); the characterization of light and heavy cognitive loads (e.g., whereas some use 1-back vs. 2-back, others use 2-back vs. 3-back, or variations of the Sternberg WM paradigm); the outcome measures reported (e.g., accuracy vs. reaction time); the extent of practice on the tasks representing working memory (e.g., whereas some give a 0-back task and a series of practice trials at the beginning of each n-back block, others give two practice blocks preceding 2- or 3-back testing); if tasks representing working memory occur in isolation or as part of a task battery; the procedures used to elevate cortisol levels (e.g., hydrocortisone vs. psychosocial manipulations); the use of manipulation checks to exclude cortisol non-responders from at least some analyses; the stage of female participants' menstrual cycle (e.g., whereas some include only female participants in the luteal phase, others include all women and do not use phase of cycle as a covariate in analyses); or if women using oral contraceptives are permitted to participate or not (Elzinga and Roelofs, 2005; Kuhlmann et al., 2005; Schoofs et al., 2008, 2013; Luethi et al., 2009; Oei et al., 2009; Weerda et al., 2010; Cornelisse et al., 2011; see Brown et al., 2014; for a detailed discussion and checklist).

In the present study, we considered the methodological choices that allowed us to best answer our questions of interest. So, for instance, we chose to use an n-back task because it appears to represent core WM processes more accurately than, for instance, digit span backwards. At the theoretical level, the n-back requires manipulation and organization of information, especially at heavier task loads (Kirchner, 1958; Veltman et al., 2003). At the level of experimental control, the n-back offers the option of including both accuracy and reaction time measures and can be of longer duration than digit span, where each difficulty level is typically tested with only two trials (Jarrold and Towse, 2006; Lynn and Irwing, 2008; Schoofs et al., 2008). Finally, at the empirical level, imaging studies indicate that n-back tasks activate the PFC reliably, and that this activation increases as cognitive load increases (Speck et al., 2000; Owen et al., 2005; Tomasi et al., 2007). Hence, even though some, based on tradition (Kane et al., 2007), or authority (Miller, 1956), justify digit span as a gold-standard measure of WM, performance on that task does not necessarily reflect the characteristics of WM as well as the n-back task does.

We also chose to control, by design rather than statistically, factors related to female hormonal states. Specifically, we excluded women using oral contraceptives, and we tested naturally-cycling women only in the late luteal phase of the menstrual cycle. As noted earlier, stressor-induced salivary cortisol levels obtained from men and from women in the late luteal phase are similar. Moreover, compared to men and to women in other phases of the cycle, women using oral contraception or who are in the follicular phase exhibit smaller stressor-induced cortisol increases (Kirschbaum et al., 1995, 1999).

Finally regarding methodological choices, we reconstituted our participant groups into those with and those without cortisol elevations. Doing so allowed us to focus exclusively on the major question of interest (i.e., relations between elevated cortisol and WM performance), and to maximize statistical power to detect differences relating to cortisol levels, rather than to stress exposure per se.

Regardless of our group reconstitution, the present pattern of cortisol data differs markedly from those reported in many other studies. The average baseline cortisol level in our sample was 1.92 (± 2.20) nmol/l. Post-manipulation, the average cortisol level in the EC groups was 5.65 (± 4.14) nmol/l. These numbers are notably smaller than those reported by most other studies using the same stress-induction method (Liu et al., 2017). Typically, studies in this field report baseline levels of over 5 nmol/l (at least one reported levels above 20 nmol/l), and post-TSST peak levels of over 10 nmol/l (at least one reported levels above 50 nmol/l; see, e.g., Kirschbaum et al., 1996; Wolf et al., 2001; Domes et al., 2004; Elzinga and Roelofs, 2005; Kuhlmann et al., 2005; Oei et al., 2006; Nater et al., 2007; Schoofs et al., 2008, 2013; Luethi et al., 2009; Cornelisse et al., 2011; Zandara et al., 2016). In the present study, the average increase in cortisol post-manipulation for the EC groups was 4.10 (± 3.47) nmol/l, a figure at the lower end of the range of increases reported by other studies, some of which describe average values of over 9 nmol/l (Kirschbaum et al., 1996; Wolf et al., 2001; Domes et al., 2004; Oei et al., 2006, 2009; Nater et al., 2007; Luethi et al., 2009; Cornelisse et al., 2011; Schoofs et al., 2013; Zandara et al., 2016). Indeed, based on the classification criteria suggested by Miller et al. (2013), 25% of participants in our EC groups (9 of 36) were ''cortisol nonresponders.''

The source(s) of differences between the pattern of cortisol data described in the present study and those described in some previously published studies cannot be easily explained. As in most other studies, our participants were undergraduate volunteers and we enforced standard exclusion criteria strictly. We were faithful to the TSST methodology, and an accredited and experienced laboratory staff analyzed our saliva samples (Pillay et al., 2008). Furthermore, previously published studies (Human et al., 2013; du Plooy et al., 2014) and unpublished theses and dissertations using similar samples consistently describe similar trends toward relatively low baseline cortisol levels and relatively small magnitudes of cortisol response.

We therefore propose a set of testable hypotheses, the core of which is this: ''These between-study cortisol differences can be traced to the fact that our samples are drawn from the southern African population, a more racially, ethnically, culturally and genetically heterogeneous population than those from which Western European and North American laboratories sample'' (Manica et al., 2007). For instance, the final sample in the present study comprised 32 white, 22 black African, and 22 colored (mixed ancestry) or Indian/Asian individuals (we use these racial terms following the descriptive nomenclature present in South African census publications; Statistics South Africa, 2015). Although previously published studies do not provide detailed racial/ethnic classification data, it is improbable that their samples were as diverse as ours.

The relatively small literature on individual racial/ethnic/cultural differences in baseline cortisol, and in magnitude of cortisol response to psychosocial stressors, may shed light on the observed discrepancies between our physiological data and those reported elsewhere. Chong et al. (2008), for example, found that, after TSST exposure, white American participants showed a significantly greater increase in cortisol levels than African-American participants. Other non-experimental studies conducted in North America report that black children have both lower basal cortisol levels and flatter cortisol curves than white children. Importantly, differences in socioeconomic status (SES) appear to strongly moderate these group differences, with black participants in those studies typically sampled from lower-SES neighborhoods or families (Chen and Paterson, 2006; Desantis et al., 2007; Dulin-Keita et al., 2012; but see Lupien et al., 2001). Although none of those studies were conducted in low- or middle-income countries (LAMICs), it is possible that low-SES individuals in LAMICs (a characterization that fits many of the participants in our sample) show diurnal cortisol rhythms (i.e., lower basal levels and flatter curves) similar to those observed in low-SES individuals in high-income countries (HICs) of North America and western Europe.

In addition, it is plausible that high-SES individuals in LAMICs have greater exposure to prenatal stress, lifetime stress and racial discrimination than high-SES individuals in HICs. Each of these adverse experiences bear a relation to both basal cortisol levels and magnitude of cortisol response to psychosocial stressors (Williams et al., 2008, 2012; Jorm and Ryan, 2014).

Furthermore, the ancestral life history (LH) strategies of individuals residing in LAMICS may differ from those residing in HICs. LH theory is an evolutionary biological framework that accounts for individual differences in reproductive strategies, suggesting that organisms inhabiting unstable environments with high mortality rates tend to produce many offspring and invest little in them (i.e., adopt faster LH strategies; see, e.g., Promislow and Harvey, 1990). Applied to humans, LH theory has guided studies seeking to understand individual differences in, for instance, the onset of puberty, age of first sexual experience and number of sexual partners (see, e.g., Belsky et al., 1991; Patch and Figueredo, 2017). Recent studies suggest cross-cultural and cross-national variations in LH strategy are associated with, for instance, variations in: (a) the parenting strategies to which individuals are exposed (Sotomayor-Peterson et al., 2012), and (b) the national prevalence of polymorphisms in the androgen receptor gene AR, the dopamine receptor gene DRD4, and the 5-HTTLPR VNTR of the serotonin transporter gene (Minkov and Bond, 2015). Of particular relevance here is that the latter study showed that nations with higher levels of socioeconomic inequality (as measured by the Gini coefficient) contained more individuals with those genetic polymorphisms, which are associated with a greater propensity toward risk-taking, dysfunctional impulsivity and unrestricted sociosexual attitudes (i.e., traits that covary with faster LH strategies). According to the World Bank (2017), South Africa has a Gini coefficient of 0.63, making it one of the most economically unequal countries in the world.

Animal studies demonstrate that variations in LH strategies are related to variations in glucocorticoid concentrations and in the strength of the HPA axis-driven physiological stress response. Although some evidence suggests that a faster LH strategy is present when basal hormone levels are lower and when stress responsiveness is dampened, there remains some debate about the precise direction and nature of this relationship (Crespi et al., 2013; Crossin et al., 2016). No published human studies address this question.

In summary, we suggest that the racial, socioeconomic and other variability we observe in our laboratory's samples serve as proxies for long-standing cultural differences that have produced remarkable inter-individual (and perhaps genetically expressed) differences in human responses to environmental challenges (Cohen et al., 1996; Sapolsky, 2017). We could disconfirm these and related hypotheses by using a quasi-experimental design that: (a) assigns individuals to groups based on factors such as measured region of origin, individual and country SES, exposure to prenatal and lifetime stress, exposure to instances of racial/ethnic/cultural discrimination, fast vs. slow LH strategy and other variables hypothesized to be critical, and then (b) examines trends toward relatively low baseline cortisol levels and relatively small magnitudes of cortisol response in specified groups.

We have highlighted potential cultural, socioeconomic and ancestral life-history differences between participants in our study and those in previously published studies. We speculate that these differences contribute to the observed betweenstudy differences in patterns of cortisol secretion (Oberlander et al., 2008; Huynh et al., 2016; Lovallo et al., 2017), and explain, at least partially, why the outcomes we present here are inconsistent with those obtained from less heterogeneous samples (e.g., Cornelisse et al., 2011; Schoofs et al., 2013). Specifically, if basal cortisol levels in our participants are relatively low, and the magnitude of cortisol response is also relatively small, then cortisol levels are unlikely to elevate to the supraphysiological levels known to affect cognitive performance significantly (Henry et al., 2014; Schultebraucks et al., 2015).

## Limitations

Although the control procedure we used is similar to that described by others (e.g., Elzinga and Roelofs, 2005; Cornelisse et al., 2011), it did not require participants to complete cognitive and physical tasks formally equivalent to those completed by participants in the TSST condition. The literature describes other, perhaps more intuitively appealing, control procedures (e.g., Het et al., 2009; Wiemers et al., 2013). There are, however, no empirical demonstrations or accepted standards providing guidance in these matters (i.e., there are no direct comparisons of different types of control procedures in studies with aims similar to ours).

Unlike several other studies in this literature (e.g., Cornelisse et al., 2011; Schoofs et al., 2013), we did not collect biomarkers of the sympathetic nervous system's response to the TSST (e.g., salivary alpha-amylase; Petrakova et al., 2015). However, our focus here was on possible sequelae of cortisol elevation, rather than on those of an overall stress response (i.e., both sympathetic and HPA-axis activation), on WM performance. Hence, the collection and analysis of additional biomarker data was superfluous. Moreover, we could not use statistical analyses more sensitive to detecting sex differences in cortisol response to the psychosocial stressor (e.g., growth curve modeling; Lopez-Duran et al., 2014) because post-manipulation cortisol sampling was not dense enough. Such analyses require samples at 5–10-min intervals in the hour immediately following stressor offset.

Finally, our group of participants was too small to run meaningful analyses on cortisol samples drawn from individuals with different cultural, racial, ethnic and/or SES backgrounds. Hence, we could not give proper statistical consideration to relations among, for instance, adverse life experiences, LH

## REFERENCES


strategies, cortisol responses and WM performance. As noted above, such consideration is important when analyzing data from studies in this research area, particularly if those studies are conducted using samples drawn from LAMICs in the global south.

## SUMMARY AND CONCLUSION

Our data analyses detected no effects of sex or cortisol variation on WM task performance. This result is inconsistent with previous reports examining sex differences, effects of stress and WM performance. However, the finding that baseline cortisol levels and magnitude of cortisol increases in our sample are substantially lower than those of other studies in the stress and cognition literature raises questions about: (a) relations among environment and physiological responses to stressors, and (b) inter-individual differences in relations between stress and cognitive performance. These questions require further investigation, especially within culturally, ethnically and socioeconomically diverse populations such as those in South Africa.

## AUTHOR CONTRIBUTIONS

RH conceptualized and designed the study, collected and analyzed data, wrote the first draft of the manuscript and was involved in re-writing and editing the final version of the manuscript. MH assisted in conceptualizing and designing the study, collected data and was involved in re-writing and editing the final version of the manuscript. WJJ was involved in re-writing and editing the final version of the manuscript. KT assisted in conceptualizing and designing the study, and was involved in re-writing and editing the final version of the manuscript.

## ACKNOWLEDGEMENTS

We thank the National Health Laboratory Services at Groote Schuur Hospital for performing all salivary cortisol analyses.


Val158Met (rs4680) genotypes on the adult cortisol response to psychological stress. Psychosom. Med. 79, 631–637. doi: 10.1097/PSY.0000000000 000481


**Conflict of Interest Statement**: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Human, Henry, Jacobs and Thomas. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Behavioral and Immunohistochemical Study of the Effects of Subchronic and Chronic Exposure to Glyphosate in Mice

#### Yassine Ait Bali , Saadia Ba-Mhamed and Mohamed Bennis \*

Laboratory of Pharmacology, Neurobiology and Behavior (URAC-37), Faculty of Sciences Semlalia, Cadi Ayyad University, Marrakech, Morocco

Many epidemiological studies have described an adolescent-related psychiatric illness and sensorimotor deficits after Glyphosate based herbicide (GBH) exposure. GBH exposure in animal models of various ages suggests that it may be neurotoxic and could impact brain development and subsequently, behavior in adulthood. However, its neurotoxic effects on adolescent brain remain unclear and the results are limited. The present study was conducted to evaluate the neurobehavioral effects of GBH following acute, subchronic (6 weeks) and chronic (12 weeks) exposure (250 or 500 mg/kg/day) in mice treated from juvenile age until adulthood. Mice were subjected to behavioral testing with the open field (OF), the elevated plus maze, the tail suspension and Splash tests (STs). Their behaviors related to exploratory activity, anxiety and depression-like were recorded. After completion of the behavioral testing, adult mice were sacrificed and the expression of tyrosine hydroxylase (TH) in the substantia nigra pars compacta (SNc) and serotonin (5-HT) in the dorsal raphe nucleus (DRN), the basolateral amygdala (BLA) and the ventral medial prefrontal cortex (mPFC) was evaluated using immunohistochemical procedure. Our results indicate that unlike acute exposure, both subchronic and chronic exposure to GBH induced a decrease in body weight gain and locomotor activity, and an increase of anxiety and depression-like behavior levels. In addition, the immunohistochemical findings showed that only the chronic treatment induced a reduction of TH-immunoreactivity. However, both subchronic and chronic exposure produced a reduction of 5-HT-immunoreactivity in the DRN, BLA and ventral mPFC. Taken together, our data suggest that exposure to GBH from juvenile age through adulthood in mice leads to neurobehavioral changes that stem from the impairment of neuronal developmental processes.

Keywords: glyphosate, locomotor activity, anxiety, depression, tyrosine-hydroxylase, serotonin, basolateral amygdala, medial prefrontal cortex

## INTRODUCTION

The widespread use of pesticides in various domains increases the risk of environmental contamination by different xenobiotics which can be potentially noxious for non-target organisms including human beings. The use of organophosphate (OP) Glyphosate (Gly) has been globally expanded. Indeed, around two-thirds of the total volume of Glyphosate-based herbicides (GBH) used has been delivered to the environment during the last decade (Myers et al., 2016),

#### Edited by:

Vivienne Ann Russell, University of Cape Town, South Africa

#### Reviewed by:

Haiyun Xu, Shantou University, China Jessica Anne Siegel, University of St. Thomas, United States Zhimin Song, University of Michigan, United States Jacqueline Samantha Womersley, Stellenbosch University, South Africa

> \*Correspondence: Mohamed Bennis mbennis@uca.ma

Received: 21 May 2017 Accepted: 21 July 2017 Published: 08 August 2017

#### Citation:

Ait Bali Y, Ba-Mhamed S and Bennis M (2017) Behavioral and Immunohistochemical Study of the Effects of Subchronic and Chronic Exposure to Glyphosate in Mice. Front. Behav. Neurosci. 11:146. doi: 10.3389/fnbeh.2017.00146 in step with the increased adoption of genetically engineered crops such as corn, soy and canola. This GBH is the active ingredient present in Roundup<sup>r</sup> (Monsanto Company, St. Louis, MO, USA), currently the most heavily used herbicide worldwide (Powles et al., 1997).

Organophosphate (OP) exposure has been reported to be linked with mood disorders, such as depression and anxiety (Landrigan et al., 1999; London et al., 2005; Lee et al., 2007). Previous studies have demonstrated the potential risks of these pesticides, including their neurotoxic effects on developing organisms (Eaton et al., 2008). Furthermore, epidemiological studies have reported that long-term exposure to OPs elicits many neurological disturbances (Steenland et al., 1994; Roldan-Tapia et al., 2006). While classified under the herbicide class phosphonomethyl amino acids, Gly is often mischaracterized as an OP. This is likely due to the molecular structure being an organic molecule containing a phosphorus atom. However, clinical reports describing incidents of human ingestion of Gly do not reflect the classic symptoms for OP poisoning (salivation, lacrimation, urination, and defecation (Costa, 2008). Conversely, Seneff et al. (2015) highlighted the strong correlations between the increasing application of Gly in agriculture and the apparent surge of several neurological diseases at different ages. Abnormal EEG activity characterized by limb tremor (akinesia and rigidity) observed in Parkinsonian syndrome have been reported after professional exposure or accidental ingestion of a commercial mixture of Gly (Barbosa et al., 2001; Malhotra et al., 2010; Wang et al., 2011). In addition, data derived from structural MRI studies highlighted in a subject exposed to Gly changes in the T2 signal in substantia nigra, periaqueductal gray and globus pallidus, revealing likely lesions in these structures (Barbosa et al., 2001). Moreover, in human studies, Gly has been detected in the brain and cerebro-spinal fluid following commercial mixtures exposure, attesting that the active component can traverse the blood brain barrier (Menkes et al., 1991; Sato et al., 2011).

Research on animal models confirmed the neurotoxic effects of Gly. Indeed, Negga et al. (2012) revealed that exposure to GBH cause neuronal death, especially of GABAergic and dopaminergic neurons in the nematode Caenorhabditis elegans. It was also reported that Gly induces acetylcholinesterase activity alteration in the brain and the muscle of fishes exposed either to pure Gly or Roundup<sup>r</sup> (Modesto and Martinez, 2010; Menéndez-Helman et al., 2012; Sandrini et al., 2013; Samanta et al., 2014). An increasing body of literature revealed that Gly was able to provoke oxidative stress in specific rat brain regions such as substantia nigra, cerebral cortex and hippocampus (Astiz et al., 2009; Cattani et al., 2014).

In addition, the exposure to the GBH Roundup<sup>r</sup> during pregnancy and lactation provokes decrease in glutamate uptake and metabolism within glial cells, and increasing glutamate release in the synaptic cleft in the hippocampus of orally exposed rat's offspring, leading to glutamate excitotoxicity (Gallegos et al., 2016).

Apart from the prenatal or postnatal period, the adolescence age is also a critical phase of brain development (Van Waes et al., 2011). Indeed, adolescence is a critical ontogenic period characterized by the maturation of brain processes that underlie higher cognitive functions, social and emotional behaviors (Spear, 2000). This period is often one of increased vulnerability and adjustment (Steinberg, 2005). In humans and rodents, adolescence is a time of extensive neural reorganization/pruning, in which long-term deleterious effects due to drug and environmental factor exposures are greater than at later times in life (Spear, 2000; Mendola et al., 2002).

The evident structural and neurochemical changes during adolescence are characterized by increases in functional connectivity between regions of the prefrontal cortex and several areas of the limbic system, especially the dopaminergic and serotoninergic systems, concomitant with changes in emotional and cognitive function regulation (Wahlstrom et al., 2010; Naneix et al., 2012; Spear, 2013). Moreover, during adolescence, brain areas continue to develop, such as the amygdala, hippocampus, and prefrontal cortex (Giedd and Rapoport, 2010). Thus, disturbing serotoninergic and dopaminergic systems in the aforementioned brain regions may result in permanent behavioral alterations associated with motor and emotional functions. In light of this, an important determinant that remains to be investigated is whether GBH exposure, following exposure to a Gly-containing herbicide, acts directly on the adolescent and adult brain resulting in subsequent sensorimotor and emotional function controlled by serotoninergic and dopaminergic systems.

## MATERIALS AND METHODS

## Animals

Male Swiss mice (1 month old) were obtained from the animal husbandry of the Faculty of Sciences, Cadi Ayyad University, Marrakech, Morocco. The choice of the use of males in this study was based on the fact that several studies have reported a difference in sensitivity of the brain to the toxicity to several substances linked either to a sex difference in the permeability of the cerebrospinal blood barrier (Kato, 1974), or to increased susceptibility to toxins in females with respect to males (Sonawane and Baksi, 1980; Rhodes and Rubin, 1999; Kim et al., 2005). The animals were housed in Plexiglas cages (6 mice/cage; 30 cm × 15 cm × 12 cm) under standard conditions of temperature (22 ± 2 ◦C) and photoperiod 12 h/12 h (lights on at 08:00 h). Food and water were available ad libitum. All procedures were conducted in accordance with approved institutional protocols, and with the provisions for animal care and use prescribed in the scientific procedures on living animals, European Council Directive: EU2010/63. All efforts were made to minimize any animal suffering. The study was approved by the Council Committee of Research Laboratories of the Faculty of Sciences, Cadi Ayyad University, Marrakech.

## Pesticide

Roundup<sup>r</sup> herbicide (Gly concentration 360 g/l in the form of Gly isopropylamine salt 486 g/l) with molecular formula C6H17N2O5P, molecular weight of 228.183 g/Mol, melting point 200◦C and density 1.218 g/cm3 was used in the liquid commercial form supplied by Monsanto Company (St. Louis, MO, USA).

## Acute, Subchronic and Chronic Toxicity Assessment of GBH

### Study Design

Healthy mice were assigned to three experimental groups (acute, subchronic and chronic: n = 18/group), and each group was subjected either to orally gavages by NaCl 0.9% (control n = 6), by 250 mg (n = 6) or 500 mg/kg/day (n = 6) of GBH. These doses were selected on the basis of Gly no-observed adverse effect level (NOAEL) of 500 mg/kg/day for subchronic and chronic toxicity (EPA, 1993). Gly solutions were prepared daily to minimize the risk of degradation since Gly half-life in water varies from 49 days to 70 days (Mercurio et al., 2014). The mice assigned to the acute group received one administration of GBH while the subchronic and chronic groups were treated daily for 6 and 12 weeks, respectively. All mice were observed and their body weights were measured daily during the exposure period. Following the end of treatment, all groups of mice were tested for their locomotor activity and affective behavior on consecutive days. On the last day of the experiment, only the subchronic and chronic treated animals were sacrificed to perform vital organs weight analysis and immunochemistry study (**Figure 1**).

#### Effects of GBH Exposure on Mouse Behavior

All treated animals were tested between 9:00 h and 12:00 h in the light cycle. Before behavioral testing, the animals were gently handled and they were individually familiarized with the testing room and the test arena for 5 min prior to the test circumstance. To minimize subjectivity, the behavior of mice was recorded and analyzed using Ethovision XT Noldus 8.5 video tracking program (Noldus Information Technology b.v., Wageningen, Netherlands), connected to a video camera (JVC). The video camera was positioned 2.5 m above the arena, inside the vertical projection of a wall, covering the entire view of the arena. Tracking of the animal was based on contrast relative to the background. Two tracking points were specified: on the head and the center of gravity of the animal.

## **Open field test**

The Open field (OF) test is the one of the commonly used test to assess locomotor activity and emotional reactivity in rodents placed into novel environment (Wilson et al., 1976). The Activity monitoring was carried out in black Plexiglas arena measuring 50 × 50 × 50 cm, equipped with the video-based Ethovision System (Noldus, Wageningen, Netherlands). The OF apparatus was illuminated by a 75W lamp placed in porthole diffusing light and located at 200 cm from the device, allowing the center of the arena to be under a dim light (100 lx). Mice were placed individually into the arena and locomotor activity was assessed for 20 min using the video-tracking system. Parameters recorded were total distance moved, velocity, and the percentage of the time spent in the arena center (15 × 15 cm). The test box was cleaned with 70% ethanol between each test.

## **Elevated-plus-maze test**

The elevated-plus-maze (EPM) is the widely paradigm used to assess anxiety-like behavior in rodents (Handley and Mithani, 1984; Lapiz-Bluhm et al., 2008). The EPM is comprised of two opposing open arms (50 × 5 cm) and two closed arms (50 × 5 × 15 cm), which are joined at a square central area (5 × 5 cm) to form a plus sign. The maze floor and the side/end walls (15 cm height) of the enclosed arms were made of clear Plexiglas. The entire apparatus was elevated to a height of 45 cm above the floor. The EPM apparatus was illuminated by a 75W lamp, located at 200 cm from the device, allowing an approximate brightness of 200 lux. For testing, the mouse was removed gently from its home cage and placed in the central arena of the EPM, facing the junction of an open and closed arm. The mouse was allowed to freely explore the maze for 5 min, while its behavior was recorded for offline analysis. The time spent in the open arm over the total time spent in both arms, the number of entries to the open arm vs. the total number of entries corresponding to the ratio of time spent and the ratio of open arm entries, respectively, and the anxiety index, which is 1 − (ratio time + ratio entry)/2, were quantified during the test session.

## **Tail suspension test**

The Tail Suspension test (TST) is one of the common behavioral tests adopted to assess the potential antidepressant-like effects in rodents (Cryan and Holmes, 2005). Each individual mouse was suspended by the tail with adhesive tape above the ground, approximately 40 cm from the floor. The course of the TST over a single 6 min session was scored. The immobility time during the last 4 min of a session was recorded.

## **Splash test**

The Splash test (ST) is widely used to evaluate the motivational deficits and self-care difficulties as symptoms of depression in rodents (David et al., 2009; Amiri et al., 2015). In this test, the decrease of grooming behavior of mice (considered as an index of self-care and set of emotional and behavioral alterations, including persistent depressed mood and loss of interest or pleasure as core symptoms: Vural et al., 2007) was measured. Briefly, a 10% sucrose solution was squirted on the dorsal coat of animals in their home cage and the time spent grooming was recorded for a period of 5 min after the sucrose application.

## Tissue Preparation

Upon completion of behavioral testing, subchronic and chronic treated mice were anesthetized with an intraperitoneal injection of urethane 40% (1 g/kg, from Sigma–Aldrich, France) and transcardially perfused with saline solution (0.9%), followed by 4% paraformaldehyde in phosphate buffered saline (PBS; 0.1 M). The brain, liver, kidneys and lung were removed and weighed. The relative weight was calculated as organ weight/body weight. The brains were removed, post-fixed in the same fixative for 12 h

and cryoprotected overnight in 30% sucrose. They were then cut on a freezing microtome (Leica Microsystems, Germany) into 40 µm frontal sections containing substantia nigra pars compacta (SNc), the dorsal raphe nucleus (DRN), the ventral part of the medial prefrontal cortex (mPFC) and the basolateral nucleus of the amygdala (BLA). These regions of interest were determined according to stereotaxic atlas of Paxinos and Franklin (2001).

#### Immunochemistry

Slides containing the region of interest were incubated with hydrogen peroxide at 1% for 30 min to block endogenous peroxidase activity, and washed five times for 10 min in PBS solution. Then they were incubated for 60–90 min in PBS solution containing bovine serum albumin-containing (BSA; SERVA ELECTROPHORESIS GmbH, Germany) and 1% Triton X-100 (ACROS ORGANICS, USA) at 0.3% (PBS-T-BSA), under stirring at room temperature. Thereafter, the sections were incubated with anti-tyrosine hydroxylase (TH) polyclonal primary antibody produced in rabbit (SIGMA ALDREICH GmbH, Germany) or with an anti-serotonin (5-HT) polyclonal antibody produced in rabbit (CALBIOCHEM, mAb (GA-5) IF03L, Germany) diluted 1:1000 and 1:5000, respectively, in PBS-T-BSA overnight, under stirring at 4◦C. Sections were washed five times in PBS for 10 min each, and then incubated with rabbit biotinylated secondary antibody diluted 1:400 in PBS-T under stirring overnight at 4◦C, and rinsed five times in TBS for 10 min each. Sections were then rinsed in PBS, and incubated for 1 h 30 at room temperature in avidin-biotin-peroxidase complex (Vectastain Elite ABC Kit, Vector Laboratories, Burlingame, CA, USA) diluted at 1:200 in PBS. After rinsing in PBS, the peroxidase activity was visualized with 3-3<sup>0</sup> -diaminobenzidine tetrahydrochloride (SIGMA ALDREICH GmbH, Germany) at 0.025% in Tris buffer (Pre-set) 0.05 M, (PROCHILAB, France) containing hydrogen peroxide (0.006%). Processing was stopped by rinsing sections for 3 × 10 min in PBS. Sections were mounted on gelatinecoated slides, dehydrated in graded series of ethanol, cleared in xylene and cover-slipped with Eukitt, and examined under a light microscope for the quantification of TH or 5-HT immunoreactivity.

## Acquisition, Image Processing and Quantification of Immunoreactive Neurons

The images of brain sections were acquired at high magnification (×100 and ×400) on an Olympus BH-2 microscope equipped with a camera Olympus DP71. The reconstruction of a complete photomicrograph association of different shots and their treatment was made through image processing software Adobe Photoshopr. For each brain region, four representative sections from anterior to posterior were chosen and counted to minimize variability. Individual means were obtained by quantification of the intensity of immunostaining for TH and 5-HT bilaterally in four sections from each region, by an experimenter blind to the treatment conditions, using ImageJ software. Sections were chosen by correspondence to the Paxinos and Franklin (2001) stereotaxic atlas. TH densitometry was quantified in SNc (Interaural = 0.88 mm, Bregma = −2.92 mm), 5-HT immunoreactive cells were quantified in the DRN (rostral part: Interaural = −0.80 mm, Bregma = −4.60 mm; caudal part: Interaural = −1.04 mm, Bregma = −4.84 mm), the ventral mPFC (Interaural = 5.78 mm, Bregma = 1.98 mm) and the BLA (Interaural = 1.98 mm, Bregma = −1.82 mm).

## Statistical Analysis

To compare data from behavioral testing with immunohistochemical findings between the treated groups and control, a statistical analysis of these different independent variables was performed by two way ANOVA (treatment and the period of treatment), using the Sigma Plot software 11.0. Data from the body and organ weight were analyzed using two way ANOVA with treatment as the factor. Post hoc analysis was performed using Holm-Sidak post hoc test. The distribution of the data related to each measure was assessed automatically for their normality by the software before performing any kind of statistical analysis. Results are presented as mean ± standard error of the mean (SEM). The significance threshold was set at p < 0.05.

## RESULTS

## GBH Effect on Body Weight Gain

Body weight gain was not affected in mice after acute exposure to the two GBH concentrations tested on day 1 (F(2.17) = 0.46; p = 0.64) and on day 7 (F(2.17) = 0.06; p = 0.94). However, subchronic and chronic exposure at doses of either 250 mg/kg or 500 mg/kg of GBH, resulted in reduced body weight gain of mice over the 15-day experimental period. On day 30, the treated animals showed a slight recovery of their body weight that remained statistically significant compared with controls (**Table 1**).

## GBH Effect on Organ Weight

The organ relative weights of both treated and control groups are presented in **Table 2**. Two-way ANOVA analysis showed a significant difference in the weight of brain, liver, kidneys and lung among the factor of treatment (F(2.17) = 13.51, p < 0.001; F(2.17) = 12.13, p < 0.001; F(2.17) = 10.76, p < 0.001; F(2.17) = 7.97, p < 0.001, respectively), and only in the brain relative weight for the period factor (F(2.17) = 39.69, p < 0.001), as well as the interaction of treatment × period for brain and liver relative weights (F(2.17) = 11.70, p < 0.001; F(2.17) = 5.15, p < 0.05). The post hoc analysis confirmed that treated groups, especially the 500 mg/kg group, exhibited significant decrease in organ's relative weight following GBH-exposure (p < 0.05; **Table 2**).

## Behavioral Changes after GBH Exposure

#### Open Field Test

As indicators of locomotor activity and anxiety-like levels, we registered the total distance traveled over the maze (activity) and the time spent in the central area. Two-way ANOVA analysis of the total distance traveled, the velocity and the percentage of the time spent in the central zone revealed significant differences among the factors of treatment (F(2.17) = 18.93, p < 0.001; F(2.17) = 3.06, p < 0.001; F(2.17) = 51.18, p < 0.001, respectively) and period (F(2.17) = 76.21, p < 0.001; F(2.17) = 71.16, p < 0.001; F(2.17) = 38.04, p < 0.001 respectively) as well as an interaction of treatment × period for velocity and the percentage of time spent in the central zone (F(2.17) = 11.02, p < 0.001; F(2.17) = 13.22, p < 0.001). The post hoc analysis using Holm-Sidak test confirmed that treated groups, especially the 500 mg/kg group, exhibited significant decrease in all parameters recorded during the OF test following subchronic and chronic GBH-exposure (p < 0.05; **Figures 2A–C**).

### Elevated-Plus-Maze Test

Two-way ANOVA test demonstrated significant difference in ratio of time spent in the OA and the anxiety index recorded during the EPM among the factors treatment (F(2.17) = 1.57, p < 0.001; F(2.17) = 17.75, p < 0.001 respectively) and period (F(2.17) = 24.35, p < 0.001; F(2.17) = 8.77, p < 0.001, respectively) as well as the interaction of treatment × period (F(2.17) = 10.48, p < 0.001; F(2.17) = 4.53, p < 0.001, respectively). However, the number of open arm entries showed no significant difference following GBH exposure (treatment (F(2.17) = 1.57, p > 0.05) and period (F(2.17) = 1.55, p > 0.05)). There was no interaction for treatment × period (F(2.17) = 0.83, p > 0.05). The post hoc comparisons confirmed that the treated groups, especially 500 mg/kg group, showed a significant decrease in the time spent in the open arm (p < 0.001) and increase in the anxiety index (p < 0.05) with respect to control (**Figures 3A–C**).

### Tail Suspension and Splash Tests

Two way ANOVA analysis of the immobility and grooming time revealed a significant differences among the factors treatment (F(2.17) = 20.12, p < 0.001; F(2.17) = 39.39, p < 0.001, respectively) and period (F(2.17) = 42.02, p < 0.001; F(2.17) = 92.41, p < 0.001, respectively) as well as in the interaction treatment × period (F(2.17) = 7.31, p < 0.001; F(2.17) = 19.89, p < 0.001, respectively). The post hoc comparisons using Holm-Sidak test confirmed that the treated groups showed a significant increase in the immobility time only following chronic treatment (p < 0.001) and a decrease in grooming time (p < 0.05) after both subchronic and chronic exposure (**Figures 4A,B**).

## Immunochemistry

#### TH-Immunoreactivity

Using TH, a key enzyme involved in dopamine synthesis, two-way ANOVA analysis of the TH immunoreactivity (TH+) revealed significant difference among the factors treatment (F(2.17) = 107.41, p < 0.001) and period (F(2.17) = 16.39, p < 0.001) as well as in the interaction of treatment × period (F(2.17) = 69.04, p < 0.001). The post hoc comparisons confirmed that the treated groups showed a significant decrease in TH<sup>+</sup> only following chronic treatment (p < 0.001; **Figure 5**).

### Serotonin Immunoreactivity

Two way ANOVA analysis of 5-HT immunoreactivity in the rostral and caudal parts of the DRN, the mPFC and basolateral amygdala (BLA) nucleus revealed significant differences for period in all regions studied (F(2.17) = 122.24, p < 0.001; F(2.17) = 73.41, p < 0.001; F(2.17) = 15.32, p < 0.001; F(2.17) = 28.98, p < 0.001, respectively), and in the rostral part of the DRN (F(2.17) = 28.16, p < 0.001; **Figures 6**, **7**). There was no significant difference in the interaction between treatment × period (p > 0.05). The post hoc comparison confirmed that the treated groups, especially 500 mg/kg group, showed a significant decrease in 5-HT densitometry in all structures studied following both subchronic and chronic treatment in respect to the control (p < 0.05; **Figures 6**, **7**).

## DISCUSSION

Because of the scarce information on the neurobehavioral effects of GBH in mammals, the present research was designed to gain insights concerning the effects of acute or repeated exposures to GBH on developing brain of juvenile and adult mice.


Results are presented as mean ± SEM. ∗∗P < 0.01, ∗∗∗P < 0.001: ANOVA analysis. ##P < 0.01, ###P < 0.001, \$P < 0.05, \$\$\$P < 0.001: the post hoc analysis. The "#" comparison to the control. The "\$" compared to the 250 mg.


Results are presented as mean ± SEM. #p < 0.05, ##p < 0.01, ###P < 0.001, \$\$P < 0.01, \$\$\$p < 0.001: the post hoc analysis. The "#" comparison to the control. The "\$" comparison to the 250 mg.

Our results showed that subchronic and chronic exposure to GBH in mice induced a significant decline in body weight gain for both treated groups. These results are in agreement with those obtained using comparable doses in adult mice (Jasper et al., 2012) and in rabbits (Yousef et al., 1995). According to Chahoud et al. (1999), weight loss is the main important indicator of toxicity, which may be associated with Roundupr's capacity to provoke reactive oxygen species production. In addition, the decrease in the food consumption and the anorexic effect of Gly could lead to lower relative organs' weight (ROW; Beuret et al., 2005). Thus, the assessment of ROW clearly revealed that GBH exposure induced a significant diminution of brain and liver weights after 6 weeks of treatment, while it affected brain, kidneys, liver and lung's weight following 12 weeks of treatment. Several studies showed that Gly residues in the kidney, liver and lung were comparable to those found in the urine. This means that the Gly does not pass through the urine without affecting the organism (Krüger et al., 2014; Seneff et al., 2015). These results are consistent with previous studies in F344/N rats and B6C3F1 mice (World Health Organisation (WHO), United Nations Environment Programme, the International Labour Organisation, 1994). Similar results have been found in animals exposed to other pesticides such as deltamethrin, fenvalerate and diazinon (Kalender et al., 2006; Kilian et al., 2007). Indeed, in the same studies, the decrease of body weight gain recorded was due to malabsorption of nutrients induced by the gastro-intestinal tract impairment or inhibition of protein synthesis.

Acute exposure to GBH showed no change in locomotor activity. In contrast, subchronic and chronic treated animals showed a hypoactive profile. Our results are similar to those obtained by Hernández-Plata et al. (2015), who showed that repeated intraperitoneal injections of 100 or 150 mg/kg of Gly causes hypoactivity in adult male rats and those of Gallegos et al. (2016) in prenatally treated rats. Likewise, rats treated with methylparathion OP (Sun et al., 2006),

FIGURE 3 | Effect of the acute and repeated exposures to GBH on anxiety-like behavior. (A) The ratio of time spent in open arm. (B) The ratio of open arm entries. (C) The anxiety index. Ratio = time spent in each arm/(time spent in open arm + time spent in closed arm). Results are presented as mean ± SEM. <sup>∗</sup>P < 0.05; ∗∗∗P < 0.001. The ∗∗∗ refers to the control vs. 250 mg/kg and 500 mg/kg group comparison and the "#" refers to the 250 mg/kg vs. 500 mg/kg group comparison. #P < 0.05: the post hoc analysis.

group comparison and the "#" refers to the 250 mg/kg vs. 500 mg/kg group comparison. #P < 0.05, ##P < 0.01: the post hoc analysis.

diisopropylfluorophosphate pesticide (Bushnell et al., 1991) or dichlorvos-exposed (Binukumar et al., 2010) exhibited decreases in locomotor activity. Corroborating these findings, several OPs have been linked to movement and coordination deficits in humans following occupational exposures (Lotti and Moretto, 2005; Ehrich and Jortner, 2010). Further, many studies have reported that the locomotor activity reduction is positively correlated with the loss of dopaminergic neurons in the substantia nigra, as well as DA receptors (Bano et al., 2014; Gallo et al., 2015). Indeed, both the nigrostriatal and mesolimbic dopaminergic systems are implicated in the control of motor and motivated behaviors, and they have been shown to be susceptible to herbicides. For example, exposure to pesticides has frequently been defined as a major contributor to the risk of PD development (Hatcher et al., 2008).

During adolescence, Peak levels of extracellular dopamine are observed (Philpot et al., 2009) and that firing of dopamine cells and neurotransmitter turnover in target regions is higher in adolescents than adults (Tarazi et al., 1998; Moll et al., 2000; Placzek et al., 2009; McCutcheon et al., 2012). Moreover, increased studies reported the vulnerability of the nigrostriatal dopaminergic system following exposure to pesticides during critical periods of neurodevelopment, thus compromising the integrity and viability of this system (Cory-Slechta et al., 2005a,b). In light of this, the alterations of the dopamine system seen in our study coupled with locomotor hypoactivity are particularly interesting, when compared with previous work that demonstrated a reduction in the TH levels that paralleled motor activity deficiency after paraquat and maneb pesticide (Thiruchelvam et al., 2000). In agreement with our results, the exposure to Gly-containing pesticide leads to degeneration of GABA and dopamine neurons in Caenorhabditis elegans (Negga et al., 2012) and induces loss of cardiolipin content and mitochondrial transmembrane potential, especially in SNc, with a concomitant increase in fatty acid peroxidation (Astiz et al., 2009). According to Jackson et al. (2002) and Ricci et al. (2003),

FIGURE 5 | Effect of subchronic and chronic exposure to GBH on dopaminergic neurons. (A) Photomicrographs of mice brain cross sections showing the tyrosine hydroxylase (TH)-immunoreactive neurons at the SNc. (B) The intensity of TH Immunoreactivity at the SNc in control and treated mice. SNc, substantia nigra pars compacta. Results are presented as mean ± SEM. ∗∗∗P < 0.001. The ∗∗∗ refers to the control vs. 250 mg/kg and 500 mg/kg group comparison and the "#" refers to the 250 mg/kg vs. 500 mg/kg group comparison. ###P < 0.001: the post hoc analysis.

mitochondrial transmembrane potential loss reflects an early stage of apoptosis and release of many mitochondrial proteins into the cytosol during the first steps of programmed cell death. Additionally, Gui et al. (2012) reported a cell death in the dopaminergic cell line PC12 by Gly through apoptotic and autophagic mechanisms. Recently, Hernández-Plata et al. (2015) showed a reduction of the D1-DA receptor binding in the nucleus accumbens, accompanied by a significant decrease in ambulatory activity observed after Gly administration. Furthermore, following only 150 mg/kg of Gly administration, it was reported that striatal extracellular DA levels decreased by 50% in the microdialysis sample (Hernández-Plata et al., 2015).

On the other hand, the results revealed that subchronic and chronic exposure to the GBH induced an anxiogenic-like behavior. Similarly to our data, Sánchez-Amate et al. (2001) and Chen et al. (2011) found that repeated exposure of rats to another OP chlorpyrifos induces anxiogenic-like behavior. Conversely, Gallegos et al. (2016) observed that prenatal exposure to 100 or 200 mg/kg of GBH produced an anxiolytic-like effect in rats. These differential results may probably due to the different exposure periods. In fact, our study focused mainly on juvenile and adult age, whereas Gallegos et al. (2016) paid more attention to the prenatal exposure period.

FIGURE 7 | Effect of subchronic and chronic exposure to GBH on serotoninergic fibers. (A) Photomicrographs of mice brain cross sections showing the 5-HT-immunoreactive fibers and the intensity of 5-HT immunoreactivity in the mPFC. (B) Photomicrographs of mice brain cross sections showing the 5-HT-immunoreactive fibers and the intensity of 5-HT immunoreactivity in the basolateral nucleus of the amygdala. mPFC, ventral medial prefrontal cortex; BLA, basolateral nucleus of amygdale; IL, the infralimbic cortex corresponding to the ventral mPFC. The arrows refer to 5-HT fibers. Results are presented as mean ± SEM. ∗∗P < 0.01; ∗∗∗P < 0.00. The ∗∗∗ refers to the control vs. 250 mg/kg and 500 mg/kg group comparison and the "#" refers to the 250 mg/kg vs. 500 mg/kg group comparison. #P < 0.05: the post hoc analysis.

Moreover, our results revealed that GBH exposed mice showed significant depression-like behavior. Our findings were in agreement with Chen et al. (2014) who showed that adolescent rats exposed to chlorpyrifos exhibited depression-like phenotype. Moreover, this result shows similarities to those of other OPs (Aldridge et al., 2004; Roegge et al., 2008). Indeed, previous researches that investigated the effects of methamidophos and diazinon at neonatal age and adulthood detected behavioral changes associated with depression (Lima et al., 2009) and changes in neurochemical markers of 5-HT function (Lima et al., 2011).

It is well established that several neuronal systems, in particular the serotoninergic system innervating cortical and limbic brain structures, are involved in the pathophysiology of anxiety and depression related behaviors (Azmitia and Segal, 1978; Vertes, 1991). Indeed, it has been shown that 5-HT system sends anatomical connections to the amygdala (Bobillier et al., 1976; Sadikot and Parent, 1990) and the ventral mPFC (Steinbusch, 1981; O'Hearn and Molliver, 1984) as pivotal structures contributing to emotional disturbances. Moreover, besides the changes in 5-HT transporters and their receptor activity, there is an increase in the release of 5-HT from the DRN in adolescence in comparison with adult age (de Jong et al., 2006), and the levels of 5-HT in several brain areas are also increased. Thus, The alteration of serotoninergic homeostasis during its development may result in a pattern of modified brain connections, leading to the loss of synapses and dendritic arborization, which normally occurs from puberty to adulthood (Barros et al., 2006; Zhang et al., 2013) and permanent behavioral alterations may be induced in adults (Whitaker-Azmitia, 2005; Ansorge et al., 2008).

Furthermore, the ventral mPFC is of special interest because it is strongly implicated in the expression of behavioral and autonomic responses to emotionally relevant stimuli, and imaging studies highlight abnormalities in the structure and function of this region in patients with mood disorder (Kennedy et al., 2001; Drevets et al., 2008). In agreement with these findings, our immunohistochemical data have shown a decrease of immunoreactive-neurons of the DRN and serotoninergic fibers in the BLA and in the ventral mPFC of mice following subchronic and chronic GBH exposure. Our results joined those obtained by Anadón et al. (2008) showing that Gly administered orally in rats decreases 5-HT and dopamine levels in the frontal cortex, midbrain and striatum. All these suggested that GBH induces depression and anxiety-like behavior observed in the treated groups.

On the other hand, the findings support a crucial role for perturbed amygdala 5-HT and enhanced glutamatergic mechanisms in the pathophysiology of mood disorders. Indeed, previous evidence supports the possibility that emotional disorders involve neuronal hyperexcitability in the amygdala (Tran et al., 2013). In support, Tran et al. (2013) showed that 5-HT depletion leads to the hyperexcitability of the central amygdala as the output of the nucleus, elucidating the anxiety-like behavior. Moreover, findings indicate that the depletion of 5-HT causes ventral mPFC neuron stimulation that excites principal BLA neurons elucidating anxiety phenotype (Russchen, 1982; McDonald, 1998). According to these findings, our results show that obvious serotoninergic circuitry disruptions observed in the DRN, the BLA and the ventral mPFC could underlie some structural and functional alterations potentiating the hyperexicitability of the amygdala and promote anxiety-like behavior.

While the mechanism underlying GBH-induced TH and 5-HT immunoreactivity reduction is not evaluated in the present study, it appears to be largely mediated by reactive oxygen species. As Gly can act as protonophore increasing mitochondrial membrane permeability to protons and Ca2<sup>+</sup> (Olorunsogo, 1990), it can trigger the production of reactive oxygen species resulting in oxidative stress (de Liz Oliveira Cavalli et al., 2013). In human cell lines, both Gly and Roundup increased necrosis and apoptosis (Gasnier et al., 2009; Mesnage et al., 2013). Similarly, in rats, treatment with Gly and Roundup generated oxidative stress, induced lipid peroxidation (Astiz et al., 2009; El-Shenawy, 2009) and thus affected the oxidation-antioxidation homeostasis, promoting apoptosis and cellular death (Shimada et al., 1998; Yang and Sun, 1998), thereby inducing the anxiety and a depression-like phenotype and reducing motor development in the animals exposed to GBH.

In conclusion, we report for the first time the neurotoxic effect of subchronic and chronic subtoxic Roundup<sup>r</sup> exposure in mice. Overall the results reported in the present study indicate that, as previously described for other OPs, GBH exposure during the juvenile and adult period leads to an alteration in the activity level of animals and their affective performances paralleled by the dopaminergic and serotoninergic system impairment.

## REFERENCES


These results support the epidemiological findings that pesticideexposed populations are susceptible to neurobehavioral changes.

## AUTHOR CONTRIBUTIONS

YAB, SB-M and MB designed the experiments; performed the analysis of the data; YAB and MB performed the experiments; assembled the figures. All authors wrote or edited and validated the manuscript.

## ACKNOWLEDGMENTS

The authors thank Pamela D. Garzone, R.Ph., M.Sc., Ph.D (Early Oncology Development and Clinical Research Pfizer ORRD, South San Francisco) for editing the manuscript.


associated with exposures: a consensus statement. Environ. Health 15:19. doi: 10.1186/s12940-016-0117-0


by prenatal stress. Neuroscience 250, 333–341. doi: 10.1016/j.neuroscience. 2013.04.031

**Conflict of Interest Statement**: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Ait Bali, Ba-Mhamed and Bennis. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Brain Metal Distribution and Neuro-Inflammatory Profiles after Chronic Vanadium Administration and Withdrawal in Mice

Oluwabusayo R. Folarin<sup>1</sup> , Amanda M. Snyder <sup>2</sup> , Douglas G. Peters <sup>2</sup> , Funmilayo Olopade<sup>3</sup> , James R. Connor <sup>2</sup> and James O. Olopade<sup>4</sup> \*

<sup>1</sup>Department of Medical Laboratory Science, Ladoke Akintola University of Technology, Osogbo, Nigeria, <sup>2</sup>Department of Neurosurgery, Pennsylvania State College of Medicine, Hershey, PA, United States, <sup>3</sup>Department of Anatomy, University of Ibadan, Ibadan, Nigeria, <sup>4</sup>Department of Veterinary Anatomy, University of Ibadan, Ibadan, Nigeria

Vanadium is a potentially toxic environmental pollutant and induces oxidative damage in biological systems including the central nervous system (CNS). Its deposition in brain tissue may be involved in the pathogenesis of certain neurological disorders which after prolonged exposure can culminate into more severe pathology. Most studies on vanadium neurotoxicity have been done after acute exposure but in reality some populations are exposed for a lifetime. This work was designed to ascertain neurodegenerative consequences of chronic vanadium administration and to investigate the progressive changes in the brain after withdrawal from vanadium treatment. A total of 85 male BALB/c mice were used for the experiment and divided into three major groups of vanadium treated (intraperitoneally (i.p.) injected with 3 mg/kg body weight of sodium metavanadate and sacrificed every 3 months till 18 months); matched controls; and animals that were exposed to vanadium for 3 months and thereafter the metal was withdrawn. Brain tissues were obtained after animal sacrifice. Sagittal cut sections of paraffin embedded tissue (5 µm) were analyzed by the Laser ablation-inductively coupled plasma-mass spectrometry (LA–ICP–MS) to show the absorption and distribution of vanadium metal. Also, Haematoxylin and Eosin (H&E) staining of brain sections, and immunohistochemistry for Microglia (Iba-1), Astrocytes (GFAP), Neurons (Neu-N) and Neu-N + 4<sup>0</sup> ,6-diamidine-2<sup>0</sup> -pheynylindole dihydrochloride (Dapi) Immunofluorescent labeling were observed for morphological and morphometric parameters. The LA–ICP–MS results showed progressive increase in vanadium uptake with time in different brain regions with prediction for regions like the olfactory bulb, brain stem and cerebellum. The withdrawal brains still show presence of vanadium metal in the brain slightly more than the controls. There were morphological alterations (of the layering profile, nuclear shrinkage) in the prefrontal cortex, cellular degeneration (loss of dendritic arborization) and cell death in the Hippocampal CA1 pyramidal cells and Purkinje cells of the cerebellum, including astrocytic and microglial activation in vanadium exposed brains which were all attenuated in the withdrawal group. With exposure into old age, the evident neuropathology was microgliosis, while progressive astrogliosis became more attenuated. We have shown that chronic administration of

#### Edited by:

Nouria Lakhdar-Ghazal, Faculty of Sciences, Mohammed V-Agdal University, Morocco

#### Reviewed by:

Paul Manger, University of the Witwatersrand, South Africa Luis Gerardo Aguayo, University of Concepción, Chile

\*Correspondence: James O. Olopade jkayodeolopade@yahoo.com

Received: 23 March 2017 Accepted: 28 June 2017 Published: 25 July 2017

#### Citation:

Folarin OR, Snyder AM, Peters DG, Olopade F, Connor JR and Olopade JO (2017) Brain Metal Distribution and Neuro-Inflammatory Profiles after Chronic Vanadium Administration and Withdrawal in Mice. Front. Neuroanat. 11:58. doi: 10.3389/fnana.2017.00058 vanadium over a lifetime in mice resulted in metal accumulation which showed regional variabilities with time. The metal profile and pathological effects were not completely eliminated from the brain even after a long time withdrawal from vanadium metal.

Keywords: vanadium, LA–ICP–MS, neuro-inflammation, neurotoxicity, withdrawal

## INTRODUCTION

Vanadium (V) is a metalloid widely distributed in the environment and it exerts potent toxic effects on a wide variety of biological systems. While some derivatives of vanadium have been found to be useful in medicine and industry (Ray et al., 2006), environmental and occupational exposure to this metal continues to be a health risk to humans and animals (Shrivastava et al., 2007).

Exposure to neurotoxic metals such as vanadium occurs through various sources including heavy metals mining (Moskalyk and Alfantazi, 2003), combustion products of vanadium bearing fuel oils (Amorim et al., 2007), forest fires and volcanic emissions; in addition, large quantities of vanadium compounds have been reported to be released into the environment mainly through the burning of fossil fuels having vanadium contaminated crude as seen in oil producing communities such as Venezuela, the Arabian Gulf, the Gulf of Mexico and the Nigerian Niger Delta (Olopade and Connor, 2011; Saxena et al., 2013; Fortoul et al., 2014). In addition, vanadium accumulates in the soil, groundwater and plants that may be consumed by both animals and humans (Pyrzy´nska and Wierzbicki, 2004).

Current opinion is that vanadium induces oxidative stress caused by reactive oxygen species (ROS) generation in vitro, as well as lipid peroxidation and oxidative damage and this action has been strongly linked to vanadate induced effect in biological systems (Evangelou, 2002; García et al., 2004), and this in turn may cause neurotoxicity in humans and animals (Haider et al., 1998; Chen et al., 2001; García et al., 2004; Olopade et al., 2011). Earlier studies have shown that vanadium crosses the blood brain barrier (García et al., 2004) to induce neuropathology including neurobehavioral (Li et al., 2013; Saxena et al., 2013; Mustapha et al., 2014; Folarin et al., 2016), neurochemical (Sasi et al., 1994; García et al., 2004) and neurocellular (Domingo, 1996; Garcia et al., 2005; Avila-Costa et al., 2006) changes. In humans, features of acute neurotoxicity include central nervous system (CNS) perturbations, (CNS) depression, tremor, impaired conditioned reflexes, as well as congestion of brain and spinal cord (Haider et al., 1998; Soazo and Garcia, 2007).

Most animal experiments on neuro-cellular studies involving vanadium metal have been based on acute exposure while in reality many people occupationally (Fortoul et al., 2014) and environmentally (Olopade and Connor, 2011) exposed to vanadium are so exposed for decades or even a life time. Few studies have reported progressive neuro-cellular and neuroinflammatory changes induced by long term vanadium exposure. Azeez et al. (2016) showed functional deficit, glial cell activation and region-dependent myelin damage in the brain of mice after 90 days of postnatal vanadium exposure. Our previous work has shown that a life time administration of vanadium in mice leads to memory deficits and progressive recovery after long period of treatment withdrawal (Folarin et al., 2016).The present study is designed to assess brain vanadium distribution patterns, and cellular (glial and neuronal) injury after a long term exposure, as well as the related progressive changes in the brain after withdrawal from treatment.

## MATERIALS AND METHODS

## Animal Experiments

All experiments were approved and carried out in accordance to the guidelines of the animal use and ethics committee of the University of Ibadan, ethical code number UI-ACUREC/App/2016/011.

## Experimental Design

A total of 85 male BALB/c mice (4 weeks old) were used for the experiment which covered a period of 18 months. The animals were bred and housed in the experimental animal house of the Neuroscience unit of the Department of Veterinary Anatomy, University of Ibadan. The animals were pellet-fed, given tap water ad libitum and were kept at 27◦C with natural light and dark cycles. The animals were assigned to one of the following animal groups: vanadium- (V-) treated, control and withdrawal groups.

## Animal Design

V-treated group consisted of six subgroups of 12 animals. The subgroups are designated as V3, V6, V9, V12, V15 and V18. The mice (from 4 weeks of age) were intraperitoneally (i.p.) administered with 3 mg/kg b.w/day of vanadium (sodium metavanadate, Sigma-Aldrich, St. Louis, MO, USA), i.p. thrice a week for 3, 6, 9, 12, 15 and 18 months. This dose and route of administration is based on the findings of García et al. (2004) as it is neurotoxic with minimal mortalities. Sample mice were sacrificed every 3 months until the animals were 18 months post exposure.

Control group consisted of six subgroups of 12 animals. The subgroups are designated as C3, C6, C9, C12, C15 and C18. The mice (from 4 weeks of age) were intraperitoneally administered with sterile water, i.p. thrice a week for 3, 6, 9, 12, 15 and 18 months which was volume matched with the V-treated group. Sample mice were sacrificed as above.

Withdrawal group consisted of five subgroups of 12 animals. The subgroups are designated as W3, W6, W9, W12 and W15. The mice (from 4 weeks old) were intraperitoneally administered with 3 mg/kg b.w./day of vanadium (sodium metavanadate Sigma-Aldrich, St. Louis, MO, USA), i.p. thrice a week only

laser power with 0.4 J/cm<sup>2</sup> confluence, 320 µm/s laser speed, 20 Hz, 0.25 s Integration. (A–D, control brains), (E–H, vanadium exposed brains), (I–K, withdrawal brains) at 3, 6, 15 and 18 months time points of treatments.

for the first 3 months and then vanadium administration was stopped. Subsequently, the animals were treated as done in controls. Sample mice were sacrificed after withdrawal from treatment every 3 months till 18 months.

## Sample Collection

The mice were anesthetized with ketamine and then perfused transcardially with 4% phosphate buffered formalin with the aid of a perfusion pressure pump and brains were removed according to the method described by Olopade et al. (2011). Briefly, the frontal, parietal and temporal bones were removed to expose the brain which was gently scooped out, post-fixed for 4 h in the same solution, then processed and embedded in paraffin blocks as described by Mustapha et al. (2014). Sections were cut on a standard microtome at 5-µm thickness from paraffin embedded tissue, sectioned at Sagittal level 17, (Bregma lateral coordinates 0.675 µm, from Allen reference Mouse Brain Atlas, 2016), containing the neocortex, basal ganglia, midbrain and brain-stem.

## Sample Preparation for LA–ICP–MS

Cut sections were mounted on silane-coated soda-glass microscope slides (StarFrostr; ProSciTech, USA). Sections were dewaxed in xylene (Sigma, USA) and decreasing concentrations of ethanol (Sigma, USA) in water according to standard protocols. Samples were finally washed in deionized water (18.2 MΩ; Merk Millipore) and dried at room temperature before analysis.

## LA–ICP–MS Imaging

The LA–ICP–MS was done according to the method described by Hare et al. (2009, 2012). Briefly, sections imaged at 80 µm spatial resolution (total area = 6.4 mm<sup>2</sup> /pixel) were ablated with a 193 nm New Wave excimer ArF source laser, which was fitted with a ''TwoVol2'' ablation cell with collection cup just above the ablation spot. The laser unit was connected to a Thermo X-Series II quadrupole ICP–MS instrument. The instrument contained Xt cones at the plasma interface. The sample cone was Ni with a Cu core and the skimmer cone was Ni. The collision cell was turned off to increase sensitivity. We tuned the quadrupole while ablating NIST 612 glass standard, attempting to maximize signal while keeping 238U/232Th ratio near 1.2 and minimizing 248ThO/232Th below 3%. We set a low energy fluence of 0.4 J/cm<sup>2</sup> and laser power of 3% combined with a high repetition rate of 20 Hz to ensure proper ablation of tissue. Ablation lines

were set parallel without any overlap. We ablated the tissue at 320 µm/s and measured <sup>13</sup>C, <sup>31</sup>P, <sup>51</sup>V, <sup>52</sup>Cr and <sup>53</sup>ClO with a sampling rate of 0.25 s so that each voxel contained a single sampling of each element. Qualitative counts for V were

vanadium exposure (M; 6EB vs. 18EB = <sup>∗</sup>p < 0.05). Magnification: ×200, inset: ×400.

compared to Cr to ClO to ensure there wasn't a false positive signal, and age matched control tissue was ablated at the same time to validate any observable difference in the experimental group.

two blinded investigators and average of N = 5 per group counts are shown per HPF (20×) in (D–F) respectively. The differences between groups (J–L) were evaluated for significance using ANOVA (∗p < 0.05; ∗∗p < 0.01). Astrocytic response decreases into old age with increasing vanadium exposure (M; 6EB vs. 18EB = <sup>∗</sup>p < 0.05). Magnification: ×200, inset: ×400.

## Immunohistochemistry

Paraffin sections were dewaxed, rehydrated and immersed in distilled water. Antigen retrieval was done in 10 mM citrate buffer (pH = 6.0) for 25 min, with subsequent peroxidase quenching in 3% H2O2/methanol. All the sections were blocked in 2% milk for 1 h and probed with the following antibodies overnight: anti-GFAP Rabbit Polyclonal antibody for astrocytic morphology (1:1000; Dako, Denmark), anti-NeuN Rabbit monoclonal antibody for neuronal morphology (1:3000; Abcam, Cambridge, MA, USA) and anti-Iba-1 Rabbit polyclonal antibody for microglia morphology (1:1000; Abcam, Cambridge, MA, USA) for 16 h at 4◦C. After washing, the sections were incubated for 2 h at room temperature in the appropriate biotinylated secondary antibodies (diluted 1:200; all purchased from Vector Labs). The sections were then reacted in avidinbiotin-peroxidase solution (ABC kit, Vectastain, Vector Labs, USA) using 3,30-diaminobenzidine as chromogen, according to manufacturer's protocol. Images were acquired with Nikon bright-field microscope equipped with digital camera. For NeuN immuno-fluorescent labeling, sections were probed with anti-NeuN mouse monoclonal antibody (1:1000; Abcam, Cambridge, MA, USA) diluted at optimal working

concentrations in 1.5% Normal Goat Serum (NGS) in PBS for 16 h at 4◦C, Sections were then incubated for 2 h with fluorescent-conjugated secondary antibodies (anti-mouse IgG Alexa Flour555 antibodies (Molecular Probes) at 1:150 dilution, 4 0 ,6-diamidine-2<sup>0</sup> -pheynylindole dihydrochloride, (DAPI; 1:1000; Invitrogen, USA) staining was used in all fluorescence staining conditions to identify nuclear DNA in the cell type.

## Image Analysis

To avoid experimenter bias during all phases of image collection and analysis, slides were coded and experimenter was blind to animal condition. Prefrontal cortex, dorsal CA1 and CA3 region, genu of corpus callosum and white matter of the cerebellum was imaged using a spinning disc laser confocal system (Nikon Eclipse 80i.) equipped with ×4 ×10 and ×40 dry and ×100 oil objectives connected to a camera (Nikon DS-Fi1, NIS-Elements BR 3.2 software). Identical light intensity and exposure settings were applied to all images taken for each experimental set. The brain regions studied were chosen because of previous studies which showed vanadium leads to memory loss (Folarin et al., 2016) and loss of motor function (García et al., 2004). The prefrontal cortex and hippocampal region are cognition centers and are associated with chronic metal pollution and neurodegenerative changes (Calderón-Garcidueñas et al., 2012).

## Stereological Analysis

Immunostaining was quantified using standardized stereological method based on ''systematic random

∗∗p < 0.01). Magnification: ×200 inset: ×400.

sampling''. Astrocytes and microglia lying within the CA1 region of the hippocampus, genu of corpus callosum and white matter of the cerebellum were counted manually by two blinded investigators using ×20 images, and compared blindly between control, vanadium treated and the withdrawal groups. For each mouse Neu-N immunolabeling was quantitatively analyzed using ×20 and ×40 images within the prefrontal and hippocampal region of CA1 and CA3 to show pyknosis and neuronal loss, the number of undegenerated purkinje cells in the cerebellar cortex from H and E stained slides were also counted and analyzed using ImageJ analysis.

## Statistical Analysis

All data were expressed as mean ± standard deviation. Comparison between groups was performed using one way analysis of variance (ANOVA). A p < 0.05 was considered significant. All statistical analyses were carried out using GraphPad Prism Version 4 (GraphPad Software, San Diego, CA, USA).

## RESULTS

The LA–ICP–MS showed evidence that vanadium crosses the Blood Brain Barrier, enters the brain tissue and is deposited in different regions (see **Figure 1**). The LA–ICP–MS results also

cortical pyramidal cells showed morphological alterations including pyknosis, cell clustering, loss of layering pattern and cytoplasmic vacuolation in the vanadium exposed groups (white arrows in B,E,H) relative to the control (red arrows in A,D) with normal neuronal morphology. The withdrawal groups (C,F,I) showed reversal effect with less cellular toxicity relative to vanadium exposed groups. The vanadium cytotoxicity was also observed in the aged control mice brain (white arrows in G), indicative of neuronal degeneration seen also at early vanadium exposure. Magnification: ×400, inset: ×600.

showed progressive increase in vanadium uptake with time in the treated brains relative to the controls while the withdrawal group showed marked elimination of vanadium metal from the brain indicated by reduced relative intensities in comparison with the treated groups.

Vanadium treatments for 3–18 months resulted in astrocytic activation in the dorsal hippocampal CA1 region and genu of corpus callosum. GFAP immunoreactive cells displayed thickened cell body with more extensive cytoplasmic processes in the vanadium exposed groups (**Figures 2**, **3B,E,H**) relative to the age matched controls (**Figures 2**, **3A,D,G**) while the withdrawal groups (**Figures 2**, **3C,F,I**) showed less reactivity relative to vanadium exposed groups. Quantitative analysis of GFAP immunostaining also confirmed these observations (**Figures 3J–L**), however in the withdrawal groups, the analysis revealed non-significant decrease in astrocytic number relative to the vanadium treated groups. This result also showed that astrocytic response decreases with increasing age and period of exposure (**Figures 2**, **3M**). IBA-1 (microglia) immunoreactive cells in the vanadium exposed groups (**Figures 4**, **5B,E,H**) were markedly larger with several short, thickened processes, relative to the control brains (**Figures 4**, **5A,D,G**) with microglial cells in resting state, while the withdrawal brains (**Figures 4**, **5C,F,I**) showed less reactivity relative to the vanadium treated brains. Microglial activation occurred progressively and reaches phagocytic state indicated by amoeboid isoform (**Figure 5M**). Microglial cell count/HPF of hippocampal CA1 region and white matter of the cerebellum (**Figures 4**, **5J–L**) were significantly elevated in vanadium exposed brains relative to the controls while the withdrawal groups showed less reactivity compared to the treated groups. This result also showed that microgliosis increased with increasing age and period of exposure (**Figures 4**, **5M**).

NeuN immunohistochemistry revealed damaged pyramidal cells of the prefrontal cortex with morphological alterations including pyknosis, cell clustering, loss of layering pattern and cytoplasmic vacuolation in the vanadium exposed brains (white arrows in **Figures 6D–F**) relative to the control (red arrows, **Figures 6A,B**) with normal neuronal morphology.

FIGURE 7 | HPF photomicrographs showing the pyramidal cells of the prefrontal cortex using the NeuN immunohistochemistry. No remarkable abnormality was observed in the cortical sections from animals of the control groups. The normal neurons were identified by their rounded and pale nuclei (see A red arrow), whereas degenerating neurons had smaller cell bodies and pyknotic nuclei (see B white arrow). There was evidence of vacuolation of neuropil surrounding the degenerating neurons. The withdrawal brain (C) showed less cellular pathology relative to the exposed brains. Quantitative analysis (D) showed that the mean % pyknosis of vanadium exposed groups were significantly (∗∗P < 0.001) elevated relative to the control brain while the withdrawal groups were significantly (∗∗P < 0.001) less than vanadium exposed groups. Magnification: ×400.

The withdrawal groups (**Figures 6G–I**) showed reversal effect relative to vanadium exposed groups. The cellular pathology observed in the vanadium exposed brain were also seen in the aged control mice brain (white arrows in **Figure 6C**). Quantification of pyknotic neurons of the total NeuN positive cells in the prefrontal cortex of vanadium exposed groups was significantly elevated relative to the controls while the withdrawal groups was significantly less than vanadium exposed group (**Figure 7**). NeuN + Dapi immuno- fluorescent staining in prefrontal cortex was used to confirm the immunohistochemistry data. This co-localization revealed pyramidal cells with intact but shrunken nucleus and cytoplasmic loss (**Figure 8**). NeuN immunohistochemistry in hippocampal CA1 revealed progressive loss of apical dendrites of the pyramidal cells in vanadium exposed mice (**Figures 9B,E,H**) in comparison with controls (**Figures 9A,D,G**) while the withdrawal groups (**Figures 9C,F,I**) showed less dendritic loss relative to vanadium exposed groups. Quantification of the total NeuN immunoreactive cells in the CA1 and CA3 neurons showed that mean total cells of vanadium exposed groups was significantly reduced relative to the control brains while the withdrawal groups was significantly higher than vanadium exposed groups (**Figures 10**, **11**). In addition, histology showed neuronal loss from the cerebellar cortex. The mean total count of undegenerated Purkinje cells of vanadium exposed groups was significantly reduced relative to the control brain (**Figure 12**) while the withdrawal groups was significantly higher than vanadium exposed groups.

## DISCUSSION

The LA–ICP–MS showed evidence that vanadium crosses the Blood Brain Barrier, enters the brain parenchyma and is deposited in different regions. Cumulative evidence has revealed that the brain barriers are subject to toxic insults from heavy metal exposure (Zhen et al., 2002). Progressive increase in vanadium accumulation in the exposed brains is indicative of an increase in vanadium uptake into the brain over time. Avila-Costa et al. (2005) showed that vanadium accumulation in the brain after exposure depends more on the duration of exposure than the concentration of administration and also strongly correlated with the CNS damage induced. They further reported that the severity of ependymal epithelium disruption after vanadium pentoxide inhalation strongly correlated with the duration of exposure.

This to the best of our knowledge is the first report of LA–ICP–MS–based vanadium distribution in the brain of mice over a lifetime and has conclusively shown that vanadium metal crosses the Blood Brain Barrier and accumulates in the brain. Our study shows evidence that with chronic exposure, vanadium has a predilection for the olfactory bulb, brain stem and cerebellum. This supports the work of Haider et al. (1998), Garcia et al. (2005) and Ngwa et al. (2014) who reported specific pathologies in the olfactory bulb, brain stem and cerebellum respectively. Further work is ongoing by us to ascertain if predilection of accumulation translates to severity of pathologies relative to other brain regions.

This study has also shown evidence of clearance of vanadium metal from the brain after cessation of exposure indicating that following withdrawal, the brains have low potential for retention of the absorbed vanadium. While transferrin is known to actively transport vanadium into the brain (Usende et al., 2016), the exact mechanism of vanadium clearance from the brain needs to be investigated.

We investigated in this study microglia activation in response to chronic vanadium neurotoxicity in the hippocampus, genus of corpus callosum and cerebellum. The degree of microglial activation correlates with the duration of vanadium exposure and neuronal damage both in the hippocampal region and the cerebellum. This is indicative of the recruitment and activation of microglial cells by vanadium induced oxidative stress (Block and Calderón-Garcidueñas, 2009) resulting from free radicals production (Tsuda et al., 2004). The microglial activation persisted throughout the duration of exposure and reached a phagocytic state indicated by amoeboid isoforms mainly in the cerebellar white matter. Microglia activation has been reported to be readily triggered by environmental toxicant (Block and Hong, 2007) and may undergo a morphological change into amoeboid shape with short or non-existent processes (Kreutzberg, 1996) to favor phagocytosis and mobility. Upregulation of glial cells has also been associated with pathogenesis of senile

neurodegenerative conditions like Alzheimer's disease. Hence, we presume that in the present study, upregulation of glial cells could be suggestive of neurodegenerative effects in old age which was prematurely induced by vanadium metal.

Astrocytes play a key role in the regulation of the neuronal, and especially synaptic, microenvironment and are key elements in the brain parenchyma defense against oxidative and toxic insults (Sofroniew and Vinters, 2010; Heller and Rusakov, 2015).

confocal microscopy. Magnification: ×200.

exposed groups (B,E,H). Magnification: ×200, inset: ×400.

The number of visible astrocytes and the complexity of their processes may also be related to the extent of neural injury. Vanadium-induced astrogliosis has previously been reported in the cerebellum and hippocampus of adult rats exposed to sodium metavanadate for 5 days (Garcia et al., 2005) indicating a rapid response of astrocytes to this challenge. Astrocyte activation was also observed in the hippocampus and corpus callosum after vanadium exposure in rats of 2 weeks of age (Olopade and Connor, 2011; Todorich et al., 2011) and 3-week-old mice (Mustapha et al., 2014), indicating that reactive astrogliosis is a component of vanadium neurotoxicity. In the present study astrocytic response was observed at the onset of the exposure increasing till 12 months but with increasing vanadium exposure into old age, this response decreased. This could be as a result of repair process carried out by the astrocytes to enhance axonal regeneration and improve functional recovery after CNS injury (Kokaia et al., 1999; Simard and Nedergaard, 2004; Erschbamer et al., 2007). The fact that microglial activation remained pronounced into old age but less so for astrocyte may indicate that some intrinsic inflammatory pathway might have been selectively switched on in microglia after chronic vanadium exposure. Lu et al. (2014) showed that while proinflammatory cytokine AP-1 is involved in LPS-induced IL-1β expression and released by microglia and

astrocytes. Resveratrol inhibits LPS-induced AP-1 activation in microglia but not astrocytes.

We had earlier reported that chronic exposure to vanadium over a life time led to memory loss (Folarin et al., 2016) after 3 months. In the present study, we detected decrease in neuronal number and apical dendrites of the hippocampal CA1 pyramidal cells which indicates cell death and severe decline in the number and availability of axonal inputs to dendritic ends. Temporary inactivation or lesions of the dorsal hippocampus have been reported to cause impairments in the acquisition and retrieval of spatial memory (Moser and Moser, 1998; Riedel et al., 1999). Thus we propose that the profound neuronal loss and hippocampal alteration observed will possibly result to memory impairment. Previous studies (Avila-Costa et al., 2004, 2006) have reported dendritic spine loss with glaring memory alteration after vanadium inhalation. Quantitative analysis of the cell count revealed significant reduction in neuronal number both in the hippocampal regions (CA1 and CA3) and the cerebellar cortex which support our previous findings (Mustapha et al., 2014; Azeez et al., 2016; Folarin et al., 2016). This result also shows more vulnerability of CA1 region to insults than CA3; this is because high intrinsic superoxide and endogenous ROS production occur in CA1 than CA3 region (Wilde et al., 1997; Wang et al., 2005). It is also reported that mitochondrial permeability transition pore of CA1 region is more sensitive to calcium homeostasis and this leads to active production of ROS (Mattiasson et al., 2003). Similar observations on relative CA1 vulnerability were reported following exposure of the hippocampus to alcohol (Tran and Kelly, 2003; Miki et al., 2004).

Chronic vanadium exposure as shown in this study produced cytotoxicity and morphological alterations of the pyramidal cells of the prefrontal cortex characterized by cell clustering, loss of layering pattern and cytoplasmic vacuolation suggesting deterioration of cell functioning which ultimately leads to destruction of cellular structures and cell death (Henics and Wheatley, 1999). The immuno-fluorescent staining of the

FIGURE 11 | Pyramidal cell loss of the hippocampal CA3 region after chronic vanadium exposure. NeuN immuno-histochemistry showed evidence of neuronal loss after 18 months of exposure. Quantitative analysis (D) revealed significant (∗∗P < 0.001) decrease in the number of pyramidal neurones in the exposed groups (B) relative to the control (A) while the withdrawal groups (C) showed reversal effect with significant (∗P < 0.05) increase in neuronal no relative to vanadium exposed groups (B). Magnification: ×200, inset: ×400.

FIGURE 12 | Haematoxylin and Eosin (H&E) stained sections of cerebellar cortex for control mice (A) and vanadium 18 months exposed mice (B) and age matched withdrawal (C). (A) Shows numerous Purkinje cells (blue arrows) intact in the Purkinje cell layer. The white arrows in (B) shows Purkinje cells detachment and sloughed cells (black arrows) from Purkinje cell layer, the withdrawal groups (C) shows less cell detachment compares with V groups. Quantitative analysis (D) revealed significantly (∗∗P < 0.001) reduced Purkinje cell count in the vanadium exposed group relative to the control, while the withdrawal group count was significantly (∗P < 0.05) high relative to vanadium group (D), scale bar 25 µm.

cortical pyramidal cells with a nuclear DNA cell marker revealed cells with intact but shrunken nucleus and cytoplasmic loss. This result supports previous observations which suggest that vanadium cytotoxicity appears more pronounced than its apoptotic effects (Mustapha et al., 2014). We also noticed in the neurons of prefrontal cortex and CA1 region that vanadium neuropathologies of dendritic spine loss, cytotoxicity after 6–9 months resembled those of geriatric control brains at 15 and 18 months. This is indicative of neuronal degeneration occurring after vanadium exposure and consistent with previous studies (Calderón-Garcidueñas et al., 2012) which suggested that chronic exposure to environmental pollutants resulted in detection of old age associated lesions in children or young adult brain and thus connotes premature aging.

This study has shown a relative reversal of vanadium neurotoxicity both in the neuronal and glial cellular alteration and severity of activation after withdrawal from exposure to the metal. This could be as a result of reducing metal oxidative assault or repair processes which is compensating for possible early CNS damage (Wo´zniak et al., 2004). The data from the LA–ICP–MS also shows marked and progressive elimination of vanadium metal from the brain with increasing duration of withdrawal from treatment. It is noteworthy that we observed almost a complete reversal of memory loss profile in mice after 6–9 months of vanadium withdrawal (Folarin et al., 2016), however the present study shows that vanadium metal load in the brain and CNS lesion induced by vanadium neurotoxicity were not completely eliminated even though the functions may be almost fully restored back to normal. This may be as a result of surviving neurons functioning to compensate for the lost cells.

## CONCLUSION

We have shown that chronic administration of vanadium over a life time in mice resulted in metal accumulation which showed regional variabilities with time. With increase in the chronicity of the exposure, microglia activation rather than astrocytic activation becomes more predominant. The vanadium metal uptake and pathological effects were not completely eliminated from the brain even after a long time withdrawal from vanadium metal.

## AUTHOR CONTRIBUTIONS

JOO and FO conceptualized this work from inception. JOO and JRC supervised ORF on aspects of this work. ORF did the mice experiments, feeding, treatment with vanadium and animal sacrifice, and slide preparation. AMS and ORF did the immunohistochemistry and double labeling while ORF and FO were involved in the processing of brain samples, histology and related data analysis. DGP, ORF and JRC were involved in the LA–ICP–MS analysis. All authors contributed to manuscript writing, analysis and correction.

## FUNDING

This work was supported in part by International Society of Neurochemistry Travel Grant of her Committee for Aid in Education in Neuroscienceand the Thomas Bassir-Thomas Biomedical Foundation grant, both to FO.

## REFERENCES


**Conflict of Interest Statement**: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Folarin, Snyder, Peters, Olopade, Connor and Olopade. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Circadian Clock Protein Content and Daily Rhythm of Locomotor Activity Are Altered after Chronic Exposure to Lead in Rat

Mariam Sabbar <sup>1</sup> , Ouria Dkhissi-Benyahya<sup>2</sup> , Abdelhamid Benazzouz 3, 4 and Nouria Lakhdar-Ghazal <sup>1</sup> \*

<sup>1</sup> Équipe de Recherche sur les Rythmes Biologiques, Neurosciences et Environnement, Faculté des Sciences, Université Mohammed V, Rabat, Morocco, <sup>2</sup> INSERM, Stem Cell and Brain Research Institute U1208, University of Lyon, Université Claude Bernard Lyon 1, Lyon, France, <sup>3</sup> Institut des Maladies Neurodégénératives, Univ. de Bordeaux, UMR5293, Bordeaux, France, <sup>4</sup> Centre National de la Recherche Scientifique, Institut des Maladies Neurodégénératives, UMR5293, Bordeaux, France

Lead exposure has been reported to produce many clinical features, including parkinsonism. However, its consequences on the circadian rhythms are still unknown. Here we aimed to examine the circadian rhythms of locomotor activity following lead intoxication and investigate the mechanisms by which lead may induce alterations of circadian rhythms in rats. Male Wistar rats were injected with lead or sodium acetate (10 mg/kg/day, i.p.) during 4 weeks. Both groups were tested in the "open field" to quantify the exploratory activity and in the rotarod to evaluate motor coordination. Then, animals were submitted to continuous 24 h recordings of locomotor activity under 14/10 Light/dark (14/10 LD) cycle and in complete darkness (DD). At the end of experiments, the clock proteins BMAL1, PER1-2, and CRY1-2 were assayed in the suprachiasmatic nucleus (SCN) using immunohistochemistry. We showed that lead significantly reduced the number of crossing in the open field, impaired motor coordination and altered the daily locomotor activity rhythm. When the LD cycle was advanced by 6 h, both groups adjusted their daily locomotor activity to the new LD cycle with high onset variability in lead-intoxicated rats compared to controls. Lead also led to a decrease in the number of immunoreactive cells (ir-) of BMAL1, PER1, and PER2 without affecting the number of ir-CRY1 and ir-CRY2 cells in the SCN. Our data provide strong evidence that lead intoxication disturbs the rhythm of locomotor activity and alters clock proteins expression in the SCN. They contribute to the understanding of the mechanism by which lead induce circadian rhythms disturbances.

Keywords: suprachiasmatic nucleus, lead, locomotor activity, clock proteins, rat, Parkinsonism

## INTRODUCTION

Lead poisoning has been reported to induce Parkinsonism and epidemiological studies suggested that lead plays a synergistic role with other heavy metals in the incidence of Parkinson's disease (PD) (Gorell et al., 1997, 1999a,b). In fact, the role that lead may play to generate Parkinsonism has been strongly supported by many studies demonstrating that this heavy metal

#### Edited by:

Bruno Poucet, Centre National de la Recherche Scientifique (CNRS), France

#### Reviewed by:

Charles N. Allen, Oregon Health and Science University, United States Concettina Fenga, University of Messina, Italy

> \*Correspondence: Nouria Lakhdar-Ghazal nlakhdarghazal@gmail.com

Received: 03 June 2017 Accepted: 08 September 2017 Published: 22 September 2017

#### Citation:

Sabbar M, Dkhissi-Benyahya O, Benazzouz A and Lakhdar-Ghazal N (2017) Circadian Clock Protein Content and Daily Rhythm of Locomotor Activity Are Altered after Chronic Exposure to Lead in Rat. Front. Behav. Neurosci. 11:178. doi: 10.3389/fnbeh.2017.00178 is affecting the dopaminergic (DA) system as reported in PD (Ehringer and Hornykiewicz, 1960). Jason and Kellogg (1981) reported that neonatal exposure to lead in rats induced an irreversible degeneration of the nigro-striatal dopaminergic system in parallel with behavioral and neurochemical abnormalities (Jason and Kellogg, 1981). Other studies have shown that lead exposure in adolescent rat also decreased dopamine levels (Kala and Jadhav, 1995) and that this decrease may not be related to the loss of tyrosine hydroxylase immunoreactive neurons in the pars compacta of substantia nigra as reported by Tavakoli-Nezhad et al. (2001).

We have recently shown that sub-chronic exposure to lead acetate induced motor deficits in parallel with a decrease in the content of both DA in the striatum and noradrenaline (NA) in the cortex (Unpublished data).

Non-motor deficits, such as anxiety disorder have also been observed in the rat following lead exposure (Sabbar et al., 2012). Similar motor and non-motor disorders have been reported in PD patients (Shulman et al., 2001; Barone et al., 2009; Breen et al., 2014; Alzahrani and Venneri, 2015), in animal models of PD (Delaville et al., 2012; Faggiani et al., 2015) and in animals exposed to manganese (Bouabid et al., 2014). Interestingly, nonmotor symptoms have been shown to precede the manifestation of motor disabilities in PD patients (Chaudhuri et al., 2006; Ishihara and Brayne, 2006). Furthermore, one of the common non-motor disorders reported in PD patients is sleep disorder, including insomnia, excessive daytime sleepiness and disturbed rapid eye movement (REM) sleep behavior (Gunn et al., 2010; Menza et al., 2010; Willison et al., 2013).

The sleep-wake cycle is one of the many daily rhythms regulated by the circadian timing system (Pace-Schott and Hobson, 2002). Circadian rhythms are controlled by the master clock localized in the suprachiasmatic nuclei (SCN) of the anterior hypothalamus in mammals (Stephan and Zucker, 1972). The circadian clock generates molecular circadian rhythms through cell-autonomous autoregulatory transcriptional/translational feedback loops consisting of the bHLH/PAS transcription factors BMAL1 and CLOCK, which heterodimerize and drive transcription of many genes, including their own negative feedback repressors, such as Period (Per1, Per2, Per3), Cryptochrome (Cry1, Cry2) and Reverb genes, which repress BMAL1/CLOCK-mediated transcription (Reppert and Weaver, 2001; Lowrey and Takahashi, 2004; Okamura, 2004).

The consequences of lead exposure on the circadian system have been investigated by few studies who showed that lead exposure affected the locomotor activity rhythm in rats (Collins et al., 1984). Using the rat model of lead-induced Parkinsonism, the present study aimed to investigate whether lead exposure affects both the circadian rhythms of locomotor activity and the molecular machinery of the SCN. The effects of a subchronic low-level lead treatment on motor behavior and coordination was first evaluated; then, we monitored the daily and circadian locomotor activity rhythm and quantified clock proteins expression; BMAL1, PER1, PER2, CRY1, and CRY2 in the SCN.

## MATERIALS AND METHODS

## Animal Housing

Male Wistar rats (Central animal service, Mohammed V University, Faculty of Sciences, Rabat, Morocco) weighing 70–80 g were used for behavioral and immunohistochemical studies. Rats were kept individually in polycarbonate cages, in a thermostatically controlled room (temperature: 24◦C, relative humidity: 45%) under a 14 h/10 h light/dark cycle (14/10 LD; lights on at 06:00 h) for 3 weeks, with access to food and water ad libitum. Body weights were monitored throughout the experiment. All experiments were performed in accordance with the European Communities Council Directive 2010/63/UE. Approval was granted by the Ethic Committee of Veterinarians of Hassan II Institute of Agronomy and Veterinary Medicine of Rabat, Morocco, and all efforts were made to minimize the number of animals used and their suffering.

## Lead Administration

Animals were randomly divided into two groups: leadintoxicated (n = 26) and control (n = 20) rats. In our experiments, lead-intoxicated rats received daily i.p. injections of lead acetate free-pyrogen solution (10 mg/kg; Sigma, France) at zeitgeber time 4 (ZT4, 4 h after light on), for 30 days. Control rats received sodium acetate (10 mg/kg; Sigma, France) solution in the same conditions.

## Behavioral Assessments

### Assessment of Spontaneous Locomotor Activity

Exploratory behavior was evaluated in the open field test as previously described by Rodrigues et al. (1996) with modification. The apparatus is a wooden box (75 cm long, 45 cm wide and 35 cm high), the surface was divided into 15 similar squares of 15 cm each side. Prior to lead or sodium acetate administration, all rats were habituated to the open field for 3 days. At the first and the last day of the i.p. injections of lead or sodium acetate, rats were tested once in a quiet and dimly lit for 5 min session. Exploratory behavior was evaluated by counting the number of crossing (transition from one square to another) and rearing (animal stands on its two legs).

## Assessment of Motor Coordination

Motor coordination was evaluated using the rotarod test. The apparatus is equipped with a rotating bar which rotates at different and adjustable speeds. Rats were placed on the rotating bar with a fixed speed of 20 rotations per minute (rpm). The performance on the rotarod was measured once a day at ZT4 (10:00 h) for 4 consecutive days, and the time was recorded until the rat failed to stay on the bar.

#### Assessment of Locomotor Activity Rhythm

Locomotor activity was continuously monitored using infrared motion captors placed over the cages and a computerized data acquisition system CAMS (Circadian Activity Monitoring System, INSERM, France) as previously described (El Moussaouiti et al., 2010), and analyzed using Clocklab software (Actimetrics, Evanston, Illinois, USA). Different parameters were analyzed: the period and the amplitude using Chi-squared periodogram, daytime and nighttime activities, onset and offset variabilities, activity onset and offset which correspond to the average clock (or circadian) time of activity onset or activity offset respectively. The diurnality index [mean activity during the light phase/(mean activity during the light phase + mean activity during the dark phase)] as previously described (Refinetti, 2006). Animals with indices above 0.5 are more active during the day than during the night. Alpha (α) was also analyzed and is defined as the duration of the active period of the animal. To assess the entrainment of the rhythm of locomotor activity by light, animals were subjected to 6 h phase advance of the 14/10 LD cycle (lights on at 00:00 h; LD AT) for 3 weeks after the last injection of either lead or sodium acetate. The phase angle and the number of days necessary to entrain to the new 14/10 LD cycle were determined for each rat. The phase angle was defined as the difference between the onset activity and the time of lights-off. In our study, the rhythm of locomotor activity is considered entrained to the new LD cycle when the onset of activity presents a stable phase relationship relative to the time of light off (±0.3 h) for at least 10 days as previously described (Lahouaoui et al., 2014).

Recording of the circadian locomotor activity was pursued in constant darkness (DD) for 15 days, and the free-running periods (Tau, τ ) were calculated using Clocklab (Actimetrics, Evanston, Illinois, USA). Then, rats were exposed to 14/10 LD cycle (lights on at 06:00 h) for 10 days.

#### Experimental Procedures

**Experiment 1:** Prior to lead or sodium acetate administration, all rats were maintained under 14/10 h LD cycle, with lights on at 06:00 h and were habituated to the open field for 3 days. On day1 (**Figure 1A**), rats were assigned randomly to receive an i.p. injections of one of the following substances: lead acetate (10 mg/kg; n = 6), and sodium acetate (10 mg/kg; n = 6) at ZT 4 (10:00 h) for 30 days. On day 30, after the last injection of lead or sodium acetate, a rat was placed into the open field box and allowed to explore the box for about 5 min. Exploratory behavior was evaluated by counting the number of crossing (transition from one square to another) and rearing (animal stands on its two legs).

Twenty-four hours after the open field test, the performance on the rotarod was evaluated in the same rats. Rats were placed on the rotating bar with a fixed speed of 20 rpm. The performance on the rotarod was measured once a day at ZT4 for 4 consecutive days, and the time was recorded until the rat failed to stay on the bar (**Figure 1A**).

**Experiment 2**: Rats were maintained under 14/10 h LD cycle, with lights on at 06:00 and daily locomotor activity rhythms of rats were recorded for 22 days, prior lead or sodium acetate treatment (LD; **Figure 1B**). On day 1, rats were assigned randomly to receive an i.p. injection of one of the following substances: lead acetate (10 mg/kg; n = 6), sodium acetate (10 mg/kg; n = 6) at ZT 4 (10:00 am) for 30 days and were maintained under 14/10 h LD cycle (LD+T; **Figure 1B**). On day 30, after the last injection of lead or sodium acetate, rats were subjected to 6 h phase advance of the 14/10 LD cycle (lights on at 00:00 h; LD AT) to examine the ability of animals to re-entrain to a new LD cycle for 22 days (**Figure 1B**). Rats underwent under total darkness (DD; **Figure 1**) to assess the circadian locomotor activity rhythm for 14 days followed by 10 days in 14/10 LD cycle (light on at 06:00 h).

## Immunohistochemistry

At the end of the locomotor activity recording period, rats were deeply anesthetized by an injection of sodium pentobarbital (100 mg/kg, i.p.) and perfused with 0.9% saline followed by 4% paraformaldehyde (PFA, 300 ml) in 0.1 M phosphate buffer (PB, pH 7.4). Brains were removed and post-fixed in 4% PFA at 4◦C for an additional 24 h, rinsed in PBS, and cryo-protected in 30% sucrose in 0.1 M PBS (phosphate-buffered saline, pH 7.4) for additional 24 h at 4◦C. Collected brains from control (n = 12) and lead-intoxicated (n = 12) rats were cut into 20 µm coronal sections and two series of three alternated sections were collected. We used one of the series for the primary antibodies; anti-PER1, anti-PER2 and anti-BMAL1, and the second serie for the primary antibodies; anti-CRY1 and anti-CRY2.

SCN sections were pre-incubated in 0.1 M PBS, 5% normal goat serum with 0.4% Triton X-100 and 1% bovine albumin serum for 1 h and then transferred to 2% normal goat serum containing primary antibody for 48 h at 4◦C (1:5,000). After washes in 0.1 M PBS (3 × 10 min), sections were incubated during 2 h in 0.1 M PBS containing 2% normal goat serum and biotinylated rabbit anti-goat IgG (1:2,000, Vector laboratories) for 2 h at room temperature. Sections were then washed in PBS (3 × 10 min), and incubated with the avidin/biotin complex (1:1,000; ABC, Vectastain kit, Vector Laboratories) for 1 h at room temperature. After several washings (two in 0.1M PBS and one in 0.05 M Tris buffer (TB, pH 7.6), immunoreactivity was visualized with 0.025% DAB (Sigma, France), 0.5% ammonium nickel sulfate (Sigma, France), in 0.1 M TB containing 0.03% hydrogen peroxide (Sigma, France), for 6–10 min. Sections were finally washed three times in 0.1 M TB and once in 0.1 M PB. After processing, tissue sections were mounted onto gelatincoated slides, dehydrated in graded ethanol, cleared in xylene and coverslipped with Eukitt (Sigma, France).

## Data Analysis

All Statistical analyses (behavioral or immunohistochemical analyses) were done using GraphPad Prism program version 6.05 (California, USA). Data are shown as mean ± SEM and statistical significance was considered for P < 0.05.

Two-way ANOVA followed by post-hoc Bonferroni were used to compare body weight, number of crossing and rearing, response latency on the rotarod (failure time) between the two groups. Period, tau, activity onset and offset variabilities, α, phase angle, and diurnality index were compared using Mann-Whitney test between controls and lead-intoxicated rats. Mean activity counts during the light and the dark phases and the total activity were also analyzed using Mann-Whitney test.

Labeled cell nuclei by PER 1-2, CRY1-2, and BMAL1 were counted manually in 4–5 sections per animal for each primary antibody) of the SCN using a computerized image system (Image J software, imagej.nih.gov/ij/) attached to a light microscope

(Leica Microsystems, Germany). The number of immunoreactive cells (ir-cells) in the SCN was counted in each section and averaged among these coronal sections. Statistical analysis was determined using the Mann-Whitney test.

## RESULTS

## Lead Intoxication Decreased Body Weight Gain

**Figure 2** shows the mean body weight (±SEM) of leadintoxicated rats and their respective controls. Two-way repeated measures ANOVA showed significant effects of lead on time [F(14, 168) = 294.6, P < 0.0001], treatment [F(1, 12) = 5.672, P < 0.05] and interaction (treatment × time) [F(14, 168) = 13.57, P < 0.0001]. At the beginning of the experiment, all animals had similar body weights (lead-intoxicated rats: 69.75 ± 3.22 vs. control rats: 71.33 ± 2.39). From day 19 to day 29 of injections, post-hoc test revealed that body weight of lead-intoxicated rats became significantly reduced compared to controls and resulted in 23% loss of body weight gain on day 29 (Bonferroni post-hoc test, P < 0.001; **Figure 2**).

## Lead Intoxication Induced Hypoactivity Measured in the Open Field Test

In the open field test, we expected that exploratory activity will be higher independently to the treatment in day 30, because the age of the rat influences the exploration rate; infant or adolescent rats had less exploratory activity than adults (Smith and Morrell,

2007). We focused, then in our data analysis in the effect of lead intoxication on the exploratory activity on day 30 compared to their controls on the same test day.

**Figure 3** shows the spontaneous locomotor activity (crossing; **Figures 3A,B**) as measured in the open field test. There were no statistically significant differences among the groups in the number of crossing and rearing movements (**Figures 3A,B**) on

sessions on the first day and the last day of lead acetate treatment. Motor coordination performance histogram represent the time that rat stayed on the rotating bar using the rotarod test at the end of lead treatment (C). Values are the mean ± SEM. Data from controls (n = 6) and lead-intoxicated rats (n = 6) were compared using two-way ANOVA followed by a post-hoc Bonferroni. \*P < 0.05, \*\*P < 0.01 in comparison with controls.

the habituation period before injections, as well as on the first day of injections.

Upon a re-introduction into the open field on day 30, the pattern of the exploratory activity in rats was notably different; the number of crossings significantly decreased in lead-intoxicated rats compared to controls [−45.59%, F(4, 40) = 3.20, P < 0.05, two-way ANOVA followed by Bonferroni posthoc test; **Figure 3A**]. However, the number of rearings did not significantly change [F(4, 40) = 0.040, P > 0.05, two-way ANOVA, **Figure 3B**].

## Lead Intoxication Impaired Motor Coordination Evaluated in the Rotarod Test

The motor coordination was assessed using the rotarod test. Latency time in seconds was recorded. Two-way repeated measures ANOVA of motor coordination after lead acetate or sodium acetate injections showed a significant effects on time [F(3, 30) = 9.400, P = 0.0002], and treatment [F(1, 10) = 6.932, P = 0.0250] but not on treatment × time interaction [F(3, 30) = 2.802, P = 0.0568]. Lead-intoxicated rats did not significantly increase their latency time on the rotating bar, in contrast to controls, in which the latency time on the rotating bar increased significantly (Bonferroni post-hoc test, P < 0.01, **Figure 3C**).

## Lead Intoxication Disturbed the Daily and the Circadian Locomotor Activity Rhythm

As mentioned earlier (see section Material and Methods), all rats were randomly divided into two groups; controls and leadintoxicated rats. The daily locomotor activity rhythm of all rats was monitored when animals were exposed to 14/10 LD cycle prior to any treatment. Each animal served as its own control. Representative actograms before treatment (LD), and during lead or sodium acetate injections (LD+T) are shown in **Figure 4**.

In LD, during the pre-treatment, all rats showed a strong daily profile of locomotor activity with the main activity was recorded during the dark phase. The average daytime activity counts was 447.1 ± 29.34 and the average nighttime activity counts was 1,390 ± 94.28. After this pre-treatment rats, we divided randomly all rats to two groups lead-intoxicated and control group and the data were re-analyzed and no significant differences were observed in the mean activity.

The activity profile for representative control and leadintoxicated rats are shown in **Figure 5A**. During lead intoxication (LD+T), daytime activity counts were significantly increased in lead-intoxicated rats when compared to their controls (666.8 ± 43.14 vs. 430.6 ± 61.68 in controls P < 0.01, Mann Whitney test, **Figure 5B**). However, no significant differences were found between lead-intoxicated rats and controls in the nighttime activity counts (1,236 ± 83.03 vs. 1,506 ± 376.5 in controls, **Figure 5C**) or in the total activity counts (1,903 ± 100.5 vs. 1,937 ± 399.2 in control rats, **Figure 5D**). The diurnality index was significantly higher in lead-intoxicated rats (0.35 ± 0.02) compared to controls (0.25 ± 0.03) (P < 0.01, Mann Whitney test; **Figure 5E**).

Periodogram analysis showed that both groups remained entrained to the 14/10 LD cycle during lead intoxication (LD+T), with similar period (23.96 ± 0.009 h in controls vs. 23.98 ± 0.006 h in lead-intoxicated rats; **Figure 5F**) but with a delay in their activity onset. The nighttime activity began in leadintoxicated rats before light OFF (onset variability was 0.94 ± 0.22 h in controls vs. 1.67 ± 0.34 h in lead-intoxicated rats; P < 0.05, Mann Whitney test; **Figure 5G**). No difference in the offset of activities was observed (offset variability was 0.49 ± 0.13 h in controls vs. 0.88 ± 0.16 h in lead-intoxicated rats; P = 0.054, Mann Whitney test, **Figure 5H**). α significantly decreased in lead-intoxicated rats with a mean value of 13.01 ± 0.16 h

by comparison to controls (13.71 ± 0.08 h; P = 0.0034, Mann Whitney test, **Figure 5I**).

light phase. The arrow indicates the beginning of lead or sodium acetate treatment.

One day after the last injection of lead or sodium acetate, rats have been exposed to a 6 h phase advance of the 14/10 LD cycle to examine their ability to entrain to a new LD cycle.

Representative actograms from control and lead-intoxicated rats are shown in **Figure 6A**. Control and lead-intoxicated rats were able to synchronize their locomotor activity rhythm to the new LD cycle. In controls, entrainment was achieved after 8.67 ± 1.17 days, whereas in lead-intoxicated rats, the entrainment was attained after 8.85 ± 1.81 days. No significant difference between the two groups was observed (Mann Whitney test, P = 0.80, **Figure 6B**).

Three weeks after cessation of lead and sodium acetate injections (LD AT; light on at 00:00), rats showed a strong daily profile of the locomotor activity with a similar period (23.96 ± 0.009 in controls vs. 23.98 ± 0.001 in lead-intoxicated rats; **Figure 7A**) but a number of parameters were altered. The variability of activity offset was significantly higher in leadintoxicated rats (0.66 ± 0.103 h vs. 0.41 ± 0.047 h in control rats; P < 0.05, Mann Whitney test; **Figure 7B**), whereas, no difference was observed between the two groups for the variability of activity onset (0.89 ± 0.15 h in lead-intoxicated rats vs. 0.72 ± 0.09 h in control rats; P > 0.05, Mann Whitney test, **Figure 7B**).

A significant decrease in the mean nighttime activity was observed in lead-intoxicated rat (1,425 ± 110.3 vs. 2,282 ± 294.8 in control rats; **Figure 7C**, P < 0.05). No difference was observed in daytime activity (501.1 ± 99.11 in controls vs. 503.1 ± 64.62 in lead-intoxicated rats; **Figure 7C**). The total activity counts were 2,783 ± 380.7 and 1,928 ± 164.4 in control and lead-intoxicated rats, respectively (**Figure 7C**) and the diurnality index was significantly increased in lead-intoxicated rats (0.26 ± 0.02) compared with control rats (0.17 ± 0.02) (**Figure 7D**, Mann Whitney test, P < 0.01).

**Figure 7E** shows the daily change in activity during the light and dark phase under the new LD cycle of rats 3 weeks after lead treatment and the analysis of the daily activity levels (mean activity counts) showed that lead profoundly induced hypoactivity during the dark phase (Two-way repeated measures ANOVA showed a significant effect of treatment [P < 0.0001, F(3, 489) = 509.6], whereas neither the time nor the interaction (time x treatment) were significant (P > 0.05).

In order to examine the impact of lead on the circadian clock, we subjected rats to constant conditions of darkness (DD) for 15 days. Under these conditions, lead-intoxicated rats exhibited a locomotor circadian rhythm with a similar free-running period (τ ) when compared to controls (23.95 ± 0.038 h in controls vs. 23.93 ± 0.087 h in lead-intoxicated rats; P = 0.72, Mann Whitney test; **Figure 8A**). The activity onsets variability was significantly higher in lead-intoxicated rats (1.15 ± 0.143 h vs. 0.69 ± 0.09 h in controls, Mann Whitney test, P < 0.05; **Figure 8B**) whereas the activity offsets variability was not significantly altered by lead intoxication (0.91 ± 0.148 h vs. 0.45 ± 0.09 h in controls, Mann Whitney test, P = 0.084, **Figure 8B**). Likewise, no differences

were found in the activity during the subjective day, in the activity during the subjective night or the total activity (**Figure 8C**).

## Effect of Lead Intoxication on the Clock Proteins Immunoreactivity in the SCN

**Figures 9**–**11** show the mean number (±SEM) of clock protein immunoreactivity (ir-) in the SCN of lead-intoxicated rats and their respective controls. Lead induced a significant decrease in the mean number of -BMAL1, -PER1, and -PER2 ir-cells in the SCN. In control rats, the average of ir-BMAL1 cells is 374 ± 25.9, whereas, in lead-intoxicated rats, the average of ir-BMAL1 cells is 303.2 ± 19.7 (Mann-Whitney test, P < 0.05; **Figure 9**).

The mean number of ir-PER1 cells is 416 ± 22.72 in controls, whereas, in lead-intoxicated rats, the mean number of ir-PER1 cells is 302.1 ± 35.93 (Mann-Whitney test, P < 0.05; **Figure 10**), and the mean number of ir-PER2 cells is 409.7 ± 19.40 in controls, whereas, in lead-intoxicated rats, the mean number of ir-PER2 cells is 277.7 ± 25.91 (Mann-Whitney test, P < 0.05; **Figure 10**).

In contrast, lead did not affect CRY1 and CRY2 immunoreactivity in the SCN (**Figure 11**). The number of ir-CRY1 cells is 227 ± 45.3 and 311.1 ± 24.5 in lead-intoxicated and control rats. For CRY2, the number of ir-CRY2 cells is 296 ± 22.7 in controls and 227 ± 28.5 in lead-intoxicated rats.

## DISCUSSION

The main interesting result obtained in this study is the alteration in the locomotor activity rhythm induced by lead toxicity which parallels with a decrease in clock protein expression. Interestingly, these alterations are obtained in all groups of rats that express neurobehavioral dysfunction.

In the current study, we have confirmed in Wistar rat (experiment 1) that lead induces progressive decline in body weight gain, decreased the exploratory activity (number of crossing in open field; **Figure 3A**) and impaired motor coordination (**Figure 3C**) as early shown by our previous findings in Sprague dawley rat (Sabbar et al., 2012) and other works (Reiter et al., 1975; Overmann, 1977; Gill et al., 2003; NourEddine et al., 2005; Sansar et al., 2011). Strangely, it seems that in lead-intoxicated rats, the number of rearing in the open filed was not significantly different compared to controls (**Figure 3B**). However, based on previous data (Shafiqur-Rehman et al., 1991), it is more likely that rearing response has decreased the first days following lead intoxication and then increased on day 30. In addition to motor symptoms (Reiter et al., 1975; NourEddine et al., 2005), non-motor symptoms like the rest/activity rhythm is also affected by lead (Collins et al., 1984). In the present study, we have shown for the first time that lead intoxication impaired the locomotor activity rhythm together with a decrease in BMAL1, PER1, and PER2 content in the SCN without inducing any changes in CRY1 and CRY2 content. These results provide strong evidence that lead disturbs circadian function by probably affecting clock protein expression in the SCN.

In the present work, we show that in lead-intoxicated rats, the 24 h rest/activity cycle is fragmented under a 14/10 LD cycle with activity predominantly expressed during the dark phase (**Figure 4**). Animals showed impaired locomotor activity rhythm consisting of abnormal phasing to the LD cycle, and strong change in the diurnality index (**Figure 5E**) than those observed in controls. Moreover, lead-intoxicated rats showed also less precision in their daily locomotor activity rhythm as reflected in the increased activity onsets variability (**Figure 5F**). No difference was however observed in the period of the 24 h locomotor activity rhythm. When the LD cycle was advanced

(n = 7) were compared using one-way ANOVA followed by Tukey's post-hoc test. \*P < 0.05, \*\*P < 0.01, \*\*\*P < 0.001 in comparison with controls.

by 6 h, both groups adjusted their daily locomotor activity to the new LD cycle; however, the offset variability was higher in lead-intoxicated rats than controls (**Figure 7B**). Our results enhance earlier observations reported by Collins et al. (1984) and Shafiq ur Rehman et al. (1986). Collins et al. (1984) showed that in pups chronically exposed to lead for many weeks, the circadian spontaneous locomotor activity was significantly affected, whereas Shafiq ur Rehman et al. (1986) reported that lead intoxication affects circadian rhythm of ambulatory activity. In addition, another study has also demonstrated that lead intoxication affects the circadian patterns of the complex stereotyped behaviors (such as rearing, preening, scratching and biting/licking) (Shafiq ur Rehman, 1999). These alterations disturb the ability of the animal to cope and interact with its environment.

Although no definitive causal links between lead intoxication and PD have been proven, there are a considerable number of epidemiological studies suggested that occupational exposure to specific metals including lead may be a high-risk factor for PD (Gorell et al., 1997; Coon et al., 2006; Weisskopf et al., 2010), we then compared 24 h locomotor activity alterations in lead-intoxicated rats to those observed in PD patients and animal models of PD and reported in the literature. Interestingly, actograms analysis revealed a significant increase in the mean daytime activity during lead intoxication followed by a decrease in the nighttime activity in lead-intoxicated rats, reflecting a poor consolidation of locomotor activity as previously reported (Collins et al., 1984). These findings reproduced those previously reported in PD patients (van Hilten et al., 1994) and in animal models of PD (Ben and Bruguerolle, 2000; Almirall et al., 2001; Boulamery et al., 2010). They strengthen the hypothesis that lead-intoxicated rats lack the ability to maintain robust locomotor activity rhythm under LD cycle, which is similar to the impairment observed in the circadian rhythms of PD patients. In fact, many studies have demonstrated that lower amplitude rest/activity rhythm affected sleep quality (Langmesser et al., 2009; Smith et al., 2009). Indeed, PD patients expressed a number of circadian rhythm alterations such as insomnia and excessive daytime sleepiness (van Hilten et al., 1994; Stevens et al., 2004; Thorpy and Adler, 2005; Ferreira et al., 2006) and nighttime sleep fragmentation (van Hilten et al., 1993; Gunn et al., 2010) strengthening the hypothesis that the alteration in circadian rhythm parameters in lead-intoxicated rats could explain partially sleep/wake cycle disturbances that occur in PD patients (Arnulf et al., 2002). Several parameters in the sleepwake pattern of animal models of PD have also been described (Fifel et al., 2016). For example, MPTP injection induced a dramatic disruption of sleep/wake architecture associated with reduced (REM) sleep and excessive daytime sleepiness in non-human primate (Barraud et al., 2009). Another study reported a drastic decrease in the latency to the onset of slow wave sleep (SWS) with REM sleep ablation in MPTPtreated rats (Lima et al., 2007). Despite lack of investigations on lead-induced sleep disturbances in human or animals, we found only one report (Kumar and Desiraju, 1992) where the authors showed that lead induced a significant reduction in the delta, theta, alpha and beta band electroencephalogram

spectral power in both wakeful and SWS stages. Moreover, our results are in accordance with previous study in PD animal model reporting that ASO (α-synuclein over-expressing) mice did not show neither entrainment deficits to 6 h changes in the LD cycle or, alteration in the locomotor activity rhythm in DD (Kudo et al., 2011). Even that lead-intoxicated rats were able to adjust their locomotor activity rhythm to new LD cycle, the precision of the daily offset of activity was altered, sign of greater fragmentation in activity. All those changes described above and those reported in the studies mentioned previously postulate that lead intoxication may affect the structures and/or functions involved in the circadian timing system. Indeed, anatomical investigations suggested that in addition to other brain region (i.e., hippocampus, cerebellum, retina. . . ), hypothalamus is a target to the neurotoxic action of lead (Wang et al., 2006; Rojas-Castaneda et al., 2011).

In mammals, SCN of the hypothalamus, the main clock pacemaker is involved in the generation and entrainment of circadian rhythms (Meijer and Rietveld, 1989; Lowrey and

3V, Third ventricle.

Takahashi, 2004). The circadian oscillations are generated by molecular mechanism based on feedback loops which is responsible of rhythmic transcription and translation of clock genes (Reppert and Weaver, 2001; Hastings and Herzog, 2004; Lowrey and Takahashi, 2004; Okamura, 2004). Bmal1 and Clock play a key role in feedback loop by acting as a positive regulator; CLOCK-BMAL1 heterodimer is able to induce a rhythmic transcription of other clock genes (Gekakis et al., 1998).

In the present study, we examined the clock protein expression, BMAL1, CRY1, CRY2, PER1, and PER2 in the SCN, and we found that BMAL1, PER1, and PER2 immunoreactivity were significantly declined in lead-intoxicated rats without any changes in CRY1 and CRY2 immunoreactivity in the SCN compared to controls (**Figure 11**). In this regard, it has been reported that in leukocytes, the expression of BMAL1 was lower in PD patients (Cai et al., 2010). Thus, the decrease in BMAL1 content in the SCN of lead-intoxicated rats could impair the molecular clock by disturbing the transcription factors of other clock genes or clock-controlled output genes. Bunger et al.

(2000) showed that under LD, locomotor activity rhythm is impaired and activity levels are reduced in Bmal1−/<sup>−</sup> mice which may explain the alteration of the circadian rhythm of locomotor activity in lead-intoxicated rats. Bunger et al. (2000) demonstrated also that in Bmal1−/<sup>−</sup> mice the expression of Per1 and Per2 were very low and not rhythmic. Zheng et al. (1999) provided evidence that Per1 gene is essential for the functioning of the circadian clock and that Per2 may regulate Per1; Per2 mutation leads to a change in the expression of other genes (Per1). This mutation displays a shorter circadian period followed by a loss of circadian rhythmicity in constant darkness (Zheng et al., 1999). In DD, lead-intoxicated rats continued to express a rhythmic locomotor activity with a period of approximately 24 h, suggesting that the decrease in PER1 and PER2 contents in the SCN may not be enough to be translated into rhythmicity loss in lead-intoxicated rats. The cryptochrome proteins, CRY1 and CRY2, act as negative regulators in the transcriptional feedback loop (van der Horst et al., 1999; Vitaterna et al., 1999), and inhibit the expression of their own genes and of period genes (Kume et al., 1999). In lead-intoxicated rats, there was no difference in

Cry1 and Cry2 content in the SCN and this strength the finding that lead-intoxicated rats were rhythmic in total darkness.

In other hand, there is relevant evidence of the implication of several neurotransmitters systems in the sleep/wake regulation/modulation. In parkinsonism, sleep/wake disturbance may result in several neurotransmission failures additionally to DAergic system, i.e., NAergic neurons in the locus coeruleus (Jellinger, 1991; Zarow et al., 2003; Fulceri et al., 2007), and serotonergic neurons in the raphe (Kish, 2003; Kish et al., 2008). Lead is also known to affect neurotransmitter systems, including NA and DA systems (Silbergeld, 1983; Sabbar et al., 2012). We did not measure concentrations of NA or DA following lead intoxication in this study, but we recently find a decrease in striatal DA concentration (unpublished data) and changes in cortical NA concentration in lead-intoxicated rats (same dose and same route of administration; 8). Since NA containing fibers and terminals were demonstrated in the SCN (Cagampang et al., 1994; Jacomy and Bosler, 1995; Vacher et al., 2003), the role of either NA or DA cannot be excluded to explain the circadian rhythm alterations during lead intoxication.

NA might modulate SCN circadian rhythms by regulating the expression of arginine-vasopressin and vasoactive intestinal peptides, two neuropeptides involved in the control of circadian rhythms as previously reported by Vacher et al. (2003). This neurotransmitter might also affect clock genes expression as reported in the astroglial cells of the SCN (Morioka et al., 2010). Furthermore, DA has been shown to modulate the expression of the clock genes (Imbesi et al., 2009) and regulate the BMAL1/CLOCK heterodimer activity (Yujnovsky et al., 2006). A decrease of DA level in the striatum in lead-intoxicated rats (unpublished data) may suggest that this heavy metal may impair the modulatory role that DA exerts and could explain partially the changes in clock protein expression our lead-intoxicated rats. This hypothesis has however to be confirmed by additional experiences.

## REFERENCES


In conclusion, we have confirmed that lead intoxication induced motor disabilities similar to those reported in animal models of PD. Moreover, we have shown that several 24 h locomotor activity parameters were altered, associated with a decrease in bmal1, per1, and per2 contents in the SCN. Tough, 24 h rest/activity disturbances, have never been extensively explored following lead intoxication, it may be interesting to investigate the mechanism(s) by which lead disrupt circadian rhythmicity, thereby providing evidence that might link lead neurotoxicity to induce Parkinsonism.

## AUTHOR CONTRIBUTIONS

NL designed the experimental protocol. MS collected, analyzed the data, wrote and edited the manuscript. OD assisted with data analysis. NL and AB edited and approved the final draft of the manuscript. All authors read and approved the final manuscript.

## FUNDING

This research was funded by the "Université Mohammed V de Rabat," Rhône-Alpes CMIRA research Grant and CNRS-CNRST Convention Adivmar 22614.

## ACKNOWLEDGMENTS

We are very grateful to Dr. Cooper Howard for sharing his expertise and helping us to set up the locomotor activity acquisition system CAMS (Circadian Activity Monitoring System). We thank Dr. Elisabeth Maywood for her generous gift of clock protein antibodies. Dr. N. Bouhaddou and our Ph.D. students Ms. M. S. Klouche and Ms. D. Salhi are thanked for their assistance during perfusion, immunohistochemistry, and animal care assistance.


in the mouse suprachiasmatic nucleus. J. Neurosci. Res. 71, 791–801. doi: 10.1002/jnr.10529


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Sabbar, Dkhissi-Benyahya, Benazzouz and Lakhdar-Ghazal. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Lead-Induced Atypical Parkinsonism in Rats: Behavioral, Electrophysiological, and Neurochemical Evidence for a Role of Noradrenaline Depletion

Mariam Sabbar 1,2,3 \*, Claire Delaville1,2, Philippe De Deurwaerdère1,2 , Nouria Lakhdar-Ghazal <sup>3</sup> and Abdelhamid Benazzouz 1,2 \*

1 Institut des Maladies Neurodégénératives, UMR 5293, Université de Bordeau, Bordeaux, France, <sup>2</sup> Centre National de la Recherche Scientifique, Institut des Maladies Neurodégénératives, UMR 5293, Bordeaux, France, <sup>3</sup> Faculté des Sciences, Equipe Rythmes Biologiques et Environnement, Université Mohammed V, Rabat, Morocco

#### Edited by:

Marco Antonio Maximo Prado, University of Western Ontario, Canada

#### Reviewed by:

Giuseppe Gangarossa, Paris Diderot University, France Patrícia Maciel, Escola de Medicina da Universidade do Minho, Portugal

#### \*Correspondence:

Mariam Sabbar sab\_mariam@yahoo.fr; mariamsabb2@gmail.com Abdelhamid Benazzouz abdelhamid.benazzouz@u-bordeaux.fr

#### Specialty section:

This article was submitted to Neurodegeneration, a section of the journal Frontiers in Neuroscience

Received: 30 September 2017 Accepted: 05 March 2018 Published: 19 March 2018

#### Citation:

Sabbar M, Delaville C, De Deurwaerdère P, Lakhdar-Ghazal N and Benazzouz A (2018) Lead-Induced Atypical Parkinsonism in Rats: Behavioral, Electrophysiological, and Neurochemical Evidence for a Role of Noradrenaline Depletion. Front. Neurosci. 12:173. doi: 10.3389/fnins.2018.00173 Background: Lead neurotoxicity is a major health problem known as a risk factor for neurodegenerative diseases, including the manifestation of parkinsonism-like disorder. While lead is known to preferentially accumulate in basal ganglia, the mechanisms underlying behavioral disorders remain unknown. Here, we investigated the neurophysiological and biochemical correlates of motor deficits induced by sub-chronic injections of lead.

Methods: Sprague Dawely rats were exposed to sub-chronic injections of lead (10 mg/kg, i.p.) or to a single i.p. injection of 50 mg/kg N-(2-chloroethyl)-N-ethyl-2-bromobenzylamine hydrochloride (DSP-4), a drug known to induce selective depletion of noradrenaline. Rats were submitted to a battery of behavioral tests, including the open field for locomotor activity and rotarod for motor coordination. Electrophysiological recordings were carried out in three major basal ganglia nuclei, the subthalamic nucleus (STN), globus pallidus (GP), and substantia nigra pars reticulata (SNr). At the end of experiments, post-mortem tissue level of the three monoamines (dopamine, noradrenaline, and serotonin) and their metabolites has been determined using HPLC.

Results: Lead intoxication significantly impaired exploratory and locomotor activity as well as motor coordination. It resulted in a significant reduction in the level of noradrenaline in the cortex and dopamine and its metabolites, DOPAC, and HVA, in the striatum. The tissue level of serotonin and its metabolite 5-HIAA was not affected in the two structures. Similarly, DSP-4, which induced a selective depletion of noradrenaline, significantly decreased exploratory, and locomotor activity as well as motor coordination. L-DOPA treatment did not improve motor deficits induced by lead and DSP-4 in the two animal groups. Electrophysiological recordings showed that both lead and DSP-4 did not change the firing rate but resulted in a switch from the regular normal firing to irregular and bursty discharge patterns of STN neurons. Neither lead nor DSP-4 treatments changed the firing rate and the pattern of GP and SNr neurons. Conclusions: Our findings provide evidence that lead represents a risk factor for inducing parkinsonism-like deficits. As the motor deficits induced by lead were not improved by L-DOPA, we suggest that the deficits may be due to the depletion of noradrenaline and the parallel disorganization of STN neuronal activity.

Keywords: lead, Parkinsonism, subthalamic nucleus, globus pallidus, pars reticulata of substantia nigra, noradrenaline, dopamine, electrophysiology

## INTRODUCTION

Parkinson's disease (PD) is a neurodegenerative disorder characterized by the manifestation of motor symptoms, which are mainly attributed to the loss of dopaminergic neurons in the substantia nigra pars compacta (SNc) (Ehringer and Hornykiewicz, 1960). Furthermore, PD is not a pure dopaminergic pathology, it is characterized by the additional degeneration of noradrenaline (NA) and serotonin (5-HT) neurons of the locus coeruleus (LC) and the dorsal raphe respectively (Bertrand et al., 1997; Kish, 2003; Fornai et al., 2007).

It is well established that the degeneration of the monoaminergic systems caused a disorganization of the neuronal activity in the subthalamic nucleus (STN), a basal ganglia structure playing a key role in the pathophysiology of PD. Indeed, the regular pattern of the neuronal activity of the STN becomes irregular and bursty in animal models of PD (Bergman et al., 1994; Ni et al., 2001; Belujon et al., 2007; Delaville et al., 2012a) and in PD patients (Hutchison et al., 1998; Benazzouz et al., 2002). It has also been reported that the external and internal globus pallidus (GPe and GPi respectively) neurons exhibited this pathological pattern in PD patients (Hutchison et al., 1994, 1998; Sterio et al., 1994) and in 6-OHDA rat model of PD (Pan and Walters, 1988; Burbaud et al., 1995; Hassani et al., 1996; Ni et al., 2000; Tai et al., 2003). Recently, Delaville et al. (2012a,b) have shown that NA depletion increased the proportion of bursty and irregular neurons in the STN without affecting the neuronal activity of globus pallidus (GP) and the pars reticulata of substantia nigra (SNr).

Epidemiological studies have indicated a potential association between exposure to lead and an increased risk of PD, suggesting a 2–3-fold increase in risk for PD following lead exposure (Duckett et al., 1977; Gorell et al., 1997; Kuhn et al., 1998; Coon et al., 2006). Furthermore, a large case—control study with biomarker data on cumulative exposure to lead, has demonstrated evidence that high cumulative exposure to lead is associated with an increased risk of PD, providing some of the strongest evidence for a role for lead in the development of PD (Weisskopf et al., 2010). To investigate the relationship between this heavy metal and PD, (Jason and Kellogg, 1981) reported that exposure to high levels of lead induced a loss of dopaminergic neurons in the striatum. However, after cessation of exposure, lead brain levels decreased and lead-induced behavioral and neurochemical abnormalities dissipate (Jason and Kellogg, 1981). Recently, in rats, after 3 weeks of lead exposure, no change in the DA tissue content was revealed by HPLC in the striatum (Sabbar et al., 2012). Interestingly, in the same lead-exposed animals, a decrease in the tissue content of NA in the cortex paralleled by a change in the firing pattern of STN neurons were found (Sabbar et al., 2012). However, it is presently unclear if the NA depletion could participate to the atypical Parkinsonism induced by lead intoxication.

In the present study, we investigated the effects of sub-chronic low-dose lead intoxication and NA depletion, induced by a systemic administration of the neurotoxin N-(2-chloroethyl)- Nethyl-2-bromobenzylamine (DSP-4) (Lapiz et al., 2000; Delaville et al., 2012a,b), on locomotor activity and motor coordination, on the tissue content of the three monoamines (dopamine, noradrenaline, and serotonin) and their metabolites, and on the neuronal activity of the three major basal ganglia nuclei, the STN, GP, and SNr.

## MATERIALS AND METHODS

## Animals and Housing

Male Sprague Dawley rats (Centre d'Elevage Depré, Saint Doulchard, France) weighing 150–170 g at the time of experiments were used for behavioral and in vivo electrophysiological studies. Rats were kept in polycarbonate cages, 3 rats/cage, in a thermostatically controlled room (temperature: 24◦C, relative humidity: 45%) on a 12 h-light/12 h-dark schedule with free access to food and water. The body weights of rats were monitored throughout the experiment. All experiments were carried out in strict accordance with the Council Directive 2010/63/EU of the European Parliament and the Council of 22 September 2010 on the protection of animals used for scientific purposes. The experimental protocol was approved by the Ethics local Committee.

## Drugs and Solutions

Lead acetate and sodium acetate (Sigma, France) were dissolved in sterile water. L-DOPA (L-3,4-dihydroxyphenylalanine methyl ester hydrochloride), DSP-4 (N-(2-Chloroethyl)-N-ethyl-2 bromobenzylamine hydrochloride), and Benserazide (Sigma, France) were dissolved in saline (0.9%). L-DOPA, the precursor of DA, remains the most effective medication for PD (Tintner and Jankovic, 2002). Benserazide, which is the peripheral decarboxylase inhibitor, was administered to the animals at least 30 min before L-DOPA injection to prevent conversion of L-DOPA to dopamine in the periphery. DSP-4 had neurotoxic actions on noradrenergic neurons and selectively damages noradrenergic terminals originating from the locus coeruleus (LC) (Lapiz et al., 2000). All drugs were dissolved just before use and administrated intraperitoneally. The doses used were 10 mg/kg for lead acetate and sodium acetate, 25 mg/kg for DSP4, 25 mg/kg for benserazide, and 12 mg/kg for L-DOPA.

## Experimental Design and Groups

The experiments were performed as reported in **Figure 1**. Lead animals (n = 18) and controls (n = 18) received daily intraperitoneal (i.p.) injection of either lead acetate or sodium acetate, respectively, during 56 days. Behavioral tests were performed every week during treatment. Another group of rats (n = 18) received an injection of DSP-4 (Sigma-Aldrich, France) at the dose of 25 mg/kg. DSP-4 solution was administered i.p. once. Behavioral tests were carried out a week after the administration of DSP-4. In a subgroup of lead animals (n = 6), L-DOPA was injected i.p. at the end of the lead treatment for 4 days. At day 4, 40 min after L-DOPA injection, locomotor behavior, and motor coordination were evaluated using the open field and the rotarod tests respectively. DSP-4 animals (n = 6) received a L-DOPA injection as lead rats and motor behavior was evaluated in the same conditions.

## Behavioral Tests

#### Open-Field Activity Test

Exploratory and locomotor activities were evaluated by an "open field" actimeter (Actitrack, Panlab, Barcelona, Spain), which consisted of a transparent cage that was connected to a photoelectric cell. Light beams detected the rat movements and the total locomotor activity of rats of each group was recorded over 20 min. Exploratory and locomotor activities were evaluated as previously described (Sabbar et al., 2012). Behavioral testing in the actimeter was done in an isolated room between 8:00 a.m. and 1:00 p.m. The spontaneous locomotor activity was recorded during 20 min for four consecutive days with habituation during the first 3 days and the test on day 4. The first 5 min session was used to establish the exploratory activity and the last 10 min was used to evaluate locomotor activity. From week 2, behavioral testing was conducted once per week during the 56 days of lead acetate or sodium acetate treatment.

## Rotarod Task

The rotarod test is widely used to evaluate the motor coordination of rodents. The rat was placed on the rod, which consisted of a 7 cm diameter that rotates at different speeds (Bioseb, in vivo Research Instruments, Spain). Before the testing sessions, the rats were habituated to stay on the stationary rod for 3 min. Habituation was repeated every day for 3 days before the beginning of lead treatment and the testing sessions. The animal was placed back on the rod immediately after falling, up to 5 times in one session. For testing, the rod was accelerated from 4 to 20 rpm over 180 s and time to fall off the apparatus was recorded for a maximum of 180 s. Two trials were conducted and times were averaged for group comparisons. Testing was conducted once per week during the 56 days of lead acetate or sodium acetate treatment.

## Electrophysiological Recordings

Extracellular single-unit recordings were performed as previously described (Ni et al., 2001; Belujon et al., 2007). STN, GP and SNr neurons were recorded in anesthetized rats (urethane 1.2 g/kg, i.p.). After being placed in a stereotaxic instrument, body temperature of rats was maintained at 37◦C with a heating pad. All procedures were carried out in accordance with European Communities Council Directive 2010/63/UE to reduce animals suffering. A single glass micropipette electrode was filled with 4% pontamine sky blue in 3 M NaCl, and the electrode tips were broken back under microscopic control until in vitro tip impedance measured 8–12 M. Microelectrodes were stereotaxically guided through drilled skull holes to the target coordinates (in mm) for the STN: AP: −3.8, L: −2.5, D: 6.8–8.2; for the GP: AP: −0.9, L: −3, D: 4.5–7.5; and for the SNr: AP: −5.3, L: −2.5, D: 6.5–9 (Paxinos and Watson, 1996). Electrical signals were passed through a preamplifier (Neurolog, Digitimer, UK), and amplified signals were monitored with an oscilloscope and transferred via a Powerlab interface (AD Instruments, Charlotte, NC, USA) to a computer equipped with Chart 5 software (AD Instruments, Charlotte, NC, USA). Only neuronal single-unit activity with a signal-to-noise ratio > 3:1 was recorded. At the end of each recording session, the recording site was marked by electrophoretic injection (Iso DAM 80, WPI, Hertfordshire, UK) of Pontamine sky blue through the micropipette at a negative current of 20 µA for 7 min. The location of the Pontamine sky blue dots was histologically verified as previously reported (Belujon et al., 2007) and only recordings from the brains in which the dot was clearly visible in the STN, GP, and SNr were used for data analysis.

#### Data Analysis

The electrical activity of each neuron was analyzed with a spike discriminator using a spike histogram program (AD Instruments, Charlotte, NC, USA). The firing rates and patterns were determined using Neuroexplorer program (AlphaOmega, Nazareth, Israel) and the method developed by Kaneoke and Vitek (1996) as previously described (Labarre et al., 2008).

## Biochemical Monoamine Assays

After electrophysiological recordings, animals were sacrificed by decapitation, the brains were rapidly removed and dissection of the striatum and the frontal cortex was performed on ice at −20◦C. Tissue samples were frozen by immersion in cold isopentane and stored at −80◦C until used for neurochemical determinations. Levels of dopamine (DA) and its metabolites, dihydroxyphenylacetic acid (DOPAC), and homovanillic acid (HVA), as well as noradrenaline (NA), serotonin (5-HT), and its metabolite 5 hydroxyindoleacetic acid (5-HIAA) were quantified by a method of high performance liquid chromatography combined with electrochemical detection (HPLC/EC) as previously described (De Deurwaerdere et al., 1995). The samples were weighed, sonicated in 0.1 N perchloric acid, centrifuged 13,000 rpm for 30 min at 4◦C. Aliquots of the supernatants were diluted in the mobile phase which contains 60 NaH2PO4, 0.1 mM disodium EDTA, and 2 octane sulfonic acid plus 7% methanol, adjusted to pH 3.9 with orthophosphoric acid. The mobile phase was filtered through a 0.22µm millipore filter and pumped through the system at a flow rate of 1.2 ml/min. Detection of monoamines and their metabolites was performed with a coulometer detector (CoulochemI, ESA) coupled to a dual-electrode analytic cell (model 5011). The potential of the electrodes was set at +350 and −270 mV.

Concentrations of monoamines were calculated using standard curves. Final results were expressed as means ± SEM in terms of monoamine contents per g of wet tissue (ng/g of tissue). DA and its metabolites (DOPAC and HVA) as well as 5-HT and its metabolite (5-HIAA) were measured in the striatum. NA, 5-HT and its metabolite (5-HIAA) were measured in the frontal cortex of all groups of rats.

## Data Processing and Statistical Analysis

Data are reported as means ± SEM (standard error of the mean). In the present study, sample size was calculated using "resource equation" method as previously reported (Charan and Kantharia, 2013). A value "E" which represents the degree of freedom of analysis of variance (ANOVA) should lie between 10 and 20 and any sample size, which keeps E between these values should be considered adequate. In the present study, we calculated E value using the formula:

E = Total number of animals − Total number of groups

The E value for all the tests used in the study was largely above 20 indicating that our sample size of animals is adequate.

All statistical procedures were conducted using the GraphPad Prism program (San Diego, USA). Behavioral results were analyzed using a two-way ANOVA comparing several groups. All significant ANOVAs were followed by Sidak's post-hoc test. Biochemical and the firing rate data were analyzed using a oneway ANOVA (followed by Tukey's post-hoc test). Changes in the proportion of different firing patterns (regular, irregular and bursty) were determined using a Chi² test. The levels of DA, NA, and 5-HT and their respective metabolites were analyzed using one-way ANOVA followed by Tukey's post-hoc test when ANOVA was significant. A P < 0.05 was considered statistically significant.

## RESULTS

## Body Weight Changes

**Figure 2A** shows that lead acetate injections significantly slowed the progression of body weight gain compared to control rats receiving sodium acetate injections (Two way ANOVA; effect on time [F(9, 45) = 250.7, P < 0.0001], treatment [F(1, 5) = 20.55, P = 0.0062] and interaction (treatment × time) [F(9, 45) = 22.12, P < 0.0001]. At the beginning of the experiment, lead rats had a mean body weight (165.8 ± 7.40 g) that was not significantly different from that of controls (149.5 ± 6.35 g). The two groups of rats grew at similar rates during the first week of treatment. However, within 14 days of treatment, the consequence was the significant reduction in body weight gain in the rats undergoing repeated lead injections (P < 0.05), such that at the end of the experiment, the mean body weight of lead rats was significantly lower than that of controls (287.5 ± 9.39 g vs. 397.5 ± 11.53 g, Sidak's post-hoc test, P < 0.0001, **Figure 2A**).

**Figure 2B** shows the mean ± SEM rate of body weight gain from the beginning of the treatment to 1 week after the end

of treatment, represented as change in body weight measured at 7 days intervals. All rats were gaining body weight at a rate between 37 and 47 g per week by the time of the first week of the treatment. Two-way repeated measures ANOVA showed significant effects of lead on time [F(9, 45) = 15.20, P < 0.0001], treatment [F(1, 5) = 53.50, P = 0.0007] and interaction (treatment × time) [F(9, 45) = 4.215, P = 0.0005]. Lead treatment caused a significant mean loss of body weight of about 26 g (**Figure 2B**; two-way ANOVA, followed by Sidak's post-hoc test, P < 0.01 and P < 0.0001). Control rats injected repeatedly with sodium acetate showed a steady growth rate from 21 to about 35 days during the treatment. Peak rate of growth was 48.5 ± 1.33 g/7 days. Thereafter, the rate of growth decreased in all rats, presumably as they approached adult rat body weight.

## Effect of Lead and DSP-4 Treatments on Exploratory and Locomotor Activities

Horizontal movements measured during the first 5 min in the actimeter were used to assess the rats' exploratory activity and those measured during the last 10 min were considered as the rats' locomotor activity. Two-way ANOVA analysis showed that lead significantly affected the exploratory activity in the open field [F (1, 214) = 56.70, P < 0.0001, **Figure 3A**] with an interaction between lead treatment and time [F (9, 214) = 3.725, P = 0.0002]. Sidak's post-hoc test showed that lead rats significantly reduced the number of their horizontal movements during the first 5 min-testing sessions, which started 14 days after the beginning of the treatment and remained lower at day 56 of lead treatment, when compared to control rats. In contrast to exploratory activity, two-way ANOVA analysis of the locomotor activity showed a delayed, but significant, effect of lead treatment on locomotor activity recorded during the last 10 min of each session [F (1, 214) = 4.639, P = 0.0324, **Figure 4A**] as well as of lead treatment × sessions interaction [F (9, 214) = 3.167, P = 0.0013]. Compared to control rats, the significant decrease in locomotor activity has been observed only at day 56 of lead treatment (Sidak's post-hoc test, P < 0.01, **Figure 4A**).

The effects of DSP-4 on the exploratory and locomotor activities were evaluated at day 56 and compared to the effect of lead treatment at the same time point. DSP-4 rats showed a decrease in their exploratory and locomotor activities as expressed by the significant reduction in the number of horizontal movements measured during the first 5 min session [One way ANOVA, F(2, 29) = 19.04; P < 0.0001 followed by Tukey's post-hoc test, P < 0.0001; **Figure 3B**] and the last 10 min session [One way ANOVA, F(2, 29) = 11.85; P = 0.0002 followed by Tukey's post-hoctest, P < 0.05; **Figure 4B**] in the open field test compared to controls. When compared to lead rats, no significant difference was observed.

The 4 days of L-DOPA treatment did not improve either the exploratory (P > 0.05, **Figure 3C**) or the locomotor (P > 0.05, **Figure 4C**) activities measured in the actimeter in lead rats as well as in DSP-4 rats.

## Effect of Lead and DSP-4 Treatments on Motor Coordination

Lead treatment affected motor coordination as shown by the decrease in the time spent on the rotated bar in the rotarod test [two-way ANOVA: effect on time F(9, 181) = 9.616, P < 0.0001, treatment, F(1, 181) = 63.51, P < 0.0001, and treatment × time interaction, F(9, 181) = 2.551, P = 0.0088). Lead rats spent less time on the bar compared to the controls since day 22 (P < 0.05; P < 0.01, P < 0.001, Sidak's post-hoc test; **Figure 5A**).

One way ANOVA analysis showed that lead as well as DSP-4 treatments significantly decreased the time spent on the rotated bar compared to controls at day 56 [F (2, 15) = 8.796, P < 0.0030, **Figure 5B**]. DSP-4 treated animals showed a significant decrease of the time spent on the rotated bar compared to controls (P < 0.01, **Figure 5B**). The effect of DSP-4 was similar to that of lead-treated rats, as no significant difference was observed between the two groups (P > 0.05, **Figure 5B**) and to lead treated animals (P < 0.01 and P < 0.01 respectively, **Figure 5B**).

L-DOPA treatment did not improve the motor coordination in lead rats as well as in DSP-4 rats as it did not improve the time spent on the rotated bar in the two groups (P > 0.05, **Figure 5C**).

## Effects of Lead and DSP-4 on the Neuronal Activity of STN, GP, and SNr Effects on STN Neurons

A total of 102 neurons were recorded in the STN of control, lead and DSP-4 rats. The firing rate of STN neurons was not significantly different in the three different groups of animals

followed by Sidak's post-hoc test. Data from controls, lead and DSP-4 rats (n = 12) on (B) were compared using the one-way ANOVA followed by Tukey's post-hoc test. Data from lead and DSP-4 rats before and after L-DOPA treatment on (C) were compared using the Mann Whitney test. \*P < 0.05, \*\*P < 0.01, \*\*\*P < 0.001, \*\*\*\*P < 0.0001, NS no significant.

[One way ANOVA, F(2, 99) = 0.4103, P = 0.66, **Figures 6A,C**] as previously reported (Delaville et al., 2012a; Sabbar et al., 2012). In control rats, the mean firing rate of STN neurons was 11.92 ± 1.74 spikes/s (n = 33), in lead treated rats 12.04 ± 2.15 spikes/s (n = 30) and in DSP-4 rats 14.01 ± 1.81 spikes/s (n = 39).

Concerning the firing pattern, sub-chronic lead treatment significantly increased the proportion of STN neurons with bursty pattern compared to controls (Chi² test, X² = 13.083, df = 2, P = 0.010, **Figures 6B,D**). In DSP-4 rats, a significant increase in the proportion of irregular neurons was found compared to controls (Chi² test, X² = 13.083, df = 2, P = 0.010; **Figure 6B**).

#### Effects on GP Neurons

A total of 138 neurons were recorded in the GP of control, lead and DSP-4 rats. Neither the lead treatment nor the DSP-4 treatment affected the firing rate of GP neurons [One way ANOVA, F(2, 135) = 0.1738, P = 0.84, **Figure 6A**]. The firing rate of GP neurons was 20.70 ± 1.57 spikes/s (n = 33) in control rats, 20.75 ± 2.08 spikes/s (n = 30) in lead treated rats and 19.53 ± 1.38 spikes/s (n = 39) in DSP-4 rats. Furthermore, the firing pattern of GP neurons was also unaffected by either lead or DSP-4 treatments (Chi² test, X² = 0.977, df = 2, P = 0.61 for lead rats, and Chi² test, X² = 5.181, df = 2, P = 0.075 for DSP-4 rats when compared to controls, **Figure 6B**).

#### Effects on SNr Neurons

We examined also the extracellular single-unit activity of SNr neurons and we recorded a total of 108 neurons in control, lead and DSP-4 rats. The firing rate of SNr neurons was not significantly different in the three different groups [One way ANOVA, F(2, 105) = 1.259, P = 0.29, **Figure 6A**]. In control rats, the mean firing rate of SNr neurons was 29.82 ± 2.44 spikes/s (n = 33), in lead treated rats 25.51 ± 2.38 spikes/s (n = 30) and in DSP-4 rats 24.88 ± 2.56 spikes/s (n = 39). Furthermore, either lead treatment nor DSP-4-induced NA depletion affected the firing pattern of SNr neurons (Chi² test, X² = 5.934, df = 2, P = 0.051 for lead rats, and Chi² test, X² = 2.594, df = 2, P = 0.27 for DSP-4 rats when compared to controls, **Figure 6B**).

## Effect of Lead and DSP-4 on Monoamine Tissue Levels in the Striatum and the Frontal Cortex

Lead as well as DSP-4 treatments significantly affected the tissue level of NA in the frontal cortex [One-way ANOVA, F(2, 18) = 6.741, P = 0.0065, **Table 1**]. Indeed, lead treatment significantly decreased the tissue level of NA in the cortex (−25.5%, P < 0.05, **Table 1**) as DSP-4 did (−39.5%, P < 0.01, **Table 1**) in comparison to controls. However, lead but not DSP-4, decreased the tissue level of DA (−64.6%, P = 0.0006) and its metabolites DOPAC (−66.2%, P < 0.0001) and HVA (−65.4%, P = 0.0105) in the striatum. Furthermore, neither lead nor DSP-4 affected the tissue contents of 5-HT and 5-HIAA in the striatum [5-HT: One way ANOVA, F(2, 18) = 0.089, P = 0.91; 5-HIAA: One way ANOVA, F(2, 18) = 0.041, P = 0.96] and in the cortex (5-HT: One way ANOVA, F(2, 18) = 1.458, P = 0.26; 5-HIAA: One way ANOVA, F(2, 18) = 3.080, P = 0.07).

## DISCUSSION

It has been reported that lead exposure can cause neurotoxicity and neurologic disorders that resembles PD. The present study is an extension of our previous work (Sabbar et al., 2012), and to our knowledge, our data provide evidence that lead intoxication induced atypical Parkinsonism that can be differentiated from PD. Here, we used a subchronic treatment of lead at the dose of 10 mg/kg, which induced a decrease in DA as well as NA tissue contents in contrast to our previous work (Sabbar et al., 2012), in which we showed that the dose of 20 mg/kg induced a decrease in NA tissue content only, without affecting the level of DA. These depletions resulted in a decrease of exploratory and locomotor activities as well as motor coordination, which were paralleled by the disorganization of the firing pattern of STN neurons without affecting the electrophysiological parameters of GP and SNr neurons. Furthermore, similar behavioral and electrophysiological results have been found after the treatment with DSP4-induced NA depletion. To confirm the atypical character of Parkinsonism induced by lead and DSP-4, we tested the behavioral effect of L-DOPA, and our data are the first to show the ineffectiveness of this dopaminergic antiparkinsonian agent on motor disabilities induced by the two neurotoxicants.

Lead is a well-established neurotoxicant heavy metal, and a growing body of data suggested that lead neurotoxicity is

the one-way ANOVA followed by Tukey's post-hoc test. Data from lead (n = 6) and DSP-4 (n = 6) rats on (C) were compared using the Mann whitney test. \*P < 0.05, \*\*P < 0.01, \*\*\*P < 0.001, NS, no significant.

associated with elevated risk of PD (Duckett et al., 1977; Gorell et al., 1997; Kuhn et al., 1998; Coon et al., 2006). Thus, in the present study, were investigated the relationship between this heavy metal and Parkinsonism.

Two weeks after a recurrent i.p. administration of 10 mg/kg of lead acetate, the treated rats showed an inhibitory effect on body weight gain and a reduced ability of rate of growth as previously reported in lead-treated pregnant rats and pups (Aprioku and Siminialayi, 2013). This effect could be attributed, at least in part, to a deficiency in energy metabolism and to the alterations of normal cell metabolism (Patel et al., 1974; Wapnir et al., 1977). More specifically, it could be attributed to lead-induced zinc deficiency (Bushnell and Levin, 1983; Taha et al., 2013). Moreover, the reduction of body weight gain can be due to a reduction in food intake as it has been shown in kids that clinical symptoms of lead exposure generally begin with loss of appetence (Lowry, 2010). Since we did not measure food intake in lead group, and there was no apparent change in food consumption in our lead rats, we can only speculate that there is metabolic dysfunction. Indeed, one of the effects of lead exposure is on glutathione metabolism (Hunaiti et al., 1995). Glutathione is an important antioxidant for quenching free radicals in the liver and glutathione metabolism considered as a highly effective compensatory mechanism to overcome metal toxicity (Hsu, 1981). However, this mechanism seems no longer effective following a long-term lead toxicity as a decrease in glutathione reductase, glutathione peroxidase, and glutathione-S-transferase levels were observed in occupationallyexposed workers that are correlated with depressed glutathione (Adonaylo and Oteiza, 1999) leading to reduce body weight gain.

Sub-chronic lead intoxication resulted in a decrease of exploratory and locomotor activities as measured in the open field test. These locomotor impairments associated with the low exploratory behavior are consistent with previous studies (Reiter et al., 1975; Moreira et al., 2001; NourEddine et al., 2005; Reckziegel et al., 2011; Sansar et al., 2011). It might be argued that the low exploratory behavior is due to a partial lack of motivation of lead-treated rats to explore the open field arena. However, it is unlikely to consider this hypothesis as in a previous study we have shown that lead did not induce "depressive-like" disorder (Sabbar et al., 2012).

Furthermore, lead rats displayed an impairment in the rotarod performance, which accounts for deficits in motor coordination and balance (Hamm et al., 1994; Rogers et al., 1997; Rozas et al.,

FIGURE 6 | Effects of lead and DSP-4 treatments on the electrical activity of STN, GP, and SNr neurons. (A) Representative examples of spike trains and the corresponding interspike interval histograms (B) showing regular, irregular and bursty neurons recorded in the STN. (C) Histograms represent the mean ± SEM of the firing rate of all neurons recorded in each experimental group (control, lead, and DSP-4 rats). Firing rate data from control, lead and DSP-4 rats were compared using the one-way ANOVA. (D) Firing pattern histograms showing the proportion (%) of STN, GP and SNr cells discharging regularly (white portion), irregularly (gray portion) or with bursts (black portion). Changes in the proportion of different firing patterns were analyzed using a Chi² test. \*P < 0.05 comparison with controls.



Lead treatment induced selective depletion of NA and DA and its metabolites DOPAC and HVA without affecting the tissue content of 5-HT and 5-HIAA. Values are tissue contents in ng/g of wet tissue presented as the mean ± SEM. Data from controls (n = 6), lead (n = 6) and DSP-4 (n = 10) rats were compared using the one-way ANOVA followed by Tukey's post-hoc test. \*P < 0.05, \*\*P < 0.01 comparison of controls vs. lead rats and controls vs. DSP-4 rats. ++P < 0.01, +++P < 0.001 comparison of lead vs. DSP-4 rats.

1997). These results are in line with our previous study (Sabbar et al., 2012), in which we showed that lead intoxication at the dose of 25 mg/kg during 3 weeks altered NAergic transmission that caused motor impairments. Indeed, i.p. injection of DSP-4 (25 mg/kg), a neurotoxin which affects primarily NA terminals arising from the LC (Jonsson et al., 1982; Fritschy and Grzanna, 1989; Fritschy et al., 1990), induced low exploratory behavior, hypolocomotor activity, and impaired motor coordination. In line with earlier reports (Berridge and Dunn, 1990; Harro et al., 1995; Delaville et al., 2012a; Sabbar et al., 2012), our results add strong evidence that the observed exploratory and motor disturbances following lead treatment are due to the depletion of NA.

Although the NAergic system is rarely associated with motor functions, a growing number of data strengthened the involvement of abnormal NAergic neurotransmission in mediating motor disabilities (see review Rommelfanger et al., 2007; Delaville et al., 2011, 2012a; Pifl et al., 2013). Indeed, Rommelfanger et al. (2007) showed that knockout mice (Dbh−/−) that lack NA display robust motor deficits including motor coordination impairment. Another study by Delaville et al. (2012a) reported similar motor impairment after DSP-4 treatment in the rat. In addition, in PD, besides the DA neurodegeneration (Ehringer and Hornykiewicz, 1960), it has been shown that PD is also characterized by the degeneration of NA neurons in the LC (Forno, 1996; Bertrand et al., 1997), which is greater than the degeneration of DA neurons in the SNc (Zarow et al., 2003).

Our results showed that 4 days of L-DOPA treatment failed to improve motor disabilities in lead as well as in DSP-4 rats. Because L-DOPA has been shown to be highly effective in the alleviation of motor disabilities in animal models of PD and PD (Navailles et al., 2010, 2014; Nevalainen et al., 2014; De Deurwaerdère et al., 2017), our results suggest that the motor deficits induced by lead and DSP-4 correspond to atypical Parkinsonism. Indeed, at the dose used in our study, L-DOPA has been shown to enhance extracellular levels of DA in lesioned and non-lesioned rats (Navailles et al., 2010; Nevalainen et al., 2014). Its effect is likely preserved in our model due to the lack of alteration of serotonergic neurons which are responsible for L-DOPA-induced DA release (Navailles et al., 2010), and would be even magnified with the loss of NA terminals (Navailles et al., 2014). Thus, the severe drop of neurochemical dopaminergic markers in the striatum (>60% decrease for DA tissue content and its metabolites) is unlikely responsible for the parkinsonism induced by lead. L-DOPA is also a 2-step metabolic precursor of NA but its effects on NA release are modest and still a matter of debate (De Deurwaerdère et al., 2017). It could not be sufficient to counteract the reduction of NA tissue contents induced by DSP-4 and lead, presumably due to a destruction of NAergic terminals. Additional experiments with pharmacological agents rescuing NAergic transmission more selectively than L-DOPA are warranted. Our data confirm that lead intoxication alters DAergic and NAergic neurotransmission (Jason and Kellogg, 1981; Lorton and Anderson, 1986; Albin et al., 1989; Selvín-Testa et al., 1994; Thach, 1998; Tavakoli-Nezhad et al., 2001) and stress a possible role of NA deficit in the atypical Parkinsonism we report.

A large number of investigations reported that lead neurotoxicity affected several brain areas that are involved in motor control, such as the cerebral cortex and basal ganglia (Tavakoli-Nezhad et al., 2001; Sabbar et al., 2012). This may underlie, at least in part, the motor deficits observed in the present study. However, it should be considered that brain regions other than the basal ganglia can be involved in the long-term motor disabilities produced by lead, including the cerebellum, a region playing a critical role in motor coordination, and balance (Rozas et al., 1997; Thach, 1998), reported to be affected after lead intoxication (Lorton and Anderson, 1986; Selvín-Testa et al., 1994).

The basal ganglia are a group of highly interconnected subcortical nuclei that are involved in motor control (Alexander et al., 1986; Albin et al., 1989; Temel et al., 2006). According to the basal ganglia functional model, the striatum represents the main input structure of the system and SNr and the entopeduncular nucleus (GPi, internal GP in primate) represent the output structures projecting mainly to the thalamus. Input and output structures are linked by a monosynaptic direct pathway and polysynaptic indirect pathway that involves GP (GPe, external GP in primate) and the STN (Albin et al., 1989; Alexander and Crutcher, 1990; Alexander et al., 1990; Burkhardt et al., 2007). Studies that have investigated the pathophysiology of PD have provided significant insight into the complex role of DA and NA depletions on the neuronal activity of basal ganglia structures (Delaville et al., 2012a,b). Indeed, STN neurons, which normally exhibit a tonic discharge pattern, become irregular, and bursty in animal models of PD (Bergman et al., 1994; Ni et al., 2001; Delaville et al., 2012a). The pathological pattern has also been reported in the STN, GPe/GPi of PD patients (Hutchison et al., 1994, 1998; Sterio et al., 1994; Benazzouz et al., 2002).

In the current study, our electrophysiological results show that chronic lead treatment did not affect the firing rate of STN neurons, in line with several previous studies where investigators were unable to detect changes in the firing rate of STN neurons in the rat model of PD (Ni et al., 2001; Tai et al., 2003; Belujon et al., 2007; Delaville et al., 2012a) and in lead-induced neurotoxicity in rats (Sabbar et al., 2012). Interestingly, DSP-4 treatment, when combined with 6-OHDA injection, did not affect the firing rate of STN neurons (Delaville et al., 2012a). Nevertheless, it should be noted that unlike the firing rate, the firing pattern is considered as the most relevant and consistent parameter in the pathophysiology of PD (Bergman et al., 1994; Ni et al., 2001; Tai et al., 2003; Meissner et al., 2005; Belujon et al., 2007; Chetrit et al., 2013). Accordingly, our results show that lead treatment induced a profound disorganization in the firing pattern of STN neurons. Indeed, NA depletion induced by DSP-4 injection provoked a switch from regular to irregular pattern whereas lead treatment induced a switch from regular to bursty and irregular pattern compared to controls. The latter results are in agreement with those observed by Delaville et al. (2012a) in which DSP-4 treatment in rats significantly increased the proportion of irregular and bursty neurons in the STN. Interestingly, and in contrast with earlier studies carried out in 6-OHDA rat model of PD (Pan and Walters, 1988; Burbaud et al., 1995; Hassani et al., 1996; Ni et al., 2000; Tai et al., 2003), we did not observe any change in the electrophysiological parameters of GP and SNr neurons following lead treatment or DSP-4 administration. These results suggest that the behavioral changes observed after lead intoxication are due to the depletion of DA, which is responsible of the changes in the STN neuronal activity without affecting GP and SNr neurons. Our results fit with those of Delaville et al. (2012b), who also showed that DSP-4 injections did not affect either the firing rate or the firing pattern of GP and SNr neurons. Nevertheless, the same authors have also reported that 6-OHDA/DSP-4 injections affected the firing rate and the firing pattern of SNr neurons (Delaville et al., 2012b). These results are different from those observed in rats intoxicated with lead in whom both DA and NA depletions occurred. These discrepancies can be explained by the fact that the observed DA/NA depletions in lead rats did not reach the critical level to induce electrophysiological changes in the SNr as reported in the 6-OHDA rat model of PD.

In our study we focused on the consequences of lead intoxication on brain functions. However, lead toxicity is multifactorial and it is unlikely that its effects are limited to

## REFERENCES

Adonaylo, V. N., and Oteiza, P. I. (1999). Lead intoxication: antioxidant defenses and oxidative damage in rat brain. Toxicology 135, 77–85. doi: 10.1016/S0300-483X(99)00051-7

brain functions. In addition to the central and peripheral nervous system, lead toxicity has shown many harmful effects. It is known to induce a broad range of physiological, biochemical and behavioral dysfunctions in humans and lab. animals, including hematopoietic system, cardiovascular system, kidneys, liver, and reproductive system (Matovic´ et al., 2015; Pal et al., 2015).

In summary, our findings provide strong evidence that leadinduced atypical Parkinsonism expressed by the impairment of exploratory and locomotor activities as well as motor coordination. These disorders are associated principally with the depletion of NA, with a low implication of DA depletion. These two monoamines are known to be involved in motor functions. As described in several reports about the pathophysiology of PD (Forno, 1996; Bertrand et al., 1997), it is possible that lead neurotoxicity affected first the NAergic system (Sabbar et al., 2012) before the DAergic system, and that the depletion of NA is essential in the manifestation of atypical Parkinsonian-like motor disabilities, which were paralleled by the irregular/busrty firing pattern of STN neurons. Our study highlights and reinforces the possible contribution of lead, as an environmental factor, in the development of atypical Parkinsonism. Further experiments are needed to determine the direct link between lead intoxication, the depletion of NA and the behavioral impairments and electrophysiological changes.

## AUTHOR CONTRIBUTIONS

MS: carried out the experiments, collected, and analyzed the data, wrote and edited the manuscript; AB: designed the experimental protocol, supervised the work, and participated in writing the paper; CD: assisted with data collection; PD: supervised the neurobiochemical analysis of monoamines; NL-G and AB: edited and approved the final draft of the manuscript. All authors read and approved the final manuscript.

## FUNDING

This research was funded by the University of Bordeaux and exchange grants of the GDRI N198 (CNRS and INSERM, France, and CNRST, Morocco), Egide-Volubilis No 20565ZM, CNRS-CNRST Convention Adivmar 22614, and NEUROMED.

## ACKNOWLEDGMENTS

We are very grateful to Dr. Roger Butterworth and Dr. Evelyne Sernagor for revising the manuscript. We thank Dr. Rabia Bouali-Benazzouz for her assistance with the rotarod test and all the staff of the Central Animal Service for care of the animals (University of Bordeaux, Neurodegenerative Disease Institute).


Trends Neurosci. 13, 266–271. doi: 10.1016/0166-2236(90) 90107-L


middle cerebral artery occlusion in the rat. Stroke 28, 2060–2065. discussion: 6. doi: 10.1161/01.STR.28.10.2060


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Sabbar, Delaville, De Deurwaerdère, Lakhdar-Ghazal and Benazzouz. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Prenatal Exposure to Paint Thinner Alters Postnatal Development and Behavior in Mice

Hanaa Malloul † , Ferdaousse M. Mahdani † , Mohammed Bennis and Saadia Ba-M'hamed\*

Laboratory of Pharmacology, Neurobiology and Behavior (URAC-37), Faculty of Sciences Semlalia, University Cadi Ayyad, Marrakech, Morocco

Occupational exposure and sniffing of volatile organic solvents continue to be a worldwide health problem, raising the risk for teratogenic sequelae of maternal inhalant abuse. Real life exposures usually involve simultaneous exposures to multiple solvents, and almost all the abused solvents contain a mixture of two or more different volatile compounds. However, several studies examined the teratogenicity due to industrial exposure to a single volatile solvent but investigating the teratogenic potential of complex chemical mixture such as thinner remains unexplored. This study was undertaken to evaluate developmental neurotoxicity of paint thinner using a mouse model. Mated female mice (N = 21) were, therefore, exposed to repeated and brief inhalation episodes of 0, 300 or 600 ppm of thinner during the entire period of pregnancy. Females weigh was recorded and their standard fertility and reproductive parameters were assessed. After birth postnatal day 1 (PND1), offspring (N = 88) length and body weight were measured in a daily basis. At PND5, the pups were assessed for their postnatal growth, physical maturation, reflex development, neuromotor abilities, sensory function, activity level, anxiety, depression, learning and memory functions. At adulthood, structural changes of the hippocampus were examined by estimating the total volume of the dentate gyrus. Except one case of thinner induced abortion at the higher dose, our results showed that the prenatal exposure to the solvent did not cause any maternal toxicity or decrease in the viability of the offspring. Therefore, a lower birth weight, decrease in the litter size and delayed reflexes ontogeny were registered in prenatally exposed offspring to both 300 ppm and 600 ppm of thinner. In addition, prenatally exposure to thinner resulted in increased anxiolytic- and depression-like behaviors. In contrast, impaired learning and memory functions and decreased hippocampal dentate gyrus volume were revealed only in the prenatally treated offspring by 600 ppm of thinner. Based on these results, we can conclude that prenatally exposure to paint thinner causes a long-lasting developmental neurotoxicity and alters a wide range of behavioral functions in mice. This shows the risk that mothers who abuse thinner paint expose their offspring.

#### Edited by:

Nilesh Bhailalbhai Patel, University of Nairobi, Kenya

#### Reviewed by:

Noriyuki Koibuchi, Gunma University, Japan Valerie J. Bolivar, Wadsworth Center, United States

> \*Correspondence: Saadia Ba-M'hamed bamhamed@uca.ma

†These authors have contributed equally to this work.

Received: 19 June 2017 Accepted: 29 August 2017 Published: 11 September 2017

#### Citation:

Malloul H, Mahdani FM, Bennis M and Ba-M'hamed S (2017) Prenatal Exposure to Paint Thinner Alters Postnatal Development and Behavior in Mice. Front. Behav. Neurosci. 11:171. doi: 10.3389/fnbeh.2017.00171

Keywords: paint thinner, prenatal exposure, fertility, reproduction, development, behavior

## INTRODUCTION

Paint thinner is a chemical mixture of different aromatic and halogenated hydrocarbons (e.g., toluene, benzene, xylene and N-hexane), commonly used in various industrial applications as solvent in removing household paints and thinning oil-based paint (Solak et al., 2006; Verma and Gomber, 2009; Singh et al., 2012; Agin et al., 2016). Despite its usefulness, there is increasing concern about occupational exposure of industrial workers to this solvent (Yilmaz et al., 2001). In addition, thinner acts as psychoactive compounds when inhaled and its inhalation by young sniffers to achieve an euphoric state continues to be a significant worldwide health problem (Cruz, 2011; Howard et al., 2011; Bowen and Cruz, 2012).

While the volatile substance abuse is documented by diverse groups throughout the world, no epidemiological studies on inhalant abuse in Africa are available. This hampers the estimation of the extent of inhalant abuse across the general population. In the United States alone, the National Survey on Drug Use and Health has documented about 22 million 12 years old Americans and older deliberately abused inhalants at least once in their life from 2002 to 2005 (Substance Abuse and Mental Health Services Administration, 2007). Data show evidence that this abuse of inhalants has dramatically increased over the past years (Bowen and Hannigan, 2013). From 2009 to 2010, the median age at first use among those aged 12–49 years was unchanged (16.9 and 16.3 years, respectively) with 68.4% being under age of 18 years (Substance Abuse and Mental Health Services Administration, 2011). In 2013, 6.3% of teens in the United States (average age of 19.2 years) reported that the first illicit drug they had ever used was the inhalant (Substance Abuse and Mental Health Services Administration, 2014). Whereas in the past, the abuse of volatile substances was once predominantly limited to males, the gap between males and females is currently reversed: more than 50% of chronic solvents abusers are young women of reproductive age (Substance Abuse and Mental Health Services Administration, 2007; Butland et al., 2012; Bowen and Hannigan, 2013). Further, while a subcategory of inhalant abusers continue to do so into adulthood (Williams and Storck, 2007; Substance Abuse and Mental Health Services Administration, 2014), there is increasing concern about the potential negative consequences of deliberately inhaled solvents on the unborn (Bukowski, 2001; Hannigan and Bowen, 2010).

In comparison to inhalant abuse, reports of occupational exposure are not well documented due to potential limitations including small sample size, selection bias, uncontrolled and/or unspecified co-drug exposures and multiple solvent exposures (Hannigan and Bowen, 2010). Standards for a permissible exposure limit (PEL) for toluene have been established by U.S. Occupational Safety and Health Administration (OSHA) at 100 ppm (375 mg/mm<sup>3</sup> , calculated over an 8 h day as a time weighted average; Donald et al., 1991). After acute exposures of at least 500 ppm, a sense of euphoria generally is achieved; inhibition, confusion, incoordination, auditory and visual hallucinations occur at 600–800 ppm (Brozosky and Winkler, 1965). In contrast to chronic low-level (500 ppm) of occupational solvent exposure (Agency for Toxic Substances and Disease Registry (ATSDR) (2000)), inhalant abusers usually inhale very repeatedly, deeply and rapidly. Therefore, abused toluene minimally reaches 800 ppm, an intoxicating exposure (Filley et al., 2004), although abuse levels in chronic abusers often are higher and exceeding 5000 ppm (50 times the OSHA PEL; Ron, 1986).

In human adults, effects observed after long-term occupational exposures to thinner include dizziness, memory deficits, dementia, depression and fatigue (Kishi et al., 1993; Wang and Chen, 1993; Lee et al., 2003; Bowen et al., 2006). These wide ranges of neurobehavioral symptoms were reported to be due to the cerebral cortical and hippocampal atrophy and to the decreased brain volume observed in solvent abusers (Fornazzari et al., 1983; Lazar et al., 1983; Zur and Yule, 1990; Kamran and Bakshi, 1998). Some of these effects have been reproduced in animal studies. Neonatal exposure to volatile inhalants leads also to persistent impacts on behaviors and brain-structural properties (Benignus, 1981; von Euler et al., 1993, 2000). However, there is a limited amount of literature on teratogenic impact of prenatal exposure to paint thinner.

Placental transfer of toluene and xylene, the largest constituents of thinner mixture (Fifel et al., 2014), has been shown in both humans and rodents (Ghantous and Danielsson, 1986; Goodwin, 1988; Hass et al., 1995). Most studies demonstrated that toluene abuse during pregnancy can result in developmental disabilities and physical anomalies in the offspring (Hersh et al., 1985; Goodwin, 1988; Hersh, 1989; Arnold et al., 1994; Arai et al., 1997). Both clinically and in animal models, maternal exposure to solvents like toluene can lead to fetal or infant death (Wilkins-Haug and Gabow, 1991; Arnold et al., 1994) and a clear evidence of crude morphological teratogenicity was reported in surviving neonates (Ng et al., 1992; Hannigan and Bowen, 2010; Bowen, 2011). The neonates are small for gestational age and microcephalic with distinct dysmorphology (e.g., spatulate fingertips, short palpebral fissures, small face, micrognathia, low-set ears, deep-set eyes and hypoplastic fingernails; Hunter et al., 1979; Toutant and Lippmann, 1979; Hersh et al., 1985; Goodwin, 1988; Hersh, 1989; Arnold et al., 1994). With age, prenatal toluene exposure results in developmental delay and behavioral disruptions (Bukowski, 2001; Hannigan and Bowen, 2010; Bowen and Hannigan, 2013).

Almost, all the studies outlined above have evaluated teratogenic potential of a single volatile solvent, toluene or xylene. However, environmental and occupational exposure to solvents involves simultaneous exposure to multiple toxicants. Based hereupon, the aim of our study was to investigate the developmental and neurobehavioral consequences of prenatal exposure to thinner. The present study used a pattern of brief and repeated thinner inhalation during gestation in mice was designed to assess the developmental neurotoxicity of thinner. A battery of behavioral tests was performed to evaluate teratogenic effect of thinner on prenatal and neonatal growth, pre-weaning behavioral development, and adult behaviors; complying with the upcoming OECD Test Guideline for Developmental Neurotoxicity Studies (OECD Test Guideline Programme, 1996).

## MATERIALS AND METHODS

## Animals

Experiments were performed on male and female Swiss mice (8–10 weeks old, weighing 25–30 g) raised in the central animal care facilities of the Cadi Ayyad University, Marrakech (Morocco). Animals were housed in Plexiglas cages with wood chip bedding under controlled environmental conditions (12:12 light/dark cycle, 22 ± 2 ◦C), with standard diet and water ad libitum. All animal procedures were in strict accordance with the guidelines of European Council Directive (EU2010/63). All efforts were made to minimize any animal suffering and the study met the ethical standards. The study received also the approval of the Council Committee of research laboratories of the Faculty of Sciences, Cadi Ayyad University of Marrakech.

Virgin female mice (N = 21) at proestrus phase were mated with male in the ratio of 2:1 overnight (12 h). The following day, the onset of pregnancy was confirmed by the observation of a vaginal plug. Confirmation of positive sperm-plug corresponded to gestational day 0 (GD0). After mating, the males were removed from the cages and the pregnant females were then assigned to one of three experimental groups in random order.

## Paint Thinner Inhalation

We exposed mice to paint thinner (Sodecso, Mohammedia, Morocco) whose chemical composition, determined by gas chromatography and single wavelength monitoring spectrometry (Chemistry Analysis and Characterization Centre, University Cadi Ayyad) in our previous study (Fifel et al., 2014), includes more than 25 distinct molecules among which the most representative are Toluene (24.46%), Xylene (15.47%), Benzene (10.67%), Dichloromethylene (6.34%) and Acetone (5.55%).

Dams were exposed daily via inhalation to either 300 ppm (N = 8) or 600 ppm of thinner (N = 5) for the whole period of gestation. An ''air-only'' control group (N = 8) was manipulated for the same period and conditions as the thinner exposure groups, but without thinner (0 ppm). The


exposure procedure and apparatus have been described in detail previously (see Fifel et al., 2014). Briefly, a dam was placed into a whole-body inhalation chamber (27 × 17.5 × 13.5 cm) and 200 µl or 400 µl of liquid thinner was added to a filter paper located on a glass petri dish covered by a wire mesh on the inhalation chamber floor to obtain an estimated thinner concentration of 300 ppm or 600 ppm respectively. Exposure to thinner vapor occurred twice a day between (first exposure between 8:00–9:00 a.m. and second exposure 8 h later), two sessions of 15 min (filter paper and liquid thinner were renewed at the beginning of each session), separated by 5 min interval in which mice were returned to the home cage (30 min of one exposure session and 60 min of total daily exposures).

## Maternal Observation

During the whole period of gestation, pregnant mice were observed daily in order to detect any changes in behavior, symptoms of poisoning or signs of morbidity. Abortion or premature delivery and body weight were recorded daily from GD0 until delivery. In addition, different parameters of fertility and reproduction were evaluated as described by Ait-Bali et al. (2016):


## Birth Measures and Developmental Assessment of the Offspring

At the day of delivery designated as postnatal day 1 (PND1), all pups from each litter (Control: N = 22, 300 ppm-treated: N = 30, 600 ppm-treated: N = 36) were counted, sexed base on the anogenital distance, and checked for apparent morphological anomalies (e.g., facial malformations, missing digits, etc). Litter size and body weight for each pup were measured at PND1, PND7, PND14 and PND21. The appearance of physical maturation landmarks were recorded, including pinna detachment, hair appearance, incisor eruption, and eye opening. In order to not disturb and to minimize the stress on the dam and the offspring, litters were not culled, but litter with less than four pups were not included in the behavioral tests. Subsequently, pups were subjected to the following battery of pre-weaning tests (**Table 1**) in order to test reflex development and neuromotor ability (Fox, 1965; Adams, 1986; Moser, 2001). These reflexologic and behavioral tests, reactive to environmental and toxic conditions with greater reliability, were used to assess the maturation of the CNS (Vaglenova et al., 2008).

### Surface Righting Reflex Test

Each pup was placed in a supine position and the latency to get back on all four paws was calculated. Each pup received one trial and a maximum of 60 s was given in each trail. The maximum latency of 60 s was assigned to the pups that did not right.

### Cliff Avoidance Test

Each pup was placed on a table edge with forepaws and snout over the edge. The time spent to turn 180◦ away from the cliff face was measured and cliff avoidance was recorded for a maximum time of 60 s.

### Homing Test

For 30 min in one holding cage, the pups were separated from the dam and kept at a temperature of 35◦C. The litter was then transferred to a Plexiglas cage (36 × 20 cm, walls 18 cm high) containing bedding from the home cage evenly distributed on one side (14 × 20 cm, nest area) and the rest of the cage covered with clean wood shavings. Each pup was placed in the middle of the cage and video recorded for 4 min. Homing performance was scored for the time spent in the area with nest litter.

### Negative Geotaxis Test

Pups were placed on 45◦ inclined plywood surface and the time taken to turn 180◦ from a head-downward position to face-upward was calculated.

## Rotarod Test

Rotarod was used to evaluate the ability of mice to remain their balance on revolving rod (15 rpm) during 5 min trial. The apparatus consisted of horizontal textured roller of 3 cm diameter placed at a height of 28 cm with automatic fall detection. All mice underwent two trials with a 15 min inter-trial interval. The latency to fall off was measured for each session. The apparatus was cleaned after each trial. Litters remained with their dams until weaning on PND21 when all offspring were re-housed in groups of five same-sex littermates or other mice from the same prenatal thinner group. All animals were unhandled between PND24 and PND60 when the behavioral tests began.

## Adult Neurobehavioral Evaluation

All animals were tested at PND60 (**Table 1**). To assess locomotor activity and anxiety-like behaviors, we used the open field test (OFT; Wilson et al., 1976) and the elevated plus maze test (EPMT; Torres and Escarabajal, 2002; Lapiz-Bluhm et al., 2008); depressive-like state of mice was evaluated by means of the tail suspension test (TST; Steru et al., 1985) and Splash test (ST; Willner, 2005; Isingrini et al., 2010); memory retention was assessed by the step-through passive avoidance task (SPAT; Lo et al., 2009), and executive function by the Puzzle box (PB; Ben Abdallah et al., 2011). The OFT and EPMT were recorded and analyzed using Ethovision XT Noldus 8.5 video tracking program (Noldus, Netherlands) connected to a video camera (JVC). The behaviors in TST, ST, SPAT and PB were video-recorded (Samsung SCO-2080R) and measured manually using the eventrecording function in the video-tracking software (Debut video capture software, NHC). All the behavioral tests were performed between 8:00 and 12:00 a.m. to avert any circadian related fluctuation in the performance of the animals. Before each mouse was introduced, the apparatus in all behavioral tests cleaned with a 75% ethanol solution to remove any trace of odor.

## Open-Field Test

The apparatus used for this test consisted of a simple square field (50 × 50 × 50 cm). A 75W lamp was placed in porthole diffusing light and located at 200 cm from the device allowing the center of the apparatus to be under a dim light (100 lx). At the beginning of each session, mice were placed in the central part (15 cm × 15 cm) of the arena and the total distance moved, velocity, and the total time spent into center were determined over a 10 min period. The center zone is 17.5 cm from the wall of the maze, corresponding to the standard area (Park et al., 2015).

### Elevated Plus Maze Test

The maze composed by two open arms (50 × 5 cm) and two enclosed arms (50 × 5 × 15 cm connected to a common central platform (5 × 5 cm). The maze floor and the side/end walls (15 cm height) of the enclosed arms were made of clear Plexiglas. The apparatus was raised to 50 cm from the floor and was under an approximate brightness of 200 lx. Each mouse was placed in the center facing an open arm and left to explore the maze for a single 5 min recorded session. The percentage of the time spent in the open arms was analyzed by calculating the ''time spent in the open arms'' divided by the ''total time spent in both the open and enclosed arms''.

#### Tail Suspension Test

Mice were suspended from a plastic rod mounted 50 cm above the surface by fastening the tail with adhesive tape. Immobility, defined as the absence of any limb or body movements, was measured during 6 min.

### Splash Test

A 10% sucrose solution was squirted on the dorsal coat of the mouse in its home cage. Because of its viscosity, the sucrose solution dirties the mouse fur and animals initiate grooming behavior. The time spent grooming body, face and paws was recorded after applying sucrose solution for a period of 5 min.

#### Step-through Passive Avoidance Learning

The step-through inhibitory avoidance apparatus consisted of bright and dark equally sized Plexiglas compartments (28.5 × 25 × 25 cm), with independent electrical grid floor, and connected by an opening guillotine door (10 × 8.5 cm). During the training session, mice were placed individually in the light chamber. Then as soon as the animal entered the dark chamber, the door was lowered and an inescapable single electric foot shock (0.5 mA) was delivered by a shocker for 5 s. Ten seconds after exposure to the foot shock, the animal was removed from the chamber to its home cage. The retention of the avoidance performance was tested 24 h later. With access to the dark chamber, each mouse was placed into the light chamber without any shock during the test session and the latency to enter the dark compartment was calculated. The mice that did not enter the dark chamber during the cut-off time (180 s) were removed from the apparatus and assigned a ceiling score of 180 s. Short latencies indicate poor retention.

## Puzzle Box

The Puzzle box is an acrylic Plexiglas arena divided into two areas by a removable barrier: a brightly-lit (300 lx) open-field area (58 × 28 × 21 cm) and a covered goal-box area (15 × 28 × 21 cm). Mice undergo a three-day protocol, consisting of nine trials (T1–T9) with three trials per day and 1.5 h of inter-trial interval, during which they were challenged to enter into the goal zone via a narrow underpass (4 cm wide). This underpass, located under the barrier, was blocked with obstacles that are increasingly difficult to overcome as testing progress. At the beginning of each trail, mouse was placed in the open-field zone facing the goal-box, and the time taken to enter the goal zone with all four paws was recorded over a period of 3 min (training and burrowing) or 4 min (plug). Three obstruction conditions were used within this task:


The performance of mice in the puzzle box permits to test the capacities of mice to remove the offered barrier in each block. This sequence allow to asses native problem-solving ability (T5 and T8), and learning/ short-term memory (T3, T6 and T9), while the repetition on the next day is used as recall and solution retention to examine the long-term memory (T4 and T7).

## Histology and Dentate Gyrus Volume

Upon completion of behavioral testing (**Table 1**), mice were deeply anesthetized by an intraperitoneal injection of lethal dose of sodium pentobarbital (>90 mg/kg) and perfused transcardially with 0.9% saline, followed by 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). Brains were removed from the skull, post-fixed overnight in the same fixative solution, cryoprotected in a 30% sucrose solution, frozen and sectioned by using a freezing microtome (Leica Microsystems, Germany). Fortymicrometer thick free-floating coronal serial sections of the dentate gyrus (DG) were collected in multiwell dishes. Sections were mounted on gelatine-coated slides and stained with Cresyl violet.

With regards to the experimental group, all area counts were conducted blind. To estimate the volume of DG, area measurement of each DG blade were made through sections of the entire DG (positioned at anterior, medial and posterior levels) at low-magnification (×6.3) by using Olympus BH-2 microscope equipped with an Olympus DP71 camera. The reconstruction of a complete photomicrograph association of different shots and their treatment has been made through image processing software Adobe Photoshop CC 2014. The area of each section was calculated by ImageJ software and the total volume of the DG was estimated by applying the Cavalieri method (Prakash et al., 1994). For each animal, 10 DG sections (along the rostrocaudal axis) were used in each analysis.

## Statistical Analysis

Statistical analyses were performed using Prism 5.0 for Windows (GraphPad software). For statistical evaluations of the different dependent variables, one-way analysis of variance (ANOVA) and two-way ANOVA followed


Data are given as means ± SD. <sup>∗</sup>P < 0.05 and ∗∗∗P < 0.001 compared to control.

by a Bonferroni post hoc for multiple comparisons were used. A difference was considered statistically significant at P ≤ 0.05.

## RESULTS

## Effect of Thinner Exposure on Fertility and Reproduction Parameters of Pregnant Mice

Maternal data for the 21 dams in the three prenatal treatment groups are presented in **Table 2**. No clinical signs of toxicity were observed in the dams during the exposure period, except a one case of thinner induced abortion and two cases of preterm births at the higher dose of thinner (600 ppm). There were no significant differences in gestation period and gestation index between the treated groups and the control group (**Table 2**). Moreover, Delivery, live-birth, viability and lactation index of treated groups did not differ compared to the control group (**Table 2**).

Concerning the body weight of pregnant females, the statistical analysis by two-way ANOVA was performed betweensubject variables: treatment and gestation duration. This analysis showed a significant effect of treatment (F(2,72) = 15.79, P < 0.001) and gestation period (F(3,72) = 135.2, P < 0.001) on dam weights. Bonferroni post hoc analyses revealed that dams exposed to 600 ppm of thinner gained significantly less weight than the control at GD18 (t = 2.95, P < 0.05); otherwise, there was a significant less weight gain for dams in all treated groups compared to control group from GD0 to GD18 (control group vs. 300 ppm group: t = 4.61, P < 0.001; control group vs. 600 ppm group: t = 5.31, P < 0.001; 300 ppm group vs. 600 ppm group: t = 1.12, NS).

#### TABLE 3 | Litter and offspring characteristics.

## Effect of Prenatal Exposure to Thinner on Physical Development

## Morphological Evaluation

By using a comparative atlas of external malformations in laboratory animals and humans (Roux, 2003), each pup was observed for signs of malformations and abnormal morphological changes. No obvious external malformations were noted in the three prenatal treatment groups.

## Body Weights of Pups

A two-way ANOVA analysis was performed considering treatment and age as main factors. Our results indicated a significant effect of treatment, age of the pups and the interaction of these two factors on body weight (F(2,288) = 83.35, P < 0.001; F(3,288) = 378.70, P < 0.001; F(6,288) = 9.53, P < 0.001; respectively). While no significant main effects were observed for thinner prenatal exposure on PND1 body weight for all pups (P > 0.50, NS), Bonferroni post hoc analysis showed that there was a significant loss in the body weight of prenatally treated offspring compared to the control at PND7 (t = 0.67, NS; t = 5.71, P < 0.001; respectively), PND14 (t = 3.03, P < 0.01; t = 7.29, P < 0.001; respectively), and PND21 (t = 4.18, P < 0.001; t = 10.44, P < 0.001; respectively). Additionally, 600 ppm pretreated group also weighs less than 300 ppm pretreated group at PND7 (t = 5.46, P < 0.001), PND14 (t = 4.78, P < 0.001), and PND12 (t = 6.99, P < 0.001; **Table 3**).

#### Body Length of Pups

There was a significant main effect of thinner exposure on body length of pups revealed by two way ANOVA analysis with treatment and age as factors (treatment: F(2,288) = 51.52, P < 0.001; age: F(3,288) = 694.30, P < 0.001; treatment × age: F(6,288) = 7.64, P < 0.001). Post hoc analysis showed that no significant main effects were found for 300 ppm prenatal inhalation on body length compared to control from PND1 to


Data are given as means ± SD. ∗∗P < 0.01 and ∗∗∗P < 0.001 compared to control. ###P < 0.001 compared to treated group.

PND21 (**Table 3**). However, prenatal exposure to 600 ppm of thinner led to a decrease in body length at PND7 (600 ppm vs. control: t = 3.95, P < 0.001; 600 ppm vs. 300 ppm: t = 1.98, NS), PND14 (600 ppm vs. control: t = 4.28, P < 0.001; 600 ppm vs. 300 ppm: t = 4.36, P < 0.001) and PND21 (600 ppm vs. control: t = 9.16, P < 0.001; 600 ppm vs. 300 ppm: t = 8.22, P < 0.001); no significant difference was observed at PND1 (**Table 3**).

#### Body Development

As reported in **Table 3**, no significant changes in indices of postnatal maturation were noted for pinna unfolding, hair appearance, incisor eruption, or age of eyes opening for thinnertreated animals compared to control-treated animals.

## Effect of Prenatal Exposure to Thinner on Sensorimotor Development

#### Surface Righting Reflex Test

Data analysis by two-way ANOVA, with treatment and gender as main factors, revealed that surface righting time was significantly affected by treatment at PND5, PND7 and PND9 (F(2,64) = 19.87, P < 0.001; F(2,64) = 23.25, P < 0.001; F(2,64) = 61.56, P < 0.001; respectively), while the sex (F(1,64) = 1.48, P = 0.23; F(1,64) = 0.34, P = 0.56; F(1,64) = 1.07, P = 0.30; respectively) and interaction of treatment and sex (F(2,64) = 0.03, P = 0.97; F(2,64) = 0.29, P = 0.75; F(2,64) = 0.06, P = 0.95; respectively) had no effect. To analyze more specifically the effects in both sexes of the maternal exposure to thinner, we tested the male and female data separately with two-way ANOVA by using treatment and age as main factors. The statistical analysis showed a significant main effect of treatment and age as shown by the improved surface righting times of animals between PND5 and PND9 [males (F(2,114) = 47.13, P < 0.001; F(2,114) = 53.78, P < 0.001; respectively); females (F(2,78) = 45.70, P < 0.001; F(2,78) = 48.33, P < 0.001; respectively)]. Maternal exposure to 600 ppm delayed significantly the development of offspring righting reflex compared to control and 300 ppm prenatally exposed pups at PND5 [males (t = 4.28, P < 0.001; t = 3.48, P < 0.01; respectively); females (t = 4.12, P < 0.01; t = 3.96, P < 0.01; respectively)], PND7 [males (t = 3.51, P < 0.01; t = 5.21, P < 0.001; respectively); females (t = 3.52, P < 0.01; t = 4.54, P < 0.001; respectively)] and PND9 [males (t = 7.28, P < 0.001; t = 6.59, P < 0.001; respectively); females (t = 7.22, P < 0.001; t = 7.68, P < 0.001; respectively)] (**Figure 1A**).

#### Cliff Avoidance Test

A significant main effect of prenatal thinner exposure was found on neonatal reflexes, as evaluated by cliff avoidance [males (F(2,38) = 150.9, P < 0.001); females (F(2,26) = 22.94, P < 0.001)]. Six hundred parts per million pretreated males and females mice take more time to avoid the cliff than controls and 300 ppm pretreated mice at PND6 [males (t = 15.76, P < 0.001; t = 14.09, P < 0.001; respectively); females (t = 5.78, P < 0.001; t = 5.89, P < 0.001; respectively)] (**Figure 1B**). Moreover, a two-way ANOVA analysis was performed considering treatment and sex as main factors. This analysis revealed that cliff avoidance time was significantly affected by treatment, sex and the interaction of these two factors (F(2,64) = 116.2, P < 0.001; F(1,64) = 5.80, P < 0.05; F(2,64) = 3.54, P < 0.05; respectively).

### Homing Test

The results showed that the prenatally treated mice presented a retardation in general development in the homing test at PND9 [males (F(2,38) = 42.55, P < 0.001); females (F(2,26) = 14.66, P < 0.001)] (**Figure 1C**), where a decreased time spent in the nest area was observed with respect to control in both sexes [males (300 ppm vs. control: t = 7.21, P < 0.001; 600 ppm vs. control: t = 8.55, P < 0.001); females (300 ppm vs. control: t = 3.99, P < 0.01; 600 ppm vs. control: t = 5.28, P < 0.001)]. Therefore, there is a significant effect only between groups (F(2,64) = 51.91, P < 0.001) independently of sex.

#### Negative Geotaxis Test

A two way ANOVA analyses with treatment and sex as factors showed that the performance of mice was affected significantly by the experimental group independently of sex at PND10 and PND12 (F(2,64) = 150.3, P < 0.001; F(2,64) = 37.41, P < 0.001; respectively). To obtain more information about the effect of treatment with the age in each sex, we applied two-way ANOVA considering treatment and age as main factors. There was a significant overall main effect of prenatal thinner treatment, age and the interaction of these two factors in the negative geotaxis test [males (F(2,76) = 110.5, P < 0.001; F(1,76) = 34.88, P < 0.001; F(2,76) = 19.27, P < 0.001; respectively); females (F(2,52) = 64.53, P < 0.001; F(1,52) = 19.05, P < 0.001; F(2,52) = 5.53, P < 0.01; respectively)]. All prenatally 600 ppm treated pups reduced their latencies to turn 180◦ in comparison with control and 300 ppm pretreated animals at PND10 [males (t = 12.41, P < 0.001; t = 13.21, P < 0.001; respectively); females (t = 8.71, P < 0.001; t = 9.07, P < 0.001; respectively)] and PND12 [males (t = 4.36, P < 0.001; t = 6.13, P < 0.001; respectively); females (t = 4.47, P < 0.001; t = 5.75, P < 0.001; respectively)] (**Figure 1D**).

## Rotarod Test

The two-way ANOVA analysis revealed that the motor coordination was significantly affected by the trial in both sexes [males (F(1,76) = 8.86, P < 0.01); females (F(1,52) = 5.83, P < 0.05)], while the treatment [males (F(2,76) = 0.76, P = 0.47); females (F(2,52) = 2.06, P = 0.14)] and interaction of treatment and trial [males (F(2,76) = 1.38, P = 0.26); females (F(2,52) = 0.43, P = 0.65)] had no effect. At PND24, our results of Rotarod test showed that no significant differences were observed between exposed and control offspring in the fall latency at trial1 [males (300 ppm vs. control: t = 0.08, NS; 600 ppm vs. control: t = 0.91, NS); females (300 ppm vs. control: t = 1.09, NS; 600 ppm vs. control: t = 1.39, NS)] and trail2 [males (300 ppm vs. control: t = 2.31, NS; 600 ppm vs. control: t = 0.65, NS); females (300 ppm vs. control: t = 1.71, NS; 600 ppm vs. control: t = 0.75, NS)] (**Figure 1E**).

FIGURE 1 | Effect of prenatal exposure to thinner on sensory-motor development. (A) Surface righting reflex at PND5, PND7 and PND9. (B) Cliff avoidance reflex at PND6. (C) Olfactory discrimination at PND9. (D) Negative geotaxis reflex at PND10 and PND12. (E) Motor coordination at PND24. Results are presented as mean ± SEM (Males: N = 13–14, Females: N = 8–11). ∗∗P < 0.01 and ∗∗∗P < 0.001 in comparison with control group. ##P < 0.01 and ###P < 0.001 in comparison with treated group.

## Effect of Prenatal Exposure to Thinner on Adults Behaviors

At PND60, the boy weight was significantly less in female (control = 24.72 ± 1.11 g; 300 ppm = 24.40 ± 1.11 g; 600 ppm = 23.72 ± 1.12 g) than male mice (control = 29.24 ± 1.03 g; 300 ppm = 27.47 ± 1.07 g; 600 ppm = 27.78 ± 1.03 g). The two way ANOVA analysis show that the difference was only significant between sexes

(Males: N = 13–14, Females: N = 8–11). <sup>∗</sup>P < 0.05, ∗∗P < 0.01 and ∗∗∗P < 0.001 in comparison with control. ##P < 0.01 and ###P < 0.001 in comparison with

(F(1,76) = 19.349, p < 0.001), whereas the treatment and interaction sex × treatment have no effect on pups body weight (F(2,76) = 0.76; p = 0, 472; F(1,76) = 0.23; P = 0.79; respectively).

#### Locomotor Activity

treated group.

Open-field analysis quantified overall locomotor activity showed that mice exposed prenatally to thinner displayed no differences in distance moved from non-exposed controls in both sexes [males (F(2,38) = 1.12, P = 0.33); females (F(2,26) = 0.67, P = 0.52)] (**Figure 2A**). However, there was a significant main effect of prenatal thinner treatment on the speed of movements only in females [males (F(2,38) = 0.11, P = 0.89); females (F(2,26) = 6.79, P < 0.01)]; Post hoc analyses revealed that the 600 ppm dose group of females showed a significant decrease in velocity over the 10 min compared to 300 ppm dose group (t = 3.69, P < 0.01). Otherwise, the velocity did not differ between both prenatally exposed females and controls (**Figure 2B**). On the other hand, two way ANOVA analyses revealed that treatment, sex and the interaction treatment × sex had no significant effects on the speed of movements (F(2,64) = 2.22, P = 0.07; F(1,64) = 0.25, P = 0.61; F(2,64) = 0.08, NS; respectively).

#### Anxiety

Anxiety-like traits were evaluated after prenatal thinner exposure using the OF and EPMTs (**Figures 2C,D**). Thinner prenatal exposure seems to cause a significant less anxiety-like behaviors in both sexes as revealed by long time spent in center of open field [males (F(2,38) = 34.20, P < 0.001); females (F(2,26) = 9.86, P < 0.001)] (**Figure 2C**) and open arms of elevated plus maze [males (F(2,38) = 29.62, P < 0.001); females (F(2,26) = 10.72, P < 0.001)] (**Figure 2D**). Furthermore, Bonferroni post hoc analysis showed that 300 ppm and 600 ppm dose groups spent significantly more time in the center of open field [males (t = 5.95, P < 0.001; t = 7.93, P < 0.001;

suspension test. (B) Grooming time calculated during splash test. (C) Latency to enter the dark chamber calculated during step-through inhibitory avoidance test. (D,E) Latency scored to reach the goal zone of males and females during the nine trials of the Puzzle box test. Results are presented as mean ± SEM (Males: N = 13–14, Females: N = 8–11). <sup>∗</sup>P < 0.05, ∗∗P < 0.01 and ∗∗∗P < 0.001 in comparison with control group. #P < 0.05, ##P < 0.01 and ###P < 0.001 in comparison with treated group.

respectively); females (t = 3.37, P < 0.01; t = 4.29, P < 0.001; respectively)] and the open arms of elevated plus maze [males (t = 2.79, P < 0.05; t = 7.62, P < 0.001; respectively); females (t = 2.79, P < 0.05; t = 4.63, P < 0.001; respectively)] compared to control group. Moreover, a significant difference was revealed between prenatally treated groups of males in EPMT (t = 4.68, P < 0.001). In addition, the two-way ANOVA revealed that the anxiety-like behavior of mice was affected significantly by the experimental group in OFT (F(2,64) = 38.65, P < 0.001) and EPMT (F(2,64) = 36.18, P < 0.001) independently of sex.

#### Depression

Tail suspension test and splash test have been used to assess depression-like behavior after prenatal thinner exposure (**Figures 3A,B**). In tail suspension test, treated mice showed a dose dependant increase in the time of immobility compared to the control mice in both sexes [males (one-way ANOVA: F(2,38) = 64.83, P < 0.001; Bonferroni post hoc: 300 ppm vs. control, t = 3.47, P < 0.01; 600 ppm vs. control, t = 11.15, P < 0.001; 600 ppm vs. 300 ppm, t = 7.48, P < 0.001); females (one-way ANOVA: F(2,26) = 31.09, P < 0.001; Bonferroni post hoc: 300 ppm vs. control, t = 4.03, P < 0.01; 600 ppm vs. control, t = 7.87, P < 0.001; 600 ppm vs. 300 ppm, t = 4.26, P < 0.001)] (**Figure 3A**). Similarly, thinner-treated groups as compared to control group exhibited a shorter duration of grooming during splash test in both sexes [males (one-way ANOVA: F(2,38) = 60.62, P < 0.001; Bonferroni post hoc: 300 ppm vs. control, t = 6.89, P < 0.001; 600 ppm vs. control, t = 10.86, P < 0.001; 600 ppm vs. 300 ppm, t = 3.77, P < 0.01); females (one-way ANOVA: F(2,26) = 26.96, P < 0.001; Bonferroni post hoc: 300 ppm vs. control, t = 4.99, P < 0.001; 600 ppm vs. control, t = 7.27, P < 0.001; 600 ppm vs. 300 ppm, t = 2.59, P < 0.05)] (**Figure 3B**). Moreover, depression-like response was affected significantly by treatment and sex in tail suspension test (F(2,64) = 96.34, P < 0.001; F(1,64) = 8.29, P < 0.01; respectively) and splash test (F(2,64) = 84.77, P < 0.001; F(1,64) = 8.14,

P < 0.01; respectively) as revealed by two-way ANOVA analyses.

#### Learning and Memory

SGZ, subgranular zone.

We further examined mice by step-through avoidance learning and puzzle box tasks which have been recognized as useful experimental paradigms for assessing learning, memory retention and executive function (**Figures 3C–E**). In step-through inhibitory avoidance test (**Figure 3C**), there is significant effect of treatment (F(2,64) = 18.47, P < 0.001), while the sex (F(1,64) = 1.32, P = 0.25) and their interaction (F(2,64) = 0.88, P = 0.42) did not affect the latency to step-through as shown by two-way ANOVA analyses. Then, analyses of the male and female data separately with one-way ANOVA revealed that the prenatal thinner exposure affect the latency to step-through during the memory retention test session in both sexes [males (F(2,38) = 13.50, P < 0.001); females (F(2,26) = 7.99, P < 0.01)]; therefore, in 600 ppm dose groups, there was a significant decrease in step-through latency compared to control group and 300 ppm dose groups [males (t = 4.82, P < 0.001; t = 4.05, P < 0.001, respectively); females (t = 3.01, P < 0.05; t = 3.76, P < 0.01, respectively)].

**Figures 3D,E** illustrate the time taken to enter the goal box in thinner-treated and control males and females mice among all trials (T1–T9) of puzzle box. Two way ANOVA showed a main effect in both sexes of treatment [males (F(2,342) = 72.34, P < 0.001); females (F(2,234) = 40.22, P < 0.001)], trials [males (F(8,342) = 45.35, P < 0.001); females (F(8,234) = 31.82, P < 0.001)] and their interaction [males (F(16,342) = 7.48, P < 0.001); females (F(16,234) = 5.66, P < 0.001)]. Mice of all groups similarly learned the puzzle task at the first day (T1–T3) and T4 as shown by their improved latencies on the repeated trials. Further, post hoc analysis displayed that all 600 ppm pre-exposed males and females mice spend more time to solve the obstacles in comparison with control and 300 ppm pre-exposed mice at T5 [males (t = 2.88, P < 0.05; t = 2.66, P < 0.05, respectively); females (t = 2.05, NS; t = 3.53, P < 0.01, respectively)], T6 [males (t = 9.34, P < 0.001; t = 6.76, P < 0.001, respectively); females (t = 6.21, P < 0.001; t = 4.76, P < 0.001, respectively)], T7 [males (t = 18.38, P < 0.001; t = 16.50, P < 0.001, respectively); females (t = 6.66, P < 0.001; t = 5.35, P < 0.001, respectively)], T8 [males (t = 2.66, P < 0.05; t = 2.68, P < 0.05, respectively); females (t = 0.37, NS; t = 2.94, P < 0.05, respectively)] and T9 [males (t = 3.19, P < 0.01; t = 3.36, P < 0.01, respectively); females (t = 2.93, P < 0.05; t = 3.74, P < 0.01, respectively)].

## Evaluation of DG Volume after Prenatal Exposure to Thinner

Cresyl violet-stained sections at the same Bregma levels, showed an important reduction in the entire volume of the hippocampus following prenatal exposure to paint thinner. The observed changes were more prominent with 600 ppm than those observed in 300 ppm and control (**Figures 4A–C**). Indeed, the analysis of the dentate gyrus volume of the hippocampus subregions by using the Cavalieri method showed a significant main effect (F(2,6) = 18.33, P < 0.01). Moreover, statistical analysis among the different experimental groups revealed that the volume of DG is sharply reduced only in the 600 ppm pre-exposed mice in comparison with control and 300 ppm pre-exposed mice (t = 5.14, P < 0.01; t = 5.33, P < 0.01; respectively). Conversely, mice that were prenatally exposed to 300 ppm of thinner displayed no differences in the volume of DG as compared to control mice (t = 0.18, NS; **Figure 4D**).

## DISCUSSION

The rationale for this study is the raise in inhalant-abuse among pregnant women, which usually include solvents containing a mixture of volatile compounds, and environmental and occupational exposures consist mostly of simultaneous exposures to various solvents. Most published works described the teratogenicity due to industrial exposure to a single volatile solvent and the teratogenic potential of chemical mixture such as thinner remains under-investigated. Hence, the current study is the first attempt evaluated developmental and behavioral effects in offspring of mice prenatally exposed briefly and repeatedly to thinner. The present results showed that, despite one case of thinner induced abortion at the higher dose, prenatal exposure to the solvent did not cause any maternal toxicity nor decrease viability of the offspring. In contrast, there were significant effects on birth weight, litter size, sensory-motor development, recognition memory, anxiolytic- and depression-like behaviors independently of sex at adult age.

## Methodological Considerations

The present study used a pattern of brief (15 min) and repeated (two sessions of 15 min twice a day, 60 min/day in total) inhalation of different concentrations of thinner (300 or 600 ppm) during pregnancy (7 days/week for 20 days) in mice. This pattern of exposure, typically found with inhaled solvent abuse, mirrors exposure levels, administration route and exposure durations. The fetuses in this study were exposed to concentrations of thinner mimicking those attained clinically. In previous study in our lab, we have used exposure parameters identical to the present study (Fifel et al., 2014). It was reported that depending on the solvents, the concentrations used during abuse episodes in human vary considerably and involves mostly high concentrations (up to 15,000 ppm; Hathaway and Proctor, 2004; Bowen et al., 2006). However, it has been demonstrated that mean solvent concentrations in the atmosphere measured in working environments in several industries (e.g., car painting, printing, fiber glass reinforced unsaturated polyester industries, and furniture industries) are in the range of 35–210 ppm (short term measurements; Dreiem et al., 2005). This indicates that the nominal concentrations used in this study are in a range that is relevant for occupational exposure; we suggested that concentrations slightly higher than 200 ppm but largely lower than the concentrations used to mimic binge patterns will lead to teratological and developmental consequences. The most common route of exposure in the working environment is by inhalation, and since the inhalant abuse is periodic or episodic rather than continuous, repeated and brief episodes of thinner exposure (4 × 15 min/day) were used. Three reasons justify our choice of this paradigm: Firstly, the analyses by gas chromatography showed that all thinner components are evaporated during the 15 min after thinner injection. Secondly, we used a static system of solvents delivery which causes the carbon dioxide (CO2) accumulation over time (Bowen et al., 2006). Therefore, in order to prevent potential CO<sup>2</sup> poisoning, the 30 min duration of each inhalation session was divided in 2 × 15 min separated by 5 min during which the animals were returned to their home cage where they could breathe fresh air. The third reason behind using 2 × 15 min sessions is based on epidemiological data showing that, the exposure to solvents during abuse episodes in human typically lasts a few minutes (<10 min; Hathaway and Proctor, 2004; Bowen et al., 2006). In addition, the exposures to inhalants began on GD8 in several previous studies (Bowen et al., 2009; Bowen and Hannigan, 2013; Callan et al., 2015) because GD8 is post implantation and limits the risk of resorption. Since the timing of the exposure period can influence outcome, in the present animal model, the thinner exposures were performed from GD0 through GD20. This animal model, that patterns exposure to model inhalant abuse practices, will be able to assess the teratogenic potential of abused solvents with highly fidelity and sensitivity to clinical situation than other patterns.

## Effects on Reproduction Parameters

The effects of thinner exposure on offspring development and behavior occurred in the absence of any obvious maternal or fetal toxicity. Despite some cases of thinner induced abortion and preterm births at the higher dose of thinner, no significant differences were observed in fertility and reproduction parameters. Previous studies of toluene exposure in rats have produced conflicting results regarding fertility and reproduction indices (Bowen et al., 2005, 2009; Roberts et al., 2007; Bowen and Hannigan, 2013). A possible explanation for the discrepancies in the results of maternal and fetal toxicity may lie in the unique methodologies used by each experimenter with regard to routes of administration, exposure duration and concentration, stage of development during exposure, maternal age, and the species used. Clinically, several cases of inhalants-related embryopathy and malformations have been reported after solvents abuse by pregnant women (e.g., toluene, 1,1,1-trichlorethane and gasoline; Arai et al., 1997; Hannigan and Bowen, 2010; Bowen, 2011). Individual cases of perinatal death and surviving neonates, showing evidence of morphological and functional teratogenicity (termed ''Fetal Solvent Syndrome''), have been reported in children born to women exposed to very high levels of toluene during their pregnancies (Arnold et al., 1994; Pearson et al., 1994).

In the present work, the daily exposure of pregnant mice to thinner vapor during gestation reduced weight gain in dams. Moreover, although body weights and length of the thinner-prenatally treated pups did not differ from non-treated pups at birth (PND1), but showed a significant decrease in the body growth from PND1 until PND21. Several studies have suggested that the current growth restrictions could be attributed to the decreased gestational body weight and possible under-nutrition of the mother (Saillenfait et al., 2007; Bowen and Hannigan, 2013; Callan et al., 2015). Others have reported that fetal and placental weights regardless of maternal food consumption decreases after prenatal exposure to solvent (Gospe et al., 1994; Gospe and Zhou, 1998). Therefore, it is possible that inhalant exposure during pregnancy may have persistent impacts on lactation and/or maternal behavior which were insufficient for postnatal maturation of the offspring.

## Postnatal Development of Pre-Exposed Pups

Following prenatal exposure to thinner, the postnatal observations showed an overall delay in the innate reflexes While there was no significant thinner-induced shift in physical maturation as evaluated by physical landmarks (i.e., pinna detachment and incisors eruption), mice offspring exposed in utero to high concentration of thinner showed significant reduction in postnatal growth up to weaning. These deficits were seen in measures of sensory-motor development and reflex ability. Pups exposed to the higher concentration of thinner exhibited delays in the righting reflex, cliff avoidance reflex and negative geotaxis reflex relative to sham-exposed pups. These results agreed with previous findings, which reported that the prenatal exposure to xylene and toluene delayed the development of those behavioral patterns (Hass et al., 1995; Hougaard et al., 1999; Bowen and Hannigan, 2013). This alteration observed in reflexologic tests in the exposed pups may indicates important neural damage for righting such as vestibular function (Secher et al., 2006), and also deficits accompanying a mouse cerebellum developmental delay (Aruga et al., 1998). The vestibular system functions already at birth (Altman and Sudarshan, 1975), but righting is not accomplished until after maturation of various placing and supporting reactions in the motor system. It is also noted that thinner-exposed pups displayed an altered response in homing test, which considered a test incorporates more complex measures of interest in social odors, cognitive and sensorimotor abilities (Scattoni et al., 2008).

## Neurobehavioral Effects on Adult Prenatally Exposed to Thinner

In addition to any changes in the body weight in both sexes, our data relative to performance in Rotarod test, distance moved, and moving velocity in the open-field test at PND60 showed that the animals exposed to thinner in utero did not differ from controls, indicating that both motor coordination and activity were not disturbed.

Interestingly, the prenatal exposure to the thinner at the used concentrations in this study induces anxiolyticand depression-like behaviors at adulthood as assessed by different behavioral tests. These findings joined those of previous studies obtained by Bowen et al. (1996), Páez-Martínez et al. (2003) and Fifel et al. (2014) showing the anxiolytic properties of different solvents (thinner, toluene, benzene, 1,1,1-trichloroethane, diethyl ether and flurothyl) in exposed adult mice. At the molecular level, the anxiolytic effect of thinner is not an unexpected finding and it could be related to the ability of its components to act as positive modulators of GABAA receptors, like other CNS depressant molecules (Bale et al., 2005; Williams et al., 2005).

The results derived from the effects of prenatal thinner exposure on the offspring cognitive functions highlight impairments in executive function and memory retention at adult age, as revealed by the Puzzle box and the step-through inhibitory avoidance tests. The Puzzle box test, considered a highly reliable test of higher-order cognitive functioning (Ben Abdallah et al., 2011), resulted in thinner 600 ppm-exposed mice exhibiting prolonged latencies to reach the goal zone relative to control mice. Moreover, the exposure in utero to 600 ppm of thinner resulted in decreased latency to enter the dark chamber calculated during step-through inhibitory avoidance test. No differences were seen between thinner 300 ppm-exposed mice and their control counterparts in the latencies of both behavioral tests. These observations are compatible with findings by Hass et al. (1995, 1999) who showed that the mnemonic processes are particularly vulnerable to xylene and toluene effects. Additionally, similar results have been demonstrated by other experimental tests (i.e., the object recognition test and the Morris water maze task), which involve short-term memory or spatial learning (Fifel et al., 2014; López-Rubalcava et al., 2014; Callan et al., 2015). No mechanistic studies have been designed to find out the molecular targets implicated in cognitive impairments, but changes in the expression of NMDA receptor (subunits NR1 and NR2 mRNA), NMDA receptor antagonism, and inhibition of neurogenesis in the hippocampus could be associated with the impaired memory observed in mice exposed to thinner during gestation (Seo et al., 2010; Huerta-Rivas et al., 2012; Win-Shwe and Fujimaki, 2012).

As the DG subregion of the hippocampus is a substrate for both cognition and mood regulation (Sahay et al., 2011), it was important to evaluate whether the prenatal exposure to thinner led to any morphological alterations in this area. Histological evidence and volumetric finding indicate that thinner may interfere with the development of the hippocampus, demonstrating by the smaller volume of the granule cell layer of the hippocampal dentate gyrus in animals exposed prenatally to 600 ppm of thinner. The finding of reduced DG volume after prenatal exposure to thinner is in accord with a study on the effects of postnatal exposure to 100 and 500 ppm toluene on the developing hippocampus in rats (12 h/day, day 1–28 postnatally; Slomianka et al., 1990). The results of those authors, demonstrated that in addition of the smaller volume observed within the area dentate of exposed animals (granule cell layer, hilus, and the commissural-associational zone of the dentate molecular layer), argyrophilic cells and pronounced granule cell degeneration were found in the granule cell layer of animals exposed to the higher dose of toluene. The effects of early developmental thinner exposure on DG volume may be secondary to a number of indirect effects, including disrupted synaptogenesis (Lin et al., 2009), altered expression of neurotrophic factors (e.g., nerve growth factor and brain-derived neurotrophic factor), increases in NMDA receptors (Lee et al., 2005), pro-inflammatory cytokines and glial markers (e.g., glial fibrillary acidic protein; Win-Shwe et al., 2012). While the behavioral alterations seen in prenatally exposed mice are consistent with hippocampal involvement, causal relationship, however, remains to be elucidated.

In summary, our experiments have provided several main findings to support our hypothesis that the exposure in utero to paint thinner results in behavioral, functional, and structural deficits. These alterations

## REFERENCES


may be irreversible as they were apparent at adult age. Real life always involves simultaneous inhalations of multiple solvents, indicating the need for further studies, with combinations of substances, on the outcomes and mechanisms of thinner-induced developmental and behavioral effects.

## AUTHOR CONTRIBUTIONS

HM, MB, FMM and SB designed the experiments; HM and FMM performed the experiments; HM, MB, FMM and SB performed the analysis of the data; HM and FMM assembled the figures. HM, SB, MB and FMM wrote and edited the manuscript. All authors validated it.

## ACKNOWLEDGMENTS

This research was supported by NEUREN Project (PIRSES-GA-2012-318997). We thank A. REGRAGUI for his support by providing us experimental animals.


**Conflict of Interest Statement**: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Malloul, Mahdani, Bennis and Ba-M'hamed. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

digital media

of impactful research

article's readership