# NON-MOTOR SYMPTOMS IN PRIMARY MOTOR NEUROLOGICAL DISORDERS: FROM MOLECULAR PATHWAYS TO CLINICAL AND THERAPEUTIC IMPLICATIONS

EDITED BY : Francesca Trojsi, Kaylena Anastasia Ehgoetz Martens, Foteini Christidi, Rosa De Micco and Cristina Moglia PUBLISHED IN : Frontiers in Neuroscience, Frontiers in Psychiatry and Frontiers in Neurology

#### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88963-389-0 DOI 10.3389/978-2-88963-389-0

#### 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

# NON-MOTOR SYMPTOMS IN PRIMARY MOTOR NEUROLOGICAL DISORDERS: FROM MOLECULAR PATHWAYS TO CLINICAL AND THERAPEUTIC IMPLICATIONS

Topic Editors:

Francesca Trojsi, University of Campania Luigi Vanvitelli, Italy Kaylena Anastasia Ehgoetz Martens, University of Waterloo, Canada Foteini Christidi, National and Kapodistrian University of Athens, Greece Rosa De Micco, Second University of Naples, Italy Cristina Moglia, University of Turin, Italy

Citation: Trojsi, F., Ehgoetz Martens, K. A., Christidi, F., De Micco, R., Moglia, C., eds. (2020). Non-Motor Symptoms in Primary Motor Neurological Disorders: from Molecular Pathways to Clinical and Therapeutic Implications. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-389-0

# Table of Contents

*04 Editorial: Non-motor Symptoms in Primary Motor Neurological Disorders: From Molecular Pathways to Clinical and Therapeutic Implications*

Foteini Christidi, Rosa De Micco, Kaylena A. Ehgoetz Martens, Cristina Moglia and Francesca Trojsi

*07 EEG Microstates Change in Response to Increase in Dopaminergic Stimulation in Typical Parkinson's Disease Patients* J. Ignacio Serrano, María Dolores del Castillo, Verónica Cortés,

Nuno Mendes, Aida Arroyo, Jorge Andreo, Eduardo Rocon, María del Valle, Jaime Herreros and Juan Pablo Romero

*16 Psychiatric Symptoms in Amyotrophic Lateral Sclerosis: Beyond a Motor Neuron Disorder*

Elisabetta Zucchi, Nicola Ticozzi and Jessica Mandrioli

*27 Notch Signalling in the Hippocampus of Patients With Motor Neuron Disease*

Ulises Gómez-Pinedo, Lucía Galán, Jordi A. Matías-Guiu, Vanesa Pytel, Teresa Moreno, Antonio Guerrero-Sola and Jorge Matías-Guiu

# *40 Sleep Disorders in Huntington's Disease*

Radoslawa Herzog–Krzywoszanska and Lukasz Krzywoszanski

#### *49 Schizophrenia and Hereditary Polyneuropathy:* PMP22 *Deletion as a Common Pathophysiological Link?*

Dominique Endres, Simon J. Maier, Christiane Ziegler, Kathrin Nickel, Anne N. Riering, Benjamin Berger, Johann Lambeck, Miriam Fritz, Birgitta Gläser, Friedrich Stock, Michael Dacko, Thomas Lange, Irina Mader, Katharina Domschke and Ludger Tebartz van Elst

#### *57 Cognitive Syndromes and C9orf72 Mutation are not Related to Cerebellar Degeneration in Amyotrophic Lateral Sclerosis*

Monica Consonni, Eleonora Dalla Bella, Anna Nigri, Chiara Pinardi, Greta Demichelis, Luca Porcu, Cinzia Gellera, Viviana Pensato, Stefano F. Cappa, Maria Grazia Bruzzone, Giuseppe Lauria and Stefania Ferraro

*64 Metabolic and Stress Response Changes Precede Disease Onset in the Spinal Cord of Mutant SOD1 ALS Mice*

Gavin Pharaoh, Kavithalakshmi Sataranatarajan, Kaitlyn Street, Shauna Hill, Jake Gregston, Bumsoo Ahn, Caroline Kinter, Michael Kinter and Holly Van Remmen

#### *83 The Gut and Parkinson's Disease—A Bidirectional Pathway* Susanne Fonseca Santos, Hadassa Loth de Oliveira, Elizabeth Sumi Yamada, Bianca Cruz Neves and Antonio Pereira Jr.

# *91 The Peripheral Nervous System in Amyotrophic Lateral Sclerosis: Opportunities for Translational Research* Francesco Gentile, Stefania Scarlino, Yuri Matteo Falzone, Christian Lunetta, Lucio Tremolizzo, Angelo Quattrini and Nilo Riva

*107 Cognitive, Emotional and Psychological Manifestations in Amyotrophic Lateral Sclerosis at Baseline and Overtime: A Review*

Soumia Benbrika, Béatrice Desgranges, Francis Eustache and Fausto Viader

# Editorial: Non-motor Symptoms in Primary Motor Neurological Disorders: From Molecular Pathways to Clinical and Therapeutic Implications

Foteini Christidi <sup>1</sup> , Rosa De Micco<sup>2</sup> , Kaylena A. Ehgoetz Martens <sup>3</sup> , Cristina Moglia<sup>4</sup> and Francesca Trojsi <sup>2</sup> \*

*<sup>1</sup> First Department of Neurology, Aeginition Hospital, Medical School, National and Kapodistrian University of Athens, Athens, Greece, <sup>2</sup> First Division of Neurology, Department of Advanced Medical and Surgical Sciences, University of Campania "Luigi Vanvitelli", Naples, Italy, <sup>3</sup> Department of Kinesiology, University of Waterloo, Waterloo, ON, Canada, <sup>4</sup> ALS Center, "Rita Levi Montalcini" Department of Neuroscience, University of Turin, Turin, Italy*

Keywords: neurodegenerative diseases, movement disorders, motor neuron disease, non-motor symptoms, C9orf72 hexanucleotide repeat expansion, neuropsychiatric disorders, sleep disorders, levodopa response

**Editorial on the Research Topic**

#### **Non-motor Symptoms in Primary Motor Neurological Disorders: From Molecular Pathways to Clinical and Therapeutic Implications**

Edited and reviewed by:

*Einar M. Sigurdsson, New York University, United States*

\*Correspondence: *Francesca Trojsi francesca.trojsi@unicampania.it*

#### Specialty section:

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

Received: *27 October 2019* Accepted: *18 November 2019* Published: *03 December 2019*

#### Citation:

*Christidi F, De Micco R, Ehgoetz Martens KA, Moglia C and Trojsi F (2019) Editorial: Non-motor Symptoms in Primary Motor Neurological Disorders: From Molecular Pathways to Clinical and Therapeutic Implications. Front. Neurosci. 13:1296. doi: 10.3389/fnins.2019.01296*

There is increasing evidence that neurodegenerative diseases imply high emotional and economic burden (GBD 2016 Neurology Collaborators, 2019), substantially influenced by a variety of nonmotor symptoms (NMS), such as gastrointestinal-, autonomic-, neuropsychiatric-, and sleep disorders, preceding classical motor signs or appearing during the disease course. A convincing explanation of these symptoms would be much better supported by the recently postulated "disease-spreading hypothesis" (Brundin et al., 2010), in that the pathology would not be limited to the initially affected cell populations, but disease would spread involving other non-motor regions, in the brain and beyond (Braak et al., 2003; Brettschneider et al., 2013). The 10 articles contributing to this Frontiers Research Topic will provide readers with an update on some of the most crucial aspects of NMS in neurodegenerative disorders, addressing both molecular and clinical issues.

Firstly, with regard to the use of animal models for investigating the pathogenesis of familial amyotrophic lateral sclerosis (fALS), Pharaoh et al. measured changes in metabolic pathways in spinal cords of the SOD1G93A mouse model of ALS using a targeted proteomic analysis. The protein content of metabolic proteins, including those involved in glycolysis, β-oxidation, and mitochondrial metabolism, was found altered in SOD1G93A mouse spinal cord before disease onset, recalling some hypotheses on the potential role of metabolism and nutrition in ALS pathogenesis (Gallo et al., 2013; Wills et al., 2014).

With regard to clinical studies, assuming the ALS as an emblematic neurodegenerative disease in which both motor and extra-motor symptoms may coexist from early stages, the neuropsychological profile described in ALS patients has been deeply reviewed by Benbrika et al. to provide readers with a clear picture of all the cognitive, emotional and psychological manifestations of ALS. Neuropsychological assessment needs the use of ALS-specific tools, such as the Edinburgh Cognitive and Behavioral ALS screen (ECAS) (Abrahams et al., 2014). Moreover, the increased understanding of the disease spectrum including ALS and frontotemporal dementia (FTD)

**4**

(Burrell et al., 2016) has led researchers to elaborate on a variety of symptoms not classically considered part of the ALS clinical picture, such as psychiatric symptoms, reviewed by Zucchi et al.. In particular, the link between ALS and schizophrenia was further supported by a substantial genetic correlation, only partially explained by pleiotropic gene variants, such as C9orf72 (McLaughlin et al., 2017).

ALS patients carrying the C9orf72 hexanucleotide repeat expansion (C9+ALS) had significantly more co-morbid behavioral variant FTD features than those without (Byrne et al., 2012) and increasing evidence has suggested that hippocampus and subcortical region degeneration also plays a role in the peculiar clinical picture of C9+ALS patients (Agosta et al., 2017). Furthermore, with regard to hippocampal involvement in sporadic ALS, Gómez-Pinedo et al. explored the expression of the intracellular pathway of Notch proteins, probably associated with amyloid precursor protein (APP) signaling pathways (Ables et al., 2011), by analyzing hippocampal samples from the autopsies of 12 patients with ALS or ALS-FTD. They reported lower levels of the Notch intracellular domain in patients with ALS than in controls, leading to increases in amyloid-β and to decrease in hippocampal neurogenesis.

Among the subcortical structures potentially involved in the cognitive status of ALS patients, Consonni et al. tested the relationship between cerebellar degeneration, cognitive syndromes, and C9orf72 mutation in ALS patients, revealing that cerebellar involvement may reflect a signature of ALS-FTD (Bede et al., 2013) other than a signature of the C9orf72 hexanucleotide repeat.

In light of the recent observation of a widespread extramotor involvement in ALS, that has implied that, beyond motor neurons, peripheral axons and nerve terminals may also be affected as early events, Gentile et al. reviewed the most relevant literature regarding common molecular pathways (i.e., impairments of axonal transport, RNA metabolism, and proteostasis) shared by ALS and peripheral axonal neuropathies. They underlined that an extensive evaluation of the molecular events occurring in the peripheral nervous system could be fundamental to understand the pathogenic mechanisms of ALS as well as other neuropsychiatric disorders. In this regard, an interesting link between hereditary polyneuropathies and psychiatric disorders, such as schizophrenia, has been discussed by Endres et al., who presented the case of a patient with schizophrenia and comorbid hereditary neuropathy with liability to pressure palsy (HNPP), due to a deletion of the peripheral myelin protein 22 gene (PMP22). This potential association has been explained by the role of PMP22, mainly expressed in the peripheral nervous system, although its mRNA has also been detected in the brain (Chanson et al., 2013). In particular,

#### REFERENCES

Ables, J. L., Breunig, J. J., Eisch, A. J., and Rakic, P. (2011). Not(ch) just development: Notch signalling in the adult brain. Nat. Rev. Neurosci. 12, 269–283. doi: 10.1038/nrn3024

PMP22 seems to play an important role in regulating cell growth and myelination (Sanahuja et al., 2005), also impaired in schizophrenia.

A widespread appearance of pathologic proteins aggregates in both the central and peripheral nervous systems, including the enteric nervous system (ENS), has been recognized also in Parkinson's disease (PD) (Goedert, 2001). In particular, Fonseca Santos et al. focused on the hypothesis of a "gut-brain axis": gut toxins could induce the formation of α-synuclein aggregates in the ENS, which may then be transmitted in a prion-like manner to the central nervous system through the vagus nerve. From the therapeutic point of view, levodopa is actually considered the best current symptomatic treatment for PD, although characterizing levodopa response may be a challenge in early stages. In this regard, Serrano et al. presented the analysis of electroencephalography microstates (EEG-MS) default-mode network changes in response to dopaminergic stimulation as a potential tool to evaluate, in a non-invasive way, the levodopa response and to assess the suitability of the patients' medication dosage.

Finally, among other common NMS, Herzog-Krzywoszanska and Krzywoszanski addressed the role of sleep abnormalities in Huntington's disease (HD). Among the major criticisms raised, the authors underlined that many medications administered to HD patients to alleviate motor and psychiatric symptoms may change sleep architecture, thereby negatively impacting sleep quality (Arnulf et al., 2008). Furthermore, more detailed knowledge of these problems in HD can also provide more profound insight into the nature of the neurodegenerative process underlying the disease (Morton, 2013).

In conclusion, the substantial message of this Frontiers Research Topic is that the identification of extra-motor abnormalities may represent a core feature for supporting the diagnosis and predicting the prognosis of many primary motor neurodegenerative disorders and for shedding light on several dysfunctional pathways in order to prompt the development of combined therapies with synergistic neuroprotective effects on several neurodegenerative pathomechanisms.

# AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

#### ACKNOWLEDGMENTS

We are deeply grateful to all authors who contributed their valuable articles and expert reviewers for their significant contribution.

Abrahams, S., Newton, J., Niven, E., Foley, J., and Bak, T. (2014). Screening for cognition and behaviour changes in ALS. Amyotroph. Lateral Scler. Frontotemporal Degener. 15, 9–14. doi: 10.3109/21678421.2013.805784

Agosta, F., Ferraro, P. M., Riva, N., Spinelli, E. G., Domi, T., Carrera, P., et al. (2017). Structural and functional brain signatures of C9orf72 in motor neuron disease. Neurobiol. Aging 57, 206–219. doi: 10.1016/j.neurobiolaging.2017.05.024


**Conflict of Interest:** 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 © 2019 Christidi, De Micco, Ehgoetz Martens, Moglia and Trojsi. 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.

# EEG Microstates Change in Response to Increase in Dopaminergic Stimulation in Typical Parkinson's Disease Patients

J. Ignacio Serrano<sup>1</sup> , María Dolores del Castillo<sup>1</sup> , Verónica Cortés<sup>2</sup> , Nuno Mendes<sup>3</sup> , Aida Arroyo<sup>2</sup> , Jorge Andreo<sup>2</sup> , Eduardo Rocon<sup>1</sup> , María del Valle<sup>4</sup> , Jaime Herreros<sup>5</sup> and Juan Pablo Romero1,6 \*

<sup>1</sup> Neural and Cognitive Engineering Group, Centre for Automation and Robotics, Spanish National Research Council – Technical University of Madrid, Madrid, Spain, <sup>2</sup> Faculty of Experimental Sciences, Francisco de Vitoria University, Madrid, Spain, <sup>3</sup> Faculty of Sciences, University of Lisbon, Lisbon, Portugal, <sup>4</sup> Department of Neurology, Fuenlabrada University Hospital, Madrid, Spain, <sup>5</sup> Department of Neurology, Infanta Leonor University Hospital, Madrid, Spain, <sup>6</sup> Brain Damage Unit, Hospital Beata Maria Ana, Madrid, Spain

#### Edited by:

Foteini Christidi, National and Kapodistrian University of Athens, Greece

#### Reviewed by:

Gabriella Santangelo, Università degli Studi della Campania Luigi Vanvitelli, Italy Ales Holobar, University of Maribor, Slovenia

> \*Correspondence: Juan Pablo Romero p.romero.prof@ufv.es

#### Specialty section:

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

Received: 16 June 2018 Accepted: 19 September 2018 Published: 15 October 2018

#### Citation:

Serrano JI, del Castillo MD, Cortés V, Mendes N, Arroyo A, Andreo J, Rocon E, del Valle M, Herreros J and Romero JP (2018) EEG Microstates Change in Response to Increase in Dopaminergic Stimulation in Typical Parkinson's Disease Patients. Front. Neurosci. 12:714. doi: 10.3389/fnins.2018.00714 Objectives: Characterizing pharmacological response in Parkinson's Disease (PD) patients may be a challenge in early stages but gives valuable clues for diagnosis. Neurotropic drugs may modulate Electroencephalography (EEG) microstates (MS). We investigated EEG-MS default-mode network changes in response to dopaminergic stimulation in PD.

Methods: Fourteen PD subjects in HY stage III or less were included, and twentyone healthy controls. All patients were receiving dopaminergic stimulation with levodopa or dopaminergic agonists. Resting EEG activity was recorded before the first daily PD medication dose and 1 h after drug intake resting EEG activity was again recorded. Time and frequency variables for each MS were calculated.

Results: Parkinson's disease subjects MS A duration decreases after levodopa intake, MS B appears more often than before levodopa intake. MS E was not present, but MS G was. There were no significant differences between control subjects and patients after medication intake.

Conclusion: Clinical response to dopaminergic drugs in PD is characterized by clear changes in MS profile.

Significance: This work demonstrates that there are clear EEG MS markers of PD dopaminergic stimulation state. The characterization of the disease and its response to dopaminergic medication may be of help for early therapeutic diagnosis.

Keywords: Parkinson's disease, electroencephalography, microstates, levodopa, diagnosis

# INTRODUCTION

Parkinson's Disease (PD) is a neurodegenerative disease affecting up to 3% of the population ≥ 65 years of age (Poewe et al., 2017). PD has been associated with several risk factors common to other age related diseases and some chemicals exposure (Beitz, 2014) but its ultimate cause is still unknown.

In typical PD, progressive degeneration of dopaminergic neurons in the substantia nigra is correlated with the wide known motor symptoms of bradykinesia, rigidity, and tremor (Kalia and Lang, 2015). However, the phenotypical profile of each patient gives rise to the identification of several subtypes such as tremordominant subtype and on the other hand bradykinesia/rigidity dominant (Marras and Lang, 2013). Dopaminergic treatment usually provides substantial alleviation in motor symptoms. Some other symptoms such as gait disturbance and postural instability do not usually have a substantial improvement (Hely et al., 2005). There are several syndromes that share some features similar to Parkinson's disease, but the progression and onset of symptoms are different. One of the main characteristics of this atypical Parkinsonism is the lack of or incomplete response to levodopa (Goetz et al., 2008; Stamelou and Hoeglinger, 2013).

Levodopa is actually considered the best current symptomatic treatment for Parkinson's Disease. One of the main obstacles for the treatment of the disease is its pharmacodynamics, meaning there is a low penetrance of the drug into the central nervous system (Khor and Hsu, 2007; Britz et al., 2010; Musso et al., 2010; Van de Ville et al., 2010; Michel and Koenig, 2017). The action of this drug determines important changes not only in motor but also in non-motor symptoms. The state where patients show a marked improvement is called "ON state" and the one with no effects is called "OFF state." Dopamine (DA) receptor agonists are also used to treat the symptoms of the disease since such drugs mimic the action of dopamine, their action is achieved by stimulation of pre-synaptic (auto receptors) and post-synaptic DA receptors (Radad et al., 2005). Levodopa equivalent dose (LED) can be calculated from dopamine receptor agonist doses so the total daily levodopa administration can be estimated (Tomlinson et al., 2010).

Dopaminergic stimulation is surely alleviating typical PD symptoms in most patients, but the degree of its effects shows major inter-individual differences. These differences in levodopa motor response are evident even between same disease-severitystage patients (Goetz et al., 2000). Every patient has their own needs of medication to reach their "ON state" and this varies according to the disease progression (Nyholm et al., 2012) and degree of denervation (Kostrzewa et al., 2005).

The clinical diagnosis of PD is currently based on clinical symptoms and other support criteria such as response to medication. This clinical diagnosis is very difficult, especially in early stages of the disease when there are no remarkable motor features.

Identification of neurophysiological variables with diagnostic value in early-stage PD would raise the chances of improving diagnostic certainty (Valls-Solé and Valldeoriola, 2002).

Electroencephalography (EEG) is a well-known technique used to record the electrical field produced by the electrical activity in the brain. This technique is characterized by a high temporal resolution and high test–retest reliability (Lopes da Silva, 1991). It has been published that quantifying EEG rhythms and their variations could be the source of biomarker for several neuropsychiatric disorders, such as schizophrenia, major depressive disorder, or even neurodegenerative diseases as Alzheimer's disease (Gandal et al., 2012; Han et al., 2013).

Electroencephalography data can be analyzed according to momentary states of the topographical brain activation, called microstates (MS). "Microstate analysis is a method in which states are defined by topographies of electrical potentials on a set of multichannel electrodes that remain stable for 80–120 ms before rapidly moving to a different microstate" (Khanna et al., 2015). Unlike other EEG processing techniques, in microstates, the simultaneous analysis of the signals from all the electrodes is used to create a global representation of a functional state. In fact, many studies have shown that time series of EEG microstates vary through behavioral states (Stevens and Kircher, 1998; Lehmann et al., 2010), personality types (Schlegel et al., 2012) and neuropsychiatric disorders (Dierks et al., 1997; Lehmann et al., 2005; Kikuchi et al., 2011; Khanna et al., 2015). Consequently, changes in the duration or frequency of appearance of specific microstates can be considered as biomarkers for different neurological and neuropsychiatric conditions.

Interestingly, several studies that have examined resting-state EEG report the same four archetypal microstates that explain most of the global topographic variance (Koenig et al., 2002; Khanna et al., 2015). The four canonical EEG microstates (A, B, C, D) seem to represent the neurophysiological correlates of four known Resting State Networks identified by fMRI, suggesting that Resting State Networks of fMRI may be the same ones that give rise to microstates (Michel and Koenig, 2017). The dynamics of these networks may imply various brain functions, and their alteration can be associated with the pathophysiology of several neurological and neuropsychiatric diseases (Khanna et al., 2015).

When the microstate time series is convolved with the restingstate fMRI BOLD signal, each microstate map correlates with the activity of particular RSNs (Britz et al., 2010; Musso et al., 2010; Yuan et al., 2016). Britz et al. (2010) showed that microstate A is associated with the phonological processing network, B with the visual network, C with the salience network, and D related to the attentional network.

There is evidence that neurotropic drugs may modulate EEG microstates (Kikuchi et al., 2007, 2011; Yoshimura et al., 2007), but there are no studies showing EEG microstate changes in response to the administration of levodopa or dopaminergic agonist drugs in typical PD patients.

The aim of the study presented in this paper is to identify EEG microstate changes that characterize levodopa response. The data obtained from this study can be used to support typical PD diagnosis in difficult clinical scenarios where a therapeutic trial of levodopa is not feasible or not well tolerated by the patient because of the gastric effects of its administration.

# MATERIALS AND METHODS

#### Participants

The protocol was approved by the CEIC Fuenlabrada Hospital, Madrid, Spain. All subjects gave written informed consent in accordance with the Declaration of Helsinki.

Fourteen patients were included in the study after signing informed consent forms (4 female: mean age 66.25 ± 12.9, range 52–80 and 10 male: mean age 66.9 ± 7.41, range 50–76).

All the patients had been diagnosed with Parkinson's Disease according to London Brain Bank criteria (mean time from onset 7.29 ± 2.33 years, range 4–13), with Hoehn and Yahr scale 2 ± 0.8 (range 1–3), and were taking levodopa or dopaminergic agonists (mean daily amount 324.28 ± 232.77 levodopa milligrams equivalent dose (LED), range 100–733) in stable dosing regimen for at least 90 days with a clear ON effect (good clinical effect). There was not head tremor in any of the patients. **Table 1** shows a description of included patients. In addition, twenty-one healthy subjects were recruited as control participants (6 female: mean age 67.4 ± 10.21, range 50–77 and 15 male: mean age 69.6 ± 10.14, range 50–93).

#### Intervention

Participants were asked to come to the hospital early in the morning without their corresponding daily levodopa or agonist intake (at least 8 h after the last levodopa or dopamine agonist dose). Resting EEG activity was recorded over 2 min by 64 electrodes placed according to the 10–20 system as depicted in **Figure 1**. They were comfortably seated with their hands on their laps, relaxed jaw and eyes open, looking at a white wall. Immediately afterward, the EEG electrodes were removed, and they took their daily Levodopa or agonist dose with a glass of water, 30 min before they had a light breakfast and were given free time. The resting EEG activity was analogously recorded 1 h after the levodopa intake, once the patient had asserted they were in their usual ON state.

#### Materials

An actiCHamp amplifier (Brain Vision LLC, NC, United States) was used to amplify and digitize the EEG data at a sampling frequency of 512 Hz. The EEG data were stored in a PC running Windows 7 (Microsoft Corporation, Washington, DC, United States). EEG activity was recorded from 64 positions with active Ag/AgCl scalp electrodes (actiCAP electrodes, Brain Vision LLC, NC, United States). The ground and reference electrodes were placed on AFz and on FCz, respectively (see **Figure 1**).

Electroencephalography acquisition was carried out by NeuroRT Studio software (Mensia Technologies SA, Paris, France). The EEG signal processing procedure was performed using MATLAB functions (MathWorks Inc., Natick MA, United States), specifically the EEGLab toolbox (Delorme and Makeig, 2004). EEG microstates were extracted and characterized by LORETA-KEY v20170220 software (the Key Institute for Brain-Mind Research, Zurich, Switzerland). Statistical analyses were performed by SPSS for Windows, version 23.0 (IBM Inc., Chicago, IL, United States).

# EEG Processing and Outcome Measurements

The continuous EEG signal for each channel was artifactcorrected by the Artifact Subspace Reconstruction algorithm (Mullen et al., 2013), disabling all parameters except the high-pass filter band width (0.25–0.75) and burst repairing (kurtosis > 5). The signal was then band-pass filtered between 2 and 31 Hz with a Finite Impulse Response (FIR) filter (order 846). Finally, a common average reference (CAR) spatial filter was applied.

The processed EEG was the input to an EEG microstate detection and characterization algorithm (Pascual-Marqui et al., 2014). The algorithm requires an initial interval for the number of microstates searched. This interval was set to 4–10. The algorithm was run independently for PD PRE, PD POST and CONTROL conditions. Mean microstate topographies in each condition were manually assigned to canonical microstates reported in previous studies. The assignment was individually performed by three judges (among the authors) by visual analysis.


<sup>∗</sup>UPDRS, Unified Parkinson's Disease Rating Scale. <sup>1</sup>Pramipexole, <sup>2</sup>Rasagiline, <sup>3</sup>Benserazide, <sup>4</sup>Ropirinole, <sup>5</sup>Rotigotine, <sup>6</sup>Safinamide, <sup>7</sup>Amantadine.

This procedure avoids the likely mislabeling introduced by the common topography correlation analysis in the presence of irregular topographies and more than four microstates (Custo et al., 2017). Labels agreed by two or more judges were assigned (all mean microstate topographies labels were agreed by two judges at least). From this algorithm, the microstates accounting for most of the variance were selected. For each microstate, the percentage of the total time in the microstate (coverage), the percentage of times entered in the microstate (frequency), the number of times entered in the microstate (occurrence) and the average duration of the microstate are calculated. In addition, the frequency and probability of change from each microstate to each other one are also calculated, giving a total of 6 features for each microstate.

# Statistical Analysis

The microstate features mentioned above are compared between the pre- and post-levodopa intake conditions. The difference of averages between pre- and post-conditions for each feature was checked by a t-test for repeated measurements with bootstrapping (n = 1000). Differences with a significance p < 0.05 and confidence intervals (lower and upper) with the same sign were considered as statistically significant.

# RESULTS

**Figure 2** shows the microstates topographies found in pre- (first column) and post- (second column) levodopa intake conditions. Control group (third column) is examined with no medication.

In the pre-condition (OFF state), canonical A, B and C microstates were found with a percentage of explained variance of 19.55, 20.34 and 18.55%, respectively. Canonical microstate D was not identified. Microstates B and C presented altered patterns with respect to the findings in the literature (Michel and Koenig, 2017). In the absence of levodopa intake, a microstate E was also

TABLE 2 | Statistically significant differences in features of Electroencephalography (EEG) microstates between pre- and post-levodopa intake in Parkinson's patients.


TABLE 3 | Statistically significant differences in features of EEG microstates between post-levodopa intake of Parkinson's patients and control participants.


EV, Explained variance; SE, Standard error; p, significance value; CI, Confidence interval.

found, in congruence with the definition of Custo et al. (2017), with an explained variance of 19.58%.

After levodopa intake (ON state), the four canonical microstates A, B, C, and D were found with normal patterns (Michel and Koenig, 2017), and explained variances of 20.12, 18.39, 20.00, and 16.67%, respectively. Microstate E was no longer present after levodopa intake. However, the microstate G, according to Custo et al. (2017), was found with an explained variance of 15.39%. Control participants presented the same latter microstates (**Figure 2**, third column), with A, B, C, D, and G percentages of global explained variance of 20.85, 22.51, 17.49, 7.06, and 12.42%, respectively.

Given that the two patient conditions only share three microstates (A, B, and C), **Table 2** presents the statistically significant differences in the features of those three microstates between the OFF and ON states. The remaining features not present in **Table 2** did not show significant differences.

According to the results in **Table 2**, the microstate A shows a decreased duration after levodopa intake. Moreover, microstate B appears more often than before levodopa intake. Finally, the frequency of transition from microstate B to A got decreased with levodopa.

**Table 3** presents the statistically significant differences in microstate features between control participants and patients after levodopa intake. No significant differences in microstate features were found between control participants and patients before levodopa intake. All microstate types except B occurred more often in post-levodopa patients than in control participants. This difference is especially high for microstate D, which also presented a higher explained variance and coverage. The probability of shifting from all microstates to D was also higher in post-levodopa patients, as is justified by the mentioned increased occurrence of the latter.

# DISCUSSION

The fact that control subjects and PD patients after taking levodopa show the same microstate types is supported by the fact that levodopa is known to restore altered motor and non-motor functions in PD patients.

A higher duration of microstate A has been related to clinical variables such as disability and cognitive fatigue in patients with multiple sclerosis (Gschwind et al., 2016), and could be related to the cognitive fatigue presented by patients with PD. Such fatigue decreases with dopaminergic stimulation, and therefore it can lead to a decrease in the duration of microstate A, as observed in the results obtained in our study. As we said earlier that Britz et al. (2010) showed that microstate B is associated with the visual network, then the observed increase in the number of times microstate B is present after levodopa or dopaminergic agonist intake might be explained by a lower

fatigability in visual monitoring and a better functioning of PD patients thanks to the medication (Lou, 2009, 2015), reflecting the tendency of visual network generators to be active in the absence of cognitive fatigue (Milz et al., 2017). Cognitive fatigability is most likely associated with neurotransmitter (dopaminergic, cholinergic, and noradrenergic) abnormalities in PD. Levodopa may be effective in treating fatigue and fatigability (Lou, 2015). The decrease of frequency of transition from microstate B to A with dopaminergic stimulation does not seem to be related to any known clinical or behavioral condition. Therefore, more research is needed on the importance and functional correlation of the transition of microstates.

There are no differences in characteristics for the microstates shared by patients before medication intake and the controls. However, after levodopa intake there are differences, mainly in the microstate D. There are studies that demonstrate reduced duration (Kikuchi et al., 2007; Nishida et al., 2013) and also a lower frequency of appearance (Lehmann et al., 2010) of microstate D in patients with schizophrenia. Schizophrenia is believed to have a dopaminergic deficit up to a certain point that could explain this common finding in PD (Van Den Brink et al., 2018). Consequently, it is to be expected that patients diagnosed with PD have a lower frequency of appearance of microstate D before taking the medication. Patients presented a greater frequency and duration of microstate D after increasing dopaminergic stimulation, as a consequence of taking the medication. This increase is even greater than in healthy controls, which, assuming that microstate D reflects dopaminergic activity, could be a result of an acute increase of this activity in the brain.

Regarding topographic considerations, microstate D is mainly due to the activation of the right inferior parietal (BA40), the right middle and superior frontal gyri and the right insula (BA13) (Custo et al., 2017). The right inferior parietal area is related to executive control and vision-guided movements (Lasaponara et al., 2018) and the insula has a direct relationship with motor planning (Beurze et al., 2007). The right middle and superior frontal areas seem to explain the changes related to the improvement in attention (Angelidis et al., 2018). The appearance of microstate D, after taking medication, is congruent and consistent with the disappearance of certain motor and non-motor symptoms after levodopa.

Apart from the four canonical microstates (A–D) (Custo et al., 2017), two additional microstates (E and G) were identified. The microstate E corresponds to the activation of the middle frontal gyrus, the dorsal part of the anterior cingulate, the cuneus and the thalamus. Dopamine has an inhibitory effect (D2 receptors). Consequently, it is plausible to attribute a relative hyperactivity in its absence to the thalamus, and that this relative hyperactivity disappears after the intake of levodopa. Besides, according to Yoo et al. (2015), the anterior cingulate and frontal areas correspond to the presence of non-motor symptoms. Therefore, this justifies the presence of microstate E in patients before dopaminergic stimulation and its disappearance after they took levodopa (Lou, 2009, 2015). Finally, the cuneus is also related with oculomotor control (Darby et al., 1996), which is well-known to also be a function specifically regulated by the basal ganglia, whose function is altered in PD (Pretegiani and Optican, 2017).

The microstate G corresponds to the activation of the right inferior parietal lobe, extending to the superior temporal gyrus and also the cerebellum (Khanna et al., 2015). Both areas are closely related to motor behavior (Pirondini et al., 2017). Therefore, the appearance of microstate G after levodopa intake is strongly consistent with the improvement of motor symptoms. Since the visible and clinically evaluable motor symptoms disappear with the medication intake, this results in the observation of the microstate G.

The present study is not without limitations. First, the sample size is relatively small. A larger population might have yielded more significant results. Second, the sample population is heterogeneous in medication terms, with different types and doses of drugs, although they are all in their optimal ON state. Heterogeneity in medication is usual in PD patient cohorts because such variations correspond to the different treatment strategies that can be initiated even in the same disease stage. Dopaminergic agonists play a key role in actual treatment of the disease and their diversity makes them comparable only by their conversion to dopaminergic equivalents as we did in our study. Besides, cognitive fatigue, related to microstates A and B, was not assessed in this study. Cognitive fatigability and cognitive fatigue are usually evaluated through self-reporting scales. Given that our measures were performed in sequential OFF and ON state (in less than 2 h), we considered that this evaluation would have had a very important bias of motor and emotional symptoms as product of dopaminergic deprivation. In addition, dopaminergic therapy optimization is one of the main management recommendations for treating fatigue in PD (Kostic, 2016 ´ ). Finally, no cognitive evaluation was performed, specially attention changes that are highly related to microstate D. Nevertheless, cognitive symptoms improvement, included attention span, has been widely reported in PD patients in response to dopaminergic treatment and can be assumed as a well stablished effect of medication. These limitations should be taken into consideration when interpreting the conclusions of the study.

# CONCLUSION

Electroencephalography microstate analysis can be performed by means of an economical and minimally invasive technique with high temporal resolution. Since the EEG microstate correlation with RSNs has been evidenced, the results obtained from microstate analysis have been interpreted based on the known findings about these networks. Our work has demonstrated that there is an alteration of EEG microstate features and occurrence in typical PD in response to levodopa administration. These changes correlate with known clinical effects of the substance on such patients and are coherent with related changes in RSNs.

In spite of the differences between controls and PD patients, the microstates found in patients after levodopa intake are closer to controls' microstates than before taking the medication. Thus, the analysis performed in this study can be considered as a means to assess the suitability of the patients' medication dosage.

Further, not every Parkinsonian patient has a good response to levodopa or dopamine agonist treatment, and patients who were non-respondent to levodopa are excluded from typical PD diagnosis. Absence of complete clinical response to Levodopa is common on atypical parkinsonian patients. This lack of effect of dopaminergic stimulation is considered a red flag and implies the exclusion for typical Parkinson's Disease diagnosis (Postuma et al., 2015). Considering our results, we would not expect a microstate "normalization" in atypical PD patients in response to dopamine administration. Consequently, the microstate analysis can be considered of great utility to characterize the levodopa response prior to making a diagnosis of typical vs. atypical Parkinsonism in a non-invasive way suitable for outpatients. Nevertheless, further studies are required to characterize EEG microstate changes due to levodopa administration on atypical PD patients.

# AUTHOR CONTRIBUTIONS

VC, JS, MDC, and JR contributed conception and design of the study. AA, JA, NM, JH, MDV, and JR organized the database. JS, MDC, and JR performed the statistical analysis. VC and JR wrote

#### REFERENCES


the first draft of the manuscript. VC, JS, MDC, ER, and JR wrote sections of the manuscript. All authors contributed to manuscript revision, read and approved the submitted version.

# FUNDING

This research was supported mainly by FEDER funds. JR was supported by the Ministry of Economy and Competitiveness (Grant DPI2015-68664-C4-3-R (MINECO/FEDER), NeuroMOD Therapy development and evaluation of motor and cognitive impact for Parkinson's disease rehabilitation). ER, MDC, and JS were supported by the Ministry of Economy and Competitiveness (Grant DPI2015-68664- C4-1-R (MINECO/FEDER), NeuroMOD and by MD project (PIE-201650E055).

#### ACKNOWLEDGMENTS

We would like to thank Lucía Esteban and Paz Fernández for their assistance with the project.

unified parkinson's disease rating scale (MDS-UPDRS): scale presentation and clinimetric testing results. Mov. Disord. 23, 2129–2170. doi: 10.1002/mds.22340


and inhibition of sensory processing in left spatial neglect. J. Neurosci. 38, 3792–3808. doi: 10.1523/JNEUROSCI.2817-17.2018


**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 Serrano, del Castillo, Cortés, Mendes, Arroyo, Andreo, Rocon, del Valle, Herreros and Romero. 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.

# Psychiatric Symptoms in Amyotrophic Lateral Sclerosis: Beyond a Motor Neuron Disorder

Elisabetta Zucchi<sup>1</sup> , Nicola Ticozzi2,3 and Jessica Mandrioli<sup>4</sup> \*

<sup>1</sup> Department of Neuroscience, Azienda Ospedaliero Universitaria di Modena, University of Modena and Reggio Emilia, Modena, Italy, <sup>2</sup> Department of Neurology and Laboratory of Neuroscience, Istituto Auxologico Italiano, Istituto di Ricovero e Cura a Carattere Scientifico, Milan, Italy, <sup>3</sup> Department of Pathophysiology and Transplantation, 'Dino Ferrari' Center, Università degli Studi di Milano, Milan, Italy, <sup>4</sup> Department of Neuroscience, Azienda Ospedaliera Universitaria Modena, St. Agostino- Estense Hospital, Modena, Italy

The historical view that Amyotrophic Lateral Sclerosis (ALS) as a pure motor disorder has been increasingly challenged by the discovery of cognitive and behavioral changes in the spectrum of Frontotemporal Dementia (FTD). Less recognized and still significant comorbidities that ALS patients may present are prior or concomitant psychiatric illness, such as psychosis and schizophrenia, or mood disorders. These non-motor symptoms disturbances have a close time relationship with disease onset, may constitute part of a larger framework of network disruption in motor neuron disorders, and may impact ALS patients and families, with regards to ethical choices and end-of-life decisions. This review aims at identifying the most common psychiatric alterations related to ALS and its prognosis, looking at a common genetic background and shared structural brain pathology.

#### Edited by:

Foteini Christidi, National and Kapodistrian University of Athens Medical School, Greece

#### Reviewed by:

Giorgia Querin, INSERM U1146 Laboratoire d'Imagerie Biomédicale, France Panagiotis Ferentinos, National and Kapodistrian University of Athens, Greece

#### \*Correspondence:

Jessica Mandrioli mandrioli.jessica@aou.mo.it

#### Specialty section:

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

Received: 14 December 2018 Accepted: 14 February 2019 Published: 11 March 2019

#### Citation:

Zucchi E, Ticozzi N and Mandrioli J (2019) Psychiatric Symptoms in Amyotrophic Lateral Sclerosis: Beyond a Motor Neuron Disorder. Front. Neurosci. 13:175. doi: 10.3389/fnins.2019.00175 Keywords: amyotrophic lateral sclerosis, frontotemporal dementia, psychosis, depression, c9orf72, psychiatric symptoms and disorders

# INTRODUCTION: AMYOTROPHIC LATERAL SCLEROSIS AND ITS PHENOTYPES

Amyotrophic Lateral Sclerosis (ALS) has traditionally been defined since the first reports as a disorder characterized by progressive degeneration of upper and lower motor neurons (UMN and LMN, respectively), leading invariably to paralysis of voluntary muscles, with a variable proportion of spasticity and atrophy. Despite a uniformly fatal outcome, extreme variability exists within ALS, with heterogeneity of initial presentation, spreading of disease, progression, and survival (Brown and Al-Chalabi, 2017; Hardiman et al., 2017). The observation of distinct patterns within ALS clinical variability has led to the recognition of homogeneous phenotypic subgroups. A first classification system is based on the differential involvement of upper and lower motor neuron, with primary lateral sclerosis (PLS) and progressing muscular atrophy (PMA) representing the extremes of the spectrum. The type of involvement of different body regions at onset is another common identifier, with bulbar patients constituting 25% of the total ALS population (Chiò et al., 2011). This phenotype is more consistently associated with cognitive alterations and displays decreased survival compared to the classic spinal-onset, "Charcot-type" phenotype (Chiò et al., 2011; Talman et al., 2016). A prevalent LMN involvement define flail arm and flail leg variants (Wijesekera et al., 2009), whereas a prevalence of UMN signs with spasticity, increased and pathological reflexes and pseudobulbar affect, identify the UMN-predominant phenotype (UMN-p); these phenotypes

display a relatively long survival (Sabatelli et al., 2008; Chiò et al., 2011). Lastly, the respiratory phenotype, characterized by an early and prominent involvement of the respiratory muscles, is characterized by the worst prognosis (Shoesmith et al., 2007). Moreover, while the majority of patients report a pattern of spreading of the disease from one limb to the contralateral, as by means of contiguity in cortical representations, there is a substantial fraction of ALS population describing a close sequential involvement of two distal sites (Walhout et al., 2017).

The clinical heterogeneity of ALS is reflected at molecular level in many ways. First of all, up to 20% of ALS patients show familiality for the disease, most typically transmitted by an autosomal dominant pattern (Gibson et al., 2014; Ryan et al., 2018). Among familial ALS (fALS), two thirds of the cases can be explained by pathogenic mutations in the C9orf72, SOD1, TARDBP, and FUS genes (Zou et al., 2017; Chiò et al., 2014), which also occur in 10–15% of sporadic ALS (sALS) (Renton et al., 2014). The relative contribution of each gene mutation to the epidemiology of the disease differs according to the population origin, with C9orf72 repeat expansion representing the most frequent alteration in ALS patients of European descent, while SOD1 mutations dominate the genetic landscape of ALS in China, Korea, and Japan (Zou et al., 2017). Overall, variable penetrance, genetic pleiotropy and the finding of double pathogenic mutations in the same patient point to an oligogenic mode of inheritance for many cases (Van Blitterswijk et al., 2012). At the histopathological level, diverse pathological signatures correspond to this fragmented genetic scenario. In the majority of sALS, as well as nearly half of frontotemporal dementia (FTD) cases, ubiquitinated and phosphorylated cytosolic TDP-43 aggregates are found in the frontal cortex (Neumann et al., 2006; Braak et al., 2010), whereas motor neurons of ALS patients harboring mutations in SOD1 or FUS display either neurofilamentous hyaline conglomerate inclusions and aggregates of misfolded SOD1 or cytoplasmic inclusions immunoreactive for FUS (Shibata et al., 1996; Kwiatkowski et al., 2009). C9orf72 associated diseases are characterized by TDP-43 pathology, but also by the presence of repeat-containing RNA (Al-Sarraj et al., 2011; Ash et al., 2013). These expanded C9orf72 RNAs form nuclear foci and can sequester various RNA-binding proteins, indirectly impairing their function on nucleic acid life cycle (Gendron et al., 2013; Zu et al., 2013). In addition, C9orf72 repeat expansions produce, via non-canonical Repeat-Associated Non-ATG (RAN) translation, several dipeptide repeat proteins (DPRs) that are highly aggregation-prone, thus compromising proteostasis (Mori et al., 2013; Kumar et al., 2017). Moreover, the human C9orf72 protein has been recently shown to play a role in endosomal degradation and lysosomal homeostasis and to target stress granules (SGs) to autophagy for clearance, acting in concert with SQSTM1 (Chitiprolu et al., 2018; Corrionero and Horvitz, 2018).

The majority of ALS-associated mutations displays an extreme variability in clinical manifestations, which may present asALSplus phenotypes in the same individuals (e.g., ALS and parkinsonism) and/or different clinical pictures in the carriers belonging to the same family (e.g., ALS, FTD, or both in C9orf72 families).

All these recent achievements in the understanding of the disease pathogenesis led to the general consensus that ALS is a multisystem disorder in which the clinical, pathological and genetic features largely overcome the boundaries of a pure motor neuron involvement.

# ALS AS AN EXTRA-MOTOR DISORDER WITH COGNITIVE INVOLVEMENT

The common notion of ALS as a disease affecting exclusively motor neurons has been initially cast into doubt by the early clinical observations of an association with FTD. In particular, clinicians observed some degrees of motor neuron diseases (MND) in FTD patients, and conversely, signs of cognitive and behavioral changes in ALS patients (Talbot et al., 1995; Neary et al., 2000).

FTD is the second most common form of early onset dementia, characteristically presenting in the fifth–sixth decade. The term FTD is used as an umbrella which encompasses a variety of clinical subtypes defined by clinical and pathological consensus criteria (Neary et al., 2000; Gorno-Tempini et al., 2011; Rascovsky et al., 2011). FTD can present as two main types, primarily affecting behavior (behavioral variant FTD, bvFTD) or language (primary progressive aphasia, PPA), the latter of which can be further divided in semantic variant (SD), nonfluent agrammatic variant (PNFA), and logopenic variant (lvPPA) (Chare et al., 2014; Finger, 2016). Although these subtypes can have very distinct neuroanatomical substrates, with time patients with bvFTD develop impairment in language functions and vice versa. Notably, ALS is most typically associated with behavioral FTD, whereas PPA variants with MND are rare (Saxon et al., 2017). In general, nearly 15% ALS patients satisfy diagnostic criteria for FTD (Raaphorst et al., 2012a,b; Phukan et al., 2012), constituting the syndrome of ALS-FTD, while larger fractions of ALS patients exhibit mild to moderate behavioral (ALSbi) and/or cognitive deficits (ALS-eci if executive impairment is present; ALS-neci if other intellectual functions are affected). An impairment of executive functions and verbal fluency has been found in 34–55% of ALS patients (Murphy et al., 2007a,b, 2016; Lillo and Hodges, 2009; Phukan et al., 2012; van Es et al., 2017), while behavioral disturbances have been observed in 14–40% of cases (Witgert et al., 2010; Phukan et al., 2012; Abrahams et al., 2014). Even more subtle cognitive and behavioral changes can be detected by recently validated batteries especially designed for screening ALS patients (Strong et al., 2017).

On the other side, almost 15% of bvFTD patients develop ALS during the course of disease, while signs of motor neuron impairment are observed in 40% of cases (Burrell et al., 2011; Bang et al., 2015). In conclusion, ALS and FTD can be regarded as the extremes of a disease continuum sharing some common histopathological and genetic background, which reflects a much extensive involvement of the sole motor neuron pathology.

The increased understanding of this diseases spectrum, has led researchers to study a variety of symptoms not classically considered part of the ALS clinical picture, the main ones being sensory and coordination impairment, pain and autonomic

involvement, sleep alteration and sphincter abnormalities. More recently, psychiatric symptoms have gained attention from several points of view: their increased presence preceding or following ALS onset, their relationship with FTD, familiality, prognosis, and treatment options.

In this review we aim to examine and elaborate on the reported aspects of psychiatric features in relation to the ALS spectrum.

## PSYCHIATRIC DISTURBANCES IN ALS AND FTD

Early clinical observations reported several cases in which psychiatric illnesses such as schizophrenia co-occurred in ALS patients, raising the hypothesis of a common genetic background (Howland, 1990). More recently, register-based nationwide studies have proven an higher occurrence of psychiatric illnesses both before and after ALS diagnosis. In particular, the presence of depression, neurotic disorders and history of drug abuse or dependence, was associated to an increased odds ratio (OR) for ALS; in-depth analysis revealed that a diagnosis of schizophrenia may also represent a risk factor for ALS (OR 5.0) (Turner et al., 2016). Moreover, the risk of presenting depression, a neurotic or stress-related disorder following the diagnosis appeared to be increased as well (Longinetti et al., 2017).

Along with these findings, family members of ALS patients, especially children, showed increased risk for manifesting psychiatric disturbances both before and after their relative's diagnosis (Longinetti et al., 2017). Further strengthening this link, aggregation studies suggest neuropsychiatric illnesses and ALS cluster in families. In a population-based cohort study the relative risk of developing a neuropsychiatric condition such as schizophrenia or psychosis was significantly higher in first or second degree relatives of ALS patients (Byrne et al., 2013; O'Brien et al., 2017). Whether this can be explained by genetic pleiotropy of few variants into several kindreds or by a shared polygenic risk between psychiatric diseases and ALS spectrum remains to be determined (O'Brien et al., 2017).

Moving to FTD, psychosis is a recognized symptom, affecting 32% of patients in the largest autoptically confirmed case series, though psychiatric disturbances are not included in the diagnostic criteria (Landqvist Waldö et al., 2015). Prevalence of hallucinations in FTD cohorts varies considerably, with auditory being the most common, and delusions affect one quarter of the patients (Hall and Finger, 2015). When retrospectively evaluating clinical features in a FTD cohort that later evolved to motor neuron disorder, the presence of delusions was the best predictor of such progression, with a hazard ratio of 4.4 (Lillo et al., 2010).

#### Psychosis and Schizophrenia

Even before the discovery of a genetic overlap between schizophrenia and ALS, a relation between the two diseases was already suggested by early historical studies (Meltzer and Crayton, 1974; Howland, 1990; Larner, 2008). Moreover, disturbances in motor neuron function both at central (Goode and Manning, 1988) and peripheral level (Crayton et al., 1977; Crayton and Meltzer, 1979) exist in schizophrenia. Populationbased studies have long corroborated the relation between single psychotic events, as well as schizophrenia, and ALS. In particular, an increased risk of hospitalization for schizophrenia could be observed in the 5 years preceding ALS, with higher statistical significance especially 1 year before onset of motor symptoms (Turner et al., 2016; Longinetti et al., 2017). This close relationship between psychotic features and motor symptoms in ALS may underlie the prodromal nature of these extra-motor symptoms in the framework of ALS pathogenesis (Turner et al., 2016). The link between ALS and schizophrenia was further supported by a large genome-wide association study which found a substantial genetic correlation, only partially explained by pleiotropic gene variants such as c9orf72 (McLaughlin et al., 2017). As previously highlighted, increased risk for schizophrenia and single psychotic episodes is observed among kindreds of c9orf72 carriers (Devenney et al., 2018). Moreover, psychosis was the presenting symptoms in 38% of c9orf72 carriers in a FTD-motor neuron disorder cohort, with florid psychotic symptoms such as delusional psychosis, somatoform psychosis or paranoid schizophrenia, and frontotemporal atrophy or hypoperfusion were noted on neuroimaging (Snowden et al., 2012). Delusions and hallucinations in this cohort of patients were reported to be mainly negative in nature, not related to their personal life experience (Snowden et al., 2012). Single psychotic episodes were also observed in elderly patients carrying c9orf72 expansion (Devenney et al., 2018). Overall, late-onset psychosis should always raise concern for familiality with motor neuron disease and thus warrant genetic testing for c9orf72 repeat expansion (Sommerlad et al., 2014). Furthermore, Snowden and colleagues noted that, though similar in appearance, in the case of c9orf72 expansion carriers bizarre behaviors and complex motor stereotypes had a distinct trait compared to those of other FTD and FTD-MND patients. According to the authors, it might therefore be hypothesized that this background activity of delusional thinking guides and reinforces the behavioral aberrancies typical of symptomatic FTD (Snowden et al., 2012).

Some of the cognitive changes associated to ALS, such as sensory behavioral disturbances, which were found in more than half of a large ALS cohort (Gibbons et al., 2008), have also been implicated in schizophrenia network dysfunction, with hyperactivation of secondary somatosensory cortex (Rains et al., 2012) and failed integration of sensory signaling (Carter et al., 2017). Regarding the pathophysiology of schizophrenia, a plethora of putative mechanisms have been implicated so far, mostly involving cerebral metabolic abnormalities in the pre-frontal cortex, anterior cingulate, caudate nucleus, basal ganglia, thalamus, and the cerebellum (Gross-Isseroff et al., 2003). Moreover, disruption of cortical inhibitory circuits by a reduction of overall GABAergic transmission has been advocated in schizophrenia by neurophysiological studies, mainly using transcranial magnetic stimulation (Fitzgerald et al., 2002), which demonstrated a reduction of long-interval cortical inhibition (LICI) especially in prefrontal cortex (Radhu et al., 2015). This not only relates to, similarly, enhanced cortical excitability in motor neuron disease (Geevasinga et al., 2016), but also to likewise affected tracts, with prefrontal cortex involvement in ALS patients showing verbal fluency, attention and executive function impairment (Lomen-Hoerth et al., 2003; Meier et al., 2010).

# Depression and Anxiety

fnins-13-00175 March 7, 2019 Time: 16:58 # 4

Exogenous depression in ALS can be partly justified by the dismal prognosis of such a diagnosis, with patients experiencing everyday continuous motor decay. Nevertheless, literature show contrasting results in terms of the prevalence of depressive disorders in ALS, partly explained by the different testing scales employed, partly by the cross-sectional or prospective nature of these studies, and partly by the emotional adjustment or concomitant cognitive symptoms affecting ALS patients. The prevalence of depression in different studies thus ranges from impressively high values as 48–75% (McElhiney et al., 2009; Körner et al., 2015; Wei et al., 2016) to as low as 0.9–12% (Ferentinos et al., 2011; Lulé et al., 2012; Rabkin et al., 2015). In a large observational study, 17% of ALS patients were diagnosed with a major depressive disorder, and more than half of them were on antidepressant medications (Thakore and Pioro, 2016). There is evidence of an increased risk of depression prior to motor symptoms in ALS patients, particularly evident 1 year before disease onset, suggesting that the mood disorder is part of the prodromal cascade (Roos et al., 2016; Turner et al., 2016; Longinetti et al., 2017). Importantly, patients receiving a diagnosis of depression have a 3.6 OR of developing ALS compared to controls within 1 year (Roos et al., 2016). Likewise, a diagnosis of depression is more probable after ALS onset, in particular within 1 year from the appearance of motor symptoms (Roos et al., 2016; Turner et al., 2016). The same increased prevalence before and after ALS onset is true also for anxiety symptoms and neuroticism (Longinetti et al., 2017). Familial history of suicide was extremely overrepresented in ALS kindreds (Byrne et al., 2013; O'Brien et al., 2017).

On the other hand, many of the behavioral alterations demonstrated in ALS patients may be confused with depressive symptoms, such as apathy, which is found in 31–88% of patients (Witgert et al., 2010; Lillo et al., 2011), self-centredness, blunting of primary emotions, and lack of concern for personal hygiene (Gibbons et al., 2008). Interestingly, some studies failed to relate these symptoms to measures of depression (Grossman et al., 2007), arguing that this may reflect a behavioral disturbance due to the intrinsic ALS-FTD pathological continuum, rather than being secondary to the mood disorder. Another potential confounder of depressive symptoms in ALS is pseudobulbar affect (PBA), a neurobehavioral phenomenon manifesting with pathological overwhelming laughter or crying which are either incongruent with or excessive for the context. However, it has been demonstrated that crying-predominant PBA is associated with depression, while laughter-predominant is not (Thakore and Pioro, 2016). This finding may be due to mutual contributions from these conditions, with depression presenting as crying in the setting of PBA, and crying from PBA reinforcing the feeling of sadness and underlying depression. Further studies will be needed to determine whether these emotional phenomena in ALS are related to a shared pathological mechanism with depression.

# Autism, Obsessiveness and Other Psychiatric Disorders Associated With ALS

Among other psychiatric disorders associated with ALS, Turner et al. found increased rate of bipolar disorder, with a relative risk of 3.2 to develop ALS within 1 year from hospitalization (Turner et al., 2016). This was later confirmed by a large registry-based study in which bipolar, neurotic and stress-related disorders, as well as a history of drug abuse/dependence, represented risk factors for subsequently developing ALS (Longinetti et al., 2017). Clustering of autism spectrum disorders has been observed within ALS and c9orf72 positive ALS-FTD kindreds (O'Brien et al., 2017; Devenney et al., 2018). Lack of empathy is commonly reported among the cognitive deficits of ALS patients (Cerami et al., 2014). Other abnormalities typically associated to autism such as stereotypical behaviors, social cognition impairment, obsessive-compulsive traits and mental rigidity have been reported in patients and corresponds to the clinical ALS-FTD continuum (Gibbons et al., 2008; Lillo et al., 2010; Mioshi et al., 2014). Theory of Mind (ToM) refers to the ability to infer mental states of oneself and others such as beliefs, emotions, intentions, and desires, thus allowing for an understanding of other people's behavior; these capacities are typically compromised in autistic patients (Hoogenhout and Malcolm-Smith, 2017). ToM processes have been further subdivided in cognitive and affective components. In ALS, 36% of patients displayed impairment in cognitive abilities, whereas 27% were dysfunctional in the affective ToM (van der Hulst et al., 2015).

Medial and orbitolateral prefrontal cortices have been involved in ToM capacity (Gallagher and Frith, 2003; Mitchell et al., 2006; Völlm et al., 2006), and several studies have largely demonstrated dysfunctional networks in these areas (Meier et al., 2010; Trojsi et al., 2017). Overall, a selective neurochemical or neuroanatomical network disruption may lie beneath ALS and autism mediated by unknown pathological mechanisms that warrant further research.

# HISTOPATHOLOGICAL SIGNATURE OF THE ALS-FTD SPECTRUM AND HINTS FOR CORRELATIONS TO PSYCHIATRIC SYMPTOMS

The most common histopathological feature in ALS is represented by TDP-43 inclusions in motor neurons, either large and round (Lewy bodies-like) or skein-like (Braak et al., 2010). TDP-43 can be mislocalized within neuronal cytoplasmatic inclusion (NCIs) or dystrophic neurites (DN), and is enriched in post-translational modifications such as ubiquitination and phosphorylation (Tan et al., 2013).

Larger works proved TDP-43 pathology to be present in about half of all FTD cases (Davidson et al., 2007; Mann and Snowden, 2017), whereas the remnant 45% is represented by protein tau, and less than 5% by FUS or other aggregate-prone proteins (Mann and Snowden, 2017).

Interestingly, based on morphology and distribution of the inclusions, TDP-43 pathology in ALS and FTD can be subclassified into four types (A, B, C, D), the first two of them displaying round intracytoplasmatic aggregates (Tan et al., 2013). Type B is the most typically observed inclusion pattern in MND, while in FTD it is observed only in patients showing concomitant motor neuron involvement (Burrell et al., 2016).

Propagation of TDP-43 pathology from its core anatomical substrate, i.e., the motor cortex, was shown in ALS from a large cross-sectional autopsy study in which four stages of progression were identified. In stage 1, TDP-43 proteinopathy can be observed in the granular motor neocortex, alpha-motoneurons of the ventral horn, and bulbar motor neurons of cranial nerves. Stage 2 is characterized by involvement of reticular formation and precerebellar nuclei. In stage 3 TDP-43 inclusions are present in the prefrontal neocortex (firstly, gyrus rectus and orbital gyri, and, secondly, sensory areas and temporal neocortical area) and basal ganglia (striatum and inferior colliculus). In the final stage (4), anteromedial areas of the temporal lobe and the hippocampal formation display signs of pathology (Brettschneider et al., 2013).

A similar mechanism of spreading has been observed for bvFTD, where involvement of the orbital gyri, gyrus rectus, and amygdala characterizes the cases at the very initial phase (stage I). At an increasing burden of disease, the middle frontal and anterior cingulate gyrus as well as anteromedial temporal lobe areas, superior and medial temporal gyri, striatum, red nucleus, thalamus, and precerebellar nuclei are involved (stage II). More advanced phases of disease are characterized by motor cortex, bulbar somatomotor neurons, and spinal cord anterior horn propagation (stage III), and ultimately, visual cortex is affected (stage IV) (Brettschneider et al., 2014).

With regard to psychosis and schizophrenia, though they are considered more as diseases of connectivity and abnormal neurochemical transmission, some histopathological studies have found small but significant areas of atrophy in hippocampus, prefrontal and superior temporal cortex, and thalamus. This is accompanied by hemispheric asymmetry, decreased cortical thickness and gyrification, and abnormalities in hippocampal shape. Moreover, an early neurodevelopmental anomaly in schizophreniamay be postulated since the discovery of abnormally placed and clustered neurons in lamina II of entorhinal cortex or in the neocortex (Harrison and Weinberger, 2005). The finding of a decreased number of dendrites and arborization at hippocampal and neocortical level further support the view of reduced or aberrant wiring in schizophrenic patients, while the major neurochemical findings at cortical levels are represented by reduced number of serotoninergic (5-HT2A) and muscarinic (M1) receptors in patients with schizophrenia (Dean et al., 2016).

A clear neuropathological hallmark similar to the ALS/FTD spectrum does not exist for schizophrenia; nevertheless, in an autopsy case series on schizophrenic patients, tau-positive glial tangles were found in the dorsal aspect of temporal horn, arcuate fibers in gyri of frontal cortex, and within parahippocampal gyrus, while neurofibrillary tangles were observed in transentorhinal cortex, entorhinal region, subiculum and anterior hippocampus, in almost one third of cases, with increasing prevalence in elderly patients (Casanova et al., 2002). Altogether, these data may point to a restricted limbic tauopathy in presenile or senile psychotic patients, with no evidence of progression unlike Alzheimer disease. Though it may not be excluded that aberrancies in tau metabolism are due to neuroleptic drugs (Wisniewski et al., 1994), it is reasonable to hypothesize neurodegenerative processes occur in schizophrenia as well, given the accelerated aging and overall atrophy resulting in severe cognitive decline and motor abnormalities from duration of untreated psychosis (Anderson et al., 2014).

A further neuropathological link between psychosis and ALS may be found in microglia activation.

In fact, ALS arises in part by non-cell-autonomous mechanisms, from a combination of damage within MNs and their glial partners (Boillée et al., 2006). During the disease course, microglia switches from a neuroprotective M2 phenotype to an activated M1 phenotype which secretes proinflammatory interleukines, cytokines and neurotoxic factors, leading to the progression of neuronal injury (Henkel et al., 2013). Similarly, in schizophrenic patients' autopsies, increased markers of microglia activation were observed in the prefrontal cortex, anterior cingulate and temporal cortex (Radewicz et al., 2000).

## NEUROIMAGING ACROSS PSYCHIATRIC AND BEHAVIORAL SYMPTOMS IN ALS

Neuroimaging studies have been crucial to better investigate functional and structural alterations in ALS patients showing psychiatric symptoms. By studying brain volume in a cohort of bvFTD and FTD-MND patients, a precise network of cortical and subcortical areas could be identified in patients with psychotic symptoms, which display bilateral medial prefrontal and occipital cortices, right thalamus and left cerebellum atrophy (Devenney et al., 2016). Sub-analysis within c9orf72 expansion carriers with psychosis prior to FTD or FTD-MND revealed that, besides presenting higher psychotic index, more extensive network disruption occurred, with volume reduction of bilateral medial frontal cortex, anterior cingulate and orbitofrontal cortex, bilateral insula, caudate, putamen and thalamic nuclei, middle, inferior and superior temporal gyrus, temporal fusiform gyrus, lateral occipital cortex and right cerebellum (Devenney et al., 2016). These areas roughly corresponds to those with the highest degree of atrophy in schizophrenic and schizoaffective patients (Amann et al., 2016). Moreover, anterior cingulate cortex and insula are strongly connected in the salience network, whose main function is to detect, analyze and integrate emotionally salient stimuli with respect to the internal environment, and which is involved in symptom generation in both FTD and schizophrenia (Seeley et al., 2007; Zhou and Seeley, 2014). In addition to schizophrenia, late-onset obsessive compulsive disorder in the setting of an upper motor neuron disease with concomitant FTD presents with bilateral hippocampal atrophy with sclerosis of right hippocampus on MRI and moderate right temporal cortex thinning at PET imaging (Bersano et al., 2018).

As already mentioned, abnormal behaviors are found in ALS patients along the FTD spectrum. Among these, apathy is one of the most commonly reported, and is correlated with

cortical thickness reduction in bilateral orbitofrontal lobe and left precentral gyrus. On the other hand, a hostile, disinhibited pattern of personality as identified by PCA analysis is more associated to thinning of temporal and cingular regions of the right hemisphere (Consonni et al., 2018).

# A COMMON GENETIC BACKGROUND

Genome wide association studies (GWAS) allowed for exploring the genetic relationship between schizophrenia and ALS through SNPs-based heritability estimates, obtaining a genetic correlation of 14% (McLaughlin et al., 2017) due to polygenic overlap. Intriguingly, conditional false discovery rate was used to investigate novel ALS-associated genomic loci, confirming some of the known pleiotropic risk loci discussed below. A further study found clustering of schizophrenia and psychosis, suicide, autism, rigid personality disorders, and alcoholism in ALS kindreds suggesting that shared pleiotropic oligogenic variants may be responsible for co-segregation of psychiatric illnesses and ALS (O'Brien et al., 2017).

#### c9orf72

In 2011 a worldwide effort identified a hexanucleotidic expansion in the c9orf72 gene as the major genetic determinant of both ALS and FTD (DeJesus-Hernandez et al., 2011; Renton et al., 2011), thus revolutionizing our knowledge of genetic pleiotropy of ALS. C9orf72 pathological expansion accounts for almost 40% of fALS, 8% of sALS, and almost 30% of familial FTD in Caucasian population (Ng et al., 2015; Ng and Tan, 2017). This prevalence is increased in ALS-FTD, where it is found in 50– 70% of familial and 15–20% of apparently sporadic cases (van der Zee et al., 2013). Noteworthy, c9orf72 repeat expansion display a high phenotypic variability, spanning from parkinsonism (Floris et al., 2012), to corticobasal degeneration (Lindquist et al., 2013), psychosis (Watson et al., 2016), and suicidal behavior (Synofzik et al., 2012). Penetrance is incomplete and agedependent (Murphy et al., 2017), with anticipation phenomena similar to other repeat expansions diseases (Van Mossevelde et al., 2017). When analyzing the clinical feature best discriminating c9orf72 carriers from non-carriers in a cohort of bvFTD patients, psychosis and familiality for ALS appeared the most reliable clues (Devenney et al., 2014). A recent study investigating the risk of psychiatric disorders in c9orf72 positive kindreds, extrapolated from FTD and ALS cohorts, revealed an association with increased risk of autism spectrum disorders (HR: 2.7), schizophrenia (HR for a family member: 4.9) or a single psychotic episode (HR: 17.9), and mood disorder (HR: 1.9) (Devenney et al., 2018). Overall, this study confirms previous reports from an aggregation study in which stratification of ALS probands in carriers and non-carriers of c9orf72 repeat expansion was associated with major risk of presenting psychiatric disturbances in family members (Byrne et al., 2013). Importantly, among the referred psychiatric disorders associated to c9orf72 expansion, obsessive-compulsive disorder seems to be excluded (Arthur et al., 2017), though rigid stereotyped behavior with obsessiveness is frequently observed in carriers (Snowden et al., 2012).

C9orf72 is an alternatively spliced gene encoding for three protein transcripts, whose functions have not been fully elucidated. Molecular studies showed that the protein localizes in the nucleus and is structurally similar to DENN (differentially expressed in normal and neoplasia) proteins, which contain a guanine nucleotide exchange factor allowing them to interact with RAB GTPase proteins and regulate membrane trafficking from the nucleus (Levine et al., 2013; Aoki et al., 2017). A striking characteristic of c9orf72 alterations is that differential repeat length is observed across different tissues (Van Blitterswijk et al., 2013), suggesting instability and possibly the occurrence of epigenetic phenomena such as hypermethylation as a potential source of this variability (Xi et al., 2014).

Healthy individuals carry up to 25 repeats of GGGGCC in c9orf72, with the majority having a couple of repeats, while in ALS and FTD cases the number of repeats ranges from 100s to 1000s (DeJesus-Hernandez et al., 2011; Renton et al., 2011; Beck et al., 2013). Uncertainty surrounds the role of intermediate length (22–30) repeats, though they seem to be associated with a higher frequency of psychiatric symptoms in FTD, FTD-ALS, and atypical parkinsonism cohorts (Ng and Tan, 2017).

Recently some studies focused on biomarkers that may help predict the so-called "phenoconversion," since genetic therapy may become an option for in C9orf72 carriers (Floeter and Gendron, 2018). Biological markers might guide pharmacological response to potential therapies, as in the case of Poly(GP) proteins (Gendron et al., 2017a), or predict the prognosis (Gendron et al., 2017b) and anticipate the onset of symptoms by a year (Benatar et al., 2018), as in the case of neurofilament heavy and light chain, respectively. Furthermore, imaging studies showed that atrophy of several cortical and subcortical structures have been observed in asymptomatic carriers, including the thalamus (Papma et al., 2017; Bertrand et al., 2018; Floeter and Gendron, 2018), the left caudate and putamen, besides diffuse cortical thinning in defined temporal, parietal, and occipital regions (Walhout et al., 2015). Similarly, white matter tracts are not spared either before symptoms onset: functional studies have shown salience and medial pulvinar networks, who are known connectivity networks prominently affected in bvFTD, to be severely disrupted in carriers already in their 40s (Lee et al., 2016). Increased radial diffusivity has been reported as well in the right anterior thalamic radiation and the right forceps, even in younger c9orf72 expansion carriers (Bertrand et al., 2018). Notwithstanding, asymptomatic carriers did not show significant atrophy before symptoms onset in a longitudinal voxel-based morphometry study (Floeter et al., 2016). When testing the hypothesis that psychiatric disturbances might be prodromal of the structural and functional brain abnormalities observed in c9orf72 presymptomatic carriers, Lee et al. (2016) found that carriers and non-carrier family members had comparable lifetime histories of psychiatric symptoms, non-carriers family members doubled the amount of psychiatric medications compared to carriers, and underwent similar rates of hospitalization for psychiatric disturbances. Overall, we cannot exclude that familiality for psychiatric diseases, which are known polygenic conditions, runs independently of the expansion among c9orf72 families, however, further studies are warranted to better explore

prodromal disturbances of thought in expansion carriers because of the high variability between personality and behavioral tests, which might not be suited to detect subtle changes in nondemented cohorts.

# Other Genes Associated With Psychiatric Disturbances

Notwithstanding the major role of c9orf72, a non-trivial residual association between ALS and psychiatric disorders persists even after excluding repeat expansion carriers from genetic analyses (Byrne et al., 2013). Isolated cases of concomitant psychiatric disorders such as schizophrenia have been found in kindreds with specific mutations in FUS (Yan et al., 2010) and TARDBP (Quadri et al., 2011). Variable rates of psychiatric illnesses, generally less common than in c9orf72 repeat carriers, were also observed in non-c9orf72 ALS-FTD cases, carrying PRGN (Hall and Finger, 2015), TBK1 (Van Mossevelde et al., 2016), and VCP (Weihl, 2011) mutation. ATXN2, has also been associated to both ALS and schizophrenia risk (Zhang et al., 2014).

It is reasonable to speculate that the numerous genetic loci known to be involved in the ALS-FTD disease spectrum, such as TBK1, PGRN, CHCHD10, TUB4A, VCP, may predispose to psychiatric illnesses by analogous mechanisms to c9orf72. The rarity of these cases, together with the relatively small populations studied, and the difficulty in discerning psychiatric disturbances from other aspects of behavioral FTD, make proving this assumption a daunting task.

# PROGNOSTIC ROLE OF PSYCHIATRIC DISTURBANCES IN ALS

Concomitant psychiatric diseases in ALS patients, whether prior or after this fatal diagnosis, may add strain on caregivers and pose important ethical challenges for support and endof-life decisions along the course of this disease. Until now, only few register-based studies have taken into account the prognostic significance of simultaneous psychiatric illness in ALS, showing a mild negative influence of anxiety symptoms and other psychiatric disturbances in univariate analysis, whose

# REFERENCES


effect was later unconfirmed in multivariate analysis (Körner et al., 2013; Mandrioli et al., 2018). Other studies examining the impact of neuropsychiatric symptoms in ALS, expressed mainly as behavioral alterations, failed to demonstrate any impact on survival (Mioshi et al., 2014; Burke et al., 2017). However, in a prospective study evaluating depression in ALS by validated outcome measures, a concurrent diagnosis of major depressive disorder corresponded to decreased survival, and any increasing score matched increased death hazard ratio (Thakore and Pioro, 2016). This discordance in findings may be related to an underestimation of depression in the ALS population. In addition, the use of more subtle evaluating psychometric measures may aid the clinician to formulate such a diagnosis and address these disorders, which are increasingly reported as a major burden for carers (Creemers et al., 2016).

# CONCLUSION

In conclusion, psychiatric diseases often anticipate the onset of motor symptoms in ALS, and their timely relation with motor neuron pathology may be due to underlying common pathogenic mechanisms affecting non-motor structures within the central nervous system. Similar changes in structural framework between ALS, ALS-FTD and schizophrenia exist, and some degree of genetic overlap between these diseases has been found, strengthening a common pathological signature. Overall, psychiatric illness do not appear to influence significantly the prognosis and survival of ALS patients, but may constitute an increased burden for caregivers and challenge ethical choices with regards to end-of-life decisions. Thus, clinicians should be aware of the tight relationship between ALS and psychiatric disorders and timely address specialist interventions to better assist ALS families.

# AUTHOR CONTRIBUTIONS

JM, EZ, and NT contributed to conceptualization, data curation, formal analysis, and methodology. EZ wrote the first draft. and JM and NT reviewed and edited it.

of hypothesized mechanisms of neurotoxicity in first-episode psychosis. Can. J. Psychiatry 59, 513–517. doi: 10.1177/070674371405901003


neurodegenerative syndromes and are more frequent than expected in the UK population. Am. J. Hum. Genet. 92, 345–353. doi: 10.1016/j.ajhg.2013.01.011



dilemma? Int. Psychogeriatr. 27, 531–539. doi: 10.1017/S104161021400 2580


a large multicenter ALS cohort. Neurology 86, 813–820. doi: 10.1212/WNL. 0000000000002305


and clinical and pathological characteristics in carriers of C9ORF72 repeat expansions (Xpansize-72): a cross-sectional cohort study. Lancet Neurol. 12, 978–988. doi: 10.1016/S1474-4422(13)70210-2


pathogenesis. Curr. Alzheimer Res. 8, 252–260. doi: 10.2174/15672051179556 3773


**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 © 2019 Zucchi, Ticozzi and Mandrioli. 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.

# Notch Signalling in the Hippocampus of Patients With Motor Neuron Disease

Ulises Gómez-Pinedo<sup>1</sup> \*, Lucía Galán<sup>2</sup> , Jordi A. Matías-Guiu<sup>2</sup> , Vanesa Pytel1,2 , Teresa Moreno<sup>2</sup> , Antonio Guerrero-Sola<sup>2</sup> and Jorge Matías-Guiu1,2

<sup>1</sup> Laboratory of Neurobiology, Institute of Neurosciences, IdISSC, Hospital Clínico San Carlos, Universidad Complutense de Madrid, Madrid, Spain, <sup>2</sup> Department of Neurology, Institute of Neurosciences, IdISSC, Hospital Clínico San Carlos, Universidad Complutense de Madrid, Madrid, Spain

Introduction: The Notch signalling pathway regulates neuronal survival. It has some similarities with the APP signalling pathway, and competes with the latter for α- and γ-secretase proteolytic complexes. The objective of this study was to study the Notch signalling pathway in the hippocampi of patients with motor neuron disease.

#### Edited by:

Foteini Christidi, National and Kapodistrian University of Athens, Greece

#### Reviewed by:

Chiara F. Valori, German Center for Neurodegenerative Diseases (DZNE), Germany Patrizia Longone, Fondazione Santa Lucia (IRCCS), Italy

> \*Correspondence: Ulises Gómez-Pinedo u.gomez.pinedo@gmail.com

#### Specialty section:

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

Received: 10 December 2018 Accepted: 15 March 2019 Published: 05 April 2019

#### Citation:

Gómez-Pinedo U, Galán L, Matías-Guiu JA, Pytel V, Moreno T, Guerrero-Sola A and Matías-Guiu J (2019) Notch Signalling in the Hippocampus of Patients With Motor Neuron Disease. Front. Neurosci. 13:302. doi: 10.3389/fnins.2019.00302 Methods: We studied biological material from the autopsies of 12 patients with motor neuron disease and 4 controls. We analysed the molecular markers of the Notch and APP signalling pathways, TDP43, tau, and markers of neurogenesis.

Results and Conclusion: Low NICD expression suggests Notch signalling pathway inactivation in neurons. Inactivation of the pathway despite increased Notch1 expression is associated with a lack of α-secretase expression. We observed increased β-secretase expression associated with activation of the amyloid cascade of APP, leading to increases in amyloid-β and AICD peptides and decreased levels of Fe65. Inactivation of the Notch signalling pathway is an important factor in decreased neurogenic response in the hippocampi of patients with amyotrophic lateral sclerosis.

Keywords: ALS, Notch, NICD, APP, ADAM10, ADAM17, BACE1, hippocampal neurogenesis

# INTRODUCTION

The Notch signalling pathway regulates cell migration and growth, synaptic plasticity, and neuronal survival (Ables et al., 2011). Notch proteins are highly conserved transmembrane receptors with such pleiotropic functions as neuronal development and organ homeostasis, and are activated by ligand binding. The ligand-receptor association triggers sequential proteolytic processes via αand γ-secretases. Proteolysis generates a Notch intracellular domain (NICD) that may translocate to the nucleus. There is an evident parallel between the Notch and amyloid precursor protein (APP) signalling pathways, which compete for proteolytic complexes; an association between both pathways has therefore been suggested.

Notch1, the most extensively studied Notch receptor, is expressed in the cortex and hippocampus, and may be involved in neurodegeneration (Woo et al., 2009). Notch1-deficient mice display memory impairment (Costa et al., 2003). It has been suggested that Notch1 participates in olfactory function (Brai et al., 2014), which is impaired in patients with Alzheimer disease (AD) (Berezovska et al., 1998; Moehlmann et al., 2002; Brai et al., 2016) and in experimental models of

**27**

familial AD secondary to presenilin mutations (Okochi et al., 2002). Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease affecting motor neurons in the brain, brainstem, and spinal cord (Rowland and Shneider, 2001). However, as recent anatomical pathology studies of ALS have shown that degeneration affects not only motor areas but also such other structures as the hippocampus (Coan and Mitchell, 2015), we aimed to determine whether the Notch pathway is active in the hippocampi of patients with ALS.

#### MATERIALS AND METHODS

We studied biological samples from the autopsies of 12 patients with ALS or ALS with frontotemporal dementia (ALS-FTD) who died between 2006 and 2017 and met diagnostic criteria for ALS (Ludolph et al., 2015). Ten patients died due to respiratory insufficiency during the terminal stage of ALS, one patient due to cardiac arrest attributed to bulbar involvement, and the remaining patient due to a concomitant cerebral haemorrhage. We also studied the brains of 4 controls, who died during hospitalisation due to non-neurological diseases and had no history of neurodegenerative disease. Only 10 brains were used for the study of neurogenesis. Patients or their families expressed in writing their consent for the brain to be donated for research. Autopsies were performed within 2–6 h after death, in accordance with our centre's protocol and Spanish national regulations. The procedure, which we described in a previous article (Gómez-Pinedo et al., 2016), is outlined in **Supplementary Material 1**. Briefly, five slides were used per patient and IHC, analysing in each slides 32 fields (per hippocampal zone: CA1, CA2, CA3, and dentate gyrus). ImageJ version 1.46r was used when the unit of measurement was the amount of labelling per field (optical density [OD]). Inclusions were calculated as the number of stained inclusions observed in neurons divided by the mean number of neurons per field (percentage of cells per mm<sup>2</sup> ). Statistical analysis was performed using the SPSS statistics software, version 20.0. GraphPad Prism version 5.0 was used to plot graphs and to calculate Pearson correlation coefficients between the parameters studied. Data are expressed as mean ± SD. Due to the small size of our sample, means were compared using the non-parametric Mann–Whitney U test. Statistical significance was set at p < 0.05.

# RESULTS

#### Notch1 Expression

Notch1 expression was observed at higher levels in patients than in controls (**Figure 1**). Mean labelling per field was 1.315 ± 0.448 OD in patients and 0.725 ± 0.061 OD in controls (p < 0.02). **Figure 1** also shows differences between patients: one patient with ALS-FTD showed lower Notch1 expression than controls, 2 patients with ALS showed similar expression to that observed in controls, and the remaining patients displayed clearly greater Notch1 expression. Patients and controls showed different labelling patterns. In controls, labelling was mainly observed in the cytoplasm and dendrites in CA1, especially in the polymorphic layer of the subgranular zone (SGZ). In patients, however, labelling was heterogeneous in all hippocampal areas, being more evident in granular neurons and remarkable in CA1, CA3, the polymorphic layer of CA4, and the SGZ; Notch1 expression was observed in neurons and to a lesser extent in astrocytes and neuronal processes, following a slight synaptic pattern in CA1 and in capillary walls (**Figure 1**).

## NICD Expression

NICD expression was variable in patients with ALS, and was observed at lower levels than in controls (**Figure 2**). Mean labelling per field was 0.228 ± 0.070 OD in patients and 1.101 ± 0.101 OD in controls (p < 0.0001). NICD expression is inversely related to Notch1 expression and shows homogeneous values, since it behaves similarly in all patients with ALS, including those with ALS-FTD; NICD labelling in patients was lower than that observed in controls. In patients with ALS, NICD labelling was mainly observed in cell nuclei, especially in granular neurons, neurons near the SGZ, and in the polymorphic layer. The SGZ contained astrocytic cells coexpressing GFAP and NICD, especially in patients with greater numbers of extracellular amyloid plaques (**Supplementary Material 2**). Patients with no plaques or isolated plaques showed no NICD labelling, except in isolated astrocytes near the SGZ (**Supplementary Material 2**). In controls, NICD was mainly expressed in neurons of the SGZ and colocalised with GFAP. Labelling was also observed in microglia (Iba1-positive cells) (**Figure 2**).

# ADAM10, ADAM17, and BACE1 Expression

ADAM10 expression was 0.199 ± 0.060 OD in patients and 0.462 ± 0.089 OD in controls (p = 0.0015); it was lower in all patients with ALS (**Figure 3**). ADAM17 expression was 0.230 ± 0.057 OD in patients and 0.542 ± 0.084 OD in controls (p = 0.0015); again, expression was lower in all patients than in controls (**Figure 3**). In patients with ALS, ADAM10 and ADAM17 expression was similar in terms of location, although ADAM10 expression was slightly more marked in the granular layer and less marked in the SGZ. ADAM17 expression was more diffuse, whereas ADAM10 expression was more defined and homogeneous (**Supplementary Material 4**). In controls, expression of both ADAM10 and ADAM17 was more marked in the SGZ (**Supplementary Material 3**), colocalised with GFAP, and was also observed in neurons. BACE1 expression was evaluated with OD and by determining the number of cells labelled by the antibody. BACE1 expression was 0.475 ± 0.114 OD in patients with ALS and 0.243 ± 0.048 OD in controls (p < 0.0001). Higher levels were observed in the majority of patients with ALS, however, three patients showed similar levels to those of controls (**Figure 4**). A total of 13.73 ± 2.97 cells per field in patients with ALS and 8.00 ± 2.588 in controls were labelled by anti-BACE1 antibody (p = 0.0047). Similarly, the antibody labelled more cells in all patients with ALS, with the exception of the 3 patients showing similar BACE1 expression to those of controls as measured by OD (**Figure 4**).

significantly higher in patients (p < 0.05). Images correspond to the following areas: A, CA1 control; B, CA2 ALS; C, CA3 ALS; D, CA1 and subgranular zone ALS; E, CA3; and F, subgranular zone ALS. Scale bar: 50 µm. Graphs express data as means (G) and standard deviation (H).

FIGURE 2 | NICD expression in the hippocampi of patients with ALS and controls. Immunohistochemical images of NICD expression obtained by confocal microscopy, showing clear differences in labelling patterns between patients and controls. The most significant differences were observed in the granular layer and the subgranular zone. Controls showed more marked labelling than patients with ALS (A–C), especially in the granular zone, Controls displayed cells (astrocytes) coexpressing GFAP and NICD in the subgranular zone (B arrows), whereas patients showed faint labelling in the granular layer (D,E), in addition to Iba1-positive cells (F, arrows). The graph in F shows quantitative data from the hippocampi (including CA1, CA2, CA3, CA4, and dentate gyrus) of all patients and controls. Values are higher in controls than in patients. The graph in G shows significant differences in mean values between patients and controls, with controls showing higher expression (p < 0.05). Images A, CA3; B, D, and E show the granular layer and subgranular zone, whereas image C and F shows the CA2. Scale bar: 50 µm. Graphs express data as means (G) and standard deviation (H).

FIGURE 3 | ADAM10 and ADAM17 expression in the hippocampus. Images of the hippocampus obtained using confocal microscopy after immunohistochemistry against ADAM10 and ADAM17 show lower marker expression in controls (A–C) than in patients (D–F). Labelling was mainly observed in the granular layer of the dentate gyrus; the subgranular zone showed weak, diffuse labelling (D–F). Labelling could not be observed in the remaining areas of the hippocampus in some patients. Controls displayed denser labelling, which was observed in the granular layer and subgranular zone, where astrocytes coexpressed ADAM10 and ADAM17 (C, arrows). Quantitative data for ADAM10 and ADAM17 are shown in graphs (G) and (H), respectively. Graphs show similar ADAM10 and ADAM17 expression in both patients and controls; this is further confirmed by graphs (I) and (J), which show statistically significant differences (p < 0.05). Images A and B correspond to representative areas of the dentate gyrus and the subgranular zone in controls, whereas images C and D display the same areas in patients. Scale bar: 25 µm. Graphs E and F express data as means, and graphs G and H as means ± standard deviation.

# APP, Amyloid-β, AICD, and Fe65 Expression

APP expression was 5689 ± 2036 OD in patients and 2544 ± 1209 OD in controls (p = 0.0076). APP intracellular domain (AICD) expression was elevated in 5 patients, whereas the remaining 5 showed similar levels to those observed in controls (**Supplementary Material 4**). Amyloid-β (Aβ) expression was elevated in all patients with ALS (38 249 ± 24 449 OD vs. 6003 ± 2231 OD in controls; p = 0.0127). AICD expression was 14.10 ± 3.07 OD in patients and 9.2 ± 3.49 OD in controls (p = 0.0153). In patients displaying Aβ expression, AICD was observed in astrocytes and microglia surrounding amyloid plaques; it was also expressed in dendrites. Elevated AICD expression was observed in all patients with ALS, with the exception of 3 who showed similar levels to those observed in controls (**Supplementary Material 4**). Fe65 expression was 13.90 ± 3.24 OD in patients and 22.60 ± 4.03 OD in controls (p = 0.0006). Decreased Fe65 expression was observed in all patients, with the exception of 3 who showed similar expression to that of controls; these were not the same patients showing increased AICD expression (**Supplementary Material 4**). Ten patients displayed amyloid plaques with varying degrees of labelling. Plaques were frequently diffuse and isolated; only 4 of these patients showed prominent plaques, mainly located in CA1, CA3, and cortical regions.

#### Expression of Markers of Neurogenesis

Shows data for 10 of the 12 patients with ALS (the remaining 2 patients were not included in this part of the study for technical reasons). Patients showed a marked decrease in proliferative and neurogenic activity in the hippocampus. Proliferation in the SGZ was analysed using the proteins PCNA and Ki67, expressed during cell division. The number of cells labelled with these 2 markers was significantly lower in patients than in controls: 0.47 ± 0.72 cells/µm<sup>2</sup> vs. 2 ± 1.41 cells/µm<sup>2</sup> for PCNA labelling, and 0.29 ± 0.72 cells/µm<sup>2</sup> vs. 1.65 ± 0.96 cells/µm<sup>2</sup> for Ki67 labelling (p < 0.01). The study of such immunohistochemical markers of human pluripotent cells as GFAPδ in the dentate gyrus identified fewer labelled cells in patients than in controls: 1.18 ± 0.83 cells/µm<sup>2</sup> vs. 5.7 ± 1.81 cells/µm<sup>2</sup> (p < 0.01). The markers DCX, TuJ1, and PSA-NCAM were used to study neuroblast expression in the dentate gyrus. PSA-NCAM expression was reduced in patients with ALS (0.39 ± 0.43 cells/µm<sup>2</sup> vs. 7.3 ± 4.29 cells/µm<sup>2</sup> ; p < 0.01). The other 2 markers could not be viewed since they only labelled small dendritic projections but no somata.

#### TDP43 and Tau Expression

Cytoplasmic TDP43 expression was evaluated using staining for phosphorylated TDP43; expression was determined at 55.75 ± 5.97 OD in patients vs. 58.8 ± 5.167 OD in controls; differences were not statistically significant. However, the number of TDP43-positive cytoplasmic inclusions per field did show statistically significant differences between patients and controls (20.30 ± 10.32 vs. 1.504 ± 0.993; p = 0.0004). All patients displayed TDP43-positive cytoplasmic inclusions. Quantification of phosphorylated tau (phospho-tau) expression revealed levels of 26 106 ± 20 413 OD in patients and 4824 ± 2896 OD in controls (p = 0.0395). Only 6 patients showed increased phosphotau expression; we observed neurofibrillary tangles mainly in CA3 and CA1, and staining of small fibres in CA1 axonal projections. The remaining patients showed similar levels of phospho-tau expression to that of controls.

#### Correlations Between Molecular Markers

**Supplementary Material 5** show the correlations between the various molecular markers. In the **Supplementary Tables 1, 2** describe the correlations between molecular markers of NOTCH pathway and the correlation with NOTCH and adult neurogenesis.

# Clinical Correlations

Patients (7 men and 5 women) were aged between 37 and 87 years at symptom onset; onset was spinal in 5 cases and bulbar in 7. Survival times ranged from 4 months to 14 years. Two patients also had dementia; one case was attributed to FTD due to the associated aphasia. Seven patients underwent a genetic study. Three patients showed a SOD1 mutation, which was pathogenic in only one case. One patient had a pathogenic mutation of the gene coding for TDP43. No patient had more than 20 repeats in C9ORF72. No correlations were found between molecular data and such clinical characteristics as age, sex, form of onset, survival time, or presence of any of the genetic variants detected (**Supplementary Table 3**).

# DISCUSSION

#### Notch1 Is Overexpressed, Whereas NICD Is Underexpressed

Notch1 is generated by a convertase enzyme in the Golgi apparatus via S1 cleavage and subsequently transported to the cell membrane, where it is expressed as a heterodimeric transmembrane protein (Lieber et al., 2002). Notch1

overexpression has previously been observed in other neurodegenerative diseases (Nagarsheth et al., 2006; Lathia et al., 2008) activation of the Notch signalling pathway has been found to play a role in ageing and memory (Ables et al., 2011; Alberi et al., 2013). Notch1's role in ALS has previously been observed in several experimental studies, which report conflicting findings. Wang et al. (2015) observed that the Notch signalling pathway is activated in vitro models and SOD1G93A mice, and that Notch suppression with a Notch1 signalling inhibitor significantly reduced neuronal apoptosis. Nonneman et al. (2018), in contrast, studied Notch signalling in the spinal cords of SOD1G93A mice and patients with

patients with ALS (B). (A) Healthy control. In the Notch signalling pathway, Notch1 activation requires direct contact between the membranes of 2 cells. Once the receptor is active, the protein undergoes two consecutive cleavages by α-secretases (ADAM10 and ADAM17), followed by transmembrane cleavage by γ-secretases. NICD is internalised into the nucleus, activating transcription factors and promoting mechanisms of cell plasticity, proliferation, and neurogenesis. The APP signalling pathway may follow two routes. The non-amyloidogenic (neuroprotective) APP pathway: after α-secretase–mediated cleavage, γ-secretases generate (Continued)

#### FIGURE 5 | Continued

fnins-13-00302 April 4, 2019 Time: 18:1 # 9

AICD and decrease APP levels. Cytosolic adaptor protein Fe65 binds to AICD; this complex translocates to the nucleus where it interacts activating signalling to promotes neurite growth and cell plasticity. The second pathway is the amyloidogenic APP pathway, where APP is processed by α-secretase and γ-secretase, forming Aβ. AICD is also generated, which may activate NICD expression via the lysosomal system or interact with Fe65. Both pathways converge in the activation of substrates by α-secretases. (B) A possible pathogenic mechanism in ALS is increased Notch1 expression due to poor processing of α-secretases, given the competition between the Notch and APP pathways. This results in poor NICD activation, leading to transcription block and the subsequent decrease in plasticity, proliferation, and neurogenesis. In the APP pathway, the competition for α-secretases (ADAM10, ADAM17) causes β-secretase activation, leading to the formation of Aβ oligomers; Aβ plaque formation in the intracellular space generates AICD, directly affecting Fe65 expression. AICD and Fe65 form a complex that translocates to the nucleus, where it may activate the apoptotic pathway. Increased AICD expression and low Fe65 activity may affect other lysosomal degradation pathways, increasing cytoplasmic TDP43 inclusions.

sporadic ALS, finding increased pathway activation in reactive GFAP-positive astrocytes. Astrocyte-specific inactivation of Jagged-1 in presymptomatic SOD1G93A mice increased the activation of the Notch signalling pathway and accelerated disease progression without affecting disease onset. In a study of Drosophila, Yang et al. (2015) observed that dipeptide repeat proteins associated with a repeat expansion in C9ORF72, present in patients with FTD and ALS, were accompanied by Notch signalling suppression. We did not observe Notch signalling pathway activation in the hippocampi of our patients: while they displayed increased Notch1 expression, NICD expression was significantly decreased. Our results are consistent with those reported by Ma et al. (2017), who observed decreased NICD signalling in spinal cord motor neurons of SOD mice; this decrease was correlated with disease expression. Disease onset occurred between 90 and 120 days of age, at which time NICD levels progressively decreased in motor neurons. After binding to one of its transmembrane ligands, Notch undergoes sequential cleavage from NICD first by α-secretase and then by γ-secretase; NICD is then internalised into the nucleus (**Figure 5**). Our results suggest that although Notch expression increases, probably in response to increased expression of its ligands (Nonneman et al., 2018). This is consistent with the findings of Ma et al. (2017), who suggest the involvement of some mechanism within the cell that interrupts activation of the pathway. However, our patients showed NICD expression in astrocytes (GFAP-positive cells) and microglia (Iba1-positive cells), suggesting Notch pathway activation in these cells. This observation coincides with the results of Nonneman et al. (2018), who report Notch signalling pathway inactivation in neurons and activation in glial cells. This is consistent with the hypothesis of neuronal loss and gliosis in ALS (Rowland and Shneider, 2001).

#### APP and Notch1 Expression

In a previous study, we observed amyloid cascade activation in the hippocampi of patients with ALS-FTD, in the form of increased APP and Aβ expression. AICD expression was variable, as the protein was not overexpressed in all patients; we also observed decreased Fe65 expression, suggesting that AICD may have bound to Fe65 and been internalised into the nucleus, as occurs when the APP signalling pathway is activated (Gómez-Pinedo et al., 2016). These results are consistent with our findings from an in vivo study of patients with ALS using PET with amyloid tracers (Matías-Guiu et al., 2016). The fact that Notch and APP signalling pathways compete with each other for α- and γ-secretase underscores the need to analyse the link between these pathways. In order to generate AICD, APP undergoes sequential cleavage, first by α- or β-secretase and subsequently by γ-secretase. When cleavage is first mediated by β-secretase, APP generates Aβ peptides, which are oligomerised in the form of aggregates. APP cleavage by γ-secretase generates AICD (**Figure 5**). In this study, we observed increased APP and Aβ expression. Similarly, a rat model of AD showed that soluble Aβ1−<sup>42</sup> suppresses Notch1 and NICD expression (Zhang et al., 2016). AICD generated by γ-secretase–mediated cleavage is rapidly degraded by the endosomal-lysosomal system. Fe65 stabilises AICD; together, both molecules localise to the nuclear compartment, where they bind the histone acetylase Tip60, forming AFT complexes (Von Rotz et al., 2004). AICD also competes with NICD within the nucleus. A study of human embryonic kidney cells found colocalisation in AFT complexes; NICD can localise to the nucleus together with Fe65 and Tip60 in the absence of AICD (Konietzko et al., 2010).

# Fe65 Expression Increases With Notch1 Overexpression

Fe65 binds to AICD; in the nucleus, this complex is involved in regulating the transcription of certain genes, including the gene coding for APP (Cao and Südhof, 2001); in vitro studies have shown colocalisation of AICD and TDP43 in the nucleus (Wang et al., 2014). Fe65 is highly expressed in the hippocampus (Kesavapany et al., 2002). Fischer et al. (2005) observed increased Notch1 expression, interaction between APP and Notch1, and NICD binding to Fe65 in the cerebral cortex of adults with Down syndrome, which is associated with enhanced APP production. NICD may colocalise with AICD, Fe65, and Tip60, interrupting the formation of the AFT complex and playing a protective role by inhibiting AFT complex–induced cell death (Kim et al., 2007). Our study shows that decreased Fe65 expression coincides with decreased NICD expression and increased Notch1 expression; this is consistent with the results of a previous study by our research group (Gómez-Pinedo et al., 2016).

# Notch Signalling Pathway Inactivation May Explain Decreased Neurogenic Response in the Hippocampus

The hippocampus, one of the classic neurogenic niches of the adult brain, constantly generates neurons throughout life (Palmer et al., 1997; Eriksson et al., 1998; Seri et al., 2004). Granule cells are born in the SGZ of the dentate gyrus; they migrate to the granular layer and integrate into the neural network.

However, these cells are estimated to be less numerous than those born in the subventricular zone. AD (Hollands et al., 2016) and other neurodegenerative diseases (Winner and Winkler, 2015) have been found to be associated with alterations in hippocampal neurogenesis. Our research group has described decreased hippocampal neurogenesis in patients with ALS, which stands in contrast with the increased neurogenesis observed in the subventricular zone in these patients (Galán et al., 2017). The Notch signalling pathway has been linked to hippocampal neurogenesis as the Notch receptor is expressed in neural stem cells (Traiffort and Ferent, 2015). It has been suggested that this pathway may alter the number of neural stem cells by acting on cell survival (Ables et al., 2010), self-renewal (Aguirre et al., 2010), and differentiation (Breunig et al., 2007), and may act as a mechanism of communication between a neural stem cell and its descendants (Semerci et al., 2017). Our results show that expression of markers of proliferation (Ki67), differentiating cells (GFAPδ), and differentiation (PSA-NCAM) is significantly correlated with NICD expression and negatively correlated with Notch1, which suggests that decreased neurogenesis in ALS (Galán et al., 2017) may be associated with Notch signalling pathway inactivation. Interestingly, some drugs that increase Notch signalling have been found to promote hippocampal neurogenesis (Xue et al., 2017).

#### Expression of α- and β-Secretases

The role of α- and β-secretases comes to the forefront in view of 3 main findings: Notch signalling inactivation despite increased Notch1 expression; decreased hippocampal neurogenesis resulting from Notch signalling inactivation; and APP signalling pathway activation (Gómez-Pinedo et al., 2016). α-Secretase acts on Notch during S2 cleavage in response to ligand binding, which induces a conformational change. The cleaved form that remains bound to the membrane is called NEXT and will be the substrate for γ-secretase–mediated proteolysis (S3 cleavage), whereas the resulting intracellular Notch fragments are short-lived. α-Secretases are membraneanchored, zinc-dependent members of the A disintegrin and metalloproteinase (ADAM) family. ADAM includes a wide range of proteins with protease and adhesive domains, which play a key role in cell-cell and cell-matrix interactions in various important biological cell processes. The role of α-secretase involves ADAM10 and ADAM17, and to a lesser extent ADAM12 and ADAM9. Little information is available on ADAM molecular expression in ALS. Our study detected decreased ADAM10 and ADAM17 expression, which may explain S2 cleavage inhibition (Groot and Vooijs, 2012). Our results do not allow us to determine the reason for reduced α-secretase expression, although the discovery of some factors involved in ADAM10 regulation permits us to propose several hypotheses. Synapse-associated protein 97 binds to ADAM10, creating a complex that enables its transportation from the endoplasmic reticulum to the cell membrane. Glucagon-like peptide-1 is associated with decreased ADAM10 expression and lower Aβ levels (Saftig and Lichtenthaler, 2015). The role of the TspanC8 subgroup of tetraspanins (including Tspan5, Tspan10, Tspan14, Tspan15, Tspan17, and Tspan33) is especially noteworthy (Dornier et al., 2012; Haining et al., 2012). TspanC8 tetraspanins interact with ADAM10 and regulate the cleavage of ADAM10 substrates. Tspan5, Tspan10, and Tspan14 regulate ADAM10-dependent Notch signalling (Zhou et al., 2014); Tspan15 promotes ADAM10-mediated N-cadherin cleavage; and Tspan14 reduces GPVI cleavage (Noy et al., 2016); changes in tetraspanin expression may alter the action of ADAM10. Another hypothesis suggests competition with other substrates where ADAM10 is also involved, such as the TNFα signalling pathway, which has been observed to be active in ALS (Tortarolo et al., 2017) and cause cell alterations (Olmos and Lladó, 2014); this is probably related to microglial activation, one of the first events to be observed in experimental models of the disease (Beers et al., 2011; Liao et al., 2012; Liu and Wang, 2017; Gómez-Pinedo et al., 2018). The observation of NICD expression in Iba1-positive cells in our patients supports this hypothesis. NICD expression is directly correlated with ADAM10 and ADAM17 expression, whereas Notch1 expression decreases with increased metalloprotease expression; this supports our hypothesis that decreased Notch1 and NICD expression is associated with Notch signalling pathway inactivation. Conversely, we observed increased β-secretase expression (BACE1) (Yan et al., 2001); this was not unexpected given the results of our previous study, which showed amyloidogenic APP pathway activation (Gómez-Pinedo et al., 2016), and other results from experimental studies into the interaction between BACE1 and TDP43 (Herman et al., 2012). It has been suggested that BACE1 may play a role in the Notch signalling pathway since it seems to be involved in maintaining the balance between hippocampal astrogenesis and neurogenesis in mice by regulating the Jag1-Notch pathway (Hu et al., 2013; He et al., 2014). In our study, BACE1 expression was significantly correlated with Notch1 expression and negatively correlated with NICD expression, which supports the hypothesis that the amyloidogenic APP pathway is activated in the hippocampi of patients with ALS, whereas the Notch signalling pathway is not.

# Lack of Correlation Between Notch Expression and Cytoplasmic TDP43 and Tau Expression

A previous study by our research group found no correlation between cytoplasmic TDP43 and APP; furthermore, the 2 proteins did not colocalise, although we did find a correlation between TDP43 expression and Aβ expression (Gómez-Pinedo et al., 2016). Zhan et al. (2013) report that TDP-43 upregulates a wide range of Notch target genes, leading to the activation of this cell differentiation pathway in vivo. Notch signalling pathway activation has been linked to prion disease (Ishikura et al., 2005), suggesting a common mechanism with ALS, considering that the C-terminal domain of TDP-43 has prion-like characteristics (Zhan et al., 2013). In our study, no correlation was observed between cytoplasmic TDP43 expression and Notch1 and NICD expression in patients; this contradicts the hypothesis that changes in the signalling pathway may be affected by TDP43. Our previous study detected tau overexpression (Gómez-Pinedo et al., 2016); we hypothesised that this finding

may be linked to increased AICD expression resulting from APP pathway activation (Kim et al., 2003; Ghosal et al., 2009). Tau overexpression has also been reported by other researchers (Vintilescu et al., 2016), and the protein's potential role in ALS has recently been reviewed (Moszczynski et al., 2018). Our study found no correlation with Notch, however.

#### Study Limitations

Our study has a number of limitations, some of which were discussed in our previous article (Gómez-Pinedo et al., 2016). Autopsies were performed 2 to 6 h after death; this time interval is longer than those used in animal studies. Although different tissue samples were used, some of the patients included in this study were included in the previous studies. However, our sample also included new patients, reducing clinical bias: the two previous studies included a large proportion of patients with bulbar-onset ALS, with very short survival times, whereas this study mostly includes patients with spinal-onset ALS, some of whom had long survival times.

#### CONCLUSION

To our knowledge, this is the first study to analyse the Notch signalling pathway in biological samples taken in tissue samples from the hippocampus from patients with ALS; In a recent paper, Nonneman et al. (2018) found that Notch signalling pathway is activated in the reactive astrocytes in the spinal cord of SOD1hG93A mice, as well as in patients with sporadic ALS and their finding of a upregulation of astrocytic Jagged-1 is in concordance with the our finding of the greater expression of Notch1 in our cases. Given the involvement of the Notch signalling pathway in cell survival, the finding that the pathway is not active in neurons (decreased NICD expression) is consistent with neuronal loss in patients with ALS. Inactivation of this pathway in neurons despite increased Notch1 expression is associated with a lack of α-secretase expression, preventing APP non-amyloidogenic signalling pathway activation; this was described in one of our previous studies. We also observed increased β-secretase expression in association with activation of the amyloid cascade of APP, leading to increased Aβ and AICD expression and decreased Fe65 expression. Inactivation of the Notch signalling pathway is an important factor in neuronal death in ALS and also plays a major role in decreased hippocampal neurogenic response in these patients, although the blockade of expression of Notch1 in oligodendrocyte precursor cells does not influence the evolution of the mutant SOD1hG93A mice (Eykens et al., 2018). Our study contributes to the

#### REFERENCES


understanding of molecular changes in cells due to ALS. Lack of α-secretase expression should be analysed in future studies, given its potential therapeutic implications.

#### ETHICS STATEMENT

All participants or their relatives gave informed consent prior to inclusion in the study (including consent to autopsy). Autopsies were performed according to the procedures established by our hospital's anatomical pathology department. This study was approved by the Clinical Research Ethics Committee of Hospital Clínico San Carlos. Data were managed in accordance with Spanish data protection legislation (Organic Law 15/1999 of 13 December). At the time the study was approved, no consent was necessary for publishing the results.

#### AUTHOR CONTRIBUTIONS

UG-P and LG contributed manuscript drafting, study concept and design, data acquisition, data analysis and interpretation, statistical analysis. JAM-G contributed manuscript drafting, study concept and design, data analysis and interpretation, statistical analysis. VP performed data analysis and interpretation, statistical analysis. AG-S performed data analysis, critical review of the manuscript. TM data acquisition, critical review of the manuscript. JM-G performed the study concept and design, data analysis and interpretation, study supervision, and manuscript drafting. All authors have read and approved the final version of the manuscript.

#### ACKNOWLEDGMENTS

The authors wish to thank María Soledad Benito-Martín for her assistance with tissue processing and the Spanish Society of Neurology's translation department for their assistance with the manuscript. The authors also wish to thank Servier Medical Art (https://smart.servier.com) for the use of their material in the preparation of the manuscript.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fnins. 2019.00302/full#supplementary-material

J. Neurosci. 30, 10484–10492. doi: 10.1523/JNEUROSCI.4721- 09.2010



Notch signaling in amyotrophic lateral sclerosis. Neurobiol. Dis. 119, 26–40. doi: 10.1016/j.nbd.2018.07.012


**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 © 2019 Gómez-Pinedo, Galán, Matías-Guiu, Pytel, Moreno, Guerrero-Sola and Matías-Guiu. 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.

# Sleep Disorders in Huntington's Disease

#### *Radoslawa Herzog–Krzywoszanska1\* and Lukasz Krzywoszanski <sup>2</sup>*

*1 General Psychology Unit, Chair of Psychology, Faculty of Pedagogy, Pedagogical University of Krakow, Krakow, Poland, 2 Neurocognitive Psychology Unit, Chair of Psychology, Faculty of Pedagogy, Pedagogical University of Krakow, Krakow, Poland*

Huntington's chorea (Huntington's disease, HD) is a genetic disorder caused by autosomal dominant mutation, leading to progressive neurodegenerative changes in the central nervous system. Involuntary movements such as chorea occur typically in HD patients, accompanied by progressive cognitive and psychiatric disturbances. Other common symptoms of HD are circadian and sleep abnormalities, which are observed from the earliest stages of the disease or even before the occurrence of clinical symptoms. The most common sleep problems reported by HD patients include insomnia, difficulties in falling asleep, frequent nocturnal awakenings, and excessive daytime sleepiness. Also, specific changes in sleep architecture have been identified in HD. In this paper, we review studies on sleep and circadian rhythm disorders in HD. We outline findings concerning sleep patterns and disturbances of circadian rhythms in HD patients, as well as the role of psychiatric disorders and motor disorders in HD patients' sleep problems. We also discuss problems related to the different methods of diagnosing sleep disorders in HD. Furthermore, the adverse effects of medication used for the treatment of core HD symptoms as one of the sources of sleep disturbances in HD are emphasized. In conclusion, the diversity and complexity of the determinants of sleep and circadian rhythm disorders in HD are highlighted. Finally, the relevance of effective treatment to improve patients' functioning and quality of life as well as the potential relief of their cognitive and emotional symptoms is addressed.

Keywords: Huntington's disease, neurodegenerative disease, sleep disorders, circadian rhythm disturbances, melatonin

# INTRODUCTION

Huntington's chorea (Huntington's disease, HD) is a genetic disorder caused by autosomal dominant mutation, leading to progressive neurodegenerative changes in the central nervous system. The disease is caused by a dynamic mutation of the HTT gene, located on the short arm of the fourth chromosome. In the unaffected population, the number of CAG repeats in the HTT gene that encodes the huntingtin (HTT) protein varies from 6 to 35 (1). The mutant version of the gene contains from 36 to 250 repeats of this nucleotide sequence; as a result, a glutamine string present in the amino acid sequence of huntingtin encoded by the HTT gene is excessively elongated. A gradual accumulation of deposits of misfolded huntingtin and other proteins as well as the death of neurons in various areas of the brain are observed during the course of the disease. Neurodegenerative changes in HD primarily occur in the striatum and globus pallidus, but also in the cerebral cortex, cerebellum, amygdala, thalamus, and hypothalamus (2–8). The age of onset of HD symptoms strongly correlates with the number of CAG trinucleotide repeats in HTT. A higher number of CAG sequence repeats

#### *Edited by:*

*Foteini Christidi, National and Kapodistrian University of Athens, Greece*

#### *Reviewed by:*

*Nicholas-Tiberio Economou, National and Kapodistrian University of Athens, Greece Alba Di Pardo, Mediterranean Neurological Institute (IRCCS), Italy*

#### *\*Correspondence:*

*Radoslawa Herzog–Krzywoszanska, radoslawa.herzog-krzywoszanska@ up.krakow.pl*

#### *Specialty section:*

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

*Received: 15 December 2018 Accepted: 26 March 2019 Published: 12 April 2019*

#### *Citation:*

*Herzog-Krzywoszanska R and Krzywoszanski L (2019) Sleep Disorders in Huntington's Disease. Front. Psychiatry 10:221. doi: 10.3389/fpsyt.2019.00221*

**40**

in the mutation carrier causes earlier onset of the disease, more severe cognitive impairments, severe progression of degenerative changes, and worse prognosis (9). The disease usually manifests in people aged 35–50, and it causes death within 15–20 years (10). It is estimated that HD frequency in Europe is 5–10 cases per 100,000 inhabitants (11).

Three main types of disorders described in HD are motor, cognitive, and psychiatric (12, 13). The most common motor symptoms in HD include chorea—uncontrolled movements of the head, neck, and limbs that may impede daily activities. Patients have difficulties maintaining balance and often stumble and fall. The clinical picture includes other motor symptoms such as bradykinesia, muscle rigidity, spasticity, myoclonus, dystonia, tics, ataxia, and athetotic movements (14).

The most common cognitive deficits in HD include attention (15, 16) and executive function disorders manifested by difficulties in taking decisions, planning and execution of complex activities, reduced flexibility of thinking and behavior, as well as a tendency to perseveration (17, 18). HD also involves implicit memory and learning disorders (19).

HD patients also develop psychiatric disorders that often occur before the onset of full-blown HD. Low mood and depression are revealed in a considerable number of the patients (20, 21), while anxiety disorders, dysphoria, emotional lability, apathy, and arousal are more rare (22, 23). In some patients, schizophrenialike psychosis and paranoid symptoms have been observed (24).

The common symptoms of HD also include sleep disturbances (25). Up to 90% of patients report sleep problems that are evaluated as important by over half of them (26). Information about sleep disorders is also provided by spouses and caregivers. The number of CAG repeats in most studies did not correlate with the occurrence of sleep disorders (27); however, they were partially correlated with the duration of the disease and the severity of clinical symptoms (28). Studies have demonstrated that some sleep disorders are found in the very early phase of HD and even in carriers of the mutation at the premorbid stage (27, 29, 30). Many studies also show a high incidence of sleep disorders in other neurodegenerative diseases such as Alzheimer's disease (AD), Parkinson's disease (PD), or spinocerebellar ataxia (31–33).

Lack of sleep causes a decrease in mental alertness, irritability, attention disturbances, deceleration of reactions, and reduced ability for logical thinking. Sleep disorders considerably affect patients' life activities, and their negative impact can be manifested especially in people with neurodegenerative disorders, who experience serious difficulties in everyday functioning due to the illness (29, 34, 35).

In this paper, we review the literature on sleep disorders and circadian rhythm disturbances in HD patients. We performed a search of online electronic databases (PubMed, MEDLINE, Scopus, and Google Scholar) that was most recently updated on October 10, 2018, followed by analysis of reference lists for additional articles.

#### SLEEP PATTERN DISORDERS IN HD

A typical sleep pattern in adults includes non-rapid eye movement sleep (NREM) divided into three stages (N1, N2, and N3): the sleep gradually deepens from N1 to N3, followed by rapid eye movement sleep (REM), during which suppression of postural muscle tone (atonia) is normally observed. NREM lasts about 60–90 min, REM lasts about 10–15 min, and the whole sleep cycle consisting of NREM and REM is repeated four or five times during a night's sleep.

There is no homogenous pattern of sleep disorders in patients with HD (36) and problems can occur in various areas. The literature usually reports insomnia, increased sleep onset latency, decrease in total sleep time, frequent nocturnal awakenings (27, 28, 37–40), REM sleep disorders (27, 38, 41), increased motor activity during sleep (27), decreased sleep efficiency (37, 38, 41, 42), and excessive daytime sleepiness (43). The problems mainly concern impaired maintenance of sleep: patients frequently experience awakenings that are accompanied by electroencephalographic abnormalities (27, 38, 41). Some studies also show that awakenings may be accompanied by anxiety and choreiform movements (44). Reduced N3 stage and an increased sleep spindle density are observed in HD patients compared to healthy controls (28). It was revealed that HD patients spend less time in the deep sleep phase, as indicated by the higher percentage of the first stage of sleep compared to the control group (27). Compared to healthy subjects, individuals with HD also spend more time in the NREM stage than in the REM stage (37), which may be related to REM sleep disturbances in HD.

Reports presenting research concerning sleep disorders during the REM stage in HD patients are ambiguous. Some authors report a lack, shortening, or delay of the REM stage (27, 37, 38), but no REM sleep disorders were detected in other studies on HD patients (28).

In the study conducted by Arnulf et al. (27), reduced duration of the REM sleep stage was noted in mutation carriers without core symptoms, as well as in patients in the first stage of the disease, for whom it intensified with progression. These authors also claim that reduced REM sleep may precede chorea and provide an early marker of the disease. According to Arnulf et al. (27), reduced REM sleep atonia in HD patients with decreased duration of REM sleep and reduced eye movement density (38) indicates that REM sleep executive systems, which are mainly located in the brainstem, are damaged in HD. This postulate is consistent with findings showing atrophy and degenerative changes in the brainstem in HD (45, 46). Another cause of REM sleep disturbance may be the dependence of REM sleep mechanisms on cerebral blood flow and oxygen metabolism, both of which are impaired in HD (27).

HD patients also demonstrate behavioral disorders related to the REM sleep stage (RBD, REM sleep behavior disorder) (43). Skeletal muscle atonia is one of the characteristic features of REM sleep, and its absence can lead to sudden body movements, e.g., hitting, kicking, tossing, and turning during sleep; these phenomena are accompanied by realistic dreams, often provoking a sense of danger (47, 48).

Some authors (27) suggest that in HD a mutant form of huntingtin may accumulate in areas that control muscle atonia during REM sleep, thus leading to their failure. These areas include the dorsolateral pons, the locus coeruleus, the reticular formation of the medulla, the pedunculopontine tegmental nucleus, and the hypothalamus. Cell loss in the locus coeruleus (49) and in the hypothalamus (45, 50, 51) in HD has been confirmed in studies.

RBD carries a risk of self-injury or injury to other persons in the bed. Therefore, systematic interviews with HD patients and their caregivers about sudden movements and aggressive actions during their sleep may be useful in medical care. A significant role can be played by counseling on RBD, securing the sleep environment and pharmacological treatment.

#### CIRCADIAN RHYTHM DISTURBANCES IN HD

Changes in the sleep–wake cycle, such as difficulty in falling asleep and maintaining sleep in HD, are considered to be a manifestation of circadian rhythm sleep disorder (CRSD) (52). Some authors argue that the REM sleep delay in HD patients could also be a symptom of circadian dysfunction (26, 53). Moreover, circadian rhythm disorders may underlie excessive sleepiness in HD. Excessive daytime sleepiness (EDS) involves a strong subjective feeling of sleepiness, difficulty in maintaining wakefulness, and a tendency to fall asleep at the wrong time or in the wrong place. The most severe form of excessive sleepiness is sleep attacks, i.e., sudden, uncontrolled incidences of falling asleep, both while performing monotonous tasks and during intensified activity such as riding a bike: a short sleep lasting a few minutes is followed by sudden awakening with amnesia of the episode (54). The results of studies on the prevalence of EDS in HD patients vary depending on the applied research method (e.g., interview with the patient, the results of the Epworth Sleepiness Scale, or multiple sleep latency tests). In some research, daytime sleepiness in the HD group was similar to that in the control group (55), whereas in the study by Videnovic et al. (43), the frequency of EDS occurrence was high and affected 50% of the patients. Higher EDS results were correlated with depression; therefore, it is believed that effective treatment of EDS in HD should take into account the coexisting depression (43). EDS may also be caused by sleep disordered breathing, but only two studies have shown abnormalities of the sleep respiration in few HD patients (56, 57), while no differences between the HD patients and control subjects with respect to sleep respiratory variables have been found in other studies (58, 59).

The circadian rhythm in mammals is regulated by the suprachiasmatic nucleus of the hypothalamus (SCN), which controls melatonin synthesis in the pineal gland. Some researchers (60, 61) have described changes in circadian melatonin secretion in HD, both in the early and late stages of the disease. In another study, mean daytime melatonin levels did not differ between the patients with HD and controls, but the evening rise in melatonin level in HD patients was delayed compared to individuals without neurological disorders (61, 62). A more recent study (63) found reduced 24-h averaged plasma melatonin concentration, flattening of the circadian rhythm of melatonin secretion, and greater spread of melatonin onset time in premanifest HD and moderate HD patients compared with control subjects.

Circadian abnormalities were manifested in an HD animal model in transgenic R6/2 mice in which daytime activity was increased and nighttime activity was reduced (64, 65). Disturbed night–day activity in R6/2 mice worsened with the progression of the disease, finally leading to complete disruption of the behavior. Circadian sleep–wake cycle disturbances in R6/2 mice might be a result of the disruption of circadian rhythm regulation by SCN. SCN is a neural stimulator of the sleep–wake rhythm, and its activity is based on the cyclic expression of genes crucial for the internal clock—the so-called clock genes. Abnormal expression of mammalian Period2 (mPer2) and brain and muscle aryl hydrocarbon receptor nuclear translocator-like protein-1 (BMAL1) clock genes in SCN, as well as in the striatum and motor cortex, was observed in R6/2 mice. The authors (64) claim that the data from the studies on R6/2 mice suggest that abnormalities in SCN activity leading to sleep disturbances can also be observed in HD patients. An effect on the circadian regulation mechanism of the pathological processes that occur in HD has not yet been recognized. It is known, however, that HD is associated with serious damage to the hypothalamus, where the SCN is located (50, 66). The normalization of the circadian rhythm slows the increase of cognitive impairments in animal models of HD and thus potentially may also have a significant effect on the inhibition of the rate of disease progression and improvement of life quality in patients with HD (67).

# PSYCHIATRIC DISORDERS AND SLEEP DISORDERS IN HD

Difficulties with falling asleep, shortness of sleep, and frequent awakenings may be the consequences of psychiatric disorders such as depression, mania, and anxiety, all of which affect many patients with HD. Depression with suicidal thoughts and tendencies develops in about 30% of people with HD and is often manifested even before the occurrence of motor symptoms (20, 21). According to some studies, the risk of depression and the rates of attempted suicide are higher in the period prior to the clinical HD phase and at its beginning (68). It is assumed that in most cases, depression in HD is endogenous and can result from damage to the medial caudate nucleus, from which projections go to the prefrontal and orbitofrontal areas. Research shows that depression in HD patients is associated with a reduction in glucose metabolism in the prefrontal and orbitofrontal cortex (69). Research on HD animal models also indicates a potential role of hypothalamus dysfunction as the neurobiological basis of depression in HD (70).

However, in addition to biological factors, psychosocial factors also play an important role in the development of depression in HD, which is an incurable disease that leads to disability and a significant reduction in quality of patients' life. Coping with the disease and the associated stress may significantly affect the incidence of depression and anxiety symptoms in patients with HD. Studies have shown that the genetic diagnosis itself is a strong stressor that can lead to suicidal attempts and psychiatric disorders requiring hospitalization, not only in people who are diagnosed as defective gene carriers, but also among those in whom the result was negative (71).

Much clinical evidence indicates a strong relationship between depression and sleep disorders among different groups of patients with extrapyramidal system diseases, e.g., PD (72). Videnovic et al.'s (43) study shows more frequent sleep disturbances in HD patients with coexisting depression than in HD patients without depression symptoms.

In the study of Aziz et al. (73), sleep disorders reported by HD patients were associated with depression but did not correlate with the severity of motor, behavioral, and cognitive symptoms that are typical of HD. In contrast, a delayed sleep phase was associated with both depression and the deterioration of cognitive and everyday functioning. In another study (74), subjective reports of sleep problems were associated with a greater severity of depressive symptoms but were not associated with neurocognitive symptoms in premanifest and symptomatic HD individuals.

Some sleep disorder symptoms noted in HD patients, such as sleep onset latency, shortness of sleep, and frequent awakenings, are consistent with the symptoms of sleep disorders observed in depression. However, there are also significant differences between the pattern of sleep disorders in HD and sleep disturbances in depression, especially regarding REM sleep. Depression is characterized by a decreased latency of REM sleep, increased REM sleep, and increased number and density of rapid eye movements (75), whereas increased REM sleep latency, reduced REM sleep, and a decreased density of rapid eye movement are observed in HD (27, 37, 38). The similarity of some symptoms and differences in other symptoms of sleep disorders in HD and in depression most likely result from heterogeneous causes of sleep disorders in HD. They can be both a consequence of damage to brain areas responsible for sleep processes, e.g., the brainstem in reference to REM sleep and endogenous depression, and the result of an emotional reaction to stressful situations.

The course and quality of sleep in HD patients can also be adversely affected by arousal resulting from mania or anxiety disorders (24, 76). As much as 10% of HD patients may experience mania or hypomania episodes, and over 50% experience anxiety (21). Excessive arousal may be the basis for a pattern of sleep disorders that occurs in some HD patients, manifested by difficulties in maintaining sleep, lack of compensatory sleep during the day, and falling asleep later compared to people not suffering from insomnia (27).

A significant relationship between sleep disorders and psychiatric disorders emphasizes the need for their early detection and treatment in HD. Effective treatment of psychiatric disorders should also lead to sleep normalization.

# MOTOR DISORDERS AND SLEEP DISORDERS IN HD

One of the causes of sleep problems in HD patients may be the involuntary movements that are one of the core symptoms of the disease (77, 78). It was originally believed that chorea ceases during sleep, but some studies have disproved this theory (37, 79). Patients are often not aware of their nocturnal movements, and information about their movements during sleep is obtained from their spouses or caregivers (72). As the disease progresses, further extrapyramidal symptoms develop, including bradykinesia, rigidity, myoclonus, and dystonic movements (involuntary movements of the limbs or the entire body). Rigidity and bradykinesia make it difficult to change position during sleep, and painful cramps and dyskinesias are the cause of awakening (73).

Studies demonstrate increased motor activity during sleep (55, 80, 81), including periodic limb movements (37, 82) and behavioral disorders during REM sleep (27) in HD patients compared to subjects not suffering from HD. More movements during sleep occurred in HD patients compared to patients with PD (79). Some HD patients suffer from restless legs syndrome (RLS), in which, due to unpleasant sensations, the patients move their legs to relieve the symptoms, as well as the syndrome of periodic limb movements in sleep (PLMS), characterized by brief, repetitive, and stereotyped limb movement during sleep. Both of these syndromes are associated with difficulties in falling asleep and frequent awakenings, which usually lead to constant fatigue, irritability, deterioration of concentration, and sleepiness during day (83). Comella (83) suggests that there is a genetic association between RLS and HD, but some studies have not revealed restless legs syndrome among examined HD patients (27). Evers and Stögbauer (84) hypothesize that dopaminergic transmission in HD may be reduced in parts of the basal ganglia, which are of particular importance for RLS. According to them, another explanation for the relationship between HD and RLS may be reduced sensitivity of the dopamine D1 and D2 receptors in HD patients with RLS. On the basis of their results, Savva et al. (85) postulate that RLS should be regarded as one of the early symptoms of HD that results from the neurodegenerative changes that occur in the disease.

# METHODS FOR DIAGNOSING SLEEP DISORDERS IN HD

The most commonly used methods used for screening and diagnosis of sleep disorders in HD are self-report tools designed for sleep assessment in the general population, such as the Pittsburgh Sleep Quality Index (PSQI), IRLS, The Berlin questionnaire, and the Epworth Sleepiness Scale (ESS) (86). However, methods for sleep assessment in the general population may not evaluate the characteristic symptoms of sleep disorders of people with HD due to the specificity of the disease. Despite the prevalence of sleep and circadian rhythm disorders in HD, the Unified Huntington Disease Rating Scale (UHDRS), which is designed to assess the functioning of HD patients, does not include questions about sleep or the circadian rhythm (87). Research results indicate that in motor disorders, including HD, non-motor symptoms such as sleep disorders are often not reported by the patient or family (82, 88), which may impede the detection and treatment of sleep disorders. For both refining diagnoses and expanding the scientific knowledge about sleep disorders in HD, it seems advisable to include sleep questions in existing assessment tools and creating new HD-specific sleep scales. The detailed self-report questionnaire for assessment of sleep problems specific for HD patients, designed similarly to the questionnaires for assessment of sleep problems in PD patients, was developed recently (29). This questionnaire contains 45 questions grouped into four subcategories: quality of sleep (e.g., difficulty falling asleep or maintaining sleep), motor activity (e.g., painful muscle cramps in the arms or legs causing waking at night), abnormal nocturnal behavior (e.g., acting out dreams, injury to self or others while dreaming), and other aspects of disturbed sleep (e.g., nocturia, numbness, paresthesia, sleep apnea, and daytime somnolence).

The use of objective methods that provide data based on physical phenomena such as polysomnography and actigraphy could be potentially beneficial in the diagnosis of sleep disorders in HD, but it is associated with a number of difficulties. Polysomnography is considered a gold-standard diagnostic tool, but the availability of this method is significantly limited by the fact that it is complex, time-consuming, and expensive. Actigraphy is designed to assess the state of wakefulness and sleep on the basis of patients' physical activity, but in the case of diseases with uncontrolled movements like HD, the test results may not be reliable. Studies using activity monitors show that during sleep, more movements and increased activity occur in HD patients compared to controls, which may mean that they either wake up at nighttime or make involuntary movements during sleep (55, 80). Fish et al. (79) observed that involuntary movements during sleep have a different character from those performed during waking—they are shorter and more fragmented. In combination with patients' declarations that they did not wake up or did so only occasionally (55), the results support the hypothesis that nocturnal HD activity is associated with involuntary movements during sleep rather than waking up. In addition, low concordance of actigraphy results with EEG records and sleep diaries was demonstrated in the sleep study in HD patients carried out by Townhill et al. (81).

Contradictory results indicating overestimation of total sleep time by activity monitors in seven HD patients was reported in the study of Maskevich et al. (89), but the subjects in the analyzed sample were in the presymptomatic and early phase of the disease and had very little chorea. Those discrepancies indicate the need for caution in the interpretation of results obtained in studies of sleep and circadian rhythms in HD patients using activity monitors. However, Adams et al. (90) suggest that further development and improvement of accelerator-based research methods will allow them to be used to accurately monitor and diagnose sleep disorders in HD.

Piano et al. (91) investigated the coherence of self-report and laboratory-based sleep assessment in 30 HD patients; the concordance of subjective sleep evaluation based on two selfreport measures, the Pittsburgh Sleep Quality Index and the Huntington Disease Sleep Questionnaire, was modest (Cohen κ = 0.375). Moreover, the results in both self-report measures demonstrated poor concordance with the Sleep Efficiently Index determined in a full-night laboratory-based videopolysomnographic recording (Cohen κ = 0.062 and 0.143, respectively) (91).

Proper diagnosis of sleep disorders in HD may require skillful combination of results from many different methods and integration of data from different sources as well as careful consideration of the specificity of this disease. The currently available methods should be considered imperfect as they require further development and improvement. There is also a need to develop diagnostic standards and guidelines for diagnosis of sleep disorders in HD.

# PHARMACOTHERAPY AND SLEEP DISORDERS IN HD

No systematic research has been conducted regarding the treatment of sleep disorders in HD patients; thus, the evidence base for the pharmacological treatment of HD is insufficient (92). The treatment is further complicated by the fact that many medications administered to HD patients to alleviate motor and psychiatric symptoms may change sleep architecture. Appropriate selection of drugs is also important because alleviating motor symptoms and depression themselves can improve sleep quality in HD patients.

Medications used in HD that are commonly associated with sleep disturbances are tetrabenazine, clonazepam, diazepam, riluzole, quetiapine, dosulepin, olanzapine, and venlafaxine. Drugs that may increase patients' activation include amantadine, sodium valproate, and L-DOPA (52, 93). Antidepressants (especially venlafaxine) should not be used in HD patients suffering from RBD because they can aggravate the symptoms. In HD patients with RBD, it is possible to administer clonazepam, but in some cases, it is not well tolerated by HD patients and can cause respiratory depression (27). All dopaminergic drugs, including L-DOPA, could have sedative effects that may lead to excessive sleepiness (94). This undesirable side effect can be mitigated by dose adjustment or changing the drugs administered to patients. If given at night, some nondopaminergic drugs with sedative side effects that are used in chorea treatment, such as olanzapine, may help HD patients with insomnia (95).

In general, drugs that can interfere with sleep should be used with caution. In cases in which there is no opportunity to change medication, stimulant drugs should be given early in the day and drugs with sedative side effects should be given in the evening.

Apart from preventing negative effects on sleep related to drugs used in HD patients, some pharmacological agents can be used in order to improve the quality of sleep. The application of short-acting benzodiazepines for insomnia in HD patients may provide improvement; however, these medications need to be used with caution because of their potential side effects. Mirtazapine, which is a noradrenergic and specific serotonergic antidepressant, is recommended for HD patients suffering simultaneously from depression and sleep disorders (96). If the HD patient has daytime sleepiness, treatment with stimulants such as modafinil can be beneficial; its effect has mainly been evaluated in PD patients, in whom it improved daytime sleepiness and cognitive functions, was well tolerated, and did not impair motor functions (53). One study evaluating the effects of modafinil on HD patients showed (97) that it did not improve cognitive function or mood, but increased alertness.

Melatonin may also be beneficial in treatment because it promotes sleep and normalizes the circadian rhythm. Research has shown its effectiveness in improving the quality of sleep in AD patients (98) and PD patients (99, 100). In HD, delayed melatonin secretion rhythm and decreasing melatonin levels may contribute to progressive neurodegeneration. Therefore, the use of melatonin or its receptors' agonists (ramelteon, agomelatine) could not only improve sleep quality but also enhance neuroprotection in HD patients (50). Regulation of the sleep–wake rhythm slows the increase of cognitive impairments in animal models of HD and thus potentially may also have a significant effect on the inhibition of the rate of disease progression and improvement of life quality in patients with HD (67).

Sleep quality in HD patients can also be improved by changing sleep behaviors. It is important to determine bedtime and wake time, avoid naps during the day, do regular physical exercise, maintain a proper diet, and limit the intake of caffeine, tobacco, and alcohol. In HD patients, relaxation training methods and cognitive-behavioral therapy can also be applied.

#### CONCLUSIONS

Different patterns of sleep disturbances are observed in HD patients: insomnia, difficulties in falling asleep, frequent nocturnal awakenings, and excessive daytime sleepiness are the most common sleep problems reported by patients with HD. In several HD studies, specific changes in sleep architecture and in circadian melatonin secretion were identified in laboratory testing.

Sleep disorders in HD have diverse and complex determinants, the most significant of which includes damage to brain areas that are responsible for the proper sleep pattern and circadian rhythm regulation. Sleep and circadian rhythm disorders in HD might also be associated with psychiatric disorders, especially depression, mania, and anxiety disorders. Another group of factors related to sleep disturbances in HD are involuntary movements and increased motor activity during sleep, and other motor disturbances. Moreover, the pattern of sleep disorders in HD is often altered by the many drugs used to alleviate the core symptoms of the disease, a substantial number of which may adversely affect sleep.

Therefore, studies on sleep disorders in HD are associated with numerous difficulties, as is also the case with other neurodegenerative diseases. However, identification of sleep disorders and more comprehensive recognition of their causes are necessary to ensure adequate care and treatment for patients.

#### REFERENCES


Disorders of sleep and the circadian rhythm in patients with HD can contribute to exacerbation of symptoms such as irritability and anxiety, as well as cognitive disorders that can interfere with daily functioning more than the motor symptoms. The effective treatment of sleep disorders in HD patients may play an important role in improving the quality of their life, helping them cope with everyday problems, and relieving the cognitive and emotional symptoms. It should also have a great impact on the quality of sleep of caregivers and thus create the opportunity to delay patients' institutionalization. More detailed knowledge of sleep problems in HD can also provide more profound insight into the nature of the neurodegenerative processes that occur in the disease (52).

Various methods are used for diagnosis of sleep disorders in HD patients. The most frequently used are self-reporting tools designed to assess sleep in the general population. However, the results obtained with these methods show low consistency with the results of laboratory sleep assessment, which makes their use in HD patients limited. Moreover, the proper diagnosis of sleep disorders in HD may require consideration of the specificity of the symptoms of this disease. The use of objective methods such as polysomnography and actigraphy might have potential benefits in the diagnosis of sleep disorders in HD patients, but the availability of polysomnography is limited because it is complex, time-consuming, and expensive, and the results of actigraphy may not be reliable in the case of diseases with uncontrolled movements, such as HD. Integration of data from various sources and results obtained from many different methods should allow the most comprehensive diagnosis of sleep disorders and circadian rhythm disorders in HD patients. The need to develop diagnostic standards and guidelines for the diagnosis of sleep disorders in HD should be considered one of the important challenges in improving the care of HD patients.

#### AUTHOR CONTRIBUTIONS

RH conceived and designed the study. RH and LK performed literature search and selection. LK created a database of literature relevant to the study. RH and LK wrote sections of the manuscript. Both authors contributed to manuscript revision and read and approved the submitted version.


impairment in Huntington's disease. *Parkinsonism Relat Disord* (2010) 16:345–50. doi: 10.1016/j.parkreldis.2010.02.009


**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 © 2019 Herzog-Krzywoszanska and Krzywoszanski. 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.*

# Schizophrenia and Hereditary Polyneuropathy: *PMP22* Deletion as a Common Pathophysiological Link?

*Dominique Endres1,2, Simon J. Maier1,2, Christiane Ziegler2, Kathrin Nickel1,2, Anne N. Riering2, Benjamin Berger3, Johann Lambeck3, Miriam Fritz3, Birgitta Gläser4, Friedrich Stock5, Michael Dacko6, Thomas Lange6, Irina Mader7,8, Katharina Domschke2 and Ludger Tebartz van Elst1,2\**

#### *Edited by:*

*Francesca Trojsi, Università degli Studi della Campania Luigi Vanvitelli Caserta, Italy*

#### *Reviewed by:*

*Antonio Lucio Teixeira, University of Texas Health Science Center at Houston, United States Daniele Coraci, Università Cattolica del Sacro Cuore, Italy Raffaele Dubbioso, Azienda Ospedaliera Universitaria Federico II, Italy*

#### *\*Correspondence:*

*Ludger Tebartz van Elst tebartzvanelst@uniklinik-freiburg.de*

#### *Specialty section:*

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

*Received: 23 December 2018 Accepted: 09 April 2019 Published: 02 May 2019*

#### *Citation:*

*Endres D, Maier SJ, Ziegler C, Nickel K, Riering AN, Berger B, Lambeck J, Fritz M, Gläser B, Stock F, Dacko M, Lange T, Mader I, Domschke K and Tebartz van Elst L (2019) Schizophrenia and Hereditary Polyneuropathy: PMP22 Deletion as a Common Pathophysiological Link? Front. Psychiatry 10:270. doi: 10.3389/fpsyt.2019.00270*

*1 Section for Experimental Neuropsychiatry, Department of Psychiatry and Psychotherapy, Medical Center—University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany, 2 Department of Psychiatry and Psychotherapy, Medical Center—University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany, 3 Department of Neurology, Medical Center—University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany, 4 Institute of Human Genetics, Medical Center—University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany, 5 Institute of Human Genetics, University Hospital Münster, Münster, Germany, 6 Department of Radiology, Medical Physics, Medical Center—University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany, 7 Department of Neuroradiology, Medical Center—University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany, 8 Department of Radiology, Clinic Schön Vogtareuth, Vogtareuth, Germany*

Background: Schizophrenic disorders are common and debilitating due to their symptoms, which can include delusions, hallucinations, and other negative symptoms. Organic forms can result from various cerebral disorders. In this paper, we discuss a potential association between schizophrenia and hereditary polyneuropathies (PNPs).

Case presentation: We present the case of a 55-year-old female patient with chronically paranoid–hallucinatory schizophrenia, severe cognitive deficits since the age of 30, and comorbid repeated focal pressure neuropathies beginning at age 20. At the age of 35, genetic testing revealed a deletion on chromosome 17p12 covering the peripheral myelin protein 22 gene (*PMP22*), which led to the diagnosis of hereditary neuropathy with liability to pressure palsy (HNPP). Cerebral magnetic resonance imaging showed internal atrophy, magnetic resonance spectroscopy found alteration of the glutamate and myo-inositol levels in the anterior cingulate cortex, neuropsychological testing showed deficits in working memory and psychomotor speed, and electrophysiological testing detected signs of sensorimotor demyelinating PNP (accentuated in the legs).

Conclusion: There may be an association between schizophrenia and HNPP. In observational studies, the deletion of interest (chromosome 17p12) was nearly 10 times more common in schizophreniform patients than in controls. This potential association could be pathophysiologically explained by the role of PMP22, which is mainly expressed in the peripheral nervous system. However, *PMP22* mRNA and protein can also be found in the brain. PMP22 seems to play an important role in regulating cell growth and myelination, functions that are disturbed in schizophrenia. Such a connection obviously cannot be clarified on the basis of one case.

**49**

Future studies should analyze whether patients with HNPP exhibit increased rates of psychotic disorders, and patients with schizophrenia and repeated focal pressure neuropathies should be examined for the *PMP22* mutation. Alternatively, the co-occurrence of schizophrenia and HNPP could be coincidental.

Keywords: hereditary polyneuropathy, schizophrenia, psychosis, *PMP22*, hereditary neuropathy with liability to pressure palsy

# BACKGROUND

Schizophrenia is a common disorder with a prevalence rate of about 1% (1). The clinical presentation is characterized by hallucinations, delusions, loss of self-boundaries, disorganized thinking and speech, cognitive deficits, lack of motivation, and social withdrawal (1). Secondary, organic forms can result from various cerebral disorders that are caused by genetic (22q11 deletion syndrome, cerebrotendinous xanthomatosis, Niemann– Pick type C, etc.), immunological (limbic encephalitis, anti-NMDA-R encephalitis, Hashimoto encephalopathy, etc.), infectious (neuroborreliosis, neurosyphilis, etc.), epileptic (paraepileptic psychosis, etc.), traumatic (traumatic brain injury), or neurodegenerative (frontotemporal dementia, etc). factors (2, 3).

The familial aggregation of schizophrenia is well established (1, 4). A variety of single-nucleotide polymorphisms (SNPs) in hundreds to thousands of genes have been implicated in the pathophysiology of the endogenous variants of schizophrenia (5). In these complex genetic forms, single genes only have a small effect size, and environmental factors are further important modulators. In contrast, secondary genetic forms are either monogenetic or oligogenetic: here, only one or a small number of genetic variants are important for the expression of clinical characteristics, such as copy number variants (CNVs), which are defined by deletions, duplications, or insertions of deoxyribonucleic acid (DNA) fragments, and chromosome aberrations (6, 7). In these forms, genetic variation in single genes has a high effect size, and environmental factors are considered to be less important (3). Secondary monogenic or oligogenic forms may potentially also occur in the context of hereditary polyneuropathies (hPNPs) (8). However, the association between schizophrenia and hPNPs is largely unknown. The hPNPs include isolated hPNPs (hereditary sensory neuropathy, hereditary sensory and autonomic neuropathy, and hereditary neuropathy with liability to pressure palsy or HNPP) and hPNPs in the context of systemic disorders (e.g., acute intermittent porphyria and Fabry's disease). These systemic disorders are characterized by a monogenic or oligogenic background similar to that of secondary schizophrenia.

Searching PubMed for "hereditary polyneuropathy AND (schizophrenia OR psychosis)" yielded only eight results (as of 23 December 2018). One observational study discussed an association between hereditary spastic paraparesis and schizophrenia (8). In addition, an association between hereditary spastic paraplegia and psychosis in a female patient due to dysmorphic changes in her corpus callosum has been described independently (9). No results were returned when the same literature research was performed using "HNPP AND (schizophrenia OR psychosis)". A nonsystematic literature search also showed that there might be a link between transthyretin-associated polyneuropathy (PNP) and schizophrenia or depression (10). Charcot–Marie–Tooth disorder, which is caused by a duplication of the *PMP22* gene, and coincident psychosis were reported in monozygotic twins (11, 12).

# CASE PRESENTATION

We present the case of a 55-year-old female Caucasian patient trained as an occupational therapist who has suffered from chronic paranoid–hallucinatory schizophrenia since the age of 30. She continuously showed positive symptoms with superimposed exacerbations. At the age of 34, she was forced to retire early from her career due to her illness. Her delusions included the idea that she had sinned and needed to die, and she perceived diverse signs as confirmation of these delusions. She suffered from auditory hallucinations (voices from God, the devil, and her dead partner or mother), visual hallucinations (visions of angels), and a loss of self-boundaries (believing that other people could read her thoughts). Negative symptoms included a lack of motivation, flattened mood, and social withdrawal. Cognitive impairment has been observed since the onset of psychotic symptoms, with inattention and increasing deficits in working memory.

Intermittently, the patient abused alcohol (at least four beers per day) and benzodiazepines, but no illegal drugs. Her consumption of these substances increased during psychotic exacerbations with social withdrawal. The early death of her life partner reinforced this withdrawal. Since the onset of the disease, she had attempted suicide 10 times. Therefore, there were frequent inpatient stays in different psychiatric hospitals. Neither various neuroleptic treatments with average or high doses of aripiprazole, amisulpride, clozapine, haloperidol, perazine, pimozide, quetiapine, and risperidone, nor anticonvulsive treatment with valproate as an augmentation strategy led to full remission. Under different combination treatments, the described symptoms persisted at a reduced level.

At age 20, the patient developed clinical signs of HNPP. Initially, she was quickly fatigued and showed transient hypoesthesia of the left arm and foot. She developed transient foot dorsi-flexor paresis twice on the right side. The symptoms occurred after mechanical pressure on the corresponding body regions. In the further course of the disease, she developed transient left brachial plexus paresis at the age of 32 and again at the age of 35. When she was 35, genetic testing revealed a deletion on chromosome 17p12 involving the peripheral myelin protein 22 gene (*PMP22*), confirming the diagnosis of HNPP. The most recent neurological examination showed a discrete foot dorsi-flexor paresis on the right side (Medical Research Council score M4), absent Achilles tendon reflexes on both sides, slight pallhypesthesia of the lower extremity accentuated on the right side (5/8 on the right versus 6/8 on the left), and an ataxic and unsteady gait.

#### Developmental, Somatic, and Family History

The patient's developmental history was negative for *in utero* or birth complications, febrile convulsions, inflammatory brain diseases, and cerebral contusions. There was no evidence of any neurodevelopmental disorder. In the first two decades, the patient's premorbid personality showed dependent traits. Her somatic medical history included only hypothyroidism, which was diagnosed at age 43. Her family history was positive for schizophrenia and HNPP. Her father's half-brother suffered from schizophrenia. The patient's older brother (transient hypesthesia of parts of one hand; side and explicit location unclear) and father (paresis of the shoulder abductors) potentially also suffered from HNPP (no genetic diagnostics performed, clinical reports not available). Her younger brother and her mother were healthy.

#### Basic Investigations and Magnetic Resonance Spectroscopy

The serum, cerebrospinal fluid, and cerebral magnetic resonance imaging (cMRI) investigations did not show any evidence of an immunological cause for schizophrenia or PNPs. The electrophysiological tests revealed sensorimotor, demyelinating PNP (accentuated in the legs). Neurosonography showed bilateral enlargement of the median and ulnar nerve at typical sides of entrapment syndromes. Neuropsychological testing showed considerable deficits in working memory and a mild deficit in psychomotor speed. The electroencephalography (EEG) showed rare intermittent slowing. The cMRI showed generalized atrophy (**Figure 1**). Magnetic resonance spectroscopy (MRS) was performed in the dorsolateral prefrontal cortex (DLPFC), the dorsal anterior cingulate cortex (dACC), and the pregenual anterior cingulate cortex (pACC) using a MEGA PRESS sequence (repetition time = 1,500 ms, echo time = 68 ms, flip angle = 90°) and the acquired spectra were quantified with the software LCModel. Glutamate concentrations in the dACC and pACC were relatively high and outside the 90% reference intervals. The myo-inositol concentrations in the pACC were relatively low and within the 90% reference interval (**Figure 2**). The DLPFC measurement displayed poor spectral quality and was therefore not analyzed. The diagnostic findings are summarized in **Table 1**.

#### Genetic Investigations: Cytogenetic and Array Analysis

MLPA (multiplex ligation-dependent probe amplification) analysis revealed a heterozygous deletion of the *PMP22* gene on chromosome 17p12. To delineate the extent of this deletion, we additionally performed conventional karyotyping and array CGH (comparative genomic hybridization).

Conventional R-banded karyotypes from the patient were analyzed according to standard protocols with a resolution of approximately 500 bphs and revealed a structurally and numerically normal female karyotype (46,XX) in all 28 metaphases examined. Furthermore, the genomic DNA of the patient was examined by microarray analysis (CytoSureTM constitutional v3 array 180k; Oxford Gene Technology) according to the manufacturer's instructions. After hybridization, the array was scanned with the SureScan microarray scanner (Agilent); the results were analyzed using CytoSure interpret software v.4.9 (Oxford Gene Technology) and the Genome Reference Consortium human genome GRCh37 (hg19). Molecular karyotyping revealed a heterozygous deletion of approximately 1.33 Mb (125 contiguous oligonucleotides) out of the chromosomal region 17p12 (karyotype after the International System for Human Cytogenetic Nomenclature, 2016: arr[GRCh37] 17p12(14111972\_15442257)x1). The deletion encompasses i.a. the genes *COX10*, *HS3ST3B1*, *PMP22*, *TEKT2*, and *CDRT4*. A deletion of that extent in 17p12 is found in nearly 80% of patients with HNPP (13). Further chromosomal deletions or duplications that might have etiologically contributed to our patient's disease were not detected.

#### Differential Diagnosis

The patient's psychiatric symptoms were compatible with paranoid–hallucinatory schizophrenia (ICD-10: F20.0). The PNPs in combination with the deletion of the *PMP22* gene led to the diagnosis of HNPP (ICD-10: G60.0). Due to the potential secondary schizophrenia in the context of HNPP, a psychotic

#### TABLE 1 | Current diagnostic findings.


*CSA, cross-sectional area*

*\*Neuropsychological assessment was interrupted repeatedly due to the patient's spontaneous delusional utterances and experiences. This might reduce the validity of test results. \*\*Below average.*

disorder with delusions due to a known physiological condition might ultimately be diagnosed (ICD-10: F06.2).

# DISCUSSION

In this paper, we present the case of a female patient with chronic paranoid–hallucinatory schizophrenia with poor response to therapy and comorbid HNPP.

#### HNPP, *PMP22* Deletion, and Central Nervous System Involvement

HNPP is an autosomal-dominant, peripheral neuropathy characterized by repeated and transient episodes of focal pressure neuropathies at compression-exposed sites (e.g., brachial plexus, sciatic nerve) (14). Our patient suffered from repeated foot flexor paresis and twice experienced brachial plexus paresis. In HNPP patients, nerve conduction study (NCS) often reveals demyelinating PNP with nerve suffering located predominantly in entrapment sites (15). In our patient, NCS showed findings compatible with sensorimotor, demyelinating PNPs with an entrapment syndrome in the right cubital tunnel. Histologically, sural nerve biopsies of HNPP patients typically display tomaculae, which are redundantly overfolded layers in the myelin coat (16). Genetically, a 1.5-Mb deletion on chromosome 17p12, including the *PMP22* gene, is found in most cases (17). Molecular karyotyping in our patient showed a heterozygous deletion of 1.33 Mb.

FIGURE 2 | Magnetic resonance spectroscopy (MRS) findings: We present the findings of the dorsal anterior cingulate cortex (dACC; the two upper rows) and the pregenual anterior cingulate cortex (pACC; lower rows). The patient's values (red line) are compared to the values of a healthy control group measured in another study (N = 43, mean age: 35, gender ratio: 14 females/29 males). The green line shows the 90% reference interval of the control group. For Glx in the dACC and Ins in the pACC the green (90% reference interval) and the red line (patient's values) are superimposed, therefore only the red line is visible. The measured metabolite concentrations have been corrected for MRS voxel composition (content of gray matter, white matter, and cerebrospinal fluid), as well as for the influence of age and differences in the signal-to-noise ratio of the MRS measurements. Abbreviations: GABA, gamma-aminobutyric acid including coedited macromolecules; NAA, N-acetylaspartate; Glx, glutamate and glutamine; Glu, glutamate; Ins, myo-inositol; GPC, glycerophosphorylcholine.

The *PMP22* gene is mainly expressed in the peripheral nervous system (PNS); however, mRNA and protein can also be found in the central nervous system (CNS) (18). Recent studies suggested the important involvement of the CNS in most patients with *PMP22* deletion (18, 19). This is supported by prolonged latencies of visual evoked potentials, neurochemical alterations with decreased *N*-acetylaspartate (NAA) and creatine (Cre) concentrations, white matter (WM) volume reduction detected by cMRI, cognitive impairment (in 70% of patients), and fractional anisotropy alteration in several WM regions (e.g., in the columns of the fornix) (18–20). A large family study described CNS involvement and WM lesions predominantly in the subcortical frontal WM (21). Therefore, in line with the studies described, we hypothesize that our patient's schizophreniform symptoms may represent CNS involvement. This consideration is clinically supported by an insufficient therapy response to neuroleptics, severe cognitive deficits, EEG slowing, internal brain atrophy, and neurometabolic alterations (high glutamate in the dACC and pACC and low myo-inositol in the pACC). Compared with the patient group from Chanson and colleagues, we also found brain atrophy and neuropsychological deficits; however, in the presented case, NAA concentration and visual evoked potentials were normal, and DTI measurements were not performed (18). A causal relationship between the poor response to therapy and the *PMP22* deletion remains speculative and is not proven by the case report. However, our hypothesis is also supported by several epidemiological and pathophysiological ideas, which we discuss in detail in the following paragraphs along with potential limitations.

#### *PMP22* Deletion in Patients With Schizophrenia

Copy number variation in the chromosomal region 17p12 has repeatedly been implicated in schizophrenia (22–25). In a study analyzing the involvement of rare CNVs in 471 patients with schizophrenia, a deletion on 17p12 was found in two patients but not in controls. Therefore, the authors reanalyzed the data from two recent, large CNV studies of schizophrenia and found a *PMP22* deletion in 6 out of 4,618 (0.13%) patients and 6 out of 36,092 (0.017%) controls (26–28). The data demonstrate that the described 17p12 deletion can be found nearly 10 times more often in schizophrenic patients compared to healthy controls (28). Moreover, there is at least one case report presenting a patient with schizophrenia in combination with mental retardation and *PMP22* deletion without hPNPs (29).

#### Pathophysiological Considerations

PMP22 is a small, hydrophobic membrane glycoprotein that is mostly expressed by Schwann cells and comprises 2–5% of PNS myelin proteins in humans (16, 30). *PMP22* mRNA and protein were also detected in the CNS, specifically in most parts of the brain (especially the corpus callosum) and the spinal cord. Changes in *PMP22* mRNA may explain myelin abnormalities because it plays a role in the regulation of cell growth (even in the absence of protein), which was first described in fibroblasts (30, 31). Reductions in *PMP22* mRNA have been observed in the hippocampus and anterior cingulate cortex in the postmortem brains of patients with schizophrenia (32). The PMP22 protein was restricted to a few areas [anterior horn and pia mater of the spinal cord, preganglionic sympathetic neurons; (33)]. In summary, *PMP22* mRNA and protein are important for the regulation of cell growth and the maintenance of myelin integrity and therefore ensure the propagation of action potential (16). Heterozygous deletion of the *PMP22* gene results in a loss-offunction phenotype (16) and altered myelination, as found in patients with schizophrenia (34).

#### Clinical Importance of Case Studies and Limitations

It is important to note that the co-occurrence of schizophrenia and HNPP could be coincidental, which is supported by the absence of schizophreniform symptoms in our patient's brother and father, who both very likely suffered from HNPP. However, genetic analyses of the patient's brother and father were not performed to our knowledge. So, we can only speculate whether the discrepancy in psychiatric symptoms between our patient and her brother and father could be due to a variable expression and an incomplete penetrance, respectively, of the genetic effect of the *PMP22* deletion. Besides environmental factors, further genetic variants might have contributed to our patient's disease, particularly since the presently identified chromosomal deletion additionally encompasses *COX10*, *HS3ST3B1*, *TEKT2*, and *CDRT4*. However, the detected radiological and neuropsychological findings can also be found in schizophrenia; therefore, they are not clearly associated with the PMP22 deletion (34).

Case studies reporting such potential associations are essential to inspire further clinical trials (35). Retrospective studies analyzing psychiatric comorbidity in patients with HNPP could demonstrate whether there is a relevant association between the disorders and confirm or refute the pathophysiological role of *PMP22* deletion in a subgroup of patients with schizophrenia.

# CONCLUSION

There may be an association between schizophrenia and HNPP, which could be explained by the role of *PMP22* in regulating cell growth and myelination. If such an association existed, which, of course, cannot be clarified from one case, then it might explain the poor control of psychiatric symptoms in our patient. But *PMP22* deletion does not necessarily mean causality for the emergence of psychotic symptoms. For further clarification, patients with psychotic disorders and repeated transient focal neurological symptoms (which could be explained by pressure neuropathies) should be examined for *PMP22* gene variation. Future studies should analyze whether patients with HNPP exhibit increased rates of schizophrenia. Alternatively, the co-occurrence of schizophrenia and HNPP could be coincidental.

#### ETHICS STATEMENT

The patient has given her signed, written informed consent for this case report, including the images presented, to be published.

# AUTHOR CONTRIBUTIONS

DE and LT treated the patient. DE performed the data research and wrote the paper. SM, KN, MD, TL, and IM performed and interpreted

#### REFERENCES


the magnetic resonance spectroscopy. FS and BG performed the genetic testing. CZ and KD supported the interpretation of genetic findings. BB, JL, and MF performed and interpreted the electrophysiological measurements, neurosonography, neurological examination, and CSF measurements. IM performed and interpreted the cMRI. AR performed the neuropsychological testing. All authors were critically involved in the theoretical discussion and composition of the manuscript. All authors read and approved the final version of the manuscript.

#### FUNDING

The article processing charge was funded by the German Research Foundation (DFG) and the University of Freiburg in the funding program Open Access Publishing.


dysfunction. *Arch Gen Psychiatry* (2003) 60(5):443–56. doi: 10.1001/ archpsyc.60.5.443

35. Albrecht J, Meves A, Bigby M. Case reports and case series from *Lancet* had significant impact on medical literature. *J Clin Epidemiol* (2005) 58(12):1227–32. doi: 10.1016/j.jclinepi.2005.04.003

**Conflicts of Interest Statement:** BB received travel grants and/or training expenses from Bayer Vital GmbH, Ipsen Pharma GmbH, Norvartis, Biogen GmbH, and Genzyme, as well as lecture fees from Ipsen Pharma GmbH, Alexion Pharma GmbH, Merck, Sanofi Genzyme, and Roche. IM received lecture fees from UCB Pharma GmbH, Germany. LTvE received advisory boards, lectures, or travel grants within the last 3 years: Eli Lilly, Janssen-Cilag, Novartis, Shire, UCB, GSK, Servier, Janssen, and Cyberonics.

The remaining 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 © 2019 Endres, Maier, Ziegler, Nickel, Riering, Berger, Lambeck, Fritz, Gläser, Stock, Dacko, Lange, Mader, Domschke and Tebartz van Elst. 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(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.*

# Cognitive Syndromes and C9orf72 Mutation Are Not Related to Cerebellar Degeneration in Amyotrophic Lateral Sclerosis

Monica Consonni<sup>1</sup> \*, Eleonora Dalla Bella<sup>1</sup> , Anna Nigri<sup>2</sup> , Chiara Pinardi<sup>2</sup> , Greta Demichelis<sup>2</sup> , Luca Porcu<sup>3</sup> , Cinzia Gellera<sup>4</sup> , Viviana Pensato<sup>4</sup> , Stefano F. Cappa5,6 , Maria Grazia Bruzzone<sup>2</sup> , Giuseppe Lauria1,7 and Stefania Ferraro<sup>2</sup>

<sup>1</sup> 3rd Neurology Unit and Motor Neuron Diseases Centre, Fondazione IRCCS Istituto Neurologico Carlo Besta, Milan, Italy, <sup>2</sup> Neuroradiology Department, Fondazione IRCCS Istituto Neurologico "Carlo Besta," Milan, Italy, <sup>3</sup> Laboratory of Methodology for Clinical Research, Oncology Department, IRCCS Istituto di Ricerche Farmacologiche Mario Negri, Milan, Italy, <sup>4</sup> Unit of Genetics of Neurodegenerative and Metabolic Diseases, Motor Neuron Diseases Centre, Fondazione IRCCS Istituto Neurologico Carlo Besta, Milan, Italy, <sup>5</sup> Institute for Advanced Study-IUSS Pavia, Pavia, Italy, <sup>6</sup> IRCCS Centro San Giovanni di Dio Fatebenefratelli, Brescia, Italy, <sup>7</sup> Department of Biomedical and Clinical Sciences "Luigi Sacco", University of Milan, Milan, Italy

#### Edited by:

Foteini Christidi, National and Kapodistrian University of Athens, Greece

#### Reviewed by:

Michael Hornberger, University of East Anglia, United Kingdom Mattia Siciliano, Second University of Naples, Italy

#### \*Correspondence:

Monica Consonni monica.consonni@istituto-besta.it; monicaconsonni@libero.it

#### Specialty section:

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

Received: 15 February 2019 Accepted: 17 April 2019 Published: 10 May 2019

#### Citation:

Consonni M, Dalla Bella E, Nigri A, Pinardi C, Demichelis G, Porcu L, Gellera C, Pensato V, Cappa SF, Bruzzone MG, Lauria G and Ferraro S (2019) Cognitive Syndromes and C9orf72 Mutation Are Not Related to Cerebellar Degeneration in Amyotrophic Lateral Sclerosis. Front. Neurosci. 13:440. doi: 10.3389/fnins.2019.00440 Objective: The notion that cerebellar pathology may contribute to cognitive impairment in ALS, especially in patients with C9orf72 repeated expansion, has been inconsistently reported. This study aimed exploring the relationship between cerebellar involvement, cognitive impairment and C9orf72 repeated expansion of patients with ALS.

Methods: Quantitative in vivo assessment of cerebellar lobules has been investigated in 66 non-demented patients with ALS and 28 healthy controls (HCs). Pathologic C9orf72 repeated expansion was found in 13 patients. Mild cognitive and/or behavioral impairment was diagnosed in 22 C9orf72 negative ALS patients. Measures of cortical volume (CV) and cortical thickness (CT) of cerebellar lobules of all participants were used for Principal Component Analysis (PCA) to identify clusters of lobular measures highly correlated with each other. PCA outcomes were used for between group comparisons and correlation analyses with neuropsychological and clinical features.

Results: Disease severity measured with ALS functional rating scale and index of disease progression rate significantly correlated with CV reduction of the second PCA cluster loading CV measures of anterior lobules. In all patients, cognitive impairment, measured with verbal fluency, was related to CV reduction of the third cluster comprising posterior lobules. No specific cortical thinning or volume reduction of cerebellar clustering patterns could be detected in ALS subgroups.

Conclusion: Our data show that specific patterns of subregional cerebellar involvement are associated with physical disability or cognitive impairment in ALS, in line with the topographic organization of the cerebellum. However, there was no specific correlation between cerebellar degeneration and cognitive syndromes or C9orf72 mutations.

Keywords: cerebellum, amyotrophic lateral sclerosis, cortical thickness, cortical volume, cognitive impairment

# INTRODUCTION

fnins-13-00440 May 10, 2019 Time: 16:37 # 2

The human cerebellum has been long recognized as playing an essential role in motor control and coordination and it has also become increasingly aligned with cognitive and affective processing (Schmahmann and Sherman, 1998; Buckner, 2013). Functional neuroimaging and connectivity studies in healthy humans suggest a topographic organization in the cerebellum: sensorimotor functions are represented in anterior lobules I–V and inferior lobule VIII, while cognitive processing is supported by posterior lobules (Stoodley and Schmahmann, 2009). Cerebellar abnormalities have been found not only in primary cerebellar injury or degeneration, but also in many psychiatric and neurological diseases without ataxia, including Amyotrophic Lateral Sclerosis (ALS) (Gellersen et al., 2017; Carass et al., 2018).

ALS is a fatal multisystemic neurodegenerative disorder mainly characterized by progressive loss of upper and lower motor neurons with cognitive and/or behavioral disturbance in at least half of cases. ALS-associated cognitive syndrome includes impairment of executive function, often with perseveration, distractibility or inattention, language deficit and social cognition impairment (Strong et al., 2017). Personality change might also occur with apathy, blunting of affect or disinhibited and inappropriate behavior (Strong et al., 2017). In a subset (15%) of patients these cognitive and neuropsychiatric changes are severe enough to fulfill criteria for the behavioral variant of frontotemporal dementia (bvFTD) (Phukan et al., 2012). ALS patients with the C9orf72 hexanucleotide repeat expansion (C9+ALS) had significantly more co-morbid bvFTD features than those without (Byrne et al., 2012). Cognitive and neuropsychiatric features of ALS have largely been attributed to changes in the frontotemporal and insular cortex (Tsujimoto et al., 2011; Agosta et al., 2016; Westeneng et al., 2016; Consonni et al., 2018a,b; Christidi et al., 2018a), although increasing evidence suggest hippocampus and subcortical region degeneration also plays a role (Bede et al., 2013b; Machts et al., 2015; Westeneng et al., 2016), especially in C9+ALS patients (Bede et al., 2013a; Agosta et al., 2017; Floeter et al., 2018; Floeter and Gendron, 2018).

Evidence of cerebellar involvement in ALS is less commonly reported (Prell and Grosskreutz, 2013). Structural changes of the cerebellum in ALS have been described as regional white matter changes (Bede et al., 2015) and decrease gray matter volume (Thivard et al., 2007; Gellersen et al., 2017; Christidi et al., 2018a,b). Cerebellar changes in ALS and FTD have been specifically linked to the C9orf72 repeat expansion (Mackenzie et al., 2013), raising the question whether cerebellar degeneration is an important feature of C9+ALS.

In order to explore the relationship between cerebellar involvement, cognitive impairment and C9orf72 repeated expansion of patients with ALS, quantitative in vivo assessment of cerebellar lobules has been investigated in a large sample of non-demented ALS patients stratified on the basis of cognitive and/or behavioral involvement and C9orf72 mutation.

# MATERIALS AND METHODS

## Participants

Sixty-six non-demented patients diagnosed with probable or definite ALS, according to El Escorial revised criteria (Brooks et al., 2000) and 28 healthy control (HC) volunteers were included in the study. Exclusion criteria were comorbid frontotemporal dementia (Rascovsky et al., 2011), Alzheimer's disease (NIA-AA), evidence of another neurologic condition affecting cognition (e.g., head trauma, hydrocephalus, and vascular disease), drug or alcohol abuse, mental retardation, primary psychiatric disorders, and other severe medical conditions. The review board of the Fondazione IRCCS Istituto Neurologico Carlo Besta of Milan approved the study. Each subject was enrolled after giving written informed consent.

# ALS Subgroups

Patients were assigned to subgroups on the basis of the genetic screening for C9orf72 mutations and on the basis of the presence of cognitive (ALSci) and/or behavioral (ALSbi) impairment as defined by recent guidelines (Strong et al., 2017). Accordingly, a cognitive screening protocol was administer to all participants (Consonni et al., 2016; **Table 1**). Our cohort was composed of 13 patients with a pathological C9orf72 expansion (C9+ALS) and 53 non-affected C9orf72 patients (C9-ALS). Among C9-ALS patients, 31 patients had normal cognitive profile (C9-ALScn) and 22 had ALSci and/or ALSbi (C9-ALSimp).

# Magnetic Resonance Imaging

All patients and healthy participants underwent a magnetic resonance imaging (MRI) session within a week of neuropsychological assessment, on a Philips Achieva 3.0 T. The MRI protocol included a 3D T1-weighted sequence (FFE, 240 sagittal slices, TR = 9.9 ms, TE = 4.6 ms, matrix 240 × 240, voxel size = 1 × 1 × 1 mm3, and flip angle = 8). In order to have a quantitative evaluation of cerebellum lobules representing the current state of the art, we used the CERES automated cerebellum parcellation algorithm (Romero et al., 2017). With this tool, the MRI images are denoised, corrected for inhomogeneity, rigid-body registered to Montreal Neurological Institute (MNI) template, cropped around the cerebellum area, and normalized to the MNI cropped cerebellum atlas. An automatic multi-atlas patch-based segmentation was then applied to obtain cerebellar cortical thickness (CT) and volume (CV), expressed in percentage of total intracranial volume, for each lobule (Manjón and Coupe, 2016).

#### Cerebellar Cluster Identification

CT and CV measures over twelve cerebellar lobules were obtained. Principal Component Analysis (PCA) was then applied as an unbiased method to reduce the number of variables for subsequent between-group and correlation analyses. PCA with oblimin rotation was used to identify the clustering pattern (of highly correlated variables) of cerebellar CT and CV measures of all subjects. For this purpose, for each CT and CV measure, z-scores were calculated by subtracting the mean CT and CV of


TABLE 1 | Demographic, clinical and neuropsychological data (mean ± standard deviation).

ALSbi, ALS patients fulfilling Strong criteria for behavioral impairment; ALSci, ALS patients fulfilling Strong criteria for cognitive impairment; ALScbi, ALS patients fulfilling criteria for ALSci and ALSbi; ALSFRS-R, Revised ALS Functional Rating Scale; FBI, Frontal Behavioral Inventory; RAVLT, Rey Auditory Verbal Learning Test; SET, Storybased Empathy Task; and ns, not significant difference (p > 0.05). ∗∗∗p < 0.001; ∗∗p < 0.005; <sup>∗</sup>p < 0.05.

TABLE 2 | Pattern matrix of PCA results for cerebellar lobar values of Cortical Thickness (CT) of 94 subjects [66 patients with ALS and 24 healthy controls (HCs)].


The loading values in bold indicate the elements that contribute the greatest variability to PCA clusters. ALScn, ALS patients with a cognitive and behavioral normal profile; ALSimp, ALS patients fulfilling Strong criteria for behavioral and/or cognitive impairment; C9+ALS, ALS patients with C9orf72 repeat expansions; and HC, Healthy Controls.

HC subjects from the patient's CT and CV, and then dividing the difference by the HC group standard deviation. All variables were normally distributed and had Kaiser-Meyer-Olkin (KMO) values higher than 0.5. Clusters with eigenvalues over the Kaiser criterion of one were retained. Cluster scores were calculated with the Anderson-Rubin method and used for between group and correlation analyses (Field, 2013).

#### Statistics

The Kolmogorov–Smirnov test was used to test the normality of the distribution of demographic, clinical, neuropsychological and neuroimaging data. Fisher Exact test, Kruskal–Wallis and Mann–Whitney U tests were used to test between group differences (i.e., C9-ALSimp, C9-ALScn, C9+ALS, and HC). The variability of cerebellar cluster loadings of groups was analyzed by ANCOVA, with age as covariates. Simple contrasts were then used to test specific group comparisons. Partial correlation analyses (correcting for age and education) were used to explore the relationship of cerebellar cluster scores with neuropsychological and clinical profiles in all patients as a whole group. Significant correlations were then bias corrected and accelerated bootstrap 95% confidence intervals (CIs) were computed with 1000 bootstrap equally sized samples obtained by randomly resampling replacement from the original data. Partial correlations with p < 0.05 and CIs not crossing "0" were reported. IBM SPSS Statistics (version 21) was used to perform analyses.

# RESULTS

# Clinical, Demographic and Neuropsychological Data

Descriptive statistics are summarized in **Table 1**. C9-ALScn, C9- ALSimp, C9+ALS, and HC groups did not differ in to age, handedness, gender, but education (H = 9.019, p = 0.029). The C9+ALS group had a slightly lower education level than the HC group (U = 93.0, p = 0.012). Disease duration, functional

disability assessed with the ALSFRS-R, progression and type of onset were similar across ALS subgroups. HC and C9- ALScn groups had similar neuropsychological performances. The occurrence of cognitive and/or behavioral impairment is higher for C9+ALS patients than C9-ALS patients (77% vs. 41%; p = 0.031). The C9-ALSimp group compared with HC and C9-ALScn groups had lower performances in all cognitive domains (**Table 1**).

#### Cerebellar Clustering Patterns

**CTz.** The KMO measure tested the sample adequacy of 12 cerebellar lobular CTz values for PCA (KMO = 0.708). The rotation converged on 23 iterations. Four clusters had eigenvalues over Kaiser criterion of one and in combination explained 78.41%

TABLE 3 | Pattern matrix of PCA results for cerebellar lobar values of Cortical Volumes (CV) of 94 subjects (66 patients with ALS and 24 HCs).



The loading values in bold indicate the elements that contribute the greatest variability to PCA clusters. ALScn, ALS patients with a cognitive and behavioral normal profile; ALSimp, ALS patients fulfilling Strong criteria for behavioral and/or cognitive impairment; C9+ALS, ALS patients with C9orf72 repeat expansions; and HC, Healthy Controls.

of the variance (**Table 2**). Cluster 1 included CTz values of lobule VIIIA, VIIIB, and IX; Cluster 2 included CTz values of Crus II and VIIB, Cluster 3 loadings were CTz values of lobule IV, V, VI, and Crus I; cluster 4 included lobule I&II and III and X.

**CVz.** The KMO measure tested the sample adequacy of 12 cerebellar lobular CV for PCA (KMO = 0.671). The rotation converged on 10 iterations. Four clusters had eigenvalues over Kaiser criterion of one and in combination explained 62.13% of the variance (**Table 3**). Cluster 1 included CVz of lobule VIIIB, IX, and X; Cluster 2 included CVz of anterior lobules (I&II, III, IV, and V); Cluster 3 loadings were CVz values of CrusII, VIIB and VIIIA; cluster 4 included CVz of lobule VI and Crus I.

We did not find group differences in cluster scores of CTz and CVz measures (**Table 4**). Correlation analyses revealed that only CVz scores of the second and third cluster were related to functional disability measured with the ALSFRS-R in all patients (r = 0.283, p = 0.024, CIs: 0.043–0.468; r = 0.281, p = 0.026, CIs: 0.053–0.467). Cluster 2 CVz scores were also related to the progression rate (r = −0.248, p = 0.050, CIs: −0.087 – −0.403). Cluster 4 CVz scores were inversely related to the phonemic fluency index (r = −0.335, p = 0.010, CIs: −0.157 – −0.535).

#### DISCUSSION

We tested the relationship between cerebellar degeneration, cognitive syndromes and C9orf72 mutation in ALS patients. The notion that cerebellum contributes to non-motor disorders, such as impairments in language, attention and social cognition is well recognized (Buckner, 2013). Consistently, we performed a quantitative in vivo assessment of cerebellar lobular CT and CV in a large sample of non-demented ALS patients stratified on the basis of cognitive and/or behavioral impairment and C9orf72 repeat expansion. To this aim, we used an unbiased reduction method, the PCA, which allows projecting data into a lower-dimensional space thus reducing the type I error and identifying clusters of highly correlated variables with each other (Field, 2013).

We identified 4 clusters of regional cerebellar CT and demonstrated their scores were similar across ALS subgroups


n.s., not significant differences (p > 0.05).

and HC individuals. Moreover, the lack of significant correlations between cerebellar CT values, neuropsychological performances and clinical features in patients suggests that the cerebellar cortical thinning has a poor sensibility to detect cognitive impairment in ALS. Cognitive and neuropsychiatric features of ALS seem to be associated instead with thinning in frontotemporal and insular cortex (Schuster et al., 2014; Agosta et al., 2016; Consonni et al., 2018a,b).

By means of lobar CV measures, we identified four clustering patterns. The clustering of the cerebellum respected lobular anatomical boundaries and functional organization. Cluster 1 included posterior regions (VIIIB-IX, X), cluster 2 the sensorymotor anterior regions (I-V), cluster 3 (CrusII-VIIIA), and 4 (VI-CrusI) included mainly cognitive posterior lobules. Contrary to our expectation, no significant group differences in cluster scores were found, but, as explorative analysis, a weak but significant correlation between CV measures of the fourth cluster and phonemic fluency indexes of ALS patients. Verbal fluency is a widely recognized marker of dysexecutive impairment in ALS (Strong et al., 2017). Specifically, poor verbal fluency was related to reduced CV of lobule VI and CrusI. This is consistent with previous studies documenting that lobules VI and Crus I are involved in executive functions, such as working memory, planning, organizing, and strategy formation (Stoodley and Schmahmann, 2009). According to a meta-analysis of voxelbased morphometry in cerebellum (Gellersen et al., 2017), the largest cluster of gray matter reduction in ALS spanned parts of the vermis and neighboring regions in left lobule VI, Crus I, and Crus II. This suggests that cerebellar involvement in ALS may be related to progressive cognitive impairment. However, as in other studies (Schönecker et al., 2018), we did not find differences in cerebellar volume within ALS cognitive subgroups, albeit specific patterns of cerebellum atrophy in ALS-FTD patients have been previously reported (Tan et al., 2014; Omer et al., 2017; Christidi et al., 2018a,b). We could not exclude that discrepancies between our findings and those reported by others are attributed to differences in sample sizes and patients' clinical features, as well as different methodological (e.g., Surface-Based Analysis vs. Voxel-Based Morphometry) and statistical approach and whether the whole cerebellum or focal cerebellar regions were addressed.

We also found that severe motor disability and high rate of disease progression were slightly related to CV reduction of the sensory-motor cerebellar lobules. This is consistent with the evidence of gradually progressive cerebellar gray matter degeneration throughout disease progression in ALS (Bede and Hardiman, 2017). A recent study showed reduced gray matter volume (lobule IV and V) in ALS patients with no cognitive defect and posterior (VIII) cerebellar involvement in ALS patients with cognitive impairment (Christidi et al., 2018a). This is in line with our correlation analyses, confirming that specific patterns of subregional involvement are differently associated with cognitive and motor impairment in ALS (Tan et al., 2014).

Unexpectedly, our investigation of the cerebellar subregions did not revealed atrophy clusters specific for C9+ALS patients, which is consistent with recent findings (Floeter et al., 2016; Schönecker et al., 2018) but not with others (Westeneng et al., 2016; Agosta et al., 2017). Since cerebellar involvement in C9orf72 carriers have been found among FTD-ALS spectrum (Bede et al., 2013a; Irwin et al., 2013; Floeter et al., 2016), it is conceivable that it may reflect a signature of ALS dementia other than a signature of the C9orf72 hexanucleotide repeat (Bede et al., 2013a). Further studies deserve larger C9+ALS samples stratified on the basis of the degree of cognitive impairment to address this issue.

The strength of the presented study was the tentative to systematically explore the cognitive contributions of cerebellar atrophy in selected cohorts of patients with ALS. Our data show that cerebellar degeneration is not specific for nondemented patients with ALS and cognitive syndromes or C9orf72 mutations. However, the relatively small number of mutation carriers and the lack of ALS-FTD patients included limit the strength for firm conclusion. Indeed, we could not exclude that overall disease progression both in terms of progressive physical disability and progressive cognitive impairment might be related to cerebellar involvement, in line with its topographic organization.

# DATA AVAILABILITY

The datasets generated for this study are available on request to the corresponding author.

# ETHICS STATEMENT

The review board of the Fondazione IRCCS Istituto Neurologico Carlo Besta of Milan approved the study. Each subject was enrolled after giving written informed consent.

# AUTHOR CONTRIBUTIONS

MC, EDB, AN, CP, CG, and SF collected the data. AN, CP, GD, MB, and SF acquired and the analyzed neuroimaging data. MC, SF, and SC designed the study. MC and LP performed the statistical analyses. MC, EDB, SC, MB, GL, and SF gave important intellectual content in data interpretation. MC, EDB, LP, and AN wrote a draft of the manuscript. MC, SC, GL, and SF revised the final version of the manuscript for intellectual content. MC, EDB, AN, CP, GD, LP, CG, VP, SC, MB, GL, and SF read and approved the final version.

# FUNDING

This work has been supported by Fondazione Regionale per la Ricerca Biomedica (Regione Lombardia), project nr. 2015-0023.

# ACKNOWLEDGMENTS

We thank neurologists of the Fondazione IRCCS Istituto Neurologico Carlo Besta of Milan, Patrizia Dacci, Daniele Cazzato, Susanna Usai, Ettore Salsano, and Davide Pareyson for referring some patients; Valentina E.C. Gessa for the neuropsychological assessment of some patients.

# REFERENCES

fnins-13-00440 May 10, 2019 Time: 16:37 # 6


study in amyotrophic lateral sclerosis: relationships with motor disability. J. Neurol. Neurosurg. Psychiatry 78, 889–892. doi: 10.1136/jnnp.2006. 101758


**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 © 2019 Consonni, Dalla Bella, Nigri, Pinardi, Demichelis, Porcu, Gellera, Pensato, Cappa, Bruzzone, Lauria and Ferraro. 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.

# Metabolic and Stress Response Changes Precede Disease Onset in the Spinal Cord of Mutant SOD1 ALS Mice

Gavin Pharaoh1,2, Kavithalakshmi Sataranatarajan<sup>1</sup> , Kaitlyn Street<sup>1</sup> , Shauna Hill1,3 , Jake Gregston<sup>1</sup> , Bumsoo Ahn<sup>1</sup> , Caroline Kinter<sup>1</sup> , Michael Kinter<sup>1</sup> and Holly Van Remmen1,2,4 \*

 Aging and Metabolism Research Program, Oklahoma Medical Research Foundation, Oklahoma City, OK, United States, Department of Physiology, University of Oklahoma Health Science Center, Oklahoma City, OK, United States, <sup>3</sup> Department of Cellular and Structural Biology, University of Texas Health Science Center at San Antonio, San Antonio, TX, United States, Oklahoma City VA Medical Center, Oklahoma City, OK, United States

#### Edited by:

Foteini Christidi, National and Kapodistrian University of Athens Medical School, Greece

#### Reviewed by:

Chiara F. Valori, German Center for Neurodegenerative Diseases (DZNE), Germany Susanne Petri, Hannover Medical School, Germany

> \*Correspondence: Holly Van Remmen Holly-VanRemmen@omrf.org

#### Specialty section:

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

Received: 10 December 2018 Accepted: 29 April 2019 Published: 31 May 2019

#### Citation:

Pharaoh G, Sataranatarajan K, Street K, Hill S, Gregston J, Ahn B, Kinter C, Kinter M and Van Remmen H (2019) Metabolic and Stress Response Changes Precede Disease Onset in the Spinal Cord of Mutant SOD1 ALS Mice. Front. Neurosci. 13:487. doi: 10.3389/fnins.2019.00487 Many Amyotrophic Lateral Sclerosis (ALS) patients experience hypermetabolism, or an increase in measured vs. calculated metabolic rate. The cause of hypermetabolism and the effects on neuronal metabolism in ALS are currently unknown, but the efficacy of dietary interventions shows promise for metabolism as an ALS therapeutic target. The goal of this study is to measure changes in metabolic pathways as a function of disease progression in spinal cords of the SOD1G93A mouse model of ALS. We conducted a comprehensive assessment of protein expression for metabolic pathways, antioxidants, chaperones, and proteases in lumbar spinal cord from male SOD1G93A mice at preonset, onset, and end-stages of the disease using targeted proteomic analysis. These results reveal that protein content of metabolic proteins including proteins involved in glycolysis, β-oxidation, and mitochondrial metabolism is altered in SOD1G93A mouse spinal cord well before disease onset. The changes in mitochondrial metabolism proteins are associated with decreased maximal respiration and glycolytic flux in SOD1G93A dermal fibroblasts and increased hydrogen peroxide and lipid hydroperoxide production in mitochondria from sciatic nerve and gastrocnemius muscle fibers at end stage of disease. Consistent with redox dysregulation, expression of the glutathione antioxidant system is decreased, and peroxiredoxins and catalase expression are increased. In addition, stress response proteases and chaperones, including those involved in the mitochondrial unfolded protein response (UPRmt), are induced before disease onset. In summary, we report that metabolic and stress response changes occur in SOD1G93A lumbar spinal cord before motor symptom onset, and are primarily caused by SOD1G93A expression and do not vary greatly as a function of disease course.

Keywords: amyotrophic lateral sclerosis (ALS), SOD1G93A ALS mouse model, metabolism, hypermetabolism, mitochondria, antioxidants, stress response, mitochondrial unfolded protein response

# INTRODUCTION

fnins-13-00487 May 29, 2019 Time: 18:23 # 2

Amyotrophic Lateral Sclerosis (ALS) is a cell non-autonomous disease affecting motor neurons as well as surrounding support cells including astrocytes, oligodendrocytes, and microglia (Ilieva et al., 2009; Rizzo et al., 2014) and skeletal muscle (Dobrowolny et al., 2008). Motor neuron cell death causes loss of neuromuscular innervation resulting in severe muscle atrophy and paralysis culminating in death by respiratory failure (Rowland and Shneider, 2001). While recent advances in sequencing technologies have led to the identification of genes underlying over 60% of familial cases (fALS) and over 10% of sporadic cases (sALS), the pathogenesis of ALS is still poorly understood delaying the development of effective treatments (Renton et al., 2014). More importantly, the only FDA approved drugs for the treatment of ALS, riluzole and edaravone, provide only minor improvements for ALS patients (Zoccolella et al., 2007; Abe et al., 2014).

The first mutated gene identified in fALS was superoxide dismutase 1 (Sod1), an antioxidant enzyme that detoxifies superoxide. Several transgenic mouse lines expressing mutant Sod1, including SOD1G93A, have been generated that recapitulate fALS symptoms and have allowed exploration of biochemical mechanisms underlying fALS pathogenesis (Ilieva et al., 2009). One phenotype of SOD1 mutant mice that also recapitulates the human disease is a hypermetabolic phenotype that occurs before the onset of motor symptoms (Dupuis et al., 2004). Approximately 50% of ALS patients experience the phenomenon of hypermetabolism, wherein their measured resting energy expenditure is increased compared to their calculated resting energy expenditure (Bouteloup et al., 2009). Patients rarely switch metabolic groups during disease progression, and the hypermetabolism phenotype does not correlate with fat free mass, age, sex, BMI, or an ALS functional test (Bouteloup et al., 2009). However, hypermetabolic patients display a trend toward decreased survival length from symptom onset (Bouteloup et al., 2009), and a phase II clinical trial suggested that high calorie diets had beneficial survival effects in ALS patients (Wills et al., 2014). In support of this, a high calorie diet extends SOD1 mutant mouse lifespan (Dupuis et al., 2004), while caloric restriction reduced lifespan (Hamadeh et al., 2005). The hypermetabolism phenotype is a potential disease modifier that can be targeted to affect patient disease progression.

We hypothesized that increased energy demands (hypermetabolism) cause an energetic stress affecting metabolism in motor neurons that contributes to disease progression. To test our hypothesis, we collected lumbar spinal cord from SOD1G93A mice and age-matched wildtype controls before and at onset of motor symptoms and at disease end-stage. We measured changes in protein abundance of major metabolic pathway enzymes at all three disease stages via a targeted proteomics approach. In addition, because mitochondrial damage occurs early in ALS pathogenesis, we measured sciatic nerve and skeletal muscle mitochondrial function and lipid hydroperoxide and hydrogen peroxide production. We also measured expression of spinal cord and skeletal muscle mitochondrial proteases and chaperones, including those involved in the mitochondrial unfolded protein response (UPRmt), a stress response pathway responsible for maintaining mitochondrial protein quality control in response to accumulation of misfolded or unfolded proteins in the mitochondria (Hill and Van Remmen, 2014). Our results reported here represent the first comprehensive measurement of major metabolic and stress response pathways in spinal cord of ALS mutant mice as a function of disease stage.

# RESULTS

# Metabolic Pathways Are Differentially Regulated in the Spinal Cord of SOD1G93A Mice, and Stress Response Proteins Are Induced

The SOD1G93A ALS mouse model recapitulates the hypermetabolism phenotype (Dupuis et al., 2004). In addition to the well characterized decline in body and muscle mass, we also observed reduced fat mass in end-stage SOD1G93A mice (**Supplemental Figure 1**). Because ALS is a neurodegenerative disease characterized by pathological changes in the spinal cord, we used targeted proteomic analysis to measure differential protein content of metabolic and stress response pathways in spinal cord homogenates from wild-type and SOD1G93A mice over the time course of the disease. We interrogated changes in 113 proteins detected in at least one genotype (**Figure 1** and **Supplemental Table 1**). We categorized proteins into panels for carbohydrate metabolism, fatty acid metabolism, oxidative mitochondrial metabolism [the tricarboxylic acid cycle and electron transport chain (ETC)], and stress response (antioxidants, chaperones, heat shock proteins, and proteases). Metabolic and stress response protein expression differences in SOD1G93A lumbar spinal cord are evident throughout disease course (**Figure 1A**). Principal component analysis (PCA) demonstrates that protein expression in lumbar spinal cord from SOD1G93A mice clusters distinctly from spinal cord from control mice beginning before onset of symptoms, but no separation occurs in clustering between disease stages (**Figure 1B**). 49% of assayed proteins exhibit a significant difference due to genotype effect, while only 10% exhibited a significant difference based on disease stage using two-way ANOVA corrected for false discovery rate with the Benjamini-Hochberg procedure. These results show that metabolism and stress response protein content is already dramatically altered in spinal cords from SOD1G93A mice before onset of motor symptoms.

#### Glycolysis and the Malate-Aspartate Shuttle

Two key proteins involved in glycolysis are elevated prior to disease onset, but we also observe a decline at end-stage in several glycolytic proteins along with a decline in some key proteins involved in the malate aspartate shuttle in spinal cord from SOD1G93A mice Specifically, levels for the first two enzymes in glycolysis (HK1 and GPI1) are increased beginning at pre-onset or disease onset. However, ALDOA and GAPDH are reduced at end-stage and PGK1 exhibits a genotype specific reduction in spinal cords from SOD1G93A mice compared to wild-type

controls (**Figure 2A**). PGAM2 only reached the threshold of detection in SOD1G93A samples (**Figure 2A**). At end-stage, protein content of key components of the malate-aspartate shuttle declines (**Figure 2B**). While SLC25A11, the mitochondrial inner-membrane α-ketoglutarate/malate carrier, is increased at onset, at end-stage the cytoplasmic and mitochondrial aspartate aminotransferases (GOT1 and GOT2) and the mitochondrial malate dehydrogenase (MDH2) are decreased compared to wildtype controls (Maglott et al., 2011; **Figure 2B**). TKT, a key enzyme in the pentose-phosphate pathway is increased in spinal cords from SOD1G93A mice (**Supplemental Table 1**). Combined these results suggest altered carbohydrate utilization in endstage spinal cords from SOD1G93A mice compared to wildtype control mice.

#### β-Oxidation

Several proteins involved in mitochondrial β-oxidation are increased in spinal cords from SOD1G93A mice. CPT2, which is involved in import of fatty acids to mitochondria, and CROT, which is involved in export of β-oxidation products from peroxisomes, are both increased in SOD1G93A samples (**Figure 3A**). The mitochondrial medium chain acyl-CoA dehydrogenase (ACADM) is increased at onset and end-stage in spinal cord from SOD1G93A mice compared to wild-type mice, and the mitochondrial β-hydroxyl acyl CoA dehydrogenase (HADH) is increased throughout the disease course in SOD1G93A mice (**Figure 3A**). Enoyl-CoA Hydratase 1 (ECH1), which is targeted to both mitochondria and peroxisomes, and DECR1, an auxiliary enzyme of mitochondrial β-oxidation are increased at disease onset in SOD1G93A mice (**Figures 3A,B**). The fatty acid transporters ABCD3 and FABP3 are upregulated beginning at disease onset (**Figure 3B**). In the peroxisome, the peroxisomal bifunctional enzyme HSD17B4 is increased in spinal cord from SOD1G93A while EHHADH is decreased throughout disease course (**Figure 3A**).

Several proteins involved in other aspects of fatty acid metabolism are differentially regulated. BDH1 is a mitochondrial dehydrogenase that interconverts two ketone bodies produced during fatty acid catabolism (Maglott et al., 2011) and is increased in onset and end-stage compared to pre-onset in spinal cords from SOD1G93A mice (**Figure 3B**). ACOT13 is a mitochondrial medium- and long-chain acyl-CoA thioesterase and is decreased at end-stage compared to wild-type (**Figure 3B**). HMGCS1 catalyzes production of HMG-CoA, an intermediate in ketogenesis and isoprenoid biosynthesis, including synthesis of a wide variety of molecules such as cholesterol, ubiquinone, and steroid hormones (Holstein and Hohl, 2004; Maglott et al., 2011). HMGCS1 expression is decreased beginning at disease onset in SOD1G93A mice (**Figure 3B**). HMGCL, a key enzyme in ketogenesis, is increased in SOD1G93A spinal cords. Taken together, these data suggest an increase in mitochondrial fatty acid metabolism in ALS spinal cords.

#### Tricarboxylic Acid (TCA) Cycle

TCA proteins are generally decreased at end-stage in spinal cords from SOD1G93A mice. Three subunits of pyruvate dehydrogenase (DLAT, PDHA1, PDHB), IDH3A, and MDH2 are decreased in end-stage in spinal cords from SOD1G93A mice compared to wild-type control mice (**Figure 4A**). However DLD, the E3 component of α-ketoglutarate dehydrogenase is increased throughout disease course. Another subunit of α-ketoglutarate dehydrogenase, DLST, and glutamate dehydrogenase (GLUD1) are increased at onset suggesting increased entry of substrates into the TCA cycle from the glutamate pathway and decreased entry from carbohydrate metabolism (**Figure 4A**).

#### Mitochondrial Electron Transport Chain (ETC)

The content of several assayed proteins in the mitochondrial ETC is increased in spinal cord from the SOD1G93A mice. Subunits of ETC Complex I (NDUFS1, NDUFV1), Complex III (UQCRC1), and Complex V (ATP5A1, ATP5B), as well as ETF (ETFB) are increased in SOD1G93A spinal cord beginning before disease onset (**Figure 4B**). COQ6, which performs a biosynthetic step of ubiquinone synthesis, is also increased in SOD1G93A spinal cord. In contrast, one subunit of Complex II (SDHB) is decreased (**Figure 4B**). Because we saw increased expression of ETC proteins in spinal cords from SOD1G93A mice, we assessed gene expression for nuclear- and mitochondrial-encoded ETC subunits and PGC1α, a transcription factor involved in mitochondrial biogenesis (Scarpulla, 2011), in spinal cords and gastrocnemius muscles from SOD1G93A mice. Expression of all assayed ETC complex subunits is increased and all exhibit a significant difference by genotype effect in spinal cords from SOD1G93A mice compared to controls (**Supplemental Table 3** and **Supplemental Figure 2A**). However, no difference due to genotype is observed in PGC1α expression. In gastrocnemius muscle, no difference is identified between genotypes for ETC subunit expression, but Ppargc1a (PGC1α) expression is reduced at end-stage compared to wild-type controls (**Supplemental Table 4** and **Supplemental Figure 2B**).

#### Antioxidant Enzymes

Oxidative stress has been implicated as an important underlying factor in ALS pathogenesis (Barber and Shaw, 2010). We found many changes in the expression of hydrogen peroxide detoxification enzymes in spinal cord from SOD1G93A mice compared to wild-type mice. The glutathione system, peroxiredoxins, and catalase are altered in spinal cords from SOD1G93A mice (**Figure 5**). PRDX2, PRDX3, CAT, and PRDX6 are increased in SOD1G93A compared to wild-type control (**Figure 5**). Simultaneously, GPX1 (glutathione peroxidase 1), GSTM1, and GSTA3 are decreased in spinal cords from SOD1G93A mice (**Figure 5**). SOD1 is upregulated in SOD1G93A compared to wild-type control, validating SOD1 overexpression in the spinal cord samples, while no change is observed in SOD2, the mitochondrial superoxide dismutase (**Figure 5**).

#### Stress Response Proteases and Chaperones

Misfolding and accumulation of ubiquitinated cytoplasmic mutant SOD1 inclusions has been observed early in disease progression and likely plays a role in pathogenesis (Ilieva et al., 2009). To understand the cellular response to accumulation of these aggregates, we measured the mRNA expression and protein content of a panel of proteases and chaperones. In spinal cord,

measured by mass spectrometry (n = 5–6). (A) Glycolysis. (B) Malate-aspartate shuttle. Data are represented as percent change relative to the same time point wild-type control. Bars represent means ± standard deviation. Phosphoglycerate mutase 2 (PGAM2) did not reach the level of detection in wild-type samples but did in SOD1G93A. "Genotype" denotes a significant genotype effect in Benjamini-Hochberg corrected two-way ANOVA with p < 0.05 but no individual Multiple Comparison test was significant. Significance symbols demonstrate p < 0.05 for a genotype effect and additional significant results for the Tukey Multiple Comparison test. <sup>∗</sup>p < 0.05 wild-type vs. SOD1G93A at the designated time point. SOD1G93A pre-onset (white), SOD1G93A onset (gray), SOD1G93A end-stage (black). DH, dehydrogenase; DHAP, Dihydroxyacetone phosphate. Protein names, IDs, and absolute expression values are detailed in Supplemental Table 1.

mRNA expression of the mitochondrial proteases Clpp, Lonp1, Afg3l2, and Oma1 is increased pre-onset, Yme1l1 is decreased at end-stage, and the chaperone Hspd1 is increased (**Figure 6B** and **Supplemental Table 3**). While CLPP is not increased at the protein level, CLPX, the chaperone subunit of the CLPX/CLPP oligomer, only reached the limit of detection in spinal cords from SOD1G93A mice (**Supplemental Table 1**). In addition, protein expression of LONP1 and LONP2 as well as the heat shock proteins HSPD1 (Hsp60), CRYAB, and HSP90B1 is increased in spinal cords from SOD1G93A mice compared to controls

beginning before disease onset (**Figure 6C**). In contrast, mRNA expression of Hspd1, Spg7, Yme1l1, and Oma1 is decreased in gastrocnemius at end-stage (**Figure 6A** and **Supplemental Table 4**). ALDH2, the mitochondrial aldehyde dehydrogenase, protects against oxidative stress (Ohta et al., 2004) and is upregulated in onset SOD1G93A compared to wild-type control (**Supplemental Table 1**). ATP2A2, an endoplasmic reticulum calcium pump, is decreased at end-stage (**Supplemental Table 1**). Together, these data suggest that regulation of stress response to SOD1G93A expression is tissue specific, and spinal cords from SOD1G93A mice experience an increase beginning before onset of motor symptoms. mRNA expression or protein content of 4 proteins (HSPD1, CLPP, LONP1, and CLPX) involved in the UPRmt was increased in spinal cords but not muscle from SOD1G93A mice.

# Mitochondrial Function in Cultured Fibroblasts, Sciatic Nerve Tissue, and Permeabilized Muscle Fibers of SOD1G93A Mice

Mitochondrial dysfunction has been broadly implicated in the pathogenesis of neurodegenerative diseases including ALS (Schon and Manfredi, 2003), and impaired mitochondrial function is observed in human fALS SOD1 mutant cultured fibroblasts (Allen et al., 2014). We analyzed oxygen consumption rate (OCR) in pre-onset SOD1G93A mouse dermal fibroblasts with the Seahorse XF24 using sequential additions of the Complex V/ATP synthase inhibitor oligomycin, protonophore FCCP, and Complex III inhibitor antimycin A to measure several parameters of mitochondrial function (**Figure 7A**). SOD1G93A mouse fibroblasts have reduced maximal respiration and reserve capacity compared to wild-type controls, but no change in basal or ATP-linked respiration and no change in proton leak (**Figure 7B**). Extracellular acidification rate (ECAR) is a measurement of glycolysis using change in pH of the media (Wu et al., 2007). ECAR is non-significantly decreased at baseline but is significantly reduced when stressed with oligomycin and FCCP in SOD1G93A mouse fibroblasts (**Figure 7D**). SOD1G93A mouse fibroblasts have a decreased ability to respond to energetic demands in response to stress using glycolytic or oxidative mitochondrial metabolism (**Figure 7C**).

Because we found changes in the expression of ETC complex proteins in spinal cord, and previous work showed altered mitochondrial function in permeabilized SOD1G93A mouse spinal cord (Cacabelos et al., 2016), we assessed mitochondrial function in end-stage nerve tissue and muscle. Due to technical limitations discussed later, we used the sciatic nerve to analyze

Tables 3 (spinal cord) and 4 (gastrocnemius). Protein names, IDs, and absolute expression values are detailed in Supplemental Table 1.

mitochondrial function in an affected nerve tissue. We measured simultaneous mitochondrial OCR and hydrogen peroxide and lipid hydroperoxide production rates in permeabilized sciatic nerves and muscle fibers using the OROBOROS Oxygraph-2k with Amplex UltraRed (AUR). Endogenous AUR reaction rate was measured directly after addition of the nerve tissue and muscle fiber bundle to the chamber. OCR and AUR reaction rate were analyzed for both tissues using a sequential addition of glutamate and malate in the absence of ADP to induce proton leak-linked respiration, ADP to stimulate Complex I-linked respiration, pyruvate as a further Complex I substrate, succinate to stimulate Complex I&II-linked respiration (OXPHOS capacity), rotenone to inhibit Complex I and measure Complex II-linked respiration, antimycin A to inhibit Complex III and measure non-mitochondrial OCR, and ascorbate and TMPD to activate CIV-linked respiration. We observe no significant difference in mitochondrial oxidative phosphorylation (OXPHOS) in either sciatic nerve or gastrocnemius muscle (**Figures 7E–H**). However, there is a significant increase in endogenous AUR reaction rate in gastrocnemius fibers from SOD1G93A mice as well as a trend for an increase in rate in sciatic nerve (**Figures 7G,H**).

output from Seahorse XF24 Flux Analyzer and calculation of measurements. (B) Oxygen consumption rate (OCR), (C) Extracellular acidification rate (ECAR), and (D) Comparison of OCR and ECAR of dermal fibroblasts isolated from wild-type (WT) and SOD1G93A mice (n = 5) measured by Seahorse XF24 Flux Analyzer. Stressed ECAR was measured in the presence of oligomycin and FCCP. Simultaneous OCR and hydrogen peroxide and lipid hydroperoxide production rates (Amplex UltraRed) were assessed in permeabilized sciatic nerves (n = 8–11) and red gastrocnemius fiber biopsies (n = 7–9) from end-stage female mice using the OROBOROS Oxygraph-2k with fluorometer. (E) Gastrocnemius fiber OCR. (F) Gastrocnemius fiber hydrogen peroxide and lipid hydroperoxide production rates. (G) Sciatic nerve OCR. (H) Sciatic nerve hydrogen peroxide and lipid hydroperoxide production rates. <sup>∗</sup>p < 0.05 wild-type vs. SOD1G93A determined by unpaired student's t-test. (I) Basal hydrogen peroxide and lipid hydroperoxide production rates for SOD1G93A gastrocnemius fibers in the OROBOROS O2K fluorometer vehicle treated, or pre-treated with 2000 U/ml Catalase or 20 µM AACOCF<sup>3</sup> (n = 7). <sup>∗</sup>p < 0.05 treated (either catalase or AACOCF3) vs. untreated control, &p < 0.05 catalase vs. AACOCF<sup>3</sup> treated. Substrate-inhibitor protocol and definitions of states are described in Materials and Methods. Bars represent means ± standard deviation. Wild-type (white) and SOD1G93A (black). CI, mitochondrial Complex I (with glutamate, malate, and ADP); CI(P), mitochondrial Complex I (with glutamate, malate, pyruvate, and ADP); CII, mitochondrial Complex II; CIV, mitochondrial Complex IV; FCCP, p-trifluoromethoxy carbonyl cyanide phenyl hydrazone; AACOCF3, Arachidonyl trifluoromethyl ketone.

While the probe AUR is generally used as a measurement for production rate of the reactive oxygen species (ROS) hydrogen peroxide, we have previously demonstrated that it also reacts with lipid hydroperoxides produced in muscle in response to loss of innervation (Bhattacharya et al., 2009). Muscles of SOD1G93A mice experience denervation and atrophy and increase expression of genes associated with loss of innervation including acetylcholine receptor subunits (Chrna1, Chrnd, Chrne) and Gadd45a (**Supplemental Figure 2C**). Sln, which we previously demonstrated is dramatically increased in sarcopenia, is also increased in gastrocnemius muscle from SOD1G93A mice along

with the chemokines Cxcl2 and Cxcl5 (Qaisar et al., 2018; **Supplemental Figure 2C**). To distinguish the source of the AUR reaction signal, we treated SOD1G93A gastrocnemius samples with either catalase or arachidonyl trifluoromethyl ketone (AACOCF3). AACOCF<sup>3</sup> is an inhibitor of calciumdependent phospholipase A2 (cPLA2) that we have previously shown inhibits lipid hydroperoxide production in denervated muscle (Bhattacharya et al., 2009). Treatment with AACOCF<sup>3</sup> more effectively inhibits the endogenous AUR reaction rate than catalase, suggesting the signal is produced primarily by reaction with lipid hydroperoxides and not hydrogen peroxide or superoxide produced by the mitochondrial ETC (**Figure 7I**). Addition of substrates and inhibitors that increase mitochondrial ETC-derived superoxide and hydrogen peroxide production demonstrates that supraphysiological concentrations of catalase entirely inhibits the increased ETCderived hydroperoxide production, while AACOCF<sup>3</sup> only significantly decreases endogenous hydroperoxide production (**Supplemental Figures 3A,B**).

# DISCUSSION

The goal of this study is to determine the changes in metabolism and stress response in lumbar spinal cord of an ALS mouse model as a function of disease course. Understanding metabolic changes in ALS may help us develop specific dietary interventions that can delay disease progression. We find significant differences in protein content of metabolic pathway enzymes, antioxidant systems, heat shock proteins, and proteases that occur early in the lifespan of the mice and before onset of motor symptoms. Expression of the SOD1G93A mutant protein also leads to tissue specific regulation of chaperones and proteases, specifically, the expression of these proteins is induced in spinal cords but not skeletal muscle from SOD1G93A mice. The differences we observed in metabolism and stress response protein content are explained primarily by expression of the SOD1G93A transgene and not greatly modulated by disease stage.

Several previous studies have utilized proteomics approaches to analyze differences in protein content in the spinal cord of pre-onset ALS rodent models (Lukas et al., 2006; Massignan et al., 2007; Li et al., 2010). Massignan et al. (2007) used the two-dimensional gel electrophoresis proteomic method to analyze protein content changes in whole spinal cords from pre-onset SOD1G93A and control mice and also observed significant changes in content of glycolytic, mitochondrial, and stress response proteins. Li et al. (2010) used both twodimensional gel electrophoresis as well as liquid chromatography followed by tandem mass spectrometry (MudPIT) to analyze changes in protein content of mitochondria isolated from spinal cords of pre-onset SOD1G93A rats and rats overexpressing wildtype SOD1 (SOD1WT). Many proteins identified in their MudPIT analysis were not assessed in our targeted approach. However, Li et al. (2010) also observed many changes pre-onset in metabolic and stress response protein content and some were also observed in our analysis including a decrease in MDH2 and an increase or trend for increase in DECR1, IDH2, and PYGB. Lukas et al. (2006) analyzed protein content of spinal cords from end-stage wildtype, SOD1WT, and SOD1G93A fractionated into mitochondrial, soluble, and particulate fractions with two-dimensional liquid chromatography coupled to tandem mass spectrometry. The authors found significant changes in metabolic and antioxidant proteins, among others (Lukas et al., 2006). Previous studies have not analyzed metabolic and stress response protein content as a function of disease course. We report here that modification of metabolic and stress response protein content is primarily due to SOD1G93A expression and not disease course.

# Metabolism Is Altered by SOD1G93A Expression Before Disease Onset

In the early stages of disease development in this ALS mouse model, there is a significant change in the content of proteins related to glycolysis. Transfection of a neuronal cell line to express SOD1G93A protein has been found to increase glycolytic flux (Valbuena et al., 2016). However, at the end stages of the disease, we find that the content of several proteins involved in the malate-aspartate shuttle, pyruvate dehydrogenase, and TCA cycle is decreased, but the content of some proteins related to mitochondrial fatty acid oxidation and the mitochondrial ETC are increased in spinal cords from SOD1G93A mice. Upon activation, astrocytes undergo metabolic changes that optimize glucose utilization to produce lactate including the upregulation of proteins involved in glycolysis (Iglesias et al., 2017). Gliosis in spinal cords from SOD1G93A mice may partially explain the observed increase in some glycolytic proteins. Activated astrocytes also increase beta-oxidation to produce ketone bodies, which is consistent with our finding that HMGCL, a key enzyme in ketogenesis, is upregulated throughout disease course glycolysis (Iglesias et al., 2017). Compared to glial cells, neurons have a higher rate of oxidative metabolism and a slower glycolytic rate (Bélanger et al., 2011). Glucose is shunted through the pentose-phosphate pathway to regenerate NADPH, which is used to reduce glutathione among other reactions in the neuronal antioxidant system (Wamelink et al., 2008). Increased glycolytic rate comes at the expense of the pentose phosphate pathway and cellular antioxidant capacity (Bélanger et al., 2011). The decrease in ECAR also suggests decreased glycolytic potential in SOD1G93A fibroblasts. Altered fatty acid metabolism has been observed in ALS animal models and patients, where hyperlipidemia is associated with increased survival (Schmitt et al., 2014). Interestingly, ACADM catalyzes the first steps of β-oxidation for medium chain fatty acids, and feeding SOD1G93A mice with a medium chain triglyceridesupplemented diet delayed paralysis and motor neuron cell death (Zhao et al., 2012). In addition to fatty acid metabolism, changes in fat mass have also previously been linked to ALS. Decreased body mass index (BMI) is associated with increased risk of ALS (Gallo et al., 2013). Furthermore, high calorie diet appeared to be beneficial in a small phase 2 clinical trial (Wills et al., 2014). Metabolism plays a key role in

ALS pathogenesis before onset of symptoms, and therapies targeting metabolism and nutrition have potential to delay disease progression.

Metabolomics experiments have been conducted in ALS patients and mouse models. Metabolomic analysis of skeletal muscle, cerebral cortex, and plasma from wildtype and SOD1G93A mice reveals a distinct metabolite profile in SOD1G93A samples in each tissue (Patin et al., 2016). ALS patients also have significantly different plasma and cerebrospinal fluid (CSF) metabolites compared to controls. Metabolites associated with glucose metabolism are significantly different in the CSF of ALS patients compared to controls (Blasco et al., 2010, 2016; Gray et al., 2015). Significantly different plasma metabolites include amino acids, carbohydrates, lipids, peptides, nucleotides, and xenobiotics (Rozen et al., 2005; Lawton et al., 2012, 2014). The difference in plasma or CSF metabolite levels has been used to predict ALS in patients in preliminary studies, although no definitive metabolite panel has been developed for diagnosis of ALS (Blasco et al., 2014; Lawton et al., 2014). Treatment with Riluzole also significantly alters concentration of several metabolites (Rozen et al., 2005). Metabolite level has also been shown to correlate with disease progression, which suggests that metabolites could be used as a biomarker to diagnose ALS (Kumar et al., 2010; Blasco et al., 2016, 2018). However, there is a significant difference in CSF metabolite profile between sporadic, familial, and familial SOD1 mutant patents (Wuolikainen et al., 2011, 2012). Though metabolite profiles are heterogenous between patient subtypes, metabolite concentrations are clearly altered in ALS patients and future studies may provide evidence of common underlying metabolic pathophysiology.

# Mitochondrial Dysfunction Is Tissue Specific

Mitochondrial dysfunction has been found to occur in numerous studies of ALS tissues, although observed differences are often not consistent between studies. A potential reason for this discrepancy in study outcomes is the diverse genetic causes and disease progressions in patients (Renton et al., 2014). Several studies have shown changes in neuronal ATP levels, ETC complex activities, mitochondrial calcium buffering capacity, and spinal cord astrocyte mitochondrial damage in ALS mouse models (Jung et al., 2002; Ilieva et al., 2009; Cacabelos et al., 2016). Fibroblasts from human patients are commonly studied as a model for mitochondrial function due to their ease of biopsy compared to the primary affected tissues in humans (spinal cord and skeletal muscle). Fibroblast studies in familial and sporadic ALS patient report differences compared to control samples as well as to each other. Familial TARDBP(p.A328T) mutant fibroblasts have decreased mitochondrial membrane potential but no change in OCR (Onesto et al., 2016). Familial C9ORF72 mutant fibroblasts conversely exhibit mitochondrial hyperpolarization, increased OCR and ATP content, increased ROS and increased mitochondrial DNA content (Onesto et al., 2016). Familial SOD1I113T mutants exhibit decreased basal and coupled OCR and decreased reserved capacity (Allen et al., 2014). Sporadic ALS patient fibroblasts presented with decreased basal and coupled OCR (Raman et al., 2015). We observed reduced maximal respiration and reserve capacity in SOD1G93A mutant mouse fibroblasts compared to controls. Together, these findings suggest that ALS pathology affects mitochondrial function in fibroblasts regardless of underlying genetic cause, but mitochondrial phenotypes may be specific to the gene mutation.

ALS is primarily considered a motor neuron disease, and a previous study reported that basal and Complex I-linked respiration were reduced in onset SOD1G93A mouse spinal cords compared to pre-onset controls (Miquel et al., 2014; Cacabelos et al., 2016). However, a major issue with measuring OCR in tissues is that time of oxygen diffusion increases exponentially with distance and the lumbar spinal cord has much larger circumference than the sciatic nerve. Therefore, we chose permeabilized sciatic nerves for nerve tissue measurements, because they have a much smaller radius than spinal cords. Here we report no difference in sciatic nerve tissue OCR but a trend (p = 0.059) for an increase in AUR reaction rate. However, protein content data measured in spinal cord demonstrated an increased ETC subunit content from ALS compared to control mice. Sciatic nerve tissue may show similar compensatory changes.

Oxidative phosphorylation has been assayed in permeabilized fibers from ALS patient M. vastus lateralis (MVL) muscle biopsy permeabilized fibers finding a reduction in CI-linked OXPHOS (Wiedemann et al., 1998). However, another study of sALS patient MVL permeabilized fibers found no difference between patient and control muscles for CI-, CII-, or CIV-linked respiration (Echaniz-Laguna et al., 2002). The mean duration of disease in this study was low compared to others, and biopsies were taken before most of the patients exhibited MVL muscle atrophy (Echaniz-Laguna et al., 2002). Here we report no significant difference in OCR of permeabilized gastrocnemius fibers from end-stage SOD1G9EA mice compared to control. However, musclespecific expression of SOD1G93A is sufficient to induce muscle dysmetabolism by inhibiting glucose uptake and shifting muscle toward using lipid metabolism, which suggests muscle metabolism also plays a significant role in ALS hypermetabolism (Dobrowolny et al., 2018).

#### Antioxidant and Stress Response Proteins Are Induced

We find that the levels of catalase and peroxiredoxins are induced, but the glutathione system expression is decreased. End-stage sciatic nerve tissue and gastrocnemius muscle fibers from SOD1G93A mice both exhibit increased reactivity ex vivo with AUR, which is primarily caused by lipid hydroperoxides in muscle. The role of lipid hydroperoxides in muscle atrophy and contractile dysfunction are still unclear, but these results are consistent with previous findings we reported in denervated muscle fibers (Bhattacharya et al., 2009). GPX4, which reduces lipid hydroperoxides, is not significantly different in spinal cords from SOD1G93A mice. In denervated muscle, we identified

cytosolic phospholipase A2 (cPLA2) as the primary producer of lipid hydroperoxides (Bhattacharya et al., 2009). Previous studies have shown an upregulation of cPLA<sup>2</sup> in sporadic ALS patients and pre-onset SOD1G93A mouse spinal cords, and inhibition of cPLA<sup>2</sup> exhibits protective effects (Solomonov et al., 2016).

Changes in mitochondrial function may play a causal role in ALS pathogenesis for both motor neuron degeneration and skeletal muscle atrophy. Mutant SOD1 localizes to the mitochondrial outer membrane, intermembrane space, and matrix, where it forms protein aggregates possibly including other mitochondrial proteins such as SOD2 (Vijayvergiya et al., 2005). SOD1 mutant protein aggregates form in spinal cord mitochondria before the onset of motor symptoms (Liu et al., 2004), and motor neuron mitochondrial morphology is altered in SOD1 mutants prior to disease onset (Wong et al., 1995; Kong and Xu, 1998). Changes in muscle mitochondrial morphology occur pre-symptomatically, and muscle-specific SOD1G93A expression is sufficient to induce changes in mitochondrial morphology and cause mutant SOD1 aggregation inside mitochondria (Dobrowolny et al., 2008; Luo et al., 2013). We find a significant increase in heat shock proteins and chaperones in spinal cords from SOD1G93A mice throughout disease course. The mitochondrial UPRmt recognizes a decline in mitochondrial protein quality and induces expression of mitochondrial stress response proteins (Hill and Van Remmen, 2014). A previous study found an upregulation of the mitochondrial intermembrane space UPRmt in lumbar spinal cord of SOD1G93A mice (Riar et al., 2017). Here we report mRNA expression of three proteins (Hspd1, Clpp, Lonp1) and protein content of 3 proteins (HSPD1, LONP1, CLPX) involved in the mitochondrial matrix UPRmt are increased throughout disease course in spinal cords of SOD1G93A mice (Wu et al., 2014). Despite increased mRNA expression, CLPP protein content is not altered in spinal cords from SOD1G93A mice. CLPP is required in C. elegans to initiate the mitochondrial UPRmt, though recent evidence suggests it is not required in mammals (Bhaskaran et al., 2018). One hypothesis for the selective vulnerability of motor neurons in ALS is a reduced ability to upregulate stress response proteins in response to metabolic or oxidative stress compared to other neuronal populations (Duan and Mattson, 1999; Batulan et al., 2003; Sinclair, 2005; Mattson et al., 2007). The inability to increase CLPP protein content despite an increase in mRNA expression may provide evidence toward this hypothesis.

Several peroxisomal proteins are differentially regulated in ALS spinal cords. ABCD3 is a peroxisomal ATP-dependent fatty acid transporter with implicated roles in peroxisomal biogenesis (Gärtner and Valle, 1993) and uptake of fatty acid overflow from mitochondrial β-oxidation (van Roermund et al., 2014). Two other proteins involved in peroxisomal quality control are upregulated at end-stage: LONP2 and catalase (CAT). LONP2 is a peroxisomal isoform of the mitochondrial LON protease (LONP1), and both degrade proteins damaged by oxidation (Pomatto et al., 2017). Both LON proteases are increased in SOD1G93A suggesting a response to oxidative damage of proteins in both mitochondria and peroxisomes. SOD1 is expressed in peroxisomes, and peroxisomes are associated with the mitochondrial vacuoles occurring in spinal cords from SOD1G93A mice (Higgins et al., 2003). While wild-type SOD1 has been found in peroxisomes, it is unknown whether SOD1 aggregates form there (Islinger et al., 2009). These results suggest a potential role of peroxisomes in ALS that is poorly understood.

Expression of mitochondrial inner-membrane proteases is also affected (Afg3l2, Yme1l1, Oma1). Yme1l1 and Oma1 are both inner-membrane proteases involved in mitochondrial fission/fusion dynamics (Quirós et al., 2015). Afg3l2 and Spg7 are subunits of the mAAA protease, which is responsible for maintaining quality of ETC subunits (Quirós et al., 2015). Alterations in mitochondrial dynamics and morphology have been widely reported in ALS patients and animal models (Muyderman and Chen, 2014). In spinal cords from SOD1G93A mice we observe an increase in protein content of some ETC complex proteins. OPA1 processing by the inner-membrane proteases YME1L1 and OMA1 affects mitochondrial fusion and cristae remodeling (Quirós et al., 2015). YME1L1 cleavage of OPA1 promotes mitochondrial fusion, while OPA1 cleavage by OMA1 is activated by mitochondrial oxidative stress, heat stress, or depolarization and inhibits mitochondrial fusion (Quirós et al., 2015). Oma1 gene expression is induced in spinal cords from SOD1G93A mice. Increased processing of L-OPA1 by OMA1 is associated with swollen mitochondrial cristae, increased mitochondrial fragmentation, and increased sensitivity to apoptosis (Anand et al., 2014), phenotypes reported in ALS (Muyderman and Chen, 2014). OMA1 activation increases turnover of OMA1 as a feedback loop to limit stress response (Quirós et al., 2015). Oma1 missense mutations were identified in sporadic ALS patients, however, its role in pathogenesis is unclear (Daoud et al., 2011).

# SOD1G93A Expression Has Tissue Specific Effects

Metabolism and stress response proteins are differentially regulated in spinal cords from SOD1G93A mice and skeletal muscle over the disease course. While UPRmt mRNA and protein content are increased in spinal cords from SOD1G93A mice, we do not observe a similar change in gastrocnemius fibers. Furthermore, mRNA expression of inner mitochondrial membrane proteases is decreased in gastrocnemius muscle. Several previous proteomics studies have been conducted in different tissues or cell lines of ALS models. Engelen-Lee et al. (2017) performed proteomics analysis in spinal cords from post-mortem sporadic ALS patients, and identified 3 differentially expressed proteins in the posterior horn and 18 in the anterior horn. While they observed changes in metabolism and protein folding pathways, no specific proteins were consistent with those reported here in the SOD1G93A familial ALS mouse model. Szelechowski et al. (2018) performed proteomic analysis in cultured primary motor neuron from preonset SOD1G93A mice as well as cultured patient fibroblasts. They found significant alterations in metabolic and misfolded protein response pathways in both cell lines. Interestingly, the

same malate-aspartate shuttle proteins (GOT1, GOT2, MDH2) we identified as decreased in end-stage spinal cords were all decreased in patient fibroblasts. Capitanio et al. (2012) examined some of the targeted proteins we identified in spinal cords in the gastrocnemius and triceps muscles of pre-onset and onset SOD1G93A mice. Only ALDH2 and CRYAB are consistently differentially regulated between spinal cord and muscle samples for stress response. αB-crystallin (CRYAB) has been extensively studied as a potential treatment for ALS treatment, because over-expression reduced SOD1 aggregates in cell culture models and knockout in SOD1 mutant mice accelerated disease onset (Xu et al., 2015). However, transgenic overexpression of αB-crystallin in SOD1 mutant mouse lines did not slow disease progression or reduce aggregate formation, making it a poor therapeutic target for ALS (Xu et al., 2015). The only significant differences in mitochondrial metabolism enzymes conserved between muscle and spinal cord are UQCRC1 and ATP5B, which are both increased in onset triceps and spinal cord (Capitanio et al., 2012). Elf et al. (2014) performed proteomics analysis in muscles of control and ALS patients. In ALS patients, ACADM, PGAM, and PRDX6 (p < 0.06) were also found to be differentially regulated. Few changes in ALS protein content compared to control appear to be conserved between tissues.

#### Caveats and Future Directions

This study provides a view of metabolism in the spinal cord of an ALS model throughout disease course. However, it is impossible to separate metabolism of different cell types in the spinal cord using this design. A recent proteomics study on cultured primary motor neurons from SOD1G93A mice found upregulation of oxidative metabolism including fatty acid β-oxidation and TCA cycle and downregulation of glycolysis (Szelechowski et al., 2018). In addition, survival of motor neurons from an ALS mouse model depended on fatty acid β-oxidation to a much greater extent than neurons from wild-type mice (Szelechowski et al., 2018). Combined, these data suggest these metabolic changes are occurring in vivo in motor neurons before onset of motor symptoms. Thus, metabolic reprogramming may play a role in early ALS pathogenesis and provides rationale for further development of dietary interventions as well as other interventions targeting metabolism in ALS patients.

This study also provides evidence for upregulation of components of the mitochondrial UPRmt in spinal cords of SOD1G93A mice before onset of symptoms. Whether upregulation of this pathway delays onset of symptoms requires further study and could provide a new avenue for development of ALS treatments.

#### MATERIALS AND METHODS

#### Animals

All animal experiments were carried out in accordance with protocol approved by Institutional Animal Care and Use Committee at Oklahoma Medical Research Foundation. The G93ASOD1 mice [B6-Tg(SOD1-G93A)1Gur/J(G93AGur1)] were purchased from Jackson Laboratories (stock number 004435, Jackson Laboratories, Bar Harbor, Maine) and were bred on a C57BL/6J background from our lab and a colony was established for use in this study. The colony is updated with mice purchased from Jackson laboratories every 10–12 months to maintain the stability of the colony. Mice were caged in a pathogen free environment with free access to standard chow and water and maintained on a 12 h light/dark cycle. Mice were weighed weekly and disease assessment was carried out as described earlier (Evans et al., 2015). Based on the severity of the disease, male and female mice were classified as pre-onset (63–69 days), onset (102–121 days), and end stage (135–152 days) and sacrificed accordingly. Wildtype littermates of same age served as controls. Mice were monitored and euthanized if symptoms became too severe. At the time of sacrifice, brain, spinal cord, sciatic nerve, epigonadal white adipose tissue (eWAT), quadriceps, soleus, and gastrocnemius were dissected, weighed and flash frozen in liquid nitrogen for biochemical analysis. One batch of mice was sacrificed separately and fresh sciatic nerve was used for the O2K measurements.

# Quantification of Protein Content Using Targeted Proteomics

Protein was isolated from lumbar mouse spinal cords of male and female wild-type and SOD1G93A (pre-onset, onset, and end-stage, n = 5–6) in RIPA buffer: 10 mM Tris-Cl (pH 8.0), 1 mM EDTA, 1% Triton X-100 (v/v), 0.1% sodium deoxycholate (w/v), 0.1% SDS (w/v), 140 mM NaCL, and 1 mM PMSF, with protease inhibitor cocktail (Calbiochem Set III, EDTA-free; EMD Millipore; Billerica, MA, United States) as previously described (Ahn et al., 2018). An aliquot containing 100 µg of each samples was taken, 8 pmol BSA was added as an internal standard, and the samples were incubated at 80◦C for 15 min in 1% SDS (w/v). Proteins were precipitated with 80% acetone (v/v) at −20◦C overnight. The dried protein pellet was reconstituted in 100 µl Laemmli sample buffer and run into a short (1.5 cm) SDS-PAGE gel. The gels were fixed and stained. Each sample was cut from the gel as the entire lane and divided into smaller pieces. The gel pieces were washed to remove the Coomassie blue then reduced, alkylated, and digested overnight with trypsin. The mixture of peptides was extracted from the gel, evaporated to dryness in a SpeedVac and reconstituted in 150 µl 1% acetic acid (v/v) for analysis.

The analyses were carried out on a TSQ Vantage triple quadrupole mass spectrometry system (ThermoScientific TSQ Vantage, San Jose, CA, United States). The HPLC was an Eksigent splitless nanoflow system with a 10 cm × 75 µm i.d. C18 reversed phase capillary column (Eksigent, Dublin, CA, United States). 7 µl aliquots were injected and the peptide eluted with a 60 min gradient of acetonitrile in 0.1% formic acid (v/v). The mass spectrometer was operated in the selected reaction monitoring mode. For each protein, the method was developed to measure 2 ideal peptides. Assays for multiple proteins were bundled together in larger panels. Data were analyzed using the program SkyLine to determine the integrated peak area of the appropriate

chromatographic peaks. The response for each protein was calculated as the geometric mean of the two peptide area. These values were normalized to the response for the BSA standard and multiple housekeeping proteins were also monitored. Entrez Gene name and ID and UniprotKB entry, name, and gene names, and protein names are listed in **Supplemental Table 1** along with results for each protein. The Skyline files containing peak identification and analysis is available at: https://zenodo.org/ record/2633276 (doi: 10.5281/zenodo.2633276).

# Quantitative Real-Time Polymerase Chain Reaction (RT-PCR)

RT-PCR was performed as previously described (Sataranatarajan et al., 2015). Total RNA was extracted from gastrocnemius and lumbar spinal cord of male and female mice using TRIzol reagent (Invitrogen, Carlsbad, CA, United States). Equal amounts of extracted RNA (1 µg) were converted to first strand cDNA using a cDNA synthesis kit (Bio-Rad, Herculus, CA, United States). 5 ng of the cDNA samples was amplified using primers for Afg3l2, Cxcl2, Cxcl5, Chrna1, Chrnd, Chrne, Clpp, Gadd45a, Hspd1, Lonp1, mt-Atp6, mt-Co2, mt-Nd1, Ndufs3, Oma1, Ppargc1a, Rfesd, Rn18s (18S rRNA), Sdha, Sdhb, Sln, Spg7, and Yme1l1 (details in **Supplemental Table 2**) and SYBER green (Invitrogen, Carlsbad, CA, United States). Real time PCR (RT-PCR) was done in Quant Studio 6 (Applied Biosystems, Foster City, CA, United States). The 11C<sup>t</sup> method was used to calculate relative mRNA expression.

#### Primary Fibroblast Cell Culture

Primary fibroblasts were isolated from tail snips of wild-type and SOD1G93A mice as previously described (Pharaoh et al., 2016). Briefly, snips were washed, minced, and incubated overnight with DMEM supplemented with Liberase DL (Roche). The next day the samples were suspended in complete DMEM (4.5 g/l glucose DMEM supplemented with 10% Fetal Bovine Serum and 1% Penicillin/Streptomycin), centrifuged at 200 ×g for 5 min, and the cell pellet was resuspended in complete DMEM and transferred to a 25 mL cell culture flask.

# Mitochondrial Function

Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) of dermal fibroblasts (30,000 cells/well, n = 5, three independent experiments) was measured with a Seahorse XF24 Flux Analyzer under basal conditions and after sequential addition of 1 µM oligomycin (Complex V inhibitor), 1 µM FCCP (proton uncoupler), and 1 µM antimycin A (Complex III inhibitor) normalized to total protein (µg per well) measured by Bradford Protein Assay with BSA standard as previously described (Pharaoh et al., 2016). All OCR measurements were corrected for non-mitochondrial respiration (NMR) using the antimycin A values; basal is the third baseline value, maximal is the highest FCCP value corrected for NMR, reserve capacity is the difference between basal and maximal respiration, ATPlinked respiration is the difference between basal and the lowest oligomycin value, and proton leak is the lowest oligomycin value. For ECAR, basal is the third baseline value, and stressed is the highest value recorded after addition of oligomycin and FCCP.

Gastrocnemius fibers and sciatic nerves were collected from end-stage female mice and placed in ice-cold buffer X for mitochondrial function measurements. OCR and the rate of hydrogen peroxide and lipid hydroperoxide production were simultaneously measured using the Oxygraph-2k (O2k, OROBOROS Instruments, Innsbruck, Austria) in gastrocnemius fibers as previously reported with minor modifications (Ahn et al., 2018). We mechanically separated ∼3–5 mg of red gastrocnemius fibers with forceps in buffer X, permeabilized their plasma membrane with saponin (30 µg/ml) for 30 min, and washed them (3 × 5 min) in buffer Z at 4◦C. We mechanically separated sciatic nerves nerve and minced them into 1–1.5 mm segments in buffer X, permeabilized their plasma membrane with saponin (50 µg/ml) for 30min, and washed them (3 × 5 min) in buffer Z at 4◦C (Zheng et al., 2011).

OCR was measured using an oxygen probe, while AUR reaction rate was measured using the O2k-Fluo LED2-Module Fluorescence-Sensor Green. Measurements were performed on permeabilized fibers and sciatic nerve sections in buffer Z media at 37◦C containing 10 µM Amplex <sup>R</sup> UltraRed (Molecular Probes, Eugene, OR), 1 U/ml horseradish peroxidase (HRP), 2.5 U/ml superoxide dismutase 1 (SOD1), and blebbistatin (25 µM). HRP catalyzes the reaction between peroxides, including hydrogen peroxide and lipid hydroperoxides, and AUR to produce the fluorescent resorufin (excitation: 565 nm; emission: 600 nM). The fluorescent signal was normalized between chambers via a hydrogen peroxide standard curve established on each day of experiments. Rates of respiration and hydrogen peroxide and lipid hydroperoxide production were measured during sequential additions of substrates and inhibitors as follows: baseline, glutamate (10 mM) and malate (2 mM) for Leak, ADP (5 mM) for Complex I-linked respiration, pyruvate (5 mM) for maximal Complex I-linked respiration, succinate (10 mM) for Complex I and II-linked respiration, rotenone (1µM) for Complex II-linked respiration, antimycin A (1 µM) to measure NMR, and TMPD (0.5 mM) immediately followed by ascorbate (5 mM) for Complex IV-linked respiration. All respiration measurements were normalized to antimycin A to account for non-mitochondrial oxygen consumption. Background resorufin production was subtracted from each experimental measurement of hydrogen peroxide and lipid hydroperoxide production. Graphs for AUR reaction rate show measurements from baseline, after addition of rotenone, and after addition of antimycin A. Data for both OCR and rates of hydrogen peroxide and lipid hydroperoxide generation were normalized by milligrams of muscle bundle wet weights. All substrates and inhibitors were prepared according to the protocols from OROBOROS instruments.

For the catalase vs. AACOCF<sup>3</sup> study, 2000 U/ml of catalase was added during permeabilization, washes, and assay, while 20 µM AACOCF<sup>3</sup> was added during permeabilization and washes. 2000 U/ml catalase was used as a final concentration for experimentation in order to have maximal inhibition. One unit of catalase will decompose 1.0 µmole of hydrogen peroxide per minute. Because our chambers are 2 ml, and we use 2000 U/ml,

our catalase concentration is capable of decomposing 4 mmoles hydrogen peroxide H2O<sup>2</sup> per minute. Our measurements are in the pmole hydrogen peroxide per minute scale.

# Statistical Analyses and Data Visualization

Statistical analysis and visualization were performed using GraphPad Prism version 7.0b for OS X (GraphPad Software, La Jolla, CA, United States<sup>1</sup> ). For RT-PCR and targeted proteomics data sets, outliers were removed using the Grubbs' outlier test (Alpha = 0.05). Statistical significance for protein content was determined by ordinary two-way ANOVA with Tukey's Multiple Comparison test and Benjamini–Hochberg procedure. "Genotype" denotes a significant genotype effect in two-way ANOVA with q < 0.05 but no individual Tukey Multiple Comparison test was significant. Significance symbols demonstrate q < 0.05 for a genotype effect and additional significant results for the Tukey Multiple Comparison test. Values for CAT, GPI1, HSPD1, and MDH1 were the average from multiple experiments.

Statistical significance for RT-PCR and longitudinal tissue masses is determined by ordinary two-way ANOVA with Tukey's Multiple Comparison test. "Genotype" denotes a significant genotype effect in Benjamini-Hochberg corrected two-way ANOVA with p < 0.05 but no individual Multiple Comparison test was significant. Significance symbols demonstrate p < 0.05 for a genotype effect and additional significant results for the Tukey Multiple Comparison test. For eWAT end-stage mass, statistical significance is determined by unpaired twotailed student's t-test (∗p < 0.05). Unpaired two-tailed student's t-test was used for comparisons between two groups in the mitochondrial function experiments, and ordinary one-way ANOVA with Tukey's Multiple Comparison Test was used for the OROBOROS O2K experiment of gastrocnemius fibers from SOD1G93A mice treated with catalase or AACOCF3.

Heatmap and the PCA plot were generated using ClustVis with default settings (Row scaling = unit variance scaling, PCA method = SVD with imputation, clustering distance for rows = correlation, clustering method for rows = average, tree ordering for rows = tightest cluster first) (Metsalu and Vilo, 2015).

<sup>1</sup> www.graphpad.com

#### REFERENCES


# ETHICS STATEMENT

This study was carried out in accordance with the recommendations of USDA guidelines by the Institutional Animal Care and Use Committee at Oklahoma Medical Research Foundation. The protocol was approved by the Institutional Animal Care and Use Committee at Oklahoma Medical Research Foundation.

# AUTHOR CONTRIBUTIONS

Study was conceived and designed by HVR. Experiments were performed and data collected by GP, KSa, KSt, SH, JG, CK, and MK. BA developed the sciatic nerve mitochondrial experimental procedure used here. Data analysis and visualization was performed by GP. The manuscript was written by GP with input from HVR. All authors edited and approved the final manuscript.

# FUNDING

This work was supported by a grant from the Pape Adams Foundation to HVR, the Oklahoma Medical Research Foundation (OMRF) Fleming Scholar Program, and the OMRF John and Mildred Carson Ph.D. Scholarship Fund and The National Institute on Aging T32 AG052363 training fellowships awarded to GP.

# ACKNOWLEDGMENTS

We would like to thank the Oklahoma Nathan Shock Center for Excellence in the Biology of Aging Multiplexing Protein Quantification Core.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fnins. 2019.00487/full#supplementary-material

phosphorylation to glycolysis. Neurobiol. Aging 35, 1499–1509. doi: 10.1016/ j.neurobiolaging.2013.11.025




lateral sclerosis detected by GC/TOFMS. PLoS One 6:e17947. doi: 10.1371/ journal.pone.0017947


**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 © 2019 Pharaoh, Sataranatarajan, Street, Hill, Gregston, Ahn, Kinter, Kinter and Van Remmen. 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.

# The Gut and Parkinson's Disease—A Bidirectional Pathway

Susanne Fonseca Santos <sup>1</sup> , Hadassa Loth de Oliveira<sup>2</sup> , Elizabeth Sumi Yamada<sup>1</sup> , Bianca Cruz Neves <sup>2</sup> and Antonio Pereira Jr. 1,3 \*

<sup>1</sup> Graduate Program in Neuroscience and Cell Biology, Institute of Biology, Federal University of Pará, Belém, Brazil, <sup>2</sup> Department of Biochemistry, Institute of Chemistry, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil, <sup>3</sup> Department of Electrical and Biomedical Engineering, Institute of Technology, Federal University of Pará, Belém, Brazil

Humans evolved a symbiotic relationship with their gut microbiome, a complex microbial community composed of bacteria, archaea, protists, and viruses, including bacteriophages. The enteric nervous system (ENS) is a gateway for the bidirectional communication between the brain and the gut, mostly through the vagus nerve (VN). Environmental exposure plays a pivotal role in both the composition and functionality of the gut microbiome and may contribute to susceptibility to neurodegenerative disorders, such as Parkinson's disease (PD). The neuropathological hallmark of PD is the widespread appearance of alpha-synuclein aggregates in both the central and peripheral nervous systems, including the ENS. Many studies suggest that gut toxins can induce the formation of α-syn aggregates in the ENS, which may then be transmitted in a prion-like manner to the CNS through the VN. PD is strongly associated with aging and its negative effects on homeostatic mechanisms protecting from inflammation, oxidative stress, and protein malfunction. In this mini-review, we revisit some landmark discoveries in the field of Parkinson's research and focus on the gut-brain axis. In the process, we highlight evidence showing gut-associated dysbiosis and related microbial-derived components as important players and risk factors for PD. Therefore, the gut microbiome emerges as a potential target for protective measures aiming to prevent PD onset.

#### Edited by:

Foteini Christidi, National and Kapodistrian University of Athens, Greece

#### Reviewed by:

Kun Lu, University of North Carolina at Chapel Hill, United States Denis Soulet, Laval University, Canada

> \*Correspondence: Antonio Pereira Jr. apereira@ufpa.br

#### Specialty section:

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

Received: 25 January 2019 Accepted: 15 May 2019 Published: 04 June 2019

#### Citation:

Santos SF, de Oliveira HL, Yamada ES, Neves BC and Pereira A Jr (2019) The Gut and Parkinson's Disease—A Bidirectional Pathway. Front. Neurol. 10:574. doi: 10.3389/fneur.2019.00574 Keywords: Parkinson's disease, enteric nervous system, microbiome, neurotoxicants, probiotics

# INTRODUCTION

Parkinson's Disease (PD) is a common neurodegenerative disorder typically associated with the progressive loss of dopaminergic neurons located in the midbrain nucleus substantia nigra pars compacta (SNpc) (1). Although the cardinal symptoms of PD are motor impairments attributed to the depletion of the neurotransmitter dopamine in the striatum, a major target of the SNpc (2), it has been long recognized [for review, see (3)] that other non-motor symptoms, including olfactory (4–6) and gastrointestinal (GI) dysfunction (4), appear during the so-called premotor phase of the disease.

The neuropathological hallmark of PD is the presence of cytoplasmic inclusions, called Lewy bodies (LB) or Lewy neurites (7–9), in SNpc neurons (10). LBs are composed mostly of α-synuclein (α-syn) aggregates (11–13), whose aberrant soluble oligomeric conformations are thought to mediate its toxic effects (14). Alpha-syn is an intrinsically disordered protein (IDP), which lacks a stable 3D structure under physiological conditions and is characterized by exacerbated structural plasticity and conformational adaptability (15). As other IDPs possessing amyloidogenic

**83**

regions (16), α-syn can turn into a promiscuous binder leading to abnormal interactions and the development of PD (17). Tuttle et al. (18) provided a detailed 3D structure of functional αsyn fibrils (see **Figure 1**), using solid-state NMR spectroscopy. The study may serve as the basis for a better understanding of molecular mechanisms involved in α-syn fibril nucleation and propagation. In addition, such structural information may provide useful insights on possible interactions of α-synuclein with other proteins and small molecules and allow the emergence of new tools with potential to facilitate both the diagnosis and treatment of PD (e.g., imaging agents and therapeutic drugs).

Aggregates of α-syn fibrils are also found in neural tissue located outside the central nervous system (CNS) of PD patients, in both the autonomic and enteric nervous system (ENS), an outcome which may be associated with the non-motor symptoms of the disease [for review, see (3)]. These findings led Braak et al. (4) to propose a staging system for the progression of the disease following a specific pattern of α-syn aggregates spreading from peripheral toward more centralized locations in the brain. The triggering event would be the invasion of vulnerable neural structures such as the olfactory epithelium and the ENS, which interface directly with the external environment (5, 22), by a neurotoxicant ("neurotropic virus") (23). While both structures (24, 25) possess immunological and physical barriers protecting them against environmental insults, these barriers steadily deteriorate with aging [for review, see (26, 27)], which is the biggest risk factor for idiopathic PD (28).

Animal studies have supported the claim that α-syn aggregates propagate in a prion-like manner [(29); for review, see (30)] via microtubule-associated transport along axons (31). In summary, the prion hypothesis of PD proposes that amyloidogenic αsyn would induce a conformational change in the endogenous protein through permissive templating, convert it into a likeness of itself (32, 33) and propagate retrogradely through the vagus nerve or the olfactory tract from the ENS or the olfactory bulb, respectively. Even though definitive proof for this prion hypothesis is still missing (30) and there is the controversial possibility that intestinal α-syn aggregates have a brain origin (34, 35), it has been shown that vagotomy is associated with a decreased risk for PD in humans (36, 37). Also, grafted neurons in PD patients develop α-syn aggregate pathology (38–40) and α-syn from PD patients can cause nigrostriatal degeneration in mice and non-human primates (41). Remarkably, exogenous αsyn fibrils, either PD patient-derived or produced in E. coli, were able to seed the formation of LB-like inclusions which spread from the GI tract to the brain through the vagus nerve in rats (31).

Prior to Braak's hypothesis, however, there was already strong evidence pointing to the role played by exogenous toxins in the etiology of sporadic PD. For instance, postencephalitic parkinsonism (von Economo's disease), which has an autoimmune basis caused by a viral illness (42), is associated with degeneration of the basal ganglia (43). Additionally, the discovery of parkinsonism induced by 1-methyl-4-phenyl-1,2,4,5-tetrahydropyridine (MPTP) through self-administration, in 1982 (44) brought to light a new class of xenobiotic substances that may cause PD-like symptoms by environmental contact. MPTP is a lipophilic compound which readily passes into the brain where it is converted by monoamine oxidase B (MAO-B) to 1-methyl-4-phenylpyridinium (MPP+) (45) which is taken up by dopaminergic cells and impairs mitochondria respiration by poisoning complex 1 (46). There are many heterocyclic molecules that structurally resemble MPTP and are found in the brain from both endogenous and exogenous sources, such as tetrahydroisoquinolines (TIQ) and β-carbolines (β-C). For instance, a TIQ derivative, salsolinol, which is produced by enterobacteria (47) and has been found in the urine of PD patients, may have a double-faced, dose-dependent effect on the nigrostriatal pathway as either a harmful or protective agent (48).

The evidence for the role played by toxins in inducing parkinsonism and the relative scarcity of familial cases (about 10%) (49) underscore the importance of environmental and lifestyle factors over genetic ones in the etiology of the disease (50–52). Some chronic diseases have been associated with a phenomenon called evolutionary mismatch when ancestral traits are no longer adaptive in modern contexts (53, 54). For instance, α-syn is involved with normal synaptic function by regulating, among other things, the size of presynaptic vesicles (55) and the assembly of SNARE proteins involved with the docking of synaptic vesicles to presynaptic membranes (56). However, as old age became common in humans after the early upper Paleolithic (57), the steady increase in longevity seen in modern times may have had a collateral effect on the protein homeostasis (proteostasis) network, which coordinates protein synthesis, folding, trafficking, disaggregation, and degradation (58, 59). The breakdown of proteostasis, which is a common feature of many neurodegenerative diseases (60), means that misfolded proteins may accumulate due to lack of clearance or failure to refold into their native structures (61). In the case of prion-like proteins, this may cause further protein misfolding (template effect) leading to protein aggregation and ultimately cell death (62).

# THE GUT-BRAIN AXIS AND PARKINSON'S DISEASE

The gut-brain axis is mediated by intense bidirectional communication between the CNS and the ENS (63). Through the ENS, the gut microbiota influences the development and function of all divisions of the nervous system (64) and this association was established very early during the evolution of multicellular organisms. The first nervous system appeared more than 500 million years ago before the divergence of cnidarians and bilaterians, the two metazoan sister groups (65). That primitive brain had a simple structure, organized as a diffuse nerve net which controlled a restricted set of basic behaviors and was the template for the subsequent evolution of the mammalian ENS (66–68), which retained many of its basic structural characteristics, such as a network of nervous ganglia distributed in the myenteric and submucous plexuses (69). Higher vertebrates went to evolve an additional set of neural structures in the central nervous system (CNS), tasked with the control of more sophisticated behaviors (70). However, the ENS and the CNS maintain intense crosstalk through reciprocal connections mediated by the VN (**Figure 1**) and pelvic nerve in mammals

based on atomic-resolution molecular data from NGL Viewer (19). Members of the gut microbiome and their extracellular compounds may trigger responses in the VN through enteroendocrine cells, which are contacted by vagus nerve terminals through specialized structures called neuropods (NP) (20). Microbial antigens can cross the gut epithelium through microfold cells, playing a central role in localized inflammatory responses [adapted from Bohórquez et al. (21)]. Toll-like receptors are microbe-sensing proteins, present in intestinal epithelial cells, mediating recognition of commensal bacteria from the harmful/inflammatory ones. ENS, enteric nervous system; M, microfold cells; NP, neuropods; PP, Peyer's patches; TLR4, Toll-like receptor 4; VN, vagus nerve.

(71, 72). As the main substrate for this information exchange, the vagus nerve is an attractive target of neurostimulation therapies for the treatment of psychiatric and gastrointestinal disorders (73, 74).

The GI tract harbors a complex microbial ecosystem (**Figure 1**), consisting of bacteria, archaea, protists, and eukaryotic and prokaryotic viruses, also known as bacteriophages (75–77). The human microbiome has coevolved with its host (78), which keeps a tight leash on the intrinsic competitive nature of the microorganisms that comprise the microbiome, through both the nervous (71, 79, 80) and the immune systems (81, 82). This arrangement maximizes the benefits the host gains from the symbiotic relationship, including protection against pathogens, improved nutrition, and mental health (81). A sub-type of intestinal epithelial cells called enteroendocrine cells, provide a signaling pathway through which the microbiome interacts with the CNS via the vagus nerve (20, 83). Enteroendocrine cells have diverse phenotypes and express a variety of peptides/hormones that can act as signaling molecules on distinct targets, both local and distant, and some are chemoreceptors responding to a variety of luminal stimuli (84, 85). As other intestinal epithelial cells, enteroendocrine cells express toll-like receptors (86), allowing them to detect bacterial products, and activate vagal afferents through basal processes called neuropods (see **Figure 1**) (20, 87).

# THE GUT MICROBIOME AND BRAIN FUNCTION

There is increasing evidence of the association between microbiome dysfunction and CNS-related co-morbidities, such as anxiety, depression, autism spectrum disorders, Alzheimer's disease and PD (88–92). This association probably arose as a by-product of natural selection forces acting on microorganisms to adapt to the host and vice-versa (93). The effect of the microbiota on the CNS can lead to behavior modifications (93–95) and even to host manipulation (96) associated with increasing fitness of its bacterial populations. For instance, the microbiome can influence social interactions by acting on the nutritional behavior of individual animals, particularly those from social species where individuals share microbes and interact around foods (97). The proximate neuro-endocrinological and inflammatory mechanisms underlying this type of host manipulation are largely shared by the microbiome and the host (98, 99). For instance, levels of many neurotransmitters that are important for the expression of social behavior, such as serotonin (5-HT), dopamine, norepinephrine (NE), γaminobutyric acid (GABA), and glutamate are either expressed or regulated by bacteria (100–102). Particularly, most of the body's serotonin (5-HT) (5-hydroxytryptamine) is produced in the gut by enterochromaffin cells (EC) under the influence of the microbiome (103). The activation of 5-HT<sup>4</sup> receptors induces the maturation of the ENS and regulates its adult function (104). In the gut, there are three major metabolic pathways leading from the essential amino acid tryptophan (Trp) to 5-HT, kynurenine (Kyn), and indole derivatives, which are under the direct or indirect control of the microbiota (105). During inflammatory states, most tryptophan is diverted to the production of Kyn and its metabolites kynurenine acid (KYNA) and quinolinic acid (QUIN) (106). While KYNA is considered neuroprotective, QUIN can cause excitotoxicity as an agonist of N-methyl-d-aspartate (NMDA) receptor and contribute to the neuropathogenesis of PD [for review, see (107)].

Although α-syn aggregates are also seen in the ENS of normally aging subjects (108), especially in the appendix (109), it is more prevalent in PD patients (110). Recent in vivo models showed that accumulation of α-syn aggregates in the ENS can be induced by alterations in the gut microbiome (111). Interestingly, Sampson et al. (112) demonstrated in mice, genetically modified to overexpress α-syn, that the presence of gut microbiota is necessary to promote pathological alterations and motor deficits similar to PD. They also demonstrated that fecal transplants from PD patients impair motor function in the same mouse strain, strongly suggesting that gut microbes may play a pivotal role in the onset of synucleinopathies such as PD (112). Underlying these findings is the fact that microbial amyloids produced by some members of the gut microbiota can be released in the extracellular space, where they can be internalized by neighboring cells, including neurons, and seed the formation of pathological aggregates of endogenous αsyn through permissive templating (113, 114). The failure of normal clearance mechanisms such as the ubiquitin-proteasome system, characteristic of both familial and idiopathic PD (115), to degrade the misfolded protein, may facilitate the seeding process.

The concept of microbial dysbiosis also comprises the bacteriophage components of the microbiome (116). Bacteriophages (phages) are viral parasites of bacteria and are important regulators of host-microbiome interactions through horizontal gene transfer and antagonistic coevolution (117, 118). Besides targeting bacteria, phages can impact human health by playing a direct role on intestinal inflammatory processes (119) and possibly causing α-syn misfolding (120). A recent study showed significant differences in the gut phagobiota of PD patients and healthy individuals and a depletion of Lactococcus bacteria (121) in the former, which is associated with the regulation of gut permeability (122) and dopamine production (102), two factors linked with the early signs of PD in the gut (123). Phage therapy has recently returned to the spotlight as an alternative antimicrobial strategy (124, 125). Eventually, it may also contribute to fighting PD through targeted approaches to manipulate the microbiome (121).

Probiotic bacteria have been linked to improved GI symptoms associated with PD (126). Probiotics affect the functionality of the CNS through beneficial interactions with the commensal gut microbiota and modulation of gut-derived inflammation (127). The microbiota of PD patients exhibits a pro-inflammatory profile (128, 129) due to increased intestinal permeability to endotoxins (lipopolysaccharide) (130). Bacterial amyloids may also favor a pro-inflammatory environment in the gut (131). A common bacterial component, the Curli fimbriae, share structural and biophysical properties with amyloids and are produced by E. coli through coordinated biosynthetic processes (132). Other components of the gut microbiome are also known to produce functional extracellular amyloids [e.g., Salmonella, Klebsiella, Citrobacter, and Bacillus species; (133)]. Since probiotic treatment induces an anti-inflammatory peripheral immune response in multiple sclerosis patients (134) there is a possibility they may also be beneficial for PD patients, although there are no reports corroborating this hypothesis. One option is to take advantage of Lactobacilli's ability to inhibit the formation of biofilms by pathogenic bacteria (135, 136). One caveat, however, is that the effects of probiotics are highly variable, being person-specific, as shown in a recent study (137). This limitation may be counteracted with the use of geneticallymodified probiotics able to deliver novel therapeutics efficiently and with site specificity (138). Despite the increasing number of probiotic products available to consumers and the aggressive marketing proclaiming their efficacy, there have been few studies addressing concerns about efficacy and, more importantly, the safety of these products (139). There is an urgent need for more studies about the therapeutic potential of specific bacterial strains to help maintain oxidative and protein homeostasis in the ENS.

# CONCLUDING REMARKS

Aging is the main risk factor for the development of PD (140) and delaying the aging process is neuroprotective to PD in animal models (141). Aging is also associated with the accumulation of neuroinflammatory sequelae and the breakdown of homeostatic mechanisms that protect against protein misfolding, oxidative stress, decreased mitochondrial function, etc. The gut, as one of the main gateways to environmental exposure to the brain, may contribute to increasing the susceptibility to these factors. The microbiome has a protective effect mediating this exposure, and dysbiosis seems to be a pivotal risk factor for PD and other neurological disorders. Thus, the adoption of preventive measures to ensure a healthy microbiome throughout the lifetime can potentially decrease the risk of developing PD and other neurodegenerative diseases. The widespread use of antibiotics, for instance, which can kill gut bacteria indiscriminately, can cause a shift of the microbiome to an alternative stable state with unknown consequences in the long term (142).

# AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

# REFERENCES


induces ascending Parkinsonism in the rat. NPJ Parkinson Dis. (2018) 4:30. doi: 10.1038/s41531-018-0066-0


systems. Neurogastroenterol Motil. (2018) 30:13234. doi: 10.1111/nmo. 13234


disease. Bull Exp Biol Med. (2017) 162:734–7. doi: 10.1007/s10517-017- 3700-7


**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 © 2019 Santos, de Oliveira, Yamada, Neves and Pereira. 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(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.

# The Peripheral Nervous System in Amyotrophic Lateral Sclerosis: Opportunities for Translational Research

Francesco Gentile<sup>1</sup> , Stefania Scarlino<sup>1</sup> , Yuri Matteo Falzone1,2, Christian Lunetta<sup>3</sup> , Lucio Tremolizzo<sup>4</sup> , Angelo Quattrini<sup>1</sup>† and Nilo Riva1,2 \* †

<sup>1</sup> Experimental Neuropathology Unit, Division of Neuroscience, Institute of Experimental Neurology – San Raffaele Scientific Institute, Milan, Italy, <sup>2</sup> Department of Neurology, San Raffaele Scientific Institute, Milan, Italy, <sup>3</sup> NeuroMuscular Omnicentre, Serena Onlus Foundation, Milan, Italy, <sup>4</sup> Neurology Unit, ALS Clinic, San Gerardo Hospital, University of Milano-Bicocca, Monza, Italy

#### Edited by:

Foteini Christidi, National and Kapodistrian University of Athens, Greece

#### Reviewed by:

Chiara F. Valori, German Center for Neurodegenerative Diseases (DZNE), Germany Yasushi Kitaoka, St. Marianna University School of Medicine, Japan

#### \*Correspondence:

Nilo Riva riva.nilo@hsr.it †These authors share last senior authorship

#### Specialty section:

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

Received: 15 February 2019 Accepted: 27 May 2019 Published: 25 June 2019

#### Citation:

Gentile F, Scarlino S, Falzone YM, Lunetta C, Tremolizzo L, Quattrini A and Riva N (2019) The Peripheral Nervous System in Amyotrophic Lateral Sclerosis: Opportunities for Translational Research. Front. Neurosci. 13:601. doi: 10.3389/fnins.2019.00601 Although amyotrophic lateral sclerosis (ALS) has been considered as a disorder of the motor neuron (MN) cell body, recent evidences show the non-cell-autonomous pathogenic nature of the disease. Axonal degeneration, loss of peripheral axons and destruction of nerve terminals are early events in the disease pathogenic cascade, anticipating MN degeneration, and the onset of clinical symptoms. Therefore, although ALS and peripheral axonal neuropathies should be differentiated in clinical practice, they also share damage to common molecular pathways, including axonal transport, RNA metabolism and proteostasis. Thus, an extensive evaluation of the molecular events occurring in the peripheral nervous system (PNS) could be fundamental to understand the pathogenic mechanisms of ALS, favoring the discovery of potential disease biomarkers, and new therapeutic targets.

Keywords: motor neuron disease, lower motor neuron syndrome, nerve, neuropathy, CMT, distal SMA, hereditary neuropathy, genetics

#### INTRODUCTION

Amyotrophic lateral sclerosis (ALS) is the most common and severe form of motor neuron disease (MND), an heterogeneous group of disorders defined by prominent motor neuron (MN) degeneration (Saberi et al., 2015; Riva et al., 2016). Recent evidence has challenged such traditional view of selective neuronal loss demonstrating widespread extra-motor involvement in ALS and implying that neuronal populations other than MNs may also be affected (Geser et al., 2008; Brettschneider et al., 2013; McCluskey et al., 2014). Furthermore, an expanded clinical spectrum has now been recognized, as overlapping phenotypes with other neuromuscular disorders (Sabatelli et al., 2016), including peripheral neuropathies, have been described (Rajabally and Jacob, 2008; Bhatt et al., 2009; Sawa et al., 2012). Although in clinical practice MNDs should be differentiated from other peripheral nervous system (PNS) disorders, sensory and autonomic neurons in the dorsal root ganglia (DRG) and lower MNs in the ventral horns share important challenges, as proper stability and functioning of their long projections throughout the body requires a protective environment and efficient communication between the central nervous system (CNS) and the outermost areas of these cells. Nonetheless, differences in

morphology and function may confer different patterns of resistance or vulnerability to specific stressors.

The aim of this review is to dissect PNS involvement in ALS, analyzing the evidence from clinical and pathological data to genetics in order to provide novel pathophysiological insights and potential implication for therapeutic strategies.

# DIAGNOSTIC CHALLENGES IN ALS AND ALS-MIMICS

In the absence of pathognomonic diagnostic biomarkers, ALS diagnosis relies on clinical findings suggestive of selective upper, and lower MN (LMN) lesion and the exclusion of alternative causes (Ludolph et al., 2015). The wide clinical heterogeneity of the disease led to the development of several phenotypic classifications, mainly dependent on the distribution of UMN and LMN lesions. In patients presenting with isolated signs of LMN involvement (LMN Syndrome – LMNS), the differential diagnosis may be more challenging since the primary disease target may be the cell soma, the axon and its myelin, the neuromuscular junction (NMJ) or the muscle (Riva et al., 2011b; Garg et al., 2017; Muller et al., 2018). Such disorders not only share the motor unit as a common target of damage, but may be potentially underpinned by common pathogenic mechanisms. Within this context, ALS, with special reference to its restricted LMN phenotypes, such as flair legs and flail arm syndromes, shows high overlap in terms of clinical presentation, targets and mechanisms of damage with hereditary axonal neuropathies, such as axonal Charcot-Marie-Tooth neuropathy (CMT2) and distal hereditary motor neuropathy (dHMN) (Fischer and Glass, 2007). Besides the different rate of progression and spreading, distinction between these disorders could be difficult especially in the early phase, as clinical and neurophysiologic studies may be strikingly similar and may not able to discern the primary lesion site. Nonetheless, a recent study supported the value of the electroneurographic split hand index has recently been suggested for distinguishing ALS from mimic disorders (Menon et al., 2014). Moreover, pes cavus, traditionally considered a distinguishing clinical sign of CMT, has also been anecdotally reported in ALS patients (Ceroni et al., 2001), and is estimated to be as high as 2% our case series (data not published). Notably, the percentage of misdiagnosis in degenerative LMNS has been reported to be as high as 19%; moreover, up to 10% of patients initially diagnosed as ALS are ultimately re-diagnosed as having another disease, including peripheral neuropathy (Davenport et al., 1996; Traynor et al., 2000; Visser et al., 2002). Therefore, in some cases, MND diagnosis remains uncertain, and only follow up can lead to a satisfying level of diagnostic confidence. Of note, we previously showed that biopsy of obturator nerve may be useful in the distinction between ALS and motor neuropathies (Corbo et al., 1997; Riva et al., 2011b). Signs of acute axonal damage, focal/multifocal fiber loss and absent or scarce axonal regeneration are consistently observed in motor nerve biopsies from patients with ALS (**Figure 1A**), while neuropathies show a higher regeneration

capacity and may present signs of myelin damage, inflammation, or pathologic deposits (Benedetti et al., 2010).

# THE PNS AS A CONVERGING POINT IN ALS

# Evidence of Peripheral Nervous System Involvement in Human ALS

Although the circumstances leading to MN damage in ALS are still unknown, different mechanisms have been proposed to explain disease onset and spread. The dying-forward hypothesis states that the primary site of damage resides in the cell soma and subsequently spreads to the peripheral compartments. In the recent years, the dying-back pattern of degeneration has obtained a large attention in the context of ALS pathophysiology (Fischer et al., 2004), suggesting that ALS is a distal axonopathy, a pattern typically seen in peripheral neuropathies with a distal to proximal gradient of damage (England and Asbury, 2004). Therefore, considering that both motor and sensory nerves share mechanisms of axonal degeneration, it may be plausible that in ALS signs of PNS dysfunction may also extend to the sensory system. Notably, other neurodegenerative diseases such as Parkinson's disease and Alzheimer's disease show evidence of distal axonopathy (Arendt, 2009; Chung et al., 2009).

While diagnostic assessment in ALS relays on the assumption that non-motor components of the PNS are usually spared, increasing amount of evidence suggests the presence of sensory and autonomic dysfunction in ALS. Sensory symptoms are reported in about 2–30% of cases (Gubbay et al., 1985; Hammad et al., 2007). Numbness, tingling and pain are the most common symptoms reported. Decreased vibration sense and abnormal pain and temperature thresholds are rarely observed during standard examination, however quantitative sensory testing shows abnormal findings in about 10% of patients, rising to almost half of the cases in advanced stages of the disease (Truini et al., 2015; Isak et al., 2017).

Autonomic dysfunction has been reported in 5–30% of cases, while quantitative autonomic testing detected abnormal sudomotor and cardiovagal response in up to 50% of patients (McCluskey et al., 2014; Piccione et al., 2015). Abnormalities in either sensory nerve amplitudes or nerve conduction velocity may be apparent in about 20% of cases (Mondelli et al., 1993; Pugdahl et al., 2008), with higher frequencies reported in patients with longer disease duration (Pugdahl et al., 2007; Isak et al., 2016).

Pathologic studies of the sural nerve in ALS have demonstrated a mild reduction of the number of large nerve fibers (**Figure 1B**) in most of the examined cases (Bradley et al., 1983; Hammad et al., 2007; Devigili et al., 2011; Luigetti et al., 2012). Clustering of mitochondria, dilation of small vesicles and accumulation of neurofilaments (NF) have also been reported (Dyck et al., 1975; Di Trapani et al., 1987; Heads et al., 1991). Skin biopsy shows in about 80% of ALS patients a reduction of intraepidermal nerve fiber density (Weis et al., 2011; Truini et al., 2015; Dalla Bella et al., 2016; Nolano et al., 2017) and focal axonal swellings, suggesting axonal degeneration (Isak et al., 2017). Interestingly, small nerve fibers in the cornea are also reduced in ALS patients, correlating with bulbar disability scores (Ferrari et al., 2014). Finally, a significant neuronal loss has been observed in DRG of ALS patients, with preferential involvement of large-sized cells (Kawamura et al., 1981). Notably, evidence of pTDP-43 pathology has been shown in Clarke's column, intermediolateral nucleus and dorsal root ganglia in a subset of cases, proving that these structures are also affected (Nishihira et al., 2008; Brettschneider et al., 2014).

# The Role of Genetics in ALS: At the Borders of the Disease Spectrum

The knowledge about ALS genetics has had a relevant acceleration in the last decade, with more than 30 genes identified as pathogenic and more than 100 as ALS-related (Riva et al., 2016). The first gene identified as a cause of ALS was SOD1 in 1993 (Rosen et al., 1993) and together with TARDBP (Sreedharan et al., 2008), FUS (Kwiatkowski et al., 2009), and chromosome 9 open reading frame 72 (C9orf72) (DeJesus-Hernandez et al., 2011; Renton et al., 2011), represent the most common ALSrelated genes, covering up to 70% of familial and 10% of sporadic cases (Riva et al., 2016; Chia et al., 2018). ALS-key genes have been associated with almost all clinical ALS phenotypes, although patients with SOD1, and FUS mutations may show preferential LMN involvement (Waibel et al., 2013; Picher-Martel et al., 2016).

The frequency of gene variants in cases with extramotor involvement is significantly higher compared to the pure ALS phenotype, and has been associated with poorer survival (McCluskey et al., 2014). Prominent sensory involvement had been described in rare mutated cases, with SOD1 and TARDBP genes being reported most frequently (Camdessanche et al., 2011).

Many of the genes involved in ALS are associated also with other neurological disorders sharing common targets of neurodegeneration, such as hereditary spastic paraplegia (HSP), axonal CMT neuropathy and dHMN. Among the genes involved in dHMN and axonal CMT there are Berardinelli-Seip Congenital Lipodystrophy 2 (BSCL2), neurofilament light (NEFL), transient receptor potential cation channel subfamily V member 4 (TRPV4), heat shock protein family B1, B3, and B8 (HSPB1, HSPB3, and HSPB8) (Martini et al., 2000; Rossor et al., 2012). Despite the discrepancy between the major genes mutated in hereditary neuropathies and ALS, each of them exerts pleiotropic functions in neuronal homeostasis, including RNA metabolism, protein quality control, axonal transport, stress response. Therefore, generation of a spectrum of clinical phenotypes from alteration in master genes, broadly involved in key neuronal metabolic pathways, could be expected (**Table 1**).

Neurofilaments (NF) are composed of three subunits defined by their molecular weight and encoded by the NF light (NEFL), medium (NEFM), and heavy (NEFH) genes (Lieberburg et al., 1989). NFs are specifically expressed in neurons and are the most abundant cytoskeletal components of myelinated axons, regulating their caliber, growth, and conduction rate (Hoffman et al., 1987). Mutations in NEFL are known to cause both axonal and demyelinating forms of CMT with different phenotypes, including pyramidal signs (Mersiyanova et al., 2000; Jordanova et al., 2003; Rebelo et al., 2016; Jacquier et al., 2017). Mutations in NEFH gene are involved in the pathogenesis of sporadic ALS (Figlewicz et al., 1994; Al-Chalabi et al., 1999) but also in CMT (Jacquier et al., 2017; Nam et al., 2017).

The cellular abundance of PI(3,5)P2, a phosphoinositide involved in the control of vesicles trafficking, is regulated by a phosphoinositide 5-phosphatase encoded by the FIG4 gene. CMT4J cases, clinically characterized by early onset and aggressive disease progression, have been associated with by an autosomal dominant pattern of transmission and by biallelic FIG4 mutations (Nicholson et al., 2011). Notably, heterozygous autosomal dominant FIG4 variants have been more recently associated with ALS and identified as ALS11 (Osmanovic et al., 2017).

The valosin containing protein (VCP) is member of the AAA ATPase family of proteins. This protein is ubiquitously expressed, and it is implicated in multiple cellular processes, such as cell survival (Vandermoere et al., 2006; Braun and Zischka, 2008), stress response and DNA and protein quality control (DeLaBarre et al., 2006; Ju et al., 2009; Weihl et al., 2009). Mutations in VCP have been described in patients with autosomal dominant inclusion body myopathy (IBM) associated with Paget disease and fronto-temporal dementia (FTD) (IBMPFD) (Watts et al., 2004), pure ALS patients (Johnson et al., 2010; Miller et al., 2012), and recently also CMT (Gonzalez et al., 2014; Jerath et al., 2015).

The exact function of senataxin (SETX) is unknown but it may be involved in RNA metabolism. Studies have shown a role in DNA transcription and repair (Suraweera et al., 2009; Skourti-Stathaki et al., 2011). Mutations in STX have been described in ataxia-ocular apraxia 2 (AOA2) (Moreira et al., 2004; Duquette et al., 2005; Fogel and Perlman, 2006; Arning et al., 2008; Airoldi et al., 2010; Fogel et al., 2014), autosomal dominant juvenile ALS (Chen et al., 2004; Zhao et al., 2009; Avemaria et al., 2011; Arning et al., 2013; Tripolszki et al., 2017), and in dHMN with pyramidal features (De Jonghe et al., 2002).

TABLE 1 | Key genes associated with ALS, hereditary neuropathy, and overlapping phenotypes.


(Continued)

**94**

#### TABLE 1 | Continued

fnins-13-00601 June 22, 2019 Time: 14:10 # 5


<sup>∗</sup>OMIM and report from literature. ALS, amyotrophic lateral sclerosis; FTD, frontotemporal dementia; CMT, charcot marie tooth; SMA, spinal muscular atrophy; SPG, spastic paraplegia; HMSN, hereditary motor and sensory neuropathy; Dhmn, distal hereditary motor neuropathy; UPS, ubiquitin proteasome system; ERAD, endoplasmic reticulum-associated protein degradation; IBMPFD, inclusion body myopathy paget disease frontotemporal dementia.

The Spastic Paraplegia 11 gene (SPG11) encodes the spatacsin protein, selectively expressed in neuron with a role in axonal growth, transport, and cytoskeletal stability (Perez-Branguli et al., 2014). SPG11 variants were first described in patients with autosomal recessive spastic paraplegia 11 with thin corpus callosum (Hehr et al., 2007). Then, descriptions in juvenile ALS (ALS5) and in CMT2X were also reported (Orlacchio et al., 2010; Daoud et al., 2012; Iskender et al., 2015; Montecchiani et al., 2016).

The dynactin subunit 1 (DCTN1) and kinesin family member 5A (KIF5A) genes encode for dynactin and kinesin subunits, involved in retrograde and anterograde axonal transport, respectively (Hirokawa et al., 2009; Kwinter et al., 2009). DCTN1 and KIF5A variants have been variably described in ALS (Munch et al., 2004, 2005; Liu et al., 2014, 2017; Brenner et al., 2018; Nicolas et al., 2018), CMT, and dHMN (Crimella et al., 2012; Lopez et al., 2015).

Finally, pleckstrin homology and RhoGEF domain containing G5 (PLEKHG5) and mitofusin 2 (MFN2) genes, whose mutations are known to cause CMT (Muglia et al., 2001; Zuchner et al., 2004a, 2006; Engelfried et al., 2006; Del Bo et al., 2008; Nicholson et al., 2008; Azzedine et al., 2013; Kim et al., 2013; Iapadre et al., 2018), have been also reported in patients with ALS or an "ALS-like" phenotype (Marchesi et al., 2011; Ozoguz et al., 2015).

#### PATHOGENIC MECHANISMS OF PNS DAMAGE IN ALS

#### Peripheral Motor and Sensory Dysfunction in ALS Models

The strongest evidence supporting the dying-back pattern of degeneration comes from transgenic mice overexpressing the mutated form of human SOD1 G93A (hSOD-1G93A), the most studied ALS model. Axonal pathology precedes spinal MN death and symptom onset, with the NMJ being the first site of morphological alteration (Fischer et al., 2004). Along with disease progression, the burden of axonal pathology overcomes the moderate MN loss in spinal cord, and suggesting that the motor phenotype in these mice models is mainly driven by the PNS damage (Gould et al., 2006; Hegedus et al., 2007). Indeed, interventions leading to complete rescue of MNs in the spinal cord of mutant SOD1 (mSOD1) ALS mice were not sufficient to halt axonal loss and motor phenotype, suggesting that mechanisms of central, and peripheral degeneration may be at least partially independent (Gould et al., 2006; Rouaux et al., 2007; Suzuki et al., 2007).

Despite the wealth amount of data generated from this model about MN degeneration, important concerns remain about the reproducibility of these findings in human ALS. The preferential involvement of spinal compared to cortical MNs, the anatomical differences in CNS organization between mice and humans and the confounding effects of artificial manipulation, such as transgenic overexpression, limit the translation of ALS preclinical research on patients (Kato, 2008; Philips and Rothstein, 2015). Nonetheless, some studies reported evidence of early NMJ alterations and axonal injury preceding MN loss also in ALS patients (Bradley et al., 1983; Dengler et al., 1990; Fischer et al., 2004; Bruneteau et al., 2015). Moreover, progressive NMJ and motor axon loss is a consistent finding also in other transgenic ALS mice with mutations in TARDBP, FUS, and C9orf72 repeat expansion (Picher-Martel et al., 2016). Overexpression of hTDP-43A315T and hTDP-43WT leads to NMJ denervation and loss of corticospinal axons, which in some cases predominate over MN al loss (Wegorzewska et al., 2009; Arnold et al., 2013; Herdewyn et al., 2014). In the hFUSP525L mouse lines, where the mutation is

conditionally expressed in MNs, the progressive degeneration is preceded by early pre-symptomatic retraction of motor axons (Sharma et al., 2016).

Reflecting evidence of sensory dysfunction in human ALS, transgenic hSOD-1G93A mice also display neurodegeneration in sensory axon, DRG and proprioceptive sensory fibers of muscle spindles. Signs of axonal damage are detected since the pre-symptomatic stage, progressing with a distal-to-proximal gradient (Guo et al., 2009; Vaughan et al., 2015). Furthermore, loss of dorsal root axons has been detected in rodent models overexpressing mutant TDP-43 and FUS (Huang et al., 2011; Arnold et al., 2013), suggesting that ALS-specific proteins may also affect sensory neurons. However, it is still unclear whether the damage comes from nearby disease spreading or arise independently from MNs. Interestingly, transfection of mutant TDP-43 and SOD1 in cultured sensory neurons exert a negative impact on neurites length and arborization after prolonged culture, suggesting a direct effect (Vaughan et al., 2018).

Therefore, the reported evidence demonstrates that all components of the PNS are affected in ALS, although the differing kinetics of damage and progression still point to differences in vulnerability between the sensory and motor axons and neurons.

# Unraveling Mechanisms of Vulnerability and Resistance of Peripheral Motor and Sensory Neurons in ALS

Studies addressing the mechanisms of PNS involvement in ALS may provide a unique opportunity for unraveling the determinants of the different patterns of vulnerability of motor and sensory neurons. This would allow the identification of specific protective or deleterious factors amenable of therapeutic intervention. Hereafter, we discuss key factors at the PNS level potentially implied in such differences.

#### Intrinsic Axonal Vulnerability

Hyperexcitability of the axolemma, due to altered Na<sup>+</sup> and K <sup>+</sup> conductance properties, has been demonstrated in motor but not sensory axons of ALS patients (Shibuya et al., 2011; Park et al., 2017; Matamala et al., 2018). The consequent increase in Ca2<sup>+</sup> influx may be selectively harmful in the motor compartment, since motor axons show a low Ca2+ buffering capacity (Jaiswal, 2013; Leal and Gomes, 2015). In turn, Ca2<sup>+</sup> overload leads to the activation of effector proteins such as calpain, a calcium-dependent protease, involved in TDP-43 fragmentation, which predisposes to aggregation, Wallerian degeneration and NMJ disassembly (Yamashita et al., 2012; Conforti et al., 2014; Campanari et al., 2016). Specific demise of MNs may also be driven by the involvement of ALS-key proteins, such as TDP-43 and FUS. They are mRNA-binding proteins, mainly localized in the nucleus, which participate in the metabolism of broad pools of mRNAs, by affecting splicing, stability, transport. Mutations in these proteins affect neurons by both loss and gain of function, leading to altered RNA processing, increasing cytoplasmic translocation and propensity to aggregation (Ratti and Buratti, 2016). They are essential during embryonic development, but only MNs in the cortex and anterior horns display sustained expression of both proteins during lifetime (Huang et al., 2010). Although most of the transcriptome is shared between motor and DRG axons, more than one third is unique (Rotem et al., 2017). It has been estimated that TDP43 binds roughly 30% of the whole transcriptome (Ling et al., 2013), including coding and non-coding RNAs, such as miRNA (Kawahara and Mieda-Sato, 2012). TDP-43 is actively recruited in RNA granules along motor axons for active transport (Fallini et al., 2012). Examples of TDP43-bound mRNAs include human neurofilaments, components of the endosomal trafficking and mitochondrial proteins (Stalekar et al., 2015; Schwenk et al., 2016; Wang et al., 2016). Over-expression of wild type and mutant TDP-43 alters the dynamics of its axonal transport (Alami et al., 2014), leading to mitochondrial toxicity and NMJ loss (Lin et al., 2011; Tripathi et al., 2014; Wang et al., 2016). Furthermore, a transient increase in TDP-43 localization along axonal routes is observed following axotomy, suggesting a role in peripheral nerve regeneration (Moisse et al., 2009). However, TDP-43G348C mice show abnormal persistence of cytoplasmic levels after peripheral injury, impairing axonal regeneration (Swarup et al., 2012). Such finding may provide a potential additional explanation to the paucity of regenerative clusters observed in ventral roots and motor nerves of human and animal ALS (Riva et al., 2011b, 2014; Luigetti et al., 2012). In contrast, sensory nerves still show some level of regeneration capacity in ALS patients, suggesting that efficient repair mechanisms may dampen progressive neurodegeneration (Hammad et al., 2007; Isaacs et al., 2007).

#### Neuromuscular Junction

The early susceptibility of the NMJ in ALS relies on peculiar features that distinguish it from other peripheral ending structures, including the sensory receptors. It has high energy demands and requires efficient mechanisms of plasticity and repair to cope with the challenges faced by muscle contraction. Oxidative stress, which contributes to ALS pathogenesis (D'Amico et al., 2013), has a great influence on NMJ function, as the early production of reactive oxygen species (ROS) in distal muscles inhibits transmitter release (Kraft et al., 2007; Naumenko et al., 2011) and leads to synaptic loss (Fischer et al., 2011, Fischer et al., 2012). High levels of oxidative stress have also been observed in muscles and spinal cord of ALS patients (D'Amico et al., 2013; Lanfranconi et al., 2017). Mitochondrial dysfunction further participates to NMJ dismantling, as its role in energy production and Ca2<sup>+</sup> buffering is essential for proper function and maintenance of the synapse (Fischer-Hayes et al., 2013). Non-neuronal factors are also involved, as terminal schwann cells (TSCs) exert important functions in NMJ stability and repair, adapting their state according to local synaptic environment. Such decoding ability depends on the muscarinic acetylcholine receptor (mAchR), a G-protein-coupled receptor (GPCR) present on the membrane of TSCs, which detects the level of Ach released in the synaptic cleft. SOD1 mutants have been shown to alter the mAchR sensitivity of these cells to local damage, leading to deficient repair of denervated NMJ in ALS mice models (Arbour et al., 2015). Still, such findings need confirmation in humans and other models of ALS, although a

relationship between TDP-43 function and TSC-related receptors has been suggested (Arbour et al., 2017).

#### PNS-Related Non-cell Autonomous Toxicity

Besides TSCs, other non-cell autonomous mechanisms in the PNS may further influence MNs vulnerability. The role of Schwann cells in the peripheral nerve has also been recently re-evaluated in ALS. Early deficits in axonal transport may be initially compensated by the transfer of polyribosomes from Schwann cells to the axonal compartment, boosting local protein synthesis as an adaptive response to mSOD1-induced injury (Verheijen et al., 2014). Moreover, Schwann cells may exert partial protection against the increased production of ROS related to axonal degeneration, as the loss of dismutase activity in these cells accelerated disease progression (Lobsiger et al., 2009). However, pathogenic gain-of-functions SOD1 mutants may affect Schwann cells functioning and protective effects as well (Wang et al., 2012). Neuroinflammatory response within the PNS and alterations of the blood-nerve barrier might offer additional insights in the pathogenesis of ALS (Kang and Rivest, 2007; Beers et al., 2008), as suggested by observations raised in mice models (Pupillo et al., 2014; Nardo et al., 2016, 2018) and in human ALS (Devigili et al., 2011; Riva et al., 2011a; Gerevini et al., 2016). Pathological studies of sural nerve biopsies of ALS patients have demonstrated, in a subset of cases, the presence of vasculitic-like inflammatory infiltrates associated with normal nerve morphology, suggesting a potential neuroprotective effect of neuroinflammation within the PNS (Devigili et al., 2011). Indeed, the recruitment of macrophages and stimulation of their phagocytic function in peripheral nerves is important to create a favorable milieu for axonal regeneration, as shown in mSOD1 ALS mice (Nardo et al., 2016).

# Common Grounds for ALS and Hereditary Neuropathies

Despite the distinctive prognosis and disease course, ALS, related MNDs and hereditary neuropathies show a non-negligible degree of overlap in terms of both clinical presentations and genetics. Therefore, a better understanding of the key cellular pathways involved in both diseases may allow a broader comprehension of the molecular events leading to axonal and MN degeneration in ALS.

#### Neuronal Cytoskeleton

Microtubules, NFs and actin are the main components of axonal cytoskeleton, and impairment in either structure or functioning of any of these components may lead to axonal atrophy, retraction and degeneration or transport defects (Gentil et al., 2015; Kevenaar and Hoogenraad, 2015). Elevated levels of light NFs (NFL) and phosphorylated heavy NFs have been observed in the cerebrospinal fluid and serum of ALS, together with lower mRNA NFL levels in MNs (Volkening et al., 2009; Gaiani et al., 2017; Feneberg et al., 2018). Accumulation of NF components in axons is a prominent feature of both human and animal ALS (Leigh and Swash, 1991; Rouleau et al., 1996; Morrison et al., 2000), and has been observed along both motor and sensory axons in the hSOD-1G93A ALS mouse model (King et al., 2012; Gentil et al., 2015; Sassone et al., 2016). Alterations of ALS-key genes, such as SOD1 and TARDBP, have been shown to affect the stability of NFL mRNA, suggesting that altered stoichiometry between the different NF components may contribute to the disease (Volkening et al., 2009). Mutations in NEFL and NEFH, leading to axonal CMT, promote protein aggregation, aberrant mitochondrial morphology and shortening of axonal length (Yoshihara et al., 2002; Rebelo et al., 2016; Nam et al., 2017). Tubulin Alpha 4a (TUBA4A), encoding for α-tubulin, has been recently identified as a novel ALS gene (Smith et al., 2014). It has been shown that TUBA4A mutations are able to alter microtubule dynamics, with depolymerization and degradation of α-tubulin (Helferich et al., 2018). Although this gene has been described only in cohorts of ALS and FTD patients, a sensory and motor neuronopathy is observed in progressive motor neuronopathy mice mutated in tubulin binding cofactor E (TBCE), a known interactor of α-tubulin (Schafer et al., 2017). Profilin 1 (PFN1) has been found as a rare cause of familial ALS (Wu et al., 2012). It is a crucial protein for the conversion of monomeric (G)-actin to filamentous (F)-actin. Interestingly, although pathogenic variants do not disrupt actin dynamics (Freischmidt et al., 2015), alterations in microtubule growth and increased propensity to cytoplasmic TDP43 aggregation have been described (Matsukawa et al., 2016; Tanaka et al., 2016; Henty-Ridilla et al., 2017), revealing potential cell-type specific mechanisms of neurodegeneration.

#### Axonal Transport

Defects in axonal transport have been clearly observed both in ALS and dHMN. SOD1 and TDP43 mutants impair anterograde and retrograde axonal transport, leading to loss of essential synaptic components, mitochondrial abnormalities, and neurite shortening (Williamson and Cleveland, 1999; Perlson et al., 2009; Alami et al., 2014). Similarly abnormal transport of organelles, such as mitochondria and the endoplasmic reticulum (ER), is found in CMT models due to NEFL mutations (Tradewell et al., 2009). The failure of axon-cell body communication may dampen the activation of repair mechanisms, thus conferring an increase vulnerability of the peripheral compartment, more pronounced for the distal segments (Fischer et al., 2004; Dadon-Nachum et al., 2011). Strategies aiming at reinforcing axonal transport may thus confer protection to a broad range of pathogenic alterations, preserving the axonal routes and its endings.

#### Protein Folding

The highly differentiated and post-mitotic state of MNs exerts a relevant pressure over the protein quality control system of the cell, which is devoted to the correct folding, function and turnover of the whole proteome. HSPB1/27, HSPB8/22, and sigma non-opioid intracellular receptor 1 (SIGMAR1) are three chaperones with a significant expression levels in MNs (Mavlyutov et al., 2010). Mutations of the small heat shock proteins have been mostly associated with a dHMN or axonal CMT2 phenotype, while SIGMAR1 gene variants have been reported also in rare ALS patients (Al-Saif et al., 2011; Rossor et al., 2012). HSPB1 mutants alter both its chaperone activity in NF organization and preservation of axonal stability

(Evgrafov et al., 2004; Ackerley et al., 2006). HSPB8 mutant mice develop a progressive motor neuropathy through specific neurite degeneration (Irobi et al., 2010; Bouhy et al., 2018). In addition, alterations in TDP-43 expression, and associated splicing is observed in the muscles of these mice (Cortese et al., 2018). Conversely, there is a significantly higher HSPB8 expression in surviving neurons in the hSOD-1G93A mouse model of ALS (Crippa et al., 2010). Finally, SIGMAR1, together with MFN2 and vesicle-associated membrane protein-associated protein B/C (VAPB), are able to modulate unfolded protein response (UPR) sensors, including protein kinase RNA-like endoplasmic reticulum kinase (PERK) (Nguyen et al., 2015).

#### Mitochondrial/Associated Membranes (MAMs)

The ER and mitochondria have evolved complex sites of interactions, defined as mitochondrial/associated membranes (MAMs), which are important for essential cellular functions, such as calcium homeostasis, autophagy, regulation of mitochondrial dynamics, and axon survival (Giorgi et al., 2015). Both ALS and hereditary neuropathies display dysfunction in one or more of MAM components. A huge number of cellular proteins participate in MAMs, including MFN2 and VAPB, which act as tethering sites between the two organelles (Bernard-Marissal et al., 2015). VAPB mutants affect the interaction with the mitochondrial protein tyrosine phosphatase-interacting protein 51 (PTPIP5) (de Brito and Scorrano, 2008; De Vos et al., 2012). Interestingly, TDP43 and FUS mutants have been shown to decrease ER-mitochondria contacts by disrupting VAPB-PTPIP5 binding (Stoica et al., 2014). MFN2 mutants alter other MAM functions, such as mitochondrial transport, fusion, and autophagosome assembly (Kim et al., 2015). Loss-of function mutations in SIGMAR1 display decreased ER-mitochondria association, resulting in MN death, and axonal degeneration due to impaired retrograde transport (Bernard-Marissal et al., 2015; Watanabe et al., 2016).

#### THERAPEUTIC PERSPECTIVES

Due to the high similarity between these neuromuscular disorders and the intimate relationship observed in the molecular pathways involved, it may not be surprising if functional proteins involved in either of the two diseases may provide benefits on their respective counterpart.

Preservation of NMJ is an attractive target to delay muscle denervation and disability in ALS. A drug-screening platform in C. elegans and zebrafish models of ALS identified pimozide, an already approved neuroleptic, as a NMJ stabilizer by blocking T-type Ca2<sup>+</sup> channels (Patten et al., 2017). Despite some benefits in neurophysiologic measures were reported in hSOD-1G93A mice and human patients in the first report, one study assessing long-term effects showed warned caution, as worsening of survival and muscle function were observed in mutant SOD1, and TDP-43 mice models (Pozzi et al., 2018). Notably, an agonist antibody directed toward musclespecific kinase (MuSK), a post-synaptic tyrosine kinase receptor essential for NMJ maintenance, preserved motor synapses, delayed muscle denervation and extended lifespan in hSOD-1 G93A ALS mice (Cantor et al., 2018). The promising findings of this study still awaits confirmation and replication in other disease models.

Reinforcing the action of factors involved in axonal stability and transport may restore soma-axon communication, delaying NMJ loss and axonal degeneration. Pharmacological inhibition of histone deacetylase 6 (HDAC6), a known client protein of HSPB1, was able to increase the stability of microtubules, rescuing axonal loss, and improving outcome in a CMT2F mouse model with mutant HSPB1 (d'Ydewalle et al., 2011). Interestingly, HDAC6 inhibition was also able to restore axonal transport defects in patient-derived MNs mutated in FUS (Guo et al., 2017). Pharmacologic agents targeting microtubule dynamics, such as noscapine and vinblastine, delayed disease onset and prolonged survival in hSOD-1G93A mice, by reducing microtubule turnover (Fanara et al., 2007).

Inhibition of p38/mitogen-associated protein kinase (MAPK) signaling in MNs was able to restore retrograde axonal transport defects in vivo (Gibbs et al., 2018), together with reduced microglia-induced neuroinflammation in the spinal cord (Kim and Choi, 2010; Zhan et al., 2015; Leal-Lasarte et al., 2017). However, it is remarkable that no significant improvement was observed on the clinical phenotype (Gibbs et al., 2018), suggesting that inhibition of this pathway alone is not sufficient to fully rescue MN function.

Reducing ER stress through either a potentiation of the chaperone system or inhibition of stress-related factors may limit the production rate of misfolded proteins. Enhancing HSPB8 activity has been shown to be protective against accumulation of TDP43 aggregates in MNs (Crippa et al., 2016; Rusmini et al., 2017), and to extend survival of hSOD-1G93A mice (Aurelian et al., 2012). Trehalose, a chemical chaperone known to induce HSPB8, reduces ER stress levels and improve autophagy, delaying disease, and prolonging MN survival in a mouse model (Zhang et al., 2014; Li Y. et al., 2015). Upon UPR activation, cells switch to a state of translational repression and stimulation of chaperones' synthesis, processes regulated by phosphorylation of the eukaryotic translation initiation factor 2A (eiF2α). Salubrinal and guanabenz, two eiF2α phosphatase inhibitors, protected against MN degeneration in ALS mice (Saxena et al., 2009; Wang et al., 2014), although one study reported that guanabenz accelerated disease progression in an ALS model (Vieira et al., 2015).

Considering their importance in cell physiology, MAMs could represent an important target for therapy in ALS and may prove beneficial for a wide array of downstream signaling cascades relying upon these structures. An agonist of SIGMAR1 was shown to improve muscle activity and motor performance in pre-symptomatic hSOD-1G93A ALS mice (Mancuso et al., 2012). Notably, overexpression of MFN2 prevented NMJ loss and delayed onset and progression in a murine model by raising calpastatin levels, a calpain inhibitor, essential for survival, and function of the NMJ (Wang et al., 2018).

Boosting axonal regeneration through either neuron- or Schwann-targeting factors may also be a valuable alternative strategy in order to preserve motor function (Gilley et al., 2017).

Studies modulating Neuregulin 1 activity, a neurotrophic factor involved in peripheral nerve development and regeneration, in murine hSOD-1G93A ALS lead to increased re-innervation and MN survival (Mancuso et al., 2016; Modol-Caballero et al., 2017). However, other approaches aimed at slowing Wallerian degeneration, such as the mutations in the Slowed Wallerian degeneration (Wld<sup>S</sup> ) and Sterile Alpha and TIR Motif Containing 1 (Sarm1) proteins, gave only modest or no results in terms of progression and survival (Fischer et al., 2005; Peters et al., 2018), suggesting that alternative mechanisms may underlie the axonal loss observed in ALS.

#### CONCLUSION

Clinical and pathological studies in both human patients and rodent models suggest that the PNS is involved in ALS pathogenetic cascade. A better comprehension of the molecular events occurring into the PNS may prove essential

#### REFERENCES


for a better understanding of MND pathomechanisms and may pave the way to the discovery of novel therapeutic targets. Considering that in ALS neurodegeneration derives from the combination of several dysfunctional pathways, it might be possible that the development of combination therapies acting on multiple cascades located in both the central and the peripheral compartments may show synergistic neuroprotective effect in order to prevent axonal degeneration, MN cell death, und ultimately in prolonging patients' survival.

#### AUTHOR CONTRIBUTIONS

FG and SS contributed to the collection of available evidence and writing. YF, AQ, NR, CL, and LT contributed to the manuscript writing and review of intellectual content of the manuscript, and approved the final version of the manuscript. AQ and NR contributed to the manuscript design.



paraplegia type 10 and in axonal Charcot-Marie-Tooth type 2. Clin. Genet. 82, 157–164. doi: 10.1111/j.1399-0004.2011.01717.x



lateral sclerosis. Neurobiol. Dis. 28, 154–164. doi: 10.1016/j.nbd.2007. 07.003



approaches in CMT neuropathies and related disorders 23-25 April 1999, Soestduinen, The Netherlands. Neuromuscul. Disord. 10, 69–74. doi: 10.1016/ s0960-8966(99)00095-4




directly affect and sensitize sensory neurons to stress. Sci. Rep. 8:16582. doi: 10.1038/s41598-018-34510-8


**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 © 2019 Gentile, Scarlino, Falzone, Lunetta, Tremolizzo, Quattrini and Riva. 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.

# Cognitive, Emotional and Psychological Manifestations in Amyotrophic Lateral Sclerosis at Baseline and Overtime: A Review

Soumia Benbrika, Béatrice Desgranges\*, Francis Eustache and Fausto Viader

Neuropsychology and Imaging of Human Memory, Normandy University-PSL Research University-EPHE-INSERM U1077, Caen University Hospital, Caen, France

It is now well recognized that, in addition to motor impairment, amyotrophic lateral sclerosis (ALS) may cause extra-motor clinical signs and symptoms. These can include the alteration of certain cognitive functions, impaired social cognition, and changes in the perception and processing of emotions. Where these extra-motor manifestations occur in ALS, they usually do so from disease onset. In about 10% of cases, the cognitive and behavioral changes meet the diagnostic criteria for frontotemporal dementia. The timecourse of behavioral and cognitive involvement in ALS is unclear. Whereas longitudinal studies have failed to show cognitive decline over time, some crosssectional studies have demonstrated poorer cognitive performances in the advanced stages of the disease. Neuroimaging studies show that in ALS, extra-motor signs and symptoms are associated with specific brain lesions, but little is known about how they change over time. Finally, patients with ALS appear less depressed than might be expected, given the prognosis. Moreover, many patients achieve satisfactory psychosocial adjustment throughout the course of the disease, regardless of their degree of motor disability. There are scant longitudinal data on extra-motor impairment in ALS, and to our knowledge, no systematic review on this subject has yet been published. Even so, a better understanding of patients' clinical trajectory is essential if they are to be provided with tailored care and given the best possible support. We therefore undertook to review the evidence for extra-motor changes and their time course in ALS, in both the cognitive, emotional and psychological domains, with a view to identifying mechanisms that may help these patients cope with their disease.

Keywords: amyotrophic lateral sclerosis, extra-motor manifestations, cognition, emotion, psychological adjustment, coping

# INTRODUCTION

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease of upper and lower motor neurons. In all sporadic and most genetic cases, proteinaceous aggregates of TAR-DNA binding protein 43 (TDP43) are found in these upper and lower motor neurons, as well as in certain glial cells (Saberi et al., 2015). The disease worsens relentlessly, and death occurs after a median

#### Edited by:

Francesca Trojsi, University of Campania Luigi Vanvitelli, Italy

#### Reviewed by:

Judith Machts, Universitätsklinikum Magdeburg, Germany Johannes Prudlo, Rostock University Hospital, Germany

> \*Correspondence: Béatrice Desgranges beatrice.desgranges@unicaen.fr

#### Specialty section:

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

Received: 28 March 2019 Accepted: 22 August 2019 Published: 10 September 2019

#### Citation:

Benbrika S, Desgranges B, Eustache F and Viader F (2019) Cognitive, Emotional and Psychological Manifestations in Amyotrophic Lateral Sclerosis at Baseline and Overtime: A Review. Front. Neurosci. 13:951. doi: 10.3389/fnins.2019.00951

**107**

duration of 3 years after onset and 2 years after diagnosis (Couratier et al., 2016).

All patients experience extensive and progressive muscle paralysis that results in severe functional disability, but as many as 50% may also have extra-motor signs and symptoms, including cognitive impairment (Abrahams et al., 2000; Murphy J. M. et al., 2007; Murphy J. et al., 2007) mainly affecting executive functions, emotion processing and social cognition, and behavior.

Despite being given such a gloomy prognosis, not all patients display symptoms of depression. Indeed, depression rates remain low, considering the severity of the disease (Rabkin et al., 2016), and patients' quality of life (QoL) is not as impaired as might be expected (Roach et al., 2009; Lulé et al., 2012; Jakobsson Larsson et al., 2017). This relatively good psychosocial adjustment raises important questions about the coping strategies used by patients.

While we now have substantial knowledge about extramotor signs of ALS at baseline, data on their clinical course remain scarce. The results of cross-sectional studies attempting to correlate disease duration and/or physical symptom severity with cognitive/psychological status suggest that, unlike motor impairment, cognitive performances are not always significantly worse in the late stages of the disease than they were in the early stages (Ringholz et al., 2005). These observations suggest that the mechanisms of psychological involvement do not simply come down to degenerative changes in the neurons of the brain. Furthermore, the fact that some studies have described patients as retaining a relatively satisfactory level of wellbeing, despite the continuous worsening of their physical condition, may seem a little surprising and deserves further consideration.

We therefore set out to review the cognitive, emotional and psychological impairments that can emerge at the beginning of the disease, and then to address the question of how they change throughout the course of the disease. We focused on ALS cases that do not meet the clinical criteria for dementia, as ALS-dementia syndromes raise different issues, and may not be as similar to ALS without dementia as is usually claimed (Lulé et al., 2018). Longitudinal studies are the most appropriate way of assessing how these impairments change, but they have several drawbacks. First, the follow-up periods are often short, and a large proportion of patients are liable to drop out, owing to the rapid progression of ALS, making it difficult to detect significant changes in extra-motor signs and symptoms over time. Second, it is difficult to find a suitable control group that accurately matches the patient sample. Third and last, the tools used to assess cognitive functioning have to be carefully chosen in order to be both adequate and usable at all stages of the disease-even the most advanced-, in order to accommodate the motor impairments that invariably interfere with testing procedures.

To fulfill the objectives of our work, we conducted a bibliographical search on PUBMED prior to the 1st October 2018 with the following keywords: Cognition, emotion, psychological adjustment, neuroimaging, longitudinal, AND ALS. We retained the articles that met our keywords, and that were written in English or in French. 190 articles were finally included in our study.

# COGNITIVE, EMOTIONAL, AND BEHAVIORAL CHANGES IN ALS

#### Cognition in ALS (Table 1) Risk Factors for Developing Cognitive Deficits

Female sex, older age at onset, and low education level seem to increase the risk of cognitive impairment (Irwin et al., 2007; Montuschi et al., 2015; Flaherty et al., 2017). The presence of a C9orf72 gene mutation is associated with more severe cognitive deficits, even in non-demented patients (Byrne et al., 2012; Montuschi et al., 2015). There are still conflicting data about the potential positive correlation between bulbar symptoms and cognitive decline, with some studies supporting this correlation (Giordana et al., 2011; Gordon et al., 2011; Strutt et al., 2012) while others don't (Sterling et al., 2010; Zalonis et al., 2012).

The presence of depressive symptoms also seems to exacerbate the executive deficit in patients, with a negative correlation between scores on depression scales and those on cognitive tests (Wei et al., 2016; Carelli et al., 2018). These findings are not surprising, given the negative impact that depression is known to have on cognitive functions, in terms of attention and memory (Bortolato et al., 2016).

#### Baseline

#### **Frequency and profile**

Cognitive impairment occurs in 30–50% of patients with ALS, depending on the study and the neuropsychological tools used to assess cognitive functions (Giordana et al., 2011; Zago et al., 2011; Byrne et al., 2012; Kasper et al., 2015; Beeldman et al., 2016; Murphy et al., 2016). The cognitive deficit profile includes impairment of executive functions, verbal fluency, language, social cognition and verbal memory (Phukan et al., 2012; Beeldman et al., 2016). In 6–14% of patients, the cognitive impairment meets the criteria for a behavioral variant of frontotemporal dementia (FTD) (Consonni et al., 2013; Montuschi et al., 2015; Murphy et al., 2016). In the study by Ringholz et al. (2005), 51% of patients were cognitively impaired, compared with 5% of controls, and 14% met the diagnostic criteria for FTD. A cluster analysis indicated four patient subgroups: 49% with intact cognition, 32% with mild cognitive impairment, 13% with moderate impairment, and 6% with severe impairment.

Executive alterations affect verbal fluency, attention monitoring, switching, working memory, cognitive flexibility and mental control, and reasoning and coordinating rules (Phukan et al., 2012; Beeldman et al., 2016). Other executive functions are also impaired, such as initiation and shifting (Kasper et al., 2015). A study by Palmieri et al. (2015) found that female patients with ALS were twice as likely as males to have dysexecutive dysfunction-an intriguing finding that is not clearly explained to date.

Cognitive alterations observed in ALS also include language impairments (Iwasaki et al., 1990; Raaphorst et al., 2010; Giordana et al., 2011). According to Woolley and Rush (2017), language may be altered in 30–40% of patients without dementia, regardless of executive dysfunction, dysarthria or respiratory

#### TABLE 1 | Cognition in ALS.

fnins-13-00951 September 6, 2019 Time: 18:1 # 3


(Continued)

#### TABLE 1 | Continued

fnins-13-00951 September 6, 2019 Time: 18:1 # 4


ALS, Amyotrophic Lateral Sclerosis; ACE, Addenbrooke's Cognitive Examination; ALS-BCA, ALS Brief Cognitive Assessment; ALS-CBS, ALS Cognitive Behavioral Screen; ALS-CFB, ALS-Computerized Frontal Battery; ALS-FTD, ALS-Frontotemporal Dementia; ALS-FTD-Q, ALS-Frontotemporal Dementia Questionnaire; BDI, Beck's Depression Inventory; Bi, Behavioral Impairment; BL, Baseline; Ci, Cognitive Impairment; CG, Caregiver; Cibi, Cognitive and Behavioral Impairment; CS, Cross-Sectional Study; ECAS, Edinburgh Cognitive and Behavioral ALS Screen; FAB, Frontal Assessment Battery; FTD, Frontotemporal Dementia; LS, Longitudinal Study; MADRS, Montgomery-Asberg Depression Rating Scale; MiND-B, Motor Neuron Disease Behavioral instrument; MMSE, Mini Mental State Examination; MRI, Magnetic Resonance Imaging; NC, Normal Controls; NMC, Controls With Neuromuscular Disease, NNMC, Non Neurological Medical Controls; NPI, Neuropsychiatric Inventory; PHQ, Patient Health Questionnaire; PNFA, Progressive Nonfluent Aphasia; Pt, Patient; SD, Semantic Dementia; STAI, State-Trait Anxiety Inventory; SNST, Stroop Neuropsychological Screening Test; TMT, Trail Making Test; VBM, Voxel-Based Morphometry; WAIS, Similarities Subtest of The Wechsler Adult Intelligence Scale.

failure. Linguistic impairments may include deficits in syntactic processing, verb naming and action verb processing, semantic and verbal paraphasias, and syntactic comprehension deficits. Kamminga et al. (2016) found that syntactic comprehension was defective in 25% of patients with ALS without dementia. Leslie et al. (2015) observed semantic deficits in 35% of not demented ALS patients, and Hervieu-Bègue et al. (2016) in as many as 60% of such patients. Woolley and Rush (2017) suggested that language alterations observed in ALS reflect the fact that ALS and nonfluent/agrammatic primary progressive aphasia lie on a pathogenesis continuum.

Memory may also be altered to some extent in ALS (Dary-Auriol et al., 1997; Abrahams et al., 2000; Neary et al., 2000; Ringholz et al., 2005), but the nature of this impairment is subject to debate. It has been suggested that defective memory is mainly the consequence of executive dysfunction, especially since memory deficits very rarely occur in isolation in ALS (Strong, 2017). There is, however, some evidence in favor of hippocampal involvement. Abdulla et al. (2014) found reduced hippocampal volume in a series of patients with ALS relative to controls, and Machts et al. (2018) showed that this reduction mostly affects the anterior part, including the CA1 field-a critical structure for episodic memory. Furthermore, Raaphorst et al. (2015) found that immediate and delayed story recall scores were below normal in 23% of patients with ALS, and these performances were correlated with hippocampal gray-matter volume. Further studies are needed to try to determine the real status of memory in ALS.

The Strong (2017) revised diagnostic criteria of ALS-FTSD classify ALS into several categories according to the type and severity of neuropsychological impairments: pure ALS (no impairment), ALSci (cognitive impairment), ALSbi (behavioral impairment), ALScibi (cognitive and behavioral impairment), all three without dementia, and ALS-FTD. According to those criteria, the diagnosis of ALS-ci requires either executive dysfunction or language impairment or both, and the diagnosis of ALS-bi requires either apathy or two of the Rascovsky criteria for FTD.

#### Over Time

Most longitudinal studies of cognitive changes in patients with ALS have a follow-up period of 6 months. A number of authors have reported an absence of decline in cognitive functions during the course of the disease (Kilani et al., 2004; Schreiber et al., 2005; Braber, 2016; Kasper et al., 2016; Burkhardt et al., 2017; Poletti et al., 2018; Woolley et al., 2018). Although patients may initially have lower performances than controls, their deficits remain stable over time. It has been argued, however, that unlike normal controls, patients with ALS do not display a practice effect in repeated assessments with the Edinburgh Cognitive and Behavioural ALS Screen (ECAS) battery, which has been interpreted as evidence of a pre-symptomatic cognitive decline (Burkhardt et al., 2017). Abrahams et al. (2005) found that while most cognitive scores remained stable over time (including written and spoken verbal fluency) patients with ALS performed a single word retrieval test increasingly slowly whereas controls performed it faster. Furthermore, the caregivers of patients report increasing cognitive dysfunction in daily life over time, unlike controls' partners. Robinson et al. (2006) found that cognitive changes occurred in 36% of patients with ALS over a 6-month period. Cognitive status seems to have a heterogeneous outcome in ALS. Behavioral symptoms may appear in patients with stable cognitive performances (Bock et al., 2017). The presence of even mild cognitive or behavioral impairment (as defined by the 2017 Strong criteria) at baseline seems to be a significant risk factor for the later appearance of a full-blown frontotemporal syndrome (Elamin et al., 2013). The choice of the neuropsychological tests used to assess cognitive status is of primary importance, as Roberts-South et al. (2012) suggested in their longitudinal study of discourse changes in a group of patients with ALS compared with healthy participants over 24 months. Subtle cognitive language deficits affecting discourse (content rather than productivity) were found to emerge early in ALS and worsen as the disease progressed. The authors concluded that language deficits are more thoroughly detected by the discourse analysis method than by standard language tasks. Gillingham et al. (2017) assessed patients with ALS at baseline and 9 months later using both standardized non-specific cognitive tests and the ALS-Computerized Frontal Battery (ALS-CFB), which was specifically designed for patients with ALS. While the basic cognitive tests failed to reveal any change over time, the ALS-CFB showed a significant decrease in cognitive performances in patients compared with controls (e.g., for verbal fluency). Finally, patients with bulbar onset seem to exhibit a progressive decline in cognitive functions (Strong et al., 1999).

It is therefore difficult to give a simple answer to the question "do cognitive functions decline over time in ALS?" owing to the heterogeneity of the patients and the difficulty of differentiating genuine cognitive deficits from the consequences of the steadily worsening motor impairment. A tentative, preliminary answer could be that when ALS is associated with a cognitive impairment at baseline, this impairment is likely to progress, and when dementia is present at diagnosis, decline is faster (Woolley and Katz, 2008). Normal cognition at baseline was associated with tendency to remain cognitively intact overtime (Elamin et al., 2013).

#### Consequences for Patients and Caregivers

Cognitive impairment in ALS is associated with a more rapid progression of the disease and a poorer prognosis, with reduced survival (Elamin et al., 2011; Giordana et al., 2011; Gordon et al., 2011; Hu et al., 2013; Xu et al., 2017). Stojkovic et al. (2016) found that the death risk was increased threefold by the presence of executive dysfunction in ALS. Poorer survival could be explained by patients' difficulty weighing up the benefits of non-invasive ventilation (Govaarts et al., 2016), or their reduced compliance in the use of medical devices (Olney et al., 2005). The correlation between the severity of bulbar symptoms and the cognitive deficit may also influence survival figures (Olney et al., 2005). Elamin et al. (2013) evaluated the clinical impact of cognitive deficit in patients with ALS in a longitudinal study. Executive impairment at the initial consultation was associated with significantly higher rates of attrition due to disability or death, and faster rates of motor functional decline, particularly

bulbar function. QoL in patients has been found to be worse in the case of cognitive impairment (Hu et al., 2013). Caregivers are the pillars of patient care. They may be the spouse, children, brothers or sisters. It is cognitive and behavioral impairment, rather than the patients' physical disability, that increases caregivers' burden and anxiety (Rabkin et al., 2009; Burke et al., 2015). This underscores the importance of screening patients with ALS for cognitive dysfunction, in order to predict disease progression and provide more adequate care.

#### Neuroimaging Correlates of Cognitive Deficits

Anatomical brain changes are more pronounced in patients with ALS who exhibit cognitive impairment than in those who do not. Gray matter volume in the frontal and temporal lobes is reduced when cognitive impairment is present (Abe et al., 1997; Giordana et al., 2011; Mioshi et al., 2013), as is that of the cerebellar cortex and basal ganglia (Christidi et al., 2018). Reduced cortical thickness is observed in the bilateral precentral gyrus, insular and cingulate cortices, and frontotemporal regions in the case of cognitive deficit (Schuster et al., 2014a; Agosta et al., 2016). Left peri-insular atrophy was found to correlate with scores on a syntactic comprehension task (Kamminga et al., 2016). There are greater white-matter changes in the corticospinal and corpus callosum tracts when cognitive functions are altered (Sarro et al., 2011), and these extend to extra-motor tracts, particularly within the frontal lobes and associative areas including the cingulum and the inferior longitudinal, inferior fronto-occipital, and uncinate fasciculi (Sarro et al., 2011; Kasper et al., 2014; Christidi et al., 2018). Verbal learning and memory test scores are correlated with white-matter values in the fornix (Sarro et al., 2011). Cerebral regional metabolism has been repeatedly studied with fluorodeoxyglucose positron emission tomography (18FDG-PET) in patients both with and without cognitive dysfunction. One of the most recent studies shows that cognitively impaired (but not demented) patients with ALS have relative hypometabolism in the right cingulate and frontal cortex, and bilaterally in the prefrontal cortex, compared with patients with no cognitive impairment (Canosa et al., 2016). These patients also have relative hypermetabolism in parts of the midbrain and corticospinal tracts (Canosa et al., 2016). In a combined MRI and PET study in patients with ALS, Buhour et al. (2017) found gray-matter atrophy, predominantly in the temporal poles, and hypometabolism in the left superior medial cortex. Hypermetabolism was also found in parts of the temporal lobes and the cerebellum. A series of negative correlations between cognitive performance and regional cerebral metabolism in functionally relevant areas suggest that hypermetabolism is more likely to reflect deleterious processes such as neuroinflammation rather than compensatory neuronal activity.

Longitudinal studies of anatomical cerebral changes in ALS with cognitive modifications remain rare, for the same reasons as those dealing with purely clinical aspects of the disease. Floeter et al. (2016) assessed brain volume changes in patients who were either sporadic or carriers of the genetic mutation C9orf72 (C9+), which is often associated with FTD or cognitive impairment. These authors found that over a 6-month period, ventricular volume increased in C9+ versus sporadic cases, suggesting that subcortical involvement influences cognitive performances in this particular group of patients. Other microstructural changes over time in ALS have been documented. A significant decline in the cortical thickness of frontal, temporal and parietal regions is observed over time, whereas the reduced cortical thickness of the precentral gyrus at the beginning of the disease remains stable (Verstraete et al., 2012; Schuster et al., 2014b). Menke et al. (2018) assessed brain changes in patients with ALS over 2 years and reported widespread changes in both white and gray matter in the cingulate gyrus, thalami, caudate nuclei, pallidum, hippocampi and parahippocampal gyri, and insula. These results indicate that over time, cerebral changes extend into extra-motor areas, but it remains difficult to draw a link between these changes and clinical cognitive implications.

#### Changes in Emotion Perception and Social Cognition (Table 2) Baseline

# **Perception of emotions**

A number of studies have shown that emotion perception is impaired in ALS (Lulé et al., 2005; Zimmerman et al., 2007; Palmieri et al., 2010; Girardi et al., 2011), with patients exhibiting deficits in emotion recognition (facial or prosodic) and emotional valence attribution, and decreased excitability when emotional material is presented. By contrast, Cavallo et al. (2011) and Papps et al. (2005) found no deterioration in either facial emotional recognition or judgments of emotional valence. Clinical features such as type of onset and disease severity may explain the heterogeneity of patients' emotional deficit profiles, as suggested by Sedda (2014). Oh et al. (2016) confirmed the presence of facial emotion recognition deficits in ALS. Bora (2017) carried out a meta-analysis of 15 studies of emotion recognition in ALS and concluded that ALS is associated with significant impairments in facial emotion recognition, especially for disgust and surprise.

Patients' processing of their own emotions (alexithymia) also seems to be altered, although there have been very few studies so far in this area. Roy-Bellina et al. (2008) reported that patients with ALS are more often alexithymic than controls. Benbrika et al. (2018) assessed a group of patients with ALS and a matched control group on the three dimensions of alexithymia: Difficulty Identifying Feelings, Difficulty Describing Feelings, and Externally-Oriented Thinking. Patients were more often alexithymic than controls and had a higher level of alexithymia, especially on the Difficulty Identifying Feelings dimension, suggesting that they have difficulty with the first stage of own emotion processing, namely recognizing one's own emotions.

#### **Social cognition**

Social cognition is a set of cognitive processes used to encode, decode, store, retrieve and use information about people in social relationships. It has several dimensions, such as theory of mind (ToM), empathy, and moral reasoning. ToM refers to the ability to infer the mental and emotional states (i.e., beliefs, preferences and intentions) of oneself and others, and contributes to the understanding of other people's behavior. It can be divided into

#### TABLE 2 | Social cognition and emotion perception in ALS.

fnins-13-00951 September 6, 2019 Time: 18:1 # 7


IGT, Iowa Gambling Task; fMRI, Functional MRI; FDG, Fluorodeoxyglucose; PET, Positron Emission Tomography; ToM, Theory of Mind; FSBS, Frontal Systems Behavior Scale; RME, Reading the Mind in The Eyes; K-MMSE, Korean Version of the Mini Mental State Examination; QoL, Quality of Life; RSN, Resting-State Network; SMA, Supplementary Motor Area. For more please refer to Table 1.

cognitive and affective ToM. Studies of ToM abilities in ALS have sometimes yielded conflicting results, owing to the heterogeneity of the tasks used and the patients' cognitive status, or the presence of depressive symptoms.

Early studies in this domain attested to a deficit in ToM, showing that patients with ALS perform more poorly than healthy individuals on the faux pas task (task assessing cognitive ToM through stories featuring appropriate and inappropriate social behavior), story comprehension task (strip stories presented to participants who have to choose the most appropriate picture) and decision making task (task evaluating decision making abilities under ambiguity in a card game) (Gibbons et al., 2007; Meier et al., 2010; Cavallo et al., 2011; Girardi et al., 2011).

More recently, van der Hulst et al. (2015) investigated whether the ToM deficit described in ALS could be further delineated as one of either affective or cognitive ToM, and explored the relationship between this social cognition deficit and the behavioral manifestations of empathy and self-awareness. Patients were evaluated on a neuropsychological battery that included a cognitive–affective judgment of preference test, a questionnaire to evaluate their self-awareness, and the Neuropsychiatric Inventory–Questionnaire (NPI-Q) to pick up any behavioral changes. The authors found deficits in both cognitive and affective ToM (affective + cognitive ToM deficit in 36% of patients, affective ToM deficit in 12%, and cognitive ToM deficit in 3%). Patients with a ToM deficit were more likely to display behavioral changes such as apathy, lack of empathy, and low self-awareness. When Carluer et al. (2015) compared 23 non-demented patients with controls matched for age, sex and educational level on cognitive ToM, by administering a falsebelief task, they found a cognitive ToM deficit in patients with impaired executive functions. Trojsi et al. (2016) assessed ToM in patients with ALS using a comprehensive battery of specific tasks: the Advanced Task of ToM (ATM), in which participants listen to stories and then have to explain the protagonists' actions; the Emotional Attribution Task (EAT), in which participants have to identify the emotions experienced by the protagonists in a story; and the Reading the Mind in the Eyes task (Eyes test or RME). Patients also underwent a cognitive battery and a QoL assessment. Patients scored lower than controls on the EAT but not on the ATM, and EAT and RME correlated positively with the education, prose memory and mental health items of the QoL questionnaire. Patients were not very cognitively impaired. By contrast, Andrews et al. (2017b) reported impaired complex facial and prosodic emotion recognition, but intact simple facial affect recognition, in a group of non-demented patients with ALS. Finally, Woolley and Rush (2017) showed that the performances of patients without dementia may be impaired on tasks of complex facial affect recognition or affective prosody recognition, and may have difficulty interpreting the gaze direction of others.

To summarize findings on social cognition in ALS, even though results are quite divergent, affective ToM seems to be systematically affected, whereas scores on cognitive ToM seem to be underpinned by patients' cognitive profile. Studies including patients with severe cognitive impairment have reported cognitive ToM deficits (Carluer et al., 2015), whereas those among patients with only mild or moderate cognitive impairment point to preserved cognitive ToM (Trojsi et al., 2016). This prompts the question of a link between social cognition and other cognitive processes.

#### **Interactions between cognitive status, social cognition, and emotion recognition**

Most studies investigating the pattern of interaction between social cognition impairment and other cognitive functions support the idea of an interaction between executive functions and cognitive ToM, with positive correlations between executive functions (verbal fluency, composite executive score, executive score of the Frontal System Behavior Scale, etc.) and ToM (decision making task, faux pas task, etc.) (Gibbons et al., 2007; Meier et al., 2010; Watermeyer et al., 2015). In Carluer et al. (2015) study, cognitive ToM scores were only partially associated with executive performances, notably shifting, inhibition, and the ability to manipulate items in working memory. This led the authors to propose that the ToM deficit observed in some cases of ALS is not simply the consequence of an executive function impairment. Nevertheless, as studies do not generally exclude patients who have only a mild cognitive impairment, it is still unclear whether any ToM deficit is associated with executive deficit in ALS or whether ToM per se is disturbed.

Results on the interaction between emotion recognition and affective ToM performances and cognitive abilities have been somewhat conflicting. In a subset of patients with bulbar ALS, Zimmerman et al. (2007), found that 62% of them had emotion recognition defects, with no correlation with cognitive symptoms. However, the cognitive assessment was done using the Mini-Mental State Examination (MMSE), which is not well suited to patients with ALS. When they calculated a composite executive score, Watermeyer et al. (2015) also found no correlation between cognitive performances and RME test scores. By contrast, Girardi et al. (2011) reported a negative correlation between RME performances and z scores for verbal fluency. Burke et al. (2016) compared the RME test performances of three groups of patients: "no cognitive impairment," "defect in only one cognitive domain," and "multi-executive deficits/cognitively impaired." The patients with no cognitive impairment performed better than those with a defect in only one cognitive domain, who in turn performed better than those in the third group. Finally, Andrews et al. (2017b) found that crossmodal (facial and prosodic integration) emotion recognition correlated with executive functions, whereas the separate modalities of emotion recognition did not. No specific cognitive function seems to be related to either affective ToM or emotion recognition, and as Trojsi et al. (2016) study suggests, the link between ToM performances and cognitive impairment may only concern the cognitive aspect of ToM.

#### Over Time

To our knowledge, only two studies have addressed changes in emotional and ToM abilities and their neural correlates in the course of ALS. Gillingham et al. (2017) assessed cognitive and emotional changes in patients with ALS at baseline and 9 months later using a specific cognitive and emotional battery (ALS-CFB) and other standardized cognitive tests, including cognitive firstorder and cognitive and affective second-order ToM tasks. At baseline, patients showed a deficit in emotion perception for happy emotions, and scored significantly lower than controls on the first-order cognitive ToM task. At 9 months, in the emotion perception task, patients were better at recognizing angry faces than controls were, but there were no changes over time in ToM. Trojsi et al. (2017) assessed cognitive and affective ToM at baseline and 6 months later. At baseline, no ToM abnormalities were found in patients, whereas at follow up, patients with bulbar onset exhibited a decline in both affective and cognitive ToM, whereas those with limb onset only displayed impairment of cognitive ToM over time. This study also included an fMRI investigation, only at baseline, that will be discussed in the next section about neuroimaging.

This result is in line with the more extensive prefrontal hypometabolism observed in patients with bulbar versus limb onset by Cistaro et al. (2012).

#### Neuroimaging Correlates of Emotional and Social Cognition Changes

Emotion recognition impairment is associated with alteration of white-matter integrity along the right inferior longitudinal fasciculus and inferior fronto-occipital fasciculus (Crespi et al., 2014). Defective emotional empathy attribution is correlated with reduced gray-matter density in the anterior cingulate cortex and right inferior frontal gyrus (Cerami et al., 2014). Palmieri et al. (2010) found a general increase in left-hemisphere activation and reduced right-hemisphere activation in patients when they were asked to attribute an emotional valence or remember an emotion. Aho-Özhan et al. (2016) assessed emotion recognition and its functional neural correlates in a group of patients with ALS and a matched control group. Patients recognized disgust and fear less accurately, and had lower activity in the hippocampus on both sides (brain regions involved in negative emotion processing) and more activity in the right inferior frontal gyrus.

Using a composite executive function score as a covariate, Carluer et al. (2015) found positive correlations between cognitive ToM performances and the metabolism of the bilateral superior frontal gyrus, bilateral middle frontal gyrus, and bilateral supplementary motor area. Buhour et al. (2017) sought to understand the metabolic dysfunction and its neurobehavioral consequences better by subjecting a sample of 37 patients with ALS to a comprehensive neuropsychological assessment and PET imaging. Significant negative correlations were found between metabolic activity within the left fusiform gyrus and performance on a false-belief task.

Longitudinal studies are very few and far between. Lulé et al. (2007) assessed emotional valence attribution, arousal, association of movement, and brain functioning when emotional material was presented to patients with ALS and healthy controls at baseline and 6 months later. Patients had an increased brain response in the right supramarginal area and a reduced brain response in extrastriatal visual areas at both measurement points, compared with healthy controls. In the patient group, a reduced brain response in the anterior insula at follow up was correlated with subjective arousal. This reduced response was tentatively interpreted as indicating reduced arousal during the course of the disease at the neural and behavioral levels. The reduced activity in extrastriatal visual areas could be similarly interpreted. The increased brain response in the right supramarginal area could represent altered sensitivity to social-emotional cues.

In the aforementioned study of ToM, Trojsi et al. (2017) assessed, only at baseline, the resting state functional connectivity with fMRI in 21 ALS patients compared to a matched group control. Subjects also underwent affective and cognitive ToM tasks both at baseline and after 6 months. Compared to controls, ALS patients exhibited abnormalities (1) within the DMN (Default mode network) with decreased connectivity in anterior node and increased connectivity in posterior node, (2) within the right FPN (Fronto-parietal network) with decreased connectivity in the supramarginal gyri and (3) within the left FPN and SLN (salient network) with decreased connectivity on the medial and dorsolateral prefrontal cortices. Positive correlations were found between affective ToM performances and functional connectivity of the posterior node of the DMN and the supramarginal gyri. ToM alterations were associated to decreased connectivity in the posterior cingulate cortex (PCC) and the occipital gyri of DMN. The authors suggest that these results support the hypothesis of the potential role of frontotemporoparietal network structures, such as the PCC and the supramarginal gyri in reasoning about the contents of another person's mind particularly in affective mentalizing and in empathetic face processing.

# Behavioral Changes (Table 3) Baseline

#### **Frequency and typology**

Behavioral changes are increasingly being recognized as a common feature in ALS, and may be similar to those observed in FTD. They occur in 24–69% of patients with ALS (Murphy J. M. et al., 2007; Gibbons et al., 2008; Witgert et al., 2010; Lillo et al., 2011; Bock et al., 2016; Burke et al., 2017), 6–25% of whom meet the criteria for FTD (Murphy J. et al., 2007; Lillo et al., 2011; Bock et al., 2016; Burke et al., 2017). In some patients, they may appear as early features, even prior to the development of the motor symptoms (Mioshi et al., 2014).

Apathy seems to be the most common behavioral change (Grossman et al., 2007; Witgert et al., 2010; Lillo et al., 2011; Kasper et al., 2015). The most frequently encountered subtype of apathy in ALS is lack of initiation, that is, a lack of motivation to self-generate thoughts (Radakovic et al., 2017). Patients in the advanced stage may display aggressiveness and obsessiveness (Marconi et al., 2012). Disinhibition, impulsivity, lack of foresight and planning, distractibility, reduced concern for hygiene, irritability, increased self-centeredness and reduced concern for the feelings and needs of others, new unusual habits, loss of insight, and blunting of the primary emotions of happiness, sadness, fear and anger have also been reported (Gibbons et al., 2008; Giordana et al., 2011; Burke et al., 2017). Finally, changes such as aspontaneity, disorganization, and mental rigidity have also been observed (Consonni et al., 2013). Insight on behavioral changes is altered in ALS-FTD, but not in ALS without dementia (Woolley et al., 2010).

#### **Implications for the disease course**

The presence of behavioral symptoms has practical implications, as it impacts the patients' psychological state (and that of their caregivers), with consequences for their QoL and indeed their prognosis.

Rabkin et al. (2016) showed that whereas cognitive status does not correlate with mood, patients with behavioral impairment report more depressive symptoms, greater hopelessness, negative mood, and more negative feedback from spouses or caregivers. Caga et al. (2018) found that apathy was associated with more depressive symptoms and a poorer QoL, especially for achievement in life and community connectedness. Aggressiveness and obsession in the advanced stages of the disease correlate with a high level of anxiety in both patients and caregivers (Marconi et al., 2012). Unglik et al. (2018) reported

#### TABLE 3 | Behavioral changes in ALS.

fnins-13-00951 September 6, 2019 Time: 18:1 # 10


#### TABLE 3 | Continued

fnins-13-00951 September 6, 2019 Time: 18:1 # 11


AES, Apathy Evaluation Scale; ALS-FRS, ALS Functional Rating Scale; BBI, Beaumont Behavioral Inventory;CBI, Cambridge Behavioral Inventory; CGBI, Caregiver Burden Inventory; DAS, Dimensional Apathy Scale; DASS, Depression,Anxiety and Stress Scale; DKEFS, Delis-Kaplan Executive Functioning Scales; EPN-31, Positive and Negative Emotionality Scale; HADS, Hospital Anxiety and Depression Scale; MND, Motor Neuron Disease; MoCA, Montreal Cognitive Assessment; PWI, Personal Wellbeing Index; WM, White Matter. For more please refer to previous tables.

that both apathetic and nonapathetic patients reported anxious and depressive symptoms, and the only significant difference between the two groups was that apathetic and anxious patients experienced more negative emotions, including sadness, shame and anger, than anxious patients without apathy.

The strongest predictor of high caregiver burden is patients' abnormal behavior (e.g., apathy and disinhibition), rather than physical disability (Lillo et al., 2012; Watermeyer et al., 2015; Tremolizzo et al., 2016). The level of depressive and anxious symptoms in caregivers is also correlated with behavioral changes (Watermeyer et al., 2015; Tremolizzo et al., 2016). Caregiver burden is further influenced by patients' everyday skills, motivation and memory, mostly because poor motivation, memory dysfunction, and difficulty performing activities of daily living require more support in the shape of direct supervision, prompting, or hands-on care (Andrews et al., 2017b).

Survival in ALS is highly influenced by the presence or absence of apathy, with a median survival time of 21.7 months in the case of moderate-to-severe apathy, 46.9 months in the case of mild apathy, and 51.9 months when apathy is absent (Caga et al., 2018). Apathy correlates negatively with survival time (Unglik et al., 2018).

#### **Associated factors**

Behavioral changes can occur either on their own or in the presence of a cognitive deficit (Murphy J. M. et al., 2007; Witgert et al., 2010). When cognitive impairment is present, behavioral symptoms seem to be greater (Witgert et al., 2010; Consonni et al., 2013). Apathy has been found to be associated with verbal fluency, leading some authors to suggest that apathy in ALS is underpinned by the medial prefrontal cortex (Grossman et al., 2007; Radakovic et al., 2017). Although Burke et al. (2017) found that social cognitive performances predicted behavioral changes, others have failed to do so (Terada et al., 2011).

Studies assessing relationships between behavioral changes and physical parameters have yielded somewhat conflicting results. Terada et al. (2011) found that apathy was correlated with the ALS condition per se, and not with the physical disability. Their population, however, consisted of only mildly disabled patients, who needed no assistance with activities of daily living, and all had normal blood gases. By contrast, in a larger sample of patients, with a wider range of physical impairments (although no details were given about their respiratory status), Ohta et al. (2017) did find a correlation between scores on the ALS-FTD Questionnaire and ALS Functional Rating Scale. Whereas the motor deficit may not directly impact behavior in mildly impaired patients, it is obviously mandatory to control for physical parameters, and first and foremost for blood gases, to avoid erroneously ascribing the consequences of hypercapnic encephalopathy to the cerebral neuronal involvement of ALS in the advanced stage.

Bock et al. (2017) found no association between bulbar or spinal onset and behavioral changes at baseline, whereas other authors (Gibbons et al., 2008; Crockford et al., 2018) have reported that bulbar palsy is associated with a higher rate of behavioral change.

#### Over Time

The way behavioral symptoms change as the disease progresses is still a matter of debate.

Most cross-sectional studies have shown no correlation between the severity of behavioral changes and time elapsed between disease onset and the time of study, suggesting that there is no significant decline over time (Woolley et al., 2011; Bock et al., 2017; Femiano et al., 2018). Crockford et al. (2018) assessed behavior in a sample of 149 patients using the ECAS caregiver behavioral interview and examined whether behavior was related to disease stages according to King's Clinical Staging System (stage 1 from stage 4 depending on the number of affected

bodily regions). Almost 40% of patients were behaviorally impaired (most of them with apathy) and a higher number of behavioral features was found across advancing stages. This result suggests that, contrary to what had been reported previously, behavioral and cognitive impairments are more severe in more severe disease stages.

Similarly, most of the longitudinal studies support the idea that behavioral manifestations over time may either increase (Bock et al., 2017) or appear as the illness progresses, with the emergence of frustration tolerance, reduced insight, mental rigidity, and lack of interest (Bock et al., 2017; Woolley et al., 2018). The severity of these disturbances increases faster over time when dementia is present in patients with the C9+ mutation (Floeter et al., 2017). Poletti et al. (2018) found no change over time in the prevalence of behavioral changes as measured with the ECAS, but did find an increase in behavioral disturbances as measured with the Frontal Behavioral Inventory, which is possibly a more sensitive tool, as it quantifies patients' performances, whereas the ECAS only indicates whether or not there are changes.

#### Neuroimaging Correlates

Most studies addressing the issue of the neural correlates of behavioral changes in ALS point to a significant correlation between apathy scores and prefrontal cortex atrophy, especially in the orbitofrontal and dorsolateral areas (Tsujimoto et al., 2011; Consonni et al., 2018), whereas disinhibition is negatively correlated with thickness of the right frontotemporal and cingulate cortices (Consonni et al., 2018).

Diffusion tensor imaging studies have shown a significant negative correlation between apathy scores and fractional anisotropy in the right anterior cingulate region, corpus callosum, bilateral amygdalae, left thalamus, and fornix, with atrophy of these brain regions (Woolley et al., 2011; Branco et al., 2018; Femiano et al., 2018). Correlations between apathy and the prefrontal cortex have also been found in other neurological diseases.

On the whole, the relationships between the anterior cingulate (and possibly some subcortical structures) and apathy, and between the anterior temporal lobe and disinhibition, appear quite consistent across studies.

# PSYCHOLOGICAL ADJUSTMENT (TABLE 4)

# Psychological Reactions and Wellbeing Baseline

Chronic diseases induce a wide range of psychological responses, such as uncertainty about the future, anxiety, and depression. These psychological responses can have a major impact on health, through for example the perceived somatic symptom burden, adherence to treatment and compliance with care, malnutrition, and mortality (Cukor et al., 2006; Katon et al., 2007), and therefore need to be detected and supported.

Patients with ALS were initially described as abnormally positive (Brown and Mueller, 1970) and, in contrast to what is usually described in other serious chronic somatic diseases, had a relatively low prevalence of depressive disorders ranging from 0% (Rabkin et al., 2000; Bungener et al., 2005) to 10% (Hammer et al., 2008; McElhiney et al., 2009; Rabkin et al., 2016; Wei et al., 2016), according to their responses to semi-structured questionnaires based on international criteria for depression.

Nevertheless, these results have to be considered in the light of several additional factors. First, when patients are assessed at the extreme stages of the disease (either soon after diagnosis or at a very advanced stage), validated scales indicate that the prevalence of depressive symptoms is around 20% (Rabkin et al., 2005), and this figure rises to above 50% when self-report questionnaires are used (Wicks et al., 2007). Second, a history of depression before the onset of ALS increases the prevalence rate from 10 to 21% when patients are assessed with a semi-structured questionnaire (McElhiney et al., 2009; Ferentinos et al., 2011). An interesting study conducted by Roos et al. (2016) found that the risk of receiving a diagnosis of depression was increased during the year before and the year after the diagnosis of ALS. This is also true for other major psychiatric disorders, namely schizophrenia, bipolar disorder, and anxiety, some of which may predate the diagnosis of ALS by as much as 5 years (Turner et al., 2016). Self-report questionnaires designed to probe depressive symptoms may have their limits, insofar as they are not able to diagnose a depressive state with certainty, leading to potential overestimation of the rate of depression, but they do have the advantage of detecting the potential presence of depressive symptoms that reflect a degree of distress. Studies using this type of instrument report higher rates of depression of above 30% (Lulé et al., 2008; Atassi et al., 2011; Grehl et al., 2011; Carvalho et al., 2016). Thus, even when the official diagnostic criteria are not met, it does not mean that patients do not feel they are affected by their disease. One must thus be aware that depression rates in ALS vary greatly, depending on the tools used to assess it. Finally, studies comparing the rate of depression in patients with ALS versus other chronically and seriously ill patients (e.g., with neuromuscular disease or receiving palliative care for cancer) have failed to find any significant differences, with 8–10% of patients in each group having a diagnosis of major depression according to DSM-IV criteria, and 50% mild-to-moderate depressive symptoms, as measured with the Beck Depressive Inventory (Taylor et al., 2010; Lulé et al., 2012).

Furthermore, more than one third of patients with ALS are on antidepressants (Pisa et al., 2015). Not all studies assessing the prevalence of depression in ALS take this fact into account, which could induce a bias and result in underestimation of the prevalence of depression in this population.

Intuitively, one might assume that the worsening of physical disability increases signs of depression. However, many studies have found either no such link or an inverse link between the severity of the motor disability and scores on mood scales (Lulé et al., 2008; Jelsone-Swain et al., 2012; Chen et al., 2015; Thakore and Pioro, 2016; Wei et al., 2016). This is only true, however, for the consequences of spinal involvement, as the presence of bulbar symptoms, or of breathing difficulty, does increase depressive

#### TABLE 4 | Psychosocial adjustment and coping in ALS.

fnins-13-00951 September 6, 2019 Time: 18:1 # 13


(Continued)

#### TABLE 4 | Continued

fnins-13-00951 September 6, 2019 Time: 18:1 # 14


(Continued)

#### TABLE 4 | Continued

fnins-13-00951 September 6, 2019 Time: 18:1 # 15


ADI, Assessment of Depression Inventory; CES-D, Center For Epidemiologic Studies Depression Scale; CISS, Coping Inventory For Stressful Situations; DSM, Diagnostic and Statistical Manual of Mental Disorders; GDS, Global Deterioration Scale; ICD-10, International Classification of Diseases; IECS, Internal-External Control Scale; HRQL, Health-Related Quality of Life; MAACL, Multiple Affect Adjective Check List; MLDL, Munich Quality-of-Life Dimensions List; MMSE, Mini-Mental State Examination; MQoL, Mcgill Quality-of-Life Questionnaire; SAS, Self-Rating Anxiety Scale; SCID, Structured Clinical Interview For DSM-IV; SDS, Zung Self-Rating Depression Scale; SeiQoL-DW, Schedule For The Evaluation of Individual Quality of Life; SPBS, Self-Perceived Burden Scale; ZDS, Zung Depression Scale. For more please refer to previous tables.

symptoms (Hillemacher et al., 2004; Goldstein et al., 2006a; Miglioretti et al., 2008; Jelsone-Swain et al., 2012).

Studies looking for a link between disease duration and the severity of depressive symptoms have reported contradictory results, with some finding a positive correlation, some a negative one (Cui et al., 2015; Hillemacher et al., 2004, and others no link at all (Atassi et al., 2011; Caga et al., 2015).

Severe somatic diseases may induce a number of other psychological reactions, such as anxiety, hopelessness, or suicidal thoughts. Kurt et al. (2007) found that the prevalence of anxiety in ALS ranged from 0 to 30%. Vignola et al. (2008) reported that almost 75% of patients experienced moderate-to-severe state anxiety at baseline, which was correlated with trait anxiety. However, Pagnini et al. (2012) found that only 20% of patients had scores above the anxiety cut off. Suicidal thoughts are not rare in patients, ranging from 19 to 39% across studies (Albert et al., 2005; Rabkin et al., 2015; Verschueren et al., 2018). This rate rises to more than 50% when assisted suicide is considered (Ganzini et al., 1998). Patients have an almost 6-fold higher risk of suicide, especially in the first year after symptom onset and when they are younger (Fang et al., 2008). A wish to die is not always associated with a depressive state (Albert et al., 2005) but it is linked to less optimism, less comfort in religion, and greater hopelessness (Albert et al., 2005; Verschueren et al., 2018).

In addition to psychological signs of distress such as depression or anxiety, the estimation of wellbeing is based on the person's satisfaction with his or her QoL. In patients with ALS, QoL is found to be high, especially when the measurement scales are adapted to the disease (Norris et al., 2010; Jakobsson Larsson et al., 2017) and avoid lending too much importance to the patient's physical state (Norris et al., 2010; Jakobsson Larsson et al., 2017). The Amyotrophic Lateral Sclerosis Quality of Life (ALSSQOL) was specifically designed for ALS, and has a revised and a short form (Felgoise et al., 2018) that give a balanced appraisal. Qol depends not only on patients' physical disabilities, but also on their religiosity/spirituality and sociability (Norris et al., 2010; Simmons, 2015). Other factors influencing QoL are anxiety and depression, the ability to cope with physical disabilities, fatigue, and hopelessness (Lou et al., 2003; Abe, 2004; Pagnini et al., 2012; Sandstedt et al., 2016; Jakobsson Larsson et al., 2017).

#### Over Time

Surprisingly, longitudinal studies of changes in patients' mood as the disease progresses report either stability (Rabkin et al., 2005; Gauthier et al., 2007; McElhiney et al., 2009; Matuz et al., 2015) or a decrease in the depression rate (McElhiney et al., 2014). When Goldstein et al. (2006a) assessed psychosocial factors influencing patients' psychological state (level of depression and anxiety) at baseline and 6 and 11 months later, they found that the quality of premarital intimacy and social support at disease onset influenced the psychological wellbeing of patients in the more advanced stages of the disease. Anxiety is particularly high in the diagnostic phase, but tends to decrease thereafter in patients, though not in caregivers (Vignola et al., 2008). Importantly, these authors also found that in caregivers, state anxiety was linked to trait anxiety, whereas in patients, it was correlated with clinical features such as a shorter disease course and the presence of depression.

Together with other factors, anxiety negatively impacts QoL (Jakobsson Larsson et al., 2017).

Caregivers' psychological distress and perceived burden is often reported to increase over time (Goldstein et al., 2006b; Gauthier et al., 2007). The burden is determined by a combination of factors, including both the caregiver's own personality and the patient's characteristics. Regarding the latter, Lillo et al. (2012) found that behavioral changes are a greater determinant of caregivers' wellbeing than physical disability, although motor impairment also plays a part, according to Chiò et al. (2005). Behavioral changes may include apathy, loss of empathy, and a lack of appreciation of the efforts made to satisfy their needs (Rabkin et al., 2009). As the disease progresses, patients' plans for future, particularly regarding life support, may also play a part and influence their caregivers' QoL. A longitudinal study by Rabkin et al. (2009) found that caregivers tended to be less depressed as the disease advanced, regardless of the outcome (death or tracheostomy), whereas the depression scores of patients remained stable.

Quality of life as reported by patients remains stable even in the advanced stages of the disease (Rabkin et al., 2000; De Groot et al., 2007; Roach et al., 2009). Religiosity, social support, level of anxiety and depression remain the mean determinants of wellbeing, despite the increase in motor impairment (Jakobsson Larsson et al., 2017).

# Adaptive Mechanisms in the Face of the Disease

To explain the fact that patients with ALS have a relatively good adaptive reaction to their illness, some authors have suggested that they are in denial of the disease, thus protecting them from depression (Brown and Mueller, 1970; Miglioretti et al., 2008), but this hypothesis has not been confirmed (Houpt et al., 1977).

Psychological adjustment refers to the psychological processes that take place in response to a stressful situation like chronic illness and associated treatment. Various models have been developed, including the stress coping model (Lazarus and Folkman, 1984), illness representation model, adaptive tasks and coping model (Moos and Holahan, 2007), and adjustment model (Moss-Morris, 2013). Based on Lazarus and Folkman's theoretical framework, Matuz et al. (2010) developed an interesting integrative model of patients' adaptation in the face of the disease, whereby patients' mood state and QoL are influenced by social support, cognitive appraisal and coping strategies. The latter include problem management, problem appraisal, emotion regulation, and emotional avoidance. Social support encompasses perceived social support, received social support, and need for social support. Matuz et al. (2010) found that patients used emotion regulation and looked for social support, and the more they used emotion-focused strategies, the lower they scored on a depression scale. When Lulé et al. (2012) compared patients with ALS and patients with cancer on psychological adjustment, they failed to find any significant differences in the use of coping strategies, even if, on average, the ALS group scored lower than the cancer group on a scale measuring active coping strategies (thinking about the situation and trying to solve it, taking an adequate step to deal with their condition). By contrast, depressive symptoms and a high level of burden are more often present in caregivers if they use emotion-focused coping strategies (Siciliano et al., 2017). Factors that have been shown to help caregivers cope with the impact of the disease include social support, as well as anticipatory coping with foreseeable difficulties (Goldstein et al., 1998).

Studies addressing changes in psychological adjustment strategies in ALS over time have yielded conflicting results. Jakobsson Larsson et al. (2016) found that the patients used the same coping strategies throughout the disease course, namely support seeking, positive action, independence, and positive thinking. By contrast, Matuz et al. (2015) reported that while patients used both problem- and emotion-focused coping strategies at disease onset, they used less emotion regulation later on. On the other hand, they continued to have higher perceived social support and an accurate assessment of their own coping potential. These two factors were correlated with depressive scores at follow up. Perceived social support, which reflects patients' view about the amount and quality of support they receive, can help patients use effective coping strategies, encourage positive health behaviors, and reduce physiological reactivity to stress. A positive subjective appraisal of one's coping potential indicates that patients feel they are keeping control over their state, which probably decreases their anxiety over physical loss.

At a time when some people advocate legalizing euthanasia for intractable disease, accurate knowledge of the mechanisms by which many patients with ALS succeed in coping with such a dreadful disease is crucial, and could greatly enhance the assistance given to both patients and caregivers in their daily struggle.

# CONCLUSION

The objective of this review was to provide readers with a clear picture of all the manifestations of ALS both at baseline and throughout the course of the disease. Unlike the relentless physical deterioration, cognitive, behavioral and psychological changes are extremely variable. The heterogeneity of clinical situations, and in particular the variety of symptoms present at baseline, seems to strongly influence patients' clinical course. Regarding their psychological reactions, even if they rarely develop a severe psychiatric illness, they can experience considerable distress, especially at the time of diagnosis and in the advanced stage of the disease. Caution must be taken not to minimize these psychological reactions and to give patients the best possible personalized support at these times. A careful examination of the psychological trajectory shows that the increase in motor disability is not the only determinant of psychological wellbeing. The quality of social support and the use of appropriate adaptive strategies enable patients to cope with their condition and to maintain good psychosocial functioning as the disease progresses.

This review also illustrates the theoretical and practical difficulties that may arise when investigating cognitive, behavioral

and psychological aspects of ALS. The first concerns the tools that are to be used. First, these must be adapted to the physical disability of patients with ALS. Potentially confounding variables, such as respiratory insufficiency, have to be controlled. As we have seen, a large number of tests have been used by the different authors. Such a diversity makes comparisons difficult and may at least partially account for the discrepancies that are often observed across studies. Investigators will have eventually to try to agree about which tests should be used, either as routine tests (such as the ECAS for the cognitive assessment, or ALS-FRS or the bulbar Norris scale for physical parameters) or for more specific purposes. It will also be interesting to further study less extensively explored cognitive domains such as memory or social cognition.

Another difficulty is that of longitudinal studies. These are particularly challenging, because of the major physical changes that occur over time. It is not clearly known, for example, whether a cognitive impairment, if any, worsens in line with the physical disability or not. Also, the short life expectancy of many patients is responsible for a substantial drop-out rate. These difficulties can be addressed by recruiting as many patients as possible, preferably in the setting of ALS centers. Bigger samples may also help to tackle the problem of physical and

#### REFERENCES


neuropsychological heterogeneity. For example, the seemingly simple question as to whether bulbar-onset patients are more cognitively affected than spinal-onset ones or not is still debated. Likewise, there have been very few studies about the relationships between the cognitive (or psychological, or behavioral, for that matter) and physical profiles of patients. Finally, morphological and functional cerebral imaging studies should be increasingly undertaken in order to learn more about the mechanisms of neuropsychological impairment, and to improve as much as possible the care of patients with ALS.

#### AUTHOR CONTRIBUTIONS

BD, SB, and FV contributed to conceptualization and methodology. SB wrote the first draft of the manuscript. BD, FV, and FE reviewed the manuscript. SB, FV, and FE edited the manuscript.

#### ACKNOWLEDGMENTS

We would like to thank Mrs. Elizabeth Portier for assisting in language editing.


profiles in amyotrophic lateral sclerosis. EJNMMI Res. 7, 21. doi: 10.1186/ s13550-017-0267-2


combined voxel-based morphometry and tract-based spatial statistics wholebrain analysis. Brain Imag. Behav. 12, 547–563. doi: 10.1007/s11682-017- 9722-y


neuroendocrine model of FTLD attenuation. Amyotroph. Lateral Scler. Front. Degener. 18, 74–85. doi: 10.1080/21678421.2016.1249487


sclerosis. Amyotroph. Lateral Scler. Mot. Neuron Disord. 5, 245–249. doi: 10. 1080/14660820410021294


lateral sclerosis (ALS). J. Neurol. 254, 519–527. doi: 10.1007/s00415-006- 0409-3


Amyotroph. Lateral Scler. Front. Degener. 16, 265–273. doi: 10.3109/21678421. 2014.980428


sclerosis: preliminary report from a serbian centre for motor neuron disease. Amyotroph. Lateral Scler. Front. Degener. 17, 543–547. doi: 10.1080/21678421. 2016.1211148



**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 © 2019 Benbrika, Desgranges, Eustache and Viader. 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.

digital media

of impactful research

article's readership