# BIOMARKERS IN LEISHMANIASIS

EDITED BY : Javier Moreno and Eugenia Carrillo PUBLISHED IN : Frontiers in Cellular and Infection Microbiology

### 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 978-2-88963-339-5 DOI 10.3389/978-2-88963-339-5

## 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

# BIOMARKERS IN LEISHMANIASIS

Topic Editors: Javier Moreno, Carlos III Health Institute, Spain Eugenia Carrillo, Carlos III Health Institute, Spain

Citation: Moreno, J., Carrillo, E., eds. (2020). Biomarkers in Leishmaniasis. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-339-5

# Table of Contents


Johan van Griensven, Bewketu Mengesha, Tigist Mekonnen, Helina Fikre, Yegnasew Takele, Emebet Adem, Rezika Mohammed, Koert Ritmeijer, Florian Vogt, Wim Adriaensen and Ermias Diro

*16 Analysis of the Antigenic and Prophylactic Properties of the* Leishmania *Translation Initiation Factors eIF2 and eIF2B in Natural and Experimental Leishmaniasis*

Esther Garde, Laura Ramírez, Laura Corvo, José C. Solana, M. Elena Martín, Víctor M. González, Carlos Gómez-Nieto, Aldina Barral, Manoel Barral-Netto, José M. Requena, Salvador Iborra and Manuel Soto


Eduardo Ontoria, Yasmina E. Hernández-Santana, Ana C. González-García, Manuel C. López, Basilio Valladares and Emma Carmelo


Ivan Best, Angela Privat-Maldonado, María Cruz, Mirko Zimic, Rachel Bras-Gonçalves, Jean-Loup Lemesre and Jorge Arévalo

*86 Comparative Evolution of Sand Fly Salivary Protein Families and Implications for Biomarkers of Vector Exposure and Salivary Vaccine Candidates*

Iliano V. Coutinho-Abreu and Jesus G. Valenzuela

*104 Biomarkers Associated With* Leishmania infantum *Exposure, Infection, and Disease in Dogs*

Carla Maia and Lenea Campino

*122 In Search of Biomarkers for Pathogenesis and Control of Leishmaniasis by Global Analyses of* Leishmania*-Infected Macrophages*

Patricia Sampaio Tavares Veras, Pablo Ivan Pereira Ramos and Juliana Perrone Bezerra de Menezes

*140 Reverse Epidemiology: An Experimental Framework to Drive* Leishmania *Biomarker Discovery* in situ *by Functional Genetic Screening Using Relevant Animal Models*

Laura Piel, Pascale Pescher and Gerald F. Späth


Micely d'El-Rei Hermida, Caroline Vilas Boas de Melo, Isadora dos Santos Lima, Geraldo Gileno de Sá Oliveira and Washington L. C. dos-Santos


Thouraya Boussoffara, Sadok Chelif, Melika Ben Ahmed, Mourad Mokni, Afif Ben Salah, Koussay Dellagi and Hechmi Louzir

*187 Phenotypic and Functional Profiles of Antigen-Specific CD4+ and CD8+ T Cells Associated With Infection Control in Patients With Cutaneous Leishmaniasis*

Adriana Egui, Darién Ledesma, Elena Pérez-Antón, Andrés Montoya, Inmaculada Gómez, Sara María Robledo, Juan José Infante, Ivan Darío Vélez, Manuel C. López and M. Carmen Thomas


# Editorial: Biomarkers in Leishmaniasis

Eugenia Carrillo\* and Javier Moreno\*

*WHO Collaborating Centre for Leishmaniasis, National Centre for Microbiology, Instituto de Salud Carlos III, Madrid, Spain*

Keywords: biomarkers, leishmaniasis, diagnostic, vaccine, cure

**Editorial on the Research Topic**

**Biomarkers in Leishmaniasis**

# INTRODUCTION

Leishmaniasis is one of the most deadly, yet neglected, of all tropical diseases (Alvar et al., 2012). The eco-epidemiological characteristics of leishmaniasis render it a complex problem. There are over 20 species of leishmaniasis pathogens transmitted by different species of sand fly (Phlebotomus spp.). Transmission may be anthroponotic or zoonotic, the latter entailing different animal reservoirs. Depending on the species of Leishmania and the immune response to it, the disease may present as cutaneous leishmaniasis (CL), mucocutaneous leishmaniasis (MCL), visceral leishmaniasis (VL), or post-kala azar dermal leishmaniasis (PKDL) (Burza et al., 2018).

The complexity of leishmaniasis means different strategies are needed if it is to be controlled and eliminated (Matlashewski et al., 2014; Rijal et al., 2019). For example:

### Edited and reviewed by:

*Jeroen P. J. Saeij, University of California, Davis, United States*

### \*Correspondence:

*Javier Moreno javier.moreno@isciii.es Eugenia Carrillo ecarrillo@isciii.es*

### Specialty section:

*This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology*

Received: *09 October 2019* Accepted: *29 October 2019* Published: *12 November 2019*

### Citation:

*Carrillo E and Moreno J (2019) Editorial: Biomarkers in Leishmaniasis. Front. Cell. Infect. Microbiol. 9:388. doi: 10.3389/fcimb.2019.00388* - Diagnostic methods need to be faster and simpler, but sensitive and robust, and allow early diagnoses to be made. Follow-up methods are also needed that confirm patient responses and help predict the risk of relapse.


Biomarkers have a central role to play in the above challenges by providing information on patient immune status, the response to treatment, exposure to vectors, the role of animal reservoirs, and the epidemiology of infection, etc. New biomarkers need to be found that will allow the development of tools for assessing the effectiveness of treatments, that can confirm when a cure has been achieved, to identify asymptomatic persons and rates of transmission in endemic areas, to develop rapid, non-invasive tests, and for checking the immune response to experimental vaccines (Ibarra-Meneses et al., submitted).

This Research Topic, entitled "Biomarkers in Leishmaniasis", is a collection of 19 articles, some of which examine the latest advances in biomarkers of the different types of leishmaniasis, while others report original research into biomarker identification and characterization.

# DIAGNOSTIC AND DISEASE PROGRESS BIOMARKERS

Several articles in the cited collection examine the identification of new biomarkers useful for understanding the pathogenesis of leishmaniasis, and for improving its diagnosis. The clinical complexity and epidemiology of leishmaniasis is a challenge in the identification of biomarkers able to track the progress of the disease. This is made clear by different review articles that focus on its different clinical forms. The work of Brodskyn and Kamhawi on biomarkers of zoonotic VL in Latin America, focuses on humans and dogs, and addresses the need to examine a combination of inflammatory mediators for the development of a tool that distinguishes between the different stages of the disease. They also discuss the use of serum antibodies against the highly immunogenic salivary proteins of Lutzomyia as biomarkers of exposure to the vector in humans and dogs.

In their review, Bahrami et al. highlight the scarcity of specific markers for CL. Apart from abnormalities in the delayed hypersensitivity test, in T cell subpopulations, cytokine levels and enzyme (e.g., adenosine deamidase and L-argininase) concentrations, these authors suggest the need to develop analyses based on comparing the transcriptome of the lesion with that of healthy skin (Christensen et al., 2016; Masoudzadeh et al., 2017). The identification of biomarkers able to predict the result of infection by different species of Leishmania is also a major challenge in CL (Patino and Ramírez, 2017). For example, the physiopathology of PKDL (which follows VL in some treated patients) is different to that of both VL and CL (Kip et al., 2015), and patients show responses to treatment that are difficult to assess (the lesions can take a long time to heal, thus responses may take time to appear). In their review, Zijlstra indicate that current biomarkers for PKDL lesions are unsatisfactory. Certainly, clinical assessment is subjective and not very precise, and while the parasite load can be determined by qPCR, serological tests such as DAT, rK39 ELISA, and rK39 RDT lack specificity since antibodies may hang over from previous bouts of VL. Moreover, the systemic and skin immune responses are different. Zijlstra also questions whether biomarkers in the blood (such as cytokines or cell populations) properly reflect skin-level changes, and declares that new avenues need to be explored. These might include 3D optical scanning and the undertaking of longitudinal studies that can provide a description of PKDL before, during and after cure.

Dogs play a major role in the transmission of the parasite to humans (Moreno and Alvar, 2002). The present collection therefore also includes an article by Maia and Campino which is exclusively devoted to canine leishmaniasis. This review discusses the latest advances in the identification of biomarkers associated with infection by L. infantum in dogs. The early detection and treatment of infected animals is a basic requirement in the control of human VL (Alvar et al., 2004). Canine leishmaniasis has a wide spectrum of manifestations, the result of complex host-parasite interactions (Reis et al., 2010), and these authors conclude that no single biomarker is able to confirm a diagnosis, reflect the effectiveness of treatment, or indicate the infectivity of affected dogs.

In their contribution, Ontoria et al. report the expression of different genes in the spleens of infected and control Balb/c mice, the final aim of their research being to better understand the immunological mechanisms that lead to protection or disease progression, and the identification of associated biomarkers. d'El-Rei Hermida et al. review the histological changes that occur in the spleen in severe VL, and record the events that eventually lead to its destruction. Garde et al. discuss markers of disease progression, reporting on the antigenicity of Leishmania antigens and the role of the eukaryotic initiation factors F2, F2B, LieIF2, and LieIF2B. These proteins, to which specific antibodies were detected in the serum of patients with VL, and in dogs with canine leishmaniasis, induce a humoral response in a murine model, along with the production of IL-10. IL-10 favors the progression of the disease and therefore could act as an indicator of the same.

Piel et al. propose experimentally infecting mice with cosmidtransfected parasites as a means of searching for new genetic markers. This might allow the identification of genetic loci associated, for example, with resistance to medications, or that might act as new treatment targets. Any factors thus identified, however, would have to be validated in specific field studies.

# BIOMARKERS OF CURE

Some of the articles included in this Research Topic focus on the identification of new biomarkers associated with the response to treatment, and that provide confirmation of cure. A biomarker that indicates a cure to have been achieved could be used to reduce treatment times and prevent relapses, help adjust doses, and be of use in research into new treatments or combinations of current medications (Alves et al., 2018). Marlais et al. report results obtained in a clinical trial involving patients with VL in which they measured Leishmania-specific IgG1 in the serum before and after treatment. Using the VL Sero K-SeT rapid diagnostic test and ELISA, they show that high post-treatment concentrations of these antibodies are associated with relapse, while low (or no) concentrations are associated with cure. They also report the above rapid diagnostic test may be of help in the diagnosis of PKDL.

In the search for biomarkers that indicate the cure of CL, Montoya et al. report the use of a hamster model to analyse skin lesions for the production of the growth factors EGF, TGFbeta1, PDGF, and FGF. They indicate an increase in TGFbeta1 to be associated with active disease, while high EGF levels are associated with the cure of the lesion. They conclude that the EGF/TFGbta1 ratio might provide an excellent biomarker of the establishment of an infection or an adequate response to treatment. Kip et al. investigate the role of neopterin, a marker of macrophage activation, and its association with the response to treatment for VL. These authors examine the plasma neopterin concentration before and after treatment and discuss its potential for identifying patients at risk of suffering an early relapse.

Botana et al. compare serological, parasitological and cellular response biomarkers in patients with different forms of leishmaniasis caused by L. infantum. Their work, which was performed with patients with active disease plus others who had been cured, shows significant differences in the results of pre- and post-treatment parasitological tests (they became negative), and in cellular immunity tests (they became positive). However, no changes to the outcome of serological tests were seen. These authors conclude peripheral blood mononuclear cell (PBMC) proliferation following stimulation with Leishmania antigens, and the secretion of IFN-gamma, to be good markers of cure of VL. However, for CL, MCL, and localized leishmanial lymphadenopathy (LLL), these same tests detected no difference between the active and cured phases.

# BIOMARKERS OF ASYMPTOMATIC INFECTION

Many of the articles in this Research Topic insist on the importance of identifying biomarkers of asymptomatic infection. This is necessary if we are to know the true prevalence of Leishmania infection in any determined area, and for designing strategies to control the disease (Alvar et al., submitted). The search for such biomarkers is limited, however, by the deficient definition of an asymptomatic patient as "someone in an endemic area who has an immune response (antibody- or cellbased) against Leishmania but who remains healthy." This partly explains why, to date, there is no reference method for detecting asymptomatic infection.

This collection of articles contains two original research papers that focus on asymptomatic infection. Best et al. report that in asymptomatic persons who had traveled to areas where American tegumentary leishmaniasis (ATL) is endemic, the expression of IFN-gamma following the stimulation of their PBMC with Leishmania antigens is directly related to the length of time spent in the area. This can provide information on how long the asymptomatic condition can last.

In very different work, Coutinho-Abreu and Valenzuela provide a comparative phylogenetic analysis of the proteins in sand fly saliva, and report differences in the amino acid sequence of those of New World and Old World flies, and indeed proteins unique to them, that might serve as biomarkers of infection by a determined species.

# BIOMARKERS FOR VACCINE ASSESSMENT

The search for biomarkers that correlate with the degree of protection achieved are vital in the development of Leishmania vaccines (Moreno, 2019). Several contributions to this Research Topic focus on the immune response to the parasite, and on parasite antigens that might be candidates for use in vaccine production. Egui et al. examine the functional and phenotypic profiles of Leishmania-specific CD4+ and CD8+ cells from patients cured of CL caused by L. panamensis, as well as those of healthy, asymptomatic patients, and report that protection against the disease is associated with an increased cytotoxic T cell response—something that could be very useful when monitoring patients. Boussoffara et al. come to a similar conclusion in their article on CL caused by L. major, and report that high levels of granzyme B are associated with better protection against the development of CL. It could therefore be a useful biomarker for assessing the effectiveness of vaccines. As mentioned by Ontoria et al., combining transcriptomic, proteomic, and metabolomic analyses may be of great interest when studying the factors involved in the cellular response, and for identifying biomarkers of infection. The review by Tavares-Veras et al. goes deeper into this question and discusses large scale studies that have identified and assessed biomarkers in infected macrophages. The parasite alters the protein profile of these cells, rendering them suitable for proteomic studies aimed at identifying new molecular biomarkers that might reveal the destiny of the host cell and pathogen (Jean Beltran et al., 2017).

# BIOMARKERS OF COINFECTION WITH HIV AND LEISHMANIA

Since many of the biomarkers identified are related to the host immune response, others need to be sought for use in situations in which no immune response occurs. Immunodepression increases the risk of developing leishmaniasis, alters the clinical spectrum of the disease, and increases the risk of therapeutic failure and relapse (van Griensven et al., 2014). Biomarkers for monitoring patients coinfected with Leishmania, and HIV, and who suffer from some other form of immunodepression, are therefore needed (Akuffo et al., 2018). The present collection contains two articles on the search for biomarkers for the former. In their work, aimed at identifying early markers of susceptibility or resistance to VL in patients with HIV, Adriansen et al. examine the serum levels of macrophage activators (sCD40L and neopterin) in such patients, and confirm that they are reduced in those with active VL, and increased in those with asymptomatic Leishmania infection. These markers could be useful when trying to predict the progress of disease in such patients. In contrast, van Griesvan et al. focus on the identification of biomarkers of therapeutic failure for VL and disease relapse in coinfected persons. This research paper shows that coinfected patients with high levels of Leishmania antigens in their urine at the moment of diagnosis of VL are at greater risk of therapeutic failure. In addition, those with high levels at the end of treatment are more likely suffer a relapse within 12 months. These results highlight the importance of antigenuria in monitoring the response to treatment and the risk of relapse in immunodepressed patients.

# CONCLUSIONS

The present collection of articles underscores the main problems faced in identifying biomarkers of leishmaniasis, and show that much work is needed to validate those already found. It is important that the knowledge we have be used in innovative ways resulting in novel clinical applications and rapid, sensitive and simple diagnostic tests.

# AUTHOR CONTRIBUTIONS

EC and JM have participated equally in the writing of this editorial.

# REFERENCES


# ACKNOWLEDGMENTS

The authors wish to thank all the authors who have sent their manuscripts to this Research Topic. We also want to thank all the reviewers who have participated in the revision of the manuscripts and have helped to improve the final result.


**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 Carrillo and Moreno. 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.

# Leishmania Antigenuria to Predict Initial Treatment Failure and Relapse in Visceral Leishmaniasis/HIV Coinfected Patients: An Exploratory Study Nested Within a Clinical Trial in Ethiopia

Johan van Griensven<sup>1</sup> \*, Bewketu Mengesha<sup>2</sup> , Tigist Mekonnen<sup>2</sup> , Helina Fikre<sup>2</sup> , Yegnasew Takele<sup>2</sup> , Emebet Adem<sup>2</sup> , Rezika Mohammed<sup>2</sup> , Koert Ritmeijer <sup>3</sup> , Florian Vogt <sup>1</sup> , Wim Adriaensen<sup>1</sup> and Ermias Diro<sup>1</sup>

<sup>1</sup> Department of Clinical Sciences, Institute of Tropical Medicine, Antwerp, Belgium, <sup>2</sup> Department of Clinical Sciences, University of Gondar, Gondar, Ethiopia, <sup>3</sup> Médecins Sans Frontières, Amsterdam, Netherlands

### Edited by:

Charles L. Jaffe, Hebrew University of Jerusalem, Israel

### Reviewed by:

Yang Zhang, University of Pennsylvania, United States Erdong Cheng, University of Pittsburgh Cancer Institute, United States

> \*Correspondence: Johan van Griensven jvangriensven@itg.be

Received: 11 January 2018 Accepted: 12 March 2018 Published: 29 March 2018

### Citation:

van Griensven J, Mengesha B, Mekonnen T, Fikre H, Takele Y, Adem E, Mohammed R, Ritmeijer K, Vogt F, Adriaensen W and Diro E (2018) Leishmania Antigenuria to Predict Initial Treatment Failure and Relapse in Visceral Leishmaniasis/HIV Coinfected Patients: An Exploratory Study Nested Within a Clinical Trial in Ethiopia.

Front. Cell. Infect. Microbiol. 8:94. doi: 10.3389/fcimb.2018.00094 Background: Biomarkers predicting the risk of VL treatment failure and relapse in VL/HIV coinfected patients are needed. Nested within a two-site clinical trial in Ethiopia (2011–2015), we conducted an exploratory study to assess whether (1) levels of Leishmania antigenuria measured at VL diagnosis were associated with initial treatment failure and (2) levels of Leishmania antigenuria at the end of treatment (parasitologically-confirmed cure) were associated with subsequent relapse.

Methods: Leishmania antigenuria at VL diagnosis and cure was determined using KAtex urine antigen test and graded as negative (0), weak/moderate (grade 1+/2+) or strongly-positive (3+). Logistic regression and Kaplan-Meier methods were used to assess the association between antigenuria and (1) initial treatment failure, and (2) relapse over the 12 months after cure, respectively.

Results: The analysis to predict initial treatment failure included sixty-three coinfected adults [median age: 30 years interquartile range (IQR) 27–35], median CD4 count: 56 cells/µL (IQR 38–113). KAtex results at VL diagnosis were negative in 11 (17%), weak/moderate in 17 (27%) and strongly-positive in 35 (36%). Twenty (32%) patients had parasitologically-confirmed treatment failure, with a risk of failure of 9% (1/11) with KAtex-negative results, 0% (0/17) for KAtex 1+/2+ and 54% (19/35) for KAtex 3+ results. Compared to KAtex-negative patients, KAtex 3+ patients were at increased risk of treatment failure [odds ratio 11.9 (95% CI 1.4–103.0); P: 0.025].

Forty-four patients were included in the analysis to predict relapse [median age: 31 years (IQR 28–35), median CD4 count: 116 cells/µL (IQR 95–181)]. When achieving VL cure, KAtex results were negative in 19 (43%), weak/moderate (1+/2+) in 10 (23%), and strongly positive (3+) in 15 patients (34%). Over the subsequent 12 months, eight out of 44 patients (18%) relapsed. The predicted 1-year relapse risk was 6% for KAtex-negative results, 14% for KAtex 1+/2+ and 42% for KAtex 3+ results [hazard ratio of 2.2 (95% CI 0.1–34.9) for KAtex 1+/2+ and 9.8 (95% CI 1.8–82.1) for KAtex 3+, compared to KAtex negative patients; P: 0.03].

Conclusion: A simple field-deployable Leishmania urine antigen test can be used for risk stratification of initial treatment failure and VL relapse in HIV-patients. A dipstick-format would facilitate field implementation.

Keywords: HIV, visceral leishmaniasis, relapse, treatment failure, antigen test, urine, prediction

# INTRODUCTION

Visceral Leishmaniasis (VL) is a vector borne protozoan disease caused by species of the Leishmania donovani complex. The parasite predominantly infects reticuloendothelial cells (van Griensven and Diro, 2012). Every year, 200,000–400,000 new VL cases are estimated to occur within approximately 70 countries. In the Mediterranean region and South America, VL is caused by L infantum. In East Africa and the Indian subcontinent, L donovani is prevalent (van Griensven and Diro, 2012).

HIV infection is one of the main risk factors for VL, and the HIV epidemic caused the re-emergence of VL in the endemic South-European countries (Desjeux and Alvar, 2003). VL/HIV coinfection is now a major problem in some low resource settings. The highest burden globally is found in North-West Ethiopia, where around 20% of VL patients are HIV co-infected (Diro et al., 2014). Management of VL/HIV patients is complicated. Besides high mortality and poor response to anti-leishmanial treatment, these patients are at high risk of VL relapse even when apparent cure is parasitologically confirmed from spleen or bone marrow aspirates (Diro et al., 2014). There are, however, only few indicators at hand to identify those at highest risk of failure or relapse, such as a history of previous VL episodes or low CD4 counts at VL diagnosis (Cota et al., 2011). Other laboratory risk factors—or biomarkers—in particular markers of Leishmania infection, have hardly been explored in resource-constrained settings.

The KAtex urine antigen test detects Leishmania antigen, which is a direct marker of infection. Its value to predict initial treatment failure has not been assessed in HIV coinfected patients. Presence of urine antigen during follow-up of HIV patients was found predictive of VL relapse in areas where L infantum is present (Riera et al., 2004), but has not been explored in L donovani endemic areas. As this test is easy to use, noninvasive, and relatively cheap, it could be particularly relevant for resource-constrained settings to help identify those HIV patients at higher risk of treatment failure or VL relapse who might hence benefit from more potent or longer treatment and close clinical follow-up after treatment.

Nested within a clinical trial on secondary prophylaxis, we conducted an exploratory study to assess whether (1) the level of Leishmania antigenuria measured at the time of diagnosis was associated with initial treatment failure and (2) the level of Leishmania antigenuria measured with the KAtex assay at the time of parasitologically confirmed cure—end of treatment—was associated with subsequent relapse in VL/HIV co-infected patients.

# METHODS

This laboratory study was nested within a clinical trial conducted between 2011 and 2015 in two VL treatment sites in North-West Ethiopia (Diro et al., 2015, 2017). We obtained approval of the trial protocol from the Ethiopian regulatory authority, the National Research Ethics Review Committee, the University of Gondar Institutional Review Board (IRB), the Ethics Review Board of Médecins sans Frontières, the IRB of the Institute of Tropical Medicine, Antwerp and the Ethics Committee of Antwerp University Hospital. All participants provided written informed consent. The protocol was registered at Clinicaltrials.gov (code NCT01360762).

The main objective of the trial was to determine the effectiveness, safety and feasibility of monthly administration of pentamidine as secondary prophylaxis to prevent VL relapse in 74 HIV-coinfected patients that had achieved microscopically confirmed parasitological cure. Secondary prophylaxis consisted of intravenous injections of 4 mg/kg of pentamidine isethionate (provided by Sanofi-Aventis) for a minimal period of 1 year. The main analysis focused on patient outcomes at 12 months after VL cure, and has been reported before. Pentamidine was found to be safe, effective and feasible to implement in resourceconstrained settings (Diro et al., 2015). In this nested study, all patients recruited in the clinical trial with (1) KAtex test results at VL diagnosis and a test of cure result after the initial treatment (objective 1) or (2) KAtex results at the end of treatment (VL cure) and follow-up for relapse in the trial were included (objective 2). KAtex results were missing for some (11 missing for objective 1; 16 for objective 2), see **Figure 1**.

The KAtex test was done at the time of VL diagnosis and achieving VL cure, and was performed as per recommendations of the manufacturer (Kalon Biological Ltd., Guilford England). The KAtex urine assay is semi-quantitative, with three levels of agglutination: 1+: weakly positive; 2+: moderately positive; 3+: strongly positive. Agglutination of any degree visible to the naked eye was considered positive, whilst no agglutination was taken as negative. Urine samples were aliquoted within 3 h after collection and kept at −20◦C, as KAtex tests were performed in batch on stored samples. The diagnosis of VL relied on parasite detection in tissue aspirates (spleen, bone marrow, or lymph node), and grading was done as reported before (WHO, 2010). Grading

of the parasite load was as follows: 6+: >100 parasites/field; 5+: 10–100 parasites/field; 4+: 1–10 parasites/field; 3+: 1– 10 parasites/10 fields; 2+: 1–10 parasites/100 fields; 1+: 1– 10 parasites/1,000 fields; 0: no parasites/1,000 fields. Initial parasitological treatment failure was defined as the presence of tissue parasites at the end of the initial treatment. Initial cure was defined as the absence of tissue parasites at the end of the initial treatment, combined with clinical improvement. Treatment consisted of AmBisome 30–40 mg/kg (with or without miltefosine for 28 days) or sodium stibogluconate 20 mg/kg for 30 days or sodium stibogluconate 20 mg/kg and paromomycin 15 mg/kg for 17 days, as reported before (Diro et al., 2015, 2017). If not yet an antiretroviral treatment (ART) at VL diagnosis, this was started during hospitalization.

Associations between patient characteristics and the level of KAtex antigenuria was assessed using the Kruskal-Wallis test for continuous variables and the Fisher's exact test for binary/categorical variables. We calculated the proportion with initial parasitological treatment failure, stratified by KAtex result. We used logistic regression to quantify the association between the KAtex test result and the risk of initial treatment failure, with the strength of association expressed as odds ratio's (OR) and 95% confidence intervals (95% CI). KAtex 1+ and 2+ results were pooled together due to the low sample size in each of these categories. There were thus three categories: negative (0); weakly/moderately positive (1+/2+); strongly positive (3+). The 1 year risk of relapse was determined using Kaplan-Meier methods, stratified by KAtex result, with follow-up time starting at the time of achieving VL cure and censored at the time of death, relapse or lost to follow-up, or after 12 months if none of these events had occurred. We used hazard ratio's (HR) and 95% CIs to quantify the association between the KAtex test result and the risk of relapse. To assess whether the KAtex result could be a non-invasive measure of the tissue parasite load, we assessed the association between the KAtex level (non-invasive test) and the parasite grading on tissue aspiration (invasive test). The association between the tissue parasite grading and the risk of initial treatment failure and relapse was assessed as well. All analysis was done using Stata version 14.

# RESULTS

Sixty-three coinfected adults were included in the analysis to assess the association between Leishmania antigenuria at VL diagnosis and initial treatment failure (**Figure 1**). The median age was 30 years [interquartile range (IQR) 27–35] and the median CD4 count was 56 cells/µL (IQR 38–113), see **Table 1**. KAtex results were negative in 11 (17%), weakly/moderately positive in 17 (27%) and strongly positive (3+) in 35 (36%). KAtex 3+ patients had the lowest CD4 counts, were more likely to have a history of VL at the time of diagnosis and had higher tissue parasite grades, although only the latter association with CD4 count and tissue parasite count reached statistical significance. Twenty (31.7%) patients had parasitologically confirmed initial treatment failure, with a risk of failure of 9% (1/11) with KAtex negative results, 0% (0/17) for KAtex 1+/2+ results and 54% (19/35) for KAtex 3+ result. Compared to KAtex negative patients, KAtex 3+ patients had a statistically significant increased risk of initial treatment failure [odds ratio 11.9 (95% confidence interval (CI) 1.4–103.0); P: 0.025], see **Table 2**.

For the analysis assessing the association between the KAtex result at the time of cure and the risk of subsequent relapse, 44 patients were included (**Figure 1**). The median age was 31 years (IQR 28–35); all but two were male (95%), see **Table 3**. Nineteen (43%) were enrolled after a primary VL episode; most (66%; n = 29) were on ART at the time VL was diagnosed. The median CD4 count at VL cure was 116 cells/µL (IQR 95–181). When achieving VL cure, KAtex results were negative in 19 (43%), weakly/moderately positive in 10 (23%) and strongly positive (3+) in 15 (34%). KAtex 3+ patients were more likely to have a history of VL, to be on ART at the time of VL diagnosis, and had higher tissue parasite grades at VL diagnosis, although only the latter association with tissue parasite count reached statistical



ART, antiretroviral treatment; IQR, interquartile range; VL, visceral leishmaniasis.

significance. Over the subsequent 12 months, four patients died, seven were lost to follow-up and eight patients relapsed. The predicted 1 year relapse risk was 6% for KAtex negative tests, 14% for KAtex 1+/2+ results, and 42% for KATEX 3+ results [hazard ratio of 2.2 (95% CI 0.1–34.9) for KAtex 1+/2+ and 9.8 (95% CI 1.8–82.1) for KAtex 3+, compared to KAtex negative patients; P: 0.03], see **Figure 2**, **Table 4**.

# DISCUSSION

To the best of our knowledge, this is the first study evaluating the value of Leishmania antigenuria at the time of VL diagnosis to predict increased risk of initial treatment failure in HIV coinfected patients. Higher levels of Leishmania antigen allowed to identify those at highest risk of failure. In addition, higher levels of Leishmania antigen in the urine at the time of VL cure were associated with an increased risk of subsequent VL relapse in Ethiopian VL-HIV coinfected patients receiving pentamidine secondary prophylaxis. In Europe, where L infantum is prevalent, a positive KAtex test during the post-treatment follow-up of HIV patients has been found predictive of VL relapse (Riera et al., 2004). Equally, detection of parasite DNA in the peripheral blood was also predictive of VL relapse in Europe (Antinori et al., 2007; Bourgeois et al., 2008; Molina et al., 2013; Nicodemo et al., 2013; Bhattacharyya et al., 2014; Cota et al., 2017; Verma et al., 2017). As far as we know, no such studies in HIV patients have been conducted on L dononavi.

The level of Leishmania antigen in the urine correlated with the level of parasites present in the tissue aspiration, which was also predictive of initial treatment failure. Consequently, this non-invasive test could be used for risk stratification to identify those at higher risk of failure in settings where invasive test such as tissue aspiration are not feasible (e.g., at the decentralized level).

A positive urine antigen test indicates the presence of viable or dead (degraded) parasites. In one European study, parasites could be cultured from the blood of asymptomatic HIV patients with positive urine antigen tests after VL treatment, indicating that viable parasites still remain detectable and in circulation in some patients after treatment (Riera et al., 2004). In L infantum endemic areas in France, continuous replication and circulation of the L infantum parasite was demonstrated over a period of up to 10 years, both during asymptomatic phases and symptomatic VL relapse episodes. This entity was defined as "active chronic VL" (Bourgeois et al., 2010). In our study, it was impossible to determine whether the positive urine antigen test found at the end of treatment indicates the presence of more live parasites remaining somewhere in the body, or whether it perhaps reflects a higher initial parasite burden which is still being cleared. Of interest is that out of the seven relapse patients with KAtex results available during the trial, six (86%) of them remained antigen positive while on PSP, whereas only 12 (44%) out of 27 patients who remained relapse-free were KAtex positive while on PSP (data not shown).



CI, confidence interval; HR, hazard ratio; OR, odds ratio; VL, visceral leishmaniasis.

<sup>a</sup>OR could not be calculated due to the zero value.

<sup>b</sup>The association remained statistically significant after accounting for a history of VL at CD4 count at VL diagnosis (OR 12.0; 95% CI 1.2-115.8; P 0.032).

TABLE 3 | Patient characteristics stratified by the level of Leishmania urine antigen at the end of VL treatment (VL cure), North Ethiopia (2011–2015).


ART, antiretroviral treatment; IQR, interquartile range; VL, visceral leishmaniasis.

In our study, there was a positive association between the tissue parasite load at the initial VL diagnosis and both the level of Leishmania antigen in the urine at VL diagnosis and at the end of treatment. A high tissue parasite load at VL diagnosis was also associated with relapse. Individuals with a higher initial parasite burden might have higher amounts of residual parasites at the end of treatment, which could be correlated with higher amounts of Leishmania urine antigen. On the other hand, this high tissue burden could also be an expression of a pronounced, persistent and more detrimental T-helper 2 cellular response, unable to clear the intracellular parasites.

We previously reported a fair diagnostic accuracy of KAtex for VL diagnosis in VL/HIV co-infection, with a sensitivity of 84% and a specificity of 99% (Vogt et al., 2017). Adding a serological marker could potentially further improve diagnostic accuracy. This could pave the way for a diagnostic work-up combining a serological test with a urine antigen test, with the latter containing prognostic information on who is most likely to fail initial treatment. Given the current limited availability of VL drugs, these patients could potentially be selected for the most effective treatment (e.g., combination therapy). The level of urine antigen at the time of cure could be used before discharge to identify those at highest risk of relapse, who could be targeted for closer medical follow-up or (prolonged) secondary prophylaxis. Development of a dipstick format would further increase field implementation. More sensitive urine antigen tests have recently been developed (Vallur et al., 2015), and their diagnostic and prognostic value in HIV coinfected patients should be evaluated as well.

There are several limitations to this study. As a small exploratory study, findings should be interpreted cautiously and remain to be confirmed in larger, prospective studies. Laboratory results were missing for some. Moreover, our findings on the risk of relapse come from patients undergoing secondary prophylaxis. However, a similar observation was made in an L infantum endemic area. Moreover, we do not see a biological reason why our observed association would only apply to patients on secondary prophylaxis. As most of the patients were on antiretroviral therapy at VL diagnosis, or received it early after VL diagnosis, the findings might not be generalizable to settings with limited antiretroviral treatment coverage. It would also have been of interest to collect urine samples during the post-treatment follow-up period and assess the association with subsequent relapse. One of the strengths of the study is that it was nested within a clinical trial, adhering to good clinical and laboratory practices, ensuring that data quality was high.

In conclusion, higher level of Leishmania antigen in the urine at the time of VL diagnosis and cure was associated with an increased risk of initial treatment failure and relapse in HIVpatients, respectively. A simple Leishmania urine antigen test that can be deployed in resource-limited settings can be used to tailor

19; KAtex 1+/2+: 10; n = x; KAtex 3+: n = 15.

TABLE 4 | Association between the level of Leishmania urine antigen at the time of VL cure and the risk of VL relapse over the subsequent 12 months, North Ethiopia (2011–2015).


CI, confidence interval; HR, hazard ratio; VL, visceral leishmaniasis.

<sup>a</sup>Due to the zero value in the reference category, these two categories were merged.

patient management. Development of a dipstick format would further increase field implementation. Newer urine antigen tests should be evaluated as well.

# ETHICS STATEMENT

This study was carried out in accordance with the recommendations of the Declaration of Helsinki 2013, the Good Clinical Practice of the WHO, and those of the Ethiopian Food, Medicine and HealthCare Administration and Control Authority (FMHACA) with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the Ethiopian regulatory authority, the National Research Ethics Review Committee, the University of Gondar Institutional Review Board (IRB), the Ethics Review Board of Médecins sans Frontières, the IRB of the Institute of Tropical Medicine, Antwerp and the Ethics Committee of Antwerp University Hospital.

# AUTHOR CONTRIBUTIONS

JvG and ED conceived the study. BM, TM, HF, YT, EA, RM, KR, FV, and WA contributed in data acquisition. Analysis was done by JvG. Interpretation of data was done by JvG, ED, and KR. JvG drafted the first draft of the manuscript. ED, BM, TM, HF, YT, EA, RM, KR, FV, and WA commented on the first draft of the manuscript. All authors read and approved the final manuscript.

# FUNDING

This trial was funded by the European Union Seventh Framework Program (FP7/2007-2013) under grant agreement n◦ 305178 via the AfriCoLeish project. Additional funding was provided by the Department of Economy, Science, and Innovation (EWI) of the Flemish government. ED has received a Ph.D. scholarship granted from the Belgian Directorate General for Development Cooperation under the ITM-DGDC framework agreement FA-III and from the European Union Seventh Framework Programme through the AfriCoLeish Project. The funders of the study had no role in study design, data collection, data analysis,

# REFERENCES


data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

## ACKNOWLEDGMENTS

We would like to thank Drs. Alan Pereira, Dhananjay Singh, Kolja Stille, and Ahmed Abdi who have contributed a lot to patient recruitment. The efforts of the ITM clinical trials unit were also highly appreciated. We thank the patients who volunteered for this clinical trial. We also highly appreciated the teams at University of Gondar Leishmaniasis Research and Treatment Center (LRTC) and at Abdurafi Health Center for supporting the trial. Our gratitude also goes to Sanofi-Aventis who donated the study drug, and to the Drugs for Neglected Diseases initiative (DNDi) for their support of the LRTC.


**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 van Griensven, Mengesha, Mekonnen, Fikre, Takele, Adem, Mohammed, Ritmeijer, Vogt, Adriaensen and Diro. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Analysis of the Antigenic and Prophylactic Properties of the *Leishmania* Translation Initiation Factors eIF2 and eIF2B in Natural and Experimental Leishmaniasis

Esther Garde<sup>1</sup> , Laura Ramírez <sup>1</sup> , Laura Corvo<sup>1</sup> , José C. Solana<sup>1</sup> , M. Elena Martín<sup>2</sup> , Víctor M. González <sup>2</sup> , Carlos Gómez-Nieto<sup>3</sup> , Aldina Barral <sup>4</sup> , Manoel Barral-Netto<sup>4</sup> , José M. Requena<sup>1</sup> , Salvador Iborra5,6,7 \* and Manuel Soto<sup>1</sup> \*

<sup>1</sup> Departamento de Biología Molecular, Facultad de Ciencias, Centro de Biología Molecular Severo Ochoa, Consejo Superior de Investigaciones Científicas (CSIC)-Universidad Autónoma de Madrid (UAM), Madrid, Spain, <sup>2</sup> Departamento de Bioquímica-Investigación, Hospital Ramón y Cajal, Instituto Ramón y Cajal de Investigación Sanitaria (IRYCIS), Madrid, Spain, <sup>3</sup> Parasitology Unit, LeishmanCeres Laboratory, Veterinary Faculty, University of Extremadura, Cáceres, Spain, <sup>4</sup> Centro de Pesquisas Gonçalo Moniz, Fundação Oswaldo Cruz-FIOCRUZ, Salvador, Brazil, <sup>5</sup> Immunobiology of Inflammation Laboratory, Department of Vascular Biology and Inflammation, Centro Nacional de Investigaciones Cardiovasculares, Madrid, Spain, <sup>6</sup> Department of Immunology, School of Medicine, Universidad Complutense de Madrid, Madrid, Spain, <sup>7</sup> Health Research Institute (imas12), Ciudad Universitaria, Madrid, Spain

### *Edited by:*

Eugenia Carrillo, Instituto de Salud Carlos III, Spain

### *Reviewed by:*

Anita Hilda Straus, Federal University of São Paulo, Brazil Herbert Leonel de Matos Guedes, Universidade Federal do Rio de Janeiro, Brazil

### *\*Correspondence:*

Salvador Iborra salvador.iborra@cnic.es Manuel Soto msoto@cbm.csic.es

*Received:* 08 February 2018 *Accepted:* 21 March 2018 *Published:* 05 April 2018

### *Citation:*

Garde E, Ramírez L, Corvo L, Solana JC, Martín ME, González VM, Gómez-Nieto C, Barral A, Barral-Netto M, Requena JM, Iborra S and Soto M (2018) Analysis of the Antigenic and Prophylactic Properties of the Leishmania Translation Initiation Factors eIF2 and eIF2B in Natural and Experimental Leishmaniasis. Front. Cell. Infect. Microbiol. 8:112. doi: 10.3389/fcimb.2018.00112 Different members of intracellular protein families are recognized by the immune system of the vertebrate host infected by parasites of the genus Leishmania. Here, we have analyzed the antigenic and immunogenic properties of the Leishmania eIF2 and eIF2B translation initiation factors. An in silico search in Leishmania infantum sequence databases allowed the identification of the genes encoding the α, β, and γ subunits and the α, β, and δ subunits of the putative Leishmania orthologs of the eukaryotic initiation factors F2 (LieIF2) or F2B (LieIF2B), respectively. The antigenicity of these factors was analyzed by ELISA using recombinant versions of the different subunits. Antibodies against the different LieIF2 and LieIF2B subunits were found in the sera from human and canine visceral leishmaniasis patients, and also in the sera from hamsters experimentally infected with L. infantum. In L. infantum (BALB/c) and Leishmania major (BALB/c or C57BL/6) challenged mice, a moderate humoral response against these protein factors was detected. Remarkably, these proteins elicited an IL-10 production by splenocytes derived from infected mice independently of the Leishmania species employed for experimental challenge. When DNA vaccines based on the expression of the LieIF2 or LieIF2B subunit encoding genes were administered in mice, an antigen-specific secretion of IFN-γ and IL-10 cytokines was observed. Furthermore, a partial protection against murine CL development due to L. major infection was generated in the vaccinated mice. Also, in this work we show that the LieIF2α subunit and the LieIF2Bβ and δ subunits have the capacity to stimulate IL-10 secretion by spleen cells from naïve mice. B-lymphocytes were identified as the major producers of this anti-inflammatory cytokine. Taking into account the data found in this study, it may be hypothesized that these proteins act as virulence factors implicated in the induction of humoral responses as well as in the production of the down-regulatory IL-10 cytokine, favoring a pathological outcome. Therefore, these proteins might be considered markers of disease.

Keywords: *Leishmania*, antigens, interleukin-10, visceral leishmaniasis, translation initiation factors, experimental murine models, vaccines

# INTRODUCTION

Leishmaniases comprise a complex group of diseases caused by the infection of protozoa of the genus Leishmania. These parasites multiply as intracellular amastigotes within macrophages of their vertebrate hosts and as extracellular promastigotes in the gut of the insect vector (phlebotomine sand flies) (Dostálová and Volf, 2012). The parasite species as well as the immune-competence state of the host determine disease spectrum. Cutaneous leishmaniasis (CL) is the less severe form of the disease. It is caused by infection, among other species, with Leishmania major in the Old World and Leishmania braziliensis in the New World. Visceral leishmaniasis (VL) is characterized by parasite dispersion to internal organs causing a form of the disease that results deadly if treatment is not administered (Rodrigues et al., 2016). It has been estimated that there are 20,000–40,000 deaths per year due to VL in the less protected regions of the world (Alvar et al., 2012). The parasite invades the patient internal organs causing episodes of fever, weight loss, anemia, and swelling of the spleen and the liver (Herwaldt, 1999; Torres-Guerrero et al., 2017). In the Mediterranean countries, Middle-East, Asia, and South America, VL it is caused by Leishmania infantum [synonym Leishmania chagasi (Maurício et al., 2000)]. Wild canids and domestic dogs are the major reservoir of these parasites playing a central role in the transmission to humans by phlebotomine sand flies (Palatnik-de-Sousa and Day, 2011; Esch and Petersen, 2013). The infection in dogs also causes a severe form of VL complicated with different cutaneous manifestations (CanVL) (Baneth et al., 2008; Solano-Gallego et al., 2011, 2017; Abbehusen et al., 2017). For both mammalian hosts, after infection some individuals can remain asymptomatic mainly because of the induction of Th1 cellular responses and IFN-γ mediated macrophage activation for destruction of intracellular parasites. On the other hand, the symptomatic forms of the disease are associated with the generation of IL-4 mediated humoral responses against parasite antigens and an IL-10 dependent inhibition of macrophage activation (Murray, 1997; Miles et al., 2005; Baneth et al., 2008).

Visceral leishmaniasis patients possess antibodies recognizing different parasite antigens including surface molecules, some secreted factors and different intracellular proteins belonging to evolutionary conserved families that play essential cell functions. These families comprise tubulins (Abanades et al., 2012), heat shock proteins (Quijada et al., 1996, 1998), histones (Soto et al., 1999; Maalej et al., 2003), or PUF proteins (Folgueira et al., 2010). Some of these proteins families are also antigenic in CL patients (Rafati et al., 2007; Souza et al., 2013; Duarte et al., 2015). The presence of high titers of anti-Leishmania antibodies is thought to be linked with pathology due to the adverse effects of deposition of the immune complexes in different tissues (García-Alonso et al., 1996; Jain et al., 2000). Moreover, the presence of IgG immune complexes correlates to the down regulation of IL-12 production and the secretion of IL-10 by macrophages in mice infected with L. major and in human VL patients (Miles et al., 2005) depending on the density of the IgG complexes formed (Gallo et al., 2010). In addition, most of these antigens are able to induce cellular responses in CL and VL human patients or dogs affected by CanVL (Probst et al., 2001; Rafati et al., 2007; Carrillo et al., 2008; Meddeb-Garnaoui et al., 2010; Baharia et al., 2014).

During the last few years, attention has been focused on translation initiation process in Leishmania, since it has emerged as an important point of regulation of gene expression in these parasites (Requena, 2012). In a recent work, the parasite eIF5A has been studied at the molecular level, although its function in Leishmania still remains unknown (Singh et al., 2014). Studies performed with the components of the Leishmania eIF4F complex (eIF4A, eIF4E, and eIF4G) (Yoffe et al., 2004, 2006, 2009; Pereira et al., 2013) and poly (A)-binding proteins (PABPs) (da Costa Lima et al., 2010) have demonstrated that translation initiation in Leishmania, as occurs in the rest of eukaryotes, depends on the interaction with the 5′ CAP structure and poly(A) tails present in the mRNAs. Similarly, the identification of functional domains in the different subunits of Leishmania eIF3 complex suggests a conserved mechanism in translation initiation in Leishmania (Rezende et al., 2014). Little is known about the components and functions of the heterotrimeric eIF2 factor in Leishmania. This factor has a GTP binding activity and it is implicated in the formation of the ternary complex that recruits the Met-tRNA to the small subunit of the ribosome for translation initiation. To be functional, the participation of the multi-subunit eIF2B complex is required. This guanine exchange factor is responsible for the GDP-GTP recycling of eIF2 (Hinnebusch, 2014). In Leishmania, phosphorylation of the eIF2α subunit inhibits GTP recycling inducing the down regulation of global translation occurring during promastigote to amastigote differentiation (Chow et al., 2011; Cloutier et al., 2012).

Protein factors implicated in the translation process in Leishmania also play relevant roles regarding parasite persistence in the vertebrate host. This is the case for the Leishmania αsubunit of the elongation factor-1 complex (EF-1), involved in regulating the rate and fidelity of protein translation. Remarkably, this factor is able to interact and activate the macrophage tyrosine phosphatase-1 (SHP-1) resulting in macrophage deactivation and parasite survival within this cell type (Nandan et al., 2002). The recognition of some other parasite proteins implicated in translation by the host immune system has been also demonstrated. In this regard, the parasite PABPs have been found to be antigenic proteins in canine and human VL patients and humans affected by mucocutaneous leishmaniasis (MCL) due to L. braziliensis infection (Guerra et al., 2011; Soto et al., 2015). In addition, the L. braziliensis initiation factor 5a (LbeIF5A) is an antigenic protein recognized by the sera from human patients infected with L. braziliensis (Duarte et al., 2015). To date, the eIF4A is the most studied parasite translation initiation factor because of its relation with the host immune system. This protein seems to help the host for fighting against infection, since it was described as an inductor of Th1 mediated responses after infection in human patients (Skeiky et al., 1995) and in experimental murine models of infection (Skeiky et al., 1998). Remarkably, it is also able to activate macrophages for controlling parasite replication by the induction of proinflammatory cytokines (Probst et al., 1997; Koutsoni et al., 2014). Importantly, experimental vaccines based on the PABPs (Soto et al., 2015) or LeIF4A, either alone (Skeiky et al., 1998) or fused with other parasite antigens forming a recombinant polyprotein (Coler et al., 2002, 2007; Bertholet et al., 2009), have the potential of inducing protection against the infection with different Leishmania species. Similarly, a recombinant version of the LbeIF5a factor was able to induce protection against L. infantum (Duarte et al., 2016a) and Leishmania amazonensis (Duarte et al., 2017) challenges when administered as part of a polyprotein vaccine.

With the aim of exploring the antigenicity and prophylactic properties of other Leishmania proteins involved in translation, in this work we have worked with the L. infantum F2 (LieIF2) and F2B (LieIF2B) proteins. With this purpose, the L. infantum putative orthologs of the α-, β-, and γ-subunits of the eukaryotic initiation factor eIF2 and the α-, β-, and δ subunits of the eukaryotic factor eIF2B were produced as recombinant proteins and tested in ELISA using human and canine VL sera. Given that these initiation factors were found to be antigenic in natural VL, analyses were extended to a highly susceptible experimental model of VL, namely hamsters infected with L. infantum (Requena et al., 2000b; Carrion et al., 2006). In addition, we have analyzed the humoral and cellular responses elicited against LieIF2 and LieF2B in murine models of VL (L. infantum-BALB/c) and CL: susceptible (L. major-BALB/c) or resistant (L. major-C57BL/6). Finally, DNA vaccines based on both factors were constructed and assayed in BALB/c and C57BL/6 CL mouse models.

# MATERIALS AND METHODS

# Cloning of *L. infantum* eIF2 (LieIF2) and eIF2B (LieIF2B) Factors Coding Regions

The DNA regions encoding the LieIF2α (LinJ.03.0960), LieIF2β (LinJ.08.0570), LieIF2γ (LinJ.09.1130), LieIF2Bα (LinJ.12.0001), LieIF2Bβ (LinJ.10.1030), and LieIF2Bδ (LinJ.27.1090) subunits were rescued from the L. infantum genome database (www. genedb.org) using the Saccharomyces cerevisiae (gene accession numbers: YJR007W, 0YPL237W, YER025W, YKR026C, YLR291C, YGR083, respectively) orthologous encoding gene sequences as queries. Genomic DNA from L. infantum (MCAN/ES/1996/BCN/150/MON-1) was employed as template for PCR, since no introns are present in Leishmania DNA (Requena, 2012). Primer sequences are available in the Supplementary Table 1. The resultant PCR products were cloned in the BamHI cut site of the pBluescript II SK vector (Stratagene. CA, U.S.A). DNA inserts containing the coding regions were obtained by BamHI digestion and subcloned in the same cut site of the pcDNA3.1/N-HA (pcDNA) eukaryotic expression vector (GenScript. NJ, U.S.A) for DNA vaccines, or in the same site of the pQE30 prokaryotic expression vector poly-linker (Qiagen, Hilden, Germany) for recombinant protein production.

# DNA Vaccines and Recombinant Proteins Production

For DNA vaccine, the pcDNA clones were transformed in the XL1-blue strain of Escherichia coli and grown. Plasmid DNA extraction was performed with the Endofree plasmid Giga Kit (Qiagen). The correct expression of the different DNA constructs was analyzed by western-blot using an anti-HA antibody (Sigma, MO, U.S.A) taking advantage of the presence of the HA tag in the N-terminal region of the recombinant proteins. Samples were obtained in COS-7 cells transfected with plasmid constructions as described in Soto et al. (2015) but using an anti-mouse IgG conjugated to horseradish peroxidase as secondary reagent (Nordic Immunological Laboratories, Tilburg, The Netherlands). E. coli M15 strain bacteria, transformed with the different pQE30 recombinant plasmids were employed for obtaining the Leishmania factors as recombinant proteins fused to an Nterminal His-tag. After over-expression of proteins in IPTGinduced cultures, the bacteria were centrifuged (5,000 × g) and total protein extracts were solubilized in binding buffer (BB; 20 mM Tris HCl pH 8.0, 0.5 M NaCl, 5 mM imidazole, 8 M urea, 1 mM 2-mercaptoethanol). For ELISA, solubilized total extracts were passed through a Ni-NTA affinity chromatography resin. Next, the column was washed in BB containing up to 20 mM imidazole. Finally, recombinant proteins were obtained from the column by passing BB containing 0.5 M imidazole. For final preparation, dialysis of the purified proteins with ELISA denaturant coating buffer was performed (EDCB: 3 M urea, 0.5 M NaCl, 5 mM imidazol, 1 mM 2-mercaptoethanol in 20 mM Tris HCl pH 8). In order to use the recombinant proteins as stimuli in cell culture assays, protein extracts were resuspended in BB. After binding to the Ni-NTA resin, interacting proteins were refolded on the affinity column as described (Shi et al., 1997). After, elution buffer was replaced by PBS using dialysis. To remove endotoxin traces, affinity chromatography polymyxin-agarose (Sigma) columns were employed. The Quantitative Chromogenic Limulus Amebocyte Assay QCL-1000 (BioWhittaker, MD, U.S.A.) was used to determine the presence of lipopolysaccharide (LPS). In all cases, proteins stocks presented <20 ng of LPS per mg of recombinant protein. Freeze-thaw total parasite extracts, namely SLA (soluble Leishmania antigens), were prepared as described elsewhere (Iborra et al., 2005). The Bio-Rad Protein Bradford assay (Bio-Rad, CA, U.S.A.) was employed for determination of protein concentrations.

# Mice and Parasites

Female BALB/c and C57BL/6 mice (6–8 week old) were purchased from Envigo (Barcelona, Spain). Two parasite species were employed: L. infantum (MCAN/ES/96/BCN150) and L. major [clone V1 (MHOM/IL/80(Friedlin)]. Promastigotes were grown by culturing at 26◦C in M3 medium (Gibco, BRL, Grand Island, NY, U.S.A) containing 10% heat inactivated fetal calf serum (FCS) and 200 U/ml penicillin and 100µg/ml streptomycin. L. major metacyclic promastigotes were isolated from stationary cultures by negative selection using peanut agglutinin (Vector Laboratories, Burlingame, CA, U.S.A). For VL murine models, BALB/c mice were infected intravenously (i.v.) with 5 × 10<sup>6</sup> stationary-phase promastigotes of L. infantum in the tail vein. For murine CL models, BALB/c or C57BL/6 mice were infected by an intradermal (i.d.) inoculation in both ears with 1 × 10<sup>3</sup> metacyclic L. major promastigotes. In some experiments, BALB/c mice were challenged subcutaneously (s.c.) in the footpad with 1 × 10<sup>5</sup> L. major promastigotes at the stationary-phase. All procedures were performed according to the Directive 2010/63/UE from the European Union and RD53/2103 from the Spanish Government. Procedures were approved by the next agencies: Severo Ochoa Molecular Biology Center Animal Care and Use Committee (CEEA-CBMSO 21/138), Bioethical Committee of the CSIC (under reference 100/2014), Government of the Autonomous Community of Madrid (Spain PROEX121/14).

# Sera Collections

Human anonymized sera samples were randomly selected from human VL diagnosed patients stored in a serum bank (LIP-CPqGM-FIOCRUZ) built from independent studies previously conducted in Brazil in a VL endemic area (n = 20) (Abanades et al., 2012). Canine symptomatic VL sera (n = 38) were collected in the Extremadura region of Spain (previously described in Coelho et al., 2009). The sera from healthy individuals [human (n = 10); dogs (n = 13)] were also employed for comparative analysis. A collection of sera from hamster obtained from animals infected with 10<sup>3</sup> (n = 5) or 10<sup>4</sup> (n = 4) L. infantum promastigotes by the intracardiac (i.c.) route, built from a previous published study was also employed (Requena et al., 2000b). All animals (n = 9) presented splenomegaly and detectable parasite burdens in the spleen and liver at the end of the assay. Sera were taken before challenge and from month 1 to month 11 post-challenge. Serum samples from healthy hamsters were employed as controls (n = 17). To analyze the antigenicity of the translation initiation factors, murine sera collection was obtained from BALB/c mice i.v. challenged with L. infantum (n = 8) and sera from BALB/c (n = 8) s.c. challenged with L. major in the left footpad, or C57BL/6 mice (n = 8) i.d. challenge with the same parasite species in both ears. All sera were taken at week 8 after challenge. Sera taken from the same mice before infection were employed as negative controls. Sera from the four different mammalian species present a positive reaction against SLA (Supplementary Figure 1).

# ELISA Assays

The ELISAs were performed on MaxiSorp plates (Nunc, Roskilde, Denmark). For coating, 0.2 µg per well of the recombinant proteins (in EDCB) or 1 µg per well of SLA (in PBS) were overnight incubated at 4◦C. Next, four washes with PBS plus 0.5% Tween 20 (PBS-Tw) were performed. For blocking, PBS-Tw buffer supplemented with 5% non-fat milk was employed [1 h at room temperature (RT)]. For the analyses of IgG responses, human, canine, hamster, or mouse sera were assayed for 2 h at RT (1/100 dilution in the same buffer employed for blocking). Four washes with PBS-Tw were performed before incubation with secondary antibodies conjugated to horseradish peroxidase (1/2,000 dilution in the blocking solution for 1 h at RT). The following antibodies, obtained from Nordic (Nordic BioSite, Täby, Sweden) were employed: anti-dog IgG, anti-human IgG, anti-hamster IgG. Finally, after four washes in PBS-Tw the TMB ELISA substrate solution was employed for reaction developing. After incubation for 15 min in the dark, reaction was stopped by addition of 2 N H2SO4. Optical densities (O.D.) were read at 450 nm in an ELISA microplate spectrophotometer (Bio-Rad). For murine samples, the IgG1 (BALB/c and C57BL6), IgG2a (BALB/c), and IgG2c (C57BL/6) antigen-specific titers were determined by ELISA following the same procedure except that serial dilutions (1/2 dilution factor) of the sera were performed. Anti-mouse IgG1, IgG2a, and IgG2c antibodies horseradish peroxidase conjugated were employed as secondary reagents (Nordic). Sera reactivity was determined as the reciprocal endpoint titer calculated as the inverse value of the highest serum dilution factor giving an absorbance value above that of the pre-immune sera.

# Analysis of the Cellular Immune Responses

For the analysis of cytokine responses, the spleens of the mice were individually processed by mechanical homogenization in complete RPMI medium (RPMI medium supplemented with 10% heat-inactivated FCS, 20 mM L-glutamine, 200 U/ml penicillin, and 100µg/ml streptomycin) and splenocytes passed through a cell strainer (70-µm pore size). Afterwards, cells were cultured in RPMI complete medium at 5 × 10<sup>6</sup> cells per ml at 37◦C in 5% CO<sup>2</sup> alone or stimulated with either SLA or either LieIF2α, LieIF2β, LieIF2γ, LieIF2Bα, LieIF2Bβ, or LieIF2Bδ recombinant proteins (all of them at 12µg/ml final concentration) for 72 h. In some experiments, a mix of the LieIF2 subunits or a mix of the LieIF2B subunits were employed (12µg/ml final concentration; 4µg/ml each subunit). The levels of IFN-γ, IL-10, or IL-4 were determined in culture supernatants by sandwich ELISA using monoclonal antibodies specific for mouse cytokines (capture and detection) provided in commercial kits (Pharmingen, San Diego, CA, U.S.A), following the manufacturer's instructions.

For the analysis of IL-10 production in naïve mice, spleen cultures established from BALB/c and C57BL/6 mice were cultured for 48 h in RPMI complete medium at 37◦C in 5% CO<sup>2</sup> with 12µg/ml of either recombinant proteins (LieIF2α, LieIF2β, LieIF2γ, LieIF2Bα, LieIF2Bβ, or LieIF2Bδ), LPS (0.1, 1, or 10 ng/ml) or medium alone. The presence of IL-10 was determined in culture supernatants as indicated above. To identify the cells producing IL-10, equivalent cultures were prepared and stimulated. For the last 6 h BD GolgiStop protein transport inhibitor was added to the cultures. Afterwards, cells were harvested, washed twice in PBS with 1% FCS (PBS-St) and incubated with Mouse Fc Block prior to surface staining. Cells were then stained with fluorochrome-conjugated antibodies against B220 AlexaFluor 647 or CD3 AlexaFluor 647 and isotypespecific control antibodies for 30 min on ice. Cells were washed twice with PBS-St and fixed for 20 min in Cytofix/Cytoperm. Next, cells were washed in PermWash Buffer and incubated with PE-conjugated anti mouse IL-10 for 30 min at 4◦C. Finally, cells were washed twice and data were collected and analyzed using a FACSCanto II Cytometer and FlowJo 10.0.7 software. All reagents and antibodies used in the cytometric assays were purchased from BD Biosciences (Franklin Lakes, NJ, U.S.A).

# Vaccination, Clinical Follow-Up, and Parasite Loads

Mice (BALB/c and C57BL/6) were s.c. inoculated in the right footpad three times, 2 weeks apart, with 200 µg of the following preparations: empty pcDNA3 (200 µg), a mixture of pcDNAs containing the genes encoding LieIF2 subunits (LieIF2α, LieIF2β, and LieIF2γ; 67 µg each) or a mixture of of pcDNAs containing the genes encoding LieIF2B subunits (LieIF2Bα, LieIF2Bβ, and LieIF2Bδ, 67 µg each) prepared in PBS. As an additional control, a group of mice received the same volume (30 µl) of the vaccine diluent. For both mouse strains, animals (n = 8 per group) were euthanized at the time of infection to analyze the immune response elicited by vaccination. Parasite challenge was carried out at the fourth week after vaccination s.c. in the left footpad of the BALB/c mice or i.d. in the ear dermis of both BALB/c and C57BL/6 mouse strains with the doses depicted above (section Mice and Parasites). For the analysis of the clinical signs of infection, the footpad swelling (thickness of the infected left footpad minus thickness of the right footpad) or the diameter of the ear lesions was weekly measured with a metric caliper. BALB/c mice s.c. infected in the footpad were euthanized at week 8 post-challenge (n = 8). The BALB/c mice i.d. infected in the ear dermis were euthanized at week 5 (n =5) and at week 8 (n = 5), whereas C57BL/6 mice were euthanized at week 5 (n = 5) and at week 9 post-challenge (n = 5). For parasite load determination, the popliteal lymph nodes (LNs) draining the footpad (for s.c. infected animals), the ears, the retromandibular draining LNs and the spleen of each mouse were individually taken and mechanically homogenized in Schneider's complete medium (Schneider's medium supplemented with 20% heat-inactivated FCS, 200 U/ml penicillin, and 100µg/ml streptomycin). Ear sheets were previously digested for 2 h at 37◦C in Dulbecco's modified Eagle medium containing Liberase TL enzyme (50µg/ml; Roche Diagnostics, Mannheim, Germany). Samples were serially diluted (1/3) in Schneider's complete medium and plated on triplicates in a 96-well plates. The number of viable parasites was calculated from the highest dilution at which promastigotes could be grown with up to 10 days of incubation at 26◦C (Buffet et al., 1995).

# Statistical Analysis

Statistical analyses were performed using the Graph-Pad Prism Program. A D'Agostino and Pearson test was employed to analyze the Gaussian distribution of the samples when n ≥ 8. The ELISAs cut-off values were calculated by comparison of the reactivity values from infected and healthy groups using a Receiver Operating Characteristic (ROC) analysis and defined as the lowest O.D. value with a 100% of specificity. The Mann-Whitney or the Kruskal-Wallis (followed by Dunn's multiple comparison analysis) tests were employed for analyzing two or more samples, respectively. Significant differences was indicated as <sup>∗</sup> , +, x for P**-**value lower than 0.05, ∗∗ , ++, xx for P**-**value lower than 0.01 or ∗∗∗∗ , +++, xxx for P value lower than 0.001.

# RESULTS

# Identification of *Leishmania infantum* eIF2 and eIF2B Factors

The genes encoding L. infantum eIF2 (LieIF2) and eIF2B (LieIF2B) factors were identified in the sequence database of L. infantum (www.genedb.org) by BLAST searches employing the S. cerevisiae orthologs sequences. We retrieved the entries annotated as the putative α, β, and γ subunits of the LieIF2 and α, β, and δ subunits of the LieIF2B complexes. As it is shown in **Table 1**, the high degree of conservation found for these proteins among different Leishmania spp. decreases when they are compared to their putative orthologs in yeast (the species selected for in silico search) and humans (one of the mammalian hosts of the parasite). To get additional evidence about the identity of the rescued proteins, an in silico search for conserved domains was performed (http://www.ebi.ac.uk/ interpro/). It was found that the L. infantum subunits of both complexes possess equivalent structural domains to those present in the human ones, including the GTP binding domain located at the amino-terminal region of the eIF2-γ subunit (Pain, 1996).


The percentages of identity (similarity) are shown, obtained using the "water" tool of the EMBOSS suite of bioinformatics. L. infantum and yeast gene accession numbers are included in the Materials and Methods section. L. major genes accession numbers are: LmjF.03.0980, LmjF.08.0550, LmjF.09.1070 LmjF.12.0010, LmjF.10.0950, LmjF.27.1210. Human NCBI reference sequence entries are: NP\_004085.1, NP\_003899.2, NP\_001406.1, NP\_001405.1, NP\_055054.1, NP\_001029288.1.

# Antigenicity of *Leishmania infantum* eIF2 and eIF2B Factors in Natural and Experimental VL

We produced the six indicated subunits of the LieIF2 and LieIF2B translation factors in E. coli and the resulting His-tagged recombinant proteins were purified by affinity chromatography. Firstly, we analyzed their antigenicity by ELISA using different sera collections. The recognition of sera collected from VL Brazilian human patients ranged from 85.0% for LieIF2α to 20.0% for LieIF2Bα. Data were summarized in **Table 2** and the details regarding the sera reactivity, determination of the cut-off values and the statistical analysis were included in Supplementary Figure 2A. The antigenicity of the subunits in the canine form of the disease was also demonstrated, ranging the percentages of positive samples between 73.7% for LieIF2γ and LieIF2Bβ to 57.9% for LieIF2α and LieIF2Bα (**Table 2**; Supplementary Figure 2B).

Next, we extended the studies to an experimental model of the VL disease, namely hamster experimentally infected with L. infantum. Eleven months after challenge, all the infected animals had circulating antibodies against the subunits of the LieIF2 and LieIF2B complexes (**Table 2**; Supplementary Figure 2C). Additionally, in L. infantum-infected hamsters, we studied the time course of appearance of anti-LieIF2 and anti-LieIF2B antibodies in parallel to anti-SLA along the infection. With this purpose, hamster sera samples were collected monthly after L. infantum challenge and analyzed by ELISA. A continuous increase in the reactivity against either LieIF2 or LieIF2B subunits was observed from the first month postinfection reaching the highest reactivity value at the end of the assay, following a similar kinetics to the SLA recognition (**Figure 1**).

# Immune Responses Elicited Against the LieIF2 and LieIF2B Subunits in Murine Models of VL and CL

To further analyze the antigenic properties of the LieIF2 and LieIF2B translation factors we take advantage of the high amino acid sequence identity shared between the orthologous subunits from L. infantum and L. major (**Table 1**). We studied the humoral


Human (n = 20), canine (n = 38), and hamster (n = 9) sera were assayed by ELISA against the different subunits, obtained as recombinant proteins.

and cellular responses elicited against the recombinant versions of these subunits using sera samples as well as spleen cells from BALB/c mice infected either with L. infantum (VL murine model) or with L. major (CL susceptible model). Samples from C57BL/6 mice infected with L. major were also analyzed (CL resistant model). A moderate humoral response was found against the LieIF2 and LieIF2B factors in these animals. All the subunits were recognized by the sera from infected mice with percentages ranging from 62.5 to 100% (**Table 3** and Supplementary Figures 3A–C).

As occurs for anti-SLA antibodies (Supplementary Figure 4A) the immunoglobulins recognizing the six subunits were predominantly of the IgG1 subclass in L. major infected BALB/c mice (**Figure 2A**). In C57BL/6 mice a weak IgG2c predominant response against the different subunits was observed (**Figure 2B**). A mixed IgG1/IgG2a response against all the subunits was observed in L. infantum infected BALB/c mice with the exception of LieIF2γ that was predominantly recognized by IgG1 antibodies (**Figure 2C**).

The analysis of the cytokine secretion after in vitro stimulation of spleen cells from infected mice with the recombinant subunits revealed a predominant IL-10 mediated response in all experimental murine models. For the three LieIF2 (**Figures 3A–C**) and LieIF2B (**Figures 3A–C**) subunits, the levels of IL-10 in culture supernatants were significantly higher than the other two assayed cytokines (IFN-γ and IL-4), with the exceptions of LieIF2β and LieIF2γ factors in the VL model (**Figure 3A**). The levels of IFN-γ detected in the supernatants of spleen cell cultures stimulated with the recombinant factors were always lower than IL-10 amounts (**Figures 3A–C**) contrasting with the parasite specific IFN-γ predominant response observed when the same cells were stimulated with SLA especially in VL and CL resistant models (Supplementary Figure 4B).

# Administration of the LieIF2 and LieIF2B Subunits as DNA Vaccines Induces a Partial Protection Against CL in Both Susceptible and Resistant Murine Models

Given the remarkable antigenicity of these Leishmania proteins in different hosts, we analyzed their immunogenicity when administered as genetic vaccines (injection of naked DNA eukaryotic expression plasmids). The correct expression of the different plasmid constructs was demonstrated by western blot using total protein extracts from COS cells transfected with the plasmid collection (**Figure 4A**). Proteins bands of expected sizes were recognized by an antibody specific for the HA tag located in the N-terminal of the recombinant Leishmania subunits produced in the mammalian cells (**Figure 4B**).

Two different DNA vaccines were assayed: the LieIF2 vaccine, composed by the same amounts of plasmids encoding the LieIF2α, LieIF2β, and LieIF2γ subunits, and the LieIF2B vaccine, containing the same amounts of the plasmids encoding the LieIF2Bα, LieIF2Bβ, and LieIF2Bδ factors. To characterize the immune response elicited by the vaccines, spleen cells

sera (= absorbance values of a given serum obtained from an infected hamster divided by the absorbance value of the pre-immune sera from the same animal).

TABLE 3 | Percentages of murine sera positive for Leishmania eIF2 and eIF2B subunits.


Sera (n = 8 per group) from BALB/c mice infected with L. infantum (VL) or L. major (CL susceptible) and C57BL/6 infected with L. major were assayed by ELISA against the different subunits, obtained as recombinant proteins.

from vaccinated and control mice (immunized with the saline excipient or the non-recombinant plasmid) were cultured in the presence or in the absence of a mixture of the subunits composing each one of the vaccines. Administration of the genetic vaccines either in BALB/c (**Figures 5A,B)** or in C57BL/6 mice (**Figures 6A,B**) resulted in the induction of both IFN-γ and IL-10 mediated responses against LieIF2 or LiF2B subunits. Remarkably, when cells of the two control mouse groups were stimulated with a mixture of either the LieIF2 or LieIF2B factors a specific IL-10 response was observed, although of lesser magnitude regarding vaccinated groups (**Figures 5**, **6**). This comportment will be analyzed in more detail in the last subsection of the results.

Additional immune-stimulation assays were done using individual recombinant subunits instead of mixtures to in vitro stimulate spleen cells from vaccinated mice. For BALB/c mice, antigen-specific production of both IFN-γ and IL-10 was elicited by all three LieF2 (**Figure 7A**) and LieIF2B (**Figure 7B**) subunits, since we did not observe significant differences in the levels of both cytokines for any of the subunits. However, we detected a slight tendency to generate higher levels of IL-10 by LieIF2α, LieIF2γ (**Figure 7A**), and LieIF2Bα (**Figure 7B**) subunits, whereas LieIF2β (**Figure 7A**) and LieIF2Bβ or LieIF2Bδ (**Figure 7B**) culture supernatants presented a slight predominance of IFN-γ. In splenocytes from vaccinated C57BL/6 mice, an antigen-specific production of IFN-γ and IL-10 was also observed. In this mice strain, a predominant IL-10 mediated response for all the subunits of the LieIF2 (**Figure 8A**) and LieIF2B (**Figure 8B**) factors was observed. This was especially marked when the LieIF2α (**Figure 8A**) and the LieIF2Bα (**Figure 8B**) subunits were employed as stimuli. In these cultures the level of IL-10 was statistically incremented with

regard to the other studied cytokines. All the subunits stimulated the production of very low levels of IL-4 in both mouse strains.

independent experiments.

Finally, we tested the immune-prophylactic properties of these DNA vaccines by challenging control and vaccinated mice with L. major. For BALB/c mice, we firstly tested the effects of vaccines in a model consisting in the inoculation of stationaryphase promastigotes into the footpad. We observed that the disease evolution of the LieIF2 vaccinated mice was slower than in the control groups showing a significant decrease in footpad swelling in the last 3 weeks of the assay (**Figure 9A**). In addition, LieIF2 vaccinated mice showed a significant reduction of the parasite burdens in the popliteal LN draining the infected footpad with respect the burdens found in both control groups (**Figure 9B**). Also, the parasite dispersion to internal organs, measured as parasite load in the spleen, was very low in the LieIF2 vaccinated mice in relation to that found for the rest of the groups (**Figure 9B**). No significant differences in disease evolution or parasite burdens were observed between mice of the LieIF2B group and mice from both control groups (**Figures 9A,B**).

We also analyzed the evolution of cutaneous lesions in control and vaccinated animals after a challenge with highly infective metacyclic promastigotes in the ear dermis of mice from the BALB/c and C57BL/6 strains. In the BALB/c mice, LieIF2 and LieIF2B vaccinated animals showed a slower evolution of the lesions than the control groups that was more marked in the LieIF2 group (**Figure 9C**). However, in any case, lesion development was not contained and at the end of the study the

IL-4 cytokines in culture supernatants was measured by sandwich ELISA. The results are represented as Whisker (min to max) plots. Asterisks mark the statistical differences among the three assayed cytokines (Kruskal–Wallis test). Results in each panel are representative of two independent experiments.

size of the lesions was similar in mice from vaccinated or control groups. Local parasite loads mirrored lesion development. At week 5 post-challenge, LieIF2 and LieIF2B vaccinated animals showed a significant reduction of the parasite burdens in the ears with respect to both control groups (**Figure 9D**). Regarding the LN cells, only LieIF2 vaccinated mice showed lower number of parasites than control groups (**Figure 9D**). However, as occurred with the lesion size, at the end of the assay (8th week after challenge) similar parasite loads were observed at the ears and retromandibular LNs (**Figure 9D**). In addition, LieIF2 and LieIF2B vaccinated animals present a remarkable capacity to control Leishmania dispersion to internal organs as revealed by the low number of parasites detected in their spleens (**Figure 9D**). A partially protective effect of vaccination was also observed in vaccinated C57BL/6 mice. Both LieIF2 and LieIF2B vaccinated mice developed lower ear lesions after L. major inoculation than mice of control groups (**Figure 9E**). Control in lesion development was correlated to the presence of lower number of parasites in the ear and retromandibular LNs 5 weeks after challenge (**Figure 9F**, 5th week). At week 9 postchallenge animals from the control groups still showed higher parasite numbers than vaccinated animals (**Figure 9F**, 9th week).

LieIF2 and LieIF2B subunits. Untransfected COS7 (lane 1) as well as cells transfected with pcDNA-HA (lane 2), pcDNA-HA-LieIF2α (lane 3), pcDNA-HA-LieIF2β (lane 4), pcDNA-HA-LieIF2γ (lane 5), pcDNA-HA-LieIF2Bα (lane 6), pcDNA-HA-LieIF2Bβ (lane 7), and pcDNA-HA-LieIF2Bδ (lane 8) for 72 h cells were harvested, lysed and separated on a 10% SDS-PAGE gel and stained with Coomassie blue (A). A western blot performed with an equivalent gel was incubated sequentially with an anti-HA antibody made in mouse and an anti-mouse IgG antibody labeled with horseradish peroxidase and revealed with 4-chloro-1-naphthol (B).

# Analysis of the Implication of LieIF2 and LieIF2B in the Induction of IL-10 Mediated Responses in Naïve Mice

As it is shown in **Figures 5**, **6**, stimulation of spleen cells from either BALB/c or C57BL/6 control mice, receiving saline (or empty pcDNA) in the immunization schedule, with a mixture of LieIF2 (**Figures 5A**, **6A**) or a mixture of LieIF2B (**Figures 5B**, **6B**) subunits induced the production of moderate levels of IL-10. To obtain more details about the subunits(s) responsible of this effect, we processed the spleens from naïve mice (n = 6) and the splenocytes were grown in the absence or in the individual presence of the six subunits, monitoring the levels of IL-10 cytokine in the culture supernatants. The highest levels of IL-10

vaccination in BALB/c mice. Three groups of BALB/c mice (n = 8 per group) were inoculated with PBS (Saline group), with the pcDNA non-recombinant vector (pcDNA group) of with a mixture of the pcDNA-LieIF2α + pcDNA-LiF2β + pcDNA-LieIF2γ plasmids (LieIF2 group) or alternatively, with a mixture of the pcDNA-LieIF2Bα + pcDNA-LiF2Bβ + pcDNA-LieIF2δ plasmids (LieIF2B group) three times, 2 weeks apart. Four weeks after the last dose, spleen cells from mice of the saline, pcDNA and LieIF2 or LieIF2B vaccinated groups were extracted and cultured in the presence or in the absence (Med), and a mixture of the LieIF2 subunits (LieIF2α, LieIF2β, and LieIF2γ) for LieIF2 vaccinated mice (A) or a mixture of the LieIF2B subunits (LieIF2Bα, LieIF2Bβ, and LieIF2δ) for LieIF2B vaccinated animals (B). Graphs show the level of the indicated cytokines in the cell culture supernatant determined by sandwich ELISA. Results are presented as Whisker (min to max) plots. Asterisks show the statistical differences between the level of cytokines in the supernatants of the stimulated and the non -stimulated cultures (Mann–Whitney test), whereas the <sup>+</sup> or the <sup>X</sup> symbols show the statistical differences among saline and vaccinated mice or pcDNA and vaccinated mice, respectively (Kruskal–Wallis test).

were found after stimulation with the LieIF2α subunit, and the three LieIF2B subunits in both BALB/c (**Figure 10A**) or C57BL/6 (**Figure 10B**) mice strains.

In a preliminary assay designed to analyze which cells were responsible for the production of IL-10, parallel cultures stimulated with these subunits, were analyzed by FACS to study the percentages of IL-10<sup>+</sup> B cells (B220+) or IL-10<sup>+</sup> T cells (CD3+). As it is shown in the **Figure 10C**, cultures stimulated with the four subunits showed higher percentages of B220+IL-10<sup>+</sup> cells than the control unstimulated cultures, although

FIGURE 6 | Cytokine response induced by LieIF2 and LieIF2B-based genetic vaccination in C57BL/6 mice. Three groups of C57BL/6 mice (n = 8 per group) were inoculated with PBS (Saline group), with de pcDNA non-recombinant vector (pcDNA group) of with a mixture of the pcDNA-LieIF2α + pcDNA-LiF2β + pcDNA-LieIF2γ plasmids (LieIF2 group) or alternatively, with a mixture of the pcDNA-LieIF2Bα + pcDNA-LiF2Bβ + pcDNA-LieIF2δ plasmids (LieIF2B group) three times, 2 weeks apart. Four weeks after the last dose, spleen cells from mice of the saline, pcDNA and LieIF2 or LieIF2B vaccinated groups were extracted and cultured in the presence, or in the absence (Med), of a mixture of the LieIF2 subunits (LieIF2α, LieIF2β, and LieIF2γ) for LieIF2 vaccinated mice (A) or a mixture of the LieIF2B subunits (LieIF2Bα, LieIF2Bβ, and LieIF2δ) for LieIF2B vaccinated animals (B). Graphs show the level of the indicated cytokines in the cell culture supernatant determined by sandwich ELISA. Results are presented, as Whisker (min to max) plots. Asterisks show the statistical differences between the level of cytokines in the supernatants of the stimulated and the non-stimulated cultures (Mann–Whitney test), whereas the <sup>+</sup> or the <sup>X</sup> symbols show the statistical differences among saline and vaccinated mice or pcDNA and vaccinated mice, respectively (Kruskal–Wallis test).

statistical significance was only attained in the cultures stimulated with either LieIF2Bα or LieIF2Bβ subunits in both mouse strains; **Figure 10C** shows the data and statistics and Supplementary Figure 5A shows representative dot-plots. On the other hand, no differences were found for CD3+IL10<sup>+</sup> cells between stimulated and unstimulated cultures (**Figure 10D**; Supplementary Figure 5B). As an additional control, stimulation of the splenocytes was carried out in the presence of amounts of LPS similar (0.1 ng/ml) and up to **two** orders of magnitude higher than those found in the protein preparations. In these conditions, we detected low levels of the cytokine in the supernatants (Supplementary Figure 5C). Moreover, no differences were found in the percentages of B220+IL10<sup>+</sup> cells between LPS stimulated or non-stimulated cultures (Supplementary Figure 5D).

# DISCUSSION

It is well-established that after Leishmania infection several intracellular parasite proteins interact with the immune system of the mammalian host. The existence of significant sequence divergence for many intracellular conserved protein families among Leishmania parasites and other eukaryotes is a common feature that may be due to the ancient position of Leishmania genus in the eukaryote phylogenetic tree (Sogin et al., 1986). From an immunological point of view, the existence of these differences is an important issue, since many of these protein families are antigenic in human and canine VL patients and the humoral and cellular responses are specific for parasite proteins without cross-reactivity with the host counterparts (Soto et al., 1999; Requena et al., 2000a; Maalej et al., 2003; Chenik et al., 2006). The similarity values obtained for the LieIF2 and LieIF2B subunits and their human orthologs range from 25 to 54% (**Table 1**). These values were comparable to those reported for other members of Leishmania translational machinery already characterized: L. major eIF3 factor subunits (20–25%, Rezende et al., 2014), Leishmania donovani eIF5A (45%, Singh et al., 2014) or L. major eIF4F subunits, (eIF4E, 22%; eIF4A, 56% and eIF4G, 25%, Dhalia et al., 2005). The data presented in this work demonstrate that the LieIF2 and LieIF2B are humoral markers of VL, since all the subunits were recognized by the sera of human and canine patients. The variability observed in the reactivity values found for each individual recombinant subunit in both mammalian hosts suggests the existence of individual differences in antigen recognition among human and canine patients. A similar behavior was observed in other studies performed with individual parasite antigenic proteins, assayed with sera collections obtained from patients naturally infected with the parasite in endemic areas (Soto et al., 1999; Maalej et al., 2003; Goto et al., 2009). Comparison of the percentages of positive individuals revealed important differences in the pattern of recognition between both species (**Table 2**). In humans, the LieIF2α subunit was the antigen recognized by a larger number of sera, whereas LieIF2Bβ and LieIF2γ were the most recognized by canine samples. Differences in canine and human immune responses to these and other Leishmania antigens may be reflecting differences in how the parasites interact with the immune system of both hosts (Goto et al., 2009).

On the other hand, when an experimental model of VL was employed, namely hamsters infected with L. infantum, we observed a 100% of positivity (**Table 2**). The high percentage value of antigenicity observed can be taken as an indication that the degree of individual variability in the recognition of both factors in this inbred experimental model is lower than that existing in human and dog natural populations. However, individual differences in the reactivity values of the hamster sera persisted, since data followed non-parametric distributions (Supplementary Figure 3C). Something similar occurred when

the sera samples employed in this work were assayed with other antigenic proteins such as the surface protein KMP-11 or intracellular antigens such as PUF proteins, the HSP20, HSP70, and HSP83 stress proteins, the H2A and H3 histones or the LiP2a and LiP2b acidic ribosomal proteins (Requena et al., 2000b; Montalvo-Alvarez et al., 2008; Folgueira et al., 2010). Inclusion of this experimental model in our work allowed us to make a longitudinal analysis of the humoral response appearance. Interestingly, similar profiles were observed for the anti-LieIF2 or anti-LieIF2B response and antibody generation against SLA extracts, representing the whole parasite antigenic repertoire. A continuous increase in the intensity of the response against the factors and SLA found in this work (**Figure 1**) was concomitant to the increase in the number of proteins recognized by the sera of the hamster as shown previously (Requena et al., 2000b). At the end of the study, most of the parasite proteins became antigenic, being this observation in accordance with the induction of exacerbated humoral responses during disease progression in symptomatic VL human and canine patients (Miles et al., 2005; Kumar and Nylén, 2012; Fernandez-Cotrina et al., 2013; Hasker et al., 2014). It can be concluded then that the production of antibodies against both factors occurs from the first moments of infection. This demonstrates an early encounter between both factors and the immunological system of the host and it is ruled out that its antigenicity is due to the induction of the polyclonal responses associated with the pathology of the VL (Deak et al., 2010).

The intracellular location of the translation factors does not seem to be an impediment for their antigenicity. Similarly, many proteins with nuclear and cytosolic location, such as protein related to the translational machinery, enzymes implicated in parasite metabolism, heat shock proteins or histones have been described as immunodominant antigens not only in human and canine VL (Requena et al., 2000a; Coelho et al., 2009; Soto et al., 2009, 2015; Ramírez et al., 2013; Baharia et al., 2014; Sundar and Singh, 2014; Siripattanapipong et al., 2017), but also in human

FIGURE 9 | Course of L. major infection in vaccinated mice. BALB/c (n = 18 per group) or C57BL/6 (n = 10 per group) were inoculated with saline, with pcDNA, with the LieIF2 or with the LieIF2B genetic vaccines three times 2 weeks apart. Four weeks after the last inoculation mice were challenge with L. major, following the next scheme: 5 × 10<sup>5</sup> stationary phase promastigotes in the footpad (BALB/c, n = 8 per group) or 1 × 10<sup>3</sup> metacyclic promastigotes in the dermis of both ears (BALB/c and C57BL/6, n = 10 per group). In (A) the mean ± SD of the difference of thickness between the infected and the uninfected footpad (n = 8 mouse per group) is represented. The asterisks symbolize the statistically decrease of footpad swelling of the LieIF2 vaccinated group with respect saline or pcDNA groups (Kruskal–Wallis test). In (C) (BALB/c) and (E) (C57BL/6), the ear lesion diameter (mean ± SD) is shown (n = 20 ears from week 2 to week 5 post-challenge and n = 10 ears from week 6 to the end of the assay). The statistically significant decrease of ear lesion found in LieIF2 (\* asterisks) or in LieIF2B (+ symbol) vaccinated mice with respect saline of pcDNA group is shown (Kruskal-Wallis test). In the three panels, and for simplicity, the lower value of significance found when data from vaccinated were compared to saline and pcDNA control data is indicated in the graph. The number of viable parasites in the left popliteal LNs and spleens (BALB/c mice infected in the footpad; B) , or the ears, the retromandibular LNs and the spleens in BALB/c (D) or C57BL/6 (F) mice challenged in the ear dermis were individually determined by limiting dilution at the indicated weeks post-challenge. Mean ± SD of the log10 of the parasite burdens in the complete organs is shown. Asterisks represent the significant differences between the indicated groups (Kruskal–Wallis test). In (B), biological samples from eight animals per group were employed. In (D,F), samples from five animals were employed at the indicated times post-challenge. All the samples were processed individually.

CL patients having moderate anti-Leishmania humoral responses (Ramírez et al., 2013; Souza et al., 2013; Duarte et al., 2015). Interaction of these proteins with B lymphocytes for antibody secretion may occur through complement mediated lysis of the non-metacyclic parasites causing the release of the whole internal cellular compounds (Mosser and Edelson, 1984; Ambrosio and De Messias-Reason, 2005; Moreno et al., 2010). Also, noninfective promastigotes can be lysed through the activity of neutrophil extracellular traps although metacyclic promastigotes are resistant to this innate immunity mechanism (Guimaraes-Costa et al., 2009, 2014; Hurrell et al., 2016). The release of different intracellular antigenic proteins contained in secreted

the non-stimulated cultures (Kruskal–Wallis test).

vesicles (Silverman et al., 2008; Cuervo et al., 2009; Torrecilhas et al., 2012) may be also an alternative form of presentation of those antigens to the host immune system, since Leishmania is able to release microvesicles in the mammalian host (Silverman and Reiner, 2012) and their contents have been found to induce humoral responses in murine susceptible hosts (Hernández-Chinea, 2007), canine (Lima et al., 2016), and human patients (Soares et al., 2015) as well as inhibitory signaling for dendritic cells activation (Markikou-Ouni et al., 2015; Iborra et al., 2016; von Stebut and Tenzer, 2018). Interestingly, whereas in patients with the active form of the disease they induce strong humoral responses (Requena et al., 2000a; Maalej et al., 2003; Rafati et al., 2007; Costa et al., 2012; Souza et al., 2013) as well as IL-10 mediated responses (Bottrel et al., 2001; de Carvalho et al., 2003; Antonelli et al., 2004; Carvalho et al., 2005) in asymptomatic or in cured patients these proteins usually induce Th1-like responses (Probst et al., 2001; Bourreau et al., 2003; Baharia et al., 2014; Jaiswal et al., 2014; Cecilio et al., 2017). For this reason, these proteins are considered adequate humoral and cellular markers of infection, and the immune response elicited against them can be useful employed to monitor the development of the infection and also the success of the treatments.

A limitation of this work is that we have not tested cellular samples from natural leishmaniasis patients to analyze the cellular responses elicited against both factors. However, as an alternative, we moved onto murine models of leishmaniasis in order to understand the relationships of the parasite LieIF2 and LieIF2B and the host cellular immune system. First we found that these factors are also antigenic in mice infected with L. infantum (BALB/c) or L. major (BALB/c and C57BL/6) (**Table 3**). As it is deduced from data included in the **Figure 2** and Supplementary Figure 4, the quality of the anti-LieIF2 or anti-LieIF2B humoral response was qualitatively similar to that observed for SLA, although with a lower intensity. The humoral response against the translation factors in the L. infantum infected BALB/c showed the mixed IgG1/IgG2a pattern found in this VL model against SLA, concomitant to the chronic infection in the spleen and the active destruction in the liver conducted by a parasitespecific T-cell dependent macrophage activation (Engwerda and Kaye, 2000; Garg and Dube, 2006; Loría-Cervera and Andrade-Narváez, 2014; Sacks and Melby, 2015). In the highly susceptible BALB/c-L. major model we found a predominant IgG1 antibody response against both factors, associated with the SLA-dependent Th2 response typically found in this mouse strain. In the resistant model C57BL/6-L. major, the low reactivity found against the factors was associated with the induction of IgG2c antibodies related to the generation of protective Th1-mediated responses (Sacks and Noben-Trauth, 2002; Sacks and Melby, 2015). The patterns of LieIF2- and LieIF2B-driven production of IFN-γ fits well with the different evolution of the disease in the three distinct murine models. Although in all cases production of this inflammatory cytokine is limited, there is a greater tendency for its secretion by cells from the C57BL/6 mice resistant to CL and in the BALB/c VL, where parasites are eliminated in the liver (**Figure 3**). On the other hand, the L. major infected susceptible animals were unable to produce LieIF2 and LieIF2B-dependent IFN-γ (**Figure 3**). Of note, a predominance of LieIF2 and LieIF2B factors-mediated IL-10 production was observed in the three experimental models of murine leishmaniasis regardless clinical evolution (**Figure 3**). This down-regulatory cytokine has been related to disease progression in mice models of CL (Noben-Trauth et al., 2003; Ronet et al., 2010; Buxbaum, 2015; Lee et al., 2017) or VL (Murphy et al., 2001; Faleiro et al., 2016) and in human CL or VL patients (Nylén et al., 2007; Carvalho et al., 2012, 2015; Gollob et al., 2014; Nabavi et al., 2018). Parasite proteins implicated in IL-10 production are being considered virulence factors and markers of disease. This is the case of the parasite KMP-11, a surface located protein that is implicated in the stimulation of IL-10 production by patients affected by CL and also in cultured murine macrophages infected in vitro by L. amazonensis (de Carvalho et al., 2003; de Mendonça et al., 2015). Similarly, the recombinant version of the papLe22 antigen is able to induce IL-10 secretion in human patients affected by VL (Suffia et al., 2000). In this work, we have found that the LieIF2- and LieIF2B-related IL-10 production occurred in the three murine models tested, although was higher in magnitude in the CL models (**Figure 3**). The immunological consequences of this production may be different. It has been previously reported that deficiency in IL-10 does not alter the final healing phenotype after L. major challenge in C57BL/6 mice (Schwarz et al., 2013). On the other hand, IL-10 plays a major role in CL disease evolution of BALB/c mice, as demonstrated by the healing phenotype shown by IL-10 deficient mice (Schwarz et al., 2013) or in the murine VL disease (Murphy et al., 2001). The fact that some of the subunits of the LieIF2 and LieIF2B factors are able to induce the secretion of IL-10 in spleen cells from both BALB/c or C57BL/6 naïve mice could be reinforcing the idea that these factors can be considered virulence factors. As mentioned above, the generation of early humoral responses to them is evidence of their rapid presentation to host B lymphocytes, cells that are also involved in the development of the disease throughout IL-10 production (Andreani et al., 2015; Silva-Barrios et al., 2017). The ability of the factors to induce the secretion of IL-10 in these cells would contribute to generate an anti-inflammatory environment in the infected tissues that would facilitate the progression of the parasite. A similar comportment has been postulated for the parasite cytosolic tryparedoxin, since it has been implicated in the induction of IL-10 by B cells of naïve mice (Cabral et al., 2008). The exposure of these intracellular Leishmania proteins may participate in immune-pathological processes by targeting B-cell to produce specific antibodies and leading to IL-10 secretion. The effect may be similar to the exposure to sand fly saliva factors that, by up-modulating IL-10 production enhance Leishmania infection in mice infected with cutaneous-tropic species (Norsworthy et al., 2004) or in human patients (Carvalho et al., 2012, 2015; Gollob et al., 2014). Data obtained in this work reinforce the implication of different parasite proteins in the modulation of the host immune system in order to facilitate the progress of infection. Previously reported examples are the LACK protein, a homolog of the receptor for activated C kinase in mammalian cells, which mounts an early IL-4 response after Leishmania infection (Launois et al., 1997) or some parasite secreted antigens that modulate C57BL/6 immune system toward a Th2 response (Tabatabaee et al., 2011). Also, the ribosomal protein S3a was found to be implicated in the induction of polyclonal expansion of B cells beside the inhibition of T cell proliferative responses (Cordeiro-Da-Silva et al., 2001).

Modulating the response against some of these immunologically active proteins by their administration in combination with adequate adjuvants was postulated as an interesting field of research for development of prophylactic or therapeutic vaccines (Badaro et al., 2001; Duarte et al., 2016b; Reguera et al., 2016). As a proof of concept, the inoculation of parasite ribosomal proteins combined with un-methylated CpG motives [ligands for the pro-inflammatory TLR-9 (Reed et al., 2013)] resulted in the protection against L. major infection inducing IFN-γ-mediated responses in C57BL/6 or BALB/c mice, correlated to the control of the humoral responses and IL-10 production driven by these parasite antigens in the last model (Iborra et al., 2008). In this work, as a proof of concept we tested DNA-vaccines based on both factors, taking advantage of the capacity of these genetic vaccines to induce IFN-γ mediated cellular responses specific for the proteins encoded in the plasmid vectors (Kaur et al., 2016; Kumar and Samant, 2016; Maspi et al., 2017). Our data showed that the LieIF2 and LieIF2B vaccinated mice produced IFN-γ in response to the corresponding subunits of both factors. However we also detected a LieIF2- and LieIF2B-specific production of IL-10 following immunization (**Figures 5**–**8**). In agreement with the results obtained in this work, the incapacity to control IL-4 or IL-10 production beside the induction of IFN-γ was considered as a bad marker for protection (Roberts et al., 2005) as it has been correlated with the inability to generate protective responses in BALB/c mice against L. major (Sjölander et al., 1998; Iborra et al., 2005, 2007) or L. infantum (Pirdel et al., 2014). The failure in protection showed in this work for the BALB/c CL model may be related to the necessity to control the responses mediated by IL-4 and by IL-10 besides generating IFN-γ as occurred with other tested vaccines against the parasite in CL (Gomes et al., 2012; Soto et al., 2015; Duarte et al., 2017) or VL models (Goto et al., 2011; Martins et al., 2017). On the other hand, protection in the C57BL/6-L. major model was associated to the induction of rapid IFN-γ mediated responses after infective challenge rather than the control of Th2 or IL-10 mediated responses as occurred with different vaccines based on parasite antigens or Leishmania live vaccines (Iborra et al., 2005; Kébaïer et al., 2006; Doroud et al., 2011; Peters et al., 2012; Solana et al., 2017). In this sense, the appearance of less severe lesions in C57BL/6 vaccinated mice after L. major infective challenge allows to reinforce the conclusion that both factors play a prominent role in the immune response after Leishmania infection.

We conclude that the subunits forming LieIF2 and LieIF2B factors are able to interact with the host immune system during Leishmania infection in different mammalian hosts. The induction of antibodies against the different subunits allows their classification as humoral markers of the disease. In addition, our findings related to the LieIF2 and LieIF2B production of IL-10 in mice infected with L. major also highlight the role of these factors as cellular markers of the disease and link them with the promotion of susceptibility against leishmaniasis. Since some of the LieIF2 and LieIF2B subunits are able to induce the secretion of IL-10 in B cells from naïve mice, they may be considered virulence factors implicated in the induction of early down-regulatory immune responses that may facilitate the

# REFERENCES


progression of the infection. The induction of partial protective responses in C57BL/6 by the administration of LieIF2 or LieIF2Bbased DNA vaccines opens the possibility of designing new formulations combining different subunits and adjuvants or new forms of antigen delivery to improve their prophylactic capacities.

# AUTHOR CONTRIBUTIONS

EG, SI, and MS: conceived and designed the experiments; EG, LC, JS, LR, and MS: performed the experiments; EG, LC, VG, MM, JR, SI, and MS: analyzed the data; CG-N, AB, MB-N, and JR: contributed reagents, materials, analysis tools; EG, JR, SI, and MS: wrote the manuscript. All authors read and approved the final version of the manuscript.

# FUNDING

The research made for this study was supported in Spain by grants from Ministerio de Ciencia e Innovación FISPI14/00366 and FISPI14/00366 (FEDER FUNDING) and the Fondo de Investigaciones Sanitarias (ISCIII-RETICRD16/0027/0008- FEDER). A Brazilian grant from CNPq within the Ciencia sem Fronteiras-PVE program (Ref: 300174/2014-4) is also acknowledged. SI is funded by grant SAF2015-74561-JIN (FEDER FUNDING) by the Spanish Ministerio de Economía y Competitividad. Finally, Institutional grants from the Fundación Ramón Areces and Banco de Santander to the CBMSO are also acknowledged. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2018.00112/full#supplementary-material

cells with an immunoregulatory phenotype. PLoS Negl. Trop. Dis. 9:e0003543. doi: 10.1371/journal.pntd.0003543


zoonosis: part one. Trends Parasitol. 24, 324–330. doi: 10.1016/j.pt.2008. 04.001


Leishmaniasis by elicitation of CD4<sup>+</sup> T cells. Infect. Immun. 75, 4648–4654. doi: 10.1128/IAI.00394-07


cannot be reverted by strong Th1 inducers. Clin. Exp. Immunol. 150, 375–385. doi: 10.1111/j.1365-2249.2007.03501.x


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

The handling Editor declared a past co-authorship with one of the authors JR.

Copyright © 2018 Garde, Ramírez, Corvo, Solana, Martín, González, Gómez-Nieto, Barral, Barral-Netto, Requena, Iborra and Soto. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Macrophage Activation Marker Neopterin: A Candidate Biomarker for Treatment Response and Relapse in Visceral Leishmaniasis

Anke E. Kip<sup>1</sup> , Monique Wasunna<sup>2</sup> , Fabiana Alves <sup>3</sup> , Jan H. M. Schellens 4,5 , Jos H. Beijnen1,4,5, Ahmed M. Musa<sup>6</sup> , Eltahir A. G. Khalil <sup>6</sup> and Thomas P. C. Dorlo<sup>1</sup> \*

<sup>1</sup> Department of Pharmacy & Pharmacology, Antoni van Leeuwenhoek Hospital/the Netherlands Cancer Institute, Amsterdam, Netherlands, <sup>2</sup> Drug for Neglected Diseases Initiative, Nairobi, Kenya, <sup>3</sup> Drug for Neglected Diseases Initiative, Geneva, Switzerland, <sup>4</sup> Division of Pharmacoepidemiology & Clinical Pharmacology, Faculty of Science, Utrecht Institute for Pharmaceutical Sciences, Utrecht University, Utrecht, Netherlands, <sup>5</sup> Department of Clinical Pharmacology, Antoni van Leeuwenhoek Hospital/the Netherlands Cancer Institute, Amsterdam, Netherlands, <sup>6</sup> Institute of Endemic Diseases, University of Khartoum, Khartoum, Sudan

The Leishmania parasite resides and replicates within host macrophages during visceral leishmaniasis (VL). This study aimed to evaluate neopterin, a marker of macrophage activation, as possible pharmacodynamic biomarker to monitor VL treatment response and to predict long-term clinical relapse of VL. Following informed consent, 497 plasma samples were collected from East-African VL patients receiving a 28-day miltefosine monotherapy (48 patients) or 11-day combination therapy of miltefosine and liposomal amphotericin B (L-AMB, 48 patients). Neopterin was quantified with ELISA. Values are reported as median (inter-quartile range). Baseline neopterin concentrations were elevated in all VL patients at 98.8 (63.9–135) nmol/L compared to reported levels for healthy controls (<10 nmol/L). During the first treatment week, concentrations remained stable in monotherapy patients (p = 0.807), but decreased two-fold compared to baseline in the combination therapy patients (p < 0.01). In the combination therapy arm, neopterin concentrations increased significantly 1 day after L-AMB infusion compared to baseline for cured patients [137 (98.5–197) nmol/L, p < 0.01], but not for relapsing patients [84.4 (68.9–106) nmol/L, p = 0.96]. The neopterin parameter with the highest predictive power for VL relapse was a higher than 8% neopterin concentration increase between end of treatment and day 60 follow-up (ROC AUC 0.84), with a 93% sensitivity and 65% specificity. In conclusion, the identified neopterin parameter could be a potentially useful surrogate endpoint to identify patients in clinical trials at risk of relapse earlier during follow-up, possibly in a panel of biomarkers to increase its specificity.

Keywords: neopterin, biomarker, visceral leishmaniasis, kala-azar, macrophage activation, pharmacodynamics, miltefosine, liposomal amphotericin B

# INTRODUCTION

Visceral leishmaniasis (VL) is a systemic disease caused by the Leishmania parasite. VL is affecting mostly the poorest of the poor and remains a devastating neglected tropical disease with high morbidity and mortality, with over 200,000 new cases and over 20,000 deaths annually (Alvar et al., 2012). New efficacious, affordable and safe treatments for this devastating disease are urgently needed.

### Edited by:

Brice Rotureau, Institut Pasteur, France

### Reviewed by:

Ricardo Silvestre, Instituto de Pesquisa em Ciências da Vida e da Saúde (ICVS), Portugal Sandra Marcia Muxel, Universidade de São Paulo, Brazil

## \*Correspondence:

Thomas P. C. Dorlo t.dorlo@nki.nl

Received: 27 February 2018 Accepted: 09 May 2018 Published: 01 June 2018

### Citation:

Kip AE, Wasunna M, Alves F, Schellens JHM, Beijnen JH, Musa AM, Khalil EAG and Dorlo TPC (2018) Macrophage Activation Marker Neopterin: A Candidate Biomarker for Treatment Response and Relapse in Visceral Leishmaniasis. Front. Cell. Infect. Microbiol. 8:181. doi: 10.3389/fcimb.2018.00181

Parasite recrudescence and clinical relapse occur in a relatively large proportion of VL patients (Collin et al., 2004; Sundar et al., 2012; Rijal et al., 2013). In a recently published clinical trial in Kenya and Sudan, 6–18% of patients relapsed within 6 months follow-up, depending on the treatment arm (Wasunna et al., 2016). As parasite recrudescence and clinical relapse is a longterm event, the follow-up period to determine efficacy of new VL treatment regimens is normally 6 or even 12 months. To speed up the process of assessing the efficacy of new treatment regimens, sensitive and specific early biomarkers are required to predict long-term clinical outcomes, e.g., to be used in an adaptive trial design with interim analysis. As yet, no longitudinal evaluations of pharmacodynamic markers have been performed in the evaluation of anti-leishmanial therapies (Kip et al., 2015).

The Leishmania parasite resides and replicates within macrophages. In experimental models, effective control of VL infection is related to the activation of macrophages by interferon-γ (IFN-γ) to produce free radicals that kill the intracellular Leishmania donovani parasites. L. donovani infection causes an increase in monocyte load in the infected organs (Murray et al., 1987; Cervia et al., 1993). The influx of immature macrophages is required to increase the capacity of macrophages to respond to IFN-γ (Murray et al., 1987; Cervia et al., 1993). As the macrophage biomass is increasing in active VL and subsequently decreases again when the parasitic infection is cleared, we hypothesized that a macrophage activation marker could potentially serve as a potential surrogate biomarker to monitor treatment response in VL.

The macrophage activation marker neopterin, a pteridine biosynthesized from guanosine triphosphate, is excreted by activated macrophages/monocytes and its production mirrors the activation of cellular immunity (Murr et al., 2002). The main stimulus for neopterin production is the pro-inflammatory IFN-γ released after T-lymphocyte activation (reviewed by Hamerlinck, 1999; Hoffmann et al., 2003). In theory, neopterin release would rise in VL due to macrophage activation and increase in macrophage load during active disease, and subsequently decrease with waning parasitic infection. After its synthesis, neopterin is metabolically stable and excreted via the kidneys by both glomerular filtration and tubular secretion, with a total clearance of 499 ± 79.7 mL/min (Estelberger et al., 1993).

Average European healthy control neopterin levels (±SD) are 6.78 ± 3.6 nmol/L (n = 263) and 5.34 ± 2.7 nmol/L (n = 359) for children (<18 years) and adults, respectively (Werner et al., 1987). In general, 10 nmol/L is taken as the upper limit of normal for healthy controls. Given that neopterin is released upon macrophage activation, increased neopterin levels are associated with a variety of conditions involving cellular mediated immunity, such as intracellular bacterial infections (tuberculosis, leprosy), parasites (malaria), and more (reviewed by Hamerlinck, 1999). Pre-treatment neopterin levels in VL patients were previously found to be significantly elevated compared to healthy controls with mean concentrations of 32 nmol/L in patients from the L. chagasi VL-endemic region Bahia in Brazil and 40 nmol/L in Dutch and Kenyan VL patients (Schriefer et al., 1995; Hamerlinck et al., 2000). Successful antimonial treatment significantly decreased neopterin levels to healthy control-levels in treatment responders at 30 days posttreatment, but not in refractory patients (Schriefer et al., 1995).

The aim of this study was to further evaluate the potential of neopterin as a reliable predictive biomarker for prediction of treatment outcome in VL in a larger patient population by longitudinal neopterin measurements during and up to 6 months after treatment. In addition, the objective of this study was to characterize the differences in neopterin kinetics over time between the treatment arms in this study: the miltefosine monotherapy and combination therapy of liposomal amphotericin B (L-AMB) and miltefosine.

# METHODS

# Ethics, Study Design, and Clinical Sample Collection

Neopterin concentrations were determined as part of a randomized multicentre trial (registered as NCT01067443) assessing the safety and efficacy of different VL treatments in Eastern Africa. The clinical results and pharmacokinetic analysis have been published elsewhere (Wasunna et al., 2016; Dorlo et al., 2017). The study was carried out at three VL treatment centers located in endemic areas: two in Sudan (Dooka and Kassab hospitals) and one in Kenya (Kimalel health center).

The study was approved by the national and local Ethics Committees in Kenya (Kenya Medical Research Institute) and Sudan (Institute of Endemic Diseases) prior to the start of the trial in each country (Wasunna et al., 2016). In addition, ethical approval was granted by the LSHTM's (London School of Hygiene and Tropical Medicine) Ethics Committee (#5543) and a "declaration of no objection" was issued by the Amsterdam Academic Medical Center Medical Ethics Committee. The study was explained to all subjects or parents/guardians in their own language. Written informed consent, or ascent in the case of minors, was obtained from all participants before enrollment in the study.

Eligible patients were primary VL cases with parasitological confirmation of VL, aged between 7 and 60 years, HIV negative, and without concomitant severe infection or comorbidities. Parasitological assessment was performed by microscopy on lymph node aspirates (Dooka, Kassab), spleen aspirates (Kimalel), or bone marrow samples (all sites). Samples originated from patients receiving either a 28-day 2.5 mg/kg/day miltefosine monotherapy (48 patients), or a combination treatment of one dose 10 mg/kg L-AMB on day 1 of treatment, followed by a 10-day 2.5 mg/kg/day miltefosine treatment (48 patients).

Patients that required rescue treatment during treatment or patients who had a fatal outcome before the end of treatment were indicated as "initial treatment failure." Final cure was determined at 6 months after end of treatment (day 210). Patients indicated as "relapse" were cured at the end of treatment, but received rescue treatment within 6 months after treatment due to reappearance of VL clinical signs and symptoms and parasite recrudescence confirmed by microscopy.

To decrease the invasiveness of sampling for patients, neopterin concentrations were quantified in the same samples Kip et al. Neopterin Candidate Visceral Leishmaniasis Biomarker

collected for miltefosine pharmacokinetic analysis (Wasunna et al., 2016; Dorlo et al., 2017). For this reason, baseline samples were taken on the first day of miltefosine treatment before the first miltefosine dose, which in the combination therapy was 1 day after the L-AMB infusion (study day 2). Real baseline neopterin concentrations were thus only available in the miltefosine monotherapy treatment arm, but were assumed to be equal in the combination therapy arm, since patients were randomized and were balanced with respect to baseline characteristics (Wasunna et al., 2016). Further sampling was performed on study days 4, 7, and 11 (combination therapy), or study days 3, 7, 14, and 28 (monotherapy). Both groups had two samples collected during follow-up at one (day 60) and 6 months (day 210) post-treatment. Plasma was collected from sodium heparin whole blood. Samples were stored and transported at nominally −20◦C until analysis.

# Analytical Method

Neopterin was determined in patient plasma samples with a commercially available ELISA kit (Demeditec, Kiel-Wellsee, Germany), following the manufacturer's instructions. Two calibration curves (0, 1.35, 4.0, 12.0, 37.0, 111 nmol/L) were included in every analysis together with two quality control samples in duplicate. Samples above the upper limit of quantitation were reanalyzed in a 10x dilution with a dilution buffer provided by the manufacturer. The optical density (OD) was measured at 450 nm by an Infinite <sup>R</sup> M200 Microplate Reader (Tecan, Männedorf, Switzerland). The OD values were converted to neopterin concentrations from the standard curve using a 4 parameter non-linear logistic regression model in Prism (version 6.0, GraphPad, La Jolla, CA, USA).

Incurred sample reanalysis was performed for 4% of all samples. The acceptance criterion was adapted from FDA guidelines for bioanalytical method validation, and stated that at least two-thirds of the analyzed concentrations should be within 20% deviation of the initially analyzed concentration (US Food and Drug Administration FDA, 2001).

Neopterin plasma stability at −20◦C was reported to be at least 6 months (in ELISA kit). As incurred sample reanalysis was performed >1.5 years after initial analysis for a proportion of samples, these results were used to assess the influence of long-term storage on neopterin quantification.

# Statistical Analysis

Data cleaning and interpretation was performed with R (version 3.1.2) and packages "ggplot2," "Hmisc," and "plyr." All values are reported as the median (IQR, interquartile range). In the display of results, nominal time points are depicted instead of actual time points.

Absolute neopterin concentrations and relative concentration changes over time—during and after treatment—were evaluated for their ability to reliably discriminate between cured and relapsed patients. In statistical comparisons, absolute and log-transformed data were checked for normality and equal variances. The two-sided t-test on log-transformed data was used when comparing groups, unless indicated otherwise.

Subsequently, a logistic regression was performed in R to evaluate the significance of the evaluated neopterin parameter as a predictor of clinical outcome. Finally, receiver-operating characteristic (ROC) curves were generated with the R package "pROC." The interplay between sensitivity and specificity of neopterin as biomarker in isolation was interpreted and the optimal cut-off value was determined with the same package.

# RESULTS

# Patient Population, Samples, and Quality Control

A total of 497 plasma samples were available from 96 patients; 48 patients in combination therapy and 48 patients in monotherapy. In both treatment arms, two patients experienced initial treatment failure. Six patients in the combination therapy arm and nine patients in the monotherapy arm relapsed within 6 months after treatment. Of patients experiencing treatment failure, samples were only included up to the day they received rescue treatment; subsequent samples were omitted (n = 4).

Patient characteristics are depicted in **Table 1**. Age distribution and gender ratio were comparable between the two treatment arms. Relapsing patients (n = 15) received rescue treatment at day 112 (median, range day 63–217), approximately 3 months after treatment.

During treatment, all actual sample collection time points were within ±15% of the nominal time points. During follow-up, the variability in actual sample collection time points was larger, with day 60 samples collected between day 54–157 and day 210 samples between day 185–345. Nonetheless, >85% of samples were collected within ±15% of the nominal time point during follow-up.

For all runs, quality control samples were within the acceptable range according to ELISA kit specifications. Incurred sample reanalysis was found to be acceptable (>95% of reanalyzed samples were within ±20% deviation of original concentration). Incurred sample reanalysis was also acceptable for the subset (n = 12) of samples analyzed >1.5 years after initial analysis (11 out of 12 within ±20% deviation). This indicates adequate stability of neopterin in plasma for at least 1.5 years when stored at −20◦C.

# Baseline Neopterin Concentrations

Baseline neopterin concentrations were elevated in all monotherapy VL patients (n = 46) at 98.8 nmol/L (IQR 63.9–135) (**Figure 1**). There was a non-significant (p = 0.448) trend toward higher neopterin baseline levels in monotherapy patients cured at the end of treatment (104 nmol/L, IQR 64.9– 154) compared to patients requiring rescue therapy during or within 6 months after treatment (75.7 nmol/L, IQR 65.4–102). There were no significant differences in baseline neopterin levels between age categories (adult/child), country and gender (p = 0.955, p = 0.620, p = 0.737, respectively).

# Neopterin Kinetics Over Time

Neopterin concentrations regressed during treatment to comparable end of treatment values of 33.6 nmol/L (IQR

### TABLE 1 | Demographics of patients included in neopterin analysis.


All values are given as median (range), unless stated otherwise.

<sup>a</sup>Fisher exact test.

<sup>b</sup>Wilcoxon u-test.

<sup>c</sup>Chi-square test.

21.3–52.0, combination therapy, day 11) and 21.9 nmol/L (IQR 16.3–40.0, monotherapy, day 28). There was, however, a difference in the rate of neopterin decline between the two treatment arms (**Figure 2**). Neopterin concentrations decreased two-fold compared to baseline within the first seven treatment days in the combination therapy arm to 55.1 nmol/L (IQR 37.2–83.2, p < 0.01), while neopterin levels remained unchanged in patients receiving monotherapy with a concentration of 91.3

combination therapy of L-AMB and miltefosine (solid line) or miltefosine monotherapy (dashed line). Error bars represent the inter-quartile range (IQR).

nmol/L (IQR 65.9–158, p = 0.807, Mann-Whitney U-test). Interestingly, for both treatment arms, day 210 neopterin concentrations were still elevated (15.5 nmol/L IQR 10.5–22.3, combination therapy, 13.5 nmol/L IQR 11.4–22.9, monotherapy) compared to the reported healthy control levels of <10 nmol/L.

# Predictive Value of Absolute Neopterin Levels for Treatment Outcome

One day after L-AMB infusion and before the first miltefosine dose, neopterin concentrations in the combination therapy arm were significantly higher (137 nmol/L, IQR 98.5–197, **Table 2**) in cured patients, compared to the baseline concentration of


TABLE 2 | Median neopterin concentration split per treatment arm and treatment outcome.

<sup>a</sup>Two-sample t-test on log-transformed neopterin concentrations.

<sup>b</sup>Wilcoxon U-test on absolute neopterin concentrations.

<sup>c</sup>Welch Two-sample t-test on log-transformed neopterin concentrations with unequal variance.

\*p < 0.05.

\*\*End of treatment.

98.8 nmol/L in the monotherapy arm (p < 0.01). This was not observed for combination therapy patients who eventually failed treatment or relapsed (84.4 nmol/L, IQR 68.9–106, p = 0.96). Despite the significantly higher day 2 neopterin concentrations in cured patients, this parameter is not a significant predictor of final cure (p = 0.0853). The ROC AUC (**Figure 3**) was 0.74 (CI 0.56–0.92) with a low sensitivity (62%, optimal cut-off value of 122 nmol/L).

In the combination treatment arm, patients that relapsed had a significantly higher day 60 neopterin concentration compared to cured patients (54.0 vs. 26.3 nmol/L, p < 0.05, **Table 2**). The same trend was observed for the monotherapy treatment arm, though not significant (**Table 2**).

The day 60 neopterin concentration was a significant predictor of relapse in both arms combined and in the combination therapy arm (p < 0.05), but not for the monotherapy arm alone. ROC curves of these parameters are depicted in **Figure 3**. In the combination therapy arm, the day 60 neopterin concentration was the best predictor of relapse with an AUC of 0.82 (CI 0.68–0.96) and optimal threshold value of 39.7 nmol/L with corresponding sensitivity of 100% and specificity of 75%. In the monotherapy arm, the absolute neopterin concentration at day 60 was a less reliable predictor of relapse (AUC 0.65), with a sensitivity of only 56% (at optimal cut-off 40.2 nmol/L, 81% specificity).

# Predictive Value of Relative Neopterin Levels for Treatment Outcome

An increase in neopterin concentrations was observed for relapsing patients between end of treatment and day 60 (**Table 2**), but not for cured patients. The D60/EoT neopterin concentration ratio (or D60/EoT ratio)—a patient's neopterin concentration on day 60 relative to the end of treatment (EoT) concentration—could be calculated for 80 patients and is shown in **Figure 4**. Relapsing patients (n = 14) experienced a significantly higher neopterin concentration increase during the first month of follow-up (D60/EoT ratio: 2.2, IQR 1.5–2.8) compared to patients that remained cured (D60/EoT ratio: 0.78, IQR 0.53–1.4) (p < 0.001, Welch ttest on log**-**transformed data). For patients that relapsed, there was no correlation between the D60/EoT ratio and the day they received rescue treatment (linear regression R <sup>2</sup> = −0.009).

The D60/EoT neopterin concentration ratio was a significant predictor of relapse for both arms combined (p < 0.001), the monotherapy (p < 0.01), and combination therapy (p < 0.05) separately. ROC curves are depicted in **Figure 5**. With a cut-off of 1.08, the D60/EoT ratio neopterin parameter had a sensitivity of 93% and specificity of 65% in predicting relapse (ROC AUC 0.84).

# DISCUSSION

This study is the first longitudinal exploration of the kinetics of neopterin in VL patients, before, during and after treatment, to identify a predictive host-related biomarker for the long-term treatment outcome and relapse in VL.

We identified several neopterin parameters that could potentially be used as early predictors of clinical relapse, evaluated based on their ROC AUC. As a general rule, AUCs

FIGURE 3 | Receiver operator characteristic (ROC) curves of absolute neopterin concentrations as predictors of clinical relapse. Combination therapy is indicated as "COMBI," monotherapy as "MONO" and data of the two arms combined as "BOTH." AUC represents the integrated area under the ROC curve. CI refers to the confidence interval of the calculated AUC. Note that day 1 (D1) neopterin concentrations are evaluated as predictor of cure (cure = 1, relapse = 0) and day 60 (D60) neopterin concentrations are evaluated as predictor of relapse (relapse = 1, cure = 0).

FIGURE 4 | D60/EoT neopterin concentration ratio for cured and relapsed patients. D60/EoT neopterin concentration ratio, for cured patients (n = 66) and patients that relapsed after treatment (n = 14). The dashed line indicates no difference within 1 month after end of treatment (combination therapy: day 11, monotherapy: day 28). Dots indicate individual observations, the horizontal lines the median per group. \*\*p < 0.001, Welch t-test on log-transformed data.

of 0.7–0.8 are considered acceptable, 0.8–0.9 excellent and >0.9 outstanding (Hosmer and Lemeshow, 2000). The D60/EoT neopterin concentration ratio was found to be a significant and

highly sensitive predictor of relapse at a cut-off of 1.08 (AUC 0.84).

As a subset of cured patients also demonstrated an increase in neopterin concentration after end of treatment, specificity was relatively low. No clinical explanation could be identified for the neopterin concentration increase on day 60 in these cured patients: there were no consistent trends in fever, hematological or clinical chemistry parameters, nor were there more co-infections or concomittant medications reported in these patients. An additional limitation of the study was the relatively small group of relapses (n = 14), which might have impeded statistical power to detect predictors for relapse.

Although increased neopterin concentrations have also been observed for other infectious diseases (Murr et al., 2002), the observed baseline neopterin concentration in this study was substantially higher than observed in other diseases, such as HIV (17–50 nmol/L) (Fuchs et al., 1990), tuberculosis (21–37 nmol/L) (Hosp et al., 1997; Cesur et al., 2014; Skogmar et al., 2015), and malaria (21–58 nmol/L) (Thuma et al., 1996; Biemba et al., 2000). It should be noted that the observed baseline concentrations in this study were higher than previously reported for VL patients (Schriefer et al., 1995; Hamerlinck et al., 2000), possibly due to different causative Leishmania subspecies or severity of disease. D60/EoT ratio might be more prone to specificity issues in co-infection, since concentrations are lower at those time points.

Depending on the purpose of use, the minimally acceptable characteristics of pharmacodynamic biomarkers concerning specificity will differ. Requirements are less strict in a clinical trial setting, as concomitant disease is often an exclusion criteria. Another solution for lack of specificity could be to use a panel of biomarkers. In routine clinical care, the implementation of the D60/EoT ratio and corresponding sampling point 1 month after treatment, is probably problematic due to the remote and/or resource-poor settings.

An advantage of neopterin as a pharmacodynamic biomarker is the relatively low cost of analysis at around 3 euro per clinical sample for a commercial kit. Nevertheless, a basic laboratory infrastructure is required, which is not always available in health centers in the resource-limited regions where VL is being treated. A simple dipstick assay is available for the semi-quantitative detection of neopterin in serum and has also been tested in VL patients (Bührer-Sekula et al., 2000), though this assay is possibly not sensitive enough to detect the relatively subtle concentration changes after treatment. Easier, cheaper and less invasive neopterin analytical methods have been developed in dried blood spots and urine, but these have not yet been evaluated in VL patients (Zurflüh et al., 2005; Svoboda et al., 2008; Opladen et al., 2011).

This study also explored differences in neopterin kinetics between treatment regimens in patients treated with either miltefosine monotherapy or miltefosine in combination with L-AMB. This longitudinal analysis revealed a different neopterin kinetic profile for the two treatment arms, possibly implying a difference in the elicited immune reaction. The initial surge in neopterin levels within 1 day after L-AMB infusion in cured patients could suggest a beneficial effect of early activation of the pro-inflammatory Th1 response initiated by the L-AMB infusion. A significant similar rise in pro-inflammatory cytokines was also observed in mice with Aspergillus flavus infection treated with L-AMB (Olson et al., 2012). Further clinical research is needed to confirm these findings and further investigate the underlying mechanisms, but one possible explanation could be that L-AMB positively reinforces and amplifies already persisting immune reactions. The stable neopterin concentrations in the first week of miltefosine monotherapy correlate with the continuous slow accumulation of the drug and potentially less effective exposure in the early phase of treatment.

In both treatment arms, neopterin concentrations were still elevated 6 months post-treatment in comparison to the reported healthy control value of <10 nmol/L. Unfortunately, healthy endemic control levels were not available in this study. No studies could be identified that investigate endemic healthy control levels in Kenya and Sudan, but a recent study in Ethiopia found a healthy control level of 3.8 nmol/L (IQR, 1.6–5.5 nmol/L) (Skogmar et al., 2015), which is in line with established average healthy control levels in European adults and children of 6.78 and 5.34 nmol/L for adults and children, respectively (Werner et al., 1987). Lingering immune activation could be a potential explanation for our observation, as was previously observed for HIV patients treated for 3–13 months with zidovudine or didanosine: neopterin concentrations remained elevated at approximately 19 nmol/L (Gisslen et al., 1997).

Currently there are no biomarkers either to identify treated VL patients at risk of relapse, or to establish final cure during the follow-up in clinical trials. This lack of early markers or test of cure is impeding the development of new antileishmanial treatment regimens. This study is the first evaluation of neopterin as a predictor of relapse in VL patients.

In conclusion, the identified 1.08 D60/EoT ratio cut-off—an >8% neopterin concentration increase between end of treatment and day 60—could serve as a surrogate endpoint identifying the patients in clinical trials who have an increased risk of relapse. Identified at-risk patients could be more intensively followed up in clinical trials, possibly using qPCR to quantify parasite loads in the blood and/or tissue to enable early detection of parasite recrudescence. The use of this neopterin parameter as a predictive biomarker for relapse in VL should be formally evaluated in a prospective trial, possibly in a panel of biomarkers to increase specificity.

# AUTHOR CONTRIBUTIONS

MW, AM, EK, FA, and TD were responsible for the clinical study conception and design. MW, AM, and EK were responsible for acquisition of the clinical data and samples. JB and JS provided the laboratory setting to perform the analysis. AK performed the laboratory analysis. AK and TD performed the data analysis and interpretation of the results. AK took the lead in writing the manuscript. All authors provided critical feedback and helped shape the research, analysis, and manuscript.

# FUNDING

This work was supported through DNDi by the Médecins Sans Frontières International; the Medicor Foundation; Department for International Development (DFID), UK; the Dutch Ministry of Foreign Affairs (DGIS), the Netherlands; Federal Ministry of Education and Research (BMBF) through KfW, Germany; Swiss Agency for Development and Cooperation (DDC-SDC), Switzerland; and other private foundations. TD was personally supported by the Netherlands Organization for Scientific Research (ZonMw/NWO), project number 91617140.

# ACKNOWLEDGMENTS

We express our gratitude to the visceral leishmaniasis patients and the parents/guardians of pediatric patients for their willingness to be part of this clinical trial. We recognize the clinical and laboratory staff of the clinical sites of Dooka, Kassab, and Kimalel for their support. We acknowledge the professional assistance we received from the DNDi Africa data centre. This clinical trial was organized by and sponsored through DNDi and was conducted within the Leishmaniasis East Africa Platform (LEAP).

# REFERENCES


and interferon-gamma, tissue immune reaction, and response to treatment with interleukin 2 and interferon-gamma. J. Immunol. 138, 2290–2297.


**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 Kip, Wasunna, Alves, Schellens, Beijnen, Musa, Khalil and Dorlo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Transcriptional Profiling of Immune-Related Genes in *Leishmania infantum*-Infected Mice: Identification of Potential Biomarkers of Infection and Progression of Disease

Eduardo Ontoria<sup>1</sup> , Yasmina E. Hernández-Santana1†, Ana C. González-García<sup>1</sup> , Manuel C. López <sup>2</sup> , Basilio Valladares <sup>1</sup> and Emma Carmelo<sup>1</sup> \*

### *Edited by:*

Javier Moreno, Instituto de Salud Carlos III, Spain

### *Reviewed by:*

Lucile Maria Floeter-Winter, Universidade de São Paulo, Brazil Herbert Leonel de Matos Guedes, Universidade Federal do Rio de Janeiro, Brazil

### *\*Correspondence:*

Emma Carmelo ecarmelo@ull.edu.es

### *†Present Address:*

Yasmina E. Hernández-Santana, National Children's Research Centre, Our Lady's Children's Hospital Crumlin, Dublin, Ireland; Department of Clinical Medicine, School of Medicine, Trinity College Dublin, Dublin, Ireland

> *Received:* 23 March 2018 *Accepted:* 28 May 2018 *Published:* 26 June 2018

### *Citation:*

Ontoria E, Hernández-Santana YE, González-García AC, López MC, Valladares B and Carmelo E (2018) Transcriptional Profiling of Immune-Related Genes in Leishmania infantum-Infected Mice: Identification of Potential Biomarkers of Infection and Progression of Disease. Front. Cell. Infect. Microbiol. 8:197. doi: 10.3389/fcimb.2018.00197 1 Instituto Universitario de Enfermedades Tropicales y Salud Pública de Canarias, Universidad de La Laguna, La Laguna, Spain, <sup>2</sup> Departamento de Biología Molecular, Instituto de Parasitología y Biomedicina "López Neyra", Consejo Superior de Investigaciones Científicas, Granada, Spain

Leishmania spp. is a protozoan parasite that affects millions of people around the world. At present, there is no effective vaccine to prevent leishmaniases in humans. A major limitation in vaccine development is the lack of precise understanding of the particular immunological mechanisms that allow parasite survival in the host. The parasite-host cell interaction induces dramatic changes in transcriptome patterns in both organisms, therefore, a detailed analysis of gene expression in infected tissues will contribute to the evaluation of drug and vaccine candidates, the identification of potential biomarkers, and the understanding of the immunological pathways that lead to protection or progression of disease. In this large-scale analysis, differential expression of 112 immune-related genes has been analyzed using high-throughput qPCR in spleens of infected and naïve Balb/c mice at four different time points. This analysis revealed that early response against Leishmania infection is characterized by the upregulation of Th1 markers and M1-macrophage activation molecules such as Ifng, Stat1, Cxcl9, Cxcl10, Ccr5, Cxcr3, Xcl1, and Ccl3. This activation doesn't protect spleen from infection, since parasitic burden rises along time. This marked difference in gene expression between infected and control mice disappears during intermediate stages of infection, probably related to the strong anti-inflammatory and immunosuppresory signals that are activated early upon infection (Ctla4) or remain activated throughout the experiment (Il18bp). The overexpression of these Th1/M1 markers is restored later in the chronic phase (8 wpi), suggesting the generation of a classical "protective response" against leishmaniasis. Nonetheless, the parasitic burden rockets at this timepoint. This apparent contradiction can be explained by the generation of a regulatory immune response characterized by overexpression of Ifng, Tnfa, Il10, and downregulation Il4 that counteracts the Th1/M1 response. This large pool of data was also used to identify potential biomarkers of infection and parasitic burden in spleen, on the bases of two different regression models. Given the results, gene expression signature analysis appears as a useful tool to identify mechanisms involved in disease outcome and to establish a rational approach for the identification of potential biomarkers useful for monitoring disease progression, new therapies or vaccine development.

Keywords: *Leishmania infantum*, transcriptional profiling, high-throughput qPCR, immune responses, regression models, biomarkers

# INTRODUCTION

The term leishmaniasis includes a spectrum of diseases caused by parasites belonging to genus Leishmania, with symptoms ranging from cutaneous lesions to fatal visceral leishmaniosis (VL) the most severe clinical form of the disease. The organs commonly affected during VL are the bone marrow, liver, and spleen and clinical symptoms include hepatosplenomegaly, long-term, low-grade fever, muscle wasting, anemia, leukopenia, polyclonal hypergammaglobulinemia, and weight loss (reviewed in Alvar et al., 1997).

Leishmania parasites are obligate intracellular pathogens in the mammalian host and therefore a successful T celldependent immune response is required to control infection. During many years, disease outcome was thought to be driven by the Th1/Th2 paradigm of resistance/susceptibility (Heinzel et al., 1989). However, identification of new cell populations, including CD4<sup>+</sup> T cell regulatory (Treg) populations, as well as further CD4<sup>+</sup> T helper (Th) populations like Th17, Th9, and T follicular helper (Tfh) cells, have certainly questioned the simplicity of the Th1/Th2 paradigm to intracellular infection (Bettelli et al., 2006; Korn et al., 2009; Jäger and Kuchroo, 2010; Crotty, 2011; Peterson, 2012; reviewed in Alexander and Brombacher, 2012). Successful immunity against Leishmania involves a complex response of several mechanisms and factors, including the migration of appropriate cell populations to the infected sites, generation of an appropriate type of immune response, cytokine microenvironment, chemokines, and others. Chemokines and their receptors have been shown to play a crucial role in determining the outcome of leishmaniasis; indeed, pathogenesis in VL is often associated with altered chemokine expression profiles and defective migration of immune cells (Stanley and Engwerda, 2006; Oghumu et al., 2010; Kong et al., 2017).

Early after infection in mice experimental model, most of the parasites appear to be phagocytized by splenic macrophages and mature DC start producing IL-12 or IL-23 to initiate protective Th1 or Th17 responses, respectively, which, in turn, will produce IFNγ, TNF or IL-17 that maximize the capacity of infected macrophages to produce NO and ROS (reviewed in Rodrigues et al., 2016). Naïve CD8 T cells are activated by DCs in the presence of IL-12 and type I IFNs and differentiate into effector cells that further contribute to the protective response by producing IFNγ and TNF (reviewed in Rodrigues et al., 2016). Nevertheless, the parasite abrogates the ability of infected DCs to initiate protective responses using several mechanisms that impair host cell function (reviewed in Arango Duque and Descoteaux, 2015; Martínez-López et al., 2018). Some of these mechanisms include exhaustion of specific CD8 T cells (in which CTLA-4 and PD-1 play a role) (reviewed in Wherry and Kurachi, 2015) and differentiation of IFN-γ and IL-10 producing Tr1 cells. In addition, spleen suffers dramatic changes in microarchitecture, including disorganization of the white and red pulp and disruption of the marginal zone, resulting in severe immunosuppression and enhancing parasite proliferation (Kaye et al., 2004; reviewed in Rodrigues et al., 2016).

One of the major hurdles for developing vaccines to either prevent or treat VL has been a limited understanding of the precise immune mechanisms required for controlling parasite growth without causing disease. Because of the intrusive techniques required to analyze tissue in VL patients, our current understanding of the host immune response during VL largely derives from studies performed in genetically susceptible mice or hamsters infected with viscerotropic species (Faleiro et al., 2014; Kong et al., 2017; Medina-Colorado et al., 2017). These animals develop distinct, organ specific immune responses as disease progresses (Engwerda and Kaye, 2000; Rodrigues et al., 2016). In the spleen, chronic infection leads to splenomegaly and results in structural alterations in the architecture of the spleen tissue which are thought to contribute to immune suppression in this organ during VL (reviewed in Faleiro et al., 2014; Rodrigues et al., 2016).

Leishmania has developed several mechanisms that influence macrophages leishmanicidal activity, altering the expression of genes coding for cytokines, chemokines, transcription factors, membrane receptors and molecules involved in signal transduction in infected cells. Different high-throughput techniques, such as transcriptome analysis or Serial Analysis of Gene Expression (SAGE) technology (Guerfali et al., 2008) and most recently transcriptional profiling using RNA-seq (Fernandes et al., 2016; Kong et al., 2017), have been applied to studying host-parasite interactions, providing important insights into the mechanism of pathogenesis. The use of this approach has been possible due to the good correlation between cytokine/chemokine mRNAs levels and protein expression observed in these experimental models (Kumar et al., 2010, 2014; Cuervo-Escobar et al., 2014; Zhang et al., 2017). As far as we know, our study is the first performed using high-throughput qPCR which analyses more than one hundred immune-related genes at the same time in a L. infantum–infection murine model, providing a large collection of differential gene expression data between infected and non-infected animals. This approach could be a useful tool to identify the mechanisms involved in disease outcome and also to establish a rational strategy for the development of immunomodulatory therapies and vaccines.

In the last few years, there is a growing need for identification of new molecules useful for monitorization of biological processes like infection or the assessment of protective immune responses. In this context, biomarkers able to predict infection or to estimate parasitic load in infected organs and its reduction upon treatment are increasingly relevant. This high-throughput approach also provides a large collection of gene expression data than can be exploited to try to identify new molecules that can be useful as biomarkers in leishmaniasis.

The aim of this work was to map global changes in gene expression patterns in the spleen, in BALB/c murine model during Leishmania infantum infection, particularly characterizing how immune system responds to infection. The results draw a global picture of how spleen reacts, in terms of gene expression, to leishmania infection at different timepoints. In addition, new potential biomarkers for leishmaniasis are identified, and their usefulness discussed upon the current knowledge.

# MATERIALS AND METHODS

# Biological Samples

All experiments involving animals were conducted in accordance to both European (2010/63/UE) and Spanish legislation (Law 53/2013), after approval by the Committee for Research Ethics and Animal Welfare (CEIBA) of the University of La Laguna (Permission code: CEIBA2015-0168).

L. infantum (JPC strain, MCAN/ES/98/LLM-724) was maintained in vivo by serial murine passages. Prior to infection, amplification of amastigote-derived promastigotes, with less than 3 passages in vitro, was carried out by culture in RPMI medium (Gibco BRL), supplemented with 20% inactivated fetal calf serum (SBFI), 100 ug/ml streptomycin (Sigma-Aldrich, St. Louis, USA) and 100 U/ml of penicillin (Biochrom AG, Berlin, Germany) at 26◦C until reaching stationary phase. Sixty-one female wild-type BALB/c mice were obtained from the breeding facilities of the Charles River Laboratories, (France) and were maintained under specific pathogen-free conditions.

# Animal Infection and Parasite Burden Determination

Mice were randomly separated in two groups: (i) non-infected control mice (n = 23) and (ii) mice infected with 10<sup>6</sup> stationaryphase L. infantum promastigotes (n = 24) via tail vein. At week 1, 2, 4, and 8 after infection, mice (n = 6 per group) were euthanized by cervical dislocation, and spleen and liver portions of each mouse were collected and used for further parasitological and immunological assays. Mice were 14–15 weeks-old when challenged with L. infantum. Spleen samples were immediately stored in RNAlater at −80◦C (Sigma-Aldrich, St. Louis, USA) for nucleic acid preservation and further mRNA extraction. Determination of parasite burden in both liver and spleen was carried out by quantitative limiting-dilution as described by Buffet et al. (1995). Fourteen more mice of the same age and origin were included for the evaluation of the potential biomarkers. These mice were divided in two groups (7 control and 7 infected mice) and all of them were infected and euthanized following the same protocols. All efforts were made to minimize animal suffering.

# RNA Isolation and Quantification

Total RNA isolation from RNA later preserved spleens (9– 11 mg) was performed by cell disruption using FastPrep <sup>R</sup> System (ProScientific, Cedex, France) and Lysing Matrix D (MP Biomedicals, Solon, USA) in TRI-Reagent (Sigma-Aldrich, St. Louis, USA). RNeasy Mini Kit (Qiagen) was subsequently used for mRNA enrichment following manufacturer's instructions. Nucleic acid purity was assessed measuring OD260/<sup>280</sup> and OD260/<sup>230</sup> ratios using NanoDrop ND-1000 (ThermoFisher Scientific). Only samples with OD260/<sup>280</sup> ratios between 2.1 and 2.2, and OD260/<sup>230</sup> ratios between 1.8 and 2.2 were included in this study. RNA integrity number (RIN) was determined using 2100 Bioanalyzer (Agilent Technologies, Santa Clara, United States). RIN was >7 for all RNA samples included in this study.

# Reverse Transcription and High-Throughput Real-Time Quantitative PCR (RT-qPCR)

Reverse transcription and high-throughput RT-qPCR were performed using the High Capacity cDNA Reverse Transcription kit and QuantStudioTM 12K Flex Real-Time PCR System (Thermo Fisher Scientific)<sup>1</sup> according to the manufacturer's protocols, as indicated in Hernandez-Santana et al. (2016). Custom TaqMan OpenArray Real-Time PCR Plates included 112 Gene Expression Assays organized in 48 subarrays. All primers and probes were commercially designed by Thermo Fisher Scientific. The complete list of genes is shown in Table S1 of Supplemental Material. All reactions were performed in triplicate. Real-time PCR and fluorescence detection were performed using QuantStudioTM 12K Flex Real-Time PCR System (Thermo Fisher Scientific) following manufacturer's instructions, which calculates Cq values using an algorithm that takes into account the efficiency of each individual curve, called Crt method. The Crt method sets a threshold for each curve individually that is based on the shape of the amplification curve, regardless of the height or variability of the curve in its early baseline fluorescence. The method first estimates a curve that models the reaction efficiency from the amplification curve. It then uses this curve to determine the relative threshold cycle (Crt) from the amplification curve, that eliminates the need of "conventional" Real-Time PCR for calculating the efficiency of each reaction. Therefore, Cq values produced by this platform are already corrected for the efficiency of the amplification (Hernandez-Santana et al., 2016).

<sup>1</sup>Crt, a relative threshold method for qPCR data analysis on the QuantStudioTM 12K Flex system with OpenArray <sup>R</sup> technology (2014). Appl. Biosyst. QuantStudioTM 12K Flex Real-Time PCR Syst. Appl. Note CO28730, 4.

## Data Analysis and Statistics

The arithmetic average quantitative cycle (Cq) was used for data analysis. The Cq values for each qPCR run were exported from QuantStudioTM 12K Flex Real-Time PCR System, as Excel files, and imported into qBase Plus 1.3 (Biogazelle NV Zulte, Belgium) to obtain Relative Quantity (RQ) and Normalized Relative Quantity (NRQ) values from the whole data set, following manufacturer's instructions (Vandesompele et al., 2002; Hellemans et al., 2007). Two genes showed the most stable expression (Stat6 and Igb2) (geNorm stability mean M-value and mean coefficient of variation lower than 0.5 and 20% respectively) and were used for normalization.

Differentially-expressed genes between infected and noninfected mice were identified using two parameters: the fold change of gene expression (FC), and the statistical significance. FC was calculated as the ratio between biological groups (infected and control mice) at each experimental timepoint and expressed as log2. Statistical significance was determined by the non-parametric Mann–Whitney U-test, considering p < 0.05 as statistically significant. In order to display changes, Volcano plots were made by plotting –log<sup>10</sup> p-value on the y-axis, and log<sup>2</sup> of FC on the x-axis. Genes passing both statistical significance threshold (–log<sup>10</sup> p > 1.3, corresponding to p = 0.05), and biological significance threshold (log<sup>2</sup> of FC > 0.6 or < −0.6, corresponding to FC > 1.5 or < −1.5), were marked in red and blue, depending on their upregulation and downregulation, respectively. Those genes were considered biologically relevant and used for further biological interpretation.

### Development of Regression Models

A logistic regression model was developed in order to assess probability of infection in mice based on NRQ values of 36 genes, all of them coding for soluble molecules that could eventually be detected in blood (Table S1). From the complete pool of 61 mice used in this model, 75% of them (47 mice), were randomly selected and used for its development and 25% of them (14 mice) were used for its evaluation. The fitting of the proposed model was evaluated using Nagelkerke R 2 as well as Hosmer and Lemeshow tests. The statistical significance of each selected variable was evaluated using Wald test. The general function for logistic regression is:

$$p = \frac{1}{1 + e^{-(\beta\_0 + \beta\_1 \ x\_1 + ... + \beta\_k \ x\_k)}}$$

in which p is the probability of infection in a given individual, β0, β1, β2, etc., represent the regression coefficients of each variable identified by the model, and x1, x2, x3, stand for the NRQ values of each variable in that mice. Finally, the predictive power of the model, as well its sensitivity and specificity, was first autoevaluated with those mice used for its development (47 mice) followed by an external evaluation using the extra 25% of animals (14 mice), and compared with the actual results (infected vs. non-infected), using standard formulae:

$$sensitivity = \frac{VP}{VP + FN}$$

$$specificity = \frac{VN}{VN + FP}$$

Additionally, a linear regression model was developed in order to estimate the parasitic burden in spleen in any infected mice. In this model, NRQ values for 36 genes (previously indicated, Table S1) were included as independent variables. Again, 75% of mice (23 infected mice) were used for the development of the model, and 25% of them (8 mice) were used for its evaluation. The fitting of the proposed model was evaluated using multiple correlation coefficient R and determination coefficient R 2 . The individual significance of each variable was analyzed using a ttest. The model was verified for collinearity, tolerance, linearity, normality, homoscedasticity and independence of errors. The general function for multivariant linear regression is regression is:

$$
\mu\_n = \beta\_0 + \beta\_1 \mathbb{x}\_{1n} + \beta\_2 \mathbb{x}\_{2n} + \dots + \beta\_p \mathbb{x}\_{pn} + e\_n
$$

in which y is the parasitic burden in spleen determined by the model in a given mouse, β0, β1, β2, etc., represent the regression coefficients of each variable identified by the model, and x1, x2, x3, etc., stand for the NRQ values of each variable in that mouse.

Statistical analyses were performed using SPSS 20 (IBM) software and the graphic representations were performed with GraphPad Prism version 5.00 (GraphPad Software, San Diego, United States).

# RESULTS AND DISCUSSION

In the last few years, differential analysis of transcriptome has raised as an increasingly relevant tool in order to identify key aspects of complex scenarios, such as infection of target organs or cells (Dillon et al., 2015; Fernandes et al., 2016; Kong et al., 2017). In this paper, an extensive real-time quantitative PCR (qPCR) analysis identified global changes in gene expression profiles of 112 immune-related genes in infected vs. non-infected spleens at four different timepoints, revealing three distinct immunopathological scenarios: early infection phase (1–2 weeks after infection), chronic phase (8 weeks after infection) and intermediate phase (4 weeks after infection) consistent with exhaustion of the immune response.

# *Leishmania infantum* Early Infection Induces a Mixed Proinflammatory and Immunosuppressive Response in the Spleens of Infected Mice

One week after parasite inoculation (1 wpi), infection is clearly stablished in both spleen (10<sup>4</sup> parasites/gr) and liver (10<sup>6</sup> parasites/gr) (**Figures 1A,B**). This infection increased during the second week of infection in both organs, demonstrating that the parasite was actively replicating and overcoming the killing by the immune system (**Figures 1A,B**). Similarly, splenomegaly

FIGURE 1 | Evolution of parasite burden in spleen (A) and liver (B) of infected mice (n = 24). Mice were inoculated with 1 × 10<sup>6</sup> promastigotes i.v. Spleen and liver parasite load were determined on week 1 (n = 6), week 2 (n = 6), week 4 (n = 6) and week 8 (n = 6) post-infection by limiting dilution assay and expressed as log<sup>10</sup> of the average parasite load per gram of tissue. Evolution of spleen weight in infected (n = 24) vs. control mice (n = 23) over the course of infection (C). The bars represent the weight in grams in infected (black bars) and non-infected control mice (white bars) at 1, 2, 4, and 8 weeks post-infection. Statistically significant differences are indicated (\*p ≤ 0.05; \*\*p ≤ 0.01).

was observed in the infected groups during the first 2 weeks after infection (**Figure 1C**). This period corresponds to the initial phase of infection, when the immune response is apparently able to control dissemination of the parasite although not to eliminate it (Engwerda et al., 2004b; Rodrigues et al., 2016).

In order to clarify the immunological complexity of host/parasite interaction at this stage, this gene expression analysis was focused on the immunological events that occur in the spleen, being that the organ where the disease becomes chronic. A high-throughput real time quantitative PCR (qPCR) gene expression analysis was carried out in spleen samples of 47 mice, comparing RNA expression levels of 112 immune system-related genes between infected and non-infected mice. Differentially-expressed genes were identified using the fold change of gene expression (FC), and the statistical significance as described in the previous section. Only genes fulfilling both statistical and biological significance thresholds were considered biologically relevant and used for further biological interpretation.

One week after challenge, 22 out of the 112 genes were differentially expressed between infected and non-infected mice, most of them upregulated (71.4%, indicated as red dots in **Figure 2**). Within the group of upregulated mRNAs, there are genes encoding for chemokines (Cxcl10, Cxcl9, and Xcl1), chemokine receptors (Ccr5 and Cxcr3), interleukins (Il1b, Il12a, Il18bp), interleukin receptors (Il12rb1, Il23r), transcription factors (Stat1), costimulatory-signal inhibitors (Ctla4) and other genes (Tnfrsf1a and Ptgs2). It is worth mentioning that three genes (Il12rb2, Il23r, and Ptgs2) are only expressed in infected mice but not in the control group, hence their high FC (**Figure 3**). The downregulated genes at 1 wpi were genes encoding one chemokine (Ccl2), four interleukin receptors (Il22ra2, Tgfbr1, Il5ra, and Il1rap), one Toll-like receptor (Tlr2), one costimulatory molecule (Cd40) and one cellular adhesion molecule (Icam2).

The upregulation of the chemokine receptor genes Ccr5 and Cxcr3, encoding chemokine receptors CCR5 and CXCR3, which are expressed in monocytes, macrophages, immature dendritic cells (DCs), natural killer (NK) cells and activated T lymphocytes (including effector and regulatory cells) (Groom and Luster, 2011) suggested the initial recruitment of these cell populations toward spleen, in agreement with the splenomegaly observed at this timepoint (**Figure 1C**). Besides, CCR5 has been described as one of the Leishmania entry-points in macrophages, contributing to infection (Majumdar et al., 2014).

Splenomegaly is also supported by upregulation of Xcl1 and Cxcr3, as well as the high transcription levels of Cxcl9 and Cxcl10 genes (both encoding CXCR3 ligands) observed at this timepoint, promoting the recruitment of T lymphocytes (TL) and DCs, which contribute to inflammation, as well as more CXCR3 expressing cells. In our assays, the upregulation of Cxcr3, Cxcl9, and Cxcl10 suggest that the mouse immune system attempts to control infection by chemoattraction of lymphocytes and DC to the spleen. On the other hand, our data also showed downregulation on mRNA levels of Mcp-1 or Ccl2, that mediates the recruitment of CC chemokine receptor 2 (CCR2) expressing cells (Ibrahim et al., 2014), which in mice includes the inflammatory monocytes subset (Gordon and Taylor, 2005; Gordon, 2007).

Taken together, these data suggest that L. infantum infection induces the preferential recruitment of T cells and CCR5-expressing cells toward spleen. This might be a parasite-induced strategy to escape from macrophages killing, since, according to different works the CCR5<sup>+</sup> macrophage subset is prone to be silently infected by Leishmania through the CCR5 receptor (Bhattacharyya et al., 2008). The parasites can use this mechanism to enter silently into the macrophages and successfully establish inside the host. This would explain the existence of parasite load and the absence of expression of Th-type cytokines in our model.

Our results also revealed upregulation of two inflammationrelated genes: interleukin 1 beta (Il1b) and prostaglandinendoperoxide synthase (Ptgs2). However, we did not observe an increased expression of characteristic Th1 genes, like interleukin 12 (Il12) or interferon gamma (Ifng). The p35 gene (IL-12a subunit) is ubiquitously expressed by most cells whereas the p40 gene (IL-12b subunit) is primarily expressed by antigen presenting cells (APC) (Ma and Trinchieri, 2001) in response to different stimuli like Cd40 and Cd40-Ligand crosstalking; however Cd40 is downregulated in our results. To be biologically

fold-change between infected and non-infected mice; the y-axis corresponds to the statistical significance, expressed as the negative logarithm of p-values. The red horizontal line indicates the cut-off for the statistical significance p = 0.05. Black vertical lines represent the log<sup>2</sup> FC of −0.6 and 06 (corresponding to FC −1.5 and 1.5 respectively) used as biological threshold to identify differentially expressed genes. The negative values correspond to down-regulated genes (indicated in blue) and the positive values are the up-regulated genes (indicated in red). Black and gray dots represent non-differentially expressed genes.

FIGURE 3 | Differential gene-expression of the 112 analyzed genes in infected (n = 6) vs. control mice (n = 6), 2 wpi. The x-axis represents log<sup>2</sup> of expression fold-change between infected and non-infected mice; the y-axis corresponds to the statistical significance, expressed as the negative logarithm of p-values. The red horizontal line indicates the cut-off for the statistical significance p = 0.05. Black vertical lines represent the log<sup>2</sup> FC of −0.6 and 0.6 (corresponding to FC −1.5 and 1.5 respectively) used as biological threshold to identify differentially expressed genes. The negative values correspond to down-regulated genes (indicated in blue) and the positive values are the up-regulated genes (indicated in red). Black and gray dots represent non-differentially expressed genes.

active and exert its biological functions, both subunits must be present and form the heterodimer (Kima, 2008); nonetheless, as shown in **Figure 2**, only Il12a is upregulated in our experiment. These results, together with the absence of Stat4 upregulation point to a blockade of IL-12 secretion (Yoshida et al., 2007). IL-12 acts on activated T lymphocytes, driving its differentiation to Th1 subclass, therefore its absence hampers Th1 differentiation and disease control. However IFN-γ production might also be stimulated by interleukin 18 (IL-18) (Gracie, 2003) and by NK cells after binding of lipophosphoglycan (LPG) to Tolllike receptor 2 (TLR2) on NK cells surface (Faria et al., 2012; Singh et al., 2012; Lemaire et al., 2013). In this sense, our results showed downregulation of Tlr2, no differential expression of Il18 and upregulation of IL-18-binding protein coding gene (Il18bp) which binds IL-18 with high-affinity and inhibits its functions (Kim et al., 2000; Gracie, 2003). All these results seem to indicate another mechanism used by Leishmania to avoid IFN-γ production and therefore hinder generation of Th1 responses. Besides IL-12 and IFN-γ, the lack of differential expression levels of Il4, Il13, Il5, Il17, Il23, Il21 and forkhead box P3 (Foxp3) genes, rule out the possibility of active Th2, Th17, nTreg and iTreg responses (Vieira et al., 2004; Gregori et al., 2012; Ma et al., 2012; Ley, 2014).

Upregulation of interleukin 12 receptor beta 1 (Il12rb1) and interleukin 23 receptor (Il23r) (whose products form the interleukin 23 (IL-23) receptor (Parham et al., 2002) might indicate and attempt to generate a Th17 response through IL-23 signaling. Nevertheless, the lack of differences on mRNA levels of transforming growth factor beta (Tgfβ) and interleukin 6 (Il6), necessary for Th17 differentiation (Bettelli et al., 2006; Mangan et al., 2006; Yoshimoto et al., 2010), exclude this possibility. Taken together, the data show that, 1 wpi, a clear adaptive cellular response has not been stablished yet, and that only an inflammatory process is taking place within the spleen. However, this inflammatory profile might be counteracted by upregulation of Cytotoxic T-Lymphocyte Antigen 4 (Ctla4). CTLA-4 is expressed by effector T lymphocytes upon activation as well as by Treg cells, being one of their immunosuppressive mechanisms (Gregori et al., 2012). It acts as a negative regulator of T cell activation, preventing appropriate T cell co-stimulation (Kaye et al., 1994). Therefore, its upregulation in our data suggests the existence of an immunosuppression process, as has been previously described (Murphy et al., 1998; Stanley and Engwerda, 2006). Therefore, 1 wpi the gene expression analysis reveals a mixed proinflammatory and immunosuppressive response within the spleen of the infected mice.

Two weeks after infection (2 wpi), only 4 out of the 112 genes were differentially expressed (**Figure 3**), probably as a consequence of the immunosuppressive signals observed at 1 wpi. This is supported by the inability to control parasite replication in target organs (**Figure 1**). Splenomegaly can be explained by upregulation of Cxcr3, indicating recruitment of a wide variety of leukocytes (reviewed in Groom and Luster, 2011).

# *L. infantum* Chronic Infection Induces an Ineffective Inflammatory Response in the Spleens of Infected Mice

As previously reported (Engwerda et al., 2004a; Rodrigues et al., 2016) after 2 wpi, the parasitic burden in the liver of the infected mice starts to go down (**Figure 1B**) due to the development of a T-cell mediated immunity and the formation of granulomas. In contrast, the parasite population in the spleen increase slowly but steadily by 4 wpi, only to rocket by the 8th week after infection (**Figure 1A**). These events mark the onset of the chronic phase in visceral leishmaniasis (Engwerda et al., 2004a).

In this early chronic phase (4 wpi), 11 out of the 112 genes were differentially expressed between infected and non-infected mice, only two of them (Ccl7 and Ccl22) upregulated and 9/11 (82%) downregulated: four interleukin receptors (Il22ra2, Tgfbr1, Il5ra, and Il1rap), one Toll-like receptor (Tlr7) and one costimulatory molecule (Cd40l) (**Figure 4**). Remarkably, this was the first time along this timecourse when there was a general downregulation of gene expression of immune related genes in Leishmania-infected spleens. The immunosuppression process revealed 1 wpi, is displayed in our results as a reduction in the number of differentially expressed genes 2 wpi, and later as an overall downregulation of immune-related gene expression (4 wpi). This possibility is also supported by the progressive reduction of the spleen weight in the infected group between the first and the fourth week following infection (**Figure 1C**). This pattern is probably a consequence of impaired cell recruitment to the spleen, induced by the absence of chemokine upregulation. L. infantum might be blocking chemokine production in an attempt to generate an adequate environment in order to insidiously stablish infection, reflected by the slow increase (100-fold) in parasite burden during this period (**Figure 1A**).

Another important finding possibly related to this apparently "dormant" state of infected spleens was the strong downregulation of Ccl7 gene expression. CCL7 is among the most pleiotropic chemokines since it recruits all major leukocyte classes, particularly monocytes and neutrophils, by binding to different chemokine receptors (CCR1, CCR2. . . ) (Menten et al., 2001; Navas et al., 2014; Melo et al., 2017). Downregulation of Ccl7 in infected spleen tissue, in addition to the pattern of increasing parasite burden, general downregulation of gene expression and no clear spleen inflammation, suggest a decreasing chemoattractant capacity of the immune system due to a L. infantum-induced immunosuppression process. This has been related to T-cell exhaustion (Joshi et al., 2009), a progressive process characterized by the loss of effector function of antigen-experienced T cells, failure to produce IFN-γ and TNF-α, and that can culminate in the physical deletion of the responding cells (Yi et al., 2010; Bhadra et al., 2011; Gigley et al., 2012; Rodrigues et al., 2014). This phenomenon can be counteracted by IL-21 produced by exhausted CD4 T cells, in an attempt to "help" the CD8 response during chronic infection (Yi et al., 2010; Gigley et al., 2012; Wherry and Kurachi, 2015). Upregulation of Il21 is remarkable in our data (**Figure 4**) therefore, a scenario of Leishmania-induced T-cell exhaustion in spleen of the infected animals at 4 wpi seems likely, despite

the lack in our data of marker genes like programmed death-1 (PD-1), T-cell immunoglobulin and mucin domain-containing protein-3 (TIM-3) and lymphocyte-activated gene-3 (LAG-3) (Rodrigues et al., 2014).

Later in the chronic phase, 8 weeks after infection (8 wpi), 33 out of the 112 genes were differentially expressed between infected and non-infected mice. Unlike what happened 4 wpi, there was an overall upregulation in gene expression, with 25 upregulated and only 8 downregulated genes (**Figure 5**). The upregulated mRNAs encoded for chemokines (Cxcl10, Cxcl9, Ccl3, Ccl4, and Xcl1), chemokine receptors (Ccr5 and Xcr1), interleukins (Il1a, Il1b, Il10, Il12a, Il18bp), interleukin receptors (Il1rn, Il2ra, Il2rg, Il12rb2), transcription factors (Stat1, Stat3), Toll-like receptors (Tlr3, Ttlr4, Tlr7, and Tlr9), cytokines (Ifng, Tnfa) and other genes (Icos and Myd88). Within the group of negatively regulated mRNAs, there were genes encoding for one chemokine (Ccl5), chemokine receptors (Ccr7), interleukins (Il4), four interleukin receptors (Il21r, Il22ra2, Il23r, and Il27ra) and one cytokine (Tgfb2). At the same time, clear differences in parasitic burden and weight are observed in infected spleens compared to healthy controls (**Figures 1A,C**).

Our analysis showed upregulation of 2 different chemokine receptors (Ccr5 and Xcr1), suggesting the recruitment of different cells including inflammatory monocytes (Gordon and Taylor, 2005; Gordon, 2007), Th1 lymphocytes, some DC (Bhattacharyya et al., 2008), NK cells (Liaskou et al., 2012), Treg cells (Yurchenko et al., 2006; Mougneau et al., 2011), and CD8<sup>+</sup> and some DCs (Crozat et al., 2011) between 4 and 8 wpi; these cell populations are probably responsible for the splenomegaly observed at this point of the infection. Upregulation of Ccr5 might be indicative of the presence of CCR5-expressing monocytes, which are highly susceptible to be infected by Leishmania (Bhattacharyya et al., 2008; Majumdar et al., 2014), and correlates with the 5000-fold increase of the parasite burden in the period between 4 and 8 wpi.

In addition to these chemokine receptors, other chemokineencoding genes (Ccl3, Ccl4, Cxcl9, Cxcl10, and Xcl1) were induced by L. infantum infection at 8 wpi. CXCL9, and CXCL10 are chemoattractant for CXCR3-expressing cells (Kima and Soong, 2013), while both CCL3 and CCL4, are chemotactic for CCR5 expressing cells. The precise role of CCL3 remains unclear, but according to some authors it seems to be important in early containment of parasite burden and the generation of an antileishmanial cytokine environment, but may be deleterious in the latter stages of chronic L. donovani infection, since it promotes parasite persistence (reviewed in Oghumu et al., 2010). This effect can be seen in our experiment since, between weeks 4 and 8 post-infection, there was a significant increase of parasite burden. This cell recruitment produced an upregulation of Leishmaniainduced proinflammatory genes (Ifng, Tnfa, Il1a, Il1b, and Il18) which also support the idea of an active inflammatory process at this point of the infection. Apart from these genes, there are other upregulated genes that are indirectly related to an inflammatory Th1-type response: Tlr3, Tlr4, and Stat1 (Flandin et al., 2006; Schindler and Plumlee, 2008; Tuon et al., 2008; Singh et al., 2012).

However, even though there was an upregulation of inflammatory genes and Ifng, a Th1 characteristic gene, the immune system was unable to contain disease progression since parasite load keeps growing. One possible explanation

for this can be found on the upregulation of Il1rn and Il18bp, coding for two anti-inflammatory cytokines which inhibit the proinflammatory effects produced by both IL-1a/IL-1b and IL-18, respectively (Correa and López, 2007). Another possibility is that IFN-γ is unable to exert its functions efficiently, for example by the existence of counteracting immune responses occurring at the same time. The lack of differences on mRNA expression levels of Il5, Il13, Il17f, Il17a, Il27, and Foxp3, along with downregulation of Il4, Il23r, Il22ra2, Tgfbr2, and Tgfb2 preclude the generation of Th2, Th17, nTreg and iTreg responses. Similarly, upregulation of FoxP3, Tgfb, Il0 and Ctla4 has been associated to the development of CD4+CD25+ regulatory T cells (Tregs) (Yamashita et al., 2006), but those markers do not correlate in this study. Nevertheless, there is a non-FOXP3 expressing Treg subset known as Tr1 that may fit our expression profile by 4 reasons: (i) Tr1 cells co-express IFN-γ and IL-10 (Wakkach et al., 2003; Nylén et al., 2007; Gregori et al., 2012; Faleiro et al., 2014) and both genes are upregulated in our results; (ii) Tr1 differentiation is STAT3-mediated (Gregori et al., 2012) and this gene is upregulated; (iii) Tr1 cells do not produce IL-4 (Wu et al., 2007; Gregori et al., 2012), and Il4 gene is downregulated by the L. infantum-infected group; (iv) Icos gene is upregulated and some authors indicate that ICOS is expressed in Tr1 cells (Häringer et al., 2009; Gregori et al., 2012). However, differentiation between a Tr1 population and other IL-10-producing T cell subsets is complicated, and other possibilities cannot be excluded (reviewed in Gregori et al., 2012).

An interesting possibility to explain the inability of the immune system to control disease progression is based on the upregulation of Tnfa and Il10 and downregulation of Ccr7, and is consistent with previous works carried out by Stanley and coworkers (Stanley and Engwerda, 2006) using L. donovani-infected murine models. During chronic infection, the spleen suffers dramatic changes in microarchitecture, including disorganization of the white pulp, hypertrophy of the red pulp and disruption of the marginal zone (Kaye et al., 2004; reviewed in Rodrigues et al., 2016). These changes are related to a TNFα-dependent, IL-10-mediated inhibition of CCR7 expression in DC, resulting in severely impaired DC migration to the periarteriolar lymphoid sheds (PALS) for antigen presentation to T cells, giving rise to a severe immunosuppression and enhancing parasite proliferation.

Both options, generation of Tr1 responses and alteration on the splenic architecture may be related events. In fact, Il10 upregulation, as a compensatory mechanism to counteract an excess of TNF-α, might have its origin on regulatory DC (rDC) from the PALS. rDCs secrete IL-10 and skew T cell development to Tr1 cells, producing both IFN-γ and more IL-10 (Wakkach et al., 2003; Nylén et al., 2007; Gregori et al., 2012; Faleiro et al., 2014). IL-10 produced by both Tr1 and rDC cells, contributes to CCR7 downregulation hampering DC migration to the PALS, avoiding their contact with naïve T cells, and therefore blocking the establishment of antigen-specific T-cell responses.

# Mechanisms Underlying Leishmaniasis Progression in Spleen Tissue Over Time

In order to identify the effect of Leishmania infection on expression of immune related genes over the course of infection, the expression fold-change between infected and control mice was plotted against time (**Figure 6**), indicating statistically significant differences (p < 0.05) between infected and control mice with black bars.

The analyses revealed that early response against Leishmania infection is characterized by the upregulation of Th1 markers and characteristic M1-macrophage activation molecules such as Ifng, Stat1, Cxcl9, Cxcl10, Ccr5, Cxcr3, Xcl1, and Ccl3 (reviewed in Martinez and Gordon, 2014). This activation does not protect spleen from infection, since parasitic burden rises along time (**Figure 1**). This marked difference in gene expression between infected and control mice disappears during intermediate stages of infection (2 and 4 wpi). This inability to control infection and the loss of those Th1/M1 activation markers, may be related to strong anti-inflammatory and immunosuppresory signals that are activated early upon infection (Ctla4) or remain activated throughout the experiment (Il18bp). That would suggest that L. infantum might be blocking chemokine production to generate an adequate environment to maintain infection during these weeks, through a T-cell exhaustion process (Rodrigues et al., 2014), that the immune system tries to overcome with the strong upregulation of Il21 at 4 wpi (Yi et al., 2010; Gigley et al., 2012; Wherry and Kurachi, 2015).

The overexpression of these Th1/M1 markers is restored later in the chronic phase (8 wpi), suggesting the generation of a classical "protective response" against leishmaniasis. Nonetheless, the parasitic burden rockets at this timepoint. This apparent contradiction can be explained by the generation of a Tr1 regulatory immune response characterized by overexpression of Ifng, Tnfa, Il10 and downregulation of Ccr7 and Il4 (**Figure 6**), that counteracts the Th1/M1 response.

This global analysis of gene expression patterns during Leishmania infection in BALB/c spleen tissue raises two interesting points. Firstly, Ifng production is not a valuable predictor for Th1 protective responses, since its action may be counteracted in many different ways and might even take part in immunosuppresory mechanisms. Secondly, the classical Th2 response, characterized by IL-4 and its regulator GATA3 overexpression among other markers (reviewed in Selvapandiyan et al., 2012), is not playing any clear role in disease progression in this experimental model, given the downregulation of Il4 and the lack of differential expression of Gata3 gene observed.

Taken together, these results highlight the need for comprehensive analysis of gene expression in infected tissues or organs, in order to avoid misinterpretation of individual data. In practice, analyses of gene expression of a limited number of genes in complex scenarios like infection with an intracellular protozoan, might generate misleading results due to "missing information."

This approach of using differential transcriptome analysis as a tool to understand Leishmania-host interactions has been successfully employed in several animal models and in vitro studies. Despite differences related to experimental models (Syrian hamster/mice/in vitro macrophages) and methodology (RNA-seq/qPCR/microarray hybridization) all of them draw a picture of mixed responses during infection and deactivation of effective parasite-controlling responses (Rabhi et al., 2012; Dillon et al., 2015; Fernandes et al., 2016; Kong et al., 2017; Medina-Colorado et al., 2017). Similar to our study, transcriptional profile of spleen samples from L. donovani-infected hamsters was analyzed 28 days after infection, revealing a strikingly proinflammatory environment and a strong expression of Ifng that did not protect against the increasing parasite burden (Kong et al., 2017). Likewise, chronic infection in hamsters revealed expression of markers of both T cell activation and inhibition, showing mixed expression of Th1 and Th2 cytokines and chemokines, and again ineffective in controlling infection (Medina-Colorado et al., 2017). Those studies that focused on the evaluation of early stages of infection (Dillon et al., 2015; Fernandes et al., 2016) revealed upregulation of both pro- and anti-inflammatory related genes similarly to our findings. In conclusion, the use of high-throughput technology on complex scenarios like the interaction between Leishmania and its animal host is opening new perspectives on immune response, and also providing large collections of data that can be useful for the identification of new potential biomarkers.

# Identification of Potential Biomarkers for Leishmaniasis Based Upon Linear and Logistic Regression Models

Different approaches have been tested to identify new potential biomarkers able to predict infection, to determine parasitic load in infected organs or its clearance upon treatment. One interesting method is the use of multivariant statistical analyses to identify markers and to develop models able to predict disease parameters like infection (or absence of it) or parasitic burden.

The logistic regression model developed to determine whether there is a relationship among some of the genes whose expression has been analyzed in this work and absence/presence of parasite in the spleens, used normalized relative expression levels (NRQ) of the 36 genes coding for soluble markers (interleukins, cytokines. . . ) from 47 randomly selected mice. The logistic regression model predicts the probability of infection in a given mouse, based on the expression levels (NRQ) of Il18bp, Cxcl1 and Il2, being Il18bp and Il2 directly correlated and Cxcl1 inversely correlated (**Table 1**).

$$p = \frac{1}{1 + e^{-(-7.89 + 10.43 \ast l l 18p - 5.35 \ast \text{Cxcl1} + 2.77 \ast l l 2)}}$$

The predictive capacity of the model was auto-evaluated by comparing the observed results and those yielded by the model (**Table 2**).

The auto-evaluation of the proposed model classified correctly 40 out of the 47 samples (85%), with a sensitivity of 87.5% and a specificity of 82.6%. The effectiveness of the proposed model was evaluated using 14 extra mice that were not included on its development (**Table 3**). In this case, the sensitivity of the model was 57% and its specificity reached 85%.

In the proposed model, the probability of infection has a positive correlation with the expression of genes coding for IL-18bp and IL-2 in spleen. IL-18bp is an inhibitor of the proinflammatory cytokine IL-18, which is a major inducing factor of IFN-γ, has multiple biological functions and is

infection: 1, 2, 4, and 8 wpi. Solid black bars indicate statistically significant differences with p ≤ 0.05.

each indicated gene, that is the ratio between the average gene expression in the infected group and non-infected-control mice. The x-axis represents time after

involved in immune regulation, anti-infection, and inflammation (Chaudhry et al., 2006). IL-18BP has been proposed as a biomarker of severity of injury after exposure to ionizing radiation in mice (Ha et al., 2016) and also a biomarker useful for differentiation of leptospirosis and dengue virus infection in humans (Conroy et al., 2014). IL-2 is and interleukin related with TL proliferation and the development of an adaptive immune response. IL-2 has been proposed as a valuable biomarker for detection of asymptomatic individuals in areas were L. infantum is endemic after whole blood stimulation with soluble Leishmania antigen (SLA), although the concentration of this biomarker is low (Ibarra-Meneses et al., 2016, 2017a). On the contrary, IL-2 did not perform as well for asymptomatic individuals from a L. donovani endemic area (Ibarra-Meneses et al., 2017b). In canine visceral leishmaniasis (CVL), serum IL-2 levels showed no correlation with disease severity (Solcà et al., 2016). Finally, in our model, expression of Cxcl1 gene presents an inverse correlation with the probability of infection. CXCL1 chemokine plays a role in inflammation and is chemoattractant for neutrophils (reviewed in Kobayashi, 2008), the first-line defense against leishmania. Impaired CXCL1 levels have been related with increased susceptibility to Klebsiella pneumoniae in mice due to low inflammatory cell recruitment, reduced CXCL2 and CXCL5


β coefficients for each selected variable, for the constant, and the results from Wald-tests including p-values.

### TABLE 2 | Logistic regression model auto-evaluation.


TN, True Negatives. FP, False Positives. FN, False Negatives. TP, True Positives.

### TABLE 3 | Logistic regression model evaluation.


TN, True Negatives. FP, False Positives. FN, False Negatives. TP, True Positives.

production and decreased activation of NF-κB and MAPKs (Cai et al., 2010). Interestingly, CXCL1 serum levels in CVL were correlated with disease severity (Solcà et al., 2016).

Another interesting issue in leishmaniasis is the evaluation of parasitic load in infected animals, a useful parameter when evaluating treatment effectiveness. After analysis of the NRQ values of the 36 genes coding for soluble markers from 23 infected mice, a linear multivariant regression model was developed based on the expression levels of Ccl3, Cxcl9, and Il18bp (**Table 4**).

The following equation predicts the parasitic burden in spleen from a given mouse based on the expression levels of Ccl3, Cxcl9 and Il18bp, being Il18bp and Ccl3 directly related and Cxcl9 inversely correlated.

$$\text{parasitic burden} = \text{3.30} + \text{1.97\*Col3} - \text{0.338\*Col3}$$

$$+ \text{1.265\*Ill18bp}$$

ANOVA test showed p = 0.0, R = 0.94, and R <sup>2</sup> = 0.89, indicating that the model is competent in the prediction of parasitic burden in spleen. The values of Tolerance and Partial correlation (**Table 4**) rejected co-linearity and partial correlation among the variables, therefore the three selected parameters are individually useful for parasitic burden determination. Durbin Watson test was 2.5, discarding independent errors.

It is interesting how the expression of these three genes may correlate with infection. Ccl3 encodes CCL-3, a chemokine chemoattractant for macrophages, the preferred leishmania host-cells (Oghumu et al., 2010) and has been shown to be overexpressed in spleen during VL (Kong et al., 2017). CCL3 has been proposed as a biomarker for a series of conditions ranging from lymphoma (Takahashi et al., 2015) to osteoarthritis (Zhao et al., 2015) and chronic obstructive pulmonary disease (Ravi et al., 2014), in which inflammation plays a role. As indicated earlier, IL-18BP inhibits IL18-induced IFN-γ production in TL and the generation of an adaptive cellular response (Chaudhry et al., 2006). Therefore, it is likely that upregulation of both markers contributes to parasitic burden increase in spleen. The linear regression analysis selected Cxcl9 gene expression as a marker negatively correlated with parasitic burden in spleen. Given its role on immunopathogenesis of the disease in mice, recruiting lymphocytes toward spleen, the selection of this marker is not surprising, and its overexpression has been reported in spleen during VL (Kong et al., 2017).

The competence of the proposed model was evaluated using data from 8 extra mice that were not included in its development, as described in the Methods section. As shown in **Table 5**, the fitting of the model is good, since the difference between

TABLE 4 | Linear multivariant regression model.


TABLE 5 | Observed and estimated parasitic burden (expressed as Log10) in the infected spleens.


Upper and lower limits indicate 95% confidence intervals of the estimated parasitic burden.

estimated and observed parasitic loads is low (<1.5 log units) in 6 out of 8 animals and the observed burden was inside the confidence interval in all of them.

The gene expression of the majority of the variables selected in these regression models as relevant for disease detection or progression have not been studied in pathological situations like Leishmania infection. Gene expression does not always translate into protein expression, and soluble factors, such as CCL3, IL-18BP or CXCL9, may or may not be secreted into blood or plasma in a fashion that correlates with gene expression. Nevertheless, the levels in serum, plasma or stimulated blood cells of some of them have been proposed as biomarkers during different conditions, suggesting that the use of some of these molecules may be useful for the monitorization of different aspects of leishmania infections in experimental models. More research will be needed to assess the real practical value of the biomarkers and the prediction models described in this manuscript, such as the actual expression levels of the proteins or their specificity during infection, but our findings outline an innovative strategy for identification of new potential biomarkers in visceral leishmaniasis.

# DATA AVAILABILITY STATEMENT

The data discussed in this publication have been deposited in NCBI's Gene Expression Omnibus (Edgar, 2002) and are

# REFERENCES


accessible through GEO Series accession number GSE112129 (for the infection experiments) (https://www.ncbi.nlm.nih.gov/ geo/query/acc.cgi?acc=GSE112129) and GSE112138 (for the identification of potential biomarkers) (https://www.ncbi.nlm. nih.gov/geo/query/acc.cgi?acc=GSE112138).

# AUTHOR CONTRIBUTIONS

EO and YH-S performed the experiments and the statistical analysis. EC, AG-G and BV contributed to conception and design of the study. AG-G and ML contributed to establishing the experimental infection model. EO wrote the first draft of the manuscript. EC wrote the final version of the manuscript. BV, ML contributed to discussion and analysis of data. All authors contributed to manuscript revision, read and approved the submitted version.

# FUNDING

This study was funded by Fundación CajaCanarias (EC) (Ref. 2015 BIO14); Fondo de Investigaciones Sanitarias (FIS)- Instituto de Salud Carlos III, Ministerio de Economía y Competitividad (BV) (N◦ PI11/02172); Red de Investigación de Centros de Enfermedades Tropicales (RICET)-Ministerio de Economía y Competitividad- Instituto de Salud Carlos III (RD16/0027/0005 (ML) and RD16/0027/0001 (EC, BV); Plan Nacional de I+D+I, Ministerio de Economía y Competitividad (ML) SAF2016-81003-R. YH-S was supported by Beca de Investigación Obra Social La Caixa- Fundación CajaCanarias para postgraduados de la Universidad de La Laguna.

# ACKNOWLEDGMENTS

We would like to acknowledge Dr. Roberto Dorta (ULL) for support on statistical analyses.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2018.00197/full#supplementary-material


NF-kappaB, and MAPKs. J. Immunol. 185, 6214–6225. doi: 10.4049/jimmunol. 0903843


interleukin-5/13-producing group 2 innate lymphocytes in murine asthma. Mol. Med. Rep. 15, 4291–4299. doi: 10.3892/mmr.201 7.6500

Zhao, X. Y., Yang, Z. B., Zhang, Z. J., Zhang, Z. Q., Kang, Y., Huang, G. X., et al. (2015). CCL3 serves as a potential plasma biomarker in knee degeneration (osteoarthritis). Osteoarthr. Cartil. 23, 1405–1411. doi: 10.1016/j.joca.2015.04.002

**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 Ontoria, Hernández-Santana, González-García, López, Valladares and Carmelo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Biomarkers of Cutaneous Leishmaniasis

### Fariborz Bahrami <sup>1</sup> , Ali M. Harandi <sup>2</sup> and Sima Rafati <sup>3</sup> \*

*<sup>1</sup> Department of Immunology, Pasteur Institute of Iran, Tehran, Iran, <sup>2</sup> Department of Microbiology and Immunology, Institute of Biomedicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden, <sup>3</sup> Department of Immunotherapy and Leishmania Vaccine Research, Pasteur Institute of Iran, Tehran, Iran*

Cutaneous leishmaniasis (CL) is an immune-mediated skin pathology caused mainly by *Leishmania* (*L*.) *major*, *Leishmania tropica*, *Leishmania braziliensis*, *L. mexicana,* and *L. amazonensis*. The burden of CL in terms of morbidity and social stigmas are concentrated on certain developing countries in Asia, Africa, and South America. People with asymptomatic CL represent a large proportion of the infected individuals in the endemic areas who exhibit no lesion and can control the infection by as yet not fully understood mechanisms. Currently, there is no approved prophylactic control measure for CL. Discovery of biomarkers of CL infection and immunity can inform the development of more precise diagnostics tools as well as curative or preventive strategies to control CL. Herein, we provide a brief overview of the state-of-the-art for the biomarkers of CL with a special emphasis on the asymptomatic CL biomarkers. Among the identified CL biomarkers so far, direct biomarkers which indicate the actual presence of the infection as well as indirect biomarkers which reflect the host's reaction to the infection, such as alterations in delayed type hypersensitivity, T-cell subpopulations and cytokines, adenosine deaminase, and antibodies against the sand fly saliva proteins are discussed in detail. The future avenues such as the use of systems analysis to identify and characterize novel CL biomarkers are also discussed.

### Edited by:

*Javier Moreno, Instituto de Salud Carlos III, Spain*

### Reviewed by:

*Sofia Cortes, Universidade NOVA de Lisboa, Portugal Shahab Babakoohi, Medical College of Wisconsin, United States*

### \*Correspondence:

*Sima Rafati s\_rafati@yahoo.com; sima-rafatisy@pasteur.ac.ir*

Received: *29 March 2018* Accepted: *11 June 2018* Published: *26 June 2018*

### Citation:

*Bahrami F, Harandi AM and Rafati S (2018) Biomarkers of Cutaneous Leishmaniasis. Front. Cell. Infect. Microbiol. 8:222. doi: 10.3389/fcimb.2018.00222* Keywords: cutaneous leishmaniasis, asymptomatic, Leishmania, biomarker, immune responses

# INTRODUCTION

# Cutaneous Leishmaniasis, an Ancient Disease With an Uneven Global Burden

Leishmaniases include multitudes of infectious diseases caused by the protozoa of Leishmania genus which pose serious public health problems in the endemic regions across 98 countries (Alvar et al., 2012). Of the two main forms of such infections, known as visceral leishmaniasis (VL) and cutaneous leishmaniasis (CL), the latter is the most common form. Leishmaniases are multifactorial diseases which their outcomes are influenced by dynamic interactions of the parasite, its reservoir(s), its vector, and eventually the immune system of its human host. The ecosystems that harbor the above factors can also affect the infectivity of the parasite. For instance, climate changes that favor expansion of deserts can provide a hospitable environment for propagation of the vectors and the reservoirs of the parasites that cause CL. The prevalence of CL in the endemic regions is on the rise due to natural favorable environmental changes, compounded by manmade influences such as global warming, deforestation, regional conflicts, mass migrations, and urbanization (González et al., 2010; Du et al., 2016). Adding to the existing complexity, recent findings have pointed to the genome instability of Leishmania in response to the environmental pressures by high aneuploidy turnover and haplotype selection mechanisms (Prieto Barja et al., 2017).

Earlier, it had been estimated that ∼75% of global CL cases are found in 10 countries, namely Afghanistan, Algeria, Colombia, Brazil, Iran, Syria, Ethiopia, North Sudan, Costa Rica, and Peru (Alvar et al., 2012). A recent comprehensive epidemiological study based on country-level data (**Table 1**) has confirmed that the burden of CL falls mostly on certain developing countries in the Middle East and North Africa in the Old World as well as on a few South American and Caribbean countries in the New World. The common factors shared among all CL endemic areas are high population densities and malnutrition, combined with poor sanitary facilities (Hotez et al., 2012, 2014; Karimkhani et al., 2016). Aside from the economic burden caused by the morbidity of CL, the social stigmatization (Hurrell et al., 2016) and emotional burden due to the ulcers, especially among the inflicted women and children represent other important sequelae associated with the disease (Chahed et al., 2016; Bennis et al., 2017a,b).

# Interaction of Leishmania and the Immune Cells

Leishmania parasite such as Leishmania major, Leishmania tropica, Leishmania braziliensis, L. Mexicana, and L. amazonensis can be transmitted by different zoonotic and anthropozoonotic cycles that may involve domestic and wild mammalian reservoirs via a digenetic life cycle, divided into extracellular, and intracellular stages. The protozoan lives as either a promastigote form inside its female blood-feeding sand fly of Diptera order, belonging to Phlebotomus (Ph.; in the Old World) or Lutzomyia (Lu.; in the New World) genera or an amastigote form inside parasitophorous vacuoles within the phagocytic cells (mostly macrophages) of its vertebrate host (Lestinova et al., 2017). The infection turns into ulcerated skin lesions when the cellmediated immune responses fail to eliminate or control the resident parasites inside the phagocytic cells.

TABLE 1 | The burden of CL in the countries with the highest prevalence in the Old/New Worlds (Karimkhani et al., 2016).


*DALY, Disability-Adjusted Life-Years.*

Phagocytic cells such as neutrophils are deemed to play a crucial role during the infection which can lead to either survival or destruction of the immunologically-evasive Leishmania parasites. Soon after the parasite entry, neutrophils are massively recruited to the site of infection. These cells mediate their defense functions through two different approaches: (i) phagocytosis and killing of the parasites (ii) formation of neutrophil extracellular traps (NETs) and releasing different microbicidal agents associated with DNA backbone (Brinkmann et al., 2004). Through NETs, neutrophils are able to entrap and inhibit the spread of the parasites (Zawrotniak and Rapala-Kozik, 2013). Based on in vitro studies, it has been shown that L. major and Leishmania donovani may survive inside the neutrophils by inhibiting the process of granule fusion with the parasites contained inside the phagosomes (Gueirard et al., 2008; Hurrell et al., 2016). Furthermore, parasites such as L. major and L. braziilensis have been shown to be capable of resisting the microbicide activities associated with NET formation through endonuclease digestion, derived either from the parasites or the sand fly components. Therefore, through different approaches, the parasite succeeds to survive inside the neutrophils (Regli et al., 2017). Depending on the Leishmania species, neutrophils can influence the disease outcome and act as important modulators of leishmaniases (Hurrell et al., 2016). Moreover, skin resident macrophages and dendritic cells (DCs) are shown to be capable of presenting the processed antigens of the amastigotes to T cells; albeit each using different machineries (Cecílio et al., 2014).

Interestingly, after encounter with Leishmania parasites, the biological behaviors of the phagocytes are altered so much as the macrophages become the host cell for the parasite and DCs appear to become mainly responsible for T cell priming and induction of protective immunity (von Stebut and Tenzer, 2017). There is no clear evidence on the fate of Leishmania parasites within the DCs. The importance of such antigenpresenting cells in parasite maintenance and induction of long term memory is also evident for the healed patients of CL and may also have functions for the asymptomatic population. A more comprehensive understanding of the versatile functions of DCs in CL and their interplay with other immune cells may provide new insights into underlying immune mechanisms that control CL.

# BIOMARKERS OF CL

In order to evaluate and compare curative or preventive strategies on people affected by CL, the researchers require defining biomarkers as indicators of normal biological processes, as well as tools to detect them. The application of biomarkers with respect to diseases such as CL can be categorized into different approaches such as diagnostics, prognosis and monitoring the disease progression or the outcome of clinical interventions (Mayeux, 2004). The methodology being used to assess the biomarkers should ideally be safe and sensitive, capable of producing consistent results for different genders and ethnic groups, objectively (Theppeang et al., 2008; Strimbu and Tavel, 2010). Since CL is mostly prevalent in the developing countries, a tool to evaluate a CL biomarker should also be relatively affordable for the afflicted societies.

Biomarkers are often divided into direct and indirect categories in the literature. Concerning CL, factors, products or conditions resulted directly from the infection can be considered as a direct biomarker. Whereas, host's quantitative or qualitative biological reactions due to the leishmanial infection can be considered as indirect biomarkers. Identifying biomarkers associated with diagnostic, treatment, and especially disease outcome may help to develop new ideas or tools for better understanding of the mechanisms behind the protective immune responses in CL. A comprehensive systematic review of potential pharmacodynamic biomarkers of different forms of leishmaniases has been published in recent years (Kip et al., 2015). While few studies on biomarkers associated with VL caused by Leishmania infantum and L. donovani have been conducted (Ibarra-Meneses et al., 2016, 2017), little is known about CL biomarkers in general and asymptomatic CL biomarkers, in particular. Although not exhaustive, we herein will briefly discuss the state-of-the-art for markers of CL. The putative biomarkers based on CL studies, divided into direct and indirect markers, are summarized in **Table 2**.

# Detection of Infection

The determination of presence of a CL-causing Leishmania parasite in tissues can be considered as the most direct biomarker of CL. However, diagnosis of CL is not an easy task due to lesion variation in term of severity, clinical appearance, and duration; therefore, development of sensitive diagnostic biomarkers is in high demand (Akhoundi et al., 2017). Occasionally, diagnostic decision is also complicated due to the similarity of the clinical symptoms, in cases such as bacterial ulcers, leprosy, sarcoidosis, and lupus vulgaris. Traditionally, definitive diagnosis was based on visualization of Leishmania on a direct smear, followed by culturing and animal inoculation. Later, molecular diagnostic methods gained prominence due to their rapidity, sensitivity, and specificity which enabled the investigators to discriminate even among different Leishmania species by using different species-specific probes. There are also non-invasive sampling methods in which the parasite DNA is isolated by sequential tape strips, followed by parasite detection, using PCR (Taslimi et al., 2017). Although there is no specific gold standard technique for detection and diagnosis of Leishmania infection, the complete sequencing of several Leishmania genomes can open new windows for revealing precise diagnostic biomarkers in future (Van der Auwera et al., 2014). A few indirect biomarkers of CL which have been clarified so far are mentioned below.

# Delayed-Type Hypersensitivity

The parasite propagation inside macrophages usually leads to typical ulcers which may last for a year. After healing, such ulcers leave disfiguring scars on the skin. Due to the importance of cell-mediated immunity (CMI) in the outcome of CL (Sacks and Noben-Trauth, 2002), the state of CMI for the suspected CL patients is generally evaluated by testing delayed-type IV hypersensitivity [DTH; (Turk, 1979)] reaction against the leishmanial antigens. Such tests can be performed by Leishmanin skin test (LST; previously known as Montenegro test) that indicates prior encounter of the host immune system with the parasite. A positive DTH response is evidenced by a small induration which forms 72 h after a 100 µl intradermal injection of Leishmanin reagent (a phenolized cultured L. major promastigotes). LST remains a powerful biomarker to distinguish the subpopulations with respect to asymptomatic CL. Moreover, LST results can be interpreted as an indicative of the developed immunological memory (Andrade-Narvaez et al., 2016).

# T-Cell Subpopulations and the Importance of Cytokines as Putative Biomarkers

Since 1980s, investigators have been trying to figure-out distinct functions of thymus-derived T cells and their cytokine profiles with respect to leishmaniases. These murine-based studies, performed with mostly L. major infections, led to the proposition of two counter-regulatory CD4+ T-cell subpopulations, known as Th1 and Th2 which are accounted for controlling resistance


*LST, Leishmanin Skin Test; ADA, Adenosine Deaminase; Treg, Regulatory T Cells; SGH, Salivary Gland Homogenate.*

Bahrami et al. Cutaneous Leishmaniasis Biomarkers

and susceptibility to the infection, respectively (Alexander and Brombacher, 2012). However, no such clear-cut Th1/Th2 phenotype has so far been demonstrated in human CL studies. Exacerbated Th1 cell-mediated immune responses during CL, such as excessive secretion of pro-inflammatory IFN-γ cytokine, have been shown to cause tissue damage and are assumed to contribute to the lesion progress (Maspi et al., 2016).

Besides the importance of Th1 and Th2 functions with respect to immunity to CL, two other T cell subpopulations, namely regulatory T cells (Treg) and Th17 cells have also been identified to regulate immunity to CL. Treg are known to function as the sustainers of tolerance and preventers of excessive damages during the inflammatory responses (Suffia et al., 2005). Evidence indicates that in CL caused by L. braziliensis and L. guyanensis, functional Treg are recruited to the lesion sites (Campanelli et al., 2006; Bourreau et al., 2009). Th17 cells, on the other hand, have been proposed to orchestrate a balance between the pro- and the anti-inflammatory cytokines during CL and also to recruit neutrophils to the site of infection (Gonçalves-de-Albuquerque et al., 2017).

Among the cytokines explored in CL studies, IL-10 appears to serve as a putative biomarker which can exhibit the treatment failure. It has been shown that mRNA level of this regulatory cytokine in CL lesion is positively correlated to unresponsiveness to the treatment (Louzir et al., 1998; Bourreau et al., 2001). The mRNA level of pro-inflammatory cytokine TNF-α in CL lesion biopsy has been shown to be positively correlated with the lesion size (Louzir et al., 1998). While the elevated serum levels of TNF-α has been suggested to be associated with the severity of mucocutaneous leishmaniasis, whether or not this holds true for CL has been a matter of debate (Barral-Netto et al., 1991; Castes et al., 1993; Vouldoukis et al., 1994; Da-Cruz et al., 1996; Kocyigit et al., 2002a,b). Furthermore, IL-6 transcript in CL lesion biopsies and IL-6 protein in the patients' sera have been associated with the lesion size in CL patients (Louzir et al., 1998; Kocyigit et al., 2002a). Unhealed lesions as well as the lesion duration were shown to be correlated with IL-12 p40 (Melby et al., 1996; Louzir et al., 1998).

# Enzymes

Adenosine deaminase (ADA) found in macrophages mediates deamination of the anti-inflammatory nucleoside adenosine to inosine. The level of ADA has been shown to be significantly increased in sera and lymphocytes of CL patients (Erel et al., 1998). In a recent study, it has been shown that serum ADA activity in patients with active VL and post kala-azar dermal leishmaniasis (PKDL) were significantly higher than their respective treated cases as well as the healthy individuals. The ADA activity in PKDL has been found to be decreased gradually during the different phases of treatment which suggests that this parameter can act as a marker of pathogenesis and prognosis of the disease (Vijayamahantesh et al., 2016).

L-arginine is a conditionally essential amino acid that plays a role in many metabolic pathways and serves as a common substrate for both arginase and nitric oxide synthase (NOS). In CL, arginine acts as a double-edged sword so that while it is needed for nitric oxide (NO)-mediated parasite killing, it can promote polyamine-mediated parasite replication (Gogoi et al., 2016). The induction of arginase activity in CL lesions causes the depletion of arginine and hence the reduction of NO content. Altogether, arginase limits the availability of arginine and becomes an agent of immune suppression and impairment of T-cell responses. In line with this notion, it has been shown that the level of arginase activity is significantly increased in sera of CL patients (França-Costa et al., 2015). Moreover, in patients infected with either L. major or L. tropica, the levels of arginase activity in the lesion of those who had acute CL (duration ≤1 year) were higher than those with chronic CL (duration ≥2 years) and also in the skin of the uninfected people (Mortazavi et al., 2016). Interestingly, arginase activity in L. major promastigotes, isolated from CL patients has been shown to be significantly higher than the activity of this enzyme in the non-pathogenic strain of this parasite (Badirzadeh et al., 2017).

# Antibodies Raised Against Sand Fly Saliva

Female phlebotomine sand flies are the only natural vectors of Leishmania species. The importance of sand fly saliva for the establishment of the infection on the feeding sites is welldocumented (Maroli et al., 2013). The skin damage caused by the biting mechanics unleashes a hostile environment for the bloodfeeding insect through hemostasis, inflammation and immune responses. The parasite in such unfavorable milieu is deemed to be counteracted by anti-hemostatic, anti-inflammatory and immunomodulatory components of the sand fly saliva. Several immunomodulatory components common among different sand fly saliva, and in particular enzymes such as apyrase, hyaluronidases, and endonucleases have been shown to regulate both the innate and the adaptive immune responses (Lestinova et al., 2017).

It has been shown that experimental exposure of mice, dogs and humans to sand fly bites induces antibody production after a few weeks; although in such cases the magnitude of the response is dependent on the number of sand fly bites (Abdeladhim et al., 2014). The immune response to specific sand fly saliva component could act as reliable biomarkers of vector exposure. Therefore, identification and characterization of salivary antigens which are specific to a particular vector species can be utilized as an exposure biomarker. The salivary antigen PpSP32 is the immunodominat target of antibody response to Phlebotomus papatasi bites in human. The antibody against recombinant PpSP32 (rPpSP32) has been recently tested as a putative biomarker of exposure to sand fly in Tunisia (Marzouki et al., 2012). However, the anti PpSP32 response has not been documented in dogs immunized with saliva of Ph. perniciosus (Marzouki et al., 2015). There are other studies in South America where it has been shown that antibody against LJM11 and LJM17 from Lutzomyia longipalpis saliva could act as putative biomarkers of vector exposure for both humans and dogs (Teixeira et al., 2010). These two proteins belong to the yellow family and are abundant in sand fly and absent from saliva of mosquitoes (Xu et al., 2011; Abdeladhim et al., 2014). It is also noteworthy that the kinetics and duration of antibody responses could also be considered as a putative biomarker of exposure to the vector or the parasite.

# Putative Biomarkers for Asymptomatic CL Infection

Although the minority of people living in the endemic areas succumb to CL and manifest ulcers, a significant proportion of individuals exhibit a positive DTH response without any skin ulcer. The underlying mechanism through which asymptomatic individuals control the infection is not fully understood. Understanding the immunological basis of CL asymptomaticity represents an unmet need, which if negotiated, could inform the development of tools to detect and control CL. Notwithstanding, it is becoming increasingly clear that addressing multifactorial complex diseases such as leishmaniases which are affected by diverse immunological, genetic, and environmental factors would benefit from an unbiased systems analysis approach.

It is well-documented that after clinical cure of certain intracellular pathogens, such as herpes viruses, Mycobacteria, Chlamydia, and Trypanosoma, a long-term persistence of these pathogens can be detected in the body (Mendonça et al., 2004). Due to the lack of adequate diagnostics tools, it is however challenging to differentiate people with persistent Leishmania infection from those with immunologically-sensitizing exposures (Andrade-Narvaez et al., 2016). The existence of live Leishmania parasites of various strains has been reported after clinical cure by chemotherapy. In one study, by applying precise molecular approaches, the presence of live parasites has been confirmed in mucocutaneous leishmaniasis caused by Leishmania species of the Viannia subgenus. The presence of Leishmania has been demonstrated in blood monocytes, tonsils and normal skin of clinically-cured as well as the asymptomatic individuals using amplification of 7SLRNA gene. As mentioned earlier, the immune responses are playing outstanding roles in this regard. For instance, the monocyte-derived macrophages from people with asymptomatic CL due to L. braziliensis are shown to control the parasite much stronger than macrophages from patients with lesions. These data besides many others may show the intrinsic differences between the innate immune responses (Giudice et al., 2012; Scorza et al., 2017).

The frequency and suppressive capacity of regulatory T cells (CD4+CD25high FoxP3+) have been shown to be comparable in the peripheral blood of the healed and the asymptomatic individuals in an endemic area of L. major infection in Iran. However, mRNA level of the Treg transcription factor FOXP3 was shown to be higher in the asymptomatic group compared to the symptomatic patients. The mRNA level of IL-17 transcription factor RORC, on the other hand, was not significantly different between the groups (Bahrami et al., 2014).

Recently, a cross-sectional immune profiling study on PBMC of individuals living in endemic areas of L. major transmission in Tunisia with documented exposure history was conducted (Kammoun-Rebai et al., 2016). The participants were subdivided according to their LST responses and the presence of scar, into three groups of healed, asymptomatic and naïve individuals. Significant Leishmania-specific responses in LST+SCAR+ and LST+SCAR- individuals, but not in LST-SCAR- (naïve subjects), were substantiated. Similarly, the levels of IL-2, IL-12p70, IL-13, and IL-18 were shown to be significantly higher in all LST+ individuals but not in LST- cohort. Once validated, such biomarkers could be potentially used to stratify patients for different clinical/sub-clinical manifestations of the disease.

Although LST is considered safe, its application requires an invasive injection of the killed parasite in a phenol solution. As a potential alternative to LST, salivary gland homogenates (SGH), or proteins thereof may be used to determine the risk of contact, vector surveillance and as such could inform the disease management programs (Oliveira et al., 2013a). In humans, the bite of sand flies has been shown to induce DTH at the biting sites and can generate specific antibody responses, detectable in serum (Oliveira et al., 2013b). As such, DTH and specific antibodies can presumably be also used as markers of exposure to the vector's bite in humans and animals. As mentioned above, the recombinant PpSP32 from Ph. papatasi could act as a serum biomarker of exposure (Marzouki et al., 2012). Hence, identification of different biomarkers from various infective sand flies combined with the development of simple and robust tools such as those based on finger-prick capillary sampling methods, and autoreactive strips with already-fixed rPpSP32, could facilitate the screening of the target population in the endemic areas.

# Asymptomatic CL Patients Serve as Potential Reservoirs

Characterization of Leishmania in asymptomatic infection is technically challenging due to the limitation in the detection of low number of the parasites. Importantly, the low number of the parasites present in the skin of the asymptomatic individuals may act as an infection reservoir. PKDL is a good example of such situations in which an asymptomatic infection can convert into PKDL (Saha et al., 2017). Therefore, development and use of precise and sensitive molecular tools could help identifying asymptomatic individuals with low-level of infection, which could in turn facilitate the development of measures to control CL.

There is a dearth of documents regarding the parasite detection in either skin or blood of the asymptomatic patients of the Old World CL. Of note, the detection of L. tropica and L. major in healed CL individuals with various durations after the recovery has recently been reported (Taslimi et al., 2017). The presence of Leishmania parasites in unaffected skin and peripheral-blood monocytes of American CL due to Leishmania (Viannia) species has also been reported (Vergel et al., 2006). The Leishmania kinetoplast DNA has been detected by PCR from scar and peripheral blood, several years after the healing. In addition, the presence of Leishmania (Viannia) parasites in tissues and its availability in tissues accessible to sand flies were demonstrated in that study. By considering two body sites consisting of the border of an active lesion and a normal skin site, viable parasites were detected in the unaffected skin of these individuals (Rosales-Chilama et al., 2015). Considering the aforementioned evidence, and the similarities of the immune responses between the healed and the asymptomatic populations in CL endemic areas, it is plausible that Leishmania species persist inside the skin of the asymptomatic individuals. To substantiate this, efforts should be made to analyze the skin and blood of the asymptomatic subjects with confirmed DTH, in the highly endemic areas.

# FUTURE DIRECTIONS

While few indicative biomarkers of CL infection and immunity have been proposed using conventional approaches, systemslevel data integrating different layers of information on host response to CL infection are scarce. Such information can provide a comprehensive understanding of the disease and may presumably identify key biomarkers of CL disease and immunity. The recent technological advances in highthroughput omics technology combined with systems biology approaches have begun to provide new insights into pathogenesis of different diseases, and to unravel the molecular signatures of vaccine-induced responses in humans. Recently, whole genome transcriptomics analysis has been applied to CL. We have reported the transcript changes in the lesion of patients infected with L. tropica as compared with the healthy normal skin (Masoudzadeh et al., 2017). We have identified key immunobiological pathways that are regulated following L. tropica infection during the acute phase of the disease. Another recent transcriptomics study of lesion biopsies from L. braziliensis patients has identified B cell activation and immunoglobulin transcript signatures in the lesions, depending on the presence or absence of the parasite transcripts (Christensen et al., 2016). Based on the rapid pace of development in the field, it is envisaged that more putative biomarkers with the power to predict the outcome of CL infection caused by different Leishmania species will be identified (Patino and Ramírez, 2017). In interpreting the biomarker data, caution should however be exercised as the identified biomarkers

# REFERENCES


may merely serve as surrogate markers of infection or immunity, and as such may not necessarily play a causal role in pathogenesis of or immunity to CL.

Once available, such key information on biomarkers could inform the design of rational diagnostics and intervention strategies to control CL. In particular, this can help to screen the populations living in the endemic areas for asymptomatic individuals, which can in turn provide invaluable information on how gender, ethnic or geographical factors would affect the asymptomaticity. To achieve this, accurate, non-invasive and rapid methods with the ability to distinguish latent CL from the asymptomatic CL are needed. It should be noted that more awareness at global and national levels is also required to pave the way for developing cost-effective diagnostics and intervention strategies for the expanding inflicted populations.

# AUTHOR CONTRIBUTIONS

FB, AH, and SR drafted the manuscript. All the authors provided critical feedback on the manuscript prior to publication and have agreed to the final content.

# FUNDING

SR is supported by research grant ID 940007 from National Elites Foundation Presidency of Islamic Republic of Iran, FB is supported by research grant ID 760 from Pasteur Institute of Iran and AH is supported by the Innovative Medicines Initiative, European Commission under the BioVacSafe (grant agreement no. 115308), VSV-EBOVAC (grant agreement no. 115842), and VSV-EBOPLUS (grant agreement no. 116068) consortia.

# ACKNOWLEDGMENTS

The authors wish to thank H2020-MSCA-RISE-2017 supported, LeiSHield-MATI consortium (grant agreement number 778298).

isolated from patients with cutaneous leishmaniasis. Parasite Immunol. 39. doi: 10.1111/pim.12454


cure of American cutaneous leishmaniasis: is there a sterile cure? J. Infect. Dis. 189, 1018–1023. doi: 10.1086/382135


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

# Biomarkers for Zoonotic Visceral Leishmaniasis in Latin America

### Claudia I. Brodskyn<sup>1</sup> \* and Shaden Kamhawi <sup>2</sup>

1 Instituto Gonçalo Moniz, Fiocruz Bahia, Salvador, Brazil, <sup>2</sup> National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health, Bethesda, MD, United States

In Latin America, zoonotic visceral leishmaniasis (ZVL) arising from infection by L. infantum is primarily transmitted by Lutzomyia longipalpis sand flies. Dogs, which are chronic reservoirs of L. infantum, are considered a significant risk factor for acquisition of ZVL due to their close proximity to humans. In addition, as a vector-borne disease the intensity of exposure to vector sand flies can also enhance the risk of developing ZVL. Traditionally, IFN-γ and IL-10 are considered as the two main cytokines which determine the outcome of visceral leishmaniasis. However, more recently, the literature has demonstrated that different mediators, such as lipid mediators (PGE-2, PGF-2 alfa, LTB-4, resolvins) and other important inflammatory and anti-inflammatory cytokines are also involved in the pathogenicity of ZVL. Analysis of a greater number of mediators allows for a more complete view of disease immunopathogenesis. Additionally, our knowledge has expanded to encompass different biomarkers associated to disease severity and healing after specific treatments. These parameters can also be used to better define new potential targets for vaccines and chemotherapy for ZVL. Here, we will provide an overview of ZVL biomarkers identified for both humans and dogs and discuss their merits and shortcomings. We will also discuss biomarkers of vector exposure as an additional tool in our arsenal to combat ZVL.

### Edited by:

Eugenia Carrillo, Instituto de Salud Carlos III, Spain

### Reviewed by:

Sima Rafati, Pasteur Institute of Iran, Iran Ricardo Silvestre, Instituto de Pesquisa em Ciências da Vida e da Saúde (ICVS), Portugal

### \*Correspondence:

Claudia I. Brodskyn brodskyn@bahia.fiocruz.br

Received: 27 April 2018 Accepted: 25 June 2018 Published: 26 July 2018

### Citation:

Brodskyn CI and Kamhawi S (2018) Biomarkers for Zoonotic Visceral Leishmaniasis in Latin America. Front. Cell. Infect. Microbiol. 8:245. doi: 10.3389/fcimb.2018.00245 Keywords: zoonotic visceral leishmaniasis, Leishmania infantum, biomarkers, cytokines/chemokines, canine visceral leishmaniasis, human visceral leishmaniasis

# INTRODUCTION

Leishmaniasis is considered a neglected tropical disease with approximately 350 million people at risk of infection, and with 2 million new cases reported annually, mainly in extremely impoverished communities (WHO/Leishmaniasis, 2014). The clinical manifestations of leishmaniasis range from cutaneous ulcers to the visceral form, one of the most severe which can be fatal if left untreated (Desjeux, 2004). Over 90% of visceral leishmaniasis (VL) cases worldwide are concentrated in six countries: India, Bangladesh, Sudan, South Sudan, Ethiopia, and Brazil. VL arises from either L. donovani in the Indian subcontinent and East Africa, or L. infantum in Europe and Latin America, and kills 20,000–40,000 people annually throughout the world (Alvar et al., 2012).

Dogs are considered the main reservoirs of L. infantum parasites and the presence of these animals in endemic areas represents a risk factor for human disease. The disease in dogs shares characterisitcs with human VL, providing a good model to study the immunopathogenesis of L. infantum infections. Canine visceral leishmaniasis (CVL) is also of veterinary interest since the disease is spreading to big cities in Latin America including Belo Horizonte, São Paulo, Natal, and Camaçari (Alves and Bevilacqua, 2004), threatening public health.

L. infantum multiplies inside macrophages in the liver, spleen and bone marrow. In human VL, 90% of infected individuals remain asymptomatic or subclinical, showing an intense cellular immune response characterized by a positive delayed-type hypersensitivity reaction to Leishmania antigens. However, in patients who progress to clinical disease, an enlargement of the spleen and liver may be observed, accompanied by hematological disorders, notably anemia, thrombocytopenia, which may result in hemorrhaging, and neutropenia. These disorders can increase host susceptibility to bacterial infection and patients with VL often suffer weight loss or fever (Werneck et al., 2003).

The search for biomarkers for prognosis of human and canine VL and measurement of the success of treatment has intensified in the last few years. This led to a better knowledge of disease immunopathogenesis and favored therapeutic and prophylactic strategies for ZVL. As L. infantum is transmitted by the bite of phlebotomine sand flies, several markers of vector exposure have also been identified and used as tools to assess success of interventions. Used in combination, biomarkers of disease and vector exposure can provide powerful tools to support efforts to control ZVL.

# HUMAN ZVL

The progression of ZVL is associated with immunosuppression, characterized by lack of a cell-mediated immune response to Leishmania antigens. Accordingly, the absence of lymphocyte blastogenesis and IFN-γ production have been associated with progression to VL (Carvalho et al., 1992). Interestingly, patients become responsive to Leishmania antigens after successful therapy (Carvalho et al., 1989). Biomarkers were evaluated in a group of ZVL patients given standard antimonial treatment (Schriefer et al., 1995). Soluble CD4 (sCD4), sCD8, and sIL-2R levels were higher in sera of patients compared to healthy controls. After treatment, levels of the above-mentioned biomarkers exhibited a significant decrease in patients who responded to therapy. Importantly, when comparing pretreatment levels of these markers among those who responded to antimonial therapy and refractory patients, the serum concentrations of sCD8, sIL-2R as well as neopterin were significantly elevated in refractory patients. Therefore, these markers represent promising indicators of a patient′ s response to antimonial therapy (Schriefer et al., 1995).

The cure for human ZVL is associated to induction of a Th1 immune response characterized by IFN–γ production. IFN– γ has an essential role in controlling the parasite load and in the development of a long-lasting immunity. Asymptomatic patients also exhibit a Th1 response, suggesting that IFN-γ activates macrophages, increasing their leishmanicidal ability and maintaining the infection under control (Kaye and Scott, 2011). In contrast, anti-inflammatory cytokines, mainly IL-10, lead to proliferation of parasites and interfere with infection control (Nylen and Sacks, 2007; Gautam et al., 2011). In a study of patients with active ZVL, in vitro stimulation of PBMC with Leishmania antigens showed an inverse pattern with low levels of IFN-γ during active disease that augmented steadily after treatment (Caldas et al., 2005). Interestingly, these patients showed elevated plasma levels of IFN-γ, IL-12p40, and IL-10 during active disease that sharply decreased after treatment (Caldas et al., 2005). Therefore, IFN-γ and IL-10 are the main hallmarks of infection by L. infantum, and the balance between these cytokines seems to be essential for control of the infection.

IL-17, a cytokine produced mainly by Th17 cells, is known for inducing the production of chemokines that recruit neutrophils to inflammatory sites. A cohort of individuals with VL caused by L. donovani showed that IL-17 seems to be protective (Pitta et al., 2009). However, in ZVL patients caused by L. infantum, high levels of this cytokine did not induce IFNγ/NO in enough concentrations to lead to a recovery from disease (Nascimento et al., 2015).

In severe forms of ZVL, there is an exaggerated inflammatory response that leads to disseminated intravascular coagulation and other manifestations such as hemorrhage (Costa et al., 2013). Children displaying an intense production of cytokines have a higher risk of death. IL-6 seems to be one of the main cytokines associated with fatal disease, but IFN-γ, IL-1β, IL-8, and TNF-α have also been associated to ZVL severity (Costa et al., 2013). In fact, L. infantum may activate inflammatory reactions via CD14, leading to a production of several cytokines such as IFN-γ, IL-27, IL-10, IL-6 as well as sCD14 (Dos Santos et al., 2016). These data reinforce previous results and denote the interdependent relationship between pro-inflammatory (IFN-γ, IL-6 and TNF-α) and anti-inflammatory (IL-10 and IL27) cytokines. The authors also showed that higher levels of IL-6 (>200 pg/ml) are associated to death (Dos Santos et al., 2016). Collectively, this highlights the complexity of finding a good biomarker for ZVL since induction of a cytokine such as IFN-γ may be associated to protection or severe disease, depending on its levels and the overall inflammatory environment.

Other biomarkers, including lipid mediators, have been associated with ZVL and could also be used to monitor the efficacy of specific therapies. Araujo-Santos et al. (2017) reported a distinct biosignature of active ZVL through increased serum levels of Prostaglandin F 2 alfa (PGF2α), Leukotrine B4 (LTB4), Resolvin D1 (RvD1), TNF-α, IL-1β, IL-6 and IL-8, IL-10, and IL-12p70, as well as decreased concentrations of TGFβ1 in comparison to healthy endemic controls, regardless of patient age or gender. Following the onset of leishmanicidal treatment, the inflammatory cytokine profile, as well as the relationships between these markers and several hematological and biochemical parameters, gradually reverted, which suggested that the observed cytokine expression profile was induced by active disease or infection (Araujo-Santos et al., 2017). Of the quantified markers, TGF-β1 concentrations were significantly elevated, while IL-6, IL-8, IL-10, and RvD1 levels substantially decreased, after 30 days of therapy in comparison to their levels during active ZVL infection.

It is worthwhile noting that common immunological signatures were observed in sera of VL patients from Brazil and Bangladesh infected with L. infantum and L. donovani, respectively. Inflammatory and regulatory cytokines (IFNγ, TNFα, IL-10, IL-17), as well as levels of growth factors (FGFfibroblast growth factor; VEGF, vascular endothelial growth factor), were elevated in the serum of VL patients from both regions (Duthie et al., 2014). Serological assays from Brazilian patients obtained during and after meglumine antimoniate treatment demonstrated that multiple parameters reverted to concentrations similar to healthy endemic controls. The authors suggested that a multi-parameter signature of the response to treatment could be useful in clinical trials to evaluate the success of therapeutic interventions (Duthie et al., 2014).

Macrophages infected by L. infantum present an M2b-like phenotype as well as a C-type lectin receptor (CLR) signature, characterized by Dectin-1, mannose receptor and DC-SIGN homolog SIGNR3 expression (Lefevre et al., 2013). Expression of Dectin-1 and the mannose receptor are essential to the leishmanicidal effect of macrophages, leading to the production of ROS and also the induction of IL-1β secretion. On the other hand, SIGNR3 was shown to favor parasite survival via the inhibition of the LTB-4-IL1β axis (Lefevre et al., 2013).

More recently, MCP-1 was shown to be a good biomarker to identify asymptomatic individuals infected by L. infantum (Ibarra-Meneses et al., 2017). This chemokine is expressed 110 times more strongly than IL-2 in cultures of whole blood stimulated with Leishmania antigens, identifying 87.5% of asymptomatic subjects; it is also significantly increased in all patients cured of ZVL (Ibarra-Meneses et al., 2017). **Table 1** summarizes the main findings about biomarkers in human ZVL.

In summary, biomarkers other than IFN-γ and IL-10 have been described more recently that reveal the complexity of ZVL. These molecules have demonstrated both their value as biomarkers of disease progression and their usefulness in monitoring the efficacy of treatment. In the future, such biomarkers may also be of value in assessing the level of protection induced by prophylactic strategies.

# CANINE VL

Dogs are one of the main urban reservoirs of L. infantum parasites and their presence in endemic areas is a risk factor for the development of human disease, due to their role in propagating infection in phlebotomine sand flies. Clinical manifestations of CVL present a wide spectrum of clinical signs that are non-specific. However, a high proportion of animals do not progress to disease, control the parasites and live for years or their entire life without any clinical signs (Foglia Manzillo et al., 2013). The presence of these dogs in the endemic area contributes to maintenance of the parasites, since they can transmit L. infantum to the sand flies. The resistance or susceptibility to CVL is directly correlated with the induction of either a Th1 response characterized by IFN- γ, IL-2, and TNF-α production, or a Th2 response with the production of IL-4, IL-5, IL-10, IL-13, and TGF-β, respectively, and the level of immune activation is considered to directly influence disease severity (Reis et al., 2010; Barbosa et al., 2011). In fact, a reduction in the burden of L. infantum was related to elevated expression of IFN-γ and TNF-α, whereas increased IL-10 and iron regulatory protein 2 (IRP2) expression and an increase in plasma albumin levels were associated with a higher parasite burden (do Nascimento et al., 2013). Accordingly, the dynamics between different aspects of the immune response and intracellular iron availability could play some role in the evolution of Leishmania infection (do Nascimento et al., 2013).


The predominance of a Th2 response in infected dogs leads to the appearance of M2 macrophages, identified by CD163 immunostaining, mainly in the spleen, muzzle, ear and popliteal and pre-capsular lymph nodes (Moreira et al., 2016). The highest proportion of M2 macrophages coincided with the highest parasite loads and were found in more susceptible organs of infected dogs such as the spleen and lymph nodes, as well as skin, considered a more resistant organ. In contrast, the liver showed low parasitism and weak immunostaining for M2 macrophages that was not significantly different between infected and negative groups of dogs. Therefore, M2 macrophages may contribute to parasite proliferation in organs (Moreira et al., 2016). Another important point that contributes to the gravity of CVL is associated with the disorganization of splenic tissue. Dogs that presented positive spleen cultures also showed more disrupted spleen architecture, with a lower concentration of serum albumin and creatinine and higher levels of aspartate aminotransferase. Together, these data suggest that the disorganization of lymphoid tissue in the spleen is linked to more severe clinical presentations of CVL (Lima et al., 2014).

The immunosuppression observed during CVL is also correlated to the occurrence of CD4+ and CD8+ T cell exhaustion and could be responsible for the absence of specific antigen blastogenesis and IFN-γ secretion. Moreover, there was a significant increase in the surface expression of programmed death 1 (PD-1) on T cell populations, mainly in CD8 T cells of symptomatic dogs in comparison to control animals. Using monoclonal antibodies able to inhibit the PD-1 ligand B7.H1 restored CD4<sup>+</sup> and CD8<sup>+</sup> T cell function and raised the levels of reactive oxygen species in cocultures of phagocytes and T cells. Consequently, these macrophages showed a reduction in the number of intracellular parasites. The T cell exhaustion resulting from symptomatic CVL may affect the response to vaccination and the efficacy of treatments used to control L. infantum (Esch et al., 2013).

Different reports in the literature have associated the parasite burden in different tissues with the host immune response, evaluating pro- and anti-inflammatory cytokines. Generally, there is a consensus that a mixed Th1/ Th2 response is observed at the time of detection of a Leishmania infection, despite the prevalence of a Th2 profile. Later in the infection, when the parasite load decreases, e.g., as a result of treatment, the Th1 profile becomes predominant. However, this is not always the case. In a study of 20 infected dogs treated with miltefosine and allopurinol, 80% of animals showed expansion of the parasite load in blood and lymph nodes and low IFNγ levels at the end of the 9–12 months study period (Manna et al., 2008). This response signals a failure to induce a Th1 response in the majority of treated dogs. Nevertheless, these animals did not show any sign of disease, suggesting they are in an asymptomatic state (Manna et al., 2008). The above highlights the complexity of CVL and the absence of reliable markers of recovery of infected dogs. In another study, quantifying TNF-α, IL-4, and IL-10 in the spleen and liver of dogs naturally infected with L. infantum, with or without clinical signs, showed that the animals exhibited higher levels of these cytokines compared to control non-infected dogs (DE F Michelin et al., 2011). Interestingly, the authors found that the liver is the main organ responsible for the production of cytokines during infection. Moreover, TNF-α was positively correlated with parasite burden and could represent a marker for disease infection, with the participation of IL-10 (DE F Michelin et al., 2011).

More recently, the parasite burden and levels of IFN-γ, TNF-α, IL-10, and TGF-β were evaluated in 5 target tissues at 6 and 16 months after infection with L. infantum in an experimental canine model (Rodriguez-Cortes et al., 2016). The data showed that the spleen and liver of infected animals exhibited a high parasite density at both time points and produced pro- and anti-inflammatory responses (Rodriguez-Cortes et al., 2016). The popliteal lymph nodes produced IFNγ both at the beginning of the infection and in the chronic phase. In contrast, an increase in IL-10 and TGF-β expression in these organs was only observed in the chronic phase. Of note, cytokines were absent in the skin, although parasites were detected at 6 months post-infection. Therefore, in the above-mentioned study, the spleen and liver of infected dogs produced diverse cytokines at early times of infection, whereas an anti-inflammatory profile was observed in peripheral tissues at later periods, considered as the chronic phase of infection (Rodriguez-Cortes et al., 2016). Another study of experimentally infected dogs, considered asymptomatic due to the few clinical signs presented, animals were maintained for six years after intradermal infection (Abbehusen et al., 2017). Although few clinical signs were noticed in these animals, most presented parasites in the lymph nodes, spleen and skin and exhibited an increase in IFN-γ, GM-CSF, IL-6, and IL-18 levels and a decrease in TNF, IL-2, and CXCL1 serum concentrations. These results seem to suggest that a persistent activation of the immune system in subclinical infections with L. infantum may possibly control parasite growth and limit disease severity (Abbehusen et al., 2017).

Our group studied 70 naturally infected dogs from an endemic area in Bahia, Brazil, grouping the animals according to a clinical score previously described by Manna et al (Manna et al., 2009) with slight modifications. In the group of dogs with severe disease (clinical score >7), we observed a reduction in the levels of serum LTB-4 and PGE-2 and an elevation in chemokine concentrations such as CXCL1 and CCL2 (Solca et al., 2016). Performing ROC curves, we observed that a combination of LTB-4, PGE-2, and CXCL-1 differentiated best among distinct groups of dogs with different clinical scores. Besides that, analysis of the interactome of the different mediators evaluated showed that LTB-4, a lipid mediator, had the highest number of interactions with other cytokines and chemokines in the group of dogs with severe disease. Although IFN-γ and IL-10 were also evaluated, we did not find an important role for these cytokines as markers of disease progression in infected dogs (Solca et al., 2016). **Table 2** summarizes the main findings about biomarkers in canine ZVL.

In CVL, the complexity of the immune response and the role played by different mediators during infection by L. infantum is clear. Although IFN-γ and IL-10 seem to be important cytokines TABLE 2 | Biomarkers in canine visceral leishmaniasis.


and are induced during infection, other mediators could be more useful as markers of disease progression or in the assessment of the response to treatment. An important point that deserves consideration is the presence of parasites in different organs of infected dogs, including skin. As such, asymptomatic dogs could represent a good source of parasites to uninfected sand flies contributing to their persistence in endemic areas. These findings indicate that the search for markers of exposure to sand flies is essential for surveillance of endemic areas.

# BIOMARKERS OF VECTOR EXPOSURE

In large part, "biomarkers" is a term mostly used to describe characteristics of a disease. However, for vector-borne diseases they can also reference exposure to the arthropod vectors that transmit them. This relatively new discipline of research has gained momentum in recent years and is proving to be of value to various approaches aiming at disease treatment or control (Andrade and Teixeira, 2012; Doucoure and Drame, 2015; Lestinova et al., 2017). In contrast to disease biomarkers, good vector exposure biomarkers may be used as preventative tools that can identify a heightened risk for contracting disease (Carvalho et al., 2015; Ya-Umphan et al., 2017). Further, good biomarkers of exposure may also be used to monitor the expansion or contraction of vector ranges over time, an important tool in light of climate change and its effect on changing the global distribution of disease vectors (Parham et al., 2015; Purse et al., 2015).

The basic requirements that define a good disease biomarker also apply to biomarkers of vector exposure. A reliable biomarker should be specific to a particular vector species (Poinsignon et al., 2008; Teixeira et al., 2010; Ali et al., 2012; Doucoure et al., 2012; Zhao et al., 2015), and recognized by the majority of the exposed or target host population (Souza et al., 2010; Marzouki et al., 2012; Doucoure and Drame, 2015; Mukbel et al., 2016). Additionally, an ideal vector biomarker should be relatively short-lived in the absence of exposure, a desired feature for use in biomonitoring (Clements et al., 2010; Gidwani et al., 2011; Noukpo et al., 2016). For arthropods that transmit pathogens by bite, which represents the majority of disease vectors including phlebotomine sand flies, saliva has been the primary target for biomarkers of vector exposure (Andrade and Teixeira, 2012; Courtin et al., 2015; Doucoure and Drame, 2015).

Most vectors bite mammalian hosts to acquire blood. As such, saliva of vectors has evolved to facilitate blood feeding (Ribeiro and Francischetti, 2003; Lestinova et al., 2017) However, in addition to their physiological effects, several salivary molecules of major disease vectors are immunogenic in humans as well as animal reservoirs and have the potential to become biomarkers of exposure. (Andrade and Teixeira, 2012; Dama et al., 2013a; Abdeladhim et al., 2014; Doucoure and Drame, 2015). Considering that vector control continues to be considered one of the most effective methods to inhibit the transmission of vector-borne diseases, developing immunogenic salivary proteins as biomarkers of vector exposure represent a powerful tool in our arsenal toward their control.

Measuring the antibody response to total saliva, though useful, has serious limitations, mostly due to cross-reactivity of some antigens in various sympatric vectors or host-biting non-vectors (Andrade and Teixeira, 2012; Dama et al., 2013a,b; Doucoure and Drame, 2015; Simo et al., 2017). There are also technical limitations to the use of total saliva preparations as marker of vector exposure including reproducibility and scale-up. This prohibits its consideration for use in largescale epidemiological surveys. To overcome such obstacles, well-defined salivary antigens or peptides that are specific to a particular vector species, while maintaining a measurable immunogenicity in the target population, are being pursued (Rohousova et al., 2005; Poinsignon et al., 2008; Teixeira et al., 2010; Ali et al., 2012; Zhao et al., 2015; Sima et al., 2016).

# Biomarkers of Exposure to Vector Sand Flies

Sand fly saliva consists of a relatively small number of secreted proteins that are injected into the skin in small quantities (Abdeladhim et al., 2014). Nevertheless, sand fly saliva is highly immunogenic and induces a potent cellular and humoral immune response in humans and animals including dogs, known reservoirs of L. infantum (Andrade and Teixeira, 2012; Abdeladhim et al., 2014; Lestinova et al., 2017). Though it has been well established that induction of a saliva-specific Th1-biased cellular immunity protects from disease, salivaspecific antibodies have been associated to both protection, for vectors of visceral leishmaniasis (Barral et al., 2000; Gomes et al., 2002; Aquino et al., 2010; Vlkova et al., 2011), as well as an enhanced risk of infection, for vectors of cutaneous leishmaniasis (Mondragon-Shem et al., 2015). To date, we have no clear understanding of the reason behind these contrasting associations.

Whether antibodies to saliva are associated to protection or risk, they remain a good indicator of the rate of exposure to bites of vector sand flies. However, as for other vectorborne diseases, there are more than one man-biting species of sand flies in endemic areas, and sympatric vectors transmitting different types of leishmaniasis are not uncommon (Rohousova et al., 2005; Clements et al., 2010; Teixeira et al., 2010; Marzouki et al., 2012). This increases the risk of cross-reactive antigens between different species of sand flies and decreases from the efficacy of total saliva as reliable biomarkers of vector exposure. For this reason, defined salivary antigens specific to a particular vector species are being developed as markers of exposure in humans and reservoirs (Andrade and Teixeira, 2012; Lestinova et al., 2017). One of the most promising markers of human exposure to vector sand flies is PpSP32, a salivary protein from Phlebotomus papatasi. PpSP32 was identified as an immunodominant antigen in a naturally exposed population in Tunisia and was later validated in largescale studies of endemic populations in Tunisia and Saudi Arabia (Marzouki et al., 2012, 2015; Mondragon-Shem et al., 2015).

# Biomarkers of Exposure to Bites of Lutzomyia longipalpis, the Principal Vector of ZVL in Latin America

Lutzomyia longipalpis, the main vector of ZVL in Latin America, has a wide distribution range extending from Mexico to Uruguay (Lainson and Rangel, 2005; Brazil, 2013). The domestic dog is considered a main reservoir host and is a major source of sand fly infection and human disease (Lainson and Rangel, 2005; Roque and Jansen, 2014). As such, it is clear that developing reliable biomarkers of exposure to L. longipalpis for humans and dogs would provide a useful epidemiological tool for monitoring ZVL in Latin America. Two studies have demonstrated the immunogenicity of L. longipalpis saliva in endemic populations, associating a positive saliva-specific antibody response to protection against ZVL (Barral et al., 2000; Aquino et al., 2010). The screening of recombinant proteins representing major secreted salivary molecules in L. longipalpis saliva revealed LJM11 and LJM17 as potential markers of vector exposure for both humans and dogs (Teixeira et al., 2010). Both LJM11 and LJM17 belong to the yellow family of salivary proteins that bind biogenic amines blocking their hemostatic-restoring activity during feeding (Abdeladhim et al., 2014). Relevant to their function as biomarkers of exposure, LJM11 and LJM17 are abundant in sand fly saliva and absent from saliva of other common insects such as mosquitoes (Abdeladhim et al., 2014). However, for use as specific markers of exposure to L. longipalpis saliva, absence of cross-reactivity with their homologs in saliva of other sympatric sand fly species needs to be demonstrated. Both LJM11 and LJM17 are not recognized by sera of humans bitten by L. intermedia, a vector of cutaneous leishmaniasis in Brazil whose distribution commonly overlaps with L. longipalpis (Teixeira et al., 2010). Further, only one of two human sera that are strongly reactive to L. longipalpis saliva weakly recognized two antigens in L. intermedia saliva (Teixeira et al., 2010). A large-scale study further validated the immunogenicity and specificity of LJM11 and LJM17 as markers of exposure to L. longipalpis saliva (Souza et al., 2010). Moreover, sensitivity of the assay was enhanced to a level comparable to that against total saliva by using a combination of both LJM11 and LJM17 (Souza et al., 2010).

In contrast to other zoonotic Leishmania infections, canids are considered the only reservoir of L. infantum infection. As such, developing a marker of vector exposure to L. longipalpis for dogs is indicated. LJM11 and LJM17 and two other salivary proteins from L. longipalpis saliva, LJL143 and LJL23, were highly immunogenic in dogs (Collin et al., 2009; Teixeira et al., 2010). LJL143, also called Lufaxin, has dual anticoagulant and anti-complement activities (Collin et al., 2012; Mendes-Sousa et al., 2017), while LJL23 is an Apyrase, an inhibitor of platelet aggregation (Teixeira et al., 2010; Abdeladhim et al., 2014). Interestingly, SP01B and SP01, Apyrase homologs from saliva of P. perniciosus, one of the primary vectors of L. infantum in Europe, were also recognized as markers capable of detecting vector exposure in dogs (Abdeladhim et al., 2014; Lestinova et al., 2017) Moreover, SP03B, a yellow protein from P. perniciosus saliva was also identified as a good marker of vector exposure in dogs (Kostalova et al., 2017). However, to our knowledge, the specificity of SP01B, SP01, and SP03B to saliva of P. perniciosus have not been tested in areas where other dog-biting sand fly species are prevalent. Though geographically separated, it is interesting to note that molecules belonging to the yellow family of proteins and Apyrases of both L. longipalpis and P. pernicious were identified as promising candidates for vector exposure to ZVL in both humans and dogs (Teixeira et al., 2010; Lestinova et al., 2017). In addition to their use to assess human and dog exposure to sand fly bites, biomarkers can also be developed for use in sentinel animals that are common around humans, such as chicken or domestic ungulates, as indicators of vector prevalence (Soares et al., 2013; Rohousova et al., 2015).

The above-mentioned defined biomarkers of L. longipalpis bites are highly promising, however, considering the richness of the sand fly fauna in Latin America, they require further validation of their specificity against salivary homologs from other dominant man- or dog-biting sand fly species. Having said that, the level of specificity of a biomarker should be primarily indicated by its intended use and by the nature of the endemic area. A biomarker can target a particular vector, several vectors, or simply sand fly bites.

Another significant utility for biomarkers of vector exposure is biomonitoring. For this, knowledge of the kinetics of antibody induction and decay is necessary. Previous studies in humans and dogs have demonstrated that saliva-specific antibodies correlate to biting intensity, fluctuate with the sand fly season and are of relatively short duration, declining significantly after bite cessation caused by sand fly seasonality, use of nets or removal from endemic areas, thus demonstrating their usefulness as tools to measure efficacy of vector control interventions (Hostomska et al., 2008; Clements et al., 2010; Gidwani et al., 2011; Vlkova et al., 2011; Marzouki et al., 2012; Kostalova et al., 2015; Quinnell et al., 2018). Nevertheless, the antibody response to saliva may be species-specific and may change in response to a defined antigen. Therefore, kinetics and duration of antibodies should be investigated for the sand fly vector species/biomarker in question. For L. longipalpis, two studies investigated the kinetics of antibody response to total saliva in dogs experimentally (Hostomska et al., 2008) or naturally (Quinnell et al., 2018) exposed to bites. In experimentallyexposed dogs, the level of saliva-specific total IgG antibodies correlated positively to the intensity of bites, increasing rapidly with each weekly exposure and declining significantly within 2 weeks of the last exposure, despite maintaining a low titer up to 6 months post-exposure (Hostomska et al., 2008). In dogs naturally exposed to L. longipalpis bites, dogs developed salivaspecific antibodies in 2 months, with antibody levels increasing during high transmission/biting intensity and declining rapidly during low transmission/biting intensity (Quinnell et al., 2018).

To date, we have developed the methodologies that identified several promising makers of exposure to L. longipalpis bites. Further studies of the utility of these defined antigens, as biomarkers of exposure to L. longipalpis need to be undertaken in natural foci to thoroughly investigate their specificity. Additionally, longitudinal studies to establish the kinetics of antibody development and decline to promising defined biomarkers will establish their value in biomonitoring and may even reveal important associations with risk of, or protection from, ZVL.

# OVERALL CONCLUSION

The complexity of ZVL and CVL challenges the reliability of a single biomarker to assess disease progression in humans and dogs, respectively. More likely, a combination of distinct inflammatory mediators will be needed to provide a tool that can distinguish relevant states of disease, also defining the role played by these different molecules in the pathogenesis of ZVL.

Despite the considerable progress made in defining important biomarkers for both ZVL and CVL, more studies are indicated, as well as an open dialogue by the scientific community, to reach a consensus for a reliable signature of distinct disease states. Another important point to consider evaluating the different biomarkers in ZVL is the possibility to find new targets to improve the treatment of the disease, increasing the chances of cure and avoiding the fatal outcome of infection. The combination of different drugs directed to several molecules would contribute to obtain an effective treatment for patients. In addition, in the future, such biomarkers may also be of value in assessing the level of protection induced by prophylactic strategies.

Together with well-defined markers of exposure to vector sand flies, such tools could become invaluable to evaluate response to treatment, and success of interventions among others.

# AUTHOR CONTRIBUTIONS

CB and SK have equal participation to write this review.

# FUNDING

This work was partly supported by the Intramural Research Program of the NIH, National Institute of Allergy and Infectious Diseases and by FAPESB (Fundação de Apoio a Pesquisa do Estado da Bahia), grant numbers SUS0036/2013 and PET0024/2013. CB is a senior investigator of the National Council of Research (CNPq).

# ACKNOWLEDGMENTS

The authors thank Andrezza Kariny and Juliana Oliveira for secretarial assistance.

# REFERENCES


activation, oxidative stress and inflammation characterize severe canine visceral Leishmaniasis. Sci. Rep. 6:32619. doi: 10.1038/srep32619


**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 Brodskyn and Kamhawi. 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.

# IFN-γ Response Is Associated to Time Exposure Among Asymptomatic Immune Responders That Visited American Tegumentary Leishmaniasis Endemic Areas in Peru

Ivan Best <sup>1</sup> \*, Angela Privat-Maldonado<sup>1</sup> , María Cruz <sup>1</sup> , Mirko Zimic<sup>2</sup> , Rachel Bras-Gonçalves <sup>3</sup> , Jean-Loup Lemesre<sup>3</sup> and Jorge Arévalo1,2

1 Instituto de Medicina Tropical Alexander von Humboldt, Universidad Peruana Cayetano Heredia, Lima, Peru, <sup>2</sup> Laboratorios de Investigación y Desarrollo, Faculty of Sciences and Philosophy, Universidad Peruana Cayetano Heredia, Lima, Peru, 3 Institut de Recherche pour le Développement (IRD), UMR177-INTERTRYP, Montpellier, France

### Edited by:

Javier Moreno, Instituto de Salud Carlos III, Spain

### Reviewed by:

Anjali Mishra, The Ohio State University, United States Aslam Khan, University of Missouri, United States

> \*Correspondence: Ivan Best ivan.best@upch.pe

### Specialty section:

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

> Received: 30 March 2018 Accepted: 30 July 2018 Published: 21 August 2018

### Citation:

Best I, Privat-Maldonado A, Cruz M, Zimic M, Bras-Gonçalves R, Lemesre J-L and Arévalo J (2018) IFN-γ Response Is Associated to Time Exposure Among Asymptomatic Immune Responders That Visited American Tegumentary Leishmaniasis Endemic Areas in Peru. Front. Cell. Infect. Microbiol. 8:289. doi: 10.3389/fcimb.2018.00289 Clinical manifestations of American Tegumentary Leishmaniasis (ATL) include cutaneous (CL) and mucous forms (ML); however, there are asymptomatic individuals who despite being infected do not present any clinical manifestations. This study characterized the cell-mediated immunity of travelers who lived in the Andean highlands of Cusco, free of leishmaniasis transmission, which eventually visited leishmaniasis endemic in the Amazonian basin and returned home without any clinical signs of the disease. Their immune response was compared with CL and ML patients who acquired the disease during their stage in the same region. Fifty-four human subjects from the highlands of Cusco (Peru), who have visited an endemic area, were enrolled: 28 of them did not show any symptoms, 12 showed CL and 14 showed ML. Ten healthy subjects from a non-endemic area (HS) were included as controls. T-cell proliferation was evaluated using peripheral blood mononuclear cells (PBMC) stimulated for 5 days with a total soluble leishmanial antigen (TSLA) of L. (V.) braziliensis. Th1/Th2/Th17 cytokines were also quantified in the supernatants by a flow cytometry multiplex assay. T-cell proliferation was expressed as stimulation index (SI) and the cut off was fixed at SI >2.47. Fifteen out of 28 subjects did not show any signs of disease (54%); subjects with an SI above the cut off. They were defined as asymptomatic immune responders (AIR). CL and ML patients presented a higher SI than HS and AIR. Among the latter group, the exposure time to Leishmania was clearly associated with the IFN-γ response. Increased levels of this cytokine were observed in individuals who remained <90 days in an endemic area of leishmaniasis. Our results evidenced two sub-populations among asymptomatic individuals, one AIR who did not develop clinical disease manifestations when they were exposed to Leishmania in endemic areas. Exposure time to Leishmania in the wild was associated with the IFN-γ response.

Keywords: American Tegumentary Leishmaniasis, asymptomatic infection, cellular immune response, T cell proliferation, Th1 response

# INTRODUCTION

American Tegumentary Leishmaniasis (ATL) is a zoonotic disease caused by parasites of genus Leishmania when people get in contact with infected sandfly vectors in the wild (Grimaldi and Tesh, 1993). The two most prevalent clinical forms of ATL in Peru are the cutaneous leishmaniasis (CL), locally known as Uta, and the severe mucocutaneous form (ML) called Espundia. The main species that causes ATL is Leishmania (V.) braziliensis, and is almost the only one associated to those 8–10% of CL patients, who later on developed the ML disease after several years of original skin lesions (Lucas et al., 1998; Davies et al., 2000). There is, however, a minority of patients who developed ML without a previous CL episode (Lindoso et al., 2009).

Among the infected people with Leishmania, not all will develop the disease (Biagi, 1953; Gonzalez and Biagi, 1968; Pampiglione et al., 1974; Follador et al., 2002; Fagundes et al., 2007a; Riera et al., 2008; Singh et al., 2014; Andrade-Narvaez et al., 2016). The asymptomatic category in the Leishmania infection was, many decades ago, proposed in both ATL and visceral leishmaniasis (VL) (Gonzalez and Biagi, 1968; Pampiglione et al., 1974).

The term asymptomatic in Leishmania infection was first used in 1953 in a Mexican cutaneous leishmaniasis area. Twelve out of 36 subjects, who showed a positive reaction to the Montenegro skin test (MST), did not develop ulcerated lesions or scars on the skin; those individuals were considered as asymptomatic carriers (Biagi, 1953). Similarly, another study in northern Italy first identified by MST, subclinical infection in an outbreak of human visceral leishmaniasis (Pampiglione et al., 1974). These results indicated that asymptomatic individuals were able to either clear the Leishmania pathogen or to host the parasite in a cryptic stage (Follador et al., 2002). The latter possibility was supported by reports which demonstrated the presence of parasite DNA among people who lived in VL endemic areas, but who never developed any clinical signs (Martín-Sánchez et al., 2004; Alborzi et al., 2008). The proportion of asymptomatic subjects could represent a large proportion of individuals living in cutaneous leishmaniasis endemic areas. A Tunisian study showed that 75% of healthy individuals, without a localized cutaneous leishmaniasis (LCL) history caused by L. major, were positive for a leishmanin skin test (LST) (Sassi et al., 1999). A previous study carried out in Peru, showed that 17% (16/94) of all infections caused by L.(V.) peruviana and evaluated by MST, corresponded to subclinical infections (Davies et al., 1995).

Defining Leishmania asymptomatic infections is very difficult because of the lack of a reliable biomarker. It is also not clear how one can discriminate parasite persistence in an asymptomatically infected individual from new infections that occur after the first episode, i.e. former parasites cleared by the immune response followed by new infecting parasite populations that will follow the same fate. Concerning asymptomatic biomarkers, a positive MST has been used as an indicator of host cell-mediated immune response against the parasite (Nogueira et al., 2008) and for the detection of asymptomatic or subclinical infection in endemic areas of leishmaniasis. However, it cannot make a distinction between active, inactive or past infections (Vega-López, 2003). There is evidence that biomarkers such cytokines and chemokines favor the identification of asymptomatic subjects in endemic areas of leishmaniasis (Sassi et al., 1999; Bittar et al., 2007; Ibarra-Meneses et al., 2017).

Concerning the limitations of having true asymptomatic individuals, the Cusco region in Peru offers a particular opportunity to study true asymptomatic cases. Members of Andean native communities living at 3,600 meters above sea level around Cusco, an area free of Leishmania transmission, descend to the Amazonian basin and become temporarily exposed to Leishmania parasites for a discrete and short period of time to carry out seasonal work or tourism activities, or they colonize leishmaniasis endemic areas. Many of the latter group returns to the highlands after several years living in contact with the Leishmania transmission cycle. Those individuals who return to the highlands, if they are Leishmania infected, are ideal to follow up and establish those cases that are true asymptomatic.

In this work we evaluated cellular immune response parameters, including inflammatory cytokines, in a group of individuals from the highlands of Cusco who went to the Amazonian basin, being exposed temporarily to leishmaniasis endemic areas. After a period of time the subjects returned to the highlands of Cusco where they were recruited to determine if there were asymptomatic immune responders (AIR) among them. Their T cell proliferation and cytokine production profile were compared with CL and ML patients. Furthermore, we assessed if these immune parameters were associated to one or more clinical or epidemiologic variables.

# MATERIALS AND METHODS

# Ethics Statement

The study was approved by the Institutional Ethics Committee of the Universidad Peruana Cayetano Heredia (Registration number: 53892), and a written informed consent was obtained from all participants.

# Study Population

The study was conducted at the Instituto de Medicina Tropical Alexander von Humboldt of the Universidad Peruana Cayetano Heredia, Lima, Peru, where the blood samples obtained from individuals living in Cusco arrived within 6 h after their bleeding, being processed the same day for the immunological assays. The study groups consisted of 12 patients with active CL, 14 with active ML, 28 individuals that visited leishmaniasis endemic areas without any disease manifestation and 10 healthy subjects from Lima who never exposed to ATL endemic areas (HS). Individuals from Cusco without clinical manifestations were recruited among habitants of the highlands of Cusco (Canchis, Paruro, Urcos, Anta, and Paucartambo). They were aware that leishmaniasis is a disease that was transmitted at the areas they had visited and therefore were concerned about a possible infection with Leishmania parasites. These subjects might have been exposed to ATL endemic areas in the neighboring forests of Cusco and Madre de Dios (Pilcopata, Calca, Santa Teresa,

Echarate) due to tourism, temporary residence or seasonal economic activities (**Figure 1**).

All CL and ML patients had a confirmed diagnosis of leishmaniasis by visualization of Leishmania amastigotes in Giemsa-stained smears, parasite culture and/or PCR test according to previous diagnostic procedures (Boggild et al., 2010). All the recruited individuals were negative to human immunodeficiency virus, hepatitis B surface antigen, hepatitis C virus, diabetes, anemia, pregnancy, and tuberculosis. None of them declared to have received corticoids treatment.

# Isolation of Mononuclear Cells

Peripheral blood mononuclear cells (PBMC) were isolated from Lithium Heparin-anticoagulated peripheral blood via density gradient centrifugation on Ficoll-Hypaque (GE Healthcare, UK), washed two times with Hanks's buffered salt solution (Gibco, USA), and one time with RPMI-1640 medium (Gibco, USA). All cells were resuspended in RPMI-1640 medium (Gibco, USA) and supplemented with 10% normal pooled human serum, 100 IU/ml penicillin (Gibco, USA), 100 mg/ml streptomycin (Gibco, USA), 2 mM L-glutamine (Gibco, USA), 1mM sodium pyruvate (Gibco, USA) and 1 mM non-essential amino acids (Gibco, USA); further referred to as complete medium. Cell viability was assessed by trypan blue dye exclusion.

# Preparation of Leishmania Antigens

Total soluble Leishmania antigen (TSLA) was prepared as follows: L. (V.) braziliensis (MHOM/BR/75/M2904) promastigotes (10<sup>9</sup> ) were resuspended in 1 ml of lysis buffer [100 ul of 20x protease inhibitor cocktail (Sigma-Aldrich, USA), 1 mM PMSF, 2 mM EDTA pH 7.4, 1 mM Tris HCl pH 7.4]. Next, the L. (V.) braziliensis parasites were disrupted by ten repeated freezing and thawing cycles (1 min at −70◦C and 2:30 min at 37◦C), then sonicated at 60 Hz. The mixture was centrifuged at 14,000 rpm for 10 min at 4◦C. The supernatant was stored at −70◦C until use. A small sample was kept to determine the protein concentration using the Qubit Protein Assay Kit (Invitrogen, USA) and SDS-PAGE was done to confirm the integrity of the isolated proteins.

# Proliferation Assays

PBMC were cultured in 96-well flat-bottomed plates (Falcon, Becton Dickinson, USA) in complete medium at 2 × 10<sup>5</sup> cells per well. The cells stimulated with 10 ug/ml TSLA, 10 ug/ml phytohemagglutinin (PHA) or complete medium alone were incubated at 37◦C in a humidified 5% CO<sup>2</sup> atmosphere for 5 days. Afterwards, 1 uCi [3H]-thymidine (Sigma-Aldrich, USA) was added to each well for the last 5 h of incubation. The cells were harvested on filter paper (Filtermat A, Perkin Elmer, Finland), washed extensively and then liquid scintillation mixture (Sigma-Aldrich, USA) was added. Incorporated [3H]-thymidine was measured with a 1205 Betaplate Liquid Scintillation Counter (Wallac, Finland). T-cell proliferation was expressed as stimulation index (SI) which is c.p.m. of stimulated cultures divided by c.p.m. of unstimulated cultures. The cut off (mean + 3 SD) for a positive response was fixed from SI of HS.

# Cytokine Measurement

Supernatants from cell cultures were collected on day 5 and analyzed with a flow cytometry multiplex assay (BD CBA Th1/Th2/Th17, Pharmingen, USA) to determine the levels of seven cytokines: IL-17A, IFN-γ, TNF-α, IL-10, IL-6, IL-2, and IL-4.

# Statistical Analysis

The chi-square test was used to analyze categorical variables and the Kruskal-Wallis H-test for continuous variables without a normal distribution. The correlation between the level of each cytokine and the time of permanence in the endemic area was estimated with the non-parametrical Spearman's rank correlation

test. AIR were classified in two groups after considering the time of permanence in the endemic area: individuals who stayed less or equal than 90 days vs. individuals who stayed longer than 90 days.

If possible, the levels of cytokines were transformed in order to have a normal distribution, confirmed with the Shapiro-Wilk W-test of normality (Boston and Sumner, 2003). The effect of the time of permanence in the endemic area on the normally-distributed transformed level of cytokine was tested after adjusting gender and age in a multiple linear regression.

# RESULTS

# Immunological Definition of Asymptomatic Immune Responders (AIR)

Among the individuals who were exposed to ATL endemic areas but did not present any clinical manifestations, there were a set of individuals called AIR who were defined because of their SI were equal or above to 2.47. Fifteen out of 28 (54%) individuals who were temporarily exposed to the Leishmania transmission endemic areas showed significantly high SI (**Figure 2**). Therefore, SI was used to identify the AIR, the sub-population of individuals who were probably exposed to Leishmania antigens and therefore to Leishmania parasites infection but no disease clinical manifestation outcome occurred. For SI, the AIR showed a median of 4.7 while the corresponding non immune responders group a median of 1.4, respectively (P < 0.001, data not shown).

# Demographic Characteristics of Study Population

According to the type of activity in the endemic areas, patients with active CL and ML were all male workers whereas the AIR group was composed by subjects who visited endemic areas for work or recreational activities, male and female individuals in equal proportions. The CL patients were significantly younger (P < 0.05) than AIR, ML patients and HS (P < 0.01, **Table 1**). No significant differences were found among these groups regarding the exposure time in the infection place (**Table 1**). Concerning the individuals that visited the leishmaniasis endemic areas there were not significant differences between AIR individuals and those ones who did not respond to Leishmania antigens regarding age, sex, occupation and time of exposure in the infection place (data not shown).

# Association of Epidemiological Parameters With the Immune Markers in the Asymptomatic Immune Responders (AIR)

No statistical significance differences were observed between the individuals that visited the leishmaniasis endemic areas without clinical manifestations and HS for all tested cytokines. There was however a trend toward higher IFN-γ levels when compared AIR with HS (P = 0.071, **Table 1**). No significant differences were found when compared AIR with corresponding non-immune responders group (data not shown).

The level of IFN-γ in AIR showed a significant correlation with the time of permanence in an endemic area [Spearman's rho (ρ) −0.64; P = 0.010, n = 15, **Table 2**]. Individuals who stayed in the endemic area for a period longer than 90 days, showed significantly lower levels of IFN-γ (median = 1.75 pg/ml) than individuals who stayed there for a shorter period (median = 31.5, P < 0.05; Kruskal-Wallis H-test). In addition, the occupation of the inhabitants belonging to each of these groups was significantly different (P < 0.01, data not shown). Temporary residence or seasonal work like mining, agriculture, and construction were the principal occupations of the subjects who stayed in the endemic area for a period longer than 90 days while the main activity for individuals who stayed there for a shorter period was tourism (87.5%). The period of permanence in the endemic area was able to explain 35.8% of the variability of IFN-γ after adjusting the age and gender in the multiple linear regression (**Table 3**). No other cytokine showed a significant correlation. When cytokine levels were compared by gender among the different groups, no differences were observed in the levels of cytokines between females and males, except for a significant increase in TNF-α levels in females compared to males in the healthy control group (P < 0.05, data not shown).

The square root transformation procedure converted the IFN-γ and the IL-10 into normally distributed variables (Boston and Sumner, 2003). The normalization was confirmed with the Shapiro-Wilk W-test (P = 0.55, 0.61; respectively). No other cytokine was able to be normalized by this statistic approach. The normally distributed and transformed IL-10 was not significantly associated, neither in the single nor in a multiple linear regression.

# T Cell Proliferation and Cytokine Response in CL and ML Patients

As shown in **Table 1**, CL and ML patients presented significantly higher SI than AIR and HS (P < 0.001). Moreover, CL patients TABLE 1 | Epidemiological parameters and immunological markers in cutaneous and mucosal leishmaniasis patients, asymptomatic immune responders and healthy subjects.


\*P-value for the comparison between CL patients and asymptomatic immune responders, \*\* P-value for the comparison between ML patients and asymptomatic immune responders. <sup>a</sup>Age, exposure time in an endemic area, stimulation index, ratio IFN-γ/IL-10 as well as IFN-γ, TNF-α, IL-10 and IL-17A levels are presented as median (Q1–Q3). <sup>b</sup>Male gender and occupation is presented as absolute numbers and percentages (between brackets). CL, cutaneous leishmaniasis; ML, mucosal leishmaniasis; Q1–Q3, first quartile–third quartile.

TABLE 2 | Spearman rank correlations between age, gender, IFN-γ, TNF-α, IL-10, IL-17A, ratio IFN-γ/IL-10, and exposure time in the infection place less or equal -and longer- than 90 days.


TABLE 3 | Univariate and multivariate comparison between subjects who stayed less or equal than 90 days and subjects who stayed longer than 90 days in an endemic area.


\*P-value for the comparison between subjects who stayed less or equal than 90 days vs. subjects who stayed longer than 90 days in an endemic area.

were observed among the different groups. In addition, CL and ML patients presented a significantly higher ratio IFN-γ/IL-10 compared to AIR and HS (P < 0.001).

## showed a significant increase of SI compared to ML patients (P < 0.01). After the evaluation of the effector response mediated by cytokines, a significant increase of IFN-γ and TNF-α production in patients with CL and ML at equivalent levels took place compared to AIR and HS (P < 0.001, **Table 1**). CL and ML patients presented higher IL-17A than AIR and HS. Patients with ML showed a significant increase of IL-10 comparing to AIR and HS (P < 0.05, **Table 1**). No differences in the remaining cytokines

# DISCUSSION

The immune cell proliferation assay of PBMC obtained with Leishmania crude antigens discriminated between AIR and other individuals who went to the same disease endemic area but were unable to mount and/or keep an immune response. This discrimination was not feasible with Th1, Th2, or Th17 cytokines, although IFN-γ showed a trend to discriminate between AIR and HS. Therefore, MST still defines the asymptomatic status of an individual. It is highly sensitive but lacks specificity, as demonstrated by the occurrence of cross reactions with other diseases (de Lima Barros et al., 2005; Fagundes et al., 2007b).

Another contribution of this study is the population under study that offers an advantage to study truly asymptomatic individuals. In general, characterization of the cell-mediated immunity of asymptomatic patients may be obscured by the fact that they are usually exposed to recurrent Leishmania infection episodes because they are permanent residents in ATL endemic areas. This work exploited, however, a particular situation found on the highlands of Cusco, were members of Andean native communities became temporarily exposed to Leishmania parasites for a discrete period of time when they descended to the Amazon basin of Cusco or Madre de Dios. They stayed either for short periods of time to carry out seasonal work or tourism activities, or were temporal residents (up to 7 years, data not shown); however, each of them left the endemic transmission areas many years ago before being analyzed. It is within this population that is was possible to identify those individuals, who belonged to the AIR group, a true asymptomatic condition. Future studies on this population, with additional cellular immune response biomarkers should permit to define better profiles of asymptomatic individuals.

This study detected that 15 out of 28 subjects who traveled to the Amazon jungle of Cusco and/or Madre de Dios, showed a positive T-cell proliferation when challenged with L. (V.) braziliensis crude antigen. Here, they are called AIR and presented a significantly lower SI as well as a lower proinflammatory cytokines production (IFN-γ, TNF-α) compared to the strong T cell response, observed during active CL and ML (**Table 1**). The AIR responded on a past infection with a moderate but significant IFN-γ when challenged with L. braziliensis TSLA. Interestingly, an increased relative proportion of IL-10 producing cells, expressed by a low IFN-γ/IL-10 ratio, were observed in the AIR compared to active CL and ML which showed high IFN-γ/IL-10 ratios. The ratio between effector and regulatory specific T cells may influence the outcome of infection. Thus, these results lead us to propose that IL-10 in AIR could counter-regulate the IFN-γ effects, thus maintaining tissue integrity with an absence of ulcers or lesions. Our results were comparable with a previous report that found a low IFN-γ/IL-10 ratio in asymptomatic Leishmania carriers compared to cured CL patients (Bittar et al., 2007). The IL-10 protective role, here proposed in ATL, should be added to other immunological factors implied in host's resistant mechanisms (Díaz et al., 2010).

Previous studies, made in murine models, showed the role of the adaptive immune response mediated by CD4 Th1 and Th2 cells in the establishment and course of the Leishmania infection (Alexander and Bryson, 2005). IFN-γ produced by Th1 cells activates infected macrophages to eliminate the Leishmania by the production of nitric oxide (NO). However, an exacerbated production of this cytokine, observed during active CL and ML (**Table 1**), could be associated to tissue damage (Liew and O'Donnell, 1993; Roberts, 2006; Sharma and Singh, 2009; Silveira et al., 2009). On the other hand, IL-10 is now considered as a regulatory cytokine involved in the persistence of the parasite in the skin while it originally was included as a Th2 cytokine that inhibit the macrophage activation and proliferation of Th1 cells (Rodriguez et al., 2007).

Consistent with a study carried out in Brazil (Gomes-Silva et al., 2007), in our study, AIR had a significantly lower degree of antigen-specific T-cell expansion manifested by a lower SI and pro-inflammatory cytokines production (IFN-γ, TNF-α), in contrast to the strong response of T cells observed during active CL and ML (**Table 1**).

Nevertheless, when the effector T cell response between the AIR and HS was compared, the specific production of all tested cytokines were unable to discriminate these groups (**Table 1**). It was only possible to observe a significant association between IFN-γ response and the nature of Leishmania exposure when the IFN-γ production was measured in AIR individuals who spent less time in the infection place, mainly engaged to tourism activities. They showed increased IFN-γ levels respect to those individuals who have spent more time in the infection place due to seasonal works. These data suggested that the exposure time and activity type, which the migrants undertook in the Amazon jungle, could influence the nature of the immune response in the absence of primary transmission of ATL. The difference in the IFN-γ levels could be explained by the way these two sub-populations interact with the Amazon jungle. Individuals involved in tourism activities are probably more susceptible and more exposed to the transmission by leishmaniasis vectors; while seasonal workers develop immunity against the parasite through its probably less aggressive activity toward the forest and their greater exposure time in an endemic area. Consistent with this hypothesis, a study carried out in Bolivia showed that migrants from the highlands have an increased risk to develop CL and ML compared to the natives who lived in an endemic area (Alcais et al., 1997). Studies with other parasites reported that individuals, who lived in areas of lower malaria transmission, had higher IgG response to Plasmodium falciparum merozoite surface protein-1 (PfMSP1–19) compared to a neighboring village with higher malaria transmission (Braga et al., 2002).

We consider that AIR correspond to true asymptomatic carriers although it might be possible that some individuals would be either subclinical diseases or people who completely cleared the infection parasite (Biagi, 1953; Gonzalez and Biagi, 1968; Pampiglione et al., 1974; Follador et al., 2002; Martín-Sánchez et al., 2004; Fagundes et al., 2007a; Alborzi et al., 2008; Riera et al., 2008). To understand the biological basis why infected people do not develop disease requires a sustained and considerable number of volunteers, a situation found in the population living in the highlands of Cusco.

It is still unclear why some infected individuals, when exposed to an endemic area of ATL, develop disease while others do not present any clinical manifestations. The latter, here called AIR, could represent individuals with a natural resistance to develop ATL diseases. The development of an appropriate immune response, which controls parasite replication and maintains tissue integrity, is the simplest and straightest explanation for this phenomenon. There is however a new option to be incorporated in future studies, the quiescent stage of the Leishmania amastigote (Kloehn et al., 2015; Jara et al., 2017; Mandell and Beverley, 2017).

# AUTHOR CONTRIBUTIONS

IB, RB-G, J-LL, and JA conceived and designed the research. MC made the inclusion of patients. IB and AP-M performed the experiments. IB and JA analyzed the data and wrote the paper. All authors have read and approved the manuscript.

# REFERENCES


# FUNDING

This study received financial support from EU's Seventh Framework Programme (FP7) (RAPSODI project, grant agreement number 223341). URL of funder's website: https:// cordis.europa.eu/guidance/archive\_en.html.

# ACKNOWLEDGMENTS

We thank the staff at the Instituto de Medicina Tropical Alexander von Humboldt of the Universidad Peruana Cayetano Heredia for their logistic support.

Montenegro skin test among patients with sporotrichosis in Rio De Janeiro. Acta Trop. 93, 41–47. doi: 10.1016/j.actatropica.2004.09.004


in peripheral blood from asymptomatic individuals at risk for parenterally transmitted infections: relationship between polymerase chain reaction results and other Leishmania infection markers. Am. J. Trop. Med. Hyg. 70, 545–548. doi: 10.4269/ajtmh.2004.70.545


cytokine production after symptomatic or asymptomatic Leishmania major infection in Tunisia. Clin. Exp. Immunol. 116, 127–132. doi: 10.1046/j.1365-2249.1999.00844.x


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

Copyright © 2018 Best, Privat-Maldonado, Cruz, Zimic, Bras-Gonçalves, Lemesre and Arévalo. 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.

# Comparative Evolution of Sand Fly Salivary Protein Families and Implications for Biomarkers of Vector Exposure and Salivary Vaccine Candidates

### Iliano V. Coutinho-Abreu\* and Jesus G. Valenzuela\*

Vector Molecular Biology Section, Laboratory of Malaria and Vector Research, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Rockville, MD, United States

### *Edited by:*

Javier Moreno, Instituto de Salud Carlos III, Spain

### *Reviewed by:*

Sandra Marcia Muxel, Universidade de São Paulo, Brazil Jere W. McBride, The University of Texas Medical Branch at Galveston, United States

### *\*Correspondence:*

Iliano V. Coutinho-Abreu vieiracoutinhoabreugomes2@nih.gov Jesus G. Valenzuela jvalenzuela@niaid.nih.gov

### *Specialty section:*

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

> *Received:* 28 April 2018 *Accepted:* 30 July 2018 *Published:* 29 August 2018

### *Citation:*

Coutinho-Abreu IV and Valenzuela JG (2018) Comparative Evolution of Sand Fly Salivary Protein Families and Implications for Biomarkers of Vector Exposure and Salivary Vaccine Candidates. Front. Cell. Infect. Microbiol. 8:290. doi: 10.3389/fcimb.2018.00290 Sand fly salivary proteins that produce a specific antibody response in humans and animal reservoirs have been shown to be promising biomarkers of sand fly exposure. Furthermore, immunity to sand fly salivary proteins were shown to protect rodents and non-human primates against Leishmania infection. We are missing critical information regarding the divergence amongst sand fly salivary proteins from different sand fly vectors, a knowledge that will support the search of broad or specific salivary biomarkers of vector exposure and those for vaccines components against leishmaniasis. Here, we compare the molecular evolution of the salivary protein families in New World and Old World sand flies from 14 different sand fly vectors. We found that the protein families unique to OW sand flies are more conserved than those unique to NW sand flies regarding both sequence polymorphisms and copy number variation. In addition, the protein families unique to OW sand flies do not display as many conserved cysteine residues as the one unique to the NW group (28.5% in OW vs. 62.5% in NW). Moreover, the expression of specific protein families is restricted to the salivary glands of unique sand fly taxon. For instance, the ParSP15 family is unique to the Larroussius subgenus whereas phospholipase A2 is only expressed in member of Larroussius and Adlerius subgenera. The SP2.5-like family is only expressed in members of the Phlebotomus and Paraphlebotomus subgenera. The sequences shared between OW and NW sand flies have diverged at similar rates (38.7 and 45.3% amino acid divergence, respectively), yet differences in gene copy number were evident across protein families and sand fly species. Overall, this comparative analysis sheds light on the different modes of sand fly salivary protein family divergence. Also, it informs which protein families are unique and conserved within taxon for the choice of taxon-specific biomarkers of vector exposure, as well as those families more conserved across taxa to be used as pan-specific vaccines for leishmaniasis.

Keywords: salivary proteins, molecular evolution, markers of exposure, vaccines, sand flies, Leishmania, Leishmaniasis

# INTRODUCTION

Vector borne diseases represent almost half of the neglected tropical infectious diseases. When attempting to get a blood meal, most vectors of disease deliver the pathogen in the host skin and together with the pathogen these arthropods deliver saliva (Coutinho-Abreu et al., 2015; de Castro et al., 2017). Blood sucking arthropods secrete a plethora of bioactive compounds in their saliva to counteract the mammalian host hemostatic system in order to get a successful blood meal. Salivary antihemostatic components such as anticoagulants (Chagas et al., 2014), vasodilators (Ribeiro et al., 1989; Lerner et al., 1991; Champagne and Ribeiro, 1994), and inhibitors of platelet aggregation (Calvo et al., 2007, 2011; Assumpcao et al., 2013) have been described and some of these proteins characterized at the molecular level (Coutinho-Abreu et al., 2015). In addition, the biological activity of some arthropod salivary proteins was shown to promote the establishment of pathogens in the mammalian host (Coutinho-Abreu et al., 2015; de Castro et al., 2017). Furthermore, immunity to specific sand fly salivary proteins was shown to protect rodents and non-human primates against leishmaniasis (Kamhawi et al., 2000; Gomes et al., 2008; Collin et al., 2009; Oliveira et al., 2015).

It is well established that humans and animal reservoirs make antibodies to proteins in the saliva of insects including those present in sand flies. These findings have prompted research groups to explore the use of sand fly salivary proteins as markers of sand fly exposure for humans and animal reservoirs (Teixeira et al., 2010; Drahota et al., 2014; Marzouki et al., 2015; Sima et al., 2016; Kostalova et al., 2017). A recombinant sand fly salivary protein of 43 kDa (rSP03B), from the sand fly P. pernicious, belonging to the yellow family of proteins was shown to be a marker of sand fly exposure in dogs living in Southern and Central Italy and in Portugal (Drahota et al., 2014; Kostalova et al., 2017). Interestingly, the yellow related protein from P. orientalis (rPorSP24) was demonstrated to be a good marker of sand fly exposure in domestic animals including dogs from a L. donovani foci in Ethiopia (Sima et al., 2016). For humans living in visceral leishmaniasis disease endemic areas in Brazil, the combination of salivary proteins LJM11 and LJM17 (yellow related proteins) was demonstrated to be the best biomarker of Lutzomyia longipalpis exposure in humans (Teixeira et al., 2010), while Linb13, a protein of 30 kDa belonging to the antigen-5 family of proteins, was shown to be the best biomarker of Lutzomyia intermedia exposure in humans living in a cutaneous leishmaniasis endemic area (Carvalho et al., 2017). The salivary protein PpSP32 from P. papatasi was demonstrated to be the marker of P. papatasi exposure for humans living in cutaneous leishmaniasis endemic areas in Tunisia (Marzouki et al., 2015) and in Saudi Arabia (Mondragon-Shem et al., 2015).

Based on their continental separation, sand flies are classified as belonging to New World (NW) and Old World (OW) groups. Thus far, over a dozen sand fly salivary gland transcriptomes have been obtained. From the OW group, 10 species have had their salivary transcriptomes decoded. These species belong to five subgenera of the genus Phlebotomus, including the subgenera Phlebotomus, Paraphlebotomus, Larroussius, Adlerius, and Euphlebotomus (Valenzuela et al., 2001; Anderson et al., 2006; Kato et al., 2006; Oliveira et al., 2006; Hostomská et al., 2009; Abdeladhim et al., 2012; Rohousova et al., 2012; Martín-Martín et al., 2013; Vlkova et al., 2014). For NW sand flies, salivary gland transcriptomes were sequenced from four species. Within the genus Lutzomyia, the salivary transcriptomes of Lu. longipalpis (subgenus Lutzomyia) and Lutzomyia ayacuchensis (subgenus Helcocyrtomyia), as well as two other transcriptomes from sand flies in the genera Nyssomyia (Nyssomyia intermedia) and Bichromomyia (Bichromomyia olmeca) have been obtained (Valenzuela et al., 2004; de Moura et al., 2013; Kato et al., 2013; Abdeladhim et al., 2016).

The use of sand fly salivary proteins as markers of vector exposure represents therefore a practical application that can be implemented in epidemiological studies as well as for vector control programs. Therefore, having a well-defined catalog of sandfly salivary proteins as well as a better understanding of the evolutionary relationship of salivary proteins from different sand fly vectors will allow us to make more precise selection for these appealing biomarkers. In the current study, we focused on the evolutionary analysis of protein families unique to Old World (OW) sand flies comparing that with the protein families shared between OW and New World (NW) sand flies as well as those unique to NW sand flies. This comparative analysis also unveiled that the some salivary protein families unique to OW sand flies emerged and diversified in different manners than their counterparts unique to NW sand flies. These findings can inform which proteins are the best candidates to be used as a pan-specific or species-specific biomarkers of sand fly exposure or as well as a pan-specific vaccine against leishmaniasis.

# MATERIALS AND METHODS

## Sequences

Nucleotide and amino acid sequences were retrieved from the NCBI databases from sand fly salivary gland transcriptomes (Valenzuela et al., 2001, 2004; Anderson et al., 2006; Oliveira et al., 2006; Hostomská et al., 2009; Abdeladhim et al., 2012, 2016; Rohousova et al., 2012; de Moura et al., 2013; Kato et al., 2013; Martín-Martín et al., 2013; Vlkova et al., 2014). Signal peptides were removed from the protein sequences whereas sequences encoding signal peptides and stop codons were removed from the nucleotide sequences for further analyses. Only sequences displaying more than 5% divergence at the amino acid level were assumed to be encoded by true paralog genes and included in the analyses, rather than being alleles of the same gene and otherwise discarded. Sand fly groups, species, and sequence accession numbers are provided in **Supplementary Table 1**.

# Sequence Alignment

Multiple sequence alignments of putative peptides were carried out using Clustal Omega built in the MacVector software 15.8 (Olson, 1994) and in MEGA7 (Kumar et al., 2016). For the construction of phylogenetic trees, the gap penalties were not taken into account in the multiple sequence alignments.


–, not expressed or not detected.

represent specific taxon: Green color represents the Larroussius and Adlerius subgenera; Blue color points to proteins of the Phlebotomus and Paraphlebotomus subgenera; and Black color indicates the proteins belonging to New World sand flies. Although the PduK84 sequence is illustrated here, it was not included in further analyses because it is truncated.

# DNA Polymorphism, Protein Divergence, and Phylogenetic Analysis

The evolutionary analyses were performed in the DnaSP 5.10 software (Librado and Rozas, 2009). The parameter ω refers to the rate of non-synonymous nucleotide polymorphisms (Ka) over the synonymous rate of nucleotide polymorphisms (Ks) (Nei, 1987). Slide window analyses of ω along the nucleotide sequences encoding such proteins were also obtained. The diversity of the protein family sequences refers to the p-distance [proportion (p) of amino acid sites at which the two sequences to be compared are different (Nei, 2000)] obtained in the MEGA7 software (Kumar et al., 2016).

The evolutionary histories of salivary protein families were inferred by using the Maximum Likelihood method and conducted in MEGA7 (Kumar et al., 2016). The amino acid substitution model was selected based on the best fit provided by the Model Selection tool built in the MEGA 7 software. The bootstrap consensus trees inferred from 1,000 replicates (Felsenstein, 1985) were taken to represent the evolutionary history of the taxa analyzed (Felsenstein, 1985). Branches corresponding to partitions reproduced in <50% bootstrap replicates are collapsed. Initial tree(s) for the heuristic search were obtained by applying the Neighbor-Joining method to a matrix of pairwise distances estimated using a JTT model (Felsenstein, 1985).

# Statistical Analysis

D'Agostino and Pearson normality test was performed to assess whether or not data follow a normal distribution, and the Mann– Whitney test was carried out in order to test the significance of the differences in protein divergence and omega (ω) values. Both tests were performed using the Prism 7 software (GraphPad).

# RESULTS

# Updated Sand Fly Salivary Protein Catalog

From 14 sand fly salivary gland transcriptomes we compiled the sand fly salivary proteins unique to either NW or OW sand flies as well as protein families common to all sand fly species and developed an updated catalog for all the analyzed sand fly salivary proteins (**Table 1**). With this extensive catalog of sand fly salivary proteins we identified 12 protein families shared between OW and NW sand flies, lufaxin, apyrase, yellow-related protein, silk-related protein or PpSP32, endonuclease, hyaluronidase, adenosine deaminase, small odorant binding-like proteins, D7 family of proteins, antigen-5 related protein, ParSP17 (39 kDa protein of unkown function), ParSP80 (16 kDa protein of unkown function) (**Figure 1** and **Supplementary Figures 4**–**15**), seven protein families unique to OW sand flies, pyrophosphatase, phospholipase A2, ParSp15 (5 kDa protein of unknown function), ParSP25 (32 kDa protein of unknown function), SP16

(14.3 kDa protein of unknown function), ParSP23 (2 kDa protein of unknown function), and SP2.5 (2.5 kDa protein of unknown function, **Figures 7**–**10** and **Supplementary Figures 1**–**3**). Two protein families unique to OW sand flies that are only shared by two sand fly species (ParSP23-like and ParSP2.5 like; **Supplementary Figures 2**,**3**). We also identified 12 protein families unique to NW sand flies: toxin, RGD, c-type lectin, 14 kDa protein family, ml domain, 5′ nucleotidase, 9 kDa protein family, salo, 11.5 kDa protein family, maxadilan, 71 kDa protein family, and 5 kDa protein family.

# Salivary Proteins Shared Between Old World and New World Sand Flies

The phylogenetic trees of 9 out of 12 salivary proteinencoding gene families shared between OW and NW sand flies (**Supplementary Figures 4**–**15**) correlated well with the sand fly species phylogeny, constructed based on ITS-2 sequences (Aransay et al., 2000). The resulting phylogenetic tree indicates that these shared gene families of sand fly salivary proteins have been evolving under purifying (negative) selection or are selective neutral (ω < 1; Barton and Etheridge, 2004). Such salivary protein families encompassed lufaxin, apyrase, yellow-related protein, antigen-5, endonuclease, hyaluronidase, ParSP80, odorant binding proteins (OBPs), and adenosine deaminase. On the other hand, Silk-related protein or PpSP32, the D7 family of proteins (**Supplementary Figures 6**,**12**; Abdeladhim et al., 2016) and the ParSP17-like family of proteins (**Figure 1**) displayed phylogenies diverging from the sand fly species phylogeny, pointing that strong selective forces related to the function and/or immunogenicity of such proteins have driven their evolution since the split of NW and OW sand flies that likely took place about 100MYs with the continental separation of the American continent from the European and African ones.

# Molecular Evolution Between OW and NW Sand Fly Salivary Proteins

The overall rates of molecular evolution of the salivary protein families unique to OW sand flies were similar to the families shared with NW sand flies (**Figures 2A,B**). The median protein divergences (% divergence) for the families unique to OW sand flies and shared with the NW ones were 41.4 and 35.5%, respectively (**Figure 2A**). Likewise, the median ratios of nonsynonymous over synonymous replacements (ω) for the families unique to OW sand flies and shared with the NW ones were 0.34 and 0.23, respectively (**Figure 2B**). This is in sharp contrast to the pattern of divergence observed in the protein families unique to NW sand flies when compared with their counterparts shared with Old World sand flies (**Figures 2C,D**), as observed elsewhere (Abdeladhim et al., 2016). In the protein families unique to NW sand flies, the median protein divergence was 54.7% whereas 39.3% divergence was noticed for the protein families shared with OW sand flies (**Figure 2C**, p < 0.0008). By the same token, the median ω value was significantly greater in the protein families unique to NW sand flies (median ω = 0.49) as compared to

those shared with OW sand flies (median ω = 0.2, p < 0.0008; **Figure 2D**).

# Molecular Evolution Within OW and NW Sand Flies

The molecular evolution of the protein families was also taken into account individually, in regard to the analyses of % divergence and the ratio of non-synonymous over synonymous replacements (ω; **Figure 3**). Among the protein families unique to OW sand flies, phospholipase A2, pyrophosphatase, and ParSP25-like displayed the lowest rates of diversification (divergence < 30%; ω < 0.35; **Figures 3A,B**) whereas ParSP2.5, ParSP23, and SP16 were among the most divergent families (divergence > 50%; ω > 0.44; **Figures 3A,B**). Within the OW sand fly protein families shared with NW sand flies, ParSP80, and hyaluronidase families showed consistently the lowest rates of sequence divergence and non-synonymous to synonymous codon replacement (divergence < 12%; ω < 0.12; **Figures 3A,B**). In contrast, the OBPs, silk-related, D7, and ParSP17 families exhibited the greatest rates of diversification (divergence > 39%; ω > 0.26; **Figures 3A,B**). Regarding the protein families unique to NW sand flies, the SALO, maxadilan, and ML domain families presented the highest rates of molecular evolution (divergence > 60%; ω > 0.62; **Figures 3C,D**; Abdeladhim et al., 2016) whereas the 5 kDa protein family presented itself as the least divergent protein family (divergence = 36%; ω = 0.26; **Figures 3C,D**; not previously assessed). Amongst the protein families of NW sand flies shared with OW species, the D7 family presented the highest levels of diversification (divergence = 43%; **Figure 3C**) whereas the antigen-5 protein family exhibited the lowest divergence at the amino acid level (divergence = 19%; **Figure 3C**). Regarding the ω values, the silk family displayed the highest

ratio of non-synonymous over synonymous replacements (ω = 0.30; **Figure 3D**) whereas the lowest ratio was exhibited by the apyrase family. Regarding the OBP family, it is actually a superfamily, encompassing at least six protein families (**Supplementary Figure 15**). Hence, the high rates of divergence for the OBPs should be view with caution, as noted below.

# Molecular Diversification Along Protein Lengths

As specific regions of a protein can be subjected to different selective constraints and in turn evolve at different rates, slide-window analyses of non-synonymous over synonymous replacements (ω) for the genes encoding the salivary protein unique to OW sand flies and shared with NW sand flies

were carried out (**Figures 4**, **5**). Contrasting to the genes encoding the protein families unique to NW sand flies, which display multiple codons under positive (diversifying) or relaxed purifying selection (ω ≥ 1; Abdeladhim et al., 2016), the gene families unique to Old World sand flies exhibited for the most part codons under purifying (negative) selection (ω < 1; **Figure 4**). In the latter group, only ParSP15 and phospholipase A2 bore a few codons under positive selection (ω ≥ 1; **Figure 4**), mostly in the 5′ portion of the genes. The gene families shared with NW sand flies also displayed low levels of sequence divergence (**Figure 3**), with exception of the ParSP17 family which bear multiple codons under positive selection (**Figure 5**). In such (ParSP17; **Figure 5A**), a greater number of codons under positive selection were noticed in the gene sequences for the sand flies belonging to the related Phlebotomus and Paraphlebotomus subgenera (P/P; **Figure 5B**) than for the sand flies belonging to related Larrossious, Adhelerius, and Euphlebotomus subgenera (L/A/E; **Figure 5C**).

# Copy Number Variation of Salivary Protein-Encoding Genes Across Species

As important as sequence polymorphism (SNPs and INDELs) for the evolution of gene families and emergence of new molecular functions are gene duplication events (Innan and Kondrashov, 2010). Although less than two gene duplication events per species were noticed for two NW sand fly gene families (yellow-related and the small OBPs) shared with OW sand flies (**Figure 6**; yellow-related, **Supplementary Figure 13**; OBPs, **Supplementary Figure 15**; Abdeladhim et al., 2016), up to five gene duplication events were accounted for in six OW sand fly gene families (antigen-5, apyrase, D7, Silk, yellowrelated, OBPs) shared with NW sand flies (**Figure 6**; antigen-5, **Supplementary Figure 4**; apyrase, **Supplementary Figure 5**; D7, **Supplementary Figure 6**; Silk, **Supplementary Figure 12**; yellow-related, **Supplementary Figure 13**; and OBPs, **Supplementary Figure 15**). In the gene families unique to NW sand flies (SALO, RGD, mannose-binding lectin, c-type lectin, and spider toxin-like) up to eight gene duplication events were detected in six gene families (**Figure 6**; SALO, RGD, mannose-binding lectin, c-type lectin, and spider toxin-like; Abdeladhim et al., 2016). In sharp contrast to the high rate of copy number variation in gene families unique to NW sand flies, no gene duplication event was noticed for the sand fly salivary protein encoding gene families unique to OW sand flies (**Figures 6**–**10** and **Supplementary Figures 1**–**3**).

# Emergence of Specific Salivary Protein Encoding Genes

Besides the differences in the rates of gene diversification and gene duplication events among protein families, the molecular evolution of salivary protein encoding gene families in sand flies has also displayed different modes of gene emergence as well as specific signatures of protein structure and divergence.

A few salivary gland gene families appear to have arisen upon duplication from a gene expressed in another tissue that subsequently acquired a specific promoter driving expression into the salivary gland, a mechanism called sub-functionalization (neofunctionalization) (Hahn, 2009; Innan and Kondrashov, 2010). Amongst the salivary gene families unique to NW sand flies, the c-type lectin, the mannose-binding lectin, and the spider toxin-like gene families share paralogs expressed in other tissues of sand flies and other unrelated arthropods (Abdeladhim et al., 2016). Similar phenomena were noticed for salivary protein families unique to OW sand flies, such as

phospholipase A2 (Tunaz et al., 2003) and pyrophosphatase (Silva et al., 2015), as well as in protein families shared with NW sand flies, like hyaluronidase (Allalouf et al., 1975), endonuclease (Broderick et al., 2014), adenosine deaminase (Dolezelova et al., 2005), and OBPs (Benoit et al., 2017). Hence, such gene families seem to have emerged by subfunctionalization.

The presence of an OBP-like domain in the D7 protein Ctermini (Hekmat-Scafe et al., 2000) points to the emergence of D7 genes by gene fusion (Kaessmann et al., 2009) between an ancient OBP and an unknown gene. In fact, conserved cysteine signatures (**Table 2**), as bore by OBPs, can unveil the mechanism of gene birth even among fast evolving genes. Multiple sand fly salivary protein families bear cysteine signatures (≥4 cysteines; **Table 2**). In the salivary protein families unique to NW sand flies, conserved cysteine residue signatures (62.5%, five out of eight protein families; **Table 2**) are present at a much higher frequency than in the protein families shared between OW and NW sand flies (50%, 6 out of 12; **Table 2**). Amongst the protein families unique to the OW sand flies, such proportion is reduced to only 28.5% of the protein families (two out of seven families; **Table 2**).

# Taxon-Specific Expression of Salivary Protein Encoding Genes

In order to adapt to the new ecological niches, sand flies have faced the hemostasis components and immune systems of the mammal and bird species indigenous to their habitat. In order to face such selective pressures, the establishment of new protein variants may have been required for the survival of the sand flies. It is important to mention, nonetheless, that even within OW sand flies, differences in the levels of polymorphisms were noticed. The rates of codons under positive selection varied within subgroups: for instance, higher rates were noticed within the P/P subgenera in ParSP17 (**Figure 5**). In addition, most of the gene duplication events have taken place in P. duboscqi for the P/P subgenera as well as in P. orientalis and P. perniciosus for the L/A/E subgenera (**Figure 6**). It is noteworthy that members of the P/P subgenera only express up to three of the salivary proteins unique to OW sand flies (SP2.5, SP16, and/or pyrophosphatase) whereas members of the L/A/E subgenera for the most part secrete salivary proteins of four to five different families (ParSP15, ParSP25, ParSP23, phospholipase A2, SP16, and/or pyrophosphatase; **Table 1**). Among the later, ParSP15 family

symbols in the legend on the right.

is unique to the Larroussius subgenus whereas phospholipase A2 is only expressed in member of Larroussius and Adlerius subgenera. Along the same lines, transcripts for hyaluronidase and endonuclease were missing in the P/P sialotranscriptomes whereas such protein are expressed in NW and L/A/E salivary glands (**Table 1**).

# DISCUSSION

The comparative analysis of salivary protein gene families between OW and NW sand flies reveals that differences in the mode of evolutionary diversification amongst gene families that could have potential implications for the application of such proteins in the selection of markers of vector exposure and vaccines as: (1) At the sequence level, protein families unique to OW sand flies have evolved at a similar pace than those shared between OW and NW species, yet at a slower pace than the families unique to NW sand flies; (2) The evolutionary rates of both the protein/gene families unique to or shared between sand fly groups range from more divergent to more conserved families; (3) Gene families unique to NW sand flies have been diverging for the most part due to positive selection whereas

alignment of Phospholipase A2 proteins. ParSP11 (P. ariasi), PorMSP129 (P. orientalis), PpeSP18 (P. perniciosus), PkanSP25 (P. kandelakki), and PabSP52 (P. arabicus). Black background shading represents identical amino acids. Gray background shading represents similar amino acids. Asterisks indicate the conserved cysteine residues. (B) The evolutionary history of Phospholipase A2 salivary protein family was inferred by using the Maximum Likelihood method based on the JTT matrix-based model (Jones et al., 1992). Sand fly species are indicated by the different symbols in the legend on the right.

PabSP288 (P. arabicus), and PagSP132 (P. argentipes). Black background shading represents identical amino acids. Gray background shading represents similar amino acids. (B) The evolutionary history of Pyrophosphatase salivary protein family was inferred by using the Maximum Likelihood method based on the Whelan And Goldman model (Whelan and Goldman, 2001). Sand fly species are indicated by the different symbols in the legend on the right. Tree branches were color-coded so as to represent specific taxon: Green color represents the Larroussius and Adlerius subgenera; Red color indicates the Euphlebotomus subgenus; Blue color points to proteins of the Phlebotomus and Paraphlebotomus subgenera.



Y, protein expressed in the salivary glands.

N, lack of expression in the salivary glands.

the families shared between NW and OW (except ParSP17) as well as those unique to OW sand flies are less divergent due to purifying selection; (4) Events of gene duplication are more often observed in the gene families unique to NW sand fly species, such as SALO, RGD, mannose-binding lectin, c-type lectin, and spider toxin-like, whereas the emergence of new gene copies in OW sand flies is more evident in the gene families shared with NW sand flies, such as antigen-5, apyrase, D7, Silk, yellow-related, and OBPs; (5) Different from NW sand flies (Abdeladhim et al., 2016), the emergence of salivary gland gene families in OW sand flies relies less often on ancient genes bearing sequences encoding conserved cysteine signatures; and (6) a few salivary protein gene families is only expressed in sand flies belonging to specific sub-genera within OW sand flies; for instance, ParSP15 is unique to the Larrossious sub-genus, and SP2.5 is only expressed in members of the Phlebotomus subgenus. Overall, these findings highlight the striking differences in the rates of molecular evolution of salivary protein encoding genes in NW and OW sand flies, which are underscored not only by sequence polymorphisms but also by copy number variation.

Also, sand fly salivary proteins are recognized by the humoral immune system of humans and animal reservoirs previously bitten by sand flies; therefore, some sand fly salivary proteins have been shown to work as biomarkers of vector exposure. For such proteins to be used as biomarker of vector exposure for the identification of taxon-specific sand fly bites, the candidates are supposed to be not only immunogenic and conserved but also expressed only within sand flies belonging to a specific taxon. For species-specific biomarkers the candidates should be divergent or with some level of homology that allows the selection of specific peptides. For specific salivary proteins to be used as vaccine components across species, we predict that the candidates need not only to be immunogenic but also conserved proteins to some extent.

Biomarkers of vector exposure have been identified among the salivary proteins of Lu. longipalpis (Teixeira et al., 2010), Lu. intermedia (Carvalho et al., 2017), P. papatasi (Marzouki et al., 2015), P. perniciosus (Drahota et al., 2014; Kostalova et al., 2017), and P. orientalis (Sima et al., 2016). For the most part, the best salivary protein candidates belong to conserved protein families. Although the less conserved Silk protein (**Figure 3**) was found to be the best biomarker of P. papatasi exposure (Marzouki et al., 2015), members of the apyrase, ParSP25, and antigen-5 conserved families have been shown as good markers of exposure in P. perniciosus (Drahota et al., 2014; Kostalova et al., 2017), P. orientalis (Sima et al., 2016), and Lu. intermedia (Carvalho et al., 2017), respectively. It is worth noting that proteins of the conserved yellow family (**Figure 3**) have been found to be the best biomarkers of vector exposure in dogs and humans amongst distantly related sand fly species, such as Lu. longipalpis, P. orientalis, and P. perniciosus (Teixeira et al., 2010).

Regarding sand fly salivary protein vaccines, SALO (Gomes et al., 2008) and lufaxin (Collin et al., 2009) from Lu. longipalpis protects hamsters and dogs, respectively, against Leishmania infantum infection. On the other hand, PdSP15 (PduM02 herein; OBP family) from P. duboscqi (and its P. papatasi ortholog) protects mice (Valenzuela et al., 2001) and non-human primates (Oliveira et al., 2015) against Leishmania major infection. As SALO is the most divergent among the unique salivary protein families in NW sand flies (**Figure 3**), it is more likely to work as a specific vaccine for L. longipalpis transmitted leishmaniasis. Regarding PdSP15 (PduM02), it belongs to the diverse OBP (super-) family, yet it only shares orthologs with P. papatasi and P. sergenti (**Supplementary Figure 15**). In fact, PdSP15 orthologs are relatively conserved (% divergence = 36.3%; ω = 0.54). Hence, a PdSP15-based vaccine is restricted to protect leishmania transmission from sand flies of the P/P subgenera. On the other hand, lufaxin is likely the best candidate for a pan-specific vaccine, as it is conserved and shared between OW and NW sand flies (**Figure 3** and **Supplementary Figure 9**).

The data pointing out yellow proteins as the best biomarkers of vector exposure across sand fly species underscores the idea that conserved salivary proteins from the saliva of different sand fly species are recognized, processed, and presented in similar ways by the host immune systems. Thereby, it gives supports to the possibility that conserved yellow epitopes can be used as a pan-biomarker for sand fly exposure. For the same reasons, it is possible that conserved lufaxin epitopes can be used as panspecific vaccines against leishmaniasis transmitted by different sand fly species.

ParSP25 proteins were shown to work as biomarkers of vector exposure for P. perniciosus and P. orientalis (Drahota et al., 2014; Sima et al., 2016; Kostalova et al., 2017). Such a protein family was among the least divergent families unique to OW sand flies (**Figures 2A,B**) and was only expressed in member of the L/A/E subgenera (**Table 1**). Thereby, it is likely that ParSP25 epitopes can be used as a taxon-specific biomarker of vector exposure to recognize bites of sand flies from L/A/E subgenera and distinguish them from bites of sand flies belonging to P/P ones that display overlapping habitats.

The evolution of salivary gland gene families unique to NW and to OW sand flies has faced completely different selective pressures, leading to a much faster pace of sequence and copy number variation in NW sand flies and a more constrained divergence in OW species. These evolutionary differences are further highlighted by the fact that the emergence of new gene families unique to NW sand flies relied upon sequences encoding specific cysteine codons, which were less evident in the genes encoding the protein families unique to OW sand flies. Regarding the protein families shared between OW and NW sand flies, the rates of sequence divergence were similar between OW and NW sand flies, yet more gene duplication events were noticed in OW species. Evolutionary differences are noticed even within OW sand flies, as more gene duplication are observed in P/P than in L/A/E, and the expression of some gene families was restricted to either P/P or L/A/E sand fly taxon.

As the salivary protein families shared between OW and NW sand flies are inter-specifically conserved, such proteins are suitable for the development of pan-specific biomarker of vector exposure and pan-specific vaccines. Along the same lines, the more conserved protein families unique to OW sand flies are ideal candidates for the development of taxon-specific biomarkers of vector exposure, as even the expression of such proteins can be restricted to a specific taxon. On the other hand, the development of species-specific biomarkers of vector exposure would require the identification of species-specific epitopes amongst the most divergent proteins families unique to either NW or OW sand flies.

# AUTHOR CONTRIBUTIONS

IC-A performed the evolutionary analysis and wrote the manuscript. JV wrote and edited the manuscript.

# ACKNOWLEDGMENTS

We would like to thank Dr. Daniel Sonenshine (Old Dominion University/LMVR-NIH) for critically reading the manuscript. Support for this work was provided by the Intramural Research Program at the National Institute of Allergy and Infectious Diseases, National Institutes of Health.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2018.00290/full#supplementary-material

Supplementary Figure 1 | Multiple sequence alignment and molecular phylogenetic analysis of the sand fly SP16 salivary protein family. (Top) Multiple sequence alignment of SP16 proteins. PPTSP14.3 (P. papatasi), PsSP98 (P. sergenti), PorASP152 (P. orientalis), PabSP63 (P. arabicus), and PagSP73 (P. argentipes). Black background shading represents identical amino acids. Gray background shading represents similar amino acids. Asterisks indicate the conserved cysteine residues. (Bottom) The evolutionary history of SP16 salivary protein family was inferred by using the Maximum Likelihood method based on the Whelan And Goldman model (Whelan and Goldman, 2001). Sand fly species are indicated by the different symbols in the legend on the right. Sand fly species are indicated by the different symbols in the legend on the right. Tree branches were color-coded so as to represent specific taxon: Green color represents the Larroussius and Adlerius subgenera; Red color indicates the Euphlebotomus subgenus; Blue color points to proteins of the Phlebotomus and Paraphlebotomus subgenera.

Supplementary Figure 2 | Multiple sequence alignment of the sand fly ParSP23 salivary protein family. ParSP23 (P. ariasi) and PabSP56 (P. arabicus). Black background shading represents identical amino acids. Gray background shading represents similar amino acids.

Supplementary Figure 3 | Multiple sequence alignment of the sand fly SP2.5-like salivary protein family. PPTSP2.5 (P. papatasi) and PduM80 (P. duboscqi). Black background shading represents identical amino acids. Gray background shading represents similar amino acids.

Supplementary Figure 4 | Multiple sequence alignment and molecular phylogenetic analysis of the sand fly Antigen-5 salivary protein family. (Top) Multiple sequence alignment of Antigen-5 proteins. PPTSP29 (P. papatasi), PduK107 and PduM48 (P. duboscqi), ParSP05 (P. ariasi), PorASP74 (P. orientalis), PpeSP07 (P. perniciosus), PkanSP14 (P. kandelakki), PabSP4 (P. arabicus), PagSP05 (P. argentipes), Luloag-5 (Lu. longipalpis), LayS79 (Lu. ayacuchensis), Linb-13 (N. intermedia), and LolAg5a (B. olmeca). Black background shading represents identical amino acids. Gray background shading represents similar amino acids. Asterisks indicate the conserved cysteine residues. (Bottom) The evolutionary history of Antigen-5 salivary protein family was inferred by using the Maximum Likelihood method based on the Whelan And Goldman model (Whelan and Goldman, 2001). Sand fly species are indicated by the different symbols in the legend on the right. Tree branches were color-coded so as to represent specific taxon: Green color represents the Larroussius and Adlerius subgenera; Red color indicates the Euphlebotomus subgenus; Blue color points to proteins of the Phlebotomus and Paraphlebotomus subgenera; and Black color indicates the proteins belonging to New World sand flies.

Supplementary Figure 5 | Multiple sequence alignment and molecular phylogenetic analysis of the sand fly Apyrase salivary protein family. (Top) Multiple sequence alignment of Apyrase. PPTSP36 (P. papatasi), PduM38 and PduM39 (P. duboscqi), PsSP42 (P. sergenti), PorMSP3 and PorMSP4 (P. orientalis), PpeSP01 and PpeSP01B (P. perniciosus), ParSP01 (P. ariasi), PtSP4 (P. tobbi), PkanSP09 (P. kandelakki), PabSP40 (P. arabicus), PagSP03 (P. argentipes), LuloApy (Lu. longipalpis), LayS17 (Lu. ayacuchensis), Linb-35 (N. intermedia), and LolApy (B. olmeca). Black background shading represents identical amino acids. Gray background shading represents similar amino acids. (Bottom) The evolutionary history of Apyrase salivary protein family was inferred by using the Maximum Likelihood method based on the Le\_Gascuel\_2008 model (Gomes et al., 2008). Sand fly species are indicated by the different symbols in the legend on the right. Tree branches were color-coded so as to represent specific taxon: Green color represents the Larroussius and Adlerius subgenera; Red color indicates the Euphlebotomus subgenus; Blue color points to proteins of the Phlebotomus and Paraphlebotomus subgenera; and Black color indicates the proteins belonging to New World sand flies.

Supplementary Figure 6 | Multiple sequence alignment and molecular phylogenetic analysis of the sand fly D7 salivary protein family. (Top) Multiple sequence alignment of D7. PPTSP28a (P. papatasi), PduK69 and Pduk103 (P. duboscqi), PsSP7 (P. sergenti), PpeSP04 and PpeSP04B (P. perniciosus), ParSP12 and ParSP16 (P. ariasi), PorMSP28 and PorMSP38 and PorMSP43 (P. orientalis), PtSP42 and PtSP54 and PtSP57 (P. tobbi), PkanSP11 and PkanSP12 (P. kandelakki), PabSP20 and PabSP54 and PabSP59 and PabSP84 (P. arabicus), LJL13 (Lu. longipalpis), LolD7 (B. olmeca), LayS101 (Lu. ayacuchensis), and Linb-42 (N. intermedia). Black background shading represents identical amino acids. Gray background shading represents similar amino acids. Asterisks indicate the conserved cysteine residues. (Bottom) The evolutionary history of D7 salivary protein family was inferred by using the Maximum Likelihood method based on the Whelan And Goldman model (Whelan and Goldman, 2001). Sand fly species are indicated by the different symbols in the legend on the right. Tree branches were color-coded so as to represent specific taxon: Green color represents the Larroussius and Adlerius subgenera; Red color indicates the Euphlebotomus subgenus; Blue color points to proteins of the Phlebotomus and Paraphlebotomus subgenera; and Black color indicates the proteins belonging to New World sand flies.

Supplementary Figure 7 | Multiple sequence alignment and molecular phylogenetic analysis of the sand fly Endonuclease salivary protein family. (Top) Multiple sequence alignment of Endonuclease. Black background shading represents identical amino acids. ParSP10 (P. ariasi), PpeSP32 (P. perniciosus), PorMSP101 (P. orientalis), PkanSP26 (P. kandelakki), PabSP49 (P. arabicus), PagSP11 (P. argentipes), LolEndo (B. olmeca), Linb-46 (N. intermedia), LJL138 (Lu. longipalpis), and LayS147 (Lu. ayacuchensis). Gray background shading represents similar amino acids. Asterisks indicate the conserved cysteine residues. (Bottom) The evolutionary history of Endonuclease salivary protein

family was inferred by using the Maximum Likelihood method based on the Whelan And Goldman model (Whelan and Goldman, 2001). Branches corresponding to partitions reproduced in <50% bootstrap replicates are collapsed. Sand fly species are indicated by the different symbols in the legend on the right. Tree branches were color-coded so as to represent specific taxon: Green color represents the Larroussius and Adlerius subgenera; Red color indicates the Euphlebotomus subgenus; Blue color points to proteins of the Phlebotomus and Paraphlebotomus subgenera; and Black color indicates the proteins belonging to New World sand flies.

Supplementary Figure 8 | Multiple sequence alignment and molecular phylogenetic analysis of the sand fly Hyaluronidase salivary protein family. (Top) Multiple sequence alignment of Hyaluronidase. PperHyal (P. perniciosus), PorASP112 (P. orientalis), PtSP125 (P. tobbi), PkanSP21 (P. kandelakki), PabSP72 (P. arabicus), Linb-54 (N. intermedia), LolHyaz (B. olmeca), LJLHYAL (Lu. longipalpis). Black background shading represents identical amino acids. Gray background shading represents similar amino acids. (Bottom) The evolutionary history of Hyaluronidase salivary protein family was inferred by using the Maximum Likelihood method based on the General Reversible Chloroplast model (Adachi et al., 2000). Sand fly species are indicated by the different symbols in the legend on the right. Tree branches were color-coded so as to represent specific taxon: Green color represents the Larroussius and Adlerius subgenera; Red color indicates the Euphlebotomus subgenus; Blue color points to proteins of the Phlebotomus and Paraphlebotomus subgenera; and Black color indicates the proteins belonging to New World sand flies.

Supplementary Figure 9 | Multiple sequence alignment and molecular phylogenetic analysis of the sand fly Lufaxin salivary protein family. (Top) Multiple sequence alignment of Lufaxin. PPTSP34 (P. papatasi), PduM04 (P. duboscqi), PsSP49 (P. sergenti), PpeSP06 (P. perniciosus), ParSP09 (P. ariasi), PtSP66 (P. tobbi), PkanSP16 (P. kandelakki), PabSP32 (P. arabicus), PagSP05 (P. argentipes), Lolaxin (B. olmeca), Lufaxin (Lu. longipalpis), LayS26 (L. ayacuchensis), and Linb-17 (N. intermedia). Black background shading represents identical amino acids. Gray background shading represents similar amino acids. Asterisks indicate the conserved cysteine residues. (Bottom) The evolutionary history of Lufaxin salivary protein family was inferred by using the Maximum Likelihood method based on the Whelan And Goldman model (Whelan and Goldman, 2001). Sand fly species are indicated by the different symbols in the legend on the right. Tree branches were color-coded so as to represent specific taxon: Green color represents the Larroussius and Adlerius subgenera; Red color indicates the Euphlebotomus subgenus; Blue color points to proteins of the Phlebotomus and Paraphlebotomus subgenera; and Black color indicates the proteins belonging to New World sand flies.

Supplementary Figure 10 | Multiple sequence alignment of the sand fly ParSP17 salivary protein family. PPTSP56.6 (P. papatasi), PsSP82 (P. sergenti), PpeSP19 (P. perniciosus), ParSP17 (P. ariasi), PtSP49 (P. tobbi), PkanSP17 (P. kandelakki), PabSP53 (P. arabicus), LJM78 (Lu. longipalpis). Black background shading represents identical amino acids. Gray background shading represents similar amino acids.

Supplementary Figure 11 | Multiple sequence alignment and molecular phylogenetic analysis of the sand fly ParSP80 salivary protein family. PduK110 (P. duboscqi), ParSP80 (P. ariasi), PabSP91 (P. arabicus), and LJS138 (Lu. longipalpis). Black background shading represents identical amino acids. Gray background shading represents similar amino acids.

Supplementary Figure 12 | Multiple sequence alignment and molecular phylogenetic analysis of the sand fly Silk salivary protein family. (Top) Multiple sequence alignment of Silk. PPTSP32 (P. papatasi), PduK45 and PduM33 (P. duboscqi), PsSP44 (P. sergenti), PorASP86 (P. orientalis), ParSP02 (P. ariasi), PpeSP05 (P. perniciosus), PtSP29 (P. tobbi), PkanSP15 (P. kandelakki), PabSP30 (P. arabicus), PagSP06 (P. argentipes), Lolsilk (B. olmeca), LJL04 (Lu. longipalpis), LayS89 (Lu. ayacuchensis), Linb-26 (N. intermedia). Black background shading represents identical amino acids. Gray background shading represents similar amino acids. (Bottom) The evolutionary history of Silk salivary protein family was inferred by using the Maximum Likelihood method based on the JTT matrix-based model (Jones et al., 1992). Sand fly species are indicated by the different symbols in the legend on the right. Tree branches were color-coded so as to represent specific taxon: Green color represents the Larroussius and Adlerius subgenera; Red color indicates the Euphlebotomus subgenus; Blue color points to proteins of the Phlebotomus and Paraphlebotomus subgenera; and Black color indicates the proteins belonging to New World sand flies.

Supplementary Figure 13 | Multiple sequence alignment and molecular phylogenetic analysis of the sand fly Yellow salivary protein family. (Top) Multiple sequence alignment of Yellow. PPTSP42 and PPTSP44 (P. papatasi), PduK06 and PduM10 (P. duboscqi), PsSP22 and PsSP26 (P. sergenti), ParSP04 and ParSP04b (P. ariasi), PorASP2 and PorASP4 (P. orientalis), PtSP37 and PtSP38 (P. tobbi), PpeSP03B and PpeSP03 (P. perniciosus), PkanSP04 (P. kandelakki), PabSP26 (P. arabicus), PagSP04 (P. argentipes), LolYLWb and LolTLWc and LolYLWA (B. olmeca), Linb-21 (N. intermedia), LayS22 and LayS118 (Lu. ayacuchensis), and LJM17 and LJM17 and LJM111 (Lu. longipalpis). Black background shading represents identical amino acids. Gray background shading represents similar amino acids. Asterisks indicate the conserved cysteine residues. (Bottom) The evolutionary history of Yellow salivary protein family was inferred by using the Maximum Likelihood method based on the Whelan And Goldman model (Whelan and Goldman, 2001). Sand fly species are indicated by the different symbols in the legend on the right. Tree branches were color-coded so as to represent specific taxon: Green color represents the Larroussius and Adlerius subgenera; Red color indicates the Euphlebotomus subgenus; Blue color points to proteins of the Phlebotomus and Paraphlebotomus subgenera; and Black color indicates the proteins belonging to New World sand flies. Names in bold represent 44 kDa protein homologs.

Supplementary Figure 14 | Multiple sequence alignment of the sand fly Adenosine deaminase salivary protein family. PduM73 (P. duboscqi), PpeAda (P. perniciosus), and LJLAda (L. longipalpis). Black background shading represents identical amino acids. Gray background shading represents similar amino acids.

Supplementary Figure 15 | Multiple sequence alignment and molecular phylogenetic analysis of the sand fly small Odorant Binding Protein (OBPs) salivary protein family. (Top) Multiple sequence alignment of OBPs. PPTSP12 and PPTSP14 and PPTSP14.2a and PPTSP15 (P. papatasi), PduM12 and PduM60 and PduM07 and PduM50 and PduM57 and PduM31 and PduM49 and PduM58 and PduM62 and PduM99 and PduM02 and PduM03 and PduM06 (P. duboscqi), PsSP14 and PsSP14 and PsSP55 and PsSP9 (P. sergenti), PtSP9 and PtSP17 and PtSP32 and PtSP31 and PtSP18 and PtSP23 (P. tobbi), PpeSP02 and PpeSP09 and PpeSP11 (P. perniciosus), PorASP28 and PorASP31 and PorASP37 and PorASP61 and PorASP61 and PorASP64 (P. orientalis), ParSP03 and ParSP01 and ParSP08 (P. ariasi), PkanSP05 and PkanSP06 and PkanSP07 (P. kandelakki), PabSP45 and PabSP2 (P. ariasi), PagSP93 and PagSP07 and PagSP01 and PagSP12 and PagSP02 and PagSP13 (P. argentipes), LolSOBPa and LolSOBPb and LolSOBPc (B. olmeca), Linb-7 and Linb-8 and Linb-28 (N. intermedia), LayS69 (Lu. ayacuchensis), and LuloOBP (Lu. longipalpis). Black background shading represents identical amino acids. Gray background shading represents similar amino acids. Asterisks indicate the conserved cysteine residues. (Bottom) The evolutionary history of OBPs salivary protein family was inferred by using the Maximum Likelihood method based on the Le\_Gascuel\_2008 model (Gomes et al., 2008). Sand fly species are indicated by the different symbols in the legend on the right. Tree branches were color-coded so as to represent specific taxon: Green color represents the Larroussius and Adlerius subgenera; Red color indicates the Euphlebotomus subgenus; Blue color points to proteins of the Phlebotomus and Paraphlebotomus subgenera; and Black color indicates the proteins belonging to New World sand flies.

Supplementary Table 1 | Nucleotide and protein accession numbers.

# REFERENCES


protein against Leishmania braziliensis infection. PLoS Negl. Trop. Dis. 7:e2242. doi: 10.1371/journal.pntd.0002242


leishmaniasis is influenced by previous exposure to sandfly bites in Saudi Arabia. PLoS Negl. Trop. Dis. 9:e0003449. doi: 10.1371/journal.pntd.0003449


**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 Coutinho-Abreu and Valenzuela. 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.

# Biomarkers Associated With *Leishmania infantum* Exposure, Infection, and Disease in Dogs

### Carla Maia\* and Lenea Campino

*Global Health and Tropical Medicine (GHTM), Instituto de Higiene e Medicina Tropical (IHMT), Universidade Nova de Lisboa (UNL), Lisbon, Portugal*

Canine leishmaniosis (CanL) is a vector-borne disease caused by the protozoan *Leishmania* (*Leishmania*) *infantum* species [syn. *L.* (*L.*) *infantum chagasi* species in the Americas] which is transmitted by the bite of a female phlebotomine sand fly. This parasitosis is endemic and affect millions of dogs in Asia, the Americas and the Mediterranean basin. Domestic dogs are the main hosts and the main reservoir hosts for human zoonotic leishmaniosis. The outcome of infection is a consequence of intricate interactions between the protozoan and the immunological and genetic background of the host. Clinical manifestations can range from subclinical infection to very severe disease. Early detection of infected dogs, their close surveillance and treatment are essential to control the dissemination of the parasite among other dogs, being also a pivotal element for the control of human zoonotic leishmaniosis. Hence, the identification of biomarkers for the confirmation of *Leishmania* infection, disease and determination of an appropriate treatment would represent an important tool to assist clinicians in diagnosis, monitoring and in giving a realistic prognosis to subclinical infected and sick dogs. Here, we review the recent advances in the identification of *Leishmania infantum* biomarkers, focusing on those related to parasite exposure, susceptibility to infection and disease development. Markers related to the pathogenesis of the disease and to monitoring the evolution of leishmaniosis and treatment outcome are also summarized. Data emphasizes the complexity of parasite-host interactions and that a single biomarker cannot be used alone for CanL diagnosis or prognosis. Nevertheless, results are encouraging and future research to explore the potential clinical application of biomarkers is warranted.

### *Edited by:*

*Javier Moreno, Instituto de Salud Carlos III, Spain*

### *Reviewed by:*

*Gabriele Rossi, Murdoch University, Australia Clarisa B. Palatnik-de-Sousa, Universidade Federal do Rio de Janeiro, Brazil*

> *\*Correspondence: Carla Maia carlamaia@ihmt.unl.pt*

### *Specialty section:*

*This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology*

*Received: 06 April 2018 Accepted: 07 August 2018 Published: 06 September 2018*

### *Citation:*

*Maia C and Campino L (2018) Biomarkers Associated With Leishmania infantum Exposure, Infection, and Disease in Dogs. Front. Cell. Infect. Microbiol. 8:302. doi: 10.3389/fcimb.2018.00302* Keywords: biomarkers, dog, exposure, infection, *Leishmania infantum*, leishmaniosis

# CANINE LEISHMANIOSIS

Canine leishmaniosis (CanL) is a vector-borne zoonotic protozoan disease caused by Leishmania (Leishmania) infantum species [syn. L. (L.) infantum chagasi species in the Americas; (Mauricio, 2018)] which is transmitted by the bite of a female phlebotomine sand fly. CanL is endemic in approximately 50 countries and affects millions of dogs in Asia, the Americas, and the Mediterranean basin (WHO, 2010; Campino and Maia, 2018).

The outcome of infection in dogs is a consequence of intricate interactions between the protozoan L. infantum and the genetic background of the host (Baneth et al., 2008; Solano-Gallego et al., 2011; Campino and Maia, 2018). In addition, several non-genetic factors of the host, such as age, breed, gender, concomitant infections, immunological, and nutritional status as well as parasite virulence and previous exposure to Leishmania parasites can also affect infection outcome (Miró et al., 2008; Saridomichelakis, 2009; Hosein et al., 2017; Campino and Maia, 2018). The presence of L. infantum parasites in dogs can manifest as chronic infection without clinical signs lasting several years, self-limiting or severe illness that can rapidly progress to death (Solano-Gallego et al., 2009; Paltrinieri et al., 2010). In fact, not all dogs exposed to the parasite develop clinical signs and asymptomatic infections are much more frequent than clinical disease. On the other hand, a subclinical infection is not necessarily permanent and the break of parasite-host equilibrium can lead to the development of patent disease (Solano-Gallego et al., 2011). The progression of disease in susceptible dogs is characterized by an exacerbated humoral immune response, a depression of cellular immune response against the parasite, and the appearance of a panoply of clinical signs and/or physiopathological alterations. On the other hand, dogs considered resistant do not present clinical signs, have low levels of specific antibodies and low parasite levels, and present a robust cell-mediated immune response (Solano-Gallego et al., 2009; Paltrinieri et al., 2010; Maia and Campino, 2012; Hosein et al., 2017).

The commonest clinical signs in dogs with CanL are atrophic myositis of masticatory muscles, cutaneous alterations, lymphadenomegaly, onychogryphosis, and lesions derived from immune-complexes deposition such as glomerulonephritis, polyarthritis, or uveitis (Ciaramella et al., 1997; Maia and Campino, 2008; Paltrinieri et al., 2010; Solano-Gallego et al., 2011; Koutinas and Koutinas, 2014; Noli and Saridomichelakis, 2014; Meléndez-Lazo et al., 2018).

According to the Biomarkers Definition Working Group (2001), biomarkers are "biological parameters that can be objectively measured and evaluated as indicators of physiological or pathological processes, or a response to a therapeutic intervention," and have been widely used in understanding several aspects of non-infectious and infectious diseases, such as exposure and susceptibility to infection (Carretón et al., 2014; Bryan, 2016). Despite clinical staging systems of CanL based on physiopathological abnormalities, clinical signs, serological alterations and/or direct detection of the parasite have been proposed (Solano-Gallego et al., 2009; Paltrinieri et al., 2010; Foglia Manzillo et al., 2013), the specific diagnosis of L. infantum is still a challenge (Maia and Campino, 2008). Apart from the confirmation of disease, other reasons for attempting laboratory diagnosis are the confirmation of Leishmania infection (in epidemiological surveys, to prevent infected dogs to be blood donors or be imported to non-endemic countries), and the monitoring of the response to treatment (Solano-Gallego et al., 2017; Campino and Maia, 2018). Therefore, the identification of reliable predictors for each purpose would represent an important tool to assist clinicians in follow-up monitoring, in determining an appropriate treatment and in giving a realistic prognosis to apparently healthy and sick dogs.

Here, we review recent research regarding the identification of biomarkers associated with L. infantum exposure, infection and disease in dogs and biomarkers useful to follow-up CanL and treatment efficacy.

# BIOMARKERS OF EXPOSURE TO *Leishmania infantum* OR *L.* (*L.*) *infantum chagasi* VECTORS

Leishmania protozoa are transmitted by the bites of infected phlebotomine sand flies; thus, as the insect takes a blood meal its saliva is injected into the vertebrate host. Sand fly salivary glands secrete a complex array of active compounds that facilitate blood feeding and modulate host immune response having important consequences on the establishment or abrogation of infection (Lestinova et al., 2017). In addition, several components present in phlebotomine sand fly saliva are immunogenic to vertebrate hosts leading to development of saliva-reactive antibodies (Collin et al., 2009; Teixeira et al., 2010; Martín-Martín et al., 2014). A positive correlation between the level of specific antibodies to Phlebotomus perniciosus and Lutzomyia longipalpis saliva and the number of phlebotomine sand flies blood-fed have experimentally been demonstrated (Hostomska et al., 2008; Vlkova et al., 2011). Data from dogs living in endemic areas of leishmaniosis suggest the use of antibody response to saliva compounds as epidemiological biomarkers for monitoring vector exposure (Teixeira et al., 2010; Solcà et al., 2016; Kostalova et al., 2017; Quinnell et al., 2018). Moreover, the intensity of vector-exposure can be monitored throughout phlebotomine sand fly season as the host anti-saliva antibody response rapidly decreases in canine sera within 1 week after the last P. perniciosus exposure (Vlkova et al., 2011). Nevertheless, whether the antigenic response to phlebotomine sand fly saliva, which is developed whether the sand fly is infected or not, could be used as a risk marker for Leishmania infection in dogs remains controversial, as positive (Kostalova et al., 2015, 2017), negative (Vlkova et al., 2011) or no (Kostalova et al., 2017) correlation between the levels of anti-P. perniciosus saliva and antibodies against the parasite have been reported. Similar results were obtained with L. longipalpis; while Solcà et al. (2016) observed an increase in the number of dogs displaying antibodies to L. longipalpis saliva along with high parasite load, and therefore with disease severity, in the work performed by Quinnell et al. (2018) no association was observed between exposure to sand fly bites and disease progression. One reason for these contradictory results could be the fact that, with the exception of the work performed by Kostalova et al. (2017), anti-saliva antibodies have been detected using the whole content of the salivary glands reducing the specificity of detection due to a higher likelihood of cross-reactivity with saliva components from other sympatric non-vector phlebotomine sand fly species (Andrade and Teixeira, 2012; Lestinova et al., 2017).

# GENETIC BIOMARKERS OF SUSCEPTIBILITY TO *Leishmania infantum* INFECTION AND DISEASE

Genetic markers can be the responsible for the phenotypic variance of susceptibility of dogs to L. infantum as they control both pro- and anti-inflammatory cytokines as well as the cellular immune response to the presence of the parasite (de Vaconcelos et al., 2017).

Mutations and polymorphisms of the natural resistanceassociated macrophage protein 1, NRAMP1 gene (synonym of the solute carrier family 11 member 1, Slc11a1) have been associated with susceptibility to disease (Altet et al., 2002; Sanchez-Robert et al., 2005, 2008). Slc11a1 gene encodes an ion transporter protein involved in the control of multiplication of Leishmania amastigotes and in macrophage activation. A predisposition to CanL has been associated with the haplotype of T antigen epitope TAG-8-141 and with two single nucleotide polymorphisms (SNPs) (A4549G in intron 6 and C4859T in exon 8) located in the Slc11a1 gene in Boxer breed (Sanchez-Robert et al., 2005) and in different dog breeds (Sanchez-Robert et al., 2008), respectively. However, no significant differences in the expression of Slc11a1 gene between phenotypically resistant and susceptible dogs or between primary canine monocyte-derived macrophages from Leishmania-free dogs with higher or lower resistance to intracellular survival of the parasites were found in the studies performed by Bueno et al. (2009) and Turchetti et al. (2015). The presence of the beta chain allele of the dog leukocyte antigen DLA–DRB1<sup>∗</sup> 01502 has been associated with susceptibility to CanL (Quinnell et al., 2003b) while the detection of SNPs 3, 4, 7 and 8 in the canine β-defensin-1 gene has been associated with susceptibility to L. infantum or L. (L.) infantum chagasi infection (da Silva et al., 2017).

A genome-wide analysis using a dataset of 115 infected and 104 sick Boxers was assessed, as it is believed that multiple loci are responsible for the progression of Leishmania infection to clinical disease (Quilez et al., 2012). More than 170,000 single SNPs distributed throughout the genome were identified and, according to the authors, the significant predictive value of this genomic information may predict with an accuracy of ∼0.29 the resistant and susceptible phenotype. Utsunomiya et al., 2015) identified two candidate loci in chromosome 1 and 2 involved in Leishmania infection using a panel of 145,000 SNPs distributed throughout the canine genome of 48 mixed-breed dogs (20 animals with PCR and ELISA positive to Leishmania and 28 negative controls). Candidate marker on chromosome 1 is related with notch signaling, which is key for macrophage activity and for T cell cluster of differentiation 4 (CD4+), while the candidate marker of chromosome 2 is related with the expression of interleukin 2 (IL-2) and IL-15, two cytokines with a pivotal role in the control and resolution of Leishmania infection (Utsunomiya et al., 2015).

# HEMATOLOGIC AND BIOCHEMICAL BIOMARKERS OF CANL

In infected dogs without or with light localized clinical signs, in which Leishmania presence was confirmed through direct methods and which have negative or low-titer anti-Leishmania antibodies, hematological and biochemical parameters are not usually changed (Solano-Gallego et al., 2009; Paltrinieri et al., 2010). On the other hand, laboratorial abnormalities are common in dogs with CanL (**Table 1**).

# Diagnostic Markers

In dogs with clinical leishmaniosis, mild to moderate normocytic and normochromic non-regenerative anemia, typical of a chronic inflammatory disease, is the most common hematological abnormality (Reis et al., 2006b; Paltrinieri et al., 2016) and may be caused by decreased erythropoiesis due to disorders in the erythroid bone marrow compartment, or a decreased eryhopoietin production due to chronic renal failure. Although not so common, if anemia is due to an increased hemolysis (demonstrated by a positive Coombs test), macrocytic hypochromic regenerative anemia can be present.

Neutrophilia is also common in dogs with CanL (Ciaramella et al., 1997; Koutinas et al., 1999; Reis et al., 2006a; Paltrinieri et al., 2010; Nicolato et al., 2013; Meléndez-Lazo et al., 2018), and may be due to the inflammatory response resulting from the presence of parasites in multiple organs (Torrecilha et al., 2016). Leishmania infantum infection causes oxidative stress [i.e., a disruption in the normal balance between the production of reactive oxygen species (ROS) and antioxidant defenses] of canine neutrophils. The release of ROS from phagocytes present in inflammatory sites, leads to the consumption of antioxidant compounds (Torrecilha et al., 2016), representing a mechanism use by the parasite to evade the immune system. In fact, dogs presenting clinical signs have reduced antioxidant levels and increased levels of oxidants with enhanced lipid peroxidation (Heidarpour et al., 2012; Almeida et al., 2017). Other less common changes observed in leukocyte populations include monocytosis, lymphopenia, eosinopenia, or leukopenia (Ciaramella et al., 1997; Koutinas et al., 1999; Reis et al., 2006a; Paltrinieri et al., 2010; Nicolato et al., 2013; Meléndez-Lazo et al., 2018). Thrombocytopenia is also a common finding in dogs with leishmaniosis (Koutinas and Koutinas, 2014; Paltrinieri et al., 2016). Other coagulation disorders such as serum hyperviscosity, thrombocytopathy, impaired secondary hemostasis and fibrinolysis, hyperfibrinogenemia, increase in prothrombin and activated partial thromboplastin times have also been documented (Ciaramella et al., 2005; Petanides et al., 2008; Paltrinieri et al., 2010).

Protein alterations such as serum polyclonal α- β- and γhyperglobulinemia, hyperproteinemia, hypoalbuminemia and decreased albumin/globulin (A/G) ratio have been associated with disease progression (Ciaramella et al., 1997; Koutinas et al., 1999; Giunchetti et al., 2008b; Meléndez-Lazo et al., 2018). In fact, hyperglobulinemia in CanL is harmful, via the production of autoantibodies (e.g., immune-mediated thrombocytopenia), and/or circulating immune complexes generated in profuse amounts (Koutinas and Koutinas, 2014).

Elevation of hepatic and renal biochemical parameters are also commonly associated with the progression of the disease (Ciaramella et al., 1997; Koutinas et al., 1999; Giunchetti et al., 2008b; Meléndez-Lazo et al., 2018). Increase of acute phase proteins-APPs (e.g., C-reactive protein-CRP, ferritin, haptoglobin, serum amyloid A) are also common laboratorial findings associated with CanL (Martinez-Subiela et al., 2014; Paltrinieri et al., 2016).


*(Continued)*


An increase of the activity of skeletal muscle enzymes (e.g., creatine kinase-CK, lactate dehydrogenase-LDH) has also been documented in diseased dogs (Vamvakidis et al., 2000). Reduced serum activity of some inflammatory markers such as adenosine deaminase and butyrylcholinesterase in infected dogs have recently been reported (Tonin et al., 2016). An increased expression or activity of other molecules, such as leptin (Di Loria et al., 2014), matrix metalloproteinases (Melo et al., 2011) or paraoxonase-1 (Martinez-Subiela et al., 2014), has also been reported in blood samples of dogs with CanL.

# Prognostic Markers

The presence of lymphopenia in diseased dogs is a marker of poor prognosis (Geisweid et al., 2012). In addition, as disease progresses increased apoptosis and reduced oxidation status and reactivity of neutrophils occurs (Gomez-Ochoa et al., 2010; Almeida et al., 2013a,b), which seems to be a critical mechanism of CanL pathogenesis (Almeida et al., 2017).

Regarding protein alterations, the severity of clinical score has been correlated with an increase of total proteins and a decrease in albumin concentration and therefore, hypoalbuminemia and hyperproteinemia are negative prognosis markers of CanL (Geisweid et al., 2012; Paltrinieri et al., 2016).

Severe (Paltrinieri et al., 2010) and very severe (Solano-Gallego et al., 2009) CanL with renal involvement should be suspected in the presence of proteinuria (i.e., when urinary protein creatinine ratio-UPC is equal or higher than 0.5) or renal azotemia (International Renal Interest Society; http://www.iriskidney.com/guidelines/index.html)<sup>1</sup> . Proteinuria without renal azotemia seems to be secondary to immune complexes deposition at the glomerular level (Zatelli et al., 2003). Azotemia only becomes evident in an advance stage of disease and may be associated with systemic hypertension (Baneth et al., 2008; Paltrinieri et al., 2016). However, and according to the results recently obtained by Meléndez-Lazo et al. (2018), azotemia is not a common finding in dogs, as this biochemical alteration was only present in 6% of the 51 diseased dogs. Gamma-glutamyl transferase (GGT) and N-acetyl-b-N-glucosaminidase are the most popular urinary enzymes to measure tubular injury, which may be present secondarily to glomerular damage (Paltrinieri et al., 2016).

# Treatment and Post-treatment Monitoring Markers

A decrease in globulin concentrations is expected in response to leishmanial treatment, however, A/G ratio will remain low in dogs with persistent glomerular damage and proteinuria as the concentration of albumin will remain low. Serum protein electrophoresis is more advisable to assess the efficacy of treatment and should not be run before 1 month of treatment initiation. A progressive decrease of globulins start to become evident after 2–6 weeks following treatment with antimonials (Rossi et al., 2014; Paltrinieri et al., 2016), and within 3 months following treatment with marbofloxacin (Rougier et al., 2012). However, the complete regression of electrophoretogram alterations requires at least 3–4 months (Torres et al., 2011).

Monitoring the concentration of APPs 1–2 weeks after the first administration of drugs provides earlier information regarding the success of treatment when other clinicopathological parameters are still abnormal (Paltrinieri et al., 2016). CRP and serum amyloid A values start to decrease within 2 weeks after treatment with meglumine antimoniate and return to previous values around 1 month (Martínez-Subiela et al., 2003; Rossi et al., 2014). Long-term treatment with allopurinol also significantly decreased the values of haptoglobin and CRP (Sasanelli et al., 2007).

Proteinuria tends to decrease in 4–8 weeks after treatment with allopurinol and meglumine antimoniate (Pierantozzi et al., 2013), butrestore of renal function after leishmanicidal treatment depends on the severity of renal damage at the time of diagnosis. According to IRIS Glomerular Disease Study Group et al. (2013) and Roura et al. (2013), "serum creatinine and proteinuria of dogs in IRIS stages 3 or 4 should be frequently tested during the treatment period, while those in IRIS stages 1 or 2 should be tested at the end of the first treatment cycle. Post-treatment evaluation of dogs in IRIS stage 1 should be done after one year, in IRIS stage 2 every 6 months, in IRIS stage 3 every 3 months and in IRIS stage 4 every 6 weeks."

# IMMUNOLOGICAL BIOMARKERS OF SUSCEPTIBILITY AND RESISTANCE TO CANL

The course of L. infantum infection in dogs is tied-up to complex interactions of the host innate and adaptive components of the immune response, which dictates the clearance or persistence and multiplication of the parasite. The innate immune response has an important role in protection against the parasite, besides instructing the development of long lasting adaptive response (Moreno and Alvar, 2002; Reis et al., 2010; Hosein et al., 2017). The ability of the host to control Leishmania infection requires a strong cellular immune response, associated with the activation of T helper (Th)-1 cells producing interferon-gamma (IFNγ ), tumor necrosis factor alpha (TNF-α), and IL-2. Th1 is responsible for the activation of macrophages and concomitant intracellular killing of the parasites (Barbiéri, 2006; Carrillo and Moreno, 2009; Reis et al., 2010; Maia and Campino, 2012; Hosein et al., 2017) while active disease is associated with a mixed Th1/Th2 response (Carrillo and Moreno, 2009). Nevertheless, the immune response to the parasite is organ/tissue-specific as in different organs Th1, Th2 or mixed Th1/Th2 immune responses were observed and correlated with the absence or presence of clinical signs and with local parasite load (Reis et al., 2009; Maia and Campino, 2012; Hosein et al., 2017).

## Bone Marrow

The absence of a specific immune response against Leishmania in bone marrow cells has been suggested as the expression of IFN-γ and IL-2, IL-4, IL-12 of asymptomatic and symptomatic

<sup>1</sup> International Renal Interest Society (IRIS): Guidelines for Staging Chronic Kidney Disease (CKD). Available online at: http://www.iris-kidney.com (Accessed March, 2018)

dogs was similar to those of healthy animals (Barbosa et al., 2011). Nevertheless, the progression of Leishmania infection has been mainly associated with a pro-inflammatory environment, characterized by an elevated expression of TNF-α and IFN-γ by bone marrow cells (Quinnell et al., 2001b; Foglia Manzillo et al., 2006; Rodríguez-Cortés et al., 2016) together with a significant positive correlation between IL-4 levels and disease severity (Quinnell et al., 2001b). The absence of clinical signs in experimentally infected dogs was related with the no detection of this cytokine, together with the expression of inducible nitric oxide synthetase (iNOS) and of pro-inflammatory (TNF-α) and regulatory/anti-inflammatory [transforming growth factorbeta (TGF-β) and IL-10] cytokines (Maia and Campino, 2012). As IL-12 is involved in the inflammatory process that activates macrophages and enhances their microbicidal activity, it is not surprisingly the significant increased of its expression by bone marrow cells following treatment of dogs with meglumine antimoniate and allopurinol (Barbosa et al., 2011). The expression of the major histocompatibility complex (MHCII+) by bone marrow monocytes was also significantly increased after treatment, probably reflecting a rise in the presentation of Leishmania antigens (Alexandre-Pires et al., 2010).

# Liver

According to Michelin et al. (2011), liver is the main cytokineproducing organ during infection as IL-4, IL-10 and TNF-α production from liver extracts was higher than in spleen extracts both in infected asymptomatic and symptomatic dogs. In the work performed by Correa et al. (2007) the production of IL-10 and TGF-β1 by liver cells of infected dogs was lower in those with clinical signs. Similar results were obtained in experimentally infected dogs, as these cytokines were expressed by the liver cells in addition to IFN-γ and iNOS (Maia and Campino, 2012). The authors suggested that a high level of parasitism might have been associated with the absence of TNF-α. In fact, the down regulation of IFN-γ, TNF-α, IL-10, IL-17 cytokines, and iNOS by hepatocytes with disease progression in naturally infected dogs was reported (Nascimento et al., 2015). According to Rodríguez-Cortés et al. (2016), the anti-inflammatory/regulatory immune response observed in the liver at 6 months after experimental infection might be responsible for the absence of clinical signs even in the presence of a high parasite load. The down regulation of IL-22 transcription in liver samples was significantly associated with clinical disease in experimental infected dogs (Hosein et al., 2015). Nascimento M. et al. (2013) observed that the impairment of the expression of chemokines ligands-CCL (CCL1, CCL17, CCL26) and chemokine receptors-CCR (CCR3, CCR4, CCR5, CCR6, and CCR8) by liver cells in animals with clinical signs might result in deficient leukocyte migration and concomitant hampering of the immune response and disease development. In the study recently performed by Rodrigues et al. (2017) it was shown that L. infantum interacts with Kupffer cells inducing an anergic state that promotes immune tolerance and parasite survival. The silence imposed by the parasite was reverted by the presence of meglumine antimoniate.

# Lymph Node

In lymph nodes, a balance between the expression of proinflammatory and anti-inflammatory cytokines expression seems to determine parasite load and clinical presentation (Alves et al., 2009; Barbosa et al., 2011; Maia and Campino, 2012). A high expression of IL-2, IL-12 (Barbosa et al., 2011), TNF-α, and IFNγ has been reported in the lymph nodes of naturally (Alves et al., 2009; de Vasconcelos et al., 2016) and experimentally (Maia and Campino, 2012) infected asymptomatic dogs. Contrarily, IL-10 and TGF-β expressions were significantly increased in dogs presenting clinical signs, suggesting a role of these cytokines in disease evolution (Alves et al., 2009; de Vasconcelos et al., 2016; Rodríguez-Cortés et al., 2016). These results diverge from the ones obtained by Barbosa et al. (2011) where the expression of IFN-γ and IL-2 was higher in the lymph nodes of symptomatic dogs. Disease progression has also be linked to down regulation of IL-17, IL-22, and forkhead box P3 protein (FoxP3) in the lymph nodes of experimental infected dogs (Hosein et al., 2015). An association of high levels of IL-6 in lymph nodes of sick dogs with disorganization of the corticomedullar region, suggest this cytokine as good marker of active disease (de Vasconcelos et al., 2016).

In popliteal lymph nodes from natural infected dogs, a significant increased number of cytotoxic T cells (CD8<sup>+</sup> T cells), together with decreased level of the cluster of differentiation (CD) CD21<sup>+</sup> B cells and upregulation of MHCII<sup>+</sup> molecules was observed (Giunchetti et al., 2008a). The levels of MHCII+ cells in lymph node lymphocytes were also increased after treatment with allopurinol and meglumine antimoniate (Alexandre-Pires et al., 2010). The CD4 T lymphocytes (CD4+) and CD8<sup>+</sup> T cells and Foxp3<sup>+</sup> regulatory T cells (Tregs) frequencies in mononuclear cells of cervical and mesenteric lymph nodes of naturally infected dogs have also been evaluated (Figueiredo et al., 2014). Infected dogs had a higher expression of CD4<sup>+</sup> and Foxp3<sup>+</sup> cells than that of controls, but no correlation of these molecules with parasite load was found. The expression of Foxp3<sup>+</sup> and CD4<sup>+</sup> T cells was significantly higher in mesenteric lymph nodes of both asymptomatic and symptomatic dogs, respectively. Alexandre-Pires et al. (2010) also observed an expansion of CD4<sup>+</sup> T cells subpopulations in popliteal and retropharyngeal lymph nodes of both symptomatic and meglumine antimoniate plus allopurinol treated dogs. The expansion of this cell subpopulation seems to be essential for the development of an efficient local immune response to Leishmania infection. In addition, the frequency of CD8<sup>+</sup> T cells was significantly lower in lymph nodes of treated dogs than in asymptomatic dogs, suggesting that these cells might contribute to the reduction of parasite load through cytotoxic mechanisms (Giunchetti et al., 2008a).

# Peripheral Blood

Despite not being the tissue of election for the multiplication and persistence of the parasite, the spectrum of cytokines and phenotypic cell profiles in peripheral blood have exhaustively been evaluated in naturally and experimentally infected dogs. However, results are often discrepant due to the use of different parameters to stage disease in dogs (i.e., based on the presence/absence of clinical signs and/or hematological and biochemical alterations) and with the different sensitivity and specificity of methodologies to determine the presence of specific antibodies and/or of the parasite (Maia and Campino, 2012).

IFN-γ expression/production by sera or by peripheral blood mononuclear cell (PBMC) lymphocytes non-stimulated or stimulated with soluble Leishmania antigen (SLA) from naturally and experimentally infected dogs was correlated with resistance to disease, asymptomatic status or mild disease (Santos-Gomes et al., 2002; Manna et al., 2006; Carrillo et al., 2007; Aslan et al., 2016; Abbehusen et al., 2017; Montserrrat-Sangrà et al., 2018)\$. The lack of its production was observed both in symptomatic infected dogs (Carrillo et al., 2007) and in sick dogs with strong humoral response, high parasitaemia, and severe clinical disease (Solano-Gallego et al., 2016; Martínez-Orellana et al., 2017). On the other hand, high levels of IFN-γ production and expression was detected in naturally and experimentally infected dogs classified as symptomatic, indicating that this cytokine does not seem to be a good marker of resistance as it was not enough to prevent disease (Travi et al., 2009; Cortese et al., 2013). However, tracking IFN-γ concentration could constitute an important prognostic tool for immune monitoring in CanL, as its concentration increases with longterm anti-Leishmania treatment with meglumine antimoniate and allopurinol (Martínez-Orellana et al., 2017). Contradictory results have also been observed regarding the detection of several other cytokines. For instance, in the work performed by Pinelli et al. (1999), IL-4 and IL-10 were only expressed by PBMC stimulated with concanavalin A of dogs with clinical signs, while in other studies (Manna et al., 2006; Carrillo et al., 2007) both cytokines were detected in asymptomatic and symptomatic dogs. In the same line of reason, increased IL-10 production by PBMC stimulated with SLA was pointed as predictive marker of canine infection evolution (Boggiatto et al., 2010) and with splenic parasite load (Aslan et al., 2016), however, in other studies this cytokine was not considered a marker of disease severity (Santos-Gomes et al., 2002; Solano-Gallego et al., 2016). Similarly, IL-6 and IL-18 cytokines seem to have no role on infection outcome (Pinelli et al., 1994; Manna et al., 2006; Carrillo et al., 2007; Aslan et al., 2016) or to be markers of active disease (Lima et al., 2007) or asymptomatic infection (Chamizo et al., 2005). IL-2 and TNF-α production by stimulated PBMC of symptomatic and control uninfected dogs was significantly lower when compared with those from infected dogs without clinical signs (Pinelli et al., 1994), while IL-12 stimulated the production of IFN-γ by PBMC from symptomatic dogs experimentally or naturally infected (Strauss-Ayali et al., 2005). IL-2 levels were negatively correlated with splenic parasite loads in experimentally infected dogs, while no correlation between IL-12 and the number of parasites in the spleen was observed (Aslan et al., 2016). Due to the role of Tregs in the suppression of host immunity against Leishmania, these cells have also been evaluated in the peripheral blood of dogs naturally infected (Cortese et al., 2013). Results revealed a reduced percentage of Tregs CD4+, CD3+and Foxp3<sup>+</sup> subsets on both asymptomatic and symptomatic infected dogs in comparison with non-infected controls.

Analyses of circulating leukocyte subpopulations pointed out the involvement of CD8<sup>+</sup> lymphocytes in resistance to CanL (Pinelli et al., 1995; Reis et al., 2006b; Coura-Vital et al., 2011; Cortese et al., 2013) as increased levels of these cells were found in dogs with low parasitism. Further studies revealed an increase of CD3<sup>+</sup> lymphocytes in infected dogs (Miranda et al., 2007), of CD5<sup>+</sup> in symptomatic dogs (Reis et al., 2006b) and of CD4<sup>+</sup> cells in dogs with a low parasite load (Reis et al., 2006b; Guerra et al., 2009). On the contrary, symptomatic dogs with high levels of parasitism in bone marrow have a decrease in CD21<sup>+</sup> B cells and CD14<sup>+</sup> monocytes and low levels of CD4<sup>+</sup> and CD8<sup>+</sup> T cells (Reis et al., 2006b). Low levels of circulating CD4<sup>+</sup> in naturally infected dogs have also been associated with a higher infectivity to phlebotomine sand flies (Guarga et al., 2000). CD4+/CD8<sup>+</sup> lymphocyte ratio has been evaluated with the rationale that development of clinical disease is accompanied by a reduction of CD4<sup>+</sup> T cells. In fact, there is a decrease of CD4<sup>+</sup> counts in the peripheral blood of sick animals, which tends to return to normal values after treatment (Moreno et al., 1999; Papadogiannakis et al., 2010). However, in some studies it was found a similar number of CD4<sup>+</sup> counts in healthy and in infected dogs with no correlation between the clinical status (Miranda et al., 2007) probably reflecting individual variability (Paltrinieri et al., 2016). Therefore, and according to Paltrinieri et al. (2016) "CD4+/CD8<sup>+</sup> ratio seems to be more suitable for monitoring the post-treatment follow-up rather than initial staging of clinical suspected dogs." Studies using the saponin enriched-Leishmune <sup>R</sup> vaccine as immunotherapy revealed a sustained or increased proportions of CD4<sup>+</sup> and CD21<sup>+</sup> B lymphocytes and an increase proportion of CD8<sup>+</sup> T cells in the peripheral blood of naturally and experimentally infected dogs (Borja-Cabrera et al., 2004, 2010; Santos et al., 2007). According to the studies performed by Araújo et al. (2008, 2009) Leishmune <sup>R</sup> promotes an increase of CD8<sup>+</sup> T-cells activation, and induces a selective pro-inflammatory pattern with the production of IFN-γ and NO by peripheral blood lymphocytes and monocytes, respectively.

The analysis of the expression of the MHCII<sup>+</sup> in peripheral blood lymphocytes has also lead to divergent results: upregulation of MHCII<sup>+</sup> expression was observed in asymptomatic (Reis et al., 2006b) and symptomatic dogs (Alexandre-Pires et al., 2010). On the other hand, the expression of this molecule was decreased in sick dogs with high parasite density in bone marrow (Reis et al., 2006b). In a cross-sectional exploratory study disease severity was characterized by an increase of the chemokine (C-X-C motif) ligand 1 (CXCL1) and CCL2 serum levels (Solcà et al., 2016). As the recruitment of neutrophils and monocytes is made by CXCL1 and CCL2 respectively, the increase production of these chemokines was probably related with enhanced parasite density (Solcà et al., 2016).

# Skin

After the inoculation of Leishmania parasites into the skin via phlebotomine sand fly female bite, several cells are involved in the activation of the innate immune system, with dendritic cells and macrophages playing a main role (Papadogiannakis and Koutinas, 2015). A Th2-biased immune response, with an increased expression of IL-4 (Brachelente et al., 2005), IL-10, and TGF-β (Rodríguez-Cortés et al., 2016) or overproduction of IL-4, IL-13, and TNF-α (Papadogiannakis and Koutinas, 2015) was associated with a high parasite burden and clinical disease. An increased parasite load was also associated with up upregulation of IL-10 and TNF-α in the skin of infected dogs (Pereira-Fonseca et al., 2017). On the other hand, a mixed Th1/Th2 cytokine profile and low levels of GATA-3 and Foxp3<sup>+</sup> transcription factors in asymptomatic dogs indicates that in the absence of clinical signs or in cases with low number of parasites in the skin, a mixed inflammatory/regulatory immune response may be crucial (Menezes-Souza et al., 2011).

Menezes-Souza et al. (2012) reported a positive association between the CCL2, CCL4, CCL5, CCL21, and CXCL8 expression by the skin cells of naturally infected dogs with high cutaneous parasitism, while CCL24 expression was negatively correlated with parasite load. The cellular immunophenotyping and skin parasitism in symptomatic dogs has also been investigated (Fondevila et al., 1997; Papadogiannakis et al., 2005). An effective local immune response is associated to the activation of epidermal Langerhans cells, to the infiltration of dermis by CD8<sup>+</sup> cells, to the upregulation of MHCII<sup>+</sup> on keratinocytes, and by the absence or presence of few parasites. In contrast, the immune response in skin of dogs with clinical disease is characterized by a high number of plasma cells outnumbering T lymphocytes in the dermal infiltrate and by a high parasite load (Papadogiannakis et al., 2005).

# Spleen

Splenic architecture disruption due to CanL is characterized by the disorganization of lymphoid tissue with eventual atrophy and loss of leukocyte diversity (Sanchez et al., 2004). As with other tissues, contradictory results regarding the expression or production of cytokines and chemokines by spleen cells have been reported. On one hand, a positive correlation between IL-10 expression by spleen cells and increased parasitism and progression of the disease was found (Lage et al., 2007; Nascimento P. et al., 2013). On the other hand, no significant changes in the expression/production of this cytokine were reported, regardless the parasite load or clinical status of the dogs (Correa et al., 2007; Strauss-Ayali et al., 2007; Silva et al., 2014; Rodríguez-Cortés et al., 2016). Increased expression of TNF-α and IFN-γ by spleen cells was associated with reduced Leishmania burden (Nascimento P. et al., 2013). However, correlation between parasite load and the production of TNFα by spleen extracts of dogs naturally infected has been documented, and according to the authors, it may represent an important marker for infection evolution (Michelin et al., 2011). Further, a worst disease prognostic was reported in dogs with a high expression or production of IFN-γ and splenic parasitism (Lage et al., 2007). In experimentally infected dogs, this cytokine was only expressed by tissues with high parasitic load (Maia and Campino, 2012), reinforcing its relation with the increase of parasitism (Lage et al., 2007). Persistence of parasites in the spleen has also been associated with early elevation of IL-4 expression by spleen cells in the presence of high levels of IFN-γ (Strauss-Ayali et al., 2007). Further, disease progression has significantly been associated to down regulation of IL-22 in the spleen of experimental infected dogs (Hosein et al., 2015), and to the down regulation of IFN-γ, IL-10, IL-17A, and iNOS in naturally infected dogs (Nascimento et al., 2015). The production of TGF-β by Tregs in the spleen was also evaluated but no correlation was found between the percentage of spleen Tregs producing this cytokine and the parasite load (Silva et al., 2014). An impairment of both pro-inflammatory and antiinflammatory cytokines induced by splenic architecture breakage due to parasite presence has recently been reported (Cavalcanti et al., 2015).

An increase of the expression levels of IP-10, MCP-1, MIP1-α, and RANTES by spleen cells was observed during the follow-up of dogs experimentally infected (Strauss-Ayali et al., 2007). The increase of the chemokines was suggested to be associated with an accumulation of monocytes attracted by MCP-1 and MIP1-α, and with CD4<sup>+</sup> Th1 and CD8<sup>+</sup> cells recruited by IP-10 (Strauss-Ayali et al., 2007). The levels of these chemokines as well of IFN-γ significantly decreased after dogs were treated with allopurinol (Strauss-Ayali et al., 2007). In infected dogs the splenic expression levels of CCL1, CCL3, CCL17, CCL20, CCL26, CXCLl9, CCR3, CCR34, CCR36, and CCR38 were found to be reduced relatively to the expression levels in uninfected animals (Nascimento M. et al., 2013). Animals with disorganized lymphoid tissue presented lower CXCL13 expression and compared to those with organized lymphoid tissue, and the expression of this chemokine was associated with a higher frequency of severe disease (Silva et al., 2012). On the other hand, an increased expression of CCL2, CCL5, and CXCL10 by spleen cells was reported in symptomatic dogs in comparison with infected animals not showing clinical signs (Nascimento M. et al., 2013), and of CXCL12 in diseased animals and in those with disruption of the white pulp (Silva-O'Hare et al., 2016). All these data reinforces that the impairment of cell migration and the induction of long-lived plasma cells favors parasite replication and progression of the disease.

CCL21 and CCL19 chemokines are expressed by endothelial venules in lymphoid cells and organs (Ato et al., 2002). The binding of these chemokines to the CCR7 receptor of mature dendritic cells (DC) allows the migration of cells from the marginal zone to the peri-arteriolar region of the spleen. In mice chronically infected with Leishmania (L.) donovani it was shown that cellular immunosuppression is mediated by failure of DC migration due to the decreased chemokine secretion by endothelium and to the reduced DCs CCR7 expression (Ato et al., 2002). The immunotherapy with DC overexpressing CCR7 efficiently controlled infection in the spleen of infected mice (Ato et al., 2002). Therefore, molecules that can prevent the inhibition of this receptor would be of great interest for the control of leishmaniosis. L. donovani nucleoside hydrolase NH36 and its C-terminal domain, the F3 peptide have recently proven to be prominent antigens in the generation of preventive immunity to visceral leishmaniosis (Nico et al., 2018). Both antigens were able to control parasite loads in the spleen and liver of mice vaccinated with both antigens and then challenged with L. (L.) infantum chagasi. The imunotherapy with F3 antigens prevented the migrating defect of DCs by restoring the expression of CCR7 receptors (Nico et al., 2018). The ability of NH36, the main antigen of Leishmune <sup>R</sup> , to prevent or control leishmaniosis in dogs and mice by restoring a Th1 response was already be proven (Aguilar-Be et al., 2005; Borja-Cabrera et al., 2012).

# Other Tissues

Figueiredo et al. (2014) evaluated CD4<sup>+</sup> and CD8<sup>+</sup> T cells and Foxp3<sup>+</sup> Tregs frequencies and cytokine expression in the mononuclear cells of the jejunum and colon of dogs naturally infected with L. (L.) infantum chagasi. Frequencies and expression of IL-10, IFN-γ, TGF-β, TNF-α, Foxp3+, CD4+, and CD8<sup>+</sup> were higher in jejunum, while IL-4 expression was significantly higher in the colon. A positive correlation between CD4<sup>+</sup> in the colon and between CD4<sup>+</sup> Foxp3<sup>+</sup> Tregs in jejunum and parasite load was found. Infected animals had reduced CD8<sup>+</sup> expression in both intestinal compartments compared to controls.

In L. (L.) infantum chagasi infected dogs CD3<sup>+</sup> T lymphocytes were found to be the major components of the inflammatory infiltrate at the choroid plexus and in the brain; according to the authors during the advanced stages of leishmaniosis leukocytes might participate in the pathogenesis of neurological disorders (Melo et al., 2009). In fact, glial reactivity in dogs with leishmaniosis was correlated with T lymphocyte infiltration of the brain (Melo and Machado, 2011). Further, up-regulation of CCL3, CCL4, and CCL5, coherent with T lymphocyte accumulation, was observed in the brain of infected dogs (Melo et al., 2015). Apparently, the presence of parasite's DNA rather itself is enough to promote the development of a local immune response as no correlation between the downregulation of expression of IL-10, IL-12p40, and TGF-β by brain cells or the upregulation of IL-1-β, IFN-γ, and TNF-α and parasitism was found (Melo et al., 2013). Nevertheless, Grano et al. (2018) have recently found in the brain of infected dogs a moderate negative correlation between the levels of IL-1β and TNF-α and the number of parasites (Grano et al., 2018). Evidence of blood-cerebrospinal fluid barrier breakdown with the passage of T lymphocytes from the blood to the brain during CanL has been reported and related with the origin and progression of the neurological disorders (Grano et al., 2016).

Despite the recent advances made on the biomarkers related to the pathogenesis of Leishmania infection in the different organs and tissues, due to the invasive sampling and to the limited access to the tools to evaluate the biological markers, most of them cannot be used in a laboratory setting. Nonetheless, and taking into account the results described above, the inclusion in the laboratory diagnosis of the evaluation of the cytokines and phenotypic cell profiles of non-invasive samples, such as peripheral blood or lymph node aspirates, would probably represent a step forward for the prognosis and for monitoring the response to treatment.

# SEROLOGICAL BIOMARKERS TO *Leishmania infantum* INFECTION AND DISEASE

CanL is often associated with a specific non-protective humoral response (Alvar et al., 2004; Maia and Campino, 2008; Miró et al., 2008; Solano-Gallego et al., 2009; Paltrinieri et al., 2016). However, the presence of antibodies to Leishmania alone is not conclusive of Leishmania infection, as it may simply reflect exposure to the parasite (Campino and Maia, 2018). Further, the production of specific antibodies is low on initial and late phase of infection and in infected dogs without clinical signs. Conversely, uncontrolled parasite dissemination is associated with gradually increase of antibody titers over time (Oliva et al., 2006), which will be high when the disease is evident (Campino, 2002; Solano-Gallego et al., 2009; Paltrinieri et al., 2016). While a direct relationship between tissue parasite density, clinical status and antibody titres is proven (Reis et al., 2006c; Dos-Santos et al., 2008; de Almeida Leal et al., 2014; Proverbio et al., 2014) lowto-medium antibody titres may also be detected in symptomatic dogs (Solano-Gallego et al., 2009, 2011, 2017; Paltrinieri et al., 2016).

Anti-Leishmania-specific canine IgG subclasses have extensively been investigated (Pinelli et al., 1994; Leandro et al., 2001; Iniesta et al., 2002, 2005; Cardoso et al., 2007) as an attempt to correlate the type of Th response, the subclass level and the clinical outcome of infection (Baneth et al., 2008; Maia and Campino, 2008). The majority of studies have focused on IgG1 and IgG2 responses and tried to link them with Th2-like susceptibility and Th1-like protective responses, respectively (Desplazes et al., 1995; Nieto et al., 1999; Iniesta et al., 2005; Cardoso et al., 2007; Rodríguez-Cortés et al., 2007a). A significant correlation between IgG, IgA, IgM (Rodríguez-Cortés et al., 2007a,b), and IgE (Iniesta et al., 2005; Reis et al., 2006c) and clinical signs have been found. After the launch of Leishmune <sup>R</sup> vaccine, which induce a strong humoral immune response, de Oliveira Mendes et al. (2003) evaluated if the production of IgG1 and IgG2 subclasses was able to distinguish the vaccinated dogs from those naturally infected. An association of IgG1 response to natural infection and IgG2 to a humoral response subsequent to the Leishmnune <sup>R</sup> vaccination was found (de Oliveira Mendes et al., 2003). In addition, IgG1/IgG2 ≥ 1 was associated to the sera of infected animals that evolve toward the disease while IgG1/IgG2 ≤1 was associated to the sera response of vaccinated dogs. Due to the low specificity of the polyclonal antisera commercially available to detect IgG subclasses results were often contradictory (Day, 2007). Thus, monoclonal antibodies to canine IgGs have been tested during natural and experimental infection. Nevertheless, a stable increase in the production of the four subclasses was observed with no indication of a practical use (Quinnell et al., 2003a; Strauss-Ayali et al., 2007).

Various quantitative serological methods such as the indirect immunofluorescence assay (IFAT) and enzymelinked immunosorbent assay (ELISA) or the qualitative rapid immunochromatographic tests (ICT) are available for CanL diagnosis (Maia and Campino, 2008; Paltrinieri et al., 2016). Due to its high sensitivity and specificity (near 100% for both) IFAT is considered the reference method for anti-Leishmania serology in dogs (Gradoni and Gramiccia, 2008; EFSA AHAW Panel (EFSA Panel on Animal Health and Welfare), 2015). Sensitivity and specificity of ELISA is also quite high, especially when recombinant proteins are used as antigen (Paltrinieri et al., 2016). ICT are very attractive due to their single-test format, ease of use and quick response time (Maia and Campino, 2008). However, they only provide a qualitative result (i.e., presence/absence of specific reactive spots/bands) and their sensitivity is variable (Maia and Campino, 2008; EFSA AHAW Panel (EFSA Panel on Animal Health and Welfare), 2015; Paltrinieri et al., 2016). Therefore, in case of a positive result, a quantitative serology to obtain a titer for follow-up monitoring should be performed. In addition, and given the moderate sensitivity of most of the ICT, a negative result obtained with these devices in a clinically suspect dog should be followed by a quantitative test. According to Solano-Gallego et al. (2009, 2017) and Paltrinieri et al. (2016), "quantitative results provide by IFAT and ELISA reflect the final antibody titer (the last 2-fold serial dilution of sample providing a positive result). For ELISA, optical density values converted based on a reference titered sample should also be used. A titer is considered high if it is 4-fold higher than the threshold value of the laboratory (Solano-Gallego et al., 2009, 2017; Paltrinieri et al., 2016). Similarly, 4-fold titer variations in sequential samples of the same dog should be expected with seroconversions." In it important clinicians to be aware that sequential samples should always be analyzed by the same method and in the same laboratory (Paltrinieri et al., 2016). Furthermore, it should be referred that serological titers not always correlate with the severity of the clinical signs (Ferrer et al., 1995; Manna et al., 2015) although asymptomatic dogs usually have low titers (Paltrinieri et al., 2010).

According to Paltrinieri et al. (2016), "in case of successful treatment, a decrease in antibody titers may be expected over time reaching values consistent with simple exposure (<4-fold the threshold value of the laboratory)," as in dogs living in endemic areas a complete disappearance of anti-leishmanial antibodies is unlikely. However, serology is not a reliable parameter to monitor treatment efficacy in the short-term (Ferrer et al., 1995; Miró et al., 2009; Torres et al., 2011; Manna et al., 2015). Albeit a significant reduction in titers can be detected 1 month post-treatment, a distinguishable decrease of titers is normally observed 6 months after initiation of therapy (Torres et al., 2011; Paltrinieri et al., 2016). In clinical relapses, a rise in antibody titers is observed (Manna et al., 2015).

As mentioned before, the presence of low levels of specific antibodies does not necessarily indicate an active infection. Further, the clinical presentation might be due to other pathologies. In these cases leishmaniosis diagnosis needs to be confirmed by the presence of the parasite or its components (direct methods) such as cytology, histology, immunohistochemistry, PCR or real-time PCR (Maia and Campino, 2008; Solano-Gallego et al., 2009; Paltrinieri et al., 2016; Campino and Maia, 2018). In addition, the vaccines available to prevent CanL have puzzled serological diagnosis, as most of the widely used tests are not able to discriminate between naturally infected and vaccinated dogs (Moreno et al., 2014; Paltrinieri et al., 2016; Solano-Gallego et al., 2017). Serological cross-reactivity with antibodies against other Leishmania species and pathogens such as Trypanosoma cruzi, Ehrlichia canis, and Leptospira interrogans (Ferreira et al., 2007; Porrozzi et al., 2007) is possible with some tests, especially those based on whole parasite antigens (Maia and Campino, 2008; EFSA AHAW Panel (EFSA Panel on Animal Health and Welfare), 2015).

# CELLULAR BIOMARKERS TO *Leishmania infantum* INFECTION AND DISEASE

It is well established that susceptibility or resistance to Leishmania infection is mediated by cellular immune responses and several tools have been used to evaluate their role on the immunology and immunopathology of CanL (Maia and Campino, 2008; Paltrinieri et al., 2010; Reis et al., 2010).

Parasite-specific cellular immunity can be assessed by the Montenegro or leishmanin skin test (LST), which induces a delayed-type hypersensitivity response in dogs. Leishmania antigen, which consists of a suspension of inactivated parasites, is intradermal inoculated. A positive reading consists of an induration of over 5 mm in diameter obtained 48–72 h after inoculation. LST is negative during active disease, while during subclinical infection, early stage of clinical disease or after successful treatment is positive (Pinelli et al., 1994; Cardoso et al., 1998; Solano-Gallego et al., 2001; Fernández-Bellón et al., 2005). A strong and long lasting cellular immune response against the parasite has also been observed in Leishmune <sup>R</sup> vaccinated dogs, as a positive DTH response was present in immunized animals up to 41 months after vaccination (da Silva et al., 2000; Borja-Cabrera et al., 2002, 2008).

Ex vivo tests to assess Leishmania-specific cell-mediated immunity include lymphocyte proliferation assay and assays measuring IFN-γ in circulating lymphocytes (such as IFN-γ cytophatic effect inhibition bioassay and IFN-γ release assay). Lymphocyte proliferation assay consists on the stimulation of PBMC with soluble Leishmania antigen (SLA) and a mitogen with non-stimulated cells representing the negative control. Cell proliferation is expressed as a stimulation index (SI), which is obtained by the ratios of stimulated cells to non-stimulated cells). SI ≥ 2 are considered positive (Cabral et al., 1992, 1998; Pinelli et al., 1994; Leandro et al., 2001; Quinnell et al., 2001a; Fernández-Pérez et al., 2003; Santos-Gomes et al., 2003; Fernández-Bellón et al., 2005). Asymptomatic and resistant dogs present a strong proliferative response to leishmanial antigens while susceptible and diseased animals fail to respond respectively, to SLA and to mitogen (Abranches et al., 1991; Pinelli et al., 1994; Rhalem et al., 1999; Quinnell et al., 2001a; Strauss-Ayali et al., 2005). Lymphoproliferation is restored after successful leishmanicidal treatment (Bourdoiseau et al., 1997; Moreno et al., 1999; Rhalem et al., 1999; Fernández-Pérez et al., 2003).

Assays measuring IFN-γ allow quantifying the level of stimulation of a specific Th1-polarity immune memory response to Leishmania antigen (Fernández-Bellón et al., 2005; Rodríguez-Cortés et al., 2007a; Moreno et al., 2014; Zribi et al., 2017). The IFN-γ cytophatic effect inhibition bioassay detects the production of IFN-γ by circulating lymphocytes in cultured supernatants incubated in the presence/absence of SLA followed by incubation with canine kidney cells. IFN-γ production is expressed as the ratio of the reciprocal of the maximum dilution that protects 50% of the cell monolayer against vesicular stomatitis virus of stimulated vs. non-stimulated cells; values ≥ 2 are considered positive. The evaluation of its usefulness is very limited (Fernández-Bellón et al., 2005; Rodríguez-Cortés et al., 2007a) due to its cumbersome nature and to the use of a virus included in list A of the OIE.

The IFN-γ release assay (IGRA) allows rapid screening of IFN-γ-secretion in whole blood challenged with SLA. It seems to be a useful tool to assess exposure to Leishmania as positive IGRA responses were seen in infected dogs without or with mild clinical signs and in dogs without or with low parasite load, whereas negative IGRAs were identified in dogs with the highest parasitism (Zribi et al., 2017). The lack of IFN-γ production in dogs with severe clinical disease and high number of parasites on blood has also been reported (Solano-Gallego et al., 2016; Martínez-Orellana et al., 2017).

Canine macrophage leishmanicidal assay is also used to evaluate cell-mediated immune response via nitric oxide (NO) analyses. NO production by macrophages is the principal effector molecule mediating intracellular killing of Leishmania amastigotes by apoptosis. Cytokines such as IFN-γ and TNF-α secreted by activated T cells have been found to induce nitric oxide synthase (iNOS) and NO production facilitating parasite control (Green et al., 1990; Wanasen and Soong, 2008). The role of NO against CanL has been demonstrated by inducing antileishmanial activity in macrophages via the L-arginine NO pathway (Vouldoukis et al., 1996). After successful antimonial therapy canine macrophages regained the ability to control the parasites via increased NO production (Vouldoukis et al., 1996). Further, canine macrophages activated by a supernatant contained IFN-γ, TNF-α, and IL-2 were able to increase NO production and anti-leishmanial activity (Pinelli et al., 2000). Similarly, canine macrophages infected in vitro by L. infantum were able to produce NO after stimulation with cytokineenriched PBMC supernatants (Panaro et al., 1998) and after stimulation with IFN-γ and bacterial lipopolysaccharide were also able to express iNOS (Sisto et al., 2001). A correlation between high iNOS expression by Leishmania infected canine macrophages and a low intracellular amastigote burden has also been reported (Zafra et al., 2008). The ability of canine macrophages to kill parasites through NO production as a measurement of long-term protection of dogs against Leishmania infection and disease has also been evaluated. Panaro et al. (2008) observed that in the first months after Leishmania diagnosis, the levels of NO produced by Leishmania-infected macrophages were higher in symptomatic dogs than in those without clinical signs. The role played by NO in leishmanicidal activity has also been demonstrated in the context of vaccination studies showing that immunized dogs develop long-lasting Th1cell-mediated immune

# REFERENCES

Abbehusen, M. M. C., Almeida, V. D. A., Solcà, M. D. S., Pereira, L. D. S., Costa, D. J., Gil-Santana, L., et al. (2017). Clinical and immunopathological findings during long term follow-up in Leishmania infantum experimentally infected dogs. Sci. Rep. 7:15914. doi: 10.1038/s41598-017-15651-8

responses against L. infantum or L. (L.) infantum chagasi. NO enhanced the anti-leishmanial activity of macrophages alone or co-cultured with IFN-γ producing autologous lymphocytes (Panaro et al., 2001; Lemesre et al., 2005; Rodrigues et al., 2007; Moreno et al., 2012, 2014), and has also been shown to mediate apoptosis of intracellular amastigotes (Holzmuller et al., 2005).

A practical and standardized assay to evaluate cellular immunity to Leishmania infection in clinical settings should be of practical use to help monitoring CanL and treatment outcome (Maia and Campino, 2008).

# CONCLUSION

The identification of biological parameters that can be indicators of pathological processes related to L. infantum infection or disease, or a response to Leishmania treatment or vaccination would represent a major progress in the control of canine leishmaniosis. Data gathered from several studies have identified potential biomarkers but none of them provided a strong evidence of their practical applicability on diagnosis. In addition, no single marker seems sufficient to be a direct correlate of resistance or susceptibility to disease, as a complex network of regulatory and counter-regulatory interactions involving several cytokines, chemokines and cell populations are involved in mounting an effective immune response. Therefore, multiple laboratorial, immunological and parasite-specific biomarkers should be considered together to obtain a general view of L. infantum infection and disease outcome. The visceral tropism of the parasites makes sampling challenging, as well as owner compliance, especially when invasive procedures need to be repeated. Therefore, the identification of biomarkers obtained from non-invasive samples is warranted. Finally, the development of vaccines to prevent CanL represents an important step forward to control the disease but it has complicated its diagnosis, so, biomarkers able to discriminate naturally infected from vaccinated dogs are urgent.

# AUTHOR CONTRIBUTIONS

CM writing and editing original draft. LC editing and review original draft.

# FUNDING

Fundação para a Ciência e a Tecnologia for funds to GHTM– UID/Multi/04413/2013. CM has the support of the Portuguese Ministry of Education and Science (via Fundação para a Ciência e a Tecnologia), through an Investigator Starting Grant (IF/01302/2015).


efficacy of a prophylactic Leishmania donovani DNA vaccine against visceral and cutaneous murine leishmaniasis. Infect. Immun. 73, 812–819. doi: 10.1128/IAI.73.2.812-819.2005


between sand flies, hosts, and Leishmania. PLoS Negl. Trop. Dis. 11:e0005600. doi: 10.1371/journal.pntd.0005600


and prevention of canine leishmaniosis. Vet. Parasitol. 165, 1–18. doi: 10.1016/j.vetpar.2009.05.022


secretion by canine macrophages resistant or susceptible to intracellular survival of Leishmania infantum. Vet. Immunol. Immunopathol. 163, 67–76. doi: 10.1016/j.vetimm.2014.11.010


**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 Maia and Campino. 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.

# In Search of Biomarkers for Pathogenesis and Control of Leishmaniasis by Global Analyses of Leishmania-Infected Macrophages

Patricia Sampaio Tavares Veras 1,2 \*, Pablo Ivan Pereira Ramos <sup>3</sup> and Juliana Perrone Bezerra de Menezes <sup>1</sup>

<sup>1</sup> Laboratory of Host-Parasite Interaction and Epidemiology, Gonçalo Moniz Institute, Fiocruz-Bahia, Salvador, Brazil, <sup>2</sup> National Institute of Tropical Disease, Brasilia, Brazil, <sup>3</sup> Center for Data and Knowledge Integration for Health, Gonçalo Moniz Institute, Fiocruz-Bahia, Salvador, Brazil

### Edited by:

Martin Olivier, McGill University, Canada

### Reviewed by:

Patricia Talamás-Rohana, Centro de Investigación y de Estudios Avanzados (CINVESTAV), Mexico Maritza Jaramillo, Institut National de la Recherche Scientifique (INRS), Canada

### \*Correspondence:

Patricia Sampaio Tavares Veras pstveras@gmail.com; patricia.veras@fiocruz.br

### Specialty section:

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 18 June 2018 Accepted: 27 August 2018 Published: 19 September 2018

### Citation:

Veras PST, Ramos PIP and de Menezes JPB (2018) In Search of Biomarkers for Pathogenesis and Control of Leishmaniasis by Global Analyses of Leishmania-Infected Macrophages. Front. Cell. Infect. Microbiol. 8:326. doi: 10.3389/fcimb.2018.00326 Leishmaniasis is a vector-borne, neglected tropical disease with a worldwide distribution that can present in a variety of clinical forms, depending on the parasite species and host genetic background. The pathogenesis of this disease remains far from being elucidated because the involvement of a complex immune response orchestrated by host cells significantly affects the clinical outcome. Among these cells, macrophages are the main host cells, produce cytokines and chemokines, thereby triggering events that contribute to the mediation of the host immune response and, subsequently, to the establishment of infection or, alternatively, disease control. There has been relatively limited commercial interest in developing new pharmaceutical compounds to treat leishmaniasis. Moreover, advances in the understanding of the underlying biology of Leishmania spp. have not translated into the development of effective new chemotherapeutic compounds. As a result, biomarkers as surrogate disease endpoints present several potential advantages to be used in the identification of targets capable of facilitating therapeutic interventions considered to ameliorate disease outcome. More recently, large-scale genomic and proteomic analyses have allowed the identification and characterization of the pathways involved in the infection process in both parasites and the host, and these analyses have been shown to be more effective than studying individual molecules to elucidate disease pathogenesis. RNA-seq and proteomics are large-scale approaches that characterize genes or proteins in a given cell line, tissue, or organism to provide a global and more integrated view of the myriad biological processes that occur within a cell than focusing on an individual gene or protein. Bioinformatics provides us with the means to computationally analyze and integrate the large volumes of data generated by high-throughput sequencing approaches. The integration of genomic expression and proteomic data offers a rich multi-dimensional analysis, despite the inherent technical and statistical challenges. We propose that these types of global analyses facilitate the identification, among a large number of genes and proteins, those that hold potential as biomarkers. The present review focuses on large-scale studies that have identified and evaluated relevant biomarkers in macrophages in response to Leishmania infection.

Keywords: biomarkers, leishmaniasis, macrophages, RNA-seq, proteomics, global analysis, functional enrichment analysis

# INTRODUCTION

Leishmaniasis is a neglected parasitic disease that is distributed worldwide and is often associated with poverty. Most cases of this disease arise in developing countries and result in 20,000–40,000 deaths per year. Leishmania, the causative agent, is transmitted to vertebrate hosts, including humans, by a bite from the sand fly during blood-feeding. Its pathogenesis involves the stimulation of different types of host immune responses that result in distinct clinical outcomes (Scorza et al., 2017), including cutaneous, mucocutaneous, and visceral manifestations, depending on the parasite species and host genetic background (Bañuls et al., 2011). Localized or mucocutaneous forms of tegumentary leishmaniasis, e.g., those caused by Leishmania braziliensis, induce activation of the host immune response, resulting in an immune-mediated pathology that manifests as localized ulcerations in human skin or disfigurement involving the nasal and oropharyngeal mucosa (Gupta et al., 2013). By contrast, the visceral form of this disease arises from parasites of the L. donovani complex and may result in severe systemic manifestations and high morbidity and mortality due to the inhibition of host inflammation and immunity (Das et al., 2014).

Neutrophils, dendritic cells, and macrophages, the main host cells that harbor parasites, are immune cells that are recruited to the infection site, where they recognize parasites that, once internalized, multiply within their phagolysosomes. In addition, these cells produce cytokines and chemokines that contribute to lymphocyte recruitment, which is critical to the disease outcome (Liu and Uzonna, 2012).

Measures designed to eradicate leishmaniasis necessitate a combination of intervention strategies, including early diagnosis and treatment. In visceral leishmaniasis, diagnostic procedures both evaluate clinical signs and employ parasitological or serological testing that is potentially capable of discriminating active visceral leishmaniasis from its asymptomatic form. By contrast, clinical evaluations are of greater importance in cutaneous and mucocutaneous leishmaniasis because serological testing is inadequate. Since leishmaniasis treatment must be affordable to ensure access by affected impoverished populations, the development of new compounds to treat leishmaniasis has attracted limited commercial interest. In addition, studies unveiling several aspects of the host response to Leishmania infection have not resulted in the discovery of effective new therapeutic interventions. Although some new alternative antileishmanials have recently emerged, none are considered ideal due to their high toxicity, prolonged treatment duration, and severe adverse reactions, which can lead to treatment abandonment and frequent cases of relapse (Aronson et al., 2017).

Biomarkers as surrogate endpoints have been recommended for use in clinical trials to aid in the early diagnosis of leishmaniasis since primary clinical markers are sparse and are only applicable after an extensive follow-up period. Biomarkers offer another advantage in that they allow measurements to be obtained more rapidly and in a less invasive context than do conventional clinical or parasitological evaluations. They could also facilitate the design of smaller, more efficient clinical studies that may lead to expedited regulatory evaluation and treatment approval. A recent elegant systematic review identified different types of direct and indirect biomarkers that were shown to be involved in Leishmania infection and disease outcome. Among the 170 studies evaluated, 53 potential pharmacodynamic biomarkers were identified, including direct, i.e., of parasite origin, and indirect, i.e., from host cells, markers of cutaneous, post-kala-azar dermal leishmaniasis, and visceral leishmaniasis (Kip et al., 2015).

The identification of a set of soluble biomarkers in host tissue has been exploited using sera of individuals with visceral leishmaniasis (Solcà et al., 2016; Araújo-Santos et al., 2017). A recent study screened a variety of soluble molecules and identified a set of inflammatory biomarkers that grouped together under a hierarchical cluster analysis (Araújo-Santos et al., 2017). A significant increase in the levels of the following inflammatory mediators was observed: resolvin D1 (RvD1), leukotriene B4 (LTB4), prostaglandin F2α (PGF2α), IL-1β, IL-6, IL-8, IL-10, IL-12p70, and TNF-α, in contrast to a decrease in TGF-β1 in the serum of patients with visceral leishmaniasis compared with an uninfected endemic control group. After 30 days of therapy, the authors observed that individuals clustered together in terms of decreases in the levels of these inflammatory molecules, distinct from the individuals with active infection, thus reinforcing the idea that this set of soluble molecules might function as biomarkers for the host response to therapy. These authors further remarked that the modulation observed in the concentrations of these markers provides evidence that "an inflammatory imbalance hallmarks active visceral leishmaniasis disease," which more importantly can greatly aid in the design of new interventions (Araújo-Santos et al., 2017). Another recent study focused on the identification of circulating biomarkers of "inflammation, immune activation, oxidative stress, and anti-sand fly saliva IgG concentrations" in canine sera to characterize biosignatures associated with the severity of visceral leishmaniasis in dogs presenting a variety of clinical manifestations (Solcà et al., 2016). These authors discovered unique biosignatures according to the frequency and intensity of clinical signs. A characteristic signature was found to be associated with animals presenting severe visceral leishmaniasis, as evidenced by a gradual decrease in LTB<sup>4</sup> and PGE<sup>2</sup> concomitant with a gradual increase in CXCL1 and CCL2. Furthermore, the quantification of three mediators, LTB4, PGE, and CXCL1, was shown to correlate with different clinical scores. This study clarified that visceral leishmaniasis severity in dogs can be associated with inflammatory profiles, which are distinguishable according to clinical presentation, via the expression of circulating eicosanoids and chemokines.

Advances in global genomic and proteomic analysis techniques have enabled the identification and characterization of pathways involved in the infection process in both parasites and the host. These approaches have been lauded due to their greater effectiveness than focusing exclusively on individual molecules, which rarely lend insight into disease pathogenesis. RNA-seq and proteomics, both large-scale techniques designed to characterize genes or proteins in a given cell line, tissue, or organism, offer the advantage of a more global and integrated view of the myriad biological processes that occur within cells (Wang et al., 2009; Veras and Bezerra De Menezes, 2016).

The analytical tools that are available for studying complex data include functional enrichment analysis (e.g., the widely adopted GSEA), in which a set of transcriptionally disturbed genes belonging to a common group of canonical pathways or biological processes reflect alterations in the pathways themselves. Gene co-expression networks can also be used to infer which genes are related to an infectious process. These networks offer the distinct advantage of enabling the discovery of previously unknown relationships by building on the notion of "guilty by association" (Huang da et al., 2009a; Kuleshov et al., 2016). A significant advantage of integrating genomic and proteomic information is that these data can be used in rich multi-dimensional analyses that allow identification from an enormous pool of expressed genes and proteins those that offer promise for use as biomarkers of different endpoints in leishmaniasis, such as disease diagnosis and treatment, in addition to markers for disease establishment and progression.

It has been clearly shown that macrophages are not only the major cells that harbor Leishmania parasites, but are also those that modulate host immune response by producing cytokines and presenting parasite antigens to T cells (Podinovskaia and Descoteaux, 2015). In addition, it has been shown that initial interactions between Leishmania parasites and macrophages contribute to the outcome of infection (Laskay et al., 1995; Scharton-Kersten and Scott, 1995). Thus, the present review focuses on recent large-scale studies detailing the host-related genes analyzed by RNA-seq and the proteins identified by proteomics, as well as describes the types of bioinformatics analyses used to integrate the large volumes of data generated by these high-throughput sequencing techniques. Due to the importance of these cells in the host response to Leishmania infection, we endeavor to review those genes and proteins expressed by macrophages in response to infection by this parasite that offer potential as future targets for use as indirect markers of pathogenesis or as targets for therapeutic intervention.

# DECODING DATA INTO KNOWLEDGE: BIOINFORMATIC STRATEGIES TO ANALYZE, INTEGRATE, AND INTERPRET HIGH-THROUGHPUT OMICS DATA

Over the last decade, the biomedical field has witnessed a tremendous increase in its capabilities to generate data. With large initiatives such as the 1000 Genomes Project (1000 Genomes Project Consortium et al., 2015) (which expanded upon the foundation established by the Human Genome Project), ENCODE (Encode Consortium, 2012), the Genotype-Tissue-Expression (GTEx) Project (GTEx Consortium, 2013), among others, the performance of large-scale omics investigations has gained more traction, and the adoption of high-throughput technologies is now widespread. The development of novel analytical strategies to decode and transform these data into knowledge is of paramount importance. In this section, we begin by reviewing traditional bioinformatics tools that can be applied to the analysis of high-throughput datasets. Next, we present more recent, complementary approaches that have yet to become entirely embraced by the community, paralleled by the development of computational techniques used by the scientific community working with leishmaniasis.

# Differentially Expressed Molecules: Only the Tip of the Iceberg

Traditional RNA-seq analyses begin by identifying genes with significantly altered expression across groups of samples, yielding a list of differentially expressed genes (DEGs). Statistical strategies for detecting DEGs based on RNA-seq-derived count data rely mainly on the use of Poisson or negative binomial distributions. The first has the advantage of being simpler, with a single parameter, λ, entirely defined by the mean, and having a variance equal to the mean. However, this property limits its application when biological replicates are available, when this assumption regarding variance does not hold because biological replicates typically present high variability (Bullard et al., 2010). In contrast, the negative binomial distribution, specified by the mean µ and variance σ 2 , is considered a more appropriate alternative since its variance is always greater than or equal to the mean. It also allows modeling of the mean-variance relationship typically observed in RNA-seq count data (Oberg et al., 2012). Computational tools that use the negative binomial include edgeR (Robinson et al., 2010), DESeq2 (Love et al., 2014), and baySeq (Hardcastle and Kelly, 2010), among others. Limitations of the negative binomial distribution include the observation that, in practice, µ and σ 2 are usually estimated from the data, which can be problematic when only a few replicates are available, as is still common practice for many high-throughput experiments, given budget constraints. These methods also suffer when the distributional assumptions in the input data do not hold, and non-parametric strategies have been proposed as more reliable alternatives in these cases (Tarazona et al., 2011; Li and Tibshirani, 2013). Other strategies, such as data transformation using voom (Law et al., 2014), in which the mean-variance trend is modeled in a non-parametric fashion, allow the subsequent use of traditional microarray packages, e.g., limma (based on normal distribution assumptions) and other microarray-specific downstream analyses. Many studies have compared the performance of these algorithms under various scenarios, including the variation in the number of replicates, sequencing depth, and the use of other tools concomitantly, such as in mapping steps (Rapaport et al., 2013; Soneson and Delorenzi, 2013; Law et al., 2014; Zhang et al., 2014; Schurch et al., 2016; Costa-Silva et al., 2017; Sahraeian et al., 2017; Williams et al., 2017). The wealth of methods available is indicative that there is no "one-tool-fits-all" approach for detecting DEGs in RNA-seq data and suggests that the a priori delineation of the experimental design together with knowledge of the biological question addressed is crucial in choosing the best set of tools and parameters.

A list of DEGs, however, is only a first step toward defining the biological processes that appear altered in a given experiment, and in most settings, the sole study of these genes will be too reductionist in nature and mostly ineffective as it must be conducted in a gene-by-gene fashion. Additionally, an intrinsic problem when detecting DEGs is the need to establish thresholds for p-values (or multiple testing corrected p-values) and foldchanges (FC), which can be largely arbitrary and lead to the loss of true DEGs (if too conservative) or their false inclusion (if too relaxed). The finding that the different DEG tools present slightly different true positive and false positive rates performance further complicates the matter (Schurch et al., 2016). To illustrate this point, recently published studies in the leishmaniasis field such as that of Christensen et al. (2016) for instance, used cutoffs of absolute FC ≥ 2 and a Benjamini-Hochberg corrected p-value ≤ 0.05 to call DEGs, as did Masoudzadeh et al. (2017). Others, such as Kong et al. (2017) have relied on a consensus strategy among different approaches, where a gene was considered differentially expressed if three methods positively identified it as a DEG, and gene lists with varying strictness for the corrected p-values (<0.001 and <0.01) were generated. To circumvent some of these issues and obtain a more expanded view of omics datasets, other analytical approaches can be used, and we detail some alternatives in the two sections that follow.

Proteomics allows the identification and quantification of many (usually thousands) of proteins present in a given sample. Recent advances in the experimental approaches available for accessing the proteome have allowed an improved resolution, with less input material, when compared to more classical techniques such as 2D gel electrophoresis (2D-GE) that can be followed by liquid chromatography coupled to mass spectrometry (LC-MS). Proteome quantification using MS can be generally classified as label-based or label-free approaches. The first relies on the differential labeling of samples using stable isotopes (such as <sup>2</sup>H, <sup>13</sup>C, <sup>15</sup>N, and <sup>18</sup>O) followed by quantification using MS. Technological improvements in the field of MS and chromatography have leveraged the development of high-throughput proteomic analyses that permit a higher proteome coverage and are collectively termed labelfree quantitative proteomics (LFQP), a highly accurate method that presents less susceptibility to technical errors. LFQP relies on measurements of individual samples by MS, and quantifies proteins based on either peak intensity or spectral counts of each peptide. Each of these broad techniques have their specificities regarding sample preparation, purification, separation, and ionization method, making the recommendation of specific computational tools for their analysis particularly more challenging than for RNA-seq data. For instance, the choice of labeling method will inform the corresponding choice of appropriate analytical packages, and a software that works well for <sup>15</sup>N label-based quantification may not be suitable for analyzing <sup>18</sup>O data (Anand et al., 2017). For this reason, we refer the reader to in-depth reviews that have tackled the methodological aspects related to the analysis of raw proteomic data (Mueller et al., 2008; Haga and Wu, 2014; Sandin et al., 2014; Kuharev et al., 2015; Navarro et al., 2016; Ramus et al., 2016; Välikangas et al., 2017), while focusing, for this review, on computational tools that use pre-processed data as input for downstream analyses. Albeit different in nature, proteomic studies suffer from similar concerns as those raised for RNA-seq data, in that the sole examination of a list of differentially expressed proteins across conditions (also constructed using ad-hoc criteria) may lead to loss of important biological aspects of the data. Rather, we argue that those working with high-throughput omics datasets will benefit from more ample analyses, such as those discussed in the following sections.

# Enrichment Analyses Allow a Contextualization of Altered Biological Processes in High-Throughput Data

While the identification of expression changes at the genelevel allows one to conveniently explain and validate small phenomena, such as by quantitative RT-PCR assays, the use of integrative approaches permits a broader understanding of the biological processes that underlie more complex questions, such as infection of the host cell by a pathogen. The contextualization of genes into pathways and other more general cellular processes effectively reduce the need to interpret causation at the gene-level and simultaneously reduce the dimensionality of the problem, as a single pathway is usually composed of several genes that act in concert to perform their cellular function.

Enrichment analysis tests whether, for a given set of events of interest (that could be DEGs or proteins, or groups of coexpressed molecules), there is over-representation (enrichment) of associated biological features than would be expected by chance. These biological features are usually cellular processes based on a common vocabulary (or ontology), including the Gene Ontology (GO) (The Gene Ontology Consortium, 2017), KEGG, and Reactome pathways (Kanehisa et al., 2017; Fabregat et al., 2018), as well as other more specific biological states such as oncogenic- and immunological-related ones, e.g., those from the MSigDB, which also comprises a myriad of other biological collections (Liberzon et al., 2015). Tools that perform enrichment analysis may utilize a single source of biological information (such as Reactome and Gene Ontology, which offer enrichment analysis tools but are restricted to their own vocabulary) or perform an integrated analysis of many sources concomitantly such as DAVID (Huang et al., 2007) and Panther (Mi et al., 2013), with the latter tools having the advantage of extracting complementary biological information available at different data sources. It is more important, however, that the utilized underlying data source be current and updated because tools based on outdated annotations can profoundly impact the results of enrichment analysis by effectively underestimating the functional significance of the gene lists used as inputs (Wadi et al., 2016).

Enrichment analysis strategies can generally be grouped into two main approaches: (1) list-based and (2) rank-based methods. The first relies on a set of biomolecules of interest that can be derived from the list of DEGs (if working with RNA-seq Veras et al. Biomarkers for Pathogenesis and Control of Leishmaniasis

data), proteins (if working with proteomics), or compounds, if the obtained data are from metabolomics experiments. To calculate the significance of the enrichment, these tools usually rely on statistical methods based on distributions, such as χ 2 (chi-squared), hypergeometric, and binomial, and evaluate whether there is an overrepresentation of biomolecules in the corresponding annotations from the data sources (such as genes in a pathway) that could be deemed statistically significant, usually after correcting for multiple hypothesis tests. One of the most used tools in this class is DAVID, which registers over 15,900 citations (Huang da et al., 2009b). A drawback of these approaches is the creation of the gene list itself, as different thresholds (such as those previously indicated for DEG identification) can be used, thus leading to gene lists of variable reliability. Additionally, genes with small expression changes, but having important biological roles, will probably not be included in such lists. Rank-based methods attempt to overcome these limitations by using the complete list of biomolecules as input when performing enrichment analysis, and the list is ranked using an appropriate metric, such as the elements on the top (or bottom) as more biologically important. Kolmogorov-Smirnovlike statistics can then be applied to calculate enrichment significance. In the case of omics studies (such as RNA-seq or quantitative proteomics), an appropriate metric could be, e.g., FC-values ordered in a decreasing manner, where the extremes represent biomolecules that are upregulated or downregulated (at the bottom) in a comparison of interest. Alternative metrics could be used for other data types, such as p-values and abundance. The Gene Set Enrichment Analysis tool (Subramanian et al., 2005) is among the most popular software for performing this class of analysis. Similar to many enrichment analysis tools, including DAVID, it was originally developed for use with microarrays, but its application to RNA-seq data is also possible. In particular, many tools that were previously restricted to use in microarray data, such as ROAST (Wu et al., 2010), can now also be employed with RNA-seq data using transformation strategies such as the previously described voom/limma pipeline (Law et al., 2014), so count-based data can be more closely related to those of microarrays.

**Table 1** provides a non-exhaustive list of some of the tools available to perform enrichment analysis fulfilling two criteria: (1) they are currently maintained, and (2) the database annotations on which they rely are updated (at most annually). However, as this field has grown substantially with the advent of high-throughput technologies, a multitude of tools have, in parallel, become available for performing these tasks, and we also refer the reader to specific reviews for a more comprehensive assessment, such as those studies from Huang da et al. (2009b), García-Campos et al. (2015), and Felgueiras et al. (2018), as well as Huang da et al. (2009a) in the **Table 1**. By focusing on 11 reports in the leishmaniasis community that used RNA-seq data, the use of the voom/limma and edgeR's pipeline for the identification of DEGs is indicative that somewhat "standard" tools are in use (**Table 2**). In particular, voom/limma allows microarray-like analysis, and its wide use is probably reminiscent of the extensive application of microarrays by the community, as exemplified by the recent parasite-focused review by Alonso et al. (2018). For enrichment analysis, however, only GSEA appears consistently among studies that performed such analyses, which may be a reflection of the multiplicity of tools available for this purpose. Thus, no clear picture emerges. Three of the 11 studies restricted their analysis to that of the DEG list (**Table 2**). In summary, the contextualization of lists of interesting biomolecules or pre-ranked sets thereof into the pathway and other cellular processes facilitate the interpretation of results derived from high-throughput data and should be used as complementary approaches to address the biological questions underlying omics experiments, thus allowing broader analyses.

# Network-Based Analyses Offer a More Global View of omics-Derived Data

While enrichment-based methods allow one to obtain a wider view of high-throughput omics experiments compared to examining a list of individual biomolecules, a complementary strategy consists of constructing networks of biomolecules. A well-known facet of biological systems is that the different elements (genes, transcripts, and proteins) and scales (genomic, transcriptomic, proteomic, and regulatory) that form these systems are intrinsically connected, such that single pathways or cellular processes seldom occur in isolation in a cell. Instead, the different cellular programs perform in a coordinated manner to achieve their biological functions. This behavior is amenable to modeling using a network-based framework. While there are various ways of applying network-based techniques, in this review, we focus on the construction of correlation networks and module detection approaches, but some methods to infer regulatory patterns are also described.

Correlation networks are being increasingly used to describe correlational patterns in omics datasets, and the elements (or nodes) that form these networks can be genes, proteins, or metabolites. In simple correlation networks, an interaction (or edge) between any two nodes is established when the value of their correlation, which can be obtained using Pearson's r or Spearman's ρ, passes a given threshold. A number of biological questions have been approached using this framework, and some applications in the context of leishmaniasis include the study of distinct states of infection with Leishmania infantum (Gardinassi et al., 2016) and the evaluation of the host-parasite interplay in localized cutaneous leishmaniasis caused by L. braziliensis (Christensen et al., 2016). Both studies used expression data as input to construct weighted gene-gene correlation networks, a particular case of a correlation network in which the edges have associated weights and no strict conditional on the correlation values is set, which characterizes a soft-thresholding approach. This method is referred to as weighted gene correlation network analysis (WGCNA) (Langfelder and Horvath, 2008) and has been used to search for biomarkers of psoriasis (Sundarrajan and Arumugam, 2016), various cancer types (Li et al., 2017; Xia et al., 2018; Yuan et al., 2018), as well as other complex, multifactorial conditions such as coronary heart disease (Huan et al., 2013).

Simple correlation networks are constructed by applying a hard-thresholding approach (i.e., reject correlations below a fixed

### TABLE 1 | Computational tools for performing functional enrichment analysis using omics datasets.


<sup>a</sup>Year of original publication.

<sup>b</sup>Date of last update relevant only to tools that rely on embedded or external databases.

<sup>c</sup>Number of citations of the original publication retrieved from Google Scholar, current as of May 2018.

d If more than one, the original work and the most recent update are cited.

\$Based on PubMed all-time search using "Ingenuity Pathway Analysis" as a query.

\*Based on PubMed searches for the first usage of the tool published in the literature.

TABLE 2 | Statistical and bioinformatics analyses performed in published articles in the leishmaniasis field that employed RNA-seq and proteomics techniques.


–, did not perform.

threshold), a strategy that may lead to a loss of information because correlations that fall even slightly below the threshold will be discarded. Defining such limits can also be overly arbitrary and dataset-specific. In contrast, correlation networks constructed using WGCNA mitigate these issues by applying a mathematical transformation to the correlation values, yielding a weighted network where the edge strengths are bounded by the transformed correlation values. The algorithm begins by first obtaining a correlation matrix from the input, usually expression data, but other omics data types can also be used. For expression data, correlation is used as a proxy for coexpression, which relates biologically to functional coupling (for instance, a group of co-expressed transcripts probably code for proteins participating in a common process) or regulatory aspects (such as activation of a transcription factor leading to increased expression of the regulated gene). The choice of correlation metric for constructing these networks has been a subject of investigation (Kumari et al., 2012; de Siqueira Santos et al., 2014), and although traditional metrics can be used, the biweight midcorrelation is recommended by the authors of WGCNA as a more robust alternative against outliers in the data (Langfelder and Horvath, 2012). Once all pairwise correlations are calculated, the correlation matrix is transformed into an adjacency matrix using a power function of the form f (x) = x β , where x represents elements in the correlation matrix, and a value of β ≥ 1 (called the soft-thresholding parameter) is chosen by the user such that the resulting correlation network adheres to a scale-free property while maintaining high connectivity (Langfelder and Horvath, 2008). Because this can lead to a range of valid β-values, an automated selection method has been proposed in the recently published CEMiTool pipeline (Russo et al., 2018). With the correlation network at hand, the next step involves detecting modules of co-expressed genes, which can be performed using hierarchical clustering per default in WGCNA or using hybrid approaches such as an additional K-means clustering step, which has been reported to improve the quality of the disclosed clusters (Botía et al., 2017). Once the modules of correlated bioelements are identified, several downstream analyses can be performed, including functional enrichment, a strategy coupled to a "guilty-by-association" paradigm that can lead to identification of novel gene functions (e.g., genes previously unrelated to a cellular pathway belonging to a module enriched for genes that belong to said pathway). Within a module, the pinpointing of "hub" genes, such as those with more connections, enables further stratification of genes that compose each module. It is also possible to calculate the module eigengene, a metric that summarizes the gene expression/abundance profiles in a module, which is defined by its first principal component (Langfelder and Horvath, 2008). The module eigengene can be correlated to trait data, such as clinical phenotypes and other associated variables in an experiment, and this module eigengene-phenotype association facilitates biomarker identification (Cui et al., 2015; Liu et al., 2015). While WGCNA is not the only tool available to create networks, its simplicity of use and biologically sound results may explain its broad acceptance, as measured by its high number of citations (**Table 3**). Alternative approaches are shown in **Table 3**, with some being independent of network inference, as exemplified by coseq (Rau and Maugisrabusseau, 2017), while others are geared toward the elucidation of regulatory



<sup>a</sup>Year of original publication.

<sup>b</sup>Number of citations of the original publication retrieved from PubMed.

c If more than one, the original work and the most recent update are cited.

interactions, such as ARACNe (Margolin et al., 2006), GENIE (Huynh-Thu et al., 2010), and the commercial alternative IPA (Krämer et al., 2014). It is important to stress that the development of computational tools for biological data analysis is a fast-moving and continuously evolving research field, and while we focused on specific tools and databases that we deemed appropriate and current, alternative solutions (either commercial or open-source) are probably available for performing many of the tasks referred here.

In summary, correlative approaches offer an alternative way of examining omics datasets in a completely data-driven fashion. Although we have focused mostly on expression data to exemplify the use of this technique, correlation networks are agnostic to the data type and can be constructed using any biomolecule with interactions that are amenable to modeling using a systems framework, including proteins (e.g., Zhang et al., 2016) and metabolites (e.g., Dileo et al., 2011) and the specificities involved in the adaptation for each data type were the subject of a recent investigation (Pei et al., 2017). The coupling of network creation and module detection with enrichment methods permits researchers to conduct more integrative analyses and extract biological insights in a much richer way than in traditional, single-gene based approaches. With the current trend of expanding the adoption of omics, particularly RNA-seq data, by the leishmaniasis scientific community (**Figure 1**), the knowledge and application of these more advanced computational techniques will be of utmost importance for progress in the field.

# TRANSCRIPTOMICS CONTRIBUTION TO UNDERSTANDING THE HOST RESPONSE TO LEISHMANIA INFECTION

Thus far, we have presented some of the analytical hurdles involved in the analysis of omics datasets. In the following sections, we focus on how the use of high-throughput approaches allowed an improved comprehension of Leishmania infection and host interplay.

Several studies have explored the advantages of transcriptome profiling using RNA-seq vs. other techniques to identify, analyze, and quantify transcriptomes from a variety of eukaryotic organisms. Most importantly, in comparison to other transcriptome sequencing techniques, RNA-seq offers improvements in terms of quality and precision regarding the level of transcripts and their isoform measures (Wang et al., 2009; Oshlack et al., 2010). A recent study highlighted the importance of RNA-seq as a tool to reveal gene expression at different stages of protozoan parasite development and to identify parasite genes modulated by vertebrate and invertebrate hosts via the simultaneous sequencing of parasite and host cell transcripts (Patino and Ramírez, 2017). A series of comprehensive studies attempted to investigate host cell signatures in response to Leishmania spp. infection by identifying not only DEGs but also modulated pathways using enrichment analysis, as discussed in section Enrichment Analyses Allow a Contextualization of Altered Biological Processes in High-Throughput Data of this review. These studies have greatly expanded our knowledge

regarding the virulence mechanisms of these parasites and their interactions with hosts (Guerfali et al., 2008; Dillon et al., 2015; Novais et al., 2015; Christensen et al., 2016; Fernandes et al., 2016). Although beyond the scope of the present review, we must acknowledge some recent studies that aimed to investigate the gene-wide transcriptional profiles of cutaneous lesions from patients infected with Leishmania braziliensis (Maretti-Mira et al., 2012; Novais et al., 2015; Christensen et al., 2016). One of these studies comparatively evaluated gene expression in lesions from patients who developed mucosal leishmaniasis and those who did not (Maretti-Mira et al., 2012). Another investigated gene expression in L. braziliensis-infected cutaneous lesions in comparison to normal skin (Novais et al., 2015), and a third report simultaneously analyzed the transcriptomic profiles of L. braziliensis amastigotes derived from skin lesions in L. braziliensis-infected patients and lesion skin samples by comparing profiles at early and late stages of disease and comparing lesions lacking detectable parasite transcripts and lesions with parasite transcripts that were readily detected, and used weighted gene-gene networks to globally assess the human host gene expression (Christensen et al., 2016). In addition, although comprehensive studies using microarray technology have contributed to the understanding of the host gene expression profile in response to parasites that cause visceral leishmaniasis (Gardinassi et al., 2016), we were unable to identify any RNA-seq studies analyzing the response to these parasite species. Another aspect that should be taken into consideration is metabolic changes induced in host cells by Leishmania parasites. While we recognize that metabolomic analysis represents an important aspect that has recently been explored in the field of leishmaniasis (Armitage et al., 2018; Cuypers et al., 2018), which certainly contributes to the understanding of disease, the results from these studies fall outside the scope of the present study. The present review instead focuses on global transcriptome analysis of macrophages in response to infection, which has been poorly investigated using RNA-seq technology (Dillon et al., 2015; Fernandes et al., 2016).

# Transcriptomic Analysis Allowing the Opportunity to Identify Possible Biomarkers in Leishmania-Infected Macrophages

Recent studies that analyzed gene expression profile in host cells have demonstrated that early stages [4 hours post-infection (hpi)], as opposed to later time points after infection (24, 48, and 72 h), seem to be ideal for the identification of DEGs or specifically modulated pathways in mouse or human macrophages (Dillon et al., 2015; Fernandes et al., 2016). This notion is supported by a well-designed protocol that used not only uninfected human macrophages as controls but also cells that engulfed inert particles to comprehensively distinguish genetic expression induced by phagocytosis from that arising upon infection, which has been shown to be indistinguishable at later infection times (Fernandes et al., 2016). Therefore, this study was able to capture the unique response of macrophages to each of the two Leishmania species investigated, Leishmania major and Leishmania amazonensis, which can potentially cause different clinical manifestations, by excluding the effect on human macrophages to inert particles. Interestingly, using principal component analysis (PCA), both human macrophages and cells incubated with latex beads for 4 hpi were shown to be clustered together, indicating that macrophages in culture can undergo phagocytosis without disturbing their steady-state transcriptome. As previously described (Vieira et al., 2002; Lee et al., 2007), phagocytosis triggers the activation of a local cascade of events that results in a cytoskeletal imbalance and formation of the phagocytic cup. By contrast, infected human macrophages seem to activate a unique transcriptional profile in response to Leishmania parasites at 4 hpi, regardless of species, since L. major- and L. amazonensis-infected cells have been shown to cluster together (Fernandes et al., 2016). This approach allowed the identification of specific genes that are expressed in response to infection, including potential macrophage biomarkers.

An evaluation of changes in the transcriptomic response to Leishmania infection over time revealed that murine and human macrophage responses to infection at early stages of infection vary significantly from those observed at later timepoints, by demonstrating that the number of DEGs, in comparison to uninfected macrophages, is higher in L. major and L. amazonensis-infected human macrophages at 4 hpi, with decreasing quantities observed at later time points. By contrast, infected human macrophages activate a similar transcriptomic response to uninfected macrophages that internalized inert particles at 24 hpi. In consonance with this finding, these two populations of macrophages, as well as uninfected control macrophages, all clustered together at 48 and 72 hpi (Fernandes et al., 2016). Evaluation of the phagocytotic effect on gene transcription demonstrated a lack of response in bead-containing macrophages at 4 hpi, with no DEGs observed between these macrophages and uninfected cells, although highly pronounced differences were detected at later time points. These findings indicate that, in contrast to the response exhibited by uninfected macrophages and macrophages that internalized the latex beads, Leishmania triggers a unique transcriptomic response shortly after phagocytosis, with reduced communication between the parasite and host cell at later stages of infection (Fernandes et al., 2016).

Similar to what was observed in human macrophages, in comparison to uninfected cells, L. major-resistant C57BL/6 macrophages were also shown to differentially modulate the variable numbers and types of genes at 4 hpi vs. later timepoints. At all tested time points, only 47 genes were up- or downregulated, which did not seem to be functionally related, except for the heavy metal transporters metallothionein 1 and 2 (Dillon et al., 2015). In L. major- and L. amazonensis-infected human macrophages, metallothionein 1 family members were also found to be some of the most upregulated (up to a 136 fold increase during L. major infection and a 196-fold increase in response to L. amazonensis infection, both compared with uninfected cells). These potential biomarkers are proteins that have previously been associated with an immunomodulatory response (Lynes et al., 1993) and are known to be activated by certain stimuli, such as exposure to reactive oxygen species (Ghoshal and Jacob, 2001), which has been confirmed to influence the host response to Listeria spp. (Emeny et al., 2015). Metallothioneins have also been found to be highly upregulated in macrophages infected with Leishmania (Chaussabel et al., 2003; Ettinger and Wilson, 2008) and have also been associated with resistance to treatment with antimonial drugs (Gómez et al., 2014). Despite this insight, the actual role played by these proteins in the establishment of Leishmania infection warrants further investigation.

Although L. major and L. amazonensis differ in several aspects of interaction with host cells (Kaye and Scott, 2011; Real et al., 2014), they surprisingly trigger a quite similar global transcriptomic response in human macrophages, with only four genes known to be differentially expressed at 4 hpi, compared to none at subsequent time points. This finding seems to indicate that human macrophages possess a nominal ability to distinguish between L. major and L. amazonensis at the transcriptional level, despite differences in several aspects of the clinical presentation of tegumentary leishmaniasis caused by these parasite species, as well as host immune response (Fernandes et al., 2016). This finding indicates that, in the search for novel biomarkers, it is likely that only those that would be similarly detected in macrophages, regardless of the parasite species that causes disease, will be identified. Notably, among the few DEGs identified between L. major- and L. amazonensisinfected macrophages, the authors reported that two were involved in the essential mechanisms of parasite establishment inside host cells: synaptotagmin family members 2 and 8 (SYT2 and SYT8), which are membrane proteins implicated in the regulation of vesicle docking and fusion in exocytosis (Baram et al., 1999; Arango Duque et al., 2013) and phagocytosis (Czibener et al., 2006; Vinet et al., 2008; Arango Duque et al., 2013). Although other synaptotagmin family members, SYT5 and SYT11, have been implicated in Leishmania infection, the roles played by SYT2 and SYT8 require further investigation. It has been proposed that the higher expression levels of SYT2 and SYT8 observed during L. major infection may be linked to differences in L. major-induced vacuole maintenance throughout the course of infection in terms of how the parasites divide within these compartments, i.e., the maintenance of a single parasite in one vacuole upon division, in contrast to L. amazonensis, which inhabits large parasitophorous vacuoles that potentially require more fusion (Veras et al., 1994, 1996) instead of fission events (Fernandes et al., 2016). Synaptotagmins are also involved in the regulation of SNARE activity by influencing membrane fusion via a Ca<sup>2</sup> -dependent mechanism (Tucker and Chapman, 2002; Andrews and Chakrabarti, 2005; Südhof and Rothman, 2009).

In comparison to uninfected cells, C57BL/6 macrophages infected with L. major upregulated two genes (Bnip3 and Bcl2a1b) related to the Bcl2 inhibitor of apoptosis, which is associated with inhibiting macrophages from resisting cell death (Dillon et al., 2015). How this finding is associated with a resistance profile in this murine model of leishmaniasis seems unclear. Previously, it was demonstrated that murine bone marrow-derived macrophages infected with L. major exhibited reduced programmed cell death when induced by stimuli, such as the deprivation of growth factors or treatment with staurosporine. Interestingly, this preventive effect was detected in both macrophages from L. major-susceptible BALB/c and L. major-resistant C57BL/6 mice, suggesting that the observed reduction in programmed cell death might be a parasite-triggered process that is seemingly independent of host genetic background and is unrelated to resistance and susceptibility to infection (Akarid et al., 2004).

# Integrative Bioinformatics Analyses Offer a Comprehensive View of Sets of Possible Biomarkers in Leishmania-Infected Macrophages

As discussed initially, the identification of pathways using database resources aids in a more complete understanding of the global response of host cells and tissues to a specific microorganism. A comprehensive analysis of these pathways using e.g., KEGG could help identify those genes that represent potential targets for disease intervention. In the C57BL/6 murine infection model involving L. major, the most highly modulated macrophage gene expression was related to the immune response, which is consistent with the resistance observed in these mice. Some of the upregulated genes that clustered together under KEGG analysis were Tnf, Hif-1, NF-kappa-B, Jak-Stat, PI3K-Akt, and Mapk, which are involved in cytokine-cytokine receptor interactions, arginine and proline metabolism, glycolysis and signaling pathways (Fernandez-Figueroa et al., 2016). Transcripts for inflammatory cytokines and their receptors were also found to be upregulated in L. major-infected mouse macrophages, including Il1, Il6, Tnf, Il1rap, Il18r1, and Nos2. In addition, KEGG enrichment analysis showed that murine macrophages infected with L. major expressed genes involved in the anti-inflammatory response, including Il11r, Il1rn, Il10, Socs3, Fos-induced growth factor (Figf), hemoxygenase1 (Hmox1), epithelium growth factor receptor (Egfr), vascular endothelial growth factor (Vegf), colony-stimulating factor 1 (Csf1), and colony-stimulating factor 3 (Csf3) (Weis et al., 2009; Luz et al., 2012; Canavese et al., 2015). Accordingly, the responses observed in human macrophages infected with L. major at 4 hpi were similar to those of murine macrophages, resulting in the upregulation of genes encoding inflammatory cytokines, including Il1 and Il6, and the upregulation of immune regulatory genes, including prostaglandin endoperoxide synthase 2 (Ptgs2), Csf1 and colony-stimulating factors 2 (Csf2), and superoxide dismutase 2 (Sod2). This finding suggests that L. majorinfected macrophages probably evolved the ability to inhibit a deleterious innate inflammatory immune response (Fleming et al., 2015); alternatively, this anti-inflammatory response could be a consequence of the effort by host macrophages to control parasite infection (Dillon et al., 2015). Consistent with these findings, in the sera of patients during the active phase of visceral leishmaniasis it was detected a significant increase in inflammatory mediators including LTB4, RvD1, PGF2α (PGF2α), IL-1β, IL-6, IL-8, IL-10, IL-12p70, and TNF-α, and a decreased level of TGF-β1 (Araújo-Santos et al., 2017).

Enrichment analysisinvolving C57BL/6 macrophages infected with L. major, conducted at 4 and 24 hpi, identified activation of the glycolysis/gluconeogenesis pathway, which contains genes that encode glycolytic enzymes, such as phosphoglycerate kinase, hexokinases, enolase, lactate dehydrogenase A, and glyceraldehyde-3-phosphate dehydrogenase. This finding seems to indicate that the glycolysis pathway represents a metabolic response arising in macrophages due to L. major infection, which, upon toll-like receptor ligation, likely results in the stimulation of an inflammatory response capable of triggering anaerobic glycolysis (Tannahill et al., 2013). Whether this metabolic response to Leishmania spp infection is typical of host macrophages or whether it is due to host resistance to infection warrants further study.

Few pathways have been found to be downregulated in the L. major murine infection model. At 4 hpi, downregulation of the lipid metabolism and biogenesis pathways was observed. In addition, in the "Fc gamma R-mediated phagocytosis" KEGG pathway, receptors and signaling molecules involved in the process of phagocytosis were downmodulated at 4 hpi. Previously, it has been demonstrated that macrophages are more permissive to IgG-opsonized-Leishmania phagocytosed by the Fc gamma receptor (Mosser, 1994). It is possible that the observed resistance to L. major could be related to a possible reduction in the uptake of L. major by C57BL/6 macrophages secondary to the downregulation of this pathway. However, the mechanism underlying this effect in this murine model and resistance to L. major by C57BL/6 macrophages in general requires further investigation.

Fernandes et al. (2016) generated transcriptomic data from infected cells and integrated those data with the database from a previous study (Dillon et al., 2015) to define a shared response that characterizes a general mammalian macrophage gene signature in response to Leishmania spp. infection. To identify known cellular processes within this signature, KEGG enrichment analysis was used to ascertain which genes were commonly up- or downregulated in infected cells. Most of the pathways identified contained upregulated genes that were related to immune activation and signaling responses. Regarding signaling pathways, KEGG analysis identified genes involved in the pathway of recognition of pathogen associated molecular patterns (PAMPs), e.g., retinoic acid-inducible gene-(RIG)-I-like receptor, nucleotide-binding oligomerization domain-NOD-like receptor, and Toll-like receptor; for the immune system signaling pathways, the detected genes were either related to the cytokine-cytokine receptor interaction pathway, including Fc epsilon RI, Jak-STAT, T cell receptor, NF-kB, mitogen-activated protein kinase (MAPK), TNF, vascular endothelial growth factor (VEGF), ErbB, FoxO, hypoxia-inducible factor 1 (HIF-1), and phosphatidylinositol 3-kinase-Akt [PI3KAkt], or related to the TGF-β signaling pathway. In addition, among the downregulated genes in both murine and human models of infection, KEGG identified pathways related to energy metabolism (glycan and amino acid degradation), lysosome structure and processes and apoptosis. KEGG enrichment analysis identified the FoxO signaling pathway among the genes that were either up- or downregulated, which is implicated in the regulation of cell growth, gluconeogenesis, and adipogenesis. The findings presented in RNA-seq technology raise the possibility of translating these pathways to biomarkers as surrogate endpoints following extensive validation studies.

# PROTEOMIC CONTRIBUTION TO UNDERSTANDING THE MACROPHAGE RESPONSE TO LEISHMANIA INFECTION

Different DNA- and RNA-based strategies have been used to provide insights into the host cell response to infection by different pathogens, including Leishmania. However, these studies did not provide information regarding translational and post-translational modifications and protein localization, which are essential to understanding gene functions. Thus, studying the proteins encoded by mRNAs is crucial for understanding the biological processes. Therefore, proteomic studies have gained significant relevance with the advancements in large-scale technologies and represent one of the most important tools for biomarker investigation. This approach has provided a wealth of protein expression data on the host response to infection by different pathogens (Chambers et al., 2000; Sundar and Singh, 2018).

Although proteomics is a known powerful tool to identify host cell protein expression (Chambers et al., 2000), only three studies have evaluated the macrophage response to Leishmania infection in the past 5 years. A previous work published by our group, using tandem liquid chromatographymass spectrometry (LC-MS/MS), was the first attempt to employ a large-scale proteomic analysis to identify host cell proteins expressed in response to Leishmania infection and, among them, potential macrophage biomarkers that could be related to a susceptibility or resistance profiles (Menezes et al., 2013). Two years after this paper was published, Singh et al. (2015) used a quantitative proteomic approach to study human monocyte-derived macrophage (THP-1) responses to L. donovani infection to investigate how the intracellular parasite manipulates the macrophage response. More recently, Goldman-Pinkovich et al. (2016) applied a phosphoproteomic analysis to understand the arginine deprivation response in infected macrophages and the underlying mechanisms.

In the first study, our group used a mouse model that was previously described as being resistant to L. major and susceptible to L. amazonensis, to identify markers that could be driving different responses of CBA mouse macrophages to Leishmania infection. A total of 62 proteins were predominantly expressed in infected macrophages. Of those, 15 proteins were found to be differentially expressed between L. amazonensis- and L. majorinfected macrophages. Thirteen of the 15 proteins exhibited reduced expression in response to L. amazonensis infection, but they were upmodulated in L. major-infected macrophages; in contrast, two proteins showed increased expression in response to L. amazonensis infection. The proteins with higher expression in L. major-infected macrophages were as follows: programmed cell death protein 5 (PDCD5), coronin 1B, HIF-1α, cytochrome C oxidase 6B (cox6B), osteoclast-stimulating factor-1 (OSTF1), protein phosphatase 2 (PP2), heterogeneous nuclear ribonucleoprotein F (HNRPF), PYD And CARD domaincontaining protein (PYCARD), RAB1, Serpin, ribosomal protein S2 (RPS2), and myosin light chain (Menezes et al., 2013). Networks constructed under the IPA framework revealed that proteins differentially expressed in CBA macrophages form part of biological modules related to cellular development and cellular metabolism, and their different modulation profiles possibly induce distinct macrophage responses, ultimately leading to disease susceptibility or control (Menezes et al., 2013). The upregulation of proteins such as HIF-1α, TRAP1, Serpin, and PYDCARD strongly suggest a modulation of the immune response after Leishmania infection. Two of these proteins, Serpin and PYDCARD, were downmodulated in L. amazonensisinfected macrophages. Serpin is a protein induced by TNFα that, together with IL-1β, is involved in the inflammatory cascade (Mishra et al., 2006). The reduced expression of Serpin in L. amazonensis-infected macrophages could be associated with a diminished inflammatory response, favoring the intracellular survival of the parasite. Additionally, the PYDCARD adapter protein also induced by TNF-α activates apoptosis via a mechanism that is dependent on NF-κB and caspases (Reed et al., 2003). These results are in accordance with a previous study performed in our laboratory, showing that CBA macrophages control L. major infection and express higher levels of TNF-α than L. amazonensis-infected macrophages (Gomes et al., 2003), which are susceptible to this parasite (Diefenbach et al., 1998).

Another critical molecule identified in this study as differentially expressed between L. amazonensis- and L. majorinfected macrophages is HIF-1α. The higher levels of this protein in macrophages infected by L. major could be associated with higher production of NO and expression of TNF, which are mediators that are known to play a role in HIF-1α regulation (Zhou et al., 2003). Additionally, investigation of the role of HIF-1α in Leishmania infection led us to the discovery of 17-AAG, a heat-shock protein-90 (HSP90) inhibitor, as a potential drug against leishmaniasis (Petersen et al., 2012; Santos et al., 2014). HIF-1α, a transcriptional factor that can potentially be modulated by specific drugs, is one of the client proteins of HSP-90, which is a very plentiful molecular chaperone in mammalian cells (Minet et al., 1999). This ATP-dependent chaperone, which is induced during stress responses, is known to play a role in the stabilization, correct folding and assembly of several client proteins, including HIF-1α. HSP90 is also expressed by protozoan parasites, which is crucial to the stabilization of heat-labile proteins inside these microbes. Treatment of L. amazonensis- or L. braziliensis-infected macrophages with 17-AAG dramatically reduced not only the percentage of infected cells, but also parasite load, in a dose- and timedependent manner together with decreases in the production of inflammatory cytokines (Petersen et al., 2012; Santos et al., 2014). More recently, we investigated the effect of modulating another identified biomarker using proteomic analysis, the peripheral benzodiazepine receptor (PBR), known as translocator protein (TSPO). We found that this mitochondrial transmembrane protein exhibited a lower relative abundance of peptides in cells infected with L. amazonensis in comparison to L. major (Menezes et al., 2013). Modulating TSPO with one of its ligand, PK11195, caused the killing of amastigotes in vitro at dosages considered non-toxic to macrophages, indicating its potential as antileishmanial (Guedes et al., 2018). In sum, these findings strengthen the potentiality of global analysis of Leishmaniainfected macrophages for the identification of biomarkers in host cells that probably participate in the pathogenesis of Leishmania infection and, subsequently, can function as targets for therapeutic intervention.

The proteomic study described herein also reveals a modulation of host cell metabolism induced by L. amazonensis. The results demonstrate that macrophages infected with L. amazonensis express higher levels of 6-phosphogluconate dehydrogenase (6PGDH), an enzyme in the pentose phosphate pathway, compared to L. major-infected cells (Menezes et al., 2013). The modulation of host cell metabolism induced by Leishmania has already been explored (Osorio Y Fortea et al., 2009; Lamour et al., 2012). The modulation of 6PGDH in cancer cells and its effect on cancer treatment are currently being studied (Zheng et al., 2017).

Another recently published study used a quantitative proteomic approach and THP-1-derived macrophages to evaluate the cell host response to L. donovani infection (Singh et al., 2015). The authors used the isobaric tag (iTRAQ) method and LC-MS/MS to compare the protein profiles of non-infected and L. donovani-infected THP-1 cells, and then performed an extensive analysis for contextualizing their results into ampler biological processes, which facilitated a global interpretation of the altered processes in response to infection. This analytical strategy is beneficial to obtain a comprehensive understanding of the studied phenomenon. The results showed that proteins involved in important metabolic pathways, such as glycolysis and fatty acid oxidation, were upregulated after L. donovani infection, suggesting that this parasite modulates host cell metabolism. The expression of proteins involved in gene transcription, RNA splicing [heterogeneous nuclear ribonucleoproteins (hnRNPs)], histones, and DNA repair and replication was also upregulated after L. donovani infection. Of note, several proteins identified in this study as differentially expressed between non-infected and L. donovani-infected macrophages had not been previously associated with the host cell response to Leishmania infection. Another exciting result of this work was the increased expression of the mitochondrial antiviral signaling protein (MAVS) after Leishmania infection. This protein is known to activate NF-κB and interferon (IFN) regulatory factors (IRF3 and IRF7), inducing the synthesis of type I interferons (IFN-α and IFN-β), which are essential during antiviral signaling. The silencing of endogenous MAVS expression by RNAi inhibits the activation of NF-κB, IRF3, and IRF7, leading to the blockade of interferon production and favoring viral infection (Yan and Tsai, 1999). These authors suggest that a crosstalk might occur between MAVS and NF-κB and IRF signaling pathways components, which would lead to the production of proinflammatory cytokines and type I IFN (Villa et al., 2003). Based on these findings, MAVS could be an interesting potential marker to investigate because it helps modulate the host inflammatory response to Leishmania infection. In addition, the modulation of host cell metabolism could be an interesting approach that could contribute to the control of Leishmania infection. Metabolomics combined with proteomic approaches represents one of the most important postgenomic analyses to investigate changes in cell metabolism and identify biomarkers during the course of infection inside macrophages (Singh et al., 2015). Several studies have demonstrated an association between host cell metabolism and response to different pathogens, including Leishmania (Lamour et al., 2012; Govinden et al., 2018; Price et al., 2018; Reddy et al., 2018).

The most recent study using a proteomic approach to better understand the host cell response to Leishmania infection applied this technology to investigate the signaling pathways involved in the upregulation of expression and activity of different transporters, such as Leishmania arginine transporter (LdAAP3), in response to arginine pool reduction in the host cell. To study phosphoproteins involved in the signaling pathway implicated in this response, the authors used a di-methylation tagging technique to investigate changes in the phosphorylation profile of Leishmania promastigotes after 5 and 15 min of arginine deprivation. Phosphoproteomic analysis revealed an increased phosphorylation of mitogen-activated protein kinase 2 (MPK2), indicating that this kinase could be involved in the arginine-deprivation response during Leishmania infection (Goldman-Pinkovich et al., 2016). Although this work did not investigate a more global cell host response to Leishmania infection, the utilized approach could be of great importance to identify potential markers that could be used for the development of new drug treatments and to understand the disease outcome.

Taken together, these few studies show that Leishmania parasites modulate the host cell proteome profile, reinforcing the idea that proteomic technology is a powerful technique that should be further explored by researchers to discover its full potential. Proteomics combined with bioinformatics represents a robust approach to investigate the global host response to infection and to identify new potential molecular markers that can control the fate of both host cell and pathogen during infection (Jean Beltran et al., 2017). In addition, further proteomic studies are required to investigate whether proteins that are modulated after Leishmania infection can be used as novel biomarkers and targets for the control of Leishmania infection.

# CONCLUSIONS

Combining the results from transcriptomic and proteomic investigations offers a more comprehensive body of information for the identification of possible biomarkers in Leishmania infection. The authors recommend compiling the findings from the studies referenced herein using macrophages, together with those obtained from blood, tissue and other cell types, and also relevant results from similar future studies, to form a complete set of potential biomarkers to aid in global analysis using transcriptomics, proteomics and metabolomics approaches. This data could be then used to identify and subsequently validate specific genes and proteins capable of enhancing the ability of researchers to identify host cell signatures at early time points in the context of leishmaniasis, in an effort to predict disease control or progression, and even the prognostic response to therapy.

# AUTHOR CONTRIBUTIONS

PV was the main responsible for conception and design and also for the formulation of the final version of this article review. PR and JdM made substantial contributions to conception and design, and also participate in drafting the article or revising it critically for important intellectual content.

# FUNDING

This work was supported by grants from Fundação de Amparo à Pesquisa do Estado da Bahia (PV http://www.fapesb.ba. gov.br), Conselho Nacional de Pesquisa e Desenvolvimento Científico (PV http://www.cnpq.br). PV holds a grant from CNPq for productivity in research (307832/2015-5). The funders had no role in study design, data collection or analysis, the decision to publish, or preparation of the manuscript.

# ACKNOWLEDGMENTS

The authors would like to thank Andris K. Walter for English language revision and manuscript copyediting assistance.

# REFERENCES


transcriptional modules associated with cutaneous immunopathology. J. Invest. Dermatol. 135, 94–101. doi: 10.1038/jid.2014.305


**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 Veras, Ramos and de Menezes. 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.

# Reverse Epidemiology: An Experimental Framework to Drive Leishmania Biomarker Discovery in situ by Functional Genetic Screening Using Relevant Animal Models

Laura Piel 1,2, Pascale Pescher <sup>1</sup> and Gerald F. Späth<sup>1</sup> \*

1 Institut Pasteur, Unité de Parasitologie Moléculaire et Signalisation, INSERM U1201, Paris, France, <sup>2</sup> Université Paris Diderot, Sorbonne Paris Cité, Paris, France

### Edited by:

Javier Moreno, Instituto de Salud Carlos III, Spain

### Reviewed by:

Elisa Azuara-Liceaga, Universidad Autónoma de la Ciudad de México, Mexico Bellisa Freitas Barbosa, Federal University of Uberlandia, Brazil

> \*Correspondence: Gerald F. Späth gerald.spaeth@pasteur.fr

### Specialty section:

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 18 June 2018 Accepted: 27 August 2018 Published: 19 September 2018

### Citation:

Piel L, Pescher P and Späth GF (2018) Reverse Epidemiology: An Experimental Framework to Drive Leishmania Biomarker Discovery in situ by Functional Genetic Screening Using Relevant Animal Models. Front. Cell. Infect. Microbiol. 8:325. doi: 10.3389/fcimb.2018.00325 Leishmania biomarker discovery remains an important challenge that needs to be revisited in light of our increasing knowledge on parasite-specific biology, notably its genome instability. In the absence of classical transcriptional regulation in these early-branching eukaryotes, fluctuations in transcript abundance can be generated by gene and chromosome amplifications, which have been linked to parasite phenotypic variability with respect to virulence, tissue tropism, and drug resistance. Conducting in vitro evolutionary experiments to study mechanisms of Leishmania environmental adaptation, we recently validated the link between parasite genetic amplification and fitness gain, thus defining gene and chromosome copy number variations (CNVs) as important Leishmania biomarkers. These experiments also demonstrated that long-term Leishmania culture adaptation can strongly interfere with epidemiologically relevant, genetic signals, which challenges current protocols for biomarker discovery, all of which rely on in vitro expansion of clinical isolates. Here we propose an experimental framework independent of long-term culture termed "reverse" epidemiology, which applies established protocols for functional genetic screening of cosmid-transfected parasites in animal models for the identification of clinically relevant genetic loci that then inform targeted field studies for their validation as Leishmania biomarkers.

### Keywords: Leishmania, biomarker discovery, reverse epidemiology, cosmid screen, functional genetics

# INTRODUCTION

Biomarkers are defined as biological characteristics that are objective and quantifiable indicators for responses to therapeutic interventions, or normal and pathogenic biological processes (Biomarkers Definitions Working Group 2001, 2001). With respect to Leishmania infection, we can distinguish direct biomarkers that are applied to determine parasite species and prevalence (e.g., parasitespecific proteins, lipids, transcripts, genetic loci), and indirect biomarkers that correspond to different correlates of the host anti-microbial response [e.g., adenosine deaminase (ADA) or cytokines such as IL-10 or TNF] (Kip et al., 2015).

Direct Leishmania biomarkers can have either purely diagnostic value (e.g., kinetoplast (k) DNA, ribosomal small sub-unit (SSU) RNA, HSP70 locus, carbohydrate antigens), or prognostic value allowing for the prediction of treatment outcome or disease evolution (e.g., dissemination in cutaneaous leishmaniasis or development of post-kala-azar dermal leishmaniasis in visceral leishmaniasis). However, despite their potentially important impact on clinical management of leishmaniasis, only few biomarker candidates with potential prognostic value are described, most of which are linked to drug resistance (Vanaerschot et al., 2012; Torres et al., 2013; Hefnawy et al., 2017; Ponte-Sucre et al., 2017). The absence of this class of markers is explained by various biological and technical constraints, some of which are linked to Leishmania genome instability that limits biomarker discovery and needs to be considered in ongoing and future biomarkers discovery campaigns.

In the absence of classical transcriptional regulation, Leishmania often regulates transcript and protein abundance by chromosome and gene copy number variations (CNVs) (Dumetz et al., 2017; Prieto Barja et al., 2017), which can drive environmental adaptation (Leprohon et al., 2009; Downing et al., 2011; Rogers et al., 2011; Brotherton et al., 2013; Mukherjee et al., 2013; Ubeda et al., 2014; Zhang et al., 2014; Laffitte et al., 2016). Our recent demonstration that karyotypic fluctuations and haplotype selection allow for fitness gain in culture reveals the importance of Leishmania genome plasticity in short-term evolutionary adaptation (Prieto Barja et al., 2017). Conceivably, the highly dynamic genomic changes occurring during culture adaptation challenge past and current protocols in Leishmania biomarker discovery, which rely on adaptation and massexpansion of field isolates in culture prior to analysis, often resulting in loss of epidemiologically relevant, genetic signals. Here, by drawing from the current literature, we propose an alternative strategy independent of long-term culture that is based on functional genetic screening in relevant animal models. Our review provides an overview on past functional screening results and their documented success in revealing genomic loci that are under environmental selection, and advocates for Leishmania biomarker discovery by combining cosmid selection and subsequent clinical validation, an experimental framework we termed "reverse" epidemiology. In the following we summarize studies that developed and applied cosmid-based approaches to identify new Leishmania factors linked to parasite pathogenicity, tropism and drug resistance, and discuss the potential epidemiological relevance of these factors where clinical data were available.

# COSMID-BASED FUNCTIONAL GENETIC SCREENING IN LEISHMANIA

Various genetic methods have been successfully applied in the past to identify Leishmania genes or genetic markers that are associated with disease outcome or clinical manifestation, including whole genome sequencing (WGS) of isolates (Downing et al., 2011; Rogers et al., 2011; Leprohon et al., 2015), random amplification of polymorphic DNA (RAPD) (Bhattacharyya et al., 1993; Schönian et al., 1996; Mkada-Driss et al., 2014), or assessment of amplified fragment length polymorphisms (AFLP) (Kumar et al., 2009, 2010a; Odiwuor et al., 2011; Jaber et al., 2018). Likewise, cosmid-based functional screens have been applied to discover clinically relevant loci. This approach is based on the genetic transfer of a given cellular phenotype (e.g., drug resistance) from a donor strain to a recipient strain via transfection of a cosmid library. While currently established WGS protocols for Leishmania biomarker discovery have been applied on clinical isolates maintained in long-term culture, causing potentially important bias, cosmidbased approaches can directly reveal clinically relevant genotypephenotype relationships, especially when applied in situ in infected animals. Even though this functional genetic approach represents a powerful tool, this technology has not been applied in a systematic way at larger scale to drive biomarker discovery.

The preparation and application of a cosmid library is a complex procedure, where genomic DNA fragments of an appropriate size are cloned into purified cosmid DNA and packaged into phages for efficient bacterial transduction, which allows for amplification of the library and assessment of its genomic coverage prior to transfection into parasites by electroporation. The generation of a first series of Leishmania shuttle cosmid vectors and the validation of a protocol that allows for genetic complementation and functional screening in these parasites using genomic cosmid libraires was established in 1993 by Beverley and collaborators (Ryan et al., 1993a) followed by Kelly and collaborators in 1994 (Kelly et al., 1994). Subsequently, this protocol was applied in various studies for the identification of Leishmania pathogenicity and drug resistance genes.

# Cosmid-Based Identification of Novel Leishmania Pathogenicity Factors

Key for Leishmania infectivity is the capacity of procyclic promastigotes to undergo differentiation into infectious metacyclic promastigotes able to resist to complement lysis encountered inside the mammal host following parasite transmission (Sacks and Perkins, 1984; Franke et al., 1985; Howard et al., 1987). Parasite resistance has been largely attributed to the surface glycolipid lipophosphoglycan (LPG), a major Leishmania virulence factor essential for L. major promastigote virulence (Späth et al., 2000, 2003a), that undergoes important modifications during metacyclogenesis (Sacks et al., 1990; Mcconville et al., 1992; Sacks, 2001). LPG biosynthetic genes and their virulence functions have been genetically identified combining cosmid screens with functional null mutant analysis and virulence assessment in macrophages and mice. LPG deficient mutants were generated by chemical mutagenesis, isolated by their failure to agglutinate in the presence of lectin (King and Turco, 1988), transfected with a cosmid library prepared from L. donovani, and screened for restoration of LPG expression using either lectin- or antibody-based agglutination assays revealing the two first LPG biosynthetic genes, a galactofuranose transferase encoded by the gene lpg1 (Ryan et al., 1993b), and an UdP galactose transporter encoded by lpg2 (Descoteaux et al., 1995). The virulence functions of both genes were confirmed in subsequent studies in L. major lpg1 and lpg2 null mutants (Späth et al., 2000, 2003b). Few years later, by combining cosmid library transfection and antibody panning, Dobson et al. identifed genes encoding arabinosyl- and galactosyltransferases that mediate developmental modifications of LPG during metacyclogenesis (Dobson et al., 2003a,b).

Cosmid-based functional screening has also been applied to gain insight into pathways that govern complement resistance in promastigotes revealing genes that are likely linked to metacyclogenesis. Based on the observation that decrease in resistance to complement lysis is a consequence of long-term maintenance in culture (Lincoln et al., 2004), Dahlin-Laborde et al. used genomic DNA from animal-derived Leishmania infantum (chagasi) promastigotes to construct a cosmid library that was transfected into long-term cultured parasites. The transfected parasites were subjected to complement lysis allowing for the selection of seven different cosmids that conferred increased complement resistance albeit at lower levels compared to short-term cultured control parasites. In-depth analysis of two cosmids revealed genomic fragments of L. infantum chromosome 36 (Dahlin-Laborde et al., 2005), with two subregions encoding, respectively, 5 and 13 genes shown to be critical for the phenotype, including an ADP-ribosylation factorlike protein and an ATP-dependent RNA helicase (Dahlin-Laborde et al., 2008). Cosmid screens were further applied by the Matlashewski team to identify virulence and visceralization factors using libraries prepared with genomic DNA from L. donovani transfected in L. major promastigotes. Transfectants expressing the heterologous library were inoculated into mice by tail vein or footpad injections and cosmids were recovered from parasites that established infection in spleen (type I), skin (type III), or both (type II). Subsequent analysis of individual ORFs by transgenic expression and infection validated an ORF encoding for an unknown protein and a 4.4 kb miniexon gene array on chromosome 36 (Zhang and Matlashewski, 2004). Unlike in L. major, overexpression of the miniexon region in L. braziliensis led to complete virulence attenuation in a hamster model (de Toledo et al., 2009), suggesting species-specific functions of this array. This is further supported by the genetic divergence of this array between new world and old world dermotropic species (Fernandes et al., 1994), which is used as a diagnostic signal for parasite genotyping (Serin et al., 2005; Ovalle-Bracho et al., 2016).

A final example documenting the power of cosmid-based approaches in identifying putative Leishmania virulence factors is represented by a complementation screen conducted using a cosmid library derived from an attenuated HSP100 null mutant that spontaneously recovered infectivity and/or pathogenicity in mice, likely by the amplification of a compensatory locus (Reiling et al., 2006). A screen conducted in mice using cosmidtransfected HSP100 null mutants and subsequent validation experiments revealed P46 as a new virulence factor (Reiling et al., 2010). A follow-up study by Bifeld et al applied a phylogenetic approach on 20 clinical isolates comparing P46 amino acid sequences thus establishing a strong correlation between P46 isoforms and their geographical origin. Transgenic parasites over-expressing three different P46 isoforms in a L. major lab strain were co-injected in BALB/c and C57BL/6 mice. Selection of different isoforms according to the mouse strain suggested that the P46 genetic polymorphism may be linked to parasite adaptation to genetically distinct, region-specific host reservoirs (Bifeld et al., 2015; **Table 1**).

# Cosmid-Based Identification of Leishmania Drug Resistance Genes

Since 1999, screening of cosmid libraries has been used as a gain-of-function strategy to identify drug resistance or drug tolerance genes (reviewed in Clos and Choudhury, 2006). Beverley and collaborators established the first proof-of-principle of this approach culturing cosmid transfected L. major parasites under pressure of the drugs methotrexate and tubercidin, which resulted in the selection of the known resistance genes DHFR-TS, PTR1, and TOR (Cotrim et al., 1999). The same study identified a new gene encoding a 63 kDa hypothetical protein located on chromosome 31 termed tubercidin-resistant protein (TRP) that is conserved in Leishmania and co-localizes in the endoplasmic reticulum in stationary phase promastigotes (Aoki et al., 2016).

Functional complementation has also been a powerful tool for the identification of transporters that can alter drug efficacy. The biopterin transporter bt1, previously named ORF G (Kundig et al., 1999), and the miltefosine (MIL) transporter LdMT (Perez-Victoria et al., 2003) were identified using Leishmania tarentolae transfected with a heterologous L. mexicana cosmid library selected under methotrexate pressure (showing that bt1 can confer resistance), and L. donovani MIL resistant parasites transfected with a L. donovani wild-type cosmid library subjected to MIL selection (showing that a non-mutated LdMT can restore susceptibility). Likewise, the cosmid approach was applied to screen for genes mediating resistance to two inhibitors of ergosterol biosynthesis, terbinafine, and itraconazole, which resulted in the selection of nine different cosmids, some of which conferred cross-resistance to both drugs, and the identification of squalene synthase 1 (SQS1) as an itraconazole resistance gene (Cotrim et al., 1999).

This approach has been recently applied to directly identify clinically relevant drug resistance loci by heterologous screening. Clos and collaborators prepared cosmid libraries from antimony SbIII/SbV resistant or SbIII sensitive/SbV resistant L. braziliensis field isolates that were transfected into SbIII sensitive/SbV resistant promastigotes. Culture under drug pressure selected for cosmids carrying a genomic fragment of chromosome 20, which also conferred drug resistance when transfected into L. infantum (Nühs et al., 2014). A competition assay with full-length or truncated derivatives of the cosmid insert validated ARM58 as a SbIII resistance gene. A more recent study performed by the same group with cosmid-transfected L. infantum extended this finding to the neighboring genes and defined a cluster of three genes, ARM58, ARM56 (previously named ARM58rel), and HSP23 at the telomere of the chromosome 34 that confer increased resistance of intracellular amastigotes against SbV (Tejera Nevado et al., 2016). Using a L. infantum cosmid library, the same team revealed a protein termed P299 that conferred increased resistance of intracellular amastigotes to MIL and reduced promastigote sensitivity to MIL and SbIII, but not pentamidin (Choudhury et al., 2008). Another gene—today annotated as cysteine leucine-rich protein (CLrP, LinJ.34.0570) was revealed causing antimony resistance in L. tarentolae transfected with a cosmid library prepared from arsenite and SbIII resistant parasites (Brochu et al., 2004), and in L. infantum axenic amastigotes (Genest et al., 2008). Brochu et al. also



\*chr, chromosome; \*\*validation refers to loss of function studies establishing a direct link between the gene and parasite pathogenicity.

reported members of the HSP70 protein family as important genes contributing to antimony tolerance, supporting recent phylogenetic evidence that HSP70 family members may allow parasite environmental adaptation with potential important consequences for drug susceptibility (Drini et al., 2016).

Recent work by the Ouellette team coupled cosmid selection and next generation sequencing for drug resistance and drug target gene discovery, proposing a high-throughput capable screening strategy the authors referred to as Cos-Seq (Gazanion et al., 2016). Screening cosmid transfected L. infantum against SbIII, amphotericin B, MIL, paramomycin or pentamidin revealed 64 enriched loci, including 12 common to at least two anti-leishmanial drugs, suggesting the existence of multi-drug resistance genes. This study validated 6 known and uncovered 7 new resistance genes in promastigotes, including two new genes causing methotrexate resistance both encoding for phosphatase 2C-like proteins (LinJ.34.2310 and LinJ.34.2320), one hypothetical protein with leucinerich repeats causing both pentamidin and paromomycin resistance (LinJ.06.1010), a serine/threonine phosphatase causing SbIII resistance (LinJ.12.0610), and phospholipid-translocating ATPase (LinJ30.2270) and C-8 sterol isomerase (LinJ.29.2250) that were revealed screening for MIL resistance (**Table 2**).

# THE FRAMEWORK OF "REVERSE" EPIDEMIOLOGY

The examples described above are testimony to the success of cosmid-based, functional screening approaches to discover genetic loci in Leishmania that are linked to parasite virulence, tissue tropism, and drug resistance. However, even though these loci may represent potential biomarkers with important prognostic value, there are no dedicated, concerted efforts for their validation in clinically relevant settings. One exception includes CLrP, whose increased abundance on RNA and protein levels were correlated with increased Sb resistance in field isolates, albeit only a small number of isolates were used in these studies (Kumar et al., 2010b; Das et al., 2015). For other loci, clinical validation of the functional screening results can be ambiguous, with for example the MRPA and PTR1 genes of the H-locus having been either strictly, partially, or not correlated to Sb resistance in different epidemiological studies (Decuypere et al., 2005, 2012; Mittal et al., 2007; Mukherjee et al., 2007; Mukhopadhyay et al., 2011). Such divergent results may be explained by the polyclonal structure of parasite field isolates and their geographic adaptation, with different resistance mechanisms being selected in genetically distinct



\*Strain used for the generation of the cosmid library; \*\*biological validation in field isolates

isolates (Decuypere et al., 2012). This possibility is supported by our recent demonstration that genetic mosaicism in an individual L. donovani strain can drive polyclonal adaptation, suggesting that different resistance mechanisms may co-exist in sub-populations of any given isolate (Prieto Barja et al., 2017). Such intra-strain specific, polyclonal fitness gain is further supported by the cosmid selection of different genetic loci in response to the same selection pressure applied on a single parasite population in vitro or during animal infection (Cotrim et al., 1999; Dahlin-Laborde et al., 2005; Gazanion et al., 2016). Indeed, such clonal phenotypic variability in a given parasite isolate has been recently documented in L. amazonensis, with important differences in culture proliferation and pathogenic potential observed in untransfected sub-clones or parasites transfected with individual cosmids selected in vivo for increased parasite infectivity (Espiau et al., 2017). Finally, other genes associated with drug resistance or susceptibility identified in cosmid screens failed to be validated in clinical studies such as LdMT, whose mutations were correlated to MIL resistance in promastigotes in culture but could not be associated with MIL resistance or treatment failure in the field (Bhandari et al., 2012). Likewise, PRP1 that has been implicated in vitro in resistance to pentamidine with reported cross-resistance to SbIII, did not show increased expression in Sb resistant field isolates (Decuypere et al., 2005, 2012).

Drawing from these examples we propose an experimental framework for the discovery of biomarker candidates by combining functional genetic screens in relevant animal models to reveal loci of interest, which then are validated by dedicated clinical and epidemiological investigations (**Figure 1**). In this approach, a cosmid library is prepared from parasites freshly derived from clinical isolates that show a phenotype of interest (donor strain). The gene(s) that express this phenotype are identified by transfecting a relevant recipient Leishmania strain and recovery of cosmids from transfectants that gained the phenotype under investigation. These genes can then be validated as biomarkers by quantitative PCR analysis directly applied on clinical samples. Thus, in contrast to classical biomarker discovery, where epidemiological field studies establish a correlation between a clinical phenotype and a genetic locus that then is validated in vitro or in animal studies, the epidemiological protocol we propose is in reverse

hamster infection (in the presence of drug in our example). The selected gene(s) of interested (GOI) will be identified by next generation sequencing (NGS). Correlating the identified genes with the clinical phenotype in dedicated epidemiological studies will then validate the new biomarker.

from lab-based studies back to the field. Even though this approach has its drawbacks (e.g., clinical manifestations caused by gene inactivation or gene deletion cannot be revealed), it provides several interesting advantages that immediately overcome important bottlenecks in Leishmania biomarker discovery. First, it is independent of long-term culture that can have an important impact on the parasite genome thus interfering with epidemiologically relevant information. Second, the screening is performed in situ in infected animals under environmental constraints that correspond to the clinical setting, thus allowing for the selection of physiologically highly relevant loci. Third, large amounts of parasite can be recovered from different tissues of the infected animals, which can be subjected to direct and even single cell sequencing, thus informing on mechanisms of polyclonal adaptation that may be relevant to the field. Finally, this approach will overcome ethical concerns associated with applying direct genome sequencing on human tissue samples as the cosmid-identified loci will be studied in clinical samples by simple qPCR analysis.

In conclusion, our reverse epidemiology approach exploits genetic amplification for biomarker discovery and thus mimics the very mechanism that has been linked to Leishmania genomic adaptation and fitness gain in the field and in culture (Dumetz et al., 2017; Prieto Barja et al., 2017). Cosmid-based functional genetic screening in situ linked to clinical validation thus represents a powerful framework that can fill an important gap in the currently rather desolate state of Leishmania biomarker discovery, which is challenged by the absence of robust protocols for direct tissue sequencing of parasites in human clinical samples, and the genetic bias caused by parasite long-term culture applied in current epidemiological investigations.

# REFERENCES


# AUTHOR CONTRIBUTIONS

LP wrote the chapter on the use of cosmid libraries regarding virulence and tropism, PP wrote the chapter on the use of cosmid libraries regarding drug resistance, GS corrected the manuscript and wrote introduction and the last chapter detailing the experimental framework.

# FUNDING

This work was supported by funds from the Institute Pasteur International Direction strategic fund to the LeiSHield project and the Laboratoire d'Excellence Integrative Biology of Emerging Infectious Diseases (Grant no. ANR-10-LABX-62-IBEID). LP was supported by the Ph.D. fellowship from the Ministère de l'Enseignement Supérieur, de la Recherche et de l'Innovation.


sequencing in Leishmania. Int. J. Parasitol. Drugs Drug Resist. 5, 26–35. doi: 10.1016/j.ijpddr.2014.09.005


**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 Piel, Pescher and Späth. 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.

# Transforming Growth Factor Beta (TGFβ1) and Epidermal Growth Factor (EGF) as Biomarkers of *Leishmania (V) braziliensis* Infection and Early Therapeutic Response in Cutaneous Leishmaniasis: Studies in Hamsters

Andrés Montoya<sup>1</sup> , Lina Yepes <sup>1</sup> , Alexander Bedoya<sup>1</sup> , Raúl Henao<sup>1</sup> , Gabriela Delgado<sup>2</sup> , Iván D. Vélez <sup>1</sup> and Sara M. Robledo<sup>1</sup> \*

<sup>1</sup> Programa de Estudio y Control de Enfermedades Tropicales (PECET), Facultad de Medicina, Universidad de Antioquia, Medellín, Colombia, <sup>2</sup> Grupo de Investigación en Inmunotoxicología, Departamento de Farmacia, Facultad de Ciencias, Universidad Nacional de Colombia, Bogotá, Colombia

### *Edited by:*

Javier Moreno, Instituto de Salud Carlos III, Spain

### *Reviewed by:*

Herbert Leonel de Matos Guedes, Universidade Federal do Rio de Janeiro, Brazil Ana Victoria Ibarra-Meneses, Instituto de Salud Carlos III, Spain

> *\*Correspondence:* Sara M. Robledo sara.robledo@udea.edu.co

### *Specialty section:*

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

*Received:* 26 March 2018 *Accepted:* 13 September 2018 *Published:* 02 October 2018

### *Citation:*

Montoya A, Yepes L, Bedoya A, Henao R, Delgado G, Vélez ID and Robledo SM (2018) Transforming Growth Factor Beta (TGFβ1) and Epidermal Growth Factor (EGF) as Biomarkers of Leishmania (V) braziliensis Infection and Early Therapeutic Response in Cutaneous Leishmaniasis: Studies in Hamsters. Front. Cell. Infect. Microbiol. 8:350. doi: 10.3389/fcimb.2018.00350 Introduction: In cutaneous leishmaniasis, the host immune response is responsible for the development of skin injuries but also for resolution of the disease especially after antileishmanial therapy. The immune factors that participate in the regulation of inflammation, remodeling of the extracellular matrix, cell proliferation and differentiation may constitute biomarkers of diseases or response to treatment. In this work, we analyzed the production of the growth factors EGF, TGFβ1, PDGF, and FGF during the infection by Leishmania parasites, the development of the injuries and the early response to treatment.

Methodology: Golden hamsters were infected with L. (V) braziliensis. The growth factors were detected in skin scrapings and biopsies every 2 weeks after infected and then at day 7 of treatment with different drug candidates by RT-qPCR. The parasitic load was also quantified by RT-qPCR in skin biopsies sampled at the end of the study.

Results: The infection by L. (V) braziliensis induced the expression of all the growth factors at day 15 of infection. One month after infection, EGF and TGFβ1 were expressed in all hamsters with inverse ratio. While the EGF and FGF levels decreased between day 15 and 30 of infection, the TGFβ1 increased and the PGDF levels did not change. The relative expression of EGF and TGFβ1 increased notably after treatment. However, the increase of EGF was associated with clinical cure while the increase of TGFβ1 was associated with failure to treatment. The amount of parasites in the cutaneous lesion at the end of the study decreased according to the clinical outcome, being lower in the group of cured hamsters and higher in the group of hamsters that had a failure to the treatment.

Conclusions: A differential profile of growth factor expression occurred during the infection and response to treatment. Higher induction of TGFβ1 was associated with active disease while the higher levels of EGF are associated with adequate response to treatment. The inversely EGF/TGFβ1 ratio may be an effective biomarker to identify establishment of Leishmania infection and early therapeutic response, respectively. However, further studies are needed to validate the utility of the proposed biomarkers in field conditions.

Keywords: growth factor, cutaneous leishmaniasis, *L. braziliensis*, EGF, TGFβI, FGF, PDGF, biomarkers

# INTRODUCTION

Leishmaniasis is a parasitic disease caused by protozoa of the genus Leishmania spp. which are transmitted by phlebotomies insects (Bates, 2007; Sharma and Singh, 2008). The disease is widely distributed in five continents and is endemic in lowincome countries in tropical and subtropical regions (Alvar et al., 2012). The disease manifests itself in three main clinical forms known as cutaneous, mucosal, and visceral leishmaniasis. Mucosal and cutaneous leishmaniasis (ML and CL, respectively) are characterized by the presence of granulomatous lesions in the mucous and dermis, respectively (Herwaldt, 1999; Handler et al., 2015; Kevric et al., 2015), that may cause disfigurement accompanied by social stigmatization and psychological disorders, altering the economic well-being of patients (Weiss, 2008; Okwor and Uzonna, 2016). There are very few options to treat CL and ML and all have numerous disadvantages that evidence the urgent need to develop new and better therapeutic alternatives. It is known that CL is manifested as skin lesions that result from the exacerbated activation of the immune system through the recruitment of abundant monocytes, neutrophils, NK cells, and CD8+ and CD4+ T lymphocytes (Brewig et al., 2009; Hrdinka et al., 2011; Nylén and Eidsmo, 2012; Soong et al., 2012). The activation of all these cells induces a chronic inflammatory response that leads to the necrosis of the tissue and therefore to the skin damage and the appearance of ulcers (in most cases). This cutaneous lesion usually resolves after specific therapy that let the elimination of the antigenic stimulus and therefore, the resolution of the inflammatory response and repair of the damaged tissue (McGwire and Satoskar, 2014; Copeland and Aronson, 2015; de Menezes et al., 2015; Aronson et al., 2016).

Although the immune response to Leishmania has been described during infection and disease, this immune response during and after treatment has been poorly described. To date, no information is available about the production of dermal and epidermal growth factors during Leishmania infection or concerning the healing of lesions caused by Leishmania spp. Studies carried out with other types of ulcers reported that during healing, the epidermal tissue is formed after transformation of the dermis into fibrous tissue (scarring) accompanied by an increase in NK cells, a decrease in circulating T cells with depletion of CD8+ T cells, and a decrease of CD4+ T cells with a decrease in the production of IFNγ (Lakhal-Naouar et al., 2015) and expression of dermal and epidermal growth factors (Kiwanuka et al., 2012).

Growth factors are polypeptides that stimulate cell proliferation, migration, and differentiation as well as survival and apoptosis (Bennett and Schultz, 1993). The most important growth factors in this process of tissue repair are fibroblast growth factor (FGF), platelet derived growth factor (PDGF), transforming growth factor beta 1 (TGFβ1) and alpha (TGFα), epidermal growth factor (EGF), connective tissue growth factor (CTGF), and vascular endothelial growth factor (VEGF; (Robson, 1991; Steed, 1997)). The source and function of these growth factors are summarized in **Table 1**. While some observations suggest an important role for TGFβ during parasite establishment in the early stages of human CL (Barral et al., 1995) and treatment failure in patients with visceral leishmaniasis (Elmekki et al., 2016) there seem to be no reports of studies for the other growth factors in the context of leishmaniasis.

Based on the important role of growth factors during healing and tissue repair, this study aimed to identify the expression levels of EGF, FGF, PDGF, and TGFβ1 during: (i) infection by Leishmania (V) braziliensis, (ii) development of CL, and (iii) early response to treatment using the experimental model for CL in the golden hamster (Mesocricetus auratus). An useful RT-qPCR method to quantify EGF, FGF, PDGF, and TGFβ1 expression was also standardized.

# MATERIALS AND METHODS

## Compounds

Propantheline bromide (CAS #50-34-0), limonin (CAS #1180- 71-8), nomilin (CAS #1063-77-0), azadirachtin (CAS #11141- 17-6), cryptolepine hydrate (CAS #480-26-2), glycyrrhizin (CAS #1405-86-3), and oleanolic acid (CAS #508-02-1) were purchased at Sigma Aldrich (St Louis, MO, USA) and were used as active pharmaceutical ingredients (API) for new formulations. Meglumine antimoniate (MA) (Sanofi-Aventis Bogotá, Colombia) was used as a reference antileishmanial drug.

# Parasites

Green fluorescent protein (GFP)-transfected L. (V) braziliensis promastigotes (MHOM/CO/88/UA301-EGFP) were maintained in biphasic Novy-MacNeal-Nicholle (NNN) medium phosphate buffer saline (PBS) enriched with glucose at pH 6.9.

# Formulations

An oil-in-water (O/W) semisolid emulsion was prepared which served as the matrix in which the API were embedded. The designed formulations were as follows: 0.5% propantheline bromide, 2.4% limonin, 2.2% nomilin, 1.1% azadirachtin, 1.4% cryptolepine hydrate, 5% glycyrrhizin, and 2.5% oleanolic acid. The concentration of the API was selected at convenience, according to the availability of the compound and previous results of the antileishmanial activity obtained in vitro.


<sup>a</sup>FGF, fibroblast growth factor; PDGF, platelet derived growth factor; TGFβ1, transforming growth factor beta 1; EGF, epidermal growth factor; CTGF, connective tissue growth factor; TGFα, transforming growth factor alpha; VEGF, vascular endothelial growth factor.

# Animals

Eighty-six week-old hamsters, male and female, with an average weight of 180 g (160–200 g) were used. Hamsters were maintained in specific pathogenic-free conditions in an animal facility at Universidad de Antioquia (Medellín, Colombia), housed in transparent cages in groups of three or four animals per cage, during the study.

# Experimental Infection

Previous anesthesia with a mixture of ketamine (40 mg/kg) and xylazine (5 mg/kg), hamsters were injected in the dorsum with 5 × 10<sup>8</sup> promastigotes of L. braziliensis in the stationary phase of growth (day 6 in culture) in 100 µl PBS. After 15 and 30 days post-infection (ID15 or ID30, respectively), one group of eight hamsters each were humanely sacrificed. Skin biopsies from the inoculation site were obtained and stored in RNA later for the RNA extraction. The remaining hamsters were kept under observation until the development of ulcers. Skin biopsies of four uninfected hamsters were obtained and they corresponded to biopsies obtained before infection day (ID0).

# Therapeutic Response

Four weeks after the infection, ulcers were formed, and treatment with each of the formulations had started. For this, the hamsters were randomly distributed in eight groups with eight hamsters each. Each group was treated topically during 20 days with 40 mg/day of one of the following formulations: 0.5% propantheline bromide, 2.4% limonin; 2.2% nomilin, 1.1% azadirachtin, 1.4% cryptolepine hydrate, 5% glycyrrhizin, and 2.5% oleanolic acid; or intramuscular MA (120 mg/kg/10 days). The hamsters were monitored daily during the study recording any change in behavior or death. The area of the injuries was measured before treatment (TD0) and 90 days after completion of the treatment (PTD90), which corresponded to the end of the study. The effectiveness of each treatment was determined according to the area of the lesion at PTD90 with respect to the size of the lesion before treatment. The response to the treatment was classified as cure (if healing of 100% of the area of the lesion did occur) or improvement (if the area of injury decreased more than 20%). On the contrary, the outcome was classified as failure (if the area of injury increased) or as relapse (if reactivation of the injury appeared after an initial cure). At the end of the study, hamsters were sacrificed humanely in a CO<sup>2</sup> chamber and necropsied. Skin samples for parasite load analysis were obtained. Prior to treatment (TD0) and on day 7 of treatment (TD7), scraping samples were taken from the lesions and deposited in RNA later (for RNA extraction).

# RNA Extraction

The RNA from the skin samples was extracted using Trizol <sup>R</sup> (Invitrogen) following the manufacturer's instructions and then quantified in a Nanodrop 1000 (Thermo scientific).

# Retrotranscription

One hundred grams of RNA were treated with 1 µl DNase I (Fermentas), 1 µl of buffer, and 8 µl of nuclease free water; the mixture was incubated in a thermocycler PTC 100TM (MJ research) for three cycles: 30 min at 37◦C, 5 min at 4◦C, and 10 min at 65◦C. Using the maximum first strand cDNA synthesis kit, the RNA was transcribed into cDNA, following the manufacturer's instructions: 4 µl master mix reaction, 2 µl RT enzyme mix, 2 µl RNA DNase I treated, and 12 µl water were mixed and incubated in the PTC 100TM (MJ research) thermocycler for three cycles: 10 min at 25◦C, 15 min at 50◦C, and 5 min at 85◦C.

# Growth Factor Expression

Specific primers for EGF, PDGF, FGF, and TGFβ1, as well as fluorescent FAM-labeled hydrolysis probes, were designed. The sequences for each forward (Fw) and reverse (Rv) primer and probe used were as follows:

FGF, Fw, 5′ -GTGTCAAGGCTGCTAGGTTT-3′ , Rv, 5′ -ACA CATTGTATCCATCCTCAA-3′ and probe 5′ -TCGCCTCAC TTCGATCCCG-3′ ; EGF, Fw, 5′ -CAGAACAAAGCCAGA AAATC-3′ , Rv, 5′ -CTGCAAGTACGTTCGTTTAACT-3′ and probe 5′ -AGACTCGCGTTGCAAGGCG-3′ ; PDGF, Fw, 5′ -GGC TCGAAGTCAGATCCATA-3′ , Rv, 5′ -CTTCTCCGGCACATG CTTAA-3′ and probe 5′ -TGGAGACAAGCCTGAGAGCC-3′ ; TGFβ1, Fw, 5′ -AGCCTGGACACACAGTACAGTA-3′ , Rv, 5′ -CTTGCGACCCACGTAGTAC-3′ and probe 5′ -AACACAACC CGGGTGCTTC-3′ ; γActin, Fw, 5′ -ACAGAGAGAAGATGA CGCA-3′ , Rv, 5′ -GCCTGAATGGCCACGTAC-3′ and probe 5′ -TTGAAACCTTCAAATGACGCA-3′ .

The reaction of amplification was carried out with 1 µl of the cDNA and using the following protocol: 600 s at 95◦C, 40 cycles of 15 s at 95◦C, and 60 s at 60◦C in a Smart Cycler II (Cepheid, Sunnyvale, CA, USA). The efficiency of the amplification reaction was determined using the LinReg program and the expression levels were calculated using the 11CT method, comparing the levels of expression for PI15, PI30 with respect to healthy skin before infection and TD7 with respect to TD0. Additionally, the levels of induction were compared in terms of clinical outcome, in terms of cure, improvement, relapse, or failure.

# Parasite Load

The parasitic load on the skin biopsy samples taken at the end of the study (PRD90) was determined by RT-qPCR using the Vero 1-step RT-qPCR SYBR Green kit, a 123 bp fragment of the DNA polymerase I gene from Leishmania was amplified. To do this, 20 ng of RNA, 12.5 µl the mix, and 100 ng of the each primer Fw 5′ -TGAGCGCATCGAGTACCT-3′ and Rv 5′ - TCCCGCTTGCCATCCTC-3′ , with a volume adjusted to 25 µl using nuclease-free water, were used to carry out the reaction with the Smart Cycler II (Cepheid, Sunnyvale, CA, USA): 50◦C 15 min, 95◦C 15 min and 40 cycles 95◦C 15 s, 60◦C 20 s and 72◦C 20 s, a final cycle 72◦C 300 s and a melting curve between 60 and 95◦C. Absolute quantification was performed using a standard logarithmic scale from 1 to 1 million parasites.

# Ethical Aspects

All the procedures were approved by the Ethics Committee for Animal Experimentation of the Universidad de Antioquia (Act No. 91 of 2014).

# Statistical Analysis

The differences between the parasitic loads according to the clinical results were determined by a one-way ANOVA and a Tukey's test for multiple comparisons. The differences in the expression of growth factors between ID15 and ID30 vs. ID0 and between TD0 and TD7 were determined by two-way ANOVA and Bonferroni's test for multiple comparisons. In addition, differences in expression levels according to the clinical results were also obtained by two-way ANOVA and Bonferroni's test.

# RESULTS

# Lesion Progression in Hamsters

After intradermal injection of promastigotes of L. (V) braziliensis promastigotes in dorsum, hamsters were monitored weekly for any sign of induration or skin damage at the site of inoculation. The diameter of the injury increased in size every week. In this way, the average size of the injuries was 1,29 ± 3.29 mm<sup>2</sup> at ID8, 3.35 ± 7.92 mm<sup>2</sup> at ID15, 12.75 ± 19 mm<sup>2</sup> at ID21, and 39.93 ± 40.73 mm<sup>2</sup> at ID30. The lesion on the skin was evidenced first as a nodule and then as an ulcer. The induration varied from mild in the second week post-infection to exacerbated in the 4 week after inoculation (**Figure 1**).

# Expression of PDGF, EGF, FGF, and TGFβ1 Associated With the Process of Infection by *L. (V) braziliensis* in the Hamster Model for Cutaneous Leishmaniasis

The expression levels of PDGF, EGF, FGF, and TGFβ1 were calculated by the 11CT method and expressed as the mean ± SD of the number of times that each factor was induced with respect to healthy skin before infection (ID0), during the development of the lesion (ID15 and ID30). Expression levels lower than the constitutive gene were found for the PDGF, EGF, TGFβ1, and FGF before infection, and these basal levels were used for the calculation of 11CT. Infection by L. (V) braziliensis induced the expression of all the growth factors in ID15, with levels of induction with respect to healthy hamsters of: PDGF 0.4 ± 0.1, EGF 3.9 ± 0.2, FGF 1.1 ± 0.5, and TGFβ1 0.4 ± 0.2. In ID30, PDGF was expressed in 3/8 hamsters with an expression value of 0.6 ± 0.1; EGF and TGFβ1 were expressed in all hamsters with an expression value of 1.5 ± 0.7 and 1.6 ± 0.5, respectively. Lastly, FGF was expressed in 2/8 hamsters and its expression value was 0.2 ± 0.0. The differences in EGF and TGFβ1 expression levels between ID15 and ID30 were statistically significant with p < 0.0001 and <0.0442, respectively.

On the contrary, the differences between PDGF and FGF were not significant (**Figure 2**).

# Expression of PDGF, EGF, FGF, and TGFβ1 in Response to Treatment

Of 64 hamsters treated with the different compounds and followed up for 3 months, 10 hamsters (15.6%) cured, 24 hamsters (37.5%) showed improvement of their ulcers, 27 hamsters (42.2%) failed, and 3 hamsters (4.6%) relapsed. The appearance of the lesions according to the clinical outcomes after treatment are shown in **Figure 3**. The levels of expression of PDGF, EGF, FGF, and TGFβ1 at TD7 were compared with respect to the levels before treatment (TD0) using the 11CT method. In the cured hamsters, the expression levels increased 0.9 ± 0.1 times for PDGF, 46.9 ± 34.2 for EGF, 0.7 ± 0.2 for FGF, and 0.8 ± 0.7 for TGFβ1 (**Figure 4**). In the hamsters with improvement of their ulcers, the expression levels increased 0.011 ± 0.017 times for PDGF, 17.5 ± 2.4, for EGF, 0.01 ± 0.03 for FGF, and 2.1 ± 0.1 for TGFβ1. In turn, in the hamster that relapsed, the levels for PDGF were 0.012 ± 0.01, 9.5 ± 8.8 for EGF, 0.8 ± 0.7 for TGFβ1, and FGF was not detected. Finally, in the hamsters that failed to the treatment, the expression levels for EGF and TGFβ1 increased 8.9 ± 2.1 and 3.1 ± 0.6 times, respectively, while PDGF and FGF were not detected. Significant differences in expression were only observed in EGF expression between cured hamsters vs. TD0 (p < 0.0001), improvement vs. TD0 (p < 0.0413), cured vs. improvement (p < 0.0003), cured vs. failure (p < 0.0001), and cured vs. relapsed (p <0.0001) (**Figure 4**).

Only EGF and TGFβ1 showed expression levels >1. When compared the proportion of EGF vs. TGFβ1 we found that the EGF was expressed 58.6 times more than the TGFβ1 in cure and 8.3 in improvement while in relapses and fails the proportion was 11.8 and 2.9, respectively. Although the EGF increases during treatment, the increase is lower in relation to a poor therapeutic

FIGURE 2 | Relative expression of growth factors during infection and development of injury. The figure shows the relative expression at ID1 (n = 8) and ID30 (n = 8) vs. before infection (ID0) calculated by the 11CT method. PDGF (black), EGF (white), FGF (pattern), and TGFβ1 (lines). \*\*\*\*p < 0.0001; \*p < 0.0442.

response. Thus, for example, the EGF was expressed 2.68 more times in cure vs. improvement, 4.93 times more in cure vs. relapse and 5.26 times more in cure vs. failure. In a similar way, the TGFβ1 was expressed 3.9 times more in failure vs. cure, 2.6 times more in improvement vs. cure.

FIGURE 3 | Treatment progress of cutaneous leishmaniasis in hamsters experimentally infected with L. (V) braziliensis*.* A representative photograph of lesion, (A) before treatment, (B) end of treatment, and (C) 90 days post-treatment. Note the complete re-epithelialization of the skin in cured hamsters, the decrease in size of the injury during improvement or the increases in the size of the injury during fail to treatment.

represent the mean value ± SD. PDGF (black), EGF (white), FGF (square), and TGFβ1 (lines). \*p < 0.0413; \*\*\*p < 0,0003; \*\*\*\*p < 0.0001, cured (n = 10), improvement (n = 24), fail (n = 27), relapse (n = 3).

The effect of each treatment in the expression of each growth factor in terms of increasing, reducing, or no changes with respect to the clinical outcome is shown in **Table 2**. The treatment with meglumine antimoniate and propantheline bromide, both drugs associated with cure of CL in hamster, showed the major increases in the expression of EGF.

# Parasite Load During Infection by *L. (V) braziliensis* and Treatment in the Golden Hamster (*Mesocricetus auratus*)

The number of parasites in the injury during infection (ID15 and ID30) and at the end of the study (PTD90) was measured by RT-qPCR and was expressed as parasites per mg of tissue. The parasite load in the 10 cured hamsters was 76.8 ± 115.8, while in the 24 animals with clinical improvement the parasite load was 238.0 ± 57.0. In turn, in the 27 animals classed as failures, the parasite load was 710.4 ± 484.5, while in the 3 hamsters that relapsed the parasitic load was 3478.0 ± 388.6. Differences were statistically significant between cured vs. fail, cured vs. relapsed, improvement vs. failure, improvement vs. relapse, and failure vs. relapsed. In all cases, the p value was < 0.0001 (**Figure 5**).

# DISCUSSION

For several decades, there have only been five drugs used for the treatment of leishmaniasis. Although these are still effective, they are far from ideal due to factors such as costs, duration of treatment, toxicity, and the appearance of resistant strains (Oliveira et al., 2011; Monge-Maillo and López-Vélez, 2013; Wolf Nassif et al., 2017). The intensive work done in the last 15 years in the search for alternative treatments of leishmaniasis has allowed for the identification of numerous candidate molecules or compounds. Some of these have advanced in their development until the stage of clinical evaluation. However, only a few are considered as real candidates for the development of medicines to date. Many candidates fail because of the poor effectiveness demonstrated in phase II clinical trials (Singh et al., 2012). The extremely high cost of clinical trials is a bottleneck in the development of new drugs; therefore, it is necessary to identify markers that allow for early identification of the response to the treatment that is being evaluated (Frank and Hargreaves, 2003; Zwierzina, 2008).

Although multiple studies have analyzed susceptibility and resistance genes associated to the development of the lesions (Castellucci et al., 2012; Sohrabi et al., 2013; Abdoli et al., 2017), the role of EGF, FGF, and PDGF in the establishment of the infection, development of the ulcer, and in early response to treatment has not been established. In contrast, although the TGFβ1 has been related to susceptibility and chronicity of the infection (Balak et al., 1992; Barral-Netto and Barral, 1994; Barral et al., 1995; Bogdan and Röllinghoff, 1998; Nieto et al., 2011; Hejazi et al., 2012; Rodrigues et al., 2014; Bhattacharya et al., 2016), its role has not been described in the context of early therapeutic response. In this work, we described the dynamics of the production of EGF, FGF, PDGF, and TGFβ from initial infection and manifestation of skin lesions to the early stages of response to different treatment compounds. We demonstrated that the expression of PDGF, EGF, FGF, and TGFβ1 was induced at ID15 during the establishment of the experimental infection by L. (V) braziliensis. This result suggests that, during infection, the cells are activated to control the tissue damage as a result of inflammation (Mast and Schultz, 1996). At ID30 when the infection is already established and the ulcer is manifested, the expression of growth factors changes drastically, not only in the profile of factors, but also in the level of induction (Mast and Schultz, 1996). FGF and PDGF were no longer expressed in more than half of the animals, but EGF and TGFβ1 showed an interesting change reversing their induction levels by increasing TGFβ1 and decreasing EGF.

The decreased expression of PDGF and FGF correlates with the appearance of lesions by ID30. It has been confirmed in the hamster model for CL that infection by Leishmania generates the activation of the Nlrp3 inflammasome with the subsequent production of IL1β (Lima-Junior et al., 2013), and this cytokine may negatively regulate the expression of PDGF (Barrientos et al., 2008; Mundy, 2009). In the case of FGF and other growth factors, such as TGFα, KGF, and VEGF, it has been shown in the experimental CL hamster model that, in some types of ulcers induced by drugs or anti-tumor therapies, the decrease in proinflammatory cytokines such as IL1β and TNFα positively regulates the expression of these factors (Araújo et al., 2015). When the inflammatory process perpetuates, it has a negative regulatory effect on these two growth factors, PDGF and FGF, in some cases, halting expression and in other cases, reducing it. Another explanation for the reduction in the expression of these growth factors is that the macrophages, keratinocytes, endothelial cells, and fibroblasts involved in the production of growth factors in the infected and inflamed tissue are in a context where


The effect of each treatment on the expression of each factor was determined according to the changes in the expression level as increase (+), no change (=), or not effect (–). C, cure; I, improvement; F, failure; R, Relapse. +, <1 increase fold; ++, >1 y <5 increase fold; + + ++, >40 increase fold, n = 8, each group.

their functions of survival, proliferation, cell differentiation, and angiogenesis have been regulated by the need to eliminate the parasite.

Interestingly, TGFβ1 tripled its expression in 8/8 hamsters during infection. In addition, other studies have associated the early expression of this factor and its increased expression levels with the regulation of the immune response, allowing for the establishment of infection and development of ulcers (Balak et al., 1992; Barral-Netto and Barral, 1994; Barral et al., 1995; Mougneau et al., 2011; Hejazi et al., 2012; Nylén and Eidsmo, 2012; Rodrigues et al., 2014; Bhattacharya et al., 2016), since TGFβ1 has the capacity to inhibit the Th1 response affecting the production of IFNγ, IL12, and key cytokines in the control and elimination of the parasite (Reed, 1999). On the other hand, the expression of EGF decreased with respect to ID15, reducing by half in 8/8 animals. This growth factor is very important for the proliferation of keratinocytes, which are essential for the establishment of a Th1 immune response that favors the control of the infection (Steed, 1997; Eming et al., 2007; Kiwanuka et al., 2012; Pikuła et al., 2015). Its increase at ID15 is likely to be an attempt to favor the Th1 response. However, when the immune response is regulated by TGFβ1 and the infection is effectively established at ID30, its expression is negatively regulated, due to the fact that the macrophages and fibroblasts, as its main sources, are in a context in which the production of EGF is diminished.

When analyzing the early expression of growth factors as biomarkers for the clinical outcome of our experimental model, we initially found differences in the expression profile of these factors. Thus, in the 3 animals that relapsed after treatment, the EGF was expressed at higher levels that TGFβ1; nevertheless, since it is only a group of three hamsters, it limits the comparisons with the other groups. In turn, in hamsters in which the treatment failed, the expressed factor profile involved EGF, and TGFβ1. In turn, in the cured hamsters and those that showed improvement of their injuries, all growth factors, even FGF, were expressed.

When comparing the production profile of growth factors in fetal and adult fibroblasts, differences have been correlated with the healing and the re-epithelialization of injuries and a better appearance of the scars, which explains why fetal scarring leaves scarcely imperceptible scars (Broker et al., 1999). However, the presence ofseveral factors is necessary in order to repair damaged tissue efficiently (Papanas and Maltezos, 2007; Buchberger et al., 2010). Some clinical trials have evaluated the use of one or several growth factors such as PDGF, EGF, and FGF for the treatment of chronic ulcers, showing good results when used individually but with better results when used in combination (Robson, 1991; Mast and Schultz, 1996; Singla et al., 2012).

On the other hand, the profile of the growth factors in our model allowed us to predict a positive clinical result of the treatment at an early stage, as well as the levels of induction of the different growth factors. In the case of FGF and PDGF, since their expression is not constant in all hamsters, it is difficult to conclude that the levels of induction of these factors are a good biomarker for early therapeutic response. However, the benefit of knowing the behavior of the expression of these growth factors and that their early induction depends on the treatment will allow for future evaluations in our animal model to identify candidates for medications that possess not only antileishmanial potential but also scarring potential.

In the case of TGFβ1, its role in pathogenesis is well described. There is an increase in the levels of TGFβ1 in peritoneal macrophages infected with the parasites (Balak et al., 1992). In human biopsies from active lesions in both the cutaneous form and the mucosal form of the disease, the presence of this factor is also detected (Barral et al., 1995), which is related to one of the functions of TGFβ1 in the regulation of the immune response allowing for the parasite to replicate (Abdoli et al., 2017). TGFβ1 appears to be a candidate biomarker for pathogenesis and for a negative therapeutic response to treatment (failure and relapse). Although the differences in the levels of expression were not statistically significant, the tendency is to increase in production when there is failure or relapse, and decrease in production when curing or improvement occur.

EGF showed significant differences in the levels of induction on TD7 with respect to TD0 and difference in the expression with respect to the clinical result among the cured animals and all the other clinical outcomes. EGF plays a very important role in the process of cell proliferation, activation, and reepithelialization (Barrientos et al., 2008). It increases in acute wounds (Shen et al., 2017) and is found to decease in chronic wounds (Mast and Schultz, 1996). In the case of CL, we confirm low levels of EGF before treatment, because it is a chronic injury. However, by implementing a treatment that induces healing, we find that the level of EGF expression increases, in comparison to the treatments that produce improvement, failure, or relapse. Notoriously, propantheline bromide and MA, which were medications associated with healing, showed a similar profile. These results suggest that this profile can help predict a curative response during treatment follow-up.

This study focused on the search for biomarkers in leishmaniasis and allowed us to establish the role of multiple cytokines in the process of infection and in pathogenesis as well as in the candidates for biomarkers related to the immune response profile in visceral leishmaniasis (Portela et al., 2018). In the case of canine visceral leishmaniasis, a recent study identified some proteins as biomarkers of the therapeutic response in dogs. Nevertheless, none of these proteins were growth factors or were related to the activation of growth factors (Martinez-Subiela et al., 2017). This is most likely because the samples used

# REFERENCES


for monitoring the therapeutic response biomarkers were taken 1 month after the end of the treatment, when the parasite is likely to have been eliminated. In this case, the samples would not indicate the effectiveness of the treatment early but would confirm whether the treatment was effective. Our results allow us to implement in our experimental model a rapid measurement of the therapeutic response to treatment and define the optimal dose or frequency, thus obtaining better therapeutic success with the new candidates that are being tested in vivo for the treatment of CL.

# CONCLUSION

A differential profile of EGF, TGFβ1, PDGF, and FGF expression was observed during the establishment of L. (V) braziliensis infection and response to treatment in our experimental CL model. Higher induction of TGFβ1 are more associated with active disease (infection and failure or relapses after treatment) while the higher levels of EGF are associated with adequate response to treatment. Thus, the inversely EGF/TGFβ1 ratio may be an effective biomarker to identify establishment of Leishmania infection and early therapeutic response, respectively. However, further studies are needed to validate the utility of these growth factors as biomarkers in the pathogenesis of human CL and response to treatment under field conditions.

# AUTHOR CONTRIBUTIONS

AM participated in the design of the project, carried out the trials, analyzed the results, and participated in writing of the manuscript for publication. SR was in charge of the project and the analysis of results, and participated in writing of the manuscript. LY, AB, and RH helped in performing the experiments. GD and IV participated in the analysis of results and in the writing of the manuscript.

# ACKNOWLEDGMENTS

The authors would like to thank the Colombian Department of Science, Technology and Innovation - Colciencias (Grant CT-695-2014) for their financial support.

FGF, KGF, and TGF- α in an oral mucositis model. PLoS ONE 10:e0116799. doi: 10.1371/journal.pone.0116799


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

The reviewer AIM and handling editor declared their shared affiliation at time of review.

Copyright © 2018 Montoya, Yepes, Bedoya, Henao, Delgado, Vélez and Robledo. 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.

# Histological Disorganization of Spleen Compartments and Severe Visceral Leishmaniasis

Micely d'El-Rei Hermida, Caroline Vilas Boas de Melo, Isadora dos Santos Lima, Geraldo Gileno de Sá Oliveira and Washington L. C. dos-Santos\*

Fundação Oswaldo Cruz, Instituto Gonçalo Moniz, Salvador, Brazil

The spleen is a secondary lymphoid organ responsible for immune surveillance against blood-circulating pathogens. Absence of the spleen is associated with increased susceptibility to systemic spread and fatal infection by different pathogens. Severe forms of visceral leishmaniasis are associated with disorganization of spleen compartments where cell interactions essential for splenic immunological function take place. White pulp atrophies, secondary lymphoid follicles and marginal zones vanish, and the boundaries separating white and red pulp blur. Leukocyte populations are reduced or disappear or are replaced by plasma cells. In this paper, we review the published data on spleen disorganization in severe forms of visceral leishmaniasis and propose a histological classification to help the exchange of information among research groups.

### Edited by:

Javier Moreno, Instituto de Salud Carlos III, Spain

### Reviewed by:

Yang Zhang, University of Pennsylvania, United States Carla Maia, Universidade NOVA de Lisboa, Portugal

\*Correspondence: Washington L. C. dos-Santos wluis@bahia.fiocruz.br

### Specialty section:

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 14 June 2018 Accepted: 19 October 2018 Published: 13 November 2018

### Citation:

Hermida M-R, de Melo CVB, Lima IdS, Oliveira GGdS and dos-Santos WLC (2018) Histological Disorganization of Spleen Compartments and Severe Visceral Leishmaniasis. Front. Cell. Infect. Microbiol. 8:394. doi: 10.3389/fcimb.2018.00394 Keywords: visceral leishmaniasis, white pulp disruption, spleen disorganization, Leishmania infantum, spleen pathology

# INTRODUCTION

Visceral leishmaniasis (VL) is a severe parasitic disease caused by infection by Leishmania infantum (syn. Leishmania chagasi) or Leishmania donovani that affects both humans and dogs. Visceral leishmaniasis is distributed in Central and South America, Asia, parts of Africa and the Mediterranean basin, with an estimated burden of 2.1 million DALY (disability adjusted life years) (Townson et al., 2005). The infection may be silent or present with slight clinical manifestations (Badaró et al., 1986). However, some patients progress to a clear pattern of clinical disease with weight loss, hepatomegaly, splenomegaly, anemia, thrombocytopenia and leukopenia, including neutropenia, and increased susceptibility to bleeding and coinfections (Badaró et al., 1986; Costa et al., 2010). The current therapeutic regimes with antimonials or with amphotericin are effective in most of the cases of VL (Herwaldt, 1999). However, the disease maintains an approximately 7% lethality in Brazil, even among patients under treatment (Ministério da Saúde do Brasil, 2010).

The spleen, bone marrow, and liver are the main organs involved in VL. The spleen is affected in all cases of the disease. Furthermore, while some control of infection is observed in the liver, the infection acquires a progressive character in the spleen throughout the course of the disease (Wilson and Streit, 1996; Carrión et al., 2006), leading to disruption of white pulp (WP) structure and replacement of normal cellularity of the red pulp (RP) by plasma cells (Silva-O'Hare et al., 2016). Such structural disorganization of the spleen is associated with severe forms of VL (Veress et al., 1977; Lima et al., 2014). In this work, we review the published data on spleen disorganization in VL and the pathways involved in the process. Our aim is to draw attention to the spleen as a potential source of biological markers for identifying patients susceptible to progress to severe forms of VL. Early identification of these patients may contribute for designing more effective therapeutic strategies for progressive forms of the disease.

# THE NORMAL SPLEEN

The spleen is a large secondary lymphoid organ composed of two compartments: the RP and the WP, and present morphological variation among different species (**Figure 1**; Steiniger and Barth, 2000; Cesta, 2006). Cords, sinuses and blood vessels mainly comprise the RP, which contains lymphocytes macrophages, erythrocytes and a small number of plasma cells. The splenic RP perform hemocateresis and keeps strict control of iron stores, reducing their availability to circulating pathogens (Mebius and Kraal, 2005). The process involves the SLC11/Nramp (natural resistance-associated macrophage protein) family whose polymorphism is associated with susceptibility to a variety of pathogens (Wessling-Resnick, 2015). The spleen is the site of differentiation and homing of inflammatory macrophages, monocytes, granulocytes, dendritic cells, natural killer cells and short-lived plasma cells (Ellyard et al., 2005; Mebius and Kraal, 2005). In the WP, take place T- and B-cell differentiation and immune responses to blood-borne antigens (Mebius and Kraal, 2005). The WP is constituted by three regions: the periarteriolar lymphocyte sheath (PALS), lymphoid follicles and marginal zones (MZ). Layers of predominantly CD3<sup>+</sup> T lymphocytes surrounding segments of small arteries forms the PALS. Contiguous with the PALS, the lymphoid follicles emerge as sparse round aggregates of predominantly B lymphocytes. There is always a variety in the primary and secondary lymphoid follicles. Primary lymphoid follicles are small nodular aggregates of small lymphocytes (Beyer and Meyer-Hermann, 2008). Secondary lymphoid follicles are large lymphoid aggregates that present germinal centers (GC). Germinal centers are composed of large proliferating lymphocytes, large macrophages sometimes containing apoptotic bodies, some follicular dendritic cells and some T lymphocytes (MacLennan, 1994). A ring of small lymphocytes (mantle zone) surrounds the follicle GC (Brozman, 1985). A diffuse layer containing predominantly B lymphocytes, some T lymphocytes and various macrophage populations form the MZ, which is more evident around the lymphoid follicles. The splenic MZ is the homing site for memory B cells responsive to T-lymphocyte dependent and T-lymphocyte independent response antigens (Kraal, 1992; Lopes-Carvalho et al., 2005).

The maintenance and organization of splenic compartments are controlled by a complex signaling network of chemokines (mostly of CC and CXC family), cytokines and adhesion molecules (den Haan et al., 2012). Lymphoid follicle architecture is dependent on CXCL13 secretion by stromal and follicular dendritic cells (Shi et al., 2001). CXCL13 interacts with the CXCR5 receptor on B cells recruiting these lymphocytes into the lymphoid follicle (Ansel et al., 2000). Lymphotoxin-α1β2 (LT-α1β2), produced by stromal cells, plays a crucial role in the early organization of the spleen (Fu and Chaplin, 1999; Mebius and Kraal, 2005). CCL21 is involved in the recruitment and retention of T lymphocytes in the PALS (Förster et al., 1999; Gunn et al., 1999). CXCL12 is responsible for plasma cell retention in the RP (Hargreaves et al., 2001). B-cell subtypes respond to antigens in a T-cell dependent (TD) manner. They enter the lymphoid follicle, interact with CD4<sup>+</sup> T cells that express CD40 ligand in the germinal centers and differentiate into IgM-, IgG- or IgA-producing plasma cells with a high-affinity to antigens. Conversely, other splenic MZ B cells may be stimulated by B cell-activating factor (BAFF) and a proliferation-inducing ligand (APRIL) in a T-cell independent (TI) response, which plays an important role against microbial antigens (Bernasconi et al., 2002; Tsuji et al., 2008; Cerutti et al., 2011). The two pathways (TI and TD) are complementary to provide a more specific and faster diversified immune response (Cerutti et al., 2011; Grant et al., 2012).

# SPLEEN IN NON-INFECTIOUS CHRONIC DISEASES

Splenic structure and function are affected in the course of many chronic diseases. Long-lasting heart failure or impairment of liver circulation may lead to spleen congestion and stromal cell hyperplasia (Pereira et al., 2002). Hemoglobinopathies frequently course with splenic alterations (Tincani et al., 1997; O'Reilly, 1998). For example, in patients with sickle cell disease, the RP appears enlarged, with high numbers of lymphocytes and nucleated red blood cells (Szczepanek et al., 2012). Erythrocyte clumping, blood vessel obstruction and infarcts may lead to fibrosis and autosplenectomy (Al-Salem, 2011). Splenomegaly has been reported in approximately 9–41% of hepatic diseases, 4–10% of congestive or inflammatory diseases, 16–44% of lymphomas (Pozo et al., 2009) and 3% of the cases of sarcoidosis (Judson, 2007). Lymphoid follicle hyperplasia is found in systemic lupus erythematous (SLE) and other autoimmune diseases (Auerbach et al., 2013). In SLE, polyclonal B-cell activation results in an increased number of immunoblasts, plasmacytoid lymphocytes and plasma cells in the RP (Mok and Lau, 2003). Spleen arterioles develop a hyperplasic onionskin aspect, composed of multiple layers of fibrosis and smooth muscle cell proliferation (Kitamura et al., 1985). In a series of SLE cases, extramedullary hematopoiesis was observed (Auerbach et al., 2013). In late stage diseases, lymphoid atrophy may follow (Li et al., 2013).

# THE SPLEEN AND INFECTION

The spleen plays a central role in defense against circulating pathogens. Absence of the spleen predisposes to devastating consequences with the dissemination of infections by viruses, bacteria and fungi (Hansen and Singer, 2001). Changes in spleen structure are common in many systemic infections caused by viruses, bacteria and parasites (Andrade et al., 1990; Kyaw et al., 2006). Some of these infections progress with lymphoid or stromal splenic cell hyperplasia, sometimes followed by lymphoid atrophy and disorganization of spleen compartments (Abreu et al., 2001; Sonne et al., 2009; Schneider et al., 2010; Glatman Zaretsky et al., 2012; Dkhil et al., 2014; Djokic et al., 2018).

Viral infections such as Mononucleosis cause splenomegaly with mononuclear cell proliferation and atypical lymphocytes (Daneshbod and Liao, 1973; Thomas et al., 1990; Won and Ethell, 2012). In patients with AIDS, there is a progressive destruction of

FDCs and concomitant germinal center loss (Fox and Cottler-Fox, 1992). Parvovirus infection in dogs causes splenomegaly, with lymphoid follicle hyperplasia, bleeding foci and congestion in the splenic parenchyma (Oliveira et al., 2009). In the course of bacterial infection acute splenitis, septic emboli originating, leading to infarcts and abscesses may occur (Mocchegiani and Nataloni, 2009; Wang et al., 2009). Infection with hemoparasites such as Plasmodium, Ehrlichia, Babesia, Toxoplasma gondii are all associated with splenomegaly, lymphoid tissue hyperplasia and eventually to disruption of WP structure. Schistosoma mansoni infection causes congestive splenomegaly and decreased number of lymphoid follicles and blurred marginal zones (Andrade and Bina, 1983; Brandt et al., 2005; Silva et al., 2012; Wang et al., 2015; Yan-Juan et al., 2017).

# STRUCTURAL DISORGANIZATION OF THE SPLEEN COMPARTMENTS

Most of the studies concerning the morphological changes of the spleen occurring in the course of infectious and non-infectious diseases have emphasized the quantitative aspect of lymphoid or stromal hyperplasia or atrophy. Only a few authors have drawn attention to the association of the morphological changes with the redistribution of leukocyte populations resulting in the remodeling of splenic microenvironments in the course of infection and inflammation. Veress et al. (1977) first used the term disorganization to describe the changes affecting the spleen compartments in patients who died of VL. The authors considered that the change in leukocyte populations with depletion of T lymphocytes would lead to impairment of the cell interactions necessary to kill the parasite (Veress et al., 1977). Morrison et al. (1981b), studying Trypanosoma brucei infection, also reported a lack of reactive germinal centers and a disarrangement of lymphoid follicles in late stages of the disease. The authors considered these alterations an indication of poor function of the lymphoid system (Morrison et al., 1981b). In fact, a general description of loss of the normal architecture of the spleen is found in many studies. However, a precise definition of the histological parameters changed by infectious diseases is frequently lacking. In the studies where a more detailed description of spleen changes is presented and hyper- or hypoplastic alterations of the white and RP of the spleen is described, an MZ effacement and macrophage emigration from the MZ into the white and RP are reported (**Table 1**). Therefore, the term disorganization is applied to the spleen for a number of distinct changes of spleen compartments.

# SPLEEN DISORGANIZATION IN VL

Although spleen changes are reported in many diseases, there are few systematic studies about the disruption of spleen compartments. We performed a systematic search in PubMed,

### TABLE 1 | Published papers mentioning spleen histological disorganization associated with parasitic infections.


Web of Science and SCOPUS databases using the following keywords: spleen, disruption, disorganized, disorganization and white pulp. The search resulted in a total of 17 articles. Nine of these articles mentioned Leishmania infection (**Table 1**).

The spleen presents sequential changes during the course of VL in all susceptible hosts. The most evident change is spleen enlargement. In humans, the spleen may reach the right lower quadrant of the abdomen and gives rise to hypersplenism, a syndrome characterized by anemia, thrombocytopenia and low white blood cell counts (Dos-Santos et al., 2014). In fact, a decrease in spleen size is used as a parameter of therapeutic response in human VL (Kip et al., 2015). Leishmania-infected macrophages are observed in all spleen compartments (Andrade and Andrade, 1966). However, parasite distribution and parasite load are heterogeneous in the spleen. Culture of fine needle aspirate collected from different areas of the spleen of dogs with VL showed a variation of 3.2 and 34.7 folds in parasite load present in different regions of the upper, middle and lower third of the organ (Bagues et al., 2018). Associated granulomas, perisplenitis and progressive changes of leukocyte populations are frequently observed (Santana et al., 2008). Initially, the WP shows hyperplastic reactive lymphoid follicles and an increased number of macrophages (Veress et al., 1983; Keenan et al., 1984; Bamorovat et al., 2014). As the disease progresses, reactive lymphoid follicles are disrupted, sometimes replaced by hyaline deposits (Veress et al., 1983). The usual boundaries between the WP compartments are progressively effaced, the mantle zone progressively vanishes, and the MZ becomes less evident (Veress et al., 1977). In chronic severe forms of the disease, the spleen presents WP atrophy, to the degree that secondary lymphoid follicles are no longer found (Santana et al., 2008; Santos et al., 2016). A clear distinction between white and RP is not always feasible and numerous plasma cell aggregates replace the normal resident cell populations of the RP (Santana et al., 2008; Silva-O'Hare et al., 2016). Therefore, the increase in the size and redistribution of cell populations corresponds to an advanced state of functional disorganization of the spleen both hematologically and of defense against infections. These changes may impair the host ability to respond to infection by Leishmania and other pathogens.

To allow a better exchange of information between groups working with the structural changes of the spleen, we proposed a morphological classification of splenic WP organization based on the level of disruption of the different compartments. This classification was based on consensus analysis of the spleens of 72 dogs from an endemic area of VL performed by three pathologists. Initially, four categories were proposed: normal, slightly disorganized, moderately disorganized and intensely disorganized (Santana et al., 2008). In subsequent studies, to improve the agreement between observers, we maintained three

FIGURE 2 | Class of spleen disorganization according to disruption of white pulp structures. Hematoxilyn-eosin staining, scale bar in the top row = 400µm; in the bottom row = 200µm.

categories, collapsing the moderately and intensely disorganized categories into a single category (**Figure 2**). Well organized spleen (spleen type 1) has a distinct periarteriolar lymphocyte sheath, lymphoid follicles and a marginal zone. Present a varied number of secondary lymphoid follicles, containing a germinal center clearly delimited by a rim of small lymphocytes (the mantle zone); slightly disorganized (spleen type 2) has either hyperplastic or hypoplastic changes blurring the boundaries between regions of the white pulp; and moderately to extensively disorganized (spleen type 3) the white pulp regions are poorly individualized or indistinct lymphoid follicles are barely distinct from the red pulp and T-cell areas and secondary lymphoid follicles are absent. As shown in **Figure 2**, the type 1 and type 3 spleens are polar categories of easily distinguished organization

FIGURE 3 | Proposed sequence of events in spleen disorganization: (A) Normal spleen with well-defined white and red pulp compartments: Red pulp (RP) with a mixed leukocyte population with lymphocytes, macrophages and some plasma cells. Marginal zone (MZ) containing lymphocyte and macrophages. Lymphoid follicle (LF) containing lymphocytes and follicular dendritic cells. Periarteriolar lymphocyte sheath (PALS) containing mostly predominantly lymphocytes. The integrity of spleen compartments is dependent on chemokines such as CCL19 and CCL21 (PALS), CXCL13 (LF). (B) After Leishmania infection amastigote-containing macrophages are observed in the different spleen compartments. (C) Antigen stimulation and polyclonal B cell activation leads to white pulp hyperplasia and plasma cell accumulation in the RP. (D) T lymphocyte and follicular dendritic cell apoptosis leads to a reduction of CXCL13 chemokine expression and white pulp disruption. Inflammatory changes of red pulp enhance B-cell activating factor (BAFF), A proliferation-inducing ligand (APRIL) and CXCL-12 chemokine expression, favoring plasma cells homing and survival.

and disorganization of the WP. All the cases that do not fit into these two categories are classified as type 2 spleen (Silva-O'Hare et al., 2016). In a survey of stray dogs collected from an endemic area of VL, 23% (48/206) of the semi-domiciled animals, most of them with Leishmania infection, presented type 3 spleens (Lima et al., 2014). Morphometric studies showed that animals with active leishmanial infection confirmed by serology or spleen culture and type 3 spleen had a smaller WP/RP proportion (6%) than animals with type 1 spleen (13%) (Silva et al., 2012). The decrease in WP size affects predominantly the lymphoid follicles and the MZs, which are 3.5 and 1.9 times smaller, respectively, in animals with type 3 than those of animals with type 1 spleen (Silva et al., 2012). CD79α+ B lymphocytes were decreased in the lymphoid follicles in the MZ and CD3+ T lymphocytes were decreased in lymphoid follicles (Silva et al., 2012). Additionally, da Silva et al. (2018) showed a decrease in number of CD4<sup>+</sup> T lymphocytes and de Lima et al. (2012) showed increased T-cell apoptosis in the disrupted white pulp of naturally infected dogs with VL (de Lima et al., 2012; da Silva et al., 2018). These observations suggest that Tcell exhaustion may play a role in the progression of splenic alterations.

These changes in the spleen are associated with the parasite burden and with a decrease in mRNA expression of proinflammatory and anti-inflammatory cytokines related to the immune response to Leishmania such as IFNγ, IL-12, IL-6, TNF, IL-10, and TGFβ (Cavalcanti et al., 2015). Follicular dendritic cells, an important source of CXCL13, are also decreased in lymphoid follicles (Silva et al., 2012). CXCL13 is a chemokine responsible for B-cell migration into the lymphoid follicle and in lymphoid follicle maintenance (Neely and Flajnik, 2015). Smelt et al. (1997) suggested that Leishmania-infected cells present in the lymphoid follicles may correlate with follicular dendritic cell death, which may explain their reduction in late stages of the disease (Smelt et al., 1997). In fact, the gene expression for CXCL13 is significantly decreased in the spleens of dogs with active L. infantum infection and type 3 spleen (Silva et al., 2012). Furthermore, changes in the extracellular matrix components such as laminin and collagen, as well as an increased expression of Metallopeptidase-9 may also participate in the process (da Silva et al., 2018).

Important changes in cell populations are also observed in splenic RP. Plasma cells progressively become the more frequent leukocyte present in this compartment (Santana et al., 2008; Silva-O'Hare et al., 2016). Plasma cells are highly specialized antibody-secreting cells derived from B lymphocytes. These cells have a complex biology arising from follicular or extrafollicular B-lymphocyte differentiation. Most plasma cells are short lived, remaining only for a few days in the WP in the spleen. Some plasmablasts migrate to the bone marrow, where a few differentiate to long-lived plasma cells (Tangye, 2011). Unfortunately, little is known about the role played by these cells in chronic inflammatory infiltrates. They are present in the spleen and are an important cell component of inflammatory infiltrates found in different organs of patients with VL (Andrade and Andrade, 1966). Spleen plasmacytosis correlates with the dysproteinemia presented by dogs with VL (Silva-O'Hare et al., 2016). Most of the plasma cells that accumulate in the spleen in VL are IgG-producing cells. The RP plasmacytosis persists after WP disorganization and complete disruption of lymphoid follicles (Silva-O'Hare et al., 2016). This finding suggests that most of these cells continue to accumulate in the spleen either by extra follicular B-cell differentiation or by an increased life span in the RP. In fact, BAFF, APRIL and CXCL12 cell factors involved in plasma cell homing and survival are increased in the spleen of dogs with active VL and type 3 spleen (Silva-O'Hare et al., 2016). This observation suggests that inflammatory changes in the RP microenvironment may favor an increase of life span of plasma cells resulting in a shift in the composition of the leukocyte population in this spleen compartment (**Figure 3**).

Spleen disorganization is associated with more severe VL presentations. Dogs with active infection by L. infantum and type 3 spleen had more frequent clinical signs of disease (alopecia, anemia, conjunctivitis, dehydration, emaciation, onychogryphosis, skin erosion and ulceration), more frequently altered laboratory biochemistry and hematological tests (lower serum albumin, higher serum AST, decreased red blood cell counts and increased number of neutrophils) and more frequently negative leishmanin skin test than animals with active Leishmania infection and type 1 spleen (Lima et al., 2014; da Silva et al., 2018).

In humans who died with severe VL, Veress et al. (1977) found WP atrophy and disorganization, absence of GCcontaining lymphoid follicles, decrease in lymphocyte number and lymphocyte replacement by plasma cells (Veress et al., 1977). Recently, we comparatively studied the spleens of patients who died with VL with the spleens of patients who died of other chronic diseases and observed that patients who died of VL had more frequent WP disruption than patients without VL (unpublished data). Furthermore, patients with VL present increased levels of BAFF a cytokine involved in plasma cell differentiation and survival in the serum (Goto et al., 2014). These data suggest that the disorganization of splenic microenvironments is also relevant to human disease. In fact, sporadic reports suggest that splenectomy may contribute to the cure of patients with chronic relapsing human VL resistant to conventional drug treatment (Rees et al., 1984; Alon and Chowers, 2012).

# THE RELEVANCE OF SPLEEN CHANGES IN VL

In this review, we draw attention to the marked changes in the organization of spleen compartments in severe forms of VL. These changes combine an increase in the size and altered distribution of cell components. The cell populations undergo change, and the appropriated sites of cell interactions and differentiation disappear or are substantially disrupted. Secondary lymphoid follicles, a site of development and refinement of antibody immune response, are no longer observed and MZs, homing sites for a variety of memory B cells are almost nonexistent. Since the spleen changes are usual in the whole course of VL, we may consider the role played by the disease in the genesis of these alterations. Furthermore, these changes in the spleen are associated with severe/terminal disease in dogs and in humans. Given the role of spleen in the protection against bloodborne pathogens the change in cell postulations and cytokine expression pattern in disorganized spleen may contribute to the progression of the disease by increasing host susceptibility to Leishmania and other pathogens. In fact, bacterial infection is among the main causes of death in patients with VL (Andrade et al., 1990; Costa et al., 2010). Therefore, the changes observed in the white and RP may impair or subvert the normal immunological functions of the spleen. A better understanding of the sequence of events and pathways involved in spleen disorganization may help to develop more sensitive tests for detecting progressive forms of VL and to design new approaches to the treatment of the disease.

# REFERENCES


# AUTHOR CONTRIBUTIONS

WD-S, GO, MH conceived the manuscript. MH, CM, IL wrote the text. WD-S, GO revised the text. WD-S, MH, produced the figures.

# FUNDING

This work was supported by Fundação de Amparo à Pesquisa do Estado da Bahia (Fapesb. http://www.fapesb. ba.gov.br) grant no. PET0053/2013, AUXPE-CAPES-FAPESB 2072/2013/PROCESSO #23038.006718/2013-19 and Fundação Oswaldo Cruz (Fiocruz. http://portal.fiocruz.br/pt-br-CNPq) PROEP grant 400905/2013-2. WD-S received a scholarship from CNPq. The funders had no role in the study design, data collection or analysis, decision to publish, or preparation of the manuscript.


response to Babesia bovis infection: an immunohistological study. Parasite Immunol. 33, 34?44. doi: 10.1111/j.1365-3024.2010.01249.x


isolated follicle formation and T cell-independent immunoglobulin A generation in the gut. Immunity 29, 261–271. doi: 10.1016/j.immuni.2008. 05.014


**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 Hermida, de Melo, Lima, Oliveira and dos-Santos. 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.

# Cellular Markers of Active Disease and Cure in Different Forms of *Leishmania infantum*-Induced Disease

### *Edited by:*

Nicolas Blanchard, INSERM U1043 Centre de Physiopathologie de Toulouse Purpan, France

### *Reviewed by:*

Diego Luis Costa, National Institute of Allergy and Infectious Diseases (NIAID), United States Tais Berelli Saito, The University of Texas Medical Branch at Galveston, United States

### *\*Correspondence:*

Juan V. San Martin juanvictor.san@salud.madrid.org

†These authors have contributed equally to this work

### *Specialty section:*

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

*Received:* 18 June 2018 *Accepted:* 09 October 2018 *Published:* 13 November 2018

### *Citation:*

Botana L, Matía B, San Martin JV, Romero-Maté A, Castro A, Molina L, Fernandez L, Ibarra-Meneses A, Aguado M, Sánchez C, Horrillo L, Chicharro C, Nieto J, Ortega S, Ruiz-Giardin JM, Carrillo E and Moreno J (2018) Cellular Markers of Active Disease and Cure in Different Forms of Leishmania infantum-Induced Disease. Front. Cell. Infect. Microbiol. 8:381. doi: 10.3389/fcimb.2018.00381 Laura Botana1†, Belén Matía2,3†, Juan V. San Martin<sup>2</sup> \*, Alberto Romero-Maté<sup>2</sup> , Alicia Castro<sup>2</sup> , Laura Molina<sup>2</sup> , Laura Fernandez <sup>1</sup> , Ana Ibarra-Meneses <sup>1</sup> , Marta Aguado<sup>2</sup> , Carmen Sánchez <sup>1</sup> , Luis Horrillo<sup>2</sup> , Carmen Chicharro<sup>1</sup> , Javier Nieto<sup>1</sup> , Sheila Ortega<sup>1</sup> , José Manuel Ruiz-Giardin<sup>2</sup> , Eugenia Carrillo1† and Javier Moreno1†

<sup>1</sup> WHO Collaborating Centre for Leishmaniasis, National Centre for Microbiology, Instituto de Salud Carlos III, Madrid, Spain, <sup>2</sup> Hospital Universitario de Fuenlabrada, Fuenlabrada, Madrid, Spain, <sup>3</sup> Programa de Doctorado en Ciencias de la Salud, Escuela Internacional de Doctorado, Universidad Rey Juan Carlos, Mostoles, Spain

Increased numbers of peripheral blood mononucleocytes (PBMC) and increased IFN-γ secretion following in vitro challenge of blood samples with soluble Leishmania antigen (SLA), have been proposed as biomarkers of specific cell-mediated immunity, indicating that treatment of visceral leishmaniasis (VL) has been successful. However, Leishmania infantum infection may manifest as cutaneous leishmaniasis (CL), and less commonly as localized leishmanial lymphadenopathy (LLL) or mucosal leishmaniasis (ML). The present work examines the value of these biomarkers as indicators of cured leishmaniasis presenting in these different forms. Blood samples were collected before and after treatment from patients living in Fuenlabrada (Madrid, Spain), an L. infantum-endemic area recently the center of a leishmaniasis outbreak. All samples were subjected to Leishmania-specific PCR, serological tests (IFAT and rK39-ICT), and the SLA-cell proliferation assay (SLA-CPA), recording PBMC proliferation and the associated changes in IFN-γ production. Differences in the results recorded for the active and cured conditions were only significant for VL. PCR returned positive results in 67% of patients with active VL and in 3% of those with cured leishmaniasis. Similarly, rK39-ICT returned a positive result in 77% of active VL samples vs. 52% in cured VL samples, and IFAT in 90% vs. 56%; in the SLA-CPA, PBMC proliferation was seen in 16% vs. 90%, and an associated increase in IFN-γ production of 14 and 84%, respectively. The present findings reinforce the idea that PBMC proliferation and increased IFN-γ production in SLA-stimulated PBMC provide biomarkers of clinical cure in VL. Other tests are urgently needed to distinguish between the cured and active forms of the other types of clinical leishmaniasis caused by L. infantum.

Keywords: visceral leishmaniasis, cutaneous leishmaniasis, lymphadenopathic leishmaniasis, mucosal leishmaniasis, biomarker, cure, cell proliferation assay, IFN-γ

# INTRODUCTION

# A Wide Spectrum of Clinical Forms Produced by *L. infantum*

Leishmaniasis is a neglected, vector-borne disease with high morbidity caused by protozoan pathogens of the genus Leishmania. Over 350 million people in some 98 countries are at risk, and an estimated 1.3 million new cases of leishmaniasis are reported every year (WHO, 2010).

Infection with Leishmania infantum is manifested in different clinical forms. The most severe is visceral leishmaniasis (VL), in which the parasite is systemically disseminated; it is fatal if untreated. Cutaneous leishmaniasis (CL) is a benign form caused by the multiplication of Leishmania in the skin; these infections usually clear up spontaneously. Both these clinical forms are seen in Mediterranean countries. During the L. infantumcaused outbreak of leishmaniasis in Fuenlabrada (Madrid, Spain), VL was seen in about a third of people with clinical infection, and CL in most of the remainder (Suárez Rodríguez et al., 2012; Arce et al., 2013); a small number presented with mucosal leishmaniasis (ML) and leishmanial localized lymphadenopathy (LLL). LLL is characterized by long term lymphadenopathy with neither fever nor any other systemic symptom (Horrillo et al., 2015). These differences in the clinical presentation of the disease seem mainly attributable to the patient's immune status. Certainly, the molecular typing of L. infantum isolates from patients in Fuenlabrada revealed no association between genotype and disease type (Chicharro et al., 2013).

# Biomarkers Are Needed That Can Identify Cured Patients Following Treatment for Leishmaniasis

Treatment for leishmaniasis is usually effective, but relapses are common, especially in immunosuppressed patients (van Griensven et al., 2014). Confirmation of a final cure in both CL and VL is still based on clinical features after followup periods of 6 months; relapses commonly occur long after treatment has ended (Rijal et al., 2013). In immunocompetent patients with VL, a successful response to therapy depends on the activation of a Th1 subset of CD4+ Leishmania-specific T cells, and the production of IFN-γ, which induces macrophage leishmanicidal activity (Kemp et al., 1993). Indeed, the proliferation of blood mononucleocytes (PBMC) after challenge with soluble leishmanial antigen (the soluble Leishmania antigen cell proliferation assay [SLA-CPA]) (Singh and Sundar, 2014; Carrillo et al., 2015), and increased IFN-γ secretion by these cells (Hailu et al., 2004; Kumar et al., 2014), provide in vitro markers that might be used to assess early response to treatment. In fact, PBMC proliferation has been shown a useful indicator of the existence of Leishmania-specific T cell memory clones in HIV+ patients, sufficient to keep the parasitic infection under control and avoid relapsing VL (Castro et al., 2016). The value of IFN-γ in monitoring the cellular immune response has previously been reported in VL caused by L. infantum (Cillari et al., 1995; Adem et al., 2016; Ibarra-Meneses et al., 2017).

The present work examines the value of these biomarkers and others as indicators of cured leishmaniasis presenting in its different forms; their sensitivity and specificity have been said to vary across the different manifestations of the disease (Kip et al., 2015). The results confirm that PBMC proliferation and increased IFN-γ production following the challenge of blood samples with SLA, can be used to identify patients cured of VL. However, new biomarkers are required that are able to indicate the same in the other forms of L. infantum-induced disease.

# MATERIALS AND METHODS

# Study Group, Blood Samples, and Tests Performed

The study subjects were 141 adult patients from Fuenlabrada (Madrid, Spain). All were diagnosed at the Hospital Universitario de Fuenlabrada between January 2013 and June 2015. Blood samples were collected during the period of active disease and after apparent cure. Active VL, CL, and ML was defined according to WHO definitions (WHO, 2010). LLL was defined as isolated adenopathy with no other systemic symptom (Ignatius et al., 2011; Horrillo et al., 2015). Blood samples were obtained from 33 patients with active VL, 27 from those with active CL, 6 from those with active LLL, and 2 from those with active ML. Post-supposed-cure blood samples were available from 61 patients originally diagnosed with active VL, 41 with CL, 21 with LLL, and 3 with ML, with cure defined as being free of leishmaniasis symptoms 6 months after the end of treatment (WHO, 2010). None of the cured patients has relapsed after 2 further years of follow-up.

All samples were analyzed by Leishmania-specific nested PCR to detect leishmanial DNA. Leishmania-specific antibodies were determined in plasma using the rK39 immunochromatographic test (rK39-ICT) and immunofluorescent antibody test (IFAT). PBMC proliferation after stimulation with SLA was measured (SLA-CPA assay), and the accompanying production of IFN-γ recorded.

# DNA Extraction and Nested PCR

Genomic DNA was extracted from 100 µl of peripheral blood using the QIAamp DNA Mini Kit (Qiagen, Germany) following the manufacturer's recommendations, and eluted in a final volume of 200 µl of PCR-grade water. The extracts were stored at 4◦C until PCR analysis (maximum of 3 days). Leishmania DNA was detected by nested PCR (LnPCR), targeting the small subunit ribosomal ribonucleic acid gene SSU-rRNA (18S RNA). A first amplification step was performed using primers R221 and R332 (van Eys et al., 1992); the PCR product was then tested in a subsequent amplification step with primers R233 and R333, as described by Cruz et al. (2002).

# Immunofluorescent Antibody Test

2 × 10<sup>5</sup> L. infantum promastigotes (JPC strain MCAN/ES/ 98/LLM-722) in PBS were fixed to glass slides. Two-fold serial dilutions of plasma—from 1/20 to 1/640 in PBS—were then added to separate preparations. The total IgG concentration was then determined by adding fluorescein isothiocyanateconjugated goat anti-human IgG (Fluoline G) (Bio-Mérieux, France) diluted 1/200. The threshold titre for positivity was set at the 1/80 plasma dilution level.

# rK39 Immunochromatographic Test

Antibody detection was performed using the dipstick format Kalazar Detect Rapid Test (InBIOS International, Seattle, WA) according to the manufacturer's instructions.

# Preparation of Soluble *L. infantum* Antigen for SLA-CPA

L. infantum antigen extract was prepared from stationary phase promastigote cultures (JPC strain, MCAN/ES/98/LLM-722) as previously described (Carrillo et al., 2015). Briefly, parasites resuspended in lysis buffer (50 mM Tris/5 mM EDTA/HCl, pH 7) were subjected to three rapid freeze/thaw cycles followed by three 20 s 40 W pulses with a sonicator. These samples were then subjected to two consecutive rounds of centrifugation at 27,000 g and 4◦C for 20 min. The supernatants were then collected, aliquoted, and stored at −80◦C until use. Protein quantification was performed using the Bradford method employing the Bio-Rad Protein Assay Kit (Bio-Rad, California, USA).

# SLA-Cell Proliferation Assay (SLA-CPA)

PBMC were isolated by density centrifugation through Ficoll-Hypaque (Rafer, Spain). The collected cells were cultured in RPMI 1640 supplemented with 10% heat-inactivated fetal bovine serum, 100 IU/ml penicillin, 100µg/ml streptomycin, 2 mM L-glutamine, 50µM 2-mercaptoethanol, and 1 mM sodium pyruvate. They were then plated in 96 well-plates and kept with RPMI 1640 medium alone (unstimulated) or with added SLA (10µg/ml). All were kept in a humidified, 5% CO<sup>2</sup> atmosphere at 37◦C for 5 days. Cell proliferation was measured by bromodeoxyuridine incorporation using the Cell Proliferation Biotrak ELISA Kit (General Electric Healthcare Life Sciences, UK). The results are shown as a stimulation index (SI).

# Cytometric Quantification of IFN-γ

IFN-γ production was determined in 50 µl of the supernatants from the above SLA-stimulated and control PBMC cultures using the BD Cytometric Bead Array Human Flex Set (Beckton Dickinson Biosciences, New Jersey, USA), as previously described (Carrillo et al., 2015). Supernatants were collected and stored at −20◦C for cytokine analysis. Data were acquired using a FACSCalibur flow cytometer and analyzed using the Flow Cytometric Analysis Program Array (Beckton Dickinson Biosciences, New Jersey, USA). IFN-γ production was expressed (in pg/ml) as the difference between the concentration in SLAstimulated and control supernatants.

# Statistical Analysis

Statistical analyses were performed using the SPSS package (Chicago, IL, USA) or GraphPad Prism 7.0 software (GraphPad Software, CA, USA). Cut-offs were determined by calculating the area under the receiver operating characteristic (ROC) curve (AUC) and the 95% confidence interval (CI). Values for variables recorded before and after treatment were compared using the Mann–Whitney U-test. Significance was set at p < 0.05.

# RESULTS AND DISCUSSION

# The Parasitological and Serological Tests Identified Cured VL, but Not Cured CL, LLL, or ML

The results of the PCR (in particular), rK39-ICT and IFAT tests for active and cured VL were significantly different, thus identifying the cured condition (**Table 1**).

In contrast, PCR detected parasite DNA in very few CL, LLL or ML blood samples with no significant differences recorded between the active and cured conditions. Similarly, the rk39-ICT and IFAT tests returned very similar results for the active and cured forms of these disease presentations (**Table 1**).

# Cell-Mediated Immunity: PBMC Proliferation and IFN-γ Production as Biomarkers of Cured VL

Analysis of cell-mediated immunity in the different clinical conditions revealed significant differences between active and cured conditions only for VL (**Table 1**): indeed, most of the patients with active VL showed no Leishmania-specific cellmediated immunity, but did so after effective treatment, with


TABLE 1 | Percentage of positive results returned by parasitological, serological, and cellular immunity tests, in patients with active (A) and cured (C) leishmaniasis.

<sup>a</sup>The stimulation index cut-offs for SLA-CPA (ROC curve) were = 2.53 for VL, 2.26 for CL, 3.16 for LLL, and 4.23 for ML.

<sup>b</sup>The cut-offs for increased IFN-γ production after SLA stimulation of blood (ROC curve) were = 133.4 pg/ml for VL, 314.1 pg/ml for CL, 406.2 pg/ml for LLL, and 406.6 pg/ml for ML. \*p < 0.050.

\*\*\*p < 0.001.

\*\*\*\*p < 0.0001.

PBMC proliferation confirmed (**Figure 1A**, p < 0.0001) and IFNγ levels increased (**Figure 1B**, p < 0.0001) (cut-offs: Stimulation index = 2.53; IFN-γ = 133.4 pg/ml). No such conversion was seen in any other disease presentation type (**Figures 1C–H**).

# Lack of Surrogate End-Points to Define Cure in Leishmaniasis

A confirmation of cure has traditionally been based on clinical features, such as the normalization of body temperature, a reduction in the size of the liver and spleen, and an increase in peripheral blood leukocytes, hemoglobin and platelets in VL, or the healing of lesions in CL. However, observations have to be made over long follow-up times since relapses may occur long after treatment ends. Unfortunately, PCR and serological tests are not good indicators of cure (WHO, 2017). PCR detects parasites in active VL, but after the first doses of treatment their numbers can fall dramatically (Castro et al., 2016) while clinical manifestations of disease remain. Further, the non-detection of peripheral blood parasites does not necessarily mean that the spleen, liver, or bone marrow are parasite-free, which might allow for the reactivation of clinical disease months later, as repeatedly reported in immunosuppressed patients (van Griensven et al., 2014). PCR results on their own cannot, therefore, provide a reliable marker of cure. Neither can serological tests (rK39-ICT and IFAT) be used to identify cure: antibodies are present in active VL but persist long after cure (Hailu et al., 2004; WHO, 2010).

# Cell-Mediated Immunity Tests for Monitoring Treatment

The development of solid cell-mediated immunity rendering patients resistant to reinfection only occurs after successful treatment and the start of healing (Kemp et al., 1993; Kumar et al., 2014). Early monitoring of response to treatment (for both VL and CL) therefore requires specific biomarkers be identified that correlate with the development of this immunity (WHO, 2017). PBMC proliferation and increased IFN-γ production upon stimulation with SLA are expressions of this immunity, and can be easily and repeatedly evaluated ex vivo without patient sensitization (Carrillo et al., 2015). In addition, these two biomarkers have added value in terms of allowing a "no relapse" prognosis to be made: both are directly associated with the immunological control of the parasite and the absence of reactivation. By way of comparison, while PCR can confirm parasite elimination in blood, it cannot predict whether the disease might be reactivated by cryptic parasites in the target organs.

The appearance of cell-mediated immunity after successful treatment has also been reported for Leishmania donovani (WHO, 2017). PBMC proliferation and an associated increase in IFN-γ production upon the stimulation of blood samples with SLA might therefore provide global markers of cure in VL (Adem et al., 2016; Castro et al., 2016). Our group has already shown that a patent cell proliferative response remains long after treatment, that it is useful for monitoring disease in immunosuppressed patients with VL (Castro et al., 2016), and that it is associated with a lack of relapse (Hailu et al., 2004; Singh and Sundar, 2014).

# Lack of Biomarkers for Identifying Cure in CL, LLL, and ML

The present results show that none of the tests examined could distinguish between active disease and cure in CL, LLL, or ML. In fact, many patients with CL and LLL returned positive SLA-CPA results (i.e., PBMC proliferation) before treatment, revealing an active cell-mediated immune response to be underway. The same response has been described in patients with active CL caused by L. major, which appears to be associated with the spontaneous healing of the lesion. In active CL, a lack of PBMC proliferation and low IFN-γ production (together with IL-4 production) following SLA challenge in vitro has been associated with severe disease and a lack of healing (Gaafar et al., 1995; Ajdary et al., 2000).

The mean stimulation index values recorded after SLA challenge shows the cell-mediated immune response against Leishmania to be strongest in patients cured of VL. Differences in this response between patients cured of VL and CL have been reported previously, and might reflect a stronger systemic Th-1 response to occur in the former given the need to clear a much heavier parasite burden (Turgay et al., 2010; Alimohammadian et al., 2012). The patients with LLL—perhaps an evolved form of CL—showed the strongest immunological response before treatment, with Leishmania parasites present in affected lymph glands near the site of the Phlebotomus bite (Horrillo et al., 2015). The intense immunological activity in lymph nodes probably results in an immune over-reaction against the parasite. No information on the immunological features of LLL has previously been reported.

The present study suffers from the limitation of the imbalance in patient numbers between the groups; indeed, the numbers of patients with active ML and LLL are too small to be able to draw safe conclusions. To our knowledge, however, this is the first time that the immune response of such a large number of individuals with different clinical forms of disease caused by L. infantum, and all from the same endemic area, has been examined.

Individual differences in the immune response to L. infantum result in different clinical outcomes. The absence of a specific cell-mediated response to the parasite in patients with active VL allows in vitro SLA-activated PBMC proliferation and the associated increase in IFN-γ production—all of which occurs after successful treatment—to act as biomarkers of cure in this disease type. Their use might improve patient follow-up and reduce the cost of clinical trials of VL treatments. In patients with active CL, LLL, and ML, the cell-mediated response observed

# REFERENCES


in a large number of patients seems capable of preventing the spread of the parasite to the internal organs, but it provides no indicator of cure since there are no differences before and after treatment. Biomarkers of cure of CL, LLL, and ML are, therefore, still needed.

# ETHICS STATEMENT

This study was approved by the Hospital de Fuenlabrada (Madrid) Ethics and Research Committee (APR 12-65 and APR 14-64). All patients gave their written informed consent to be involved.

# AUTHOR CONTRIBUTIONS

JS, AR-M, JR-G, EC, and JM contributed conception and design of the study. BM, AR-M, AC, LM, MA, and LH recruited patients and collected samples. LB, LF, AI-M, CS, CC, JN, and SO did immunological tests. LB, BM, and JS organized the database and performed the statistical analysis. LB and BM wrote the first draft of the manuscript. JS, EC, and JM wrote sections of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

# FUNDING

This work was funded by the Instituto de Salud Carlos III via the Red de Enfermedades Tropicales, Subprograma RETICS del Plan Estatal de I+D+I 2013-2016, which is co-funded by FEDER Una manera de hacer Europa funds, via projects RD16/0027/0017 and RD16CIII/0003/0002.

# ACKNOWLEDGMENTS

We thank the staff of the Hospital Universitario de Fuenlabrada for their assistance in the collection of blood samples.


and monitoring Leishmania infantum infection in patients co-infected with human immunodeficiency virus. Trans. R. Soc. Trop. Med. Hyg. 96 (Suppl. 1), S185–S189. doi: 10.1016/S0035-9203(02)90074-X


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

Copyright © 2018 Botana, Matía, San Martin, Romero-Maté, Castro, Molina, Fernandez, Ibarra-Meneses, Aguado, Sánchez, Horrillo, Chicharro, Nieto, Ortega, Ruiz-Giardin, Carrillo and Moreno. 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.

# Immunity Against *Leishmania major* Infection: Parasite-Specific Granzyme B Induction as a Correlate of Protection

Thouraya Boussoffara1,2, Sadok Chelif 1,2, Melika Ben Ahmed1,2, Mourad Mokni <sup>3</sup> , Afif Ben Salah1,2†, Koussay Dellagi 1,2† and Hechmi Louzir 1,2 \*

<sup>1</sup> Laboratory of Transmission, Control and Immunobiology of Infections, Pasteur Institute of Tunis, Tunis, Tunisia, <sup>2</sup> Université de Tunis El Manar, Tunis, Tunisia, <sup>3</sup> Department of Dermatology, Hospital La Rabta, Tunis, Tunisia

Zoonotic cutaneous leishmaniasis (ZCL) caused by Leishmania (L.) major infection is characterized by different clinical presentations which depend in part on the host factors. In attempt to investigate the impact of the host's immune response in the outcome of the disease, we conducted a prospective study of 453 individuals living in endemic foci of L. major transmission in Central Tunisia. Several factors were assessed at the baseline including (i) the presence of typical scars of ZCL, (ii) in vivo hypersensitivity reaction to leishmanin, and (iii) the in vitro release of granzyme B (Grz B) by peripheral blood mononuclear cells (PBMC) in response to stimulation with live L. major promastigotes. After one season of parasite's transmission, repeated clinical examinations allowed us to diagnose the new emerging ZCL cases. Heterogeneity was observed in terms of number of lesions developed by each individual as well as their size and spontaneous outcome, which led us to establish the parameter "severity of the disease." The efficacy of the presence of typical ZCL scar, the leishmanin skin test (LST) positive reactivity and the high levels of Grz B (≥2 ng/ml), in the protection against the development of ZCL were 29, 15, and 22%, respectively. However, these factors were more efficient against development of intermediate or severe forms of ZCL. Levels of Grz B >2 ng/ml showed the best efficacy of protection (equals to 72.8%) against development of these forms of ZCL. The association of such parameter with the positivity of the LST exhibited a better efficacy (equals to 83.6%). In conclusion, our results support the involvement of Leishmania-specific cytotoxic cellular immune response in host protection against Leishmania-infection. This factor could be of great interest in monitoring the success of vaccination against human leishmaniasis.

Keywords: *Leishmania major*, specific-cytotoxicity, granzyme B, correlate of protection, re-infection

# INTRODUCTION

Cutaneous leishmaniasis are parasitic infections caused by different species of Leishmania parasite and grouping distinct clinical manifestations. In Tunisia, ZCL is due to infection by the parasite L. major zymodeme MON-25 and transmitted by Phlebotomus papatasi (Ben Ismail and Ben Rachid, 1989). ZCL takes place as seasonal epidemics with an annual prevalence ranging from 2 to 10

*Edited by:*

Eugenia Carrillo, Instituto de Salud Carlos III, Spain

### *Reviewed by:*

Claudia Ida Brodskyn, Instituto Gonçalo Moniz (IGM), Fiocruz Bahia, Brazil Sima Rafati, Pasteur Institute of Iran (PII), Iran

### *\*Correspondence:*

Hechmi Louzir hechmi.louzir@pasteur.tn

### *†Present Address:*

Afif Ben Salah, Department of Family and Community Medicine, College of Medicine and Medical Sciences, Arabian Gulf University (AGU), Manama, Bahrain Koussay Dellagi, Direction Internationale, Institut Pasteur, Paris, France

### *Specialty section:*

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

*Received:* 19 June 2018 *Accepted:* 22 October 2018 *Published:* 13 November 2018

### *Citation:*

Boussoffara T, Chelif S, Ben Ahmed M, Mokni M, Ben Salah A, Dellagi K and Louzir H (2018) Immunity Against Leishmania major Infection: Parasite-Specific Granzyme B Induction as a Correlate of Protection. Front. Cell. Infect. Microbiol. 8:397. doi: 10.3389/fcimb.2018.00397 thousand cases (Bettaieb et al., 2014; Bettaieb and Nouira, 2017). Transmission of the parasite occurs during the summer months, and the emergence of active lesions in humans is recorded during the autumn and winter months. The clinical features of ZCL are rather polymorphic, ranging from benign self-limited to wide cutaneous lesions that may cause severe disfigurement. Nevertheless, human's infection by Leishmania might be asymptomatic, particularly in endemic areas (Ben Salah et al., 2005). The outcome of the Leishmania-infection depends partly on the type and intensity of the host immune response. Moreover, the Leishmania-specific immunity associated with healing often provides a resistance to subsequent infection (Guirges, 1971; Davies et al., 1995).

Healing of cutaneous leishmaniasis was mostly associated with the development of a type 1 CD4<sup>+</sup> cell-specific immune response (Sassi et al., 1999) as well as a positive leishmanin skin test (LST) reactivity (Liew and O'Donnell, 1993; Reithinger et al., 2007). Therefore, vaccine efficacy is currently evaluated through Leishmania-specific Delayed Type Hypersensitivity (DTH), PBMC proliferation, or interferon (IFN)-γ production in response to stimulation with Leishmania antigens. These indicators of Th1 response were usually used for the selection of naïve individuals and as correlate of protection during the clinical trials (Reithinger et al., 2007; Duthie et al., 2012). LST is usually used for clinical diagnosis of leishmaniasis and epidemiological surveys (Ben Salah et al., 2005). It is considered as a good protective correlate when evaluating the efficacy of vaccines against leishmaniasis in humans (de Luca et al., 2001; Alimohammadian et al., 2002). However, previous trials evaluating anti-Leishmania vaccines, consisting of the heatkilled or autoclaved L. major parasite showed that there is no protective effect despite the conversion of the LST in vaccinated individuals (Sharifi et al., 1998; Khalil et al., 2000; Armijos et al., 2004). Obviously, the positive LST reactivity observed after vaccination with killed parasite is generally not predictive of protection against ZCL, despite a lower prevalence of disease in individuals with positive LST. Thus, although LST conversion might be an indicator of Leishmania-specific immunity, it may not, however, be considered as an accurate correlate of protection against the disease (Momeni Boroujeni et al., 2013), hence the need to define the immunological basis of resistance to infection with Leishmania parasites. Besides the key role of CD4<sup>+</sup> Th1 cells, which is closely associated with LST reactivity, several studies pointed to the involvement of CD8<sup>+</sup> T cells in acquiring immunity against leishmaniasis in mice model (Belkaid et al., 2002; Rhee et al., 2002). Such cells mediate effectors mechanisms through the secretion of cytokines and chemokines and also through cytotoxic activity (Ruiz and Becker, 2007). Several studies have shown that human Leishmania-specific cell-mediated cytotoxicity is part of the Leishmania-specific acquired immunity (da Conceição-Silva et al., 1994; Brodskyn et al., 1997; Marry et al., 1999; Russo et al., 1999; Bousoffara et al., 2004). Accordingly, in a previous study, we have demonstrated a cytotoxicity of peripheral blood lymphocytes toward L. major-infected macrophages as well as a significant increase of Grz B activity in patients with active infection or those healed from ZCL (Bousoffara et al., 2004). The role of cytotoxic cells during cutaneous leishmaniasis is still controversial. Indeed, it seems that cytotoxic activity might contribute to infection clearance but also to skin ulcer development in patients with CL (Stäger and Rafati, 2012).

In attempt to evaluate the impact of host Leishmania-specific cytotoxic immune response on the outcome of ZCL, we have conducted a prospective study of a large cohort of subjects living in endemic foci in Tunisia. Elucidation of the specific effective immunological mechanisms responsible for the resistance to leishmaniasis is crucial for vaccine development and evaluation.

# MATERIALS AND METHODS

# Study Population and Samples

A total of 59 subjects have been enrolled in the first part of this study consisting on the optimization of Grz B test. Subjects were categorized on four groups basing on clinical, epidemiological and immunological criteria:


The second part consists on a prospective study carried out in two different endemic foci for L. major (Zymodeme MON-25) infection in Tunisia: an old endemic focus (OEF) situated in the district of El Gtar, governorate of Gafsa (south west Tunisia); and a new endemic focus (NEF) situated in the district of Souk Ejjdid, governorate of Sidi Bouzid (Central Tunisia). This study was carried out as part of the research project "Advances in epidemic parameters of ZCL to validate tools for surveillance and control." At enrolment of the study population between April and May 2001, before the parasite transmission season, the donors had a clinical examination looking for active ZCL lesions or typical scars, a blood sampling and an intradermal administration of leishmanin for LST reactivity. Repeated clinical examinations were performed during the season of emergence of lesions (between September 2001 and April 2002) to evaluate the clinical course of L. major infection, occurring after the transmission season (**Figure 1**). The protocols were approved by the institutional review board as detailed

### TABLE 1 | Clinical and immunological data of subjects included in the study.


LST, Leishmanin Skin Test; ZCL, Zoonotic Cutaneous Leishmaniasis; ND, Not done.

*<sup>a</sup>*Number of subject with ZCL scar/Total number of subject tested.

*<sup>b</sup>*Number of subject with positive LST/Total number of subject tested.

Positive LST, Induration's diameter ≥5 mm.

*<sup>c</sup>*Number of subject with positive Leishmania-specific lymphoproliferative response (1cpm ≥ 5,000)/Total number of subject tested.

below. A total of 453 subjects were enrolled in this study (age, mean ± SD: 7.82 ± 1.9 years [range, 3–15 years]; male: female ratio, 0.96). The study population was selected from four primary schools in the district of El Gtar (n = 212) and from three primary schools in the district of Souk Ejjdid (n = 187). Children <6 years old and therefore, of pre-school age, from the region of souk Ejjdid were also recruited (n = 54; **Table 2**).

# Ethics Statements

Before starting the study, written consent were obtained from participants or from their parents in case of children. Subjects enrolled in this study were verbally informed of the nature of the research project and written consent was obtained for the clinical follow-up, leishmanin skin test, and blood sampling. Consent was prepared in the native language (Arabic). The protocol was approved by the Bio-Medical Ethics Committee (BMEC) of Pasteur Institute of Tunis. The leishmanin test applied to the study population is made of leishmanin approved by WHO norms and regulation. It was previously used for thousands of individuals worldwide without significant hazards.

# Leishmanin Skin Test

LST was performed by intradermal injection of 100 µL of leishmanin suspension as previously described by Sassi et al. (1999). An induration's diameter ≥5 mm indicates a positive LST reactivity.

# Parasite Culture

In the present study, we used L. major (zymodeme MON25; MHOM/TN/94/GLC94) obtained from a human ZCL lesion (Kébaïer et al., 2001). Parasites were cultivated at 26◦C in RPMI 1640 medium (Sigma, St. Louis, MO) containing 2 mM L-glutamine, 100 U/mL penicillin, 100 mg/mL streptomycin, and 10% heat-inactivated fetal calf serum. Stationary-phase


TABLE 2 | Clinical and immunological data of subject included on the prospective study.

LST, Leishmanin Skin Test;

<sup>a</sup>Number of subject with scars/Total number of subject tested.

<sup>b</sup>Number of subjects with positive LST/Total number of subject tested.

OEF, Old endemic foci; NEF, New endemic foci.

promastigotes were used for preparation of soluble Leishmania antigens (SLA) as previously described (Sassi et al., 1999) and for stimulation of PBMCs.

# Lymphoproliferative Test

PBMCs were separated from heparinized blood samples using Ficoll-Paque (GE Healthcare) density gradient centrifugation. PBMCs were cultured in 96-well plates at a concentration of 1 × 10<sup>6</sup> cells/mL in a final volume of 200 µL of RPMI 1640 medium (Sigma) supplemented with 2 mM Lglutamine (Sigma), 100 U/mL penicillin (Sigma), 100µg/mL streptomycin (Sigma), and 10% heat-inactivated human AB serum (Sigma). PBMCs were stimulated with SLA (10µg/mL) for 5 days and lymphoproliferative response was evaluated after adding 1µCi/well of [3H]-thymidine (Amersham, France) for the last 6 h using a liquid scintillation counter (Rack Beta, LKB Wallace, Australia). Results were expressed as 1cpm, difference between mean counts of triplicates in SLAstimulated PBMC and mean counts of triplicates in unstimulated PBMC.

# Measure of Granzyme B and IFN-γ Level in Culture Supernatant

Measurement of Grz B levels was performed using ELISA described by Spaeny-Dekking et al. (1998). The Grz B captured by an anti-Grz B monoclonal antibody (MoAb GB11), fixed on microtiter plate was revealed by a biotin-labeled anti-Grz B monoclonal antibody (GB10 MoAb). Recombinant Grz B of known concentration was used for tracing a standard curve. Briefly, wells of the microtiter plates (Maxisorp, NUC) were coated overnight at 4◦C with 100 µl/well of MoAb GB11 diluted on carbonate buffer, 0.1 M NaHCO<sup>3</sup> pH 9.6, at 2µg/ml. Plates were subsequently washed five times with PBS containing 0.02% Tween 20. The free sites were saturated by incubation for 45 min at room temperature (RT) with 150 µl/well of PBS containing 2% tween 20. The recombinant Grz B and the samples, diluted in HPE (high performance ELISA buffer, Central laboratory of the Red Cross Blood Transfusion Service) and 100 µl were added and incubated for 1 h at RT. After washing, 100 µL of GB10 MoAb diluted at a concentration of 0.5µg/mL in HPE buffer containing 1% (v/v) of normal murine serum (NMS). After incubation for 1 h, the wells were washed and then incubated for 20 min with 100 µl of streptavidin coupled to peroxidase (Amersham, Saclay, France) diluted at 1/10,000. Five washes were carried out before adding 100 µl/well of TMB solution (3,3′ , 5,5′ tetramethylbenzidine, Merck, Darmstadt, Germany) at a concentration of 100µg/mL containing 0.003% (v/v) H2O2, diluted in 0.11M sodium acetate buffer pH 5.5. After stopping the reaction with 2N H <sup>2</sup>SO4, optical density was measured using an automated plate reader (Titerteck Multiscan) at 450 nm/540 nm. Grz B levels were determined by reference to the standard curve and results were expressed in pg/mL. Detection threshold was fixed at 1.42 pg/mL.

Measurement of IFN-γ level within supernatant was carried out using OptEIA set ELISA (BD Biosciences, San Jose, CA) according to the manufacturer's recommendations. The results were interpolated from the standard curve established using recombinant IFN-γ and expressed in pg/mL. Detection threshold was fixed at 6.02 pg/mL.

# Statistical Analysis

Statistical analysis was performed using SPSS (version 20.0; SPSS) software. The association between two parameters was assessed by the Spearman's Rank Correlation Coefficient test. The correlation is considered statistically significant if r > 0.3 with a significance level p < 0.05.

The optimal and recommended cut-off points (where applicable) were used to calculate the numbers of subject positive and negative for each test. The comparison between two groups for a studied parameter was carried out by Student's t-test (for the cohort study) or by the Mann-Witney U-test (in the validation step).

For the prospective study, further analysis was performed, including the χ2 tests applied to analyze the associations between the categorical results (LST reactivity/Scar; LST/Grz B; Scar/Grz B). Odds ratios (OR) were then computed to determine the nature of the association between different parameters by stratifying subjects as described by van Belle et al. (2004). A Fisher's exact P-value was also obtained for each association. A positive association is indicated by an odds ratio >1 (significant when p < 0.05 and the 95% confidence limits both exceed 1). Odds ratios significantly <1 indicate a negative association. Estimation of the efficacy of the immune status of the host on the protection against the development of the ZCL was performed through the calculation of the relative risk (RR) and then the preventive fraction (Rothman et al., 2008). In our case, the analyzed factors consist of positive LST reactivity, presence of typical ZCL scar and high levels of Grz B (>2 ng/ml).

### Boussoffara et al. Granzyme B Induction as a Correlate of Protection

# RESULTS

# Levels of Grz B Produced by PBMCs Stimulated With *L. major* Promastigotes

Before starting the prospective study, we optimized the conditions for the analysis of Leishmania-specific cytotoxic immune response. This was done by measurement of Grz B level in supernatants of PBMCs stimulated for 5 days with promastigotes of L. major at a ratio of 3 parasites per cell. The analysis was carried out for 15 subjects with active ZCL, 24 healed ZCL subjects selected basing on the presence of typical scar and positive reactivity to LST (Scar+LST+), and a total of 20 apparently healthy donors. The latter group includes 11 subjects living in endemic areas with no history of ZCL and negative LST (Scar−LST−), and 9 subjects living outside these area (**Table 1**). As shown in **Figure 2A**, the levels of Grz B detected within supernatants of PBMC from subjects cured of the disease (mean ± SD: 20693.9 ± 19576.8 pg/mL) were significantly higher compared to those measured in ZCL patients (mean ± SD: 779.62 ± 1231.48 pg/mL; p < 0.001). In addition, levels of Grz B measured in culture supernatants of cells from apparently healthy and naïve (Scar−LST−) individuals but living in endemic region of L. major transmission were significantly higher (mean ± SD: 7088 ± 8837.11 pg/mL) compared to those detected in the supernatants of PBMCs from healthy and naïve individuals living outside endemic areas (mean ± SD: 305.06 ± 589.67 pg/mL; p = 0.002). This result is very important and may suggest that in endemic area, even in the absence of LST reactivity other marker of specific immune response (like Grz B production) could indicate a previous contact with the Leishmania parasite.

We thus used the results obtained with healthy and naïve individuals living outside endemic areas as reference for the definition of the cut-off of positivity for Grz B levels and calculated as the mean + 3 times the standard deviation of the Grz B levels produced by PBMCs from healthy negative controls living outside endemic area. According to the cut-off, fixed at 2,000 pg/ml, we found that three among the 15 subjects with active ZCL, 19 among the 24 subjects with a history of ZCL and five among the 11 healthy individuals (Scar−LST−) living in endemic area were Grz B+. However, none of the healthy negative controls living outside endemic areas had Grz B levels greater than the cut-off (**Figure 2B**).

Otherwise, a positive correlation was found between Grz B levels and those of IFN-γ measured in culture supernatants (Spearman rank correlation coefficient r = 0.583, p = 0.001; **Figure 2C**). Furthermore, a concordance of 61.36 and 45.09% was found between Grz B levels and results of LST and SLAspecific lymphoproliferation, respectively (**Figures 2D,2E**).

# Evaluation of the Grz B Production as Marker of Cytotoxic Immune Response for Subjects Included in a Prospective Survey

In a next step, we evaluated the Leishmania-specific cytotoxic immune response at the baseline (before the transmission season of the parasite) within donors included in the prospective study (n = 453). Results of the clinical examination as well as those of the LST collected at the baseline showed that 167 subjects had a positive LST, among them 75 individuals exhibited ZCL scars (**Table 2**). Among the remaining individuals with a negative LST, 30 individuals showed ZCL scars. Interestingly, the percentage of individuals with ZCL scars is higher in the old focus (28%; 60/212) compared to the new one (18.7%; 45/240). In addition, similar percentages of subjects with positive LST reactivity were observed in both of focus (33.9% in OEF and 39.5% in NEF). However, in the NEF, this percentage is lower within subjects under 6 years old (27%) compared to that of subjects older than 6 years (43%) indicating a limited contact with the parasite for this age group.

Globally, Grz B was detected within 72.18% (327/453) of culture supernatants at variable levels ranging from 1.43 to 211 560.4 pg/mL (mean ± SD; 9158.5 ± 19246.6 pg/mL). Production of Grz B by PBMCs stimulated with Leishmania parasite was not associated with the age or the gender of subjects. Indeed, no correlation was found between subject's age and the Grz B levels (spearman rank correlation r = 0.175; p = 0.002). Likewise, the difference between Grz B levels measured within PBMCs from male subjects and those from female ones was not statistically significant (p = 0.449). Furthermore, no significant difference was found between Grz B levels detected within individuals from El Gtar (OEF; mean ± SD; 6,946 ± 20,295 pg/mL) and those within subjects from Souk Ejjdid (NEF; mean ± SD; 6,269 ± 13,181 pg/mL; p > 0.05). However, in the new endemic focus, Grz B levels were significantly higher within individuals over 6 years old comparing to those within children <6 years old (p < 0.0001; **Figure 3A**). Moreover, according to the cut-off of positivity of the Grz B levels, the percentage of individuals with high levels of Grz B (35. 8%) was greater in the group over 6 years old than in this under 6 years old (9.3%; **Figure 3B**). This could reflect the importance of frequency of the contact with the parasite (on time and intensity) on the development of the host's immune response.

# Association Between Clinical and Immunological Markers During ZCL

We investigated association between Grz B levels, results of LST and the presence of typical ZCL scars. Statistical analysis showed a positive correlation between Grz B levels and the delayed hypersensitivity test (Spearman's rho correlation r = 0.523, p < 0.0001; **Figure 4A**). However, some discrepancy was noted between these tests. Indeed, we noted that 37 individuals showed variable levels of Grz B despite a negative LST (induration diameter < 5 mm) and conversely 66 individuals showed Grz B levels <2,000 pg/ml although they exhibited positive LST reactivity. As expected, levels of Grz B produced by PBMCs from individuals with a positive LST were higher comparing to those from individuals with negative LST (p < 0.0001; **Figure 4B**). Similarly, a significant difference was found between Grz B levels within individuals with ZCL scars (Scar+) and those without scar (Scar−; p < 0.0001; **Figure 4C**). Moreover, significant differences were found between Grz B

individuals living in endemic focus for transmission of L. major or outside these areas. Horizontal bars represent median values. (B) Histograms represent the number of subjects categorized in Grz B<sup>+</sup> and Grz B−, basing on the cut-off fixed at 2,000 pg/mL. (C) Association between Grz B and IFN-γ levels measured in the same supernatants (D,E) Association between Grz B levels and results of leishmanin skin test (LST), expressed as diameter of induration and the lymphoproliferative response to soluble Leishmania antigens (SLA), expressed as 1cpm. Horizontal dashed lines indicate cut-off for Grz B level, LST and proliferation fixed at 2,000 pg/ml, 5 mm and 5,000 cpm, respectively.

levels measured in individuals with a negative LST and no ZCL scar (LST−Scar−) and those with a positive LST with scars (LST<sup>+</sup> Scar+) or without scars (LST<sup>+</sup> Scar−; p < 0.0001; **Figure 4D**).

In addition, we found that 138 individuals among the 453 individuals tested showed Grz B levels >2,000 pg /ml. These high levels were specifically found within individuals with positive LST and those with ZCL scars (**Figures 4E,4F**). As described in **Table 3**, statistical analysis showed a positive association between levels of Grz B and the positivity of LST (χ <sup>2</sup> = 112.49, df = 1, p < 0.0001) and the presence of scars (χ <sup>2</sup> = 49.26, df = 1, p < 0.0001).

# Incidence of ZCL Among Subjects of the Prospective Study After One Transmission Season

<2,000 pg/mL (Grz B−). OEF: old endemic focus; NEF, New endemic focus.

To investigate the protective effect of the host's immune status on the development of ZCL, we performed a clinical follow-up of the 453 individuals, between the month of September 2001 and April 2002, to diagnose new lesions of ZCL. We noticed that, after one season of transmission, 90 among the 453 subjects (19.9%) developed new ZCL lesions. Number of lesions ranged from 1 to 9 (**Figure 5A**) and were mostly localized within limbs or face, with a total area ranging from 12.5 to 5985.6 mm<sup>2</sup> (mean ± SD = 492.7 ± 784.8 mm<sup>2</sup> ) and a mean duration of progression ranging from 71 to 439 days (mean ± SD = 143.7 ± 62 days).

To refine the analysis, we defined a new variable for the expression of the severity of the disease calculated basing on the total number of lesions, the total surface of lesions (taken at their maximum during the various visits) as well as the average duration of evolution of the lesions. The severity score is determined by the formula:

[Average duration of lesions (days) <sup>∗</sup>Total lesion′ s area (mm<sup>2</sup> )] /[Number of lesions<sup>∗</sup> 1000]

We thus categorized ZCL patients into two groups: (GI) with a lesion's severity score comprised within the interval [0–20] and a second group (GII) with a severity score >20. Among the 90 individuals who developed a new ZCL lesion and for whom information on the spontaneous evolution and size of lesions were available (n = 84), 45 developed a mild form of the disease (GI) while 39 individuals presented intermediate or severe forms of the disease (GII; **Figure 5B**).

# Impact of the Immune Response in Protection Against Infection by *L. major*

Finally, we evaluated the efficacy of the host's factors in protecting against L. major infection. These factors include the presence of typical ZCL scar, the positive LST reactivity and the induction of high levels of Grz B (≥2000 pg/ml). These factors showed a weak efficacy of protection against the development of ZCL lesion. Indeed, protective efficacy was 29, 15, and 22% for the presence of scar, positive LST reactivity and high Grz B levels with a non-significant p (**Table 4**).

When ZCL patients were categorized into two groups (GI and GII) according to the lesion's severity score, we showed that none of these parameters was protective against the development of mild ZCL (RR > 1). However, all these parameters were highly protective against the development of intermediate or severe leishmaniasis (**Table 4**). Accordingly, the incidence of the latter forms of leishmaniasis was significantly lower in individuals with typical ZCL scars (relative risk RR = 0.38, Fisher's exact p = 0.047) and presence of scar showed a protective efficacy of 62% (**Table 4**). Similarly, positive LST reactivity seems to be efficient in protecting against the development of intermediate or severe disease with a protective efficacy equal to 60% (RR = 0.39, Fisher's exact p = 0.009; **Table 4**). Strikingly, only 2.9% (4/137) of individuals with Grz B levels >2,000 pg/ml developed intermediate or severe forms of ZCL (Severity > 20) while 13.1% (18/137) developed a mild form of the disease with severity score comprised within the interval [0–20]. Thus high levels of Grz B showed an efficacy of 72.8% (relative risk RR = 0.27, Fisher's exact p = 0.003) to protect against such forms of leishmaniasis. Interestingly, together, the positivity of the LST and level of Grz B >2,000 pg/ml, leads to a better efficacy of protection against development of intermediate or severe leishmaniasis (efficacy is 83.6%). Indeed, only 2% (2/100) of LST<sup>+</sup> individuals, with Grz B levels >2,000 pg/ml, developed intermediate or severe ZCL (**Figure 6**).

## Discussion

Conventionally, protection during leishmaniasis is associated with the development of a Th1-type immune response that can be demonstrated in humans by positive LST reactivity and specific

FIGURE 4 | Association between granzyme B levels, LST reactivity and presence of typical ZCL scar. (A) Grz B level expressed in function of result of LST. n number of individual for each group. (B–D) Histograms represent mean values of Grz B levels measured in culture supernatants of PBMCs from individuals, categorized basing on categorical results of LST (LST+/LST−) and presence or absence of ZCL scars (scar+/scar−). (E,F) Categorial result for Grz B expression within individuals categorized basing on LST results (LST+/LST−) or presence or absence of scar (scar+/scar−). Grz B+: Grz B levels ≥2 ng/ml, Grz B−: Grz B levels <2 ng/ml. LST+, Diameter of induration ≥5 mm; LST−, Diameter of induration <5 mm.

TABLE 3 | Odds ratios and confidence intervals describing the associations between putative markers of previous ZCL and the immune status at baseline for individuals of the study.


LST, Leishmanin skin test; Cic, Presence/ansence of ZCL scar; Grz B, levels of granzyme measured within supernatant of PBMCs stimulated with L.major promastigotes; N, number of subject included in the study;

\*Statistically significant (p< 0.05).

in vitro T cell responses associated with IFN-γ production (Liew and O'Donnell, 1993; Sassi et al., 1999; Reithinger et al., 2007). We and others have demonstrated the involvement of cytotoxic T cell (CTL) response as part of the acquired adaptive immune response developed against the parasite Leishmania (da Conceição-Silva et al., 1994; Brodskyn et al., 1997; Marry et al., 1999; Russo et al., 1999; Bousoffara et al., 2004). However, the role of the CTL response during Leishmania infections is still controversial. In humans, clinical studies show the presence of a large number of CD8<sup>+</sup> T cells in the lesions as well as in blood of ZCL patients during the acute phase, but also during the healing process (Da-Cruz et al., 1994, 2002, 2005; Gaafar et al., 1999; Bottrel et al., 2001). In addition, L. major infection induces Th1 and CD8<sup>+</sup> T cells in human patients and both responses are associated with disease resolution (Nateghi Rostami et al., 2010). Nevertheless, other studies have shown that cytotoxicity is one of the main mechanisms underlying disease induced by L. braziliensis infection (Faria et al., 2009; Dantas et al., 2013; Novais et al., 2013; Santos et al., 2013; Cardoso et al., 2015; Ferraz et al., 2015, 2017; Novais and Scott, 2015). Herein, we used the release of Grz B by PBMC in response to live L.major promastigotes as marker of cytotoxicity and clearly demonstrated that Grz B is a good marker of immunity to L.major infection, but, most importantly it constitutes a good correlate for protection against intermediate and severe form of CL due to L. major infection. This result is extremely important for vaccine development and evaluation.

In this study, we first showed that in endemic foci of L. major infection subjects having a previous contact with the parasite showed the presence of Leishmania-specific cytotoxicity. Indeed, compared to healthy individuals living outside endemic area for L. major transmission, those living in these areas, including subjects with active or healed ZCL but also those apparently healthy with no ZCL scars and no LST reactivity (Scar<sup>−</sup> LST−) showed high levels of Grz B. Interestingly, even in the absence of LST reactivity, Grz B induction may constitute an indicator of a previous contact with the parasite and may reflect an eventual role in control of Leishmania infection. This data is contradictory with those described by Cordosa and collaborators showing the involvement of CD8+T cells producing IFN-g rather than those

with cytotoxic activity in the control of Leishmania infection (Cardoso et al., 2015). One explanation could be a difference in the nature of Grz B-producing cells between subjects with history of ZCL and those with subclinical infection. Accordingly, we have previously shown that in healed ZCL subjects (Scar+LST+), the main source of GrB produced in response to stimulation with Leishmania-antigens is CD4<sup>+</sup> T cells, including T regulatory cells. These CD4<sup>+</sup> T cells were different from the Th1 cells producing IFN-γ (Naouar et al., 2014).

Furthermore, the higher peripheral Grz B levels in individuals healed from ZCL compared to those with active disease might be attributed to the homing of Grz B-producing cells to the site of infection in the latter group. Accordingly, we have demonstrated high levels of Grz B mRNA as well as the presence of double positive CD8<sup>+</sup> Grz B<sup>+</sup> T cells within active ZCL lesions (Boussoffara et al., submitted). In American cutaneous leishmaniasis (ACL), the percentage of CD8<sup>+</sup> T cells were

FIGURE 6 | Impact of the host's immune response on the prevalence of ZCL. Expression of severity score in function of result of LST reactivity, expressed as diameter of induration (mm) and those of Grz B induction expressed in pg/mL.


TABLE 4 | The relative risk (RR) of the different tests.

<sup>a</sup>Number of subject positive for the test/Number of individuals negative for the test.

<sup>b</sup>Number of subject who developed ZCL lesions/Number of subjects without any ZCL lesion.

<sup>c</sup>RR, Relative Risk, RR < 1 indicates a decrease in the risk of ZCL and vice versa.

<sup>d</sup>Number of subject who developed an intermediate or severe ZCL (Sv)/Number of subjects without any ZCL lesion.

\*Statistically significant (p < 0.05)

N, Number of tested subjects; LST, Leishmanin Skin Test; Scar +/−, presence/absence of ZCL scar; GrzB +/−, Grz B level ≥2 ng/ml or <2 ng/ml.

+/−Corresponding to the number of subject positive for the test/Number of subject negative for the test.

higher in lesions compared to blood (Conceição-Silva et al., 1990; Da-Cruz et al., 2005). This, associated with an increase in Leishmania-specific T cells in lesions, have been attributed to the migratory process between these compartments Moreover, a predominance of CD4<sup>+</sup> and CD8<sup>+</sup> effector memory T cells (TEM-CD45RO+CCR7−) have been shown in ACL lesions. An enrichment of TEM for both CD4<sup>+</sup> and CD8<sup>+</sup> T cells sets were observed comparing to blood (de Oliveira Mendes-Aguiar et al., 2016).

In a next step, the measurement of Grz B levels in culture supernatants of PBMCs stimulated with L. major promastigotes was performed during the prospective study of a cohort of 453 individuals living in L.major transmission area. We, thus, recovered in the beginning of the study (before the season of L. major transmission) the associations between the different parameters indicators of an earlier contact with the Leishmania parasite (presence of typical ZCL scars, LST reactivity and production of Grz B). Our results showed some discrepancies between production of Grz B and scars or LST reactivity. The discrepancy between production of Grz B and the presence or absence of scars might be explained by the fact that lesions of ZCL could disappear without leaving scars and that scars of other pathologies could be taken for ZCL scars. However, the discrepancy between Grz B and LST reactivity seems to be more relevant. Thus, a large fraction of individuals (n = 134) living in the endemic area of leishmaniasis showed a diameter of induration equals to zero but variable levels of Grz B which are sometimes very high. This is consistent of the striking results found during the validation of the Grz B test. Indeed, we were struck by the differences in Grz B levels between the so-called "naïve" (Scar<sup>−</sup> LST−) living in an endemic area of leishmaniasis vs. those living outside the endemic areas. These results are extremely important for the selection of naive populations in endemic areas, especially when including adults in vaccination trials. In fact, the selection of individuals Scar−LST<sup>−</sup> does not exclude an eventual previous contact with the parasite which could be tracked otherwise by the Grz B analysis.

Secondly, we explored the protective role of the cytotoxic effector against the development of the disease. Since ZCL is heterogeneous in terms of disease severity (number and size of lesions, duration of spontaneous healing), we hypothesized that host immune response will not protect against the development of leishmaniasis but rather against the severe forms of it. Our objective was achieved through the monitoring of the cohort included in our prospective study. We analyzed the predictive value of a cytotoxic response (investigated by the Grz B assay) in resistance to L. major infection, as well as the positive LST reactivity, usually used as a protective correlate. Immune response was evaluated at the baseline (before the parasite transmission season) and the development of new ZCL cases was monitored actively during the season of emergence of the disease. A gradation of the disease was also made by calculating a new parameter, severity of the disease, basing on the size of lesions as well as their total duration of evolution.

During the season of emergence of the disease, we diagnosed a total of 90 new cases of ZCL among the 453 individuals tested, with mild, intermediate or high severity score. Interestingly, we noticed a significant number (n = 45) of individuals that develop a simple ZCL characterized by a unique or low number of lesion(s) and a rapid spontaneous healing (severity score <20). Surprisingly, we found that neither the positivity of the LST, nor the presence of ZCL scar, nor the cytotoxic activity (Grz B level ≥2,000 pg/ml), were protective against the development of the disease. This can be explained by the fact that in our study, patients have been actively monitored allowing the diagnosis of mild forms of ZCL, often ignored by the patient himself. However, the positive LST reactivity, the presence of ZCL scar and also the cytotoxic activity were protective against the development of intermediate or severe forms of ZCL. Accordingly, we showed that a Grz B level >2,000 pg/ml has a protective efficacy of 72.8% against the development of intermediate or severe forms of ZCL. Such efficacy was greater than those obtained with the presence of ZCL scar and the positivity of LST (62 and 60%, respectively). Interestingly, together positivity of the LST and a cytotoxic response (Grz B level ≥2,000 pg/ml) confer an excellent protection (>80% efficiency) against intermediate or severe forms of ZCL.

Altogether, our study clearly demonstrated that Grz B production may be an indicator of a previous contact with L.major although the negativity of LST and the absence of scars. Such parameter could be used in vaccination trials for the selection of naive populations living in endemic areas. Moreover, our data show that the preexistence of a parasite-specific cytotoxic immune response may confer protection against the development of intermediate or severe forms of ZCL. These results are of crucial importance for anti-Leishmania vaccine design and more generally for the evaluation of their protective efficacy.

# AUTHOR CONTRIBUTIONS

TB, KD, HL, and AB conceived and designed the experiments. TB performed the experiments. MM carried out the clinical monitoring. TB and SC performed the statistical analyses. TB, MB, and HL wrote the manuscript with the input and approval of all other coauthors.

# FUNDING

This work was supported by the Tunisian Ministry of Scientific Research and Technology and the European Commission (Research and Technology Development, RTD, project; grant ICFP599A3PR01).

# ACKNOWLEDGMENTS

We are indebted to subjects living in endemic area in Tunisia, who actively collaborated in this study. We are grateful to the health-care workers in regional hospitals and to Mr A. Zaatour for their helpful assistance. We are also thankful to Pr. P.P. Tak, Pr. R.J. ten Berg, and Pr. C.E. Hack (CLB Blood Bank, Amesterdam) for their valuable help on the analysis of Grz B production.

# REFERENCES


production after symptomatic or asymptomatic Leishmania major infection in Tunisia. Clin. Exp. Immunol. 116, 127–132.


**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 Boussoffara, Chelif, Ben Ahmed, Mokni, Ben Salah, Dellagi and Louzir. 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.

# Phenotypic and Functional Profiles of Antigen-Specific CD4<sup>+</sup> and CD8<sup>+</sup> T Cells Associated With Infection Control in Patients With Cutaneous Leishmaniasis

### Edited by:

Eugenia Carrillo, Instituto de Salud Carlos III, Spain

### Reviewed by:

Hira Nakhasi, Center for Biologics Evaluation and Research (FDA), United States Javier Moreno, Instituto de Salud Carlos III, Spain Vicente Larraga, Department of Cellular and Molecular Biology, Consejo Superior de Investigaciones Científicas (CSIC), Spain

### \*Correspondence:

Manuel C. López mclopez@ipb.csic.es M. Carmen Thomas mcthomas@ipb.csic.es

### Specialty section:

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 25 April 2018 Accepted: 19 October 2018 Published: 19 November 2018

### Citation:

Egui A, Ledesma D, Pérez-Antón E, Montoya A, Gómez I, Robledo SM, Infante JJ, Vélez ID, López MC and Thomas MC (2018) Phenotypic and Functional Profiles of Antigen-Specific CD4<sup>+</sup> and CD8<sup>+</sup> T Cells Associated With Infection Control in Patients With Cutaneous Leishmaniasis. Front. Cell. Infect. Microbiol. 8:393. doi: 10.3389/fcimb.2018.00393 Adriana Egui <sup>1</sup> , Darién Ledesma<sup>1</sup> , Elena Pérez-Antón<sup>1</sup> , Andrés Montoya<sup>2</sup> , Inmaculada Gómez <sup>1</sup> , Sara María Robledo<sup>2</sup> , Juan José Infante<sup>3</sup> , Ivan Darío Vélez <sup>2</sup> , Manuel C. López <sup>1</sup> \* and M. Carmen Thomas <sup>1</sup> \*

<sup>1</sup> Molecular Biology Department, Instituto de Parasitología y Biomedicina "López Neyra", Consejo Superior de Investigaciones Científicas, Granada, Spain, <sup>2</sup> Programa de Estudio y Control de Enfermedades Tropicales, Facultad de Medicina, Universidad de Antioquia, Medellín, Colombia, <sup>3</sup> Bionaturis Group, Bioorganic Research and Services, S.A., Jerez de la Frontera, Spain

The host immunological response is a key factor determining the pathogenesis of cutaneous leishmaniasis. It is known that a Th1 cellular response is associated with infection control and that antigen-specific memory T cells are necessary for the development of a rapid and strong protective cellular response. The present manuscript reports the analysis of the functional and phenotypic profiles of antigen-specific CD4<sup>+</sup> and CD8<sup>+</sup> T cells from patients cured of cutaneous leishmaniasis (CL), patients with an active process of cutaneous leishmaniasis, asymptomatic individuals with a positive Montenegro test and healthy donors (HD). Peripheral blood mononuclear cells (PBMCs) from the patients exhibited a lymphoproliferative capacity after stimulation with total soluble protein from either Leishmania panamensis (SLpA) or Leishmania infantum (SLiA) or with a recombinant paraflagellar rod protein-1 (rPFR1). Higher frequencies of antigen-specific TNAIVE cells, mainly following stimulation with rPFR1, were observed in asymptomatic and cured patients than in patients with active cutaneous leishmaniasis, while T cells from patients with active cutaneous leishmaniasis showed a higher percentage of effector memory T cells (TEM for CD4<sup>+</sup> T cells and TEMRA for CD8<sup>+</sup> T cells). The amount of antigen-specific CD57+/CD8<sup>+</sup> TEMRA cells in patients with active cutaneous leishmaniasis was higher than that in cured patients and asymptomatic subjects. Regarding functionality, a more robust multifunctional CD8<sup>+</sup> T cell response was detected in cured patients than in those with active cutaneous leishmaniasis. Moreover, cured patients showed a significant increase in the frequency of cells expressing a Th1-type cytotoxic production profile (IFN-γ <sup>+</sup>/granzyme-B/+perforin+). Patients with an active leishmaniosis process had a significantly higher frequency of CD8<sup>+</sup> T cells expressing the inhibitory CD160 and 2B4 receptors than did cured patients. The expression profile observed in cured patients could be indicative of an imbalance toward a CD8<sup>+</sup> Th1 response, which could be associated with infection control; consequently, the determination of this profile could be a useful tool for facilitating the clinical follow-up of patients with cutaneous leishmaniasis. The results also suggest a possible exhaustion process of CD8<sup>+</sup> T cells associated with the evolution of Leishmania infection.

Keywords: Leishmania, leishmaniasis, paraflagellar rod protein-1, biomarkers, CD8<sup>+</sup> and CD4<sup>+</sup> T-cells, phenotype, inhibitory receptors, Th1-cytokines

# INTRODUCTION

Leishmaniasis is caused by intracellular parasites belonging to Leishmania species, and 1.5–2 million new cases are reported annually worldwide (WHO, 2016). Therapeutic options are limited, and there is no effective vaccine (Alvar et al., 2012). Depending on the infecting Leishmania species, different clinical manifestations occur, with cutaneous leishmaniasis (CL) being the most prevalent clinical form (de Vries et al., 2015). In South and Central America, Leishmania parasites of the subgenus Viannia are the most prevalent etiologic agents of human CL (Castilho et al., 2010). Systematic studies carried out in different areas of Colombia since the 1980s showed the presence of six species belonging to the genus Leishmania, with a higher prevalence of Leishmania panamensis (50.8–74.5%) and Leishmania braziliensis (15.3–30.3%) isolates than isolates of the other species. It is thought that 97% of the pathologies caused by Leishmania spp. in Colombia correspond to CL (Corredor et al., 1990; Ovalle et al., 2006; Ramírez et al., 2016).

The infections caused by Leishmania species are, in many cases, self-healing, so it is assumed that the host immune response is a key factor that determines the pathogenesis of the infection. It has been widely reported that the Th1 response is critical for the control of Leishmania infection, since this response creates a cytokine environment that promotes the clearance of the parasite by macrophages (Kaye and Scott, 2011). The development of Leishmania infection in IFN-γ- and TNF-α-deficient murine models increased the lesion sizes and the parasite burdens (Theodos et al., 1991; Wilhelm et al., 2001; Pinheiro and Rossi-Bergmann, 2007). CD4<sup>+</sup> and CD8<sup>+</sup> T cells play a central role in the Th1 response by producing IFN-γ, TNF-α, and other Th1 cytokines that are essential for controlling parasite growth (da Silva Santos and Brodskyn, 2014). Thus, the cellular immune functions performed by these T cells are fundamental for eliminating the parasites, although there is evidence that CD8<sup>+</sup> cytotoxic T lymphocytes (CTL) are involved in tissue damage in CL patients through cytotoxic mediators (Faria et al., 2009; Santos Cda et al., 2013). It is equally important to note that T lymphocytes play a critical role in protection against reinfection by Leishmania species. In this sense, after primary infection, longlived memory T cell populations are maintained in the absence of antigens and are able to mediate immunity against a second infection (Glennie and Scott, 2016). It has been reported that cured patients who have overcome an episode of CL harbor specific effector memory T cells (TEMs) that produce IFN-γ and central memory T cells (TCMs) that produce IL-2 in response to stimulation with soluble leishmania antigens (Keshavarz Valian et al., 2013).

During the chronic stage of Leishmania infection, antigenspecific T cells become functionally impaired, as has been observed in other protozoan diseases (Gigley et al., 2012; Rodrigues et al., 2014). This dysfunctional process, known as T cell exhaustion, occurs gradually, with the upregulation of both the expression and coexpression of inhibitory receptor molecules in the membrane of T cells. It has been reported that CD8<sup>+</sup> T cells from patients with visceral leishmaniasis exhibit an increased expression of the inhibitory receptors CTLA-4 and PD-1 (Gautam et al., 2014). In experimental models of Leishmania donovani infection, the blockade of the PD-1/PD-L1 pathway partially restored CD8<sup>+</sup> T cell immune functions and significantly reduced the splenic parasite burden (Joshi et al., 2009; Hernández-Ruiz et al., 2010). Nevertheless, further information is needed to understand this exhaustion process in the context of Leishmania infection and its impact on the progression of leishmaniasis.

A systematic review of biomarkers for monitoring therapeutic responses in leishmaniasis (Kip et al., 2015) stated that sensitive and specific markers that are capable of assessing therapeutic efficacy and are able to predict long-term clinical outcomes using noninvasive sampling methods are urgently needed. The paraflagellar rod proteins (PFRs) represent a family of relevant trypanosomatid antigens that are located in the paraflagellar pocket of these parasites (Cachon et al., 1988). Knockout assays in Leishmania mexicana demonstrated that the proteins encoded by PFR genes play a critical role in the mobility and survival of the parasite (Santrich et al., 1997). Some members of the PFR antigen family stand out due to their high immunogenicity (Michailowsky et al., 2003). Additionally, a very recent study highlighted the potential of a recombinant PFR1 antigen for the serological diagnosis of Leishmania infantum infection (Ledesma et al., 2017). In the context of trypanosomatid infection by Trypanosoma cruzi, the immunization of mice with a protein from the PFR family (PFR2) fused to HSP70 as a DNA vaccine provided a protective response against T. cruzi experimental infection by inducing an increase in the expression of IL-2 and IFN-γ by splenic CD8<sup>+</sup> T cells and by the generation of antigenspecific T cells (Morell et al., 2006).

In the present work, the cellular mechanisms that are triggered in human patients with different clinical statuses related to Leishmania infection were studied. For this purpose, a phenotypic and functional characterization of antigen-specific CD4<sup>+</sup> and CD8<sup>+</sup> T cells from patients cured of CL, patients with active CL (CL), asymptomatic individuals with a positive Montenegro test and healthy donors (HD) was conducted. Additionally, the degree of exhaustion and the senescence profile of these antigen-specific T cells were evaluated.

# MATERIALS AND METHODS

# Peripheral Blood Mononuclear Cells Isolation of the Study Populations

Peripheral blood was collected from Montenegropositive subjects (n = 4), cutaneous leishmaniasis patients (n = 4), patients cured of leishmaniasis (n = 5), and healthy donors from endemic (n = 4), and nonendemic areas (n = 6). All individuals included in this study live in Colombia. Peripheral blood (∼20 mL) from each individual was collected by venipuncture into EDTA-containing tubes (BD Vacutainer), and peripheral blood mononuclear cells (PBMCs) were purified as previously described (Egui et al., 2012). The PBMCs were suspended in inactivated fetal bovine serum (iFBS) (Gibco, Grand Island, NY) containing 10% DMSO and cryopreserved in liquid nitrogen until use.

Individuals with evidence of exposure to Leishmania but without clinical manifestations of disease were identified by a positive Montenegro skin test. To this end, 0.1 mL of leishmanin (provided by PECET, Antioquia University, Medellin-Colombia) was subcutaneously inoculated into the arms of healthy subjects living in a leishmaniasis-endemic region of Colombia (Caldas Department) where the most widespread species of Leishmania is L. panamensis (Corredor et al., 1990; Ovalle et al., 2006; Ramírez et al., 2016). Subjects positive to Montenegro skin test were carefully clinically evaluated. They were considered asymptomatic when it was confirmed that they never presented lesions or scarring due to Leishmania infection in their clinical history, and were never treated against this disease. The diagnosis of CL patients was based on combined procedures, using direct microscopic examination, parasite growth, and molecular (PCR) methods. Smear preparations from the skin lesions were fixed and stained with Giemsa for microscopic observation and also cultivated in NNN media (Robinson et al., 2002). The identification of the Leishmania species causing the infection was carried out by PCR-RFLP following a previously published protocol (Montalvo et al., 2008). All the collected samples corresponded to the L. panamensis species. L. panamensis infection requires specific treatment. There are no data that report spontaneous cure in patients infected with L. panamensis in Colombia. All the cutaneous leishmaniasis patients presented from two to six cutaneous lesions. Blood was collected from these patients before the administration of antiparasitic treatment. The enrolled patients who had been cured of leishmaniasis had not presented any cutaneous lesions compatible with leishmaniasis for at least the previous 2 years.

# Ethical Considerations

The Human Research Bioethics Committee of the University of Antioquia, Medellin (reference: 16-05-727) and the Consejo Superior de Investigaciones Científicas (CSIC) in Spain (reference: 094/2016) approved the protocols used in this study. All subjects gave written informed consent in accordance with the guidelines of the Declaration of Helsinki.

# Overexpression and Purification of the Recombinant PFR1 Protein

The cloning, expression, and purification of the L. infantum PFR1 antigen were performed as described previously (Ledesma et al., 2017). The overexpression of the recombinant PFR1 protein was induced by the addition of 0.2 mM isopropylbeta-D-thiogalactopyranoside (IPTG) for 3 h at 37◦C. Protein was solubilized in a buffer containing 0.3 M NaCl and 50 mM Na2HPO<sup>4</sup> and purified by Ni2+-NTA affinity chromatography. The purified protein was visualized via 10% SDS-PAGE followed by Coomassie blue staining. The protein concentration was measured using a Micro BCA Protein Assay Kit (Thermo Fisher Scientific). The purified PFR1 protein was tested by an E-Toxate reaction kit (Limulus amebocyte lysate [LAL], Sigma), which showed that the endotoxin levels were below the detection limit of the kit (0.1 endotoxin units/mL).

# Isolation of Leishmania panamensis and Leishmania infantum Total Soluble Antigens

L. panamensis and L. infantum promastigotes were grown in modified RPMI 1640 medium supplemented with 20% FBS and gentamycin (50µg/mL). Promastigotes were collected by centrifugation and washed with 1 × PBS at pH 7.2. The parasites were suspended at 10<sup>9</sup> parasites/mL in lysis buffer (50 mM Tris-HCl at pH 7.4, 50 mM NaCl, 0.005% NP-40, 1 mM PMSF, and 1µg/mL leupeptin) and sonicated 3 times with pulses of 50– 62 kHz for 40 s at time intervals of 20 s. The soluble protein extracts (SLpA and SLiA) were collected by centrifugation at 10,000 rpm for 20 min at 4◦C. The protein concentration was determined using the Micro BCA Protein Assay Kit (Thermo Fisher Scientific), and the proteins were visualized by 10% SDS-PAGE followed by Coomassie blue staining (Gibco).

# Cell Proliferation Assay

PBMCs (2 × 10<sup>5</sup> cells/well) from the subjects under study were split in 96-well flat-bottom plates containing 200 µL/well RPMI-20% iFBC medium in the presence of the recombinant PFR1 antigen (5µg/mL), SLpA (10µg/mL), SLiA (10µg/mL), or 2µg/mL concanavalin A (ConA) in triplicate. Polymyxin B (50 U/mL) was added to a parallel set of wells. The plates were incubated at 37◦C in a CO<sup>2</sup> atmosphere for 120 h. The cells stimulated with SLpA or SLiA were pulsed with BrdU (10µM) and incubated for 20 h at 37◦C. Cell proliferation was determined using a nonradioactive ELISA technique, according to the manufacturer's instructions (Cell Proliferation ELISA BiotrakTM System, version 2, GE Healthcare, AmershamTM, UK). The results were expressed as optical density (OD) values, and the stimulation index (SI) was calculated using the following formula:

$$SI = \frac{\left[ \text{OD (stimulated culture)} - \text{OD (control culture)} \right]}{\text{OD (control culture)}}$$

The cells stimulated with rPFR1 were pulsed with methyl-3H] thymidine (0.5 µCi/well) and incubated for an additional 20 h at 37◦C. The cells were immobilized in glass fiber filter mats using a FilterMate harvester (Perkin Elmer). <sup>3</sup>H incorporation was measured in a Wallac 1450 MicroBeta counter device. The results were expressed as counts per minute (cpm), and the stimulation index (SI) was calculated using the following formula:

SI = - cpm (stimulated culture) - cpm (control culture) cpm (control culture)

# Monoclonal Antibodies for Cell Surface Staining

The following conjugated antibodies were included in the panel for cell surface staining: anti-CD3-Pacific Blue (clone UCHT1), anti-CD4-AlexaFluor700 (clone RPA-T4), anti-CD8 allophycocyanin-H7 (clone SK1), anti-CD27-FITC (clone M-T271), anti-CCR7-PE (clone 150503), anti-CD45RA-PerCP-Cy5.5 (clone HI100), anti-CD57-APC (clone NK-1), anti-2B4- FITC (clone 2-69), anti-TIM-3-PE-CF594 (clone 7D3), anti-CD160-AlexaFluor647 (clone BY55) (all from BD Pharmingen, San Diego, CA), and anti-PD-1-PE (clone J105; eBioscience, San Diego, CA). For the intracellular staining, the following conjugated antibodies were used: anti-CTLA-4-PE-Cy5 (clone BNI3), anti-IFN-γ-PE-Cy7 (clone B27), anti-granzyme B-PE-CF594 (clone GB11), anti-IL-2-APC (clone MQ1-17H12), anti-TNF-α-AlexaFluor488 (clone Mab11) (all from BD Pharmingen, San Diego, CA), and anti-perforin-PE (Clone B-D48; Abcam, Cambridge, U.K.). Live cells were isolated by using the Fixable Aqua Dead Cell Stain viability marker (LIVE/DEAD) (Invitrogen, Eugene, OR).

# Detection of Intracellular Cytokines, Cytotoxic Molecules, and Inhibitory Receptors by Flow Cytometric Assays

PBMCs were cultured in RPMI 1640 supplemented with 2 mM L-glutamine, 10% iFBS, and 50µg/mL gentamicin. PBMCs were cultured at 1 × 10<sup>6</sup> cells/mL with anti-CD28 (1µg/mL) and anti-CD49d (1µg/mL) antibodies (purified clones CD28.2 and 9F10, BD Pharmingen, San Diego, CA) in the presence of SLpA (1µg/mL), in the absence of SLpA (considered the basal response) or in the presence of 10µg/mL L. infantum recombinant PFR1 protein (with 30µg/mL of polymyxin B). The cells were incubated for 14 h at 37◦C in a humidified atmosphere with 5% CO2. For intracellular staining, cells were cultured in the presence of GolgiPlug and GolgiStop according to the manufacturer's instructions (BD Pharmingen, San Diego, CA). After stimulation, the cells were washed and incubated in PBS (containing 5% iFBS) for 10 min at RT. PBMCs were stained with a viability marker, namely, LIVE/DEAD Fixable Aqua (Invitrogen, Eugene, OR), for 20 min in the dark at RT. Prior to the addition of the surface staining antibodies, the cells were washed three times with PBS (containing 5% iFBS). For the memory and phenotypic characterizations, the cells were stained with anti-CD3, anti-CD4, anti-CD8, anti-CD45RA, anti-CCR7, anti-CD27, and anti-CD57 antibodies for 20 min in the dark at 4◦C. To analyze inhibitory receptor expression, the cells were stained with anti-CD3, anti-CD4, anti-CD8, anti-PD-1, anti-2B4, anti-TIM-3, and anti-CD160 antibodies for 20 min in the dark at 4◦C and washed with PBS (containing 5% iFBS). Then, the cells were fixed and permeabilized with Cytofix/Cytoperm (BD Biosciences) for 20 min at 4◦C, washed following the manufacturer's instructions and stained with anti-CTLA-4 for 30 min in the dark at 4◦C. To evaluate the intracellular production of cytokines and cytotoxic molecules, cells were stained with anti-CD3, anti-CD4, and anti-CD8 antibodies. After washing, the cells were fixed and permeabilized with Cytofix/Cytoperm (BD Biosciences). Intracellular staining was performed with anti-granzyme B, anti-IFN-γ, anti-IL-2, anti-TNF-α, and anti-perforin antibodies for 30 min at 4◦C. Finally, after the last staining step, the cells were washed, suspended in 1 × PBS and analyzed in an LSRFortessa cytometer (BD Biosciences). At least 100,000 events were acquired, and analyses were performed using FlowJo 9.3.2 (TreeStar, Ashland, OR), Pestle 1.7 and SPICE 5.3 (National Institutes of Health, Bethesda, MD). Positivity for each marker was determined using fluorescence minus one (FMO) controls. Each antibody was titrated prior to the assays. Dead and doublet cells were excluded from the analysis.

# Statistical Analysis

The statistical analysis was carried out using Prism version 6.0 (GraphPad Software, La Jolla, CA). Statistical significance was calculated using nonparametric tests, such as the Mann–Whitney U-test (pairwise comparisons) or the Kruskal–Wallis test with the Dunn correction to identify differences between groups of cellular subpopulations or between more than two groups of subjects (ANOVA was used for comparisons among > 2 groups). Differences were considered statistically significant for p < 0.05, and the symbols used are <sup>∗</sup>p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001. The box plot graphs were generated by GraphPad Prism version 6.0 software and represent all values (minimum to maximum). The boxes represent the 25–75th percentiles.

# RESULTS

# Leishmanial Antigens and Lymphoproliferative Responses

The total soluble Leishmania proteins (SLpA and SLiA) and the recombinant PFR1 protein (rPFR1) used in the assays are shown in **Supplementary Figure 1**. A single and intensely stained band corresponding to the PFR1 antigen with an expected molecular mass of approximately 70 kDa was observed (lane 3). The purity was higher than 95%, as determined by a band densitometry analysis following Coomassie blue staining. PBMCs from patients with active leishmaniasis showed a proliferative response after stimulation with SLpA or SLiA, with proliferation indexes varying between 4.4 and 0.8 and between 4.3 and 0.8, respectively (**Table 1**). PBMCs from asymptomatic subjects with a positive Montenegro test showed proliferation indexes between 3.1 and 0.7 when stimulated with SLpA and between 1.5 and 0.1 when stimulated with SLiA (**Table 1**). The proliferation indexes of PBMCs from cured patients ranged within tighter intervals, between 1.3 and 0.1 and between 0.6


The data show the proliferation index range after the stimulation of PBMCs with soluble protein extracts of L. panamensis (SLpA), L. infantum (SLiA) and the recombinant PFR1 protein (rPFR1).

<sup>a</sup>Asymptomatic, asymptomatic subjects with positive Montenegro skin test. Cured, patients cured of cutaneous leishmaniasis. Active CL, patients with active cutaneous leishmaniasis. \*Nd, not detected.

All experiments were performed in triplicate.

The cut-off point is 0. Values above 0 in the three replicates are considered positive values. Values 0 are referred as not detected (Nd).

and 0.02 after stimulation with SLpA or SLiA, respectively (**Table 1**). A proliferative response was also detected after the stimulation of PBMCs with rPFR1. This response was stronger in cells from patients with active CL, with values ranging from 2.3 to 0.4, than in PBMCs from cured patients, with values between 0.8 and 0.02 (**Table 1**). Stimulation with rPFR1 did not lead to a detectable proliferative response in PBMCs from asymptomatic subjects. In general terms, and in spite of the limitation regarding to the wide range of proliferative response after antigen stimulation (SLpA SLiA or rPFR1), the obtained results allowed to detect which group of the subject responded with a higher and lower value of the proliferative index.

# Phenotypic Characterization of Antigen-Specific Memory CD4<sup>+</sup> and CD8<sup>+</sup> T Cells

To evaluate the existence of a possible association between the distribution of antigen-specific memory T cells and the clinical status of CL patients, we analyzed the phenotypes of CD4<sup>+</sup> and CD8<sup>+</sup> T cells from PBMCs isolated from all recruited subjects except the HD (13 individuals). The expression of CD45RA, CCR7, CD27, and CD57 in PBMCs stimulated with either SLpA or rPFR1 was analyzed by flow cytometry after staining with conjugated antibodies.

As shown in **Figure 1A**, within the subset of SLpA-specific CD4<sup>+</sup> T cells from either asymptomatic individuals with a positive Montenegro test or cured patients, the proportion of cells with a TNAIVE phenotype (CD45RA+CD27+CCR7+) was significantly higher than that of cells with a terminal effector memory phenotype (TEMRA cells, CD45RA+CD27−CCR7−) (p < 0.05). This pattern was also observed in SLpA-specific CD8<sup>+</sup> T cells from both groups of subjects (**Figure 1B**), although the difference was not statistically significant. Patients with active leishmaniasis showed a higher percentage (statistically nonsignificant) of SLpA-specific CD8<sup>+</sup> T cells with a TEMRA phenotype than of cells with a TNAIVE phenotype. Remarkably, the frequency of TEMRA CD8<sup>+</sup> T cells was greater in patients with active CL than in cured patients (**Figure 1B**). In addition, the frequency of CD4<sup>+</sup> and CD8<sup>+</sup> effector memory T cells (TEM, CD45RA−CCR7−CD27−) was significantly higher than the frequency of CD4<sup>+</sup> and CD8<sup>+</sup> central memory T cells (TCM, CD45RA−CCR7+CD27+) in both asymptomatic and cured patients (p < 0.05 and p < 0.01, respectively; **Figures 1A,B**).

Regarding the PFR1-specific CD4<sup>+</sup> T cells, a similar phenotypic profile with a significantly higher percentage of TNAIVE cells than of TEMRA cells was observed in the three groups of subjects (p < 0.05 and p < 0.01; **Figure 1C**). Interestingly, in PFR1-specific CD8<sup>+</sup> T cells, the difference between the proportions of TNAIVE and TEMRA cells was less pronounced in patients with active CL (**Figure 1D**) than in patients in the other two groups. Within this subset of CD8<sup>+</sup> T cells, statistically significant differences between the proportions of TNAIVE and TEMRA cells were detected only in the group of cured patients (p < 0.01; **Figure 1D**). Furthermore, in this group of patients, a significantly higher frequency of CD8<sup>+</sup> T cells with a TEM phenotype than of CD8<sup>+</sup> T cells with a TCM phenotype was observed (p < 0.01; **Figure 1D**).

Since the expression of the CD57 marker has been associated with an irreversible senescence state (replicative senescence) in cells, which is more prevalent in effector memory T cells and especially in those with a highly differentiated phenotype (Xu and Larbi, 2017), CD57 expression was evaluated in effector memory T cells (TEM and TEMRA). The results indicate that the frequency of CD4<sup>+</sup> and CD8<sup>+</sup> T cells expressing CD57 tended to be higher (although not statistically significant) in TEMRA cells from patients with active CL than in TEMRA cells from either asymptomatic individuals with a positive Montenegro test or cured patients (**Supplementary Figure 2**). Remarkably, this difference was statistically significant within both subsets of SLpA- and PFR1-specific CD8<sup>+</sup> T cells when the evaluation was made by considering the absolute number of CD57<sup>+</sup> cells within the subset of TEMRA cells (p < 0.05; **Figure 2**). The CD4<sup>+</sup> T cells exhibited a similar behavior, although the observed differences were not statistically significant.

# Functional Capacity of Circulating CD4<sup>+</sup> and CD8<sup>+</sup> T Cells in Response to Leishmania Antigens

To determine if there was a relationship between the pattern of functional responses and disease status, the functional responses of antigen-specific CD4<sup>+</sup> and CD8<sup>+</sup> T cells from asymptomatic subjects with a positive Montenegro skin test, patients cured of leishmaniasis and patients with active CL were evaluated. The production of intracellular cytokines (IL-2, IFN-γ, and TNF-α) and cytotoxic molecules (granzyme B and perforin) after stimulation with SLA from L. panamensis was determined. The results showed that the frequency of CD4<sup>+</sup> T cells with multifunctional capacity tended to be greater in both asymptomatic individuals and patients with active CL than in cured patients (**Figure 3**), based on the greater proportions of antigen-specific CD4<sup>+</sup> T cells performing four (1.3 and 3.6% vs. 0.9%, respectively) or three (10.5 and 15.8% vs. 6.5%, respectively) functions. Asymptomatic patients showed

According to CD45RA, CD27, and CCR7 expression, the cells were grouped as TNAIVE (CD8+CD45RA+CCR7+), TEMRA (CD8+CD45RA+CCR7−), TEM (CD8+CD45RA−CCR7−), and TCM (CD8+CD45RA−CCR7+). The boxes (25th−75th percentiles) and whisker plots (using the Tukey's method) show the median frequency and range of the CD4<sup>+</sup> and CD8<sup>+</sup> T cells. Statistical analyses were carried out using the Mann–Whitney U-test. Statistically significant differences were indicated by \*p < 0.05 and \*\*p < 0.01.

a higher frequency of bifunctional CD4<sup>+</sup> T cells (40.7%) and of CD4<sup>+</sup> T cells expressing both of the cytotoxic molecules, namely, granzyme B and perforin (**Figure 3A**). Remarkably, the evaluation of the multifunctional response in CD8<sup>+</sup> T cells led to different results. The frequency of multifunctional CD8<sup>+</sup> T cells was a greater although not statistically significant in cured patients than in either patients with active CL or asymptomatic patients. In fact, CD8<sup>+</sup> T cells performing five functions (granzyme B+, IFN-γ <sup>+</sup>, IL-2+, perforin+, and TNF-α <sup>+</sup>) were detected only in the group of cured patients (**Figure 3B**). Similarly, higher frequencies of CD8<sup>+</sup> T cells expressing four (3.3% vs. 1.7 and 0.8%) and three markers (18.5% vs. 6.7% and 7.5%) were observed in patients cured of CL than in either the asymptomatic Montenegro-positive individuals or those with active CL, respectively (**Figure 3B**). Furthermore, cured patients showed a greater frequency of CD8<sup>+</sup> T cells expressing IFNγ, perforin and granzyme B than did either asymptomatic Montenegro-positive individuals or patients with active CL (11.8 vs. 6.7 and 1.2%, respectively; **Figure 3**). In addition, while the subjects cured of leishmaniasis exhibited monofunctional cells

Montenegro skin test (Mont +, n = 4), patients cured of cutaneous leishmaniasis (Cured, n = 5) and patients with active cutaneous leishmaniasis (Active CL, n = 4). The boxes (25th−75th percentiles) and whisker plots (using Tukey's method) show the median and range of the CD4<sup>+</sup> and CD8<sup>+</sup> T cells. Statistical analyses were

that individually expressed all the analyzed molecules, CD8<sup>+</sup> T cells from patients with active CL preferentially expressed granzyme B or IL-2 (**Figure 3**).

carried out using the Mann–Whitney U-test. Statistically significant differences are indicated by \*p < 0.05.

# Expression of Inhibitory Receptors in Specific CD8<sup>+</sup> T Cells

As shown in **Figure 4**, patients with active CL had a higher proportion of CD4<sup>+</sup> T cells that simultaneously produced the cytokines IFN-γ, TNF-α, and IL-2 than did either asymptomatic individuals or cured patients (8.2 vs. 0.8%; **Figure 4A**). In cured patients, the monofunctional CD4<sup>+</sup> T cells mainly expressed TNF-α or IFN-γ (**Figure 4A**), while the frequency of CD8<sup>+</sup> T cells expressing three (1.7% vs. 0.5 and 0.4%) or two cytokines (14.7% vs. 9.5 and 5.1%) was higher than that in either asymptomatic individuals or patients with active CL. Conversely, patients with active CL showed a greater frequency of monofunctional CD8<sup>+</sup> T cells (94.6%) than did either asymptomatic or cured patients (90 and 83.6%, respectively; **Figure 4B**). Our results showed that the different groups of subjects under examination might be differentiated not only based on the frequency of multifunctional CD4<sup>+</sup> or CD8<sup>+</sup> T cells that produce several cytokines but also based on the patterns of the expressed cytokines. For example, compared with its production in either asymptomatic individuals or cured patients, IL-2 was mainly produced by monofunctional CD4<sup>+</sup> and CD8<sup>+</sup> T cell populations from patients with active CL (**Figures 4A,B**). Moreover, the frequency of CD8<sup>+</sup> T cells expressing both of the cytotoxic molecules, namely, granzyme B and perforin, was significantly lower in cured patients than in either asymptomatic subjects or patients with active CL (11.8% vs. 25.5 and 31.3%, respectively; **Figure 5**).

The next question to be evaluated was whether the functional response patterns of antigen-specific CD8<sup>+</sup> T cells were associated with the expression or coexpression of inhibitory receptors. Consequently, the expression of PD-1, CTLA-4, 2B4, CD160, and TIM-3 was evaluated ex vivo in PBMCs from 10 HD, 4 asymptomatic subjects, 5 patients cured of cutaneous leishmaniasis and 4 patients with active CL. PBMCs were stimulated with SLpA and stained using conjugated antibodies, as described in section Materials and Methods. Patients with active CL showed significantly higher frequencies of CD8<sup>+</sup> T cells expressing CD160 (p < 0.01), CTLA-4 (p < 0.05), PD-1 (p < 0.05), and TIM-3 (p < 0.01) than did HD (**Figure 6A**). The frequency of CD8<sup>+</sup> T cells expressing CD160 was also significantly higher in patients with active CL than in cured patients (p < 0.05). The same pattern of differences between these two pairs of patient groups was found for the frequency of CD8<sup>+</sup> T cells that expressed 2B4, although the differences were not statistically significant. Further examination of the differences among samples from the different subjects revealed that the frequency of CD8<sup>+</sup> T cells expressing CTLA-4 (p < 0.05 and p < 0.01) and TIM-3 (p < 0.01) was higher in both asymptomatic individuals and cured patients than in HD (**Figure 6A**). An evaluation of the concurrent expression of inhibitory receptors showed a significantly higher frequency of CD8<sup>+</sup> T cells that coexpressed two, three, or four inhibitory receptors in patients with active CL than in HD (p < 0.01; **Figure 6B**). There was a higher frequency of CD8<sup>+</sup> T cells that coexpressed four

subjects with a positive Montenegro skin test (Mont +, n = 4), patients cured of cutaneous leishmaniasis (Cured, n = 5) and patients with active cutaneous leishmaniasis (Active CL, n = 4). This profile was determined using a five-function assay to simultaneously measure the expression of IFN-γ, IL-2, TNF-α, granzyme B, and perforin after stimulation with SLA from L. panamensis. The color of each portion of the pie charts depicts the number of molecules produced by CD4<sup>+</sup> and CD8<sup>+</sup> T cells in response to L. panamensis antigens. The arcs of the pie charts represent the proportions of cells expressing each of the analyzed molecules. The percentage of unstimulated CD4<sup>+</sup> and CD8<sup>+</sup> T cells expressing these molecules (basal response) was subtracted from the value obtained following antigen stimulation.

molecules in asymptomatic individuals than there was in HD (p < 0.01; **Figure 6B**).

# DISCUSSION

Understanding the immunological mechanisms that underlie disease outcomes after the infection of a subject with Leishmania (Viannia) is essential given that there are asymptomatic patients, parasite persistence after treatment (de Oliveira Camera et al., 2006; Figueroa et al., 2009) and disease reactivation (Saravia et al., 1990). It has been reported that the clinical cure of leishmaniasis is mainly dependent on T cell populations that secrete Th1-related cytokines (Coutinho et al., 1998; Da-Cruz et al., 2005). Most knowledge obtained from the development of vaccines against CL comes from experimental infections with Leishmania major in murine models and cannot be fully extended to the subgenus Viannia. Likewise, clinical studies indicate that the human response to infection with Leishmania (Viannia) parasites differs from the response to infection caused by other members of the Leishmania subgenus (Laskay et al., 1991; Melby et al., 1994; Silveira et al., 2009). In addition, an evaluation of the treatment efficacy of several drugs against CL showed that healing is a dynamic process that involves changes in the frequency and quality of antigen-specific T cells (Da-Cruz et al., 1994; Lakhal-Naouar et al., 2015). These facts reinforce the importance

of evaluating the quality of the immune response triggered by infection with Leishmania (Viannia) species (De Luca and Macedo, 2016).

In the present manuscript, the phenotype and functionality of CD4<sup>+</sup> and CD8<sup>+</sup> T cells from CL patients, cured patients and asymptomatic individuals with a positive Montenegro skin test were characterized. In addition, the existence of a possible CD8<sup>+</sup> T cell exhaustion process associated with the evolution of Leishmania infection was evaluated. The results showed that cells from these patients exhibited lymphoproliferative capacity following stimulation with both SLpA and SLiA and, to a lesser extent, with the rPFR1 protein. In active CL patients, a similar proliferative response was observed in T cells stimulated with either SLpA or SLiA. In cured and asymptomatic patients, the greatest proliferative response was observed following cell stimulation with SLpA. The cell proliferation assay with soluble antigens of L. infantum showed a proliferative response in all subjects, although with lower values compared to those obtained with SLpA. The proliferative capacity of the subjects against the antigens of L. infantum, was performed as the rPFR1 protein came from L. infantum. These data are expected since all patients were infected with L. panamensis, the most prevalent infecting strain in the area where the patients under study live (Corredor et al., 1990; Ovalle et al., 2006; Ramírez et al., 2016). Consequently, subsequent cellular assays were performed using SLpA as the stimulation agent.

The analysis of the phenotype of the stimulated CD4<sup>+</sup> and CD8<sup>+</sup> T cells from asymptomatic individuals and cured patients showed that TNAIVE cells, which represented 30–40% of the total T cell population, were the subset with the highest

FIGURE 5 | Cytotoxic profile of Leishmania-specific CD8<sup>+</sup> T cells. Functional activity of CD8<sup>+</sup> T cells determined by perforin and granzyme B expression after stimulation with SLA from L. panamensis in asymptomatic subjects with a positive Montenegro skin test (Mont +, n = 4), patients cured of cutaneous leishmaniasis (Cured, n = 5) and patients with active cutaneous leishmaniasis (Active CL, n = 4). The colors in the pie charts depict the proportion of cells expressing each of the analyzed molecules.

(25th−75th percentiles) and whisker plots (using Tukey's method) show the median and range of the frequency of inhibitory receptors on T cells. Statistical analyses were carried out using the Mann–Whitney U-test. Statistically significant differences are indicated by \*p < 0.05, \*\*p < 0.01, and \*\*\*p < 0.001.

frequency, mainly following stimulation with rPFR1. The effector memory (TEM) population represented ∼10% of the total T cells, which was higher than the frequency of central memory T cells (TCM). This pattern was observed in T cells stimulated with both the soluble L. panamensis antigen (SLpA) and the L. infantum rPFR1 antigen. These central and effector memory T cell profiles have been previously reported in other parasitic and viral human infections. Thus, asymptomatic and cardiac Chagas disease patients have a significantly higher proportion of TEM cells than TCM cells in the population of CD4<sup>+</sup> and CD8<sup>+</sup> T cells (Fiuza et al., 2009) both ex vivo and after stimulation with T. cruzi antigens. A similar profile was also reported in memory HIVspecific CD8<sup>+</sup> T cells, for which TEM cells comprised 71.8% of the population and TCM cells comprised 4.1% (Champagne et al., 2001). It has been reported that after resolution of Leishmania infection, a small number of persistent parasites can act as source of continuous antigen stimulation to host immune cells (Sacks and Noben-Trauth, 2002), which leads to the maintenance of effector memory T cells (Aebischer et al., 1993; Mendonça et al., 2004; Okwor and Uzonna, 2008). These effector memory T cells can be maintained for prolonged durations, exhibiting increased resistance to apoptosis and playing a key role in protection against reinfection (Sallusto et al., 2004). Correspondingly, it has been reported that both central and effector memory T cells mediate long-term resistance to leishmaniasis (Zaph et al., 2004). The T cells from the studied patients with active CL showed a higher proportion of effector memory T cells (TEM for CD4<sup>+</sup> T cells and TEMRA for CD8<sup>+</sup> T cells) than TNAIVE cells after stimulation with SLpA. These profiles are consistent with a shift in the activation phenotype associated with the persistence of the lesions. However, the patients with active CL had similar frequencies of CD4<sup>+</sup> TNAIVE and TEM cells after stimulation with rPFR1. The rPFR1-specific CD8<sup>+</sup> T cells presented a slightly higher proportion of TNAIVE cells than of TEMRA or TEM cells. The higher frequency of effector cells detected in patients with active CL than in either asymptomatic or cured patients could be caused by the existence of a high parasite load and the consequent activation of an antigen-specific immune response. In this context, it has been reported that chronic Chagas disease patients in the asymptomatic stage have a higher proportion of TNAIVE antigen-specific CD8<sup>+</sup> T cells, while chronic Chagas patients with cardiac symptomatology present a greater frequency of effector memory cells (TEM and TEMRA) (Egui et al., 2015). The T cell-mediated immune response has also been analyzed in the skin lesions of CL patients by other authors. Thus, in CL patients infected with Leishmania braziliensis, a higher percentage of CD4<sup>+</sup> and CD8<sup>+</sup> TEM cells than others T cell subsets (TNAIVE and TCM cells) was found at the lesion level than in the blood (de Oliveira Mendes-Aguiar et al., 2016). The higher percentage of these cells together to activated cytotoxic cells was taken as an indication of the contribution of TEM to immunemediated tissue damage. It is known that the proliferative capacity of effector T cells expressing CD57 is severely and irreversibly compromised. Interestingly, the number of antigenspecific CD8<sup>+</sup> CD57<sup>+</sup> TEMRA cells from patients with active CL was statistically significantly higher than that observed in cured patients. Likewise, a significantly higher number of CD4<sup>+</sup> TEMRA cells that expressed the CD57 molecule was found in patients with active CL than in cured patients and asymptomatic subjects. These TEMRA cells, which have a terminally differentiated phenotype, are very susceptible to apoptosis and produce a large amount of cytotoxic molecules (Khamesipour et al., 2012). The expression of CD57 has been strongly correlated with the simultaneous production of granzyme and perforin (Chattopadhyay et al., 2009). This phenotypic pattern is consistent with the cytotoxic profile observed in antigen-specific CD8<sup>+</sup> T cells from patients with active CL. These results suggest that persistent exposure to the parasite might push the effector cells toward a terminally differentiated and senescent cell phenotype. In active leishmaniasis, the prolonged parasite presence would lead to persistent cytotoxic activity against intracellular protozoa, which could lead to a higher prevalence of senescent effector CD8<sup>+</sup> T cells. The results show that CD4<sup>+</sup> T cells exhibit a similar behavior, although these differences are not statistically significant. This fact may be related to the CD8<sup>+</sup> T cells, as they are the cells endowed with cytotoxic activity, which is the main characteristic of CD8<sup>+</sup> T cells.

CD8<sup>+</sup> T cells from cured patients showed a higher multifunctional capacity than did those from patients with active CL. A multifunctional capacity of antigen-specific CD8<sup>+</sup> T cells, which was only observed in cured patients, was indicated by the ability of the cells to simultaneously perform the five examined functions (IFN-γ <sup>+</sup>, TNF-α <sup>+</sup>, IL-2+, granzyme B+, and perforin+). In addition, cured patients showed a higher proportion of CD8<sup>+</sup> T cells expressing three or four of these cytokines and cytotoxic molecules. Similarly, cured patients, in particular, showed a higher frequency of CD8<sup>+</sup> T cells expressing IFN-γ <sup>+</sup> and TNF-α <sup>+</sup> than did patients with active CL, suggesting a relevant role of these cytokines in infection control. In murine models of leishmaniasis, CD8<sup>+</sup> T cells have been shown to exert a curative role, which has been attributed to the production of Th1-related cytokines such as IFN-γ (Ruiz and Becker, 2007; Jayakumar et al., 2011). Moreover, IFN-γ-deficient mice were more susceptible to Leishmania infection, with larger lesions and higher parasite burdens than their wild-type counterparts (Pinheiro and Rossi-Bergmann, 2007). Additionally, it has been reported that other mediators produced by CD8<sup>+</sup> T cells (such as perforin, granzymes, and chemokines) contribute to host defense. Conversely, an exacerbation of the CD8<sup>+</sup> T cell immune response and the high production of cytotoxic molecules has been associated with tissue damage (Santos Cda et al., 2013). In fact, it has been reported that the CD8<sup>+</sup> T cell immune response is involved in both the resolution of disease and the pathology caused by Leishmania (Viannia) infection in humans, highlighting the functional heterogeneity of these cells (Coutinho et al., 1998; Bangham, 2009). Interestingly, herein, it was shown that only cured patients showed a Th1-type cytotoxic profile, as indicated by the simultaneous expression of cytotoxic molecules and IFN-γ (CD8<sup>+</sup> T cells, IFN-γ <sup>+</sup>granzyme B+perforin+). It is well-documented that resistance to leishmaniasis is related to the development and production of Th1-type proinflammatory cytokines, which leads to the activation of macrophages and to parasite killing (Sacks and Noben-Trauth, 2002; Von Stebut et al., 2003). While cytotoxic activity induces target cell death, Th1 cytokines (such as IFN-γ and TNF-α) are involved in the development of an inflammatory response that modulates the activity of macrophages and dendritic cells. However, when these pathways are not properly regulated, inflammatory disorders and tissue damage could develop (Ribeiro-de-Jesus et al., 1998; Follador et al., 2002; Arias et al., 2014). At the lesion level, there is elevated expression of granzyme B, perforin and CD107a (Novais et al., 2013; Cardoso et al., 2015), which has been positively correlated with lesion size (Santos Cda et al., 2013). Correspondingly, CD8<sup>+</sup> T cells from patients with active CL presented a higher expression level of cytotoxic molecules (granzyme B+perforin+) than did those from cured patients or asymptomatic subjects.

In the context of CD4<sup>+</sup> T cells, different functional profiles of the immune response were detected among the different groups of studied patients. Thus, a higher frequency of CD4<sup>+</sup> T cells with multifunctional activity was detected in patients with active CL than in cured patients. However, cured patients had more monofunctional cells expressing cytokines related to the Th1 immune response, such as IFN-γ and TNF-α. These results are consistent with those of studies carried out in CL patients, which showed that after treatment, there is a reduction in both the frequency of multifunctional CD4<sup>+</sup> T cells and the amount of cytokines secreted by these cells (Lakhal-Naouar et al., 2015). In fact, the depletion of effector CD4<sup>+</sup> T cells in murine models did not appear to influence resistance to infection, while the depletion of CD8<sup>+</sup> T effector cells reversed the protective effect, which led to a Th2 cytokine environment related to infection susceptibility (Jayakumar et al., 2011). Although little is known regarding the role of the cytotoxic activity of CD4<sup>+</sup> cells in Leishmania infection, it has been reported that a high frequency of CD4<sup>+</sup> T cell subsets producing granzyme B is found in individuals who previously had contact with L. major (Naouar et al., 2014). Consistently, in the present study, the production of cytotoxic molecules by CD4<sup>+</sup> T cells was mainly detected in asymptomatic subjects. However, further work is needed to determine the potential roles of these cells and to understand the implications for protection and damage.

It has been reported that the healing process of CL lesions at the end of therapy is associated with an increase in the proportion of L. braziliensis-reactive-CD8<sup>+</sup> T cells and a decline in CD4+-specific-reactive T cells (Da-Cruz et al., 2002). Consistent with these findings, the results of this study show a higher proportion of multifunctional L. panamensisspecific CD8<sup>+</sup> T cells in treated cured patients than in asymptomatic subjects and active CL patients. However, a lower frequency of multifunctional CD4<sup>+</sup> T cells was found in cured patients than in asymptomatic subjects and active CL patients.

In chronic diseases, the persistence of pathogen-derived antigens induces an increase in the expression of inhibitory receptors and a gradual loss of the response capacity of antigenspecific T cells. This process, known as the exhaustion process, has been mainly described in CD8<sup>+</sup> T cells (Kahan et al., 2015). Very little is known about the exhaustion process in infections caused by Leishmania (Viannia) subgenus species. In the context of Chagas disease, this process has been well-documented in CD8<sup>+</sup> T cells, with a higher level of dysfunction associated with a severe stage of the disease (Lasso et al., 2015). It was also recently reported that anti-Trypanosoma cruzi treatment partially reverses the exhaustion process in CD8<sup>+</sup> T cells, thus improving antigen-specific functionality (Mateus et al., 2017). The results obtained in the present study showed higher expression and coexpression of inhibitory receptors in CD8<sup>+</sup> T cells from patients with L. panamensis infection than in HD. Patients with active CL showed higher expression of CD160 and PD-1 than did cured patients and asymptomatic subjects. A significantly higher coexpression of inhibitory receptors by CD8<sup>+</sup> T cells was only observed when patients with active CL were compared with HD. The relatively lower level of inhibitory receptor coexpression in cured patients than in patients with active CL, which was associated with the high multifunctional capacity of the CD8<sup>+</sup> T cells of cured patients, suggests the critical importance of the functional response of CD8<sup>+</sup> T cells in the healing process. Even the impaired or dysfunctional CD8<sup>+</sup> T cell response induced by the exhaustion process and/or the senescent cell phenotype could contribute to the maintenance of an active lesion. Thus, in experimental model of CL, mice lacking PD-L1 were markedly more resistant to infection, and the lesions they developed were smaller (Liang et al., 2006). Furthermore, it has been reported in diffuse cutaneous leishmaniasis that an increase in PD-L1 expression by monocytes is a possible mechanism used by the parasites to evade the immune response (Barroso et al., 2018). In patients with visceral leishmaniasis, elevated expression of the CTLA-4 and PD-1 inhibitory receptors has been reported (Esch et al., 2013; Gautam et al., 2014; Chiku et al., 2016). Interestingly, the expression of such markers decreased in patients with Leishmania infection after treatment (Gautam et al., 2014). In this context, the high production of IFN-γ and TNF-α by CD8<sup>+</sup> T cells and the greater proportion of multifunctional CD8<sup>+</sup> T cells observed in patients cured of CL could be associated with the lower expression of PD-1 found in these patients, which might be related to a partial reversion of the exhaustion process induced by treatment. Consistent with that theory, in experimental models of visceral leishmaniasis, it has been reported that the blockade of PD-1 reduces the parasite burden, restores the T cell proliferation capacity and increases the production of IFN-γ and TNF-α (Mou et al., 2013; Chiku et al., 2016).

The findings presented in this work improve our understanding of the poorly elucidated exhaustion process that occurs in the context of L. panamensis infection. In addition, the data provide evidence that antigen-specific immune response markers can be correlated with the clinical status of human patients with CL caused by L. panamensis and might consequently be a useful tool for facilitating the clinical follow-up of patients with CL.

# AUTHOR CONTRIBUTIONS

MCT carried out the conceptualization. MCT and MCL designed the study. DL, AM, AE, EP-A, and IG obtained the samples and materials. AE, DL, and EP-A performed the experiments. AE, EP-A, DL, and MCL analyzed the results. AE and MCL performed the data visualization. AE, EP-A, MCL, and MCT discussed the results. JI, MCL, MCT, and IV acquired funding. MCL and SR supervised the study. DL, AE, and EP-A wrote the original draft. MCT, MCL, and JI wrote and edited the final version of the manuscript. All authors read and approved the final manuscript.

# FUNDING

This work was supported by grants RTC-2016-50005-1, SAF2016-80998-R, and SAF2016-81003-R from the Programa Estatal I+D+i (MINECO) and by the Network of Tropical Diseases Research RICET (RD16/0027/0005) funded by ISCIII and FEDER.

# ACKNOWLEDGMENTS

We thank the patients and healthy volunteers who participated in this study. We are very grateful to A. López-Barajas from IPBLN-CSIC for her technical assistance. DL was supported by an FPU predoctoral fellowship (FPU-AP2012-1662) from the Spanish Ministry of Education, Culture, and Sport, and EP-A was supported by a predoctoral fellowship from FUNCCET.

# REFERENCES


# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2018.00393/full#supplementary-material


**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 Egui, Ledesma, Pérez-Antón, Montoya, Gómez, Robledo, Infante, Vélez, López and Thomas. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Serum Levels of Soluble CD40 Ligand and Neopterin in HIV Coinfected Asymptomatic and Symptomatic Visceral Leishmaniasis Patients

Wim Adriaensen<sup>1</sup> \*, Saïd Abdellati <sup>1</sup> , Saskia van Henten<sup>1</sup> , Yonas Gedamu<sup>2</sup> , Ermias Diro<sup>2</sup> , Florian Vogt <sup>1</sup> , Bewketu Mengesha<sup>2</sup> , Emebet Adem<sup>2</sup> , Luc Kestens <sup>3</sup> and Johan van Griensven<sup>1</sup>

### Edited by:

José (Pepe) Alcami, Instituto de Salud Carlos III, Spain

### Reviewed by:

Alejandro Vallejo, Instituto Ramón y Cajal de Investigación Sanitaria, Spain Jose Angelo Lauletta Lindoso, Institute of Tropical Medicine, University of São Paulo, Brazil

> \*Correspondence: Wim Adriaensen wadriaensen@itg.be

### Specialty section:

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 29 March 2018 Accepted: 28 November 2018 Published: 11 December 2018

### Citation:

Adriaensen W, Abdellati S, van Henten S, Gedamu Y, Diro E, Vogt F, Mengesha B, Adem E, Kestens L and van Griensven J (2018) Serum Levels of Soluble CD40 Ligand and Neopterin in HIV Coinfected Asymptomatic and Symptomatic Visceral Leishmaniasis Patients. Front. Cell. Infect. Microbiol. 8:428. doi: 10.3389/fcimb.2018.00428 <sup>1</sup> Unit of NTDs, Department of Clinical Sciences, Institute of Tropical Medicine, Antwerp, Belgium, <sup>2</sup> Department of Internal Medicine, Leishmaniasis Research and Treatment Centre, University of Gondar, Gondar, Ethiopia, <sup>3</sup> Unit of Immunology, Department of Biomedical Sciences, Institute of Tropical Medicine, Antwerp, Belgium

Human Immunodeficiency Virus (HIV) co-infection drastically increases the risk of developing overt visceral leishmaniasis (VL). The asymptomatic Leishmania infection window constitutes an opportunity to identify those HIV patients at highest risk by defining early markers associated with disease susceptibility or resistance. As intracellular parasite killing is essential, we investigated whether serum markers of macrophage activation were notably affected in HIV patients with an asymptomatic Leishmania infection or overt visceral leishmaniasis disease. Serum levels of soluble CD40 ligand and neopterin were assessed in 24 active VL-HIV patients, 35 HIV patients with asymptomatic Leishmania infection and 35 HIV endemic controls. All patients were recruited in L. donovani endemic regions of North-West Ethiopia. The serum levels of sCD40L and neopterin significantly decreased and increased in HIV patients with active VL compared to HIV patients with asymptomatic Leishmania infection, respectively. No statistically significant differences could be detected in neopterin and sCD40L levels between Leishmania asymptomatically infected HIV patients and endemic HIV control patients. However, an inverse trend, between Leishmania antibody positivity or VL development and neopterin levels could be seen. The CD4+ T-cell count was inversely correlated with serum neopterin levels, but not with sCD40L levels. Our results in HIV coinfected patients, correspond with the postulated protective role of sCD40L in VL and underline the importance of the CD40-CD40L pathway in resistance against the parasite. Neopterin levels suggest an increased macrophage activation upon infection and could have a value in clinical algorithms to, although non-specifically, improve prediction of VL development in HIV patients with asymptomatic Leishmania infection.

Keywords: visceral leishmaniasis, kala-azar, HIV, sCD40L, neopterin, asymptomatic

# INTRODUCTION

Human Immunodeficiency Virus-1 (HIV-1) has been identified as one of the emerging challenges for Visceral Leishmaniasis (VL) control, an important yet neglected vector-borne disseminated infection caused by the protozoan Leishmania donovani spp. complex (van Griensven et al., 2014b). The anthroponotic form of VL is caused by Leishmania donovani and is prevalent in the Indian subcontinent (300,000 cases/year) and East Africa (30,000 cases/year), mainly Sudan and Ethiopia (van Griensven and Diro, 2012). Untreated, overt disease is universally lethal. HIV is one of the strongest risk factors to develop VL. In contrast with the zoonotic Leishmania infantum endemic regions in Europe, where introduction of anti-retroviral therapy (ART) resulted in a significant reduction in the incidence of VL-HIV (Desjeux and Alvar, 2003), scaling-up of ART did so far not yield similar effects in East-Africa. In Ethiopia, close to 30% of patients with VL are co-infected with HIV, and an increasing number and proportion of VL cases are now seen in individuals on ART, including primary VL episodes (Diro et al., 2014). In addition, once the infection has evolved to active VL in HIV patients (typically within 6–9 months after infection) it is characterized by low cure rates, higher drug toxicity, frequent VL relapse and high case-fatality rates (van Griensven et al., 2010). The asymptomatic Leishmania infection window constitutes an opportunity to define early markers associated with disease control or progression (van Griensven et al., 2014a). To date, our knowledge on early immunopathology of VL is limited, and very scarce in HIV coinfected patients (Okwor and Uzonna, 2013). As both infections are clearly associated with immune deficiency, simple serum markers of a deteriorating immune response may allow an early detection of those at high risk for progression to VL. Chronic immune activation is a typical characteristic of HIV disease progression and several biomarkers also proved informative for coinfection progression (Sokoya et al., 2017). In a similar manner, VL showed to be an independent source of chronic immune activation in VL-HIV patients (Casado et al., 2015). Therefore, we investigated whether serum markers of immune activation, in particular macrophage activation, were notably affected in HIV patients with an asymptomatic Leishmania infection and overt VL disease.

One marker of interest is CD40 ligand (CD40L or CD154). This membrane glycoprotein is primarily expressed on activated CD4+ T-cells, platelets and a small proportion on CD8+ Tcells (Kornbluth, 2000). It binds and activates CD40 on antigen presenting cells, thereby enhancing the survival of the APC and promotes secretion of pro-inflammatory cytokines and synthesis of nitric oxide (NO) (Subauste, 2009). A T-helper 1 (Th1) cell-mediated immune response with high interferon(IFN)-y production activating macrophages to produce NO is reported to be protective in Leishmania-infected murine models (Rodrigues et al., 2016). Vice versa, a Th2-skewed response with high levels of IL-10 was shown to be detrimental. This dichotomy is not so clear in human VL patients, let alone in HIV coinfected patients (McMahon-Pratt and Alexander, 2004). Irrespective, the production of both IL-10 and IFN-y is dependent on this costimulatory pathway. Several studies in mice and human lymphocytes underlined the central role of the CD40-CD40L pathway in the generation of effective T-cell responses and protection against Leishmania and other parasitic infections (Subauste, 2009).

sCD40L, the soluble derivate of CD40L, is a functional trimer which retains its biological function after cleavage of the Tcell membrane, allowing it to interact with and activate cells expressing CD40, such as macrophages. This soluble form was associated with clinical resolution of VL (de Oliveira et al., 2013). On top of a gradual increase in serum sCD40L levels during treatment, levels were also negatively correlated with spleen size and parasite load. The same authors recently showed that sCD40L from sera of exposed subjects could indeed increase production of inflammatory cytokines and improve control of the parasite in human L. infantum infected macrophages (de Oliveira et al., 2015). The variation in sCD40L levels and its prognostic value in asymptomatic Leishmania infection or a concurrent HIV coinfection is unknown.

Increased serum levels of neopterin are associated with immune activation and showed to be one of the better soluble predictors of adverse outcomes in HIV patients (disease progression, ART activity or inflammation-associated comorbidities), at least comparable to that of the number of CD4+ T-cells (Nyamweya et al., 2012; Eisenhut, 2013; Bipath et al., 2015). Neopterin is a purine nucleotide derivate from guanosine triphosphate (GTP) and produced by human and primate IFN-y-activated macrophages (Hamerlinck et al., 2000). Hence, neopterin levels are increased in pathologies associated with a Th1 dominated immune response and usually correlate well with the disease stage. The fact that neopterin is produced by the common target cell of Leishmania and HIV (cf. macrophage), we investigated whether neopterin levels can be used as a marker of T-cell activation and produced oxidative stress inducing intracellular Leishmania parasite killing in HIV patients. Because previous studies have reported increased neopterin levels in the early phases of viral infections (e.g., EBV, CMV, and parvovirus B19), we investigated the asymptomatic Leishmania infection phase in particular (Reibnegger et al., 1988; Murr et al., 2002).

This study is the first to assess the association of serum sCD40L and neopterin concentrations with asymptomatic and symptomatic Leishmania infection status in HIV patients living in VL-endemic regions.

# METHODS

# Study Design and Population

All patients were recruited in L. donovani endemic regions of North-West Ethiopia (Abdurafi, Metema and Gondar). Active VL-HIV patients were selected from a pentamidine secondary prophylaxis clinical trial for VL relapse in HIV coinfected patients (NCT01360762) in which 24 patients had available serum samples from their baseline visit (before initial treatment). Baseline samples from 35 asymptomatic Leishmania antibody positive HIV patients and 35 endemic HIV controls with no antibodies against Leishmania were selected from an observational cohort study on asymptomatic Leishmania infection in HIV patients (NCT02839603). All 34 of 35 asymptomatic Leishmania antibody positive HIV patients remained disease free for a median of 12 months (IQR: 9–12) and one patient developed VL 9 months later. In contrast to CD4+ T-cell counts <200 cells/mL in the majority of active VL-HIV cases, we expected higher heterogeneity in HIV history and CD4+ T-cell counts in non-diseased Leishmania antibody positive and negative HIV patients that could affect sCD40L and neopterin serum concentrations. For this reason, non-diseased individuals with and without antibodies against Leishmania were individually matched on sex, months on ART, ART regimen, and CD4+ T-cell count. Antibody positivity was tested with rK39- Rapid diagnostic test (RDT) (Kalazar Detect Rapid Test, InBios International Inc., Seattle).

# Serum Markers of Macrophage Activation

Concentrations of human neopterin were measured by enzyme immunoassay (ELISA, IBL international, Germany), with an upper limit of 29,400 pg/mL (no left-over sample for further dilution). Likewise, concentrations of human sCD40L were measured in serum samples by enzyme immunoassay (ELISA, IBL international, Germany).

# Covariates

An antibody-detecting direct agglutination test (DAT, Institute of Tropical Medicine, Antwerp) was performed on all serum samples and a titer ≥1:200 was considered positive in case of asymptomatic infection and ≥1:6,400 in case of active VL-HIV patients. Three VL-HIV patients had missing DAT values. Urine samples were used to perform the KAtex urine antigen test (Kalon Biological Ltd, Guildford, UK). Four VL-HIV patients had missing KAtex values. Microscopy for malaria and parasitic infections was performed in whole blood and stool samples, respectively.

# Statistical Analyses

Continuous data are presented as medians and interquartile ranges (IQR). Categorical data are presented as numbers and frequencies. Comparisons between asymptomatic Leishmania antibody positive cases and active VL-HIV cases were performed using the chi-square test and Mann–Whitney U-test for continuous data. Comparisons between the matched HIV patients with and without Leishmania antibody positivity were performed using robust conditional logistic regression and McNewar Chi2 test. p < 0.05 was considered to be statistically significant. Dot plots are shown with median and IQR. Spike curves showed the individual change from case to control in each matched pair. Correlations between CD4+ T-cell counts and our markers of interest were plotted and the corresponding Pearson correlation coefficients were calculated. The statistical analyses were performed using STATA 14 (StataCorp, College Station, TX, United States) and GraphPad Prism 7 (GraphPad Software, San Diego, CA, United States).

# RESULTS

The matched case-control study consisted of 70 HIV patients living in a VL endemic area in North-West Ethiopia, 50% with confirmed positive antibody test against rK39 antigen. All 34 of 35 asymptomatic Leishmania antibody positive HIV patients remained disease free for a median of 12 months (IQR: 9–12) and one patient developed VL 9 months later. Of all cases and controls, 78.6% lived in the endemic area for more than 10 years and showed potential risk factors for Leishmania infection, with 66 (94.3%) having animals in or around the house, 49 (70%) were sleeping outside and most patients were male (85.7%) daily laborers of farmers working on the fields (78.3%) (**Table 1**). With respect to their HIV infection, the majority were on ART (91.4%) for more than 2 years (60%) with fairly good CD4+ T-cell counts (**Table 1**). In general, patients were rather malnourished with 40% having a body mass index below 18.5 kg/m<sup>2</sup> ; 14 (20.6%) had intestinal parasites in their stool.

Of the 35 Leishmania antibody positive asymptomatic cases, five (14.3%) had a previous VL episode (**Table 1**). Fifteen (42.9%) patients also tested antibody positive on DAT in serum and only one tested rK39 antigen positive on a latex agglutination test (KAtex) in the urine. Besides the infection markers, asymptomatic HIV patients with Leishmania antibody positivity were not statistically significantly different from HIV patients without Leishmania antibody positivity (**Table 1**).

Twenty-four HIV patients with overt VL disease from the same region were included in the study. The serum levels of sCD40L and neopterin significantly decreased and increased in active VL-HIV cases [325 pg/mL (IQR: 0–1097.5), 29,400 pg/mL (IQR: 13,550–29,400)] compared to asymptomatic Leishmania antibody positive HIV patients [2,490 pg/mL (IQR: 1,605–3,405), 4,610 pg/mL (IQR: 3,430–7,700)], respectively (**Figure 1**). The single asymptomatic Leishmania infected patient that developed VL 9 months later had sCD40L and neopterin concentrations of 1,955 and 29,400 pg/mL, respectively (hexagon in **Figure 1**). The VL-HIV patients consisted almost exclusively of male patients (95.8%), who were significantly younger (p = 0.036), had a lower CD4+ T-cell count (p = 0.001), with only 50% on ART (p < 0.001) with a shorter time on ART (p = 0.016), compared to the asymptomatic Leishmania antibody positive HIV patients (**Table 1**). No difference in serum immune activation markers could be detected in pre-ART and ART patients among active VL-HIV patients (**Table 2**). In contrast to sCD40L levels, where no association could be detected, neopterin levels were inversely associated with the CD4+ T-cell count (**Figure 3**).

To account for the higher heterogeneity in HIV history and CD4+ T-cell counts in non-diseased HIV patients with and without Leishmania antibodies, matched analyses were performed to be able to detect small differences due to the parasitic infection in the levels of sCD40L and neopterin independent from the concurrent HIV/ART stage. No significant differences could be detected in sCD40L and neopterin levels between Leishmania asymptomatically infected HIV patients and endemic HIV control patients (**Figure 1**). Because matched analyses were performed, group median values shown in **Figure 1** for non-diseased HIV patients with and without Leishmania antibodies are less informative. Alternatively, spike curves were plotted to show the changes in levels among case-control pairs (**Figure 2**). The median change from control to case in serum sCD40L and neopterin levels was −255 pg/mL (IQR: −1,445,

### TABLE 1 | Patient characteristics.


\*p-value comparing the non-infected cases against the asymptomatic cases by robust conditional logistical regression (continuous variables) or McNemar Chi2 test (categorical variables). \*\*p-value comparing the active cases (24 VL-HIV patients) against the asymptomatic cases (35 rK39+ patients) by chi2-test.

\*\*\*Matching criteria.

965) and 1,860 pg/mL (IQR: −550, 4,700), respectively. An inverse trend, although not statistically significant, between Leishmania antibody positivity and neopterin levels could be observed (**Figures 1**, **2**).

# DISCUSSION

Because the activation of macrophages for intracellular parasite killing is essential with regard to VL progression or resistance, we investigated the association of serum macrophage activation markers with the status of Leishmania infection in HIV coinfected patients.

Comparable levels of sCD40L were found in the serum of asymptomatic Leishmania antibody positive HIV patients and endemic healthy controls. In the latter, levels were slightly higher compared to previously published endemic healthy controls in Ethiopia (Gadisa et al., 2017). In particular, serum levels of sCD40L were significantly decreased in diseased VL-HIV

TABLE 2 | Median and interquartile ranges for sCD40L and neopterin levels with regard to ART status in active VL-HIV patients.


\*p-value Mann Whitney U–test.

patients. This is in clear contrast with previous studies that detected high levels of sCD40L in chronic hepatitis C HIV coinfected patients (Lapinski et al., 2014) and suggested high shedding of sCD40L due to T-cell turn-over as a marker of immune activation and disease, associated with T-cell exhaustion and poor prognosis of HIV infection (Kornbluth, 2000; Miller et al., 2015). Higher levels of sCD40L have also been reported in untreated HIV patients than in ART-treated HIV patients (Olmo et al., 2012), but this could not be confirmed in our VL-HIV

group (**Table 2**). Previously published data in VL patients also showed very low sCD40L levels (de Oliveira et al., 2013; Gadisa et al., 2017), supporting the hypothesis of a specific parasitedriven inhibition of the CD40 costimulatory pathway. L. major amastigotes were shown to modulate the CD40-CD40L pathway downstream by inducing ERK1/2 and IL-10 production, which inhibits the p38MAPK/IL12 pathway resulting in persistence of infection (Subauste, 2009; de Oliveira et al., 2015). A continuous loop could be created as it has also been reported that IL-10 is among those mediators that reduces sCD40L expression (Daoussis et al., 2004).

We believe sCD40L could induce a strong CD4+ Tcell independent activation of macrophages, especially in CD4+CD40L+ T-cell deprived HIV conditions, resulting in IFN-y and NO production followed by parasite clearance. In line with the proposed CD4+ T-cell independent activation, no correlation was found between sCD40L levels and the CD4+ Tcell count in our study population (**Figure 3**) nor in previous studies among HIV patients (Kalayjian et al., 2010; Lapinski et al., 2014). Moreover, de Oliveira and coworkers recently showed that sCD40L from sera of exposed subjects could indeed increase

production of inflammatory cytokines and improve control of the parasite in human L. infantum infected macrophages (de Oliveira et al., 2015). In addition, high levels of sCD40L in non-diseased non-HIV individuals living in high risk endemic settings in Brazil (however with unknown infection status), compared to very low levels in non-endemic controls, suggests a protective role of sCD40L in Leishmania infection and disease (de Oliveira et al., 2013). We obtained similar results in HIV coinfected controls.

In experimental models of leishmaniasis, strong CD40-CD40L signaling induced IL-12 production by macrophages whereas weak signaling induced IL-10 production (Mathur et al., 2006). The CD40-CD40L interaction thus seems to steer resistance or susceptibility to infection and indicates a potential relevance of sCD40L supplementing in control of the early infection. Although all monomeric, dimeric, and trimeric forms of soluble CD40L can bind to CD40, the soluble trimeric form of CD40L has the most potent biological activity through oligomerization of cell surface CD40 (Manzoor, 2015). For this reason, recombinant trimeric sCD40L could be considered as a potential adjuvant for a therapeutic or prophylactic vaccine approach in HIV patients. The latter is supported by the ability of recombinant sCD40L to potentiate vaccine-induced immunity against L.major infection (Gurunathan et al., 1998; Chen et al., 2001). Previous studies in knock out mice and in vitro models using recombinant trimeric (s)CD40L or anti-CD40 mAb also showed the important role of the CD40 costimulatory pathway in protection against L.major and L.donovani infection (Campbell et al., 1996; Soong et al., 1996; Murray et al., 2003; Murray, 2005). Although a larger effect could be anticipated in an HIV population with decreased CD4+CD40L+ T-cells numbers, we cannot extrapolate these findings to HIV individuals with a suppressed immunity as the mechanisms behind the improved parasite killing remain unknown.

In addition, neopterin levels were studied as a marker of the total effect of immunological interactions on the populations of macrophages (Sucher et al., 2010). Neopterin levels are produced upon macrophage activation mainly by IFN-y and to a lesser extent by IFN-α and β, reflecting an activated cellular immunity. Besides an observed increase in active VL-HIV patients due to an activated Th1-mediated immune response and the herewith connected development of oxidative stress against intracellular Leishmania parasites, neopterin levels have also been constitutively reported to be elevated in active HIV patients having an increased number of activated CD4+ T-cells which are highly susceptible for HIV infection. Hence, in contrast to sCD40L, neopterin levels seem to reflect inflammation due to HIV and were inversely associated with the CD4+ T-cell count (**Figure 3**; Chadha et al., 2013). Linked to the CD4+ Tcell count recovery, a decrease in neopterin levels is reported after successful ART although these do not normalize completely in HIV infected individuals (Amirayan-Chevillard et al., 2000; Chadha et al., 2013). As normal serum values of neopterin range until 2,200 pg/ml in healthy people, a median of 3,960 pg/ml fits with a slightly elevated level of neopterin in our population of stable HIV patients on ART. Active VL-HIV patients with a median of 29,400 pg/ml showed comparable levels to newly diagnosed HIV patients (Amirayan-Chevillard et al., 2000) and previous VL patients (Hamerlinck et al., 2000).

Accounting for the CD4+ T-cell count association, no statistically significant difference could be found between nondiseased HIV patients with Leishmania antibodies and those without Leishmania antibodies in our CD4-matched case-control study, with a small trend toward higher levels in Leishmania antibody positive HIV patients (**Figures 1**, **2**). This suggests the lack of a Leishmania-specific effect on serum levels of neopterin. Although an increasing number and proportion of VL cases are now seen in individuals on ART, HIV patients with and without Leishmania antibodies studied here were rather stable ART patients with high CD4+ T-cell counts, as we selected longterm residents of the study area in stable follow-up at the ART clinic. Respectively, only one male patient presented with VL 9 months after his baseline sampling (see hexagon in **Figure 1**). Interestingly, this patient had the highest level of neopterin at baseline and a rather average concentration of sCD40L. This single case observation corresponds with the postulated value of neopterin levels to, although non-specifically, predict opportunistic infections in HIV patients but have less value as a specific marker of early Leishmania immunopathology in HIV patients. Neopterin production was for instance also reported to be increased in adults with TB-HIV coinfection (Skogmar et al., 2015). Although therefore less sensitive to screen for VL progression in particular, neopterin levels could be a valuable marker in a clinical predictive algorithm for resource-constrained settings. Compared to other cytokines, neopterin also has a higher stability in body fluids allowing easy sample handling. In addition, a rapid test is being constructed and urinary neopterin levels could also be stable under field conditions as a non-invasive marker of disease progression (Heistermann and Higham, 2015).

Extensive clinical data was missing to investigate the impact of chronic inflammation from other coinfections (Trypanosomiasis, helminths, etc.) on the levels of neopterin as well as sCD40L, but these results suggest some degree of additional immune cell activation during an asymptomatic Leishmania infection in HIV patients. It remains to be seen whether such a high T-cell activation environment could be beneficial in the initial stages of infection and whether this remains true in chronic relapse cases. The value of neopterin as an alternative test of cure in VL/HIV was not investigated here. Because neopterin is only produced by humans and primates, literature in experimental models of VL is nonexistent. To our knowledge, only two older studies from the 90's and one recent study investigated the value of serum neopterin levels as markers of cure during treatment in CL and VL patients (Schriefer et al., 1995; Hamerlinck et al., 2000). Serum levels only appeared increased in VL patients before treatment, indicating a restricted association with a systemic infection. Only 1 out of 7 patients followed for 6– 12 months after treatment died of leishmaniasis and showed a gradual increase in levels of neopterin. Vice versa, values in the other 6 patients decreased to normal values during treatment. These results confirmed the findings in 20 VL patients by Schriefer and coworkers (Schriefer et al., 1995). Likewise, Kip and coworkers recently confirmed the pharmacodynamic potential of neopterin to identify Sudanese and Kenyan VL patients at risk for VL relapse (Kip et al., 2018). Unfortunately, the longitudinal evaluation of neopterin in coinfected patients has not yet been reported.

Both molecules should be further explored as useful markers in a clinical algorithm for indirectly monitoring and predicting initial Leishmania infection progression in HIV patients, as recently proposed by Van Griensven and coworkers (van Griensven et al., 2014a). Antibody positivity has been suggested as an indicator of poor control (Th2 response). Nevertheless, all but one Leishmania antibody positive HIV patients remained asymptomatic for a median follow-up time of 1 year. To evaluate this hypothesis, longitudinal observational cohort studies with a large number of HIV patients in care living in endemic regions should comprehensively study asymptomatic infection with markers of Th1 immunity (Leishmanin Skin Test, T-cell functionality, etc.), Th2 immunity (antibody-based tests, etc.) as well as antigen detection (KAtex, Loop-mediated isothermal amplication (LAMP), RT-PCR, etc.), to discriminate past, latent or active Leishmania infection. Such studies would allow a simultaneous investigation of the utility of these cytokines and derive cut-off levels to suggest progression or resistance to VL. Advantages of cytokine measurements is that they allow for easy sample collection, analysis at low cost and require little technical competence, potentially using a partial or fully automated ELISA procedure.

# CONCLUSION

Our results match with the stated protective role of sCD40L in VL and indicate the importance of the CD40-CD40L pathway in early human immune responses against leishmaniasis, also in CD4+CD40L+ deprived HIV patients. Recombinant sCD40L could counteract the parasite's regulatory influence on host immunity and should be further explored with regard to resistance to Leishmania infection. On the other hand, neopterin levels could indicate general progression of disease, although non-specifically, and could be explored as a marker of a prognostic algorithm to predict VL progression in HIV patients.

# ETHICS STATEMENT

This study was carried out in accordance with the recommendations of the Declaration of Helsinki 2013, the Good Clinical Practice of the WHO, and those of the Ethiopian Food, Medicine and HealthCare Administration and Control Authority (FMHACA) with written informed consent from all subjects. The protocol was approved by the National Research Ethics Review Committee of Ethiopia, the University of Gondar Institutional Review Board (IRB), the Ethics Review Board of Médécins Sans Frontiers, the IRB of the Institute of Tropical Medicine, Antwerp and the Ethics Committee of Antwerp University Hospital.

# AUTHOR CONTRIBUTIONS

WA conceived the study and drafted the manuscript. SA, EA, and FV contributed in sample measurements and data acquisition. SvH, YG, BM, ED, and EA helped in sample collection and daily coordination. Interpretation of the data was done by WA, LK, ED, and JvG. SA, SvH, YG, ED, FV, BM, EA, LK, and JvG commented on the draft. All authors read and approved the final manuscript.

# FUNDING

Funding was provided by the Department of Economy, Science, and Innovation (EWI) of the Flemish government and Belgian Directorate General for Development Cooperation under the ITM-DGDC framework agreement FA-IIII. WA is personally supported by a Research Foundation-Flanders postdoctoral fellowship.

# ACKNOWLEDGMENTS

We would like to thank the patients who volunteered for the clinical trial and cohort study. We also highly appreciated the teams at University of Gondar Leishmaniasis Research and Treatment Center (LRTC), Abdurafi Health Center and Metema District Hospital for supporting the trial. The efforts of ITM colleagues and statisticians were also highly appreciated. Special thanks go to the Drugs for Neglected

Diseases initiative (DNDi) and Médécins Sans Frontiers for their support of the LRTC and Abdurafi Health Center, respectively.

# REFERENCES


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

Copyright © 2018 Adriaensen, Abdellati, van Henten, Gedamu, Diro, Vogt, Mengesha, Adem, Kestens and van Griensven. 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.

# Visceral Leishmaniasis IgG1 Rapid Monitoring of Cure vs. Relapse, and Potential for Diagnosis of Post Kala-Azar Dermal Leishmaniasis

### *Edited by:*

*Javier Moreno, Instituto de Salud Carlos III, Spain*

### *Reviewed by:*

*Suresh Kumar Kalangi, Indrashil University, India Kevin M. Tyler, University of East Anglia, United Kingdom*

### *\*Correspondence:*

*Tegwen Marlais tegwen.marlais@lshtm.ac.uk*

*†These authors have contributed equally to this work*

### *Specialty section:*

*This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology*

*Received: 09 May 2018 Accepted: 28 November 2018 Published: 13 December 2018*

### *Citation:*

*Marlais T, Bhattacharyya T, Singh OP, Mertens P, Gilleman Q, Thunissen C, Hinckel BCB, Pearson C, Gardner BL, Airs S, de la Roche M, Hayes K, Hafezi H, Falconar AK, Eisa O, Saad A, Khanal B, Bhattarai NR, Rijal S, Boelaert M, El-Safi S, Sundar S and Miles MA (2018) Visceral Leishmaniasis IgG1 Rapid Monitoring of Cure vs. Relapse, and Potential for Diagnosis of Post Kala-Azar Dermal Leishmaniasis. Front. Cell. Infect. Microbiol. 8:427. doi: 10.3389/fcimb.2018.00427* Tegwen Marlais <sup>1</sup> \* † , Tapan Bhattacharyya1†, Om Prakash Singh<sup>2</sup> , Pascal Mertens <sup>3</sup> , Quentin Gilleman<sup>3</sup> , Caroline Thunissen<sup>3</sup> , Bruno C. Bremer Hinckel 3,4, Callum Pearson<sup>1</sup> , Bathsheba L. Gardner <sup>1</sup> , Stephanie Airs <sup>1</sup> , Marianne de la Roche<sup>1</sup> , Kiera Hayes <sup>1</sup> , Hannah Hafezi <sup>1</sup> , Andrew K. Falconar <sup>5</sup> , Osama Eisa<sup>6</sup> , Alfarazdeg Saad<sup>6</sup> , Basudha Khanal <sup>7</sup> , Narayan Raj Bhattarai <sup>7</sup> , Suman Rijal <sup>8</sup> , Marleen Boelaert <sup>9</sup> , Sayda El-Safi<sup>6</sup> , Shyam Sundar <sup>2</sup> and Michael A. Miles <sup>1</sup>

*<sup>1</sup> Faculty of Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine, London, United Kingdom, <sup>2</sup> Department of Medicine, Institute of Medical Sciences, Banaras Hindu University, Varanasi, India, <sup>3</sup> Coris BioConcept, Gembloux, Belgium, <sup>4</sup> Department of Biomedical Sciences, University of Antwerp, Antwerp, Belgium, <sup>5</sup> Departamento de Medicina, Universidad del Norte, Barranquilla, Colombia, <sup>6</sup> Faculty of Medicine, University of Khartoum, Khartoum, Sudan, <sup>7</sup> Department of Microbiology, B.P. Koirala Institute of Health Sciences, Dharan, Nepal, <sup>8</sup> Department of Internal Medicine, B.P. Koirala Institute of Health Sciences, Dharan, Nepal, <sup>9</sup> Department of Public Health, Institute of Tropical Medicine, Antwerp, Belgium*

Background: There is a recognized need for an improved diagnostic test to assess post-chemotherapeutic treatment outcome in visceral leishmaniasis (VL) and to diagnose post kala-azar dermal leishmaniasis (PKDL). We previously demonstrated by ELISA and a prototype novel rapid diagnostic test (RDT), that high anti-*Leishmania* IgG1 is associated with post-treatment relapse versus cure in VL.

Methodology: Here, we further evaluate this novel, low-cost RDT, named VL Sero K-SeT, and ELISA for monitoring IgG1 levels in VL patients after treatment. IgG1 levels against *L. donovani* lysate were determined. We applied these assays to Indian sera from cured VL at 6 months post treatment as well as to relapse and PKDL patients. Sudanese sera from pre- and post-treatment and relapse were also tested.

Results: Of 104 paired Indian sera taken before and after treatment for VL, when deemed clinically cured, 81 (77.9%) were positive by VL Sero K-SeT before treatment; by 6 months, 68 of these 81 (84.0%) had a negative or reduced RDT test line intensity. ELISAs differed in positivity rate between pre- and post-treatment (*p* = 0.0162). Twenty eight of 33 (84.8%) Indian samples taken at diagnosis of relapse were RDT positive. A comparison of Indian VL Sero K-SeT data from patients deemed cured and relapsed confirmed that there was a significant difference (*p* < 0.0001) in positivity rate for the two groups using this RDT. Ten of 17 (58.8%) Sudanese sera went from positive to negative or decreased VL Sero K-SeT at the end of 11–30 days of treatment. Forty nine of 63 (77.8%) PKDL samples from India were positive by VL Sero K-SeT. Conclusion: We have further shown the relevance of IgG1 in determining clinical status in VL patients. A positive VL Sero K-SeT may also be helpful in supporting diagnosis of PKDL. With further refinement, such as the use of specific antigens, the VL Sero K-SeT and/or IgG1 ELISA may be adjuncts to current VL control programmes.

Keywords: visceral leishmaniasis, serology, treatment, relapse, cure, IgG1, RDT, PKDL

# INTRODUCTION

Visceral leishmaniasis (VL; kala-azar), is caused by the protozoan parasites Leishmania donovani in Asia, Africa and the Middle East and Leishmania infantum in Europe and South America. These parasites are transmitted by blood-feeding female phlebotomine sand flies. Symptomatic VL is usually fatal if untreated. Symptoms include prolonged fever >14 days, wasting, splenomegaly, hepatomegaly and anemia (Sundar and Rai, 2002). While VL is present in about 75 countries, the majority (90%) of cases in 2015 occurred in India, Sudan, South Sudan, Ethiopia, Somalia, Kenya, and Brazil (World Health Organization, 2017), where it is closely linked to poverty, both as cause and effect (Boelaert et al., 2009; Sarnoff et al., 2010).

Following clinical suspicion of VL, serology is used for diagnosis. Techniques vary by region and include the immunofluorescence antibody test (IFAT), direct agglutination test (DAT), enzyme linked immunosorbent assay (ELISA), and detection of IgG against recombinant antigens rK39 or rK28 (Singh and Sundar, 2015). In India the DAT and rK39 serology are used, with a positive result in either test indicative of exposure to infection with L. donovani. For confirmatory parasitological diagnosis, seropositive individuals undergo spleen, bone marrow or lymph node biopsy to search for the intracellular amastigote stage in films of Giemsa-stained aspirates. These are invasive, costly and potentially hazardous techniques with low and variable sensitivities ranging from 53 to 99% (Singh and Sundar, 2015).

VL is treated with antimonials, miltefosine, paromomycin, amphotericin B, liposomal amphotericin (AmBisome) or drug combinations (World Health Organization, 2010). Currently, post-treatment outcome is determined by assessment of clinical signs and symptoms, initially on the last day of drug treatment and, in India, again 6 months after administration of the last dose (World Health Organization, 2010). Possible outcomes are: cure; relapse; death (by VL or not); post kala-azar dermal leishmaniasis (PKDL); loss to follow up. However, recent studies from India and Nepal have reported relapse rates of between 1.4 and 20%, including up to and beyond 12 months following the end of treatment (Burza et al., 2013, 2014; Rijal et al., 2013). In Sudan, relapse rates around 6% have been reported (Gorski et al., 2010; Atia et al., 2015). Patients who relapse face a further biopsy procedure to confirm the presence of parasites.

PKDL is a non-painful sequela of VL occurring in over 50% of cases in Sudan (Zijlstra et al., 2003) but is far less prevalent in South Asia (Zijlstra et al., 2003; Uranw et al., 2011). PKDL is less reported in L. infantum endemic regions where cases have mostly been associated with HIV/AIDS (Ridolfo et al., 2000; Bittencourt et al., 2003; Celesia et al., 2014), other co-infections (Trindade et al., 2015) or immune suppression (Roustan et al., 1998). PKDL manifests between 0.5 months to one or more years after apparently successful VL treatment (Musa et al., 2002; Uranw et al., 2011; Singh et al., 2012; Moulik et al., 2017) and may occasionally occur without a prior episode of VL (el Hassan et al., 1992; Zijlstra et al., 2003; Das et al., 2012, 2016). PKDL is suspected based on dermal manifestations that are non-specific and diagnosis is made on previous VL treatment history and confirmed parasitologically by microscopy of slit skin smear or biopsy or PCR (Zijlstra et al., 2017). Conventional serology is likely to remain positive from the earlier VL and there is no test in use to predict PKDL (Gidwani et al., 2011). The high parasite density in PKDL skin provides a source of infection to sand flies and thus sustains long term transmission and endemicity (Molina et al., 2017; Mondal et al., 2018).

An unresolved crucial question is how to identify asymptomatic infected individuals simply and reliably (as defined by seropositivity, lack of clinical symptoms and no prior history of VL) who will progress to active VL. High DAT and/or rK39 ELISA titres have been associated with increased risk of progression in the Indian subcontinent but as yet there is no single rapid test in use for this purpose (Hasker et al., 2014; Chapman et al., 2015). To improve outcome monitoring of VL and disease control, the World Health Organization has identified the vital need for a marker of post-chemotherapeutic cure, and the high priority incorporation of such a biomarker into a point-of-care rapid diagnostic test (RDT) (World Health Organization, 2012). Such a test should meet the "ASSURED" criteria of being: affordable; sensitive (few false negatives); specific (few false positives); user-friendly, requiring minimal training; rapid; robust, not requiring cold-storage; equipmentfree, and deliverable to those who need it (Peeling et al., 2006).

We have previously shown that high anti-Leishmania IgG1 ELISA titers are associated with treatment failure, whereas, in cases deemed to be cured following chemotherapy, IgG1 levels diminish significantly by 6 months post-treatment and only IgG1 gave this level of discrimination (Bhattacharyya et al., 2014a). We demonstrated this by ELISA against L. donovani whole cell lysate, and then adapted the assay to a prototype lateral flow immunochromatographic RDT. Here, we present further evaluation of this RDT, called VL Sero K-SeT, to indicate cure after VL treatment in a larger, paired, sample set and to confirm relapse. We also performed western blot on the same sample set. Additionally we show the potential utility of VL Sero K-SeT and other IgG1 assays to confirm PKDL.

### Marlais et al. IgG1 to Monitor Visceral Leishmaniasis

# METHODS

## Ethics Statement

In India, the collection of samples was approved by the Ethics Committee of Banaras Hindu University, Varanasi. In Sudan approval was by the Ethical Research Committee, Faculty of Medicine, University of Khartoum and the National Health Research Ethics Committee, Federal Ministry of Health, Sudan. Written informed consent was obtained from adult subjects included in the study or from the parents or guardians of individuals <18 years of age. In Nepal, informed consent was obtained from all the participants and the ethical committee of the B.P. Koirala Institute of Health Sciences (BPKIHS) approved the study. This research was also approved, as part of the EC NIDIAG project, by the London School of Hygiene and Tropical Medicine Ethics Committee.

# Sources of Sera/Plasma

We retrospectively selected sera or plasma from an archive of different VL disease states. Samples had been collected in VL endemic regions, namely Muzaffarpur in Bihar, India after 2007 and in 2013 in Gedaref, Sudan. Sample sizes used during this evaluation were dependent on availability of appropriate samples and reagents.

In India, cases of VL had been diagnosed by positive rK39 serology and/or parasitologically by microscopy of splenic aspirates. In Sudan active cases of VL had been diagnosed by microscopy of bone marrow or lymph node aspirates in conjunction with serological assays. These diagnoses were made according to their respective national procedures, prior to the present study. Sera/plasma were stored at −80◦C until use. All patients were HIV negative. We have previously observed that serum and plasma derived from the same sample show no difference in IgG titer in ELISA against L. donovani lysate (unpublished observations), although we have not specifically assessed IgG1 with both sample types.

## India

Indian sample types are described in **Table 1**. We have previously found that in Indian VL, IgG1 titer up to day 30 post-treatment initiation is not statistically significantly different from pretreatment (Bhattacharyya et al., 2014a) and therefore we consider these as "pre-treatment" in paired samples for the purposes of this study. Treatment of VL was with single-dose AmBisome alone or with 10 days of miltefosine. PKDL was treated with miltefosine for 84 days. DAT and rK39 ELISA were conducted prior to the present study as part of standard diagnostic procedures in India.

# Sudan

Sudanese paired serum samples (n = 17 pairs) were taken on day of diagnosis of VL and at the end of treatment at 11 days (AmBisome), 17 days (sodium stibogluconate (SSG) + paromomycin), or 30 days (SSG only). These samples were previously tested for IgG1 by ELISA (Bhattacharyya et al., 2014a). Additional Sudanese serum samples used in the present study were unpaired treated individuals (n = 2) taken an unknown time


after treatment, and relapse (n = 1). Sudanese Endemic Healthy Control (EHC) samples had previously been tested by the IgG1 ELISA using the same antigen and were negative (Bhattacharyya et al., 2014a) but were not retested here.

# Antigen Production

Whole cell lysate of L. donovani strain MHOM/IN/80/DD8 isolated from India, and MHOM/SD/97/LEM3458 isolated from Sudan, was prepared as described previously (Bhattacharyya et al., 2014a). Lysate antigen was used for VL Sero K-SeT development (strain LEM3458), ELISA and western blot (strain DD8). Antigen preparation for western blot strips contained 50 µl of protease inhibitor cocktail (P8340, Sigma, UK) per 1 ml of L. donovani cells; centrifugation after sonication was 16,160 × g for 45 min at 4◦C.

# ELISA for IgG1 Anti *L. donovani*

Duplicate ELISA plates were coated overnight at 4◦C with L. donovani DD8 strain antigen prepared as above, at 2µg/ml in coating buffer (35 mM NaHCO3, 15 mM NaCO3, pH 9.6), 100 µl/well. Plates were washed 3 times with phosphate buffered saline (PBS) + 0.05% Tween 20 (PBST) prior to blocking with 200 µl/well PBS + 2% w/v non-fat milk powder (Premier International Foods, UK) (PBSM) for 2 h at 37◦C, followed by three PBST washes. Sera/plasma were diluted 1/100 in PBST+ 2% w/v non-fat milk powder (PBSTM) and applied at 100 µl/well, incubated for 1 h at 37◦C then washed 6 times with PBST. Mouse anti human IgG1-horse radish peroxidase (HRP) (ab99774, Abcam, UK) was diluted 1:5,000 in PBSTM and incubated at 100 µl/well, 37◦C for 1 h. Plates were washed 6 times with PBST before the addition of 100 µl/well of freshly prepared substrate solution (50 mM citric acid, 50 mM Na2HPO4, 2 mM o-phenylenediamine HCl, 0.009% H2O2). Plates with substrate were incubated in the dark at room temperature for 10–15 min when the reaction was stopped with 100 µl/well of 1 M H2SO<sup>4</sup> and absorbance read at 490 nm. Each plate contained an EHC sample as a negative serological control for determining the positivity cut-off and a known seropositive VL sample as positive control. All ELISA results reported are the mean A<sup>490</sup> of duplicate plates.

# RDT Production and Use

Whole cell lysate was prepared as described previously (Bhattacharyya et al., 2014a) from L. donovani strain MHOM/SD/97/LEM3458. The VL Sero K-SeT lateral flow immunochromatographic tests were developed at Coris BioConcept and consisted of a cassette with a nitrocellulose membrane, a sample pad, a conjugate pad and an absorbent pad, backed with a plastic strip. The nitrocellulose membranes were sensitized with the L. donovani lysate antigen and anti-human IgG1-specific antibody labeled with colloidal gold was dried on the conjugate pad. This strip was housed in a plastic cassette with two windows: the smaller buffer well and the long central test window.

To perform the test, 3.5 µl of serum/plasma was applied to the test window at the point indicated by a dot (·) on the cassette, followed immediately by 120 µl of supplied running buffer to the buffer well (**Figure 1**). Devices were incubated flat, at ambient temperature for 15 min before being assessed visually. Any test line at position T was considered a positive result if a control line was also present at position C. Positive test line intensity was assessed visually for samples from pre- and post-treatment VL (**Figure 1**). A subset of samples was tested on different batches of the VL Sero K-SeT. Readers of the RDTs were blinded to all the corresponding ELISA results.

# Western Blotting

Western blots were performed to visualize antigen recognition in patients from the different clinical groups, as described in **Supplementary Material S1**. Briefly, tricine SDS-PAGE gels were made as per Schägger (2006). L. donovani DD8 lysate was used as antigen with sera/plasma diluted 1 in 300 (Sudan) or 1 in 400 (India) and detection was by mouse anti human IgG1-HRP.

# Statistical Analysis

We performed a two-tailed Fisher's exact test on Indian VL Sero K-SeT and IgG1 ELISA data to calculate p-values between samples from pre- and matched 6 months post-treatment (deemed cured), separately between post-treatment and relapse, and between post-treatment and PKDL. Cut-off for the IgG1 ELISA was calculated as the mean absorbance of the EHC samples plus 3 standard deviations.

# RESULTS

# IgG1 Diminishes by 6 Months in Cured VL Patients

Samples taken from Indian patients before or at the outset of therapy, were compared by VL Sero K-SeT and IgG1 ELISA with paired samples taken 6 months later when the individuals were deemed cured. Both IgG1 assay methods showed a statistically significant difference in positivity rate between pre- and posttreatment samples (p = 0.0162 and p < 0.0001 for ELISA and RDT, respectively) (**Figure 2**). A consistent and strongly significant difference was also observed between cured vs. relapsed samples (p < 0.0001), again with both IgG1 assay methods (**Figure 2**).

A subset of pre- and post-treatment of the cured pairs samples (n = 87) was tested on different batches of the VL Sero K-SeT, with agreement between individual RDTs of 92.0% (80/87).

# Changes in IgG1 Levels by ELISA and VL Sero K-SeT

ELISA absorbance and corresponding VL Sero K-SeT results for individual samples are given in **Supplementary Material S2**. Of the 80 Indian paired samples tested for anti L. donovani IgG1 by ELISA, 54 (67.5%) were positive before treatment. Of these, 51/54 declined in titer: 21/51 (representing 26.3% of the total 80) went from positive to negative and 30/51 (representing 37.5% of the total 80) had reduced IgG1 at 6 months when deemed cured (**Figure 2** and **Table 2**). Twenty one (26.3%) paired cured sera were negative by IgG1 ELISA before treatment and remained so at 6 months.

Overall, including those negative at the start, at 6 months after treatment 79/104 (76.0%) were negative by VL Sero K-SeT (**Table 2**). VL Sero K-SeT results were additionally assessed according to whether the Indian 6 month post-treatment (cure) sample had a decreased or not decreased test line intensity compared to the paired pre-treatment sample. Of the 104 paired samples tested by VL Sero K-SeT from deemed cured Indian VL patients, 81 (77.9%) were positive at start of treatment (**Table 2**); of these, 68/81 (84.0%) had either become negative or had a visibly reduced test line intensity at 6 months when deemed cured. Thirteen (12.5%) initially RDT positive individuals showed no visible decrease in RDT band intensity at 6 months, despite being deemed cured, and none became positive from negative.

Ninety four percent of samples positive by ELISA at pretreatment, decreased in seropositivity; for VL Sero K-SeT, this proportion was 84%. However, at 6 months post-treatment, the ELISA was more likely to remain positive than the RDT, using the cut-off value established for the IgG1 ELISA.

Seventy nine Indian samples were tested by both VL Sero K-SeT and ELISA. Of these samples, the RDT was more likely than the ELISA to be positive with pre-treatment samples (78.5 vs. 67.1%) and negative with 6 month samples (78.5 vs. 53.2%). Of samples which remained positive at 6 months by both methods (n = 14), the change in intensity of RDT test line generally mirrored the change in ELISA absorbance value for the same sample. Three of the Indian samples increased markedly in IgG1

titer by ELISA at 6 months (**Figure 2**). Two of these accorded with a corresponding rise in VL Sero K-SeT test line intensity; for the third sample, both pre-treatment and 6 month RDTs were negative (**Figure 2A**).

Sudanese paired samples taken before and immediately after treatment (11–30 days later) were similarly assessed (**Table 2**). For Sudanese paired samples prior to treatment, 13/17 (76.5%) were positive by VL Sero K-SeT and at completion of treatment, 10/13 (76.9%) had a negative or reduced test line intensity (**Table 2**). If taken as a single time point at the end of treatment, 10/17 (58.8%) Sudanese VL patients had negative VL Sero K-SeT result. Four (23.5%) of the Sudanese treated individuals were negative pre-treatment, similar to the proportion of Indian samples (22.1%). Two additional un-paired treated Sudanese samples were negative by RDT (not shown).

# IgG1 Western Blot Confirmed Negative/Declined RDT in Cure

For a subset of 25 of the paired Indian samples, western blots mirrored the VL Sero K-SeT RDT findings, in that IgG1 declined dramatically in all but one VL patient at 6 months follow up after treatment (**Supplementary Material S3**). As with the RDT, the blots showed that samples that were positive and detecting many antigens before treatment had become negative or reduced

TABLE 2 | Change in IgG1 response of pre- and post-treatment paired samples from India and Sudan.


in intensity by 6 months. Corresponding RDT images are shown in **Supplementary Material S4**.

# Elevated IgG1 in VL Relapse

For 33 Indian patients for whom we had unpaired samples at the time of relapse, the VL Sero K-SeT was 84.8% (28/33) positive and ELISA 91.3% (21/23) positive, confirming relapse. Of the 23 samples tested by ELISA that were also tested by RDT, 19 gave the same result by both assays (**Supplementary Material S2**). The single available Sudanese relapse sample was IgG1 positive (**Supplementary Material S5**). Twenty five of the Indian samples and the single Sudanese sample were also tested on western blot for IgG1 against L. donovani lysate antigen and showed concordance between the RDT and blots (**Supplementary Materials S2**, **S5**). For two of the 33 Indian relapse samples, a paired pre-treatment sample was available. Both individuals were VL Sero K-SeT positive at both time points.

All samples from other diseases, namely malaria, tuberculosis, dengue fever, rheumatoid arthritis, and multiple myeloma were negative by VL Sero K-SeT, as were all samples from endemic healthy controls.

# VL Sero K-SeT Can Provide Evidence for PKDL but Not for Its Cure

Of the 63 PKDL samples tested, 49 (77.8%) were positive by VL Sero K-SeT and of the subset of 45 tested by IgG1 ELISA, 43 (95.6%) were positive (**Supplementary Material S2**). A subset of 10 VL Sero K-SeT-positive PKDL samples were tested by western blot, of which 9 showed discernible bands. Images of the blots and their corresponding VL Sero K-SeT RDTs are shown in **Supplementary Material S6**. There was a highly statistically significant difference between post-treatment cured samples at 6 months and PKDL by both VL Sero K-SeT and IgG1 ELISA (Fisher's exact p < 0.0001 for both assays).

Seventeen of the 63 individuals with PKDL provided between 1 and 5 additional sequential follow-up samples over intervals ranging from 15 to 365 days post-treatment. These PKDL posttreatment sequential samples retained the initial RDT result in 12/17 (70.6%) cases, decreased in 3/17 (17.6%), increased slightly in one case (5.9%) and varied between positive and negative over time in one case (5.9%).

# IgG1 Can Indicate Progression From Asymptomatic Status

When samples from asymptomatic seropositive individuals who later progressed to symptomatic disease (progressors, n = 4) were tested on the VL Sero K-SeT, all gave a positive test result (**Supplementary Material S7**). In contrast, 4 individuals who were seropositive but did not develop symptomatic VL were negative by VL Sero K-SeT. Thus, in our limited sample size, elevated IgG1 levels, as detected by VL Sero K-SeT, were associated with progression to symptomatic disease. This result was corroborated by ELISA and western blot (**Supplementary Material S7**).

# DISCUSSION

Conventional serology for VL diagnosis relies on detecting the overall IgG response. This has been reported to remain elevated, often for years, after treatment (Bhattarai et al., 2009; Gidwani et al., 2011; Srivastava et al., 2013). This makes current serology unsuitable for timely monitoring of treatment outcome. We have previously found using ELISA that a decreased or negative anti Leishmania IgG1 titer at 6 months post-treatment can be indicative of VL cure, whereas elevated IgG1 levels are associated with post-chemotherapeutic relapse (Bhattacharyya et al., 2014a).

# Monitoring of Post-treatment Outcomes

Here we used a larger panel of paired samples to assess the IgG1 response as detected by the rapid test, VL Sero K-SeT, where 77.9% of Indian samples were positive before treatment and of these 69.1% had become negative 6 months later when deemed cured (**Table 2**). In total, 76% of 6 month samples were negative, a significant difference from pre-treatment (p < 0.0001). Of those still positive at 6 months using this RDT, we found that a diminished test line intensity was also consistent with cure. This decline was corroborated by ELISAs, and despite slight differences in the antigen preparations. We have found no difference in performance of the VL Sero K-SeT when DD8 strain antigen is used instead of LEM3458 (unpublished observations). Thus, the VL Sero K-SeT is a promising innovation, although there is a need to improve further its discriminative capacity.

Sudanese samples declined from positive to negative or decreased VL Sero K-SeT test line intensity in 76.9% of patients immediately after treatment, no more than 30 days after the first sample. This apparently rapid drop in IgG1 was not seen in Indian samples and could be due to the overall lower IgG titer observed in Sudanese samples (Bhattacharyya et al., 2014b; Abass et al., 2015). Thus, a small drop in IgG1 titer could have taken these samples below the detection limit of the VL Sero K-SeT. This may suggest that the VL Sero K-SeT can be used before 6 months to indicate cure or relapse in eastern Africa. The unexpectedly low sensitivity of the VL Sero K-SeT at the start of treatment for both Indian (77.9%) and Sudanese (76.5%) samples does not hinder the subsequent assessment of cure at 6 months, because a negative IgG1 result at 6 months can indicate cure. In addition, we do not propose to use IgG1 assays as a diagnostic for active VL but rather to assist with confirming cure, relapse and PKDL, all of which currently lack an appropriate diagnostic test. With Indian samples, where there was discrepancy between VL Sero K-SeT and ELISA, the RDT was generally more accurate, being positive with pre-treatment and negative with 6 month samples (**Supplementary Material S2**). As for the Indian sera, the strength of RDT test line intensity broadly corresponded with ELISA signal for an individual sample.

Elevated levels of IgG1 were associated with VL relapse in both assays here for Indian samples. Likewise, the single Sudanese relapse patient was positive by VL Sero K-SeT, whilst 2 treated individuals were negative. We do not know the length of time between treatment and relapse for relapsed individuals (India and Sudan), or the outcome of treated Sudanese individuals. Burza et al. (2014) advised that patient follow-up should be extended from 6 to 12 months as 50–85% of relapses have been found to occur 6 to ≥12 months post-treatment (Rijal et al., 2013; Burza et al., 2014). Our evaluations of a limited number Nepalese relapse samples eluted from filter paper indicated that, although encouraging, elution volumes and conditions need further optimisation before they can be more extensively used with VL Sero K-SeT (data not shown).

We found that in Indian cases who relapsed, the RDT positivity rate was significantly different from 6 month samples from patients deemed cured (p < 0.0001). Thus, the VL Sero K-SeT, with Indian samples, can contribute to distinguishing patients deemed cured from those who have relapsed. Of the 13 Indian patients deemed cured at 6 months but who had no clear decrease in VL Sero K-SeT test line intensity (**Table 2**), none is known to have relapsed with VL. However, the quantitative ELISA did detect an IgG1 decrease in these samples, consistent with cure. Apparent relapses might however, occasionally include re-infections given the highly endemic locations (Morales et al., 2002). Although beyond the scope of the present study, the inclusion of parasite genotyping in a future study would be an advantage.

Cases co-infected with HIV and VL were not included in the present study. Serological diagnosis is less reliable in HIV/VL co-infection (Cota et al., 2012; Abass et al., 2015) and the dynamics of IgG1 response in HIV/VL co-infections need to be determined. Other techniques such as a loop mediated isothermal amplification (LAMP) or qPCR detecting parasite DNA might have the potential to discriminate cure from relapse in HIV/VL patients but are currently less accessible than immunological tests (Mukhtar et al., 2018).

# PKDL

Indian individuals with PKDL tested here were defined as being with or without a previous history of VL, presenting with a dermal macular, papular or nodular rash often starting on the face with further spread to other parts of the body without loss of sensation. VL Sero K-SeT and IgG1 ELISA results suggest that these assays might contribute to PKDL case detection, as found by a study by Saha et al. (2005), whereas conventional serology may be of limited utility (Gidwani et al., 2011). Our data did not assess the predictive value of IgG1 for development of PKDL.

Where the information was available with our sample set, we did not observe an association between elevated IgG1 and macular vs. polymorphic PKDL presentation, this is in contrast to the report of Mukhopadhyay et al. (2012). For a subset of these PKDL samples, we also tested sequential samples taken up to 1 year after the initial sample. We did not observe a consistent decrease in IgG1 after PKDL treatment.

# Progression From Asymptomatic to Active VL

Asymptomatic, seropositive cases outnumber active VL cases (Bern et al., 2007; Ostyn et al., 2011; Hasker et al., 2013; Hirve et al., 2016; Saha et al., 2017) but a proportion are at elevated risk of progressing to active VL (Gidwani et al., 2009; Topno et al., 2010; Ostyn et al., 2011). Asymptomatics have been reported to occasionally have detectable parasites by PCR or culture of blood (le Fichoux et al., 1999; Costa et al., 2002; Bhattarai et al., 2009; Srivastava et al., 2013). Therefore, neither standard seropositivity nor parasitaemia are indicators of progression to clinical disease. Gidwani et al. (2009) found that this progression to VL occurred up to 2 years after serological positivity.

Our limited sample size of seropositive asymptomatic individuals were identified during a community serological screening study, before the present study. Those who later progressed to clinical VL were positive by VL Sero K-SeT and ELISA, whilst those who did not progress were negative by both assays. High titres in both DAT and rK39 ELISA have been indicative of progression in larger studies (Ostyn et al., 2011; Hasker et al., 2014). However, this combination of tests requires laboratory facilities, therefore it would be desirable to have an RDT that could predict progression.

Additional validation of the VL Sero K-SeT should compare larger cohorts who do and do not progress to VL.

# Potential Clinical Application of IgG1 Tests

On the basis of the IgG1 responses reported here by VL Sero K-SeT and ELISA, we propose that IgG1 levels may contribute to monitoring the therapeutic outcome of VL, irrespective of whether there is a pre-treatment sample or result. With further development and validation, IgG1 assays, including the VL Sero K-SeT, which can be produced in large-scale at a cost of a few Euros per test, can be used as an adjunct to the clinical assessment of VL status following treatment. A negative, or defined decrease in IgG1 result at 6 months post treatment in India could be supportive of the clinical assessment of cure. Conversely, an un-paired positive or non-decreased paired positive result at 6 months could indicate the need for additional follow-up. In Sudan, the test may be applicable for defining cure before 6 months. A positive IgG1 result in suspected PKDL or relapse could support the presence of leishmaniasis compared to differential diagnoses. Although western blots were supportive of the use of IgG1, we did not specifically assess banding patterns, and do not propose their use in VL diagnosis. However, we are investigating the discriminative diagnostic potential of antigens separated on acrylamide gels.

# Recommendations for Further Validation of IgG1 Assays

We propose that a prospective study, with extended follow-up of a larger cohort of treated VL patients, should be used to validate the use of IgG1 ELISA and the VL Sero K-SeT for confirming cure in all endemic areas and defining the optimal time for testing, which may differ between regions. This longer follow-up would also indicate the potential of elevated IgG1 to predict relapse and PKDL and in turn, link these with different treatment regimens. A more extensive study of PKDL is required to determine the potential role of IgG1 in identifying PKDL as distinct from leprosy and fungal skin diseases (Saha et al., 2005; Mondal and Khan, 2011). In addition, use of the IgG1 assays on a much larger panel of seropositive asymptomatic individuals would help to define its role in predicting progression to VL. In all cases, comparison with existing diagnostics, including definitive parasitological methods, would directly assess the advantage of IgG1 assays.

Technical refinement of the VL Sero K-SeT should consider the use of electronic RDT readers to give an objective assessment of test band intensity. In addition, the identification of specific antigens suitable to replace the use of parasite lysate would obviate issues regarding batch-to-batch variation. These developments could improve precision of IgG1 readings and reproducibility. A comparison of whole blood and serum/plasma is also required for point-of-care use, although a study in Bangladesh on various VL RDTs did find high agreement between the two sample types (Ghosh et al., 2015).

# CONCLUSION

IgG1 assays, particularly in the VL Sero K-SeT RDT format, may be a useful adjunct in the assessment of VL treatment outcome and diagnosis of PKDL, which have been identified as research priorities for VL (World Health Organization, 2012). With additional refinement and validation, the VL Sero K-SeT and IgG1 ELISA could contribute to life-saving follow-up of treated patients and to control programme monitoring, surveillance, and targeting of strategies for long-term control of VL.

# AUTHOR CONTRIBUTIONS

MM conceived and designed the study. TM, TB, and MM wrote the manuscript. TM, TB, CP, BG, SA, MdlR, KH, and HH performed the experiments, collected data. TM, TB, CP, SA, MdlR, KH, and HH analyzed data. OS, PM, QG, CT, BH, AF, OE, AS, BK, NB, SR, SE-S, and SS provided materials. AF, MM, OS, and PM provided feedback on final draft. MM, TB, and TM supervised the project. MB and MM obtained funding.

# FUNDING

This work was part of the NIDIAG network research partnership supported by the European Commission under the Health Cooperation Work Programme of the 7th Framework Programme (Grant agreement no. 260260, http://cordis. europa.eu/fp7/home\_en.html). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. TM is funded by the Sir Halley Stewart Trust (http://www.sirhalleystewart.org.uk/). The views expressed within this report are those of the authors and not necessarily those of the Trust. TM was additionally supported by the John Henry Memorial Fund (http://www.johnhenrymf.org/). BH is funded by the European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No. 642609. (https://ec.europa.eu/programmes/ horizon2020/).

# ACKNOWLEDGMENTS

We thank Abhishek Kumar Singh and Keshav Rai for their kind assistance during the laboratory work.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2018.00427/full#supplementary-material

Supplementary Material S1 | Detailed SDS-PAGE and western blotting methods.

Supplementary Material S2 | Spreadsheet with ELISA, RDT & blot results for Indian cured paired, PKDL and relapse samples.

Supplementary Material S3 | Images of western blots for Indian paired cured samples.

Supplementary Material S4 | Images of VL Sero K-SeT for Indian paired cured samples, corresponding to those in S3.

Supplementary Material S5 | Images of VL Sero K-SeT and western blots for Indian and Sudanese relapsed samples.

Supplementary Material S6 | Images of VL Sero K-SeT and western blots for Indian PKDL samples.

Supplementary Material S7 | Images of VL Sero K-SeT and western blots for Indian asymptomatic progressors and non-progressors.

# REFERENCES


treating visceral leishmaniasis under routine conditions in eastern Sudan. Trop. Med. Int. Health 20, 1674–1684. doi: 10.1111/tmi. 12603


relapse in visceral leishmaniasis, and adaptation to a rapid diagnostic test. PLoS Negl. Trop. Dis. 8:e3273. doi: 10.1371/journal.pntd.0003273


of Malda district, West Bengal, India. PLoS Negl. Trop. Dis. 11:e0005391. doi: 10.1371/journal.pntd.0005391


study (2000-2010). PLoS Negl. Trop. Dis. 5:e1433. doi: 10.1371/journal.pntd.00 01433


**Conflict of Interest Statement:** PM, QG, and CT are employed by Coris BioConcept which developed the VL Sero K-SeT.

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 © 2018 Marlais, Bhattacharyya, Singh, Mertens, Gilleman, Thunissen, Hinckel, Pearson, Gardner, Airs, de la Roche, Hayes, Hafezi, Falconar, Eisa, Saad, Khanal, Bhattarai, Rijal, Boelaert, El-Safi, Sundar and Miles. 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.

# Biomarkers in Post-kala-azar Dermal Leishmaniasis

Eduard E. Zijlstra\*

Rotterdam Centre for Tropical Medicine, Rotterdam, Netherlands

Post-kala-azar dermal leishmaniasis (PKDL) follows visceral leishmaniasis (VL, kala-azar) in 10–60% of cases. It is characterized by an asymptomatic skin rash, usually starting in the face and consisting of macules, papules, or nodules. Diagnosis is difficult in the field and is often made clinically. There is an extensive differential diagnosis, and parasitological confirmation is preferred particularly when drug treatment is considered. The response to treatment is difficult to assess as this may be slow and lesions take long to heal, thus possibly exposing patients unnecessarily to prolonged drug treatment. Biomarkers are needed; these may be parasitological (from microscopy, PCR), serological (from blood, or from the lesion), immunological (from blood, tissue), pathological (from cytology in a smear, histology in a biopsy), repeated clinical assessment (grading, photography), or combinations. In this paper, we will review evidence for currently used biomarkers and discuss promising developments.

### Edited by:

Eugenia Carrillo, Carlos III Health Institute, Spain

## Reviewed by:

Dinesh Mondal, International Centre for Diarrhoeal Disease Research (ICDDR), Bangladesh Prakash Ghosh, International Centre for Diarrhoeal Disease Research (ICDDR), Bangladesh Carlos Henrique Costa, Federal University of Piauí, Brazil

\*Correspondence:

Eduard E. Zijlstra e.e.zijlstra@roctm.com

### Specialty section:

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

> Received: 09 April 2019 Accepted: 11 June 2019 Published: 31 July 2019

### Citation:

Zijlstra EE (2019) Biomarkers in Post-kala-azar Dermal Leishmaniasis. Front. Cell. Infect. Microbiol. 9:228. doi: 10.3389/fcimb.2019.00228 Keywords: biomarkers, clinical, parasitological, biochemical, immunological, post-kala-azar dermal leishmaniasis

# INTRODUCTION

Visceral leishmaniasis (VL, kala-azar) is most common in Asia (India, Bangladesh, Nepal), East Africa (Sudan, South Sudan, Ethiopia, Kenya, Uganda), where it is caused by Leishmania donovani, and South America (Brazil), where Leishmania infantum is the causative parasite. Interestingly, almost exclusively, VL cases caused by L. donovani may be followed by post-kala-azar dermal leishmaniasis (PKDL), be it not in a uniform manner. In Africa (Sudan), PKDL is much more common (up to 50–60% of VL cases) with mainly papulonodular lesions, compared with Asia (5– 20%) where most cases show macular lesions (Zijlstra et al., 2003, 2017) (**Figures 1**, **2**). In addition, the interval between VL and PKDL is short in Africa (< 12 months), whereas in Asia, it is often 3– 5 years or more (Zijlstra et al., 2003). The underlying mechanism that determines development of PKDL is not completely known but may very well lie in factors that influence the evolving immune response to the parasites that can be found in the skin lesions (**Figure 3**). In VL patients the predominant immune response switches from a Th2 into a Th1 profile as the result of treatment, and PKDL patients are thought to have a dissociated immune response. While systemically this will be mainly Th1, the response in the skin may be still be Th2, possibly under the influence of UV light; persistence of IL-10 plays a prominent role (Gasim et al., 1998, 2000; Ismail et al., 2006; Zijlstra, 2016). The immune response varies according to clinical type and is stronger in macular PKDL than in papulonodular PKDL (Haldar et al., 1983; Saha et al., 2007; Katara et al., 2011; Mukhopadhyay et al., 2012).

Treatment induces a change in the immune response both by its antileishmanial effect and by intrinsic effects of the drugs used (Ansari et al., 2008; Mukhopadhyay et al., 2011). It follows that the effect of treatment or self-healing will be reflected in changes in clinical features, the parasite load,

FIGURE 1 | Typical papular rash in a patient from Sudan.

FIGURE 2 | A macular rash in a patient from Bangladesh; the macules vary in size and some are confluent.

immune parameters in cell-mediated immunity (cell profile, cytokines, chemokines), as well as in humoral parameters such as antibody levels. In drug treatment, it is likely that parasitological cure precedes immunological cure that in its own right precedes clinical cure, with unknown intervals (Zijlstra et al., 2017). Parameters that indicate these changes may therefore be sought in each of these categories and as such have a different clinical meaning. In the case of self-healing, this process is spontaneous but most likely induced by the immune response; here, the intervals between events are equally unknown. Experience from various studies indicates that the healing process is slow, particularly in macular lesions. The reduction in size of the macular lesion is difficult to appreciate as the repigmentation process is slow; this process may take months if not years to complete (WHO, 2013; Verma et al., 2015).

Biomarkers (or biological markers) are a broad category of medical signs that reflect the medical state from outside the body and may include physical signs found on examination of the patient, basic chemical measurements, and more complex tests of blood and other tissues (Strimbu and Tavel, 2010).

FIGURE 3 | Parasites can be seen in a skin biopsy taken from a PKDL lesion.

Identifying biomarkers in PKDL is hampered by the lack of adequate studies on diagnosis of PKDL that show considerable heterogeneity. This is due to lack of consistent reference standards, emphasizing the need for well-designed trials to assess diagnostic accuracy (Adams et al., 2013). In addition, it is essential to interpret identified biomarkers of leishmania infection in the context of the pathophysiology of PKDL, which is different from VL or cutaneous leishmaniasis (Kip et al., 2015). In recent years, considerable progress has been made in research on diagnosis, pathophysiology, and immunology of PKDL and VL, which has increased our understanding of these conditions and how they relate to each other, before and during treatment.

In practice, under field conditions, neither parasitological assessment (microscopy, PCR), biochemical parameters (antigen or antibody-based tests in blood or urine), nor immunological markers [cytokines, lymphocyte subsets, leishmanin skin test (LST)] are used and clinical assessment is often the only tool available. In hospital-based (research) laboratories, some of these tools are routinely used or under investigation.

## Special Types of PKDL

PKDL is more frequent and more severe in HIV co-infection (Abongomera et al., 2019); skin lesions may precede, accompany, or follow VL in HIV co-infection, some of which may be referred to as PKDL (Zijlstra, 2014). As the pathophysiology is completely different, this category will not be discussed.

In this paper, we will review available information on (potential) biomarkers in classical PKDL (i.e., following {successful} treatment of VL) and discuss promising developments.

## CLINICAL BIOMARKERS

## Baseline Assessment

Clinical assessment of PKDL at first presentation includes recording and description of individual lesions or groups of lesions. In addition, the presence or absence of systemic symptoms and signs is recorded; in 10% of patients, PKDL occurs concomitantly with VL, as evidenced by fever, splenomegaly, hepatomegaly or lymphadenopathy, and poor nutritional status (para-kala-azar dermal leishmaniasis) (Zijlstra et al., 2003). This distinction is important as the approach to treatment may be different: in case of suspected concomitant VL, parasitological confirmation needs to be sought and systemic treatment is given, by which the PKDL lesions are also treated simultaneously. In addition, in the case of PKDL without systemic VL, in Asia, all patients are treated, while in Africa (Sudan), only those with severe PKDL are treated as the majority of cases will self-heal (Musa et al., 2002; Zijlstra et al., 2003).

The differential diagnosis may be different in Asia and Africa but usually includes leprosy, vitiligo, and miliaria rubra (WHO, 2013). To assist field workers, guidelines for diagnosis, a PKDL atlas and an online self-teaching course have been designed, all by WHO (2012, 2013, 2017). Misdiagnosis is common (el Hassan et al., 2013) and reported in up to 26% of cases in India (Ramesh et al., 2015a). Histopathological examination of biopsies will also be of use to distinguish between differential diagnoses (el Hassan et al., 1992; Singh and Ramesh, 2013; Verma et al., 2015).

In any PKDL patient, typically the appearance of the lesions is described as macules, papules, nodules, plaques, or a mixed from. There are major differences between regions (Zijlstra et al., 2003; WHO, 2013). In Africa (mainly Sudan), a maculopapular rash (90% of cases) is most common, and in advanced cases, the papules will increase to form nodules or plaques; a pure macular rash is uncommon. In Asia, a macular rash is more common (90% of cases in Bangladesh); in hospital settings, the most common presentation may be mixed/polymorphic (53%), followed by macular lesions (23%) and papulonodular lesions (21%); unusual forms include the erythrodermic, and fibroid type, or presentations with plaques or ulcerations (Ramesh et al., 2015a; Verma et al., 2015). In contrast with Africa, advanced cases with massive lesions have been described in India (WHO, 2013; Sethuraman et al., 2017).

While the principal localization of the rash is often in the face, starting around the mouth, the rash may spread to the upper chest and arms, often corresponding with areas of the body that are not covered by clothing. Over time, or in severe cases, all parts of the body may be covered by the rash, with varying degrees of density. To describe the rash in terms of density and distribution, a grading system was designed for semi-quantitative assessment. The first was developed in Sudan and focused on distribution; later, this was refined to also include density and to describe discrepancies between these two observations. For example, a patient may have lesions all over the body (distribution grade 3), but with mostly normal skin in between (density grade 1). This would be called grade 3.1 (**Table 1**) (Zijlstra and el-Hassan, 2001; Musa et al., 2008).

Alternatively, the distribution of the lesions can be plotted on a manikin as was designed in Bangladesh and the severity is assessed by counting the number of affected squares (Mondal et al., 2016) (**Figure 4**).

TABLE 1 | The PKDL grading system as reported from Sudan (Zijlstra et al., 2000; Zijlstra and el-Hassan, 2001).


Monitoring During Treatment or Follow-Up While clinically, improvement or cure may be defined as flattening of lesions, improvement of dyschromia, and healing of complications, a more accurate assessment is needed particularly in drug treatment studies (Abongomera et al., 2016).

To improve accuracy of clinical (visual) description, various auxiliary methods were introduced such as clinical photography, clinical recording of severity, and distribution on various body parts, by comparing lesions as plotted on a manikin, or by comparing a clinical score in combination with a parasitological score.

Regular two-dimensional photography was introduced and standardized to accurately document lesions and allow objective comparison. This includes using the same camera, the standardized distance between camera and patient, the background of the patient, lighting in the studio, etc. The images may be interpreted by two or more independent observers who remain blinded to the patient data.

In the MSF program at Fulbaria, Bangladesh, AmBisome was used on an outpatient basis with 6 infusions of 5 mg/kg administered over 3 weeks (total dose, 30 mg/kg). After 12 months' follow-up, 34% were cured and 40% had 70–80% reduction of lesions. All nodular and papular lesions showed complete recovery while complete or significant repigmentation of macular lesions was observed in 86.5% of patients. In a subset of the latter group (n = 20), PCR was done on slit skin smears; all were negative. In this study, two-dimensional photography was used for documentation and comparison (Zijlstra et al., 2017; den Boer et al., 2018).

In another study in Bangladesh by MSF, with 15 mg/kg AmbiSome total dose, the lesions were also photographed by a standardized method and analyzed by three experienced physicians. In addition, a weighed core was calculated. At baseline, a percentage was assigned for the relative measure in which each affected body part (face, torso, arms, legs) contributed to the total burden of lesions.

The improvement of lesions was compared to the baseline for each affected body part and was recorded in percentages. The relative improvement was calculated for each body part: Relative improvement = {contribution at baseline (%) × improvement (%)}/100 (den Boer et al., 2018).

The overall improvement for each patient was calculated by: overall improvement = the sum of the relative improvement of each body part.

Final cure was defined on clinical grounds as complete resolution of nodular and papular lesions, and complete to major repigmentation of macular lesions. At 12 months, the final outcome was scored after carefully evaluating all photos of each patient in one of the following descriptive categories. This was done independently by each member of the study team (three persons) as well as by an independent external evaluator.


In the study from Bangladesh by Mondal et al., a more refined method was used. Here, skin lesions were plotted by visual assessment in squares on a manikin (**Figure 4**) and the total amount of squares with lesions is counted before and after treatment. The percentage of skin lesions affected is then counted at various timepoints as a percentage: total number of squares free from lesions after treatment/total number of squares affected by the lesions at baseline (Mondal et al., 2016) (**Figure 4**).

In India, another method for assessment was used. In a study on the effect of miltefosine, papules and nodules were assessed rather than macular lesions and at least at three sites of the body. Two efficacy parameters were considered: a clinical score (from 0 to 6) based on numbers of papules and/or nodules, and a parasitological score (from 0–5) based on numbers of amastigotes (from 0/1000 fields to >10/field) (Sundar et al., 2013). Clinical cure was defined as follows: at 12 months, clinical score = 0 for all three locations and a parasitological score = 0 when last measured after treatment.

Three dimensional optical scanning is a novel tool to measure PKDL lesions as was demonstrated from Sudan. Optical 3-D scanning may potentially be used for surface scanning of any body part and has been tested in the assessment of burn injuries as well as mycetoma (Telfer and Woodburn, 2010; Retrouvey et al., 2018; Siddig et al., 2018). It may be a useful tool for accurate computerized measurement of lesions in patients with other tropical dermatological conditions. The 3-D scanning images may be quantified in terms of surface, circumference, and diameter, with an accuracy of 0.5 mm. In addition, the height of a lesion can be measured and sequential images can be compared to quantify changes in size and color including repigmentation as in macular lesions.

# LABORATORY BIOMARKERS

# Parasitological Biomarkers

A confirmed diagnosis of PKDL is preferred and is mandatory in research studies.

This may be done by demonstration of leishmanial parasites by microscopy in a slit skin smear, micro-biopsy, fine needle aspirate (FNA), or conventional biopsy, with limited sensitivity of 32–50%. Parasites can more easily be found in up to 95% of mixed papulonodular lesions and in only up to 40% in macular lesions; biopsies from the buccal mucosa or the tongue have higher yield (Ramesh et al., 2015b; Verma et al., 2015).

In a recent study, comparing the slit skin smear technique (SSS) and tissue biopsy, all cases with macular lesions (n = 4) were negative in microscopy, while in papular lesions, 2/17 and 10/20 were positive in microscopy, in SSS and biopsy, respectively; for nodular lesions, 13/26 and 20/26 patients were microscopy positive in SSS and biopsy, respectively (Bhargava et al., 2018). The technique of performing an SSS is clearly demonstrated in Verma et al. (2013b).

PCR has higher sensitivity than microscopy, but a wellequipped lab is required. PCR has been first explored in the diagnosis of PKDL in Sudan by Verma et al. (2013b) and extensive further studies were done in Asia (Osman et al., 1998; Salotra et al., 2001; Mondal et al., 2010). Later developments included RFLP analysis and nested PCR that increased sensitivity from 69 to 93% (Schonian et al., 2003; Sreenivas et al., 2004). More recent studies show that parasite DNA is detected by PCR in a slit skin smear or biopsy in 96–100% of cases (Ramesh et al., 2015a,b; Sundar et al., 2015). PCR proved more sensitive than immunohistochemistry in biopsies from PKDL patients (Salotra et al., 2003).

qPCR or real-time PCR allows detection and quantification of a number of parasites. A summary of studies performed is given in **Table 2**.

There are few studies on the value of qPCR as a tool to detect parasites after treatment. Patients with a higher parasite load as measured by qPCR are at higher risk of relapse; in 30 patients studied, 26/30 were negative in qPCR 1 month after treatment, while 4 showed residual parasites, of whom 2 relapsed (Ramesh et al., 2015b) (**Table 2**). In a study in patients treated with miltefosine, of 15 patients sampled after 60 or 90 days post-treatment, all had become parasitologically negative as determined by qPCR from a slit skin smear (Ramesh et al., 2011). In another study, 17/19 patients became negative in qPCR 1 month after treatment with SSG or miltefosine; in 2 patients, a residual parasite load was found (7 and 8 parasites/µl slit aspirate, respectively); 1 of these relapsed (Verma et al., 2013b). Moulik et al. (2018) measured the parasite load by qPCR after treatment with miltefosine for 3 months; all had become negative by qPCR. In contrast, those treated with LAMB had higher parasite loads after 3 weeks of treatment that increased 6 months after treatment, thus predicting relapse (Moulik et al., 2018) (**Table 2**).

A field-friendly adaptation for DNA detection such as (closed tube) loop-mediated isothermal application (LAMP) is currently being explored (Verma et al., 2013a, 2017).

# Biochemical and Immunological Biomarkers

Antibody-based serological diagnosis with the Direct Agglutination Test (DAT) or rK39 ELISA is not helpful as antibodies may persist from the preceding VL episode. The rK39 rapid diagnostic test may be done directly on the lesions with good sensitivity but unknown specificity (Verma et al., 2013b). It has not been evaluated after treatment. A novel application is the use of the rK39 RDT in sweat samples of VL and PKDL patients with 96.6 and 94.7% sensitivity and specificity (as measured in healthy controls), respectively, with 100% concordance with blood specimens (Topno et al., 2018).

In a study from India, there was a non-significant decrease in DAT positivity rate of 75 and 66%, before and after treatment of PKDL. An assay for the migration inhibition factor showed that 70 and 100% of patients were positive, before and after treatment, respectively (Verma et al., 2015). Newer tests include the circulating immune complexes (CICs) containing glycoproteins; this test was found useful in the monitoring of PKDL patients and in distinguishing between drug-responsive and drug-unresponsive patients (Jaiswal et al., 2018).

An early study showed that IgG1 and IgG3 were significantly raised in polymorphic PKDL, while in macular PKDL, only IgG1 was elevated (Mukhopadhyay et al., 2012). The measurement of IgG1 by ELISA or by a novel rapid diagnostic test named VL sero K-SeT have recently shown to predict relapse or cure in treated VL patients. Evaluation in PKDL showed that the VL sero K-set and IgG1 ELISA supported PKDL diagnosis with a strong correlation with post VL samples, suggesting persistence of antibodies after VL. No association was found between elevated IgG1 and macular or polymorphic PKDL lesions (Marlais et al., 2018). IgG1 levels did not show a consistent decrease 1 year after PKDL treatment (Marlais et al., 2018).

Immunological parameters include measurement of immune cells (lymphocytes, monocytes, macrophages) as well as cytokines or chemokines, direct in the serum or after stimulation of peripheral blood mononuclear cells (PBMCs), or in the skin by immunohistochemistry. Serum cytokines may be measured quantitively and expressed as a ratio, referred to as the cytokine polarization index (CPI). A CPI of interferon-γ (IFN-γ) vs. IL-10 was significantly lower in PKDL cases, compared to asymptomatic VL cases, while a CPI of TNF-α vs. IL-10 was also lower, but this did not reach statistical significance (Singh et al., 2018). A similar study showed that the ratio of TNF-α (inflammatory)/IL-10 (antiinflammatory) message was 2.66 and 1.18 in PKDL (skin biopsies) and VL (bone marrow aspirates), respectively, showing the importance of the dynamics of the cytokine profiles in various disease manifestations (Ansari et al., 2006). One study noted elevated IFN-γ transcripts after miltefosine treatment of PKDL, whereas this is not noted after antimonial therapy, suggesting that the immune response may differ according to the immunomodulatory properties of the treatment given (Ansari et al., 2008; Mukhopadhyay et al., 2011; Zijlstra, 2016). Similar effects were described for serum arginase activity (decreased) and increased serum nitrate (increased) after miltefosine treatment of PKDL, possibly indicating a macrophage activating effect (Mukhopadhyay et al., 2011).

Matrix metalloproteases (MMPs) are chemokines that are involved in tissue remodeling and leukocyte recruitment, and inhibitors thereof such as tissue inhibitor of matrix metalloproteases 1 (TIMP1) have been studied in PKDL. Using serum, MMP9 levels and the ratio of MMP9/TIMP1 were found elevated in active PKDL while patients with resolved PKDL lesions had levels similar to controls (Islam et al., 2013). This is interesting as MMP9 influences collagen degradation and mediates basement membrane modeling and TIMP inhibits activated MMP, thus possibly reflecting steps in the healing process of PKDL (Islam et al., 2013).

Adenosine deaminase (ADA) activity is an aspecific marker of the immune response and is present in all tissues (isoenzyme ADA-1) or in monocytes and macrophages (ADA-2). Serum ADA levels are raised at diagnosis of PKDL and gradually decrease during treatment. It was suggested to use the test with an rK39 RDT (specific for leishmaniasis) to increase specificity (Vijayamahantesh et al., 2016).

Cell-mediated immunity may be assessed by the in vivo LST. In Sudan, the LST did not discriminate between patients who, after VL, developed PKDL and those who did not as 49% and 42% of patients, respectively, had a positive LST, while all patients had been LST negative during VL (Zijlstra et al., 2000). However, patients who had parasites demonstrated in a lymph node or bone marrow aspirate during PKDL diagnosis had a positive LST in 11%, while those with a negative aspirate were

TABLE 2 | Summary of studies that use qPCR in diagnosis of PKDL, including those that use qPCR during follow-up.


(Continued)

TABLE 2 | Continued


\*Number of parasites detected by microscopy: grade 1: 1–10 per 1,000 fields; grade 2: 1–10 per 100 fields; grade 3: 1–10 per 10 fields. Difference not significant (p = 0.2457).

significantly more likely to be LST positive (37%), suggesting a developing immune response associated with clearance of parasites (Zijlstra et al., 2000). A positive LST was associated with grade: those with PKDL grade 1 were LST positive in 39%, while those with grade 2 and 3 were LST positive in 25 and 24%, respectively (Zijlstra et al., 2000). In the only paper on the natural history of PKDL by Musa et al., 134 patients with PKDL were followed up for 12 months, using clinical assessment (grading), DAT, and LST as biomarkers. All patients had a negative LST at diagnosis. At 12 months' follow-up, those who showed self-healing had lower DAT titers over time and were more likely to develop a positive LST (89.4%), reflecting a Th1 response, whereas those with persistent PKDL after 12 months had high DAT titers and a negative LST, reflecting a Th2 response (Musa et al., 2002).

There is evidence of a stronger immunological response in macular PKDL with strong cell-mediated immunity, low numbers of parasites, and low antibody levels (only IgG1 is elevated), whereas in polymorphic PKDL, the CMI is low due to the effect of TGF-β and IL-10, with higher levels of markers for regulatory T cells, more parasites, and high antibody levels, including both IgG1 and IgG3 (markers for IL-10) (Haldar et al., 1983; Saha et al., 2007; Mukhopadhyay et al., 2012).

In another study from Sudan, the effect of immunochemotherapy in PKDL patients with persistent lesions was studied; patients were allocated standard treatment with SSG and placebo or SSG with alum/ALM+BCG vaccination. Cure was assessed by comparing grade scores before and after treatment, and by the ratio of IFN-γ/IL-10, and conversion in the LST (Musa et al., 2008). Cure rates were higher in the vaccinated group that showed increased IFN-γ production followed by conversion in the LST. The slow responders had unchanged IFN-γ and increased IL-10 levels, suggesting that IL-10 blocks the action of IFN-γ, leading to persistence of lesions; most had a non-reactive LST (Musa et al., 2008).

Other tools for cell-mediated immunity include PBMCs; increased IFN-γ may be found after ex vivo stimulation of blood samples with soluble leishmania antigen (SLA). In VL, these markers indicate clinical cure, but no studies have been done in PKDL (Botana et al., 2018).

# DISCUSSION

The evolution of PKDL lesions is difficult to assess accurately, either during self-healing or after treatment. Clinical assessment as biomarker is unsatisfactory because of subjectivity and thus limited accuracy. The use of two-dimensional photograpy has limitations as it is difficult to standardize, and interpretation is subjective. Novel three-dimensional optical scanning shows promise with objective measurement of index lesions with superior accuracy and monitoring changes in size and color. Further prospective studies are awaited. Laboratory tests offer options for qualitative assessment but still have limitations, although qPCR seems the tool of choice in drug trials to assess parasite load. There is considerable heterogeneity in results from various studies, including measurements in clinical types (macular, papulonodular); this may be due to, among others, regional differences, genes targeted, clinical characteristics (duration, size, self-healing), and sampling technique. Slit skin smear or aspirate is the preferred method and is more sensitive and patient-friendly than a biopsy. Using qPCR, the parasite load can be monitored during treatment. If still positive after treatment, it is not clear what this indicates in relation to study treatment duration and whether this test will become negative over time; not all who have residual parasites as measured by qPCR will relapse. Longitudinal studies are essential in this respect. Adaptations of qPCR for use in the field are eagerly awaited.

Serological tests such as DAT, rK39 ELISA or rK39 RDT lack specificity as antibodies persist from previous VL. RDT rK39 direct on skin lesions has not been evaluated during or after treatment. Assessment of the developing immune response would be most useful as this may predict the risk of cure or relapse. In vitro measurement of the cytokines, chemokines, or lymphocyte subsets should be explored, and a CPI or ratio may be examined further to identify the most accurate immunological profile associated with cure. In vivo application of the LST deserves further study; the leishmanin needs to be well standardized and validated and requires production under good manufacturing practice.

In summary, assessment of the parasite load by qPCR in a slit skin smear or aspirate seems the most preferred biomarker to objectively measure the response to treatment and is more

# REFERENCES


patient-friendly than tissue biopsy. However, a simpler and less invasive method is preferred. Given the differences of the systemic immune reponses and those of the skin, it remains to be seen if systemic biomarkers in the blood sufficiently reflect the changes in the skin.

# CONCLUSIONS

The current biomarkers in PKDL lesions are unsatisfactory and new approaches need to be explored.

For early detection of cure, a combination of a parasitological assessment using qPCR and/or an immunological assessment such as a cytokine, chemokine profile, or lymphocyte subset profile would be preferred as these are intrinsically related. Longitudinal studies are needed to describe the dynamics of this interaction in the pathophysiology of PKDL, before, during, and after cure.

# AUTHOR CONTRIBUTIONS

The author confirms being the sole contributor of this work and has approved it for publication.

features, pathology and treatment. Trans. R. Soc. Trop. Med. Hyg. 86, 245–248. doi: 10.1016/0035-9203(92)90294-M


with indian PKDL: a study from endemic districts of west bengal, india. PLoS ONE. 13:e0192302. doi: 10.1371/journal.pone.0192302


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

Copyright © 2019 Zijlstra. 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