Regulatory role of B cell-derived transforming growth factor-β1 expression in autoimmune neuroinflammation

Thesis

Reference

Regulatory role of B cell-derived transforming growth factor-β1 expression in autoimmune neuroinflammation

BJARNADOTTIR, Kristbjorg

Abstract

Data indicate a role for TGF-β1 expression in regulatory B cell functions, although this mechanism has not been tested in autoimmune neuroinflammation. Transgenic mice deficient for TGF-β1 expression in B cells (B–TGF-β1–/–) were tested in EAE induced by recombinant mouse MOG (rmMOG) or adoptive transfer EAE (at-EAE). In rmMOG EAE, B–TGF-β1–/–

mice showed an earlier disease onset compared to littermate controls. Exacerbated EAE susceptibility was associated with augmented CNS T helper cell responses. Additionally, selective B cell TGF-β1–deficiency increased the frequencies and activation of myeloid DCs.

Lack of TGF-β1 production by B cells did not influence the EAE course in at-EAE. Collectively our data suggest that in rmMOG EAE B cells can regulate the function of APCs, and in turn encephalitogenic Th1/17 responses, via TGF-β1. An in vitro study of human B cells further indicated that TGF-β1 is downregulated when B cells are activated.

BJARNADOTTIR, Kristbjorg. Regulatory role of B cell-derived transforming growth factor-β1 expression in autoimmune neuroinflammation. Thèse de doctorat : Univ.

Genève, 2016, no. Sc. 5028

URN : urn:nbn:ch:unige-916298

DOI : 10.13097/archive-ouverte/unige:91629

Available at:

http://archive-ouverte.unige.ch/unige:91629

Disclaimer: layout of this document may differ from the published version.

UNIVERSITE DE GENEVE

Département de Biologie Cellulaire FACULTE DES SCIENCES Professeur Didier Picard Département de Pathologie et Immunologie FACULTE DE MEDECINE

Professeur Patrice Lalive ______________________________________________________________

Regulatory role of B cell-derived transforming growth factor-β1 expression in autoimmune

neuroinflammation

THESE

présentée à la Faculté des sciences de l’Université de Genève pour obtenir le grade de Docteur ès sciences, mention biologie

par

Kristbjörg Bjarnadóttir Reykjavík (Islande)de

Thèse n° 5028

Atelier d’impression ReproMail

Genève 2016

1 Table of Contents

ABBREVIATIONS 4

RESUME EN FRANÇAIS 6

ABSTRACT 12

2 INTRODUCTION 16

2.1 EXPERIMENTAL AUTOIMMUNE ENCEPHALOMYELITIS 16 2.1.1 EXPERIMENTAL AUTOIMMUNE ENCEPHALOMYELITIS AS A MODEL FOR

T CELL-MEDIATED AUTOIMMUNITY 16

2.1.2 PEPTIDE MYELIN OLIGODENDROCYTE GLYCOPROTEIN (MOG)-

INDUCED EAE IN C57BL/6 MICE 17

2.1.3 RECOMBINANT MOUSE MOG PROTEIN-INDUCED EAE IN C57BL/6

MICE 19

2.1.4 ADOPTIVE TRANSFER EAE MODEL IN C57BL6/J MICE 20 2.2 TRANSFORMING GROWTH FACTOR (TGF)-BETA 21

2.2.1 TGF-ΒETA ISOFORMS 21

2.2.2 TGF-ΒETA RECEPTORS AND SIGNALLING PATHWAYS 22 2.2.3 IMPORTANCE OF TGF-BETA ISOFORMS IN EMBRYONIC DEVELOPMENT

23

2.2.4 TGF-BETA1 FUNCTION IN IMMUNITY 24

2.2.5 ROLE OF TGF-ΒETA1 IN EAE 28

2.3 B CELLS 29

2.3.1 MULTIFACTORIAL B CELLS IN AUTOIMMUNE NEUROINFLAMMATION 29

2.3.2 REGULATORY B CELLS 31

2.3.3 REGULATORY IL-10-PRODUCING B CELLS 32 2.3.4 REGULATORY IL-35-PRODUCING B CELLS 36

2.3.5 TGF-ΒETA1 EXPRESSION BY B CELLS 37

2.3.6 HUMAN BREGS 39

3 RESULTS 41

3.1 CHAPTER I 42

3.2 CHAPTER II 63

4 DISCUSSION 95

5 REFERENCES 106

ACKNOWLEDGEMENTS 118

Abbreviations

AAD – Allergic airway disease AIA – Antigen-induced arthritis Ag – Antigen

APC – Antigen-presenting cell

at-EAE – Adoptive transfer experimental autoimmune encephalomyelitis BCDT – B cell depleting therapy

BCR – B cell receptor Bregs – Regulatory B cells

cDC – Conventional dendritic cell

CCIA – Chronic collagen-induced arthritis CFA – Complete Freund’s adjuvant CNS – Central nervous system dLN – Draining lymph node

EAE – Experimental autoimmune encephalomyelitis

GITR – Glucocorticoid-induced TNF receptor family-related protein hLN – Hilar lymph node

IL – Interleukin i.p. – Intraperitoneal i.v. – Intravenous KO – Knock out

LAP – Latent associated polypeptide LC – Langerhans cells

LIT – Local inhalation tolerance LTBP – Latent TGF-β binding protein lyDC – Lymphoid dendritic cell mLN – Mesenteric lymph node

MOG – Myelin oligodendrocyte glycoprotein myDC – Myeloid dendritic cell

nTregs – Natural regulatory T cells OVA – Ovalbumin

PAMPs – Pattern-associated molecular patterns pDC – Plasmacytoid dendritic cells

PD-L1 – Programmed death-ligand 1 PRR – Pattern-recognition receptors Ptx – Pertussis toxin

rmMOG – Recombinant mouse myelin oligodendrocyte glycoprotein s.c. – Subcutaneous

SNC – Système nerveux central SEP –Sclérose en plaques

SLE – Systemic lupus erythematosus TβR – TGF-β receptor

TCR – T cell receptor

TGF-β – Transforming growth factor β Th – T helper cells

TLR – Toll-like receptor Tregs – Regulatory T cells

T1DM – Type 1 diabetes mellitus

RESUME EN FRANÇAIS Introduction

La sclérose en plaques (SEP) est une maladie chronique inflammatoire démyélinisante et neurodégénérative du système nerveux central (SNC) qui touche approximativement 2,5 millions d’individus à travers le monde et pour laquelle il n’existe pas de traitement curatif.

De nombreuses observations indiquent qu’un type particulier de cellules immunitaires, les lymphocytes T, joue un rôle pathogénique essentiel dans la maladie. Plus récemment, les lymphocytes de type B ont également montré un rôle prépondérant dans la pathogenèse de la SEP et de son modèle murin, l’encéphalite auto-immune expérimentale (EAE). Il a en effet été démontré que l’administration de nouveaux traitements éliminant les lymphocytes B chez les patients atteints de SEP conduisait à une amélioration spectaculaire de la maladie. La question actuelle est de définir le rôle pathogénique exact que les lymphocytes B jouent dans le contexte de la SEP.

Des études entreprises chez les patients atteints de SEP ou dans l’EAE ont récemment démontré qu’il existe de nombreux types de lymphocytes B. Certains favorisent l’inflammation, alors que d’autres, les lymphocytes B régulateurs, peuvent la freiner. Si les thérapies visant l’élimination des cellules B se sont avérées très efficaces dans des essais cliniques avancés, leur sécurité à long terme reste à déterminer. Les thérapies peuvent éliminer une partie des cellules B régulatrices avec les cellules B pathogéniques donc en obtenant une meilleure idée de qui sont les cellules anti-inflammatoires et pro-inflammatoires, nous pouvons espérer

développer des traitements qui sont tout aussi efficaces et potentiellement encore plus sûrs.

Buts de l’étude

Les études entreprises dans l’EAE ont démontré que plusieurs populations de cellules B régulatrices co-existent, possédant non seulement des phénotypes distincts mais également des mécanismes d’action différents.

L’objectif de cette thèse est de déterminer un possible rôle régulateur de la production par les cellules B du facteur de croissance transformant bêta 1 (TGF-β1), une cytokine reconnue pour ses fonctions anti-inflammatoires.

Pour aborder cette problématique, nous avons développé deux approches complémentaires, l’une dans l’EAE, le modèle animal de la SEP, l’autre à l’aide de cellules B humaines. Le premier projet a fait l’objet d’une publication qui est intégrée dans cette thèse. Le second aspect de ce projet fait actuellement l’objet d’un manuscrit en préparation, qui figure également dans cette thèse.

Résultats

L’expression du facteur de croissance transformant bêta 1 par les lymphocytes B limite la phase d’induction de l’encéphalomyélite auto- immune expérimentale

De façon comparable aux actions suppressives exercées par les cellules T régulatrices (Tregs), la production de cytokines immuno- régulatrices puissantes a été notée dans les cellules B régulatrices. Bien que la fonction protectrice des cellules B dans l’EAE a été principalement associée à la production de l'interleukine (IL)-10 et de l’IL-35, d'autres données suggèrent que la production de TGF-β1 peut être tout aussi

importante dans la fonction des cellules B régulatrices. En effet un rôle de l'expression du TGF-β (il en existe trois isoformes) dans les activités régulatrices des lymphocytes a été suggéré dans des modèles de diabète auto-immun et d’inflammation intestinale. Ce mécanisme n'a pas encore été testé dans des modèles de SEP.

Dans le cadre de ce travail de thèse, nous avons démontré, et ce pour la première fois, un rôle régulateur des cellules B produisant le TGF-β1, une cytokine reconnue pour ses fonctions immuno-régulatrices. Pour évaluer l'importance de la production de TGF-β1 par les cellules B dans la régulation des réponse auto-immunes neuro-inflammatoires, nous avons conçu des souris transgéniques déficientes pour l'expression de TGF-β1 dans les cellules B (souris B-TGF-β1^–/–) et les avons testées dans un modèle d'EAE dans lequel l’action des cellules B contribue fonctionnellement au développement de la maladie. Par rapport aux souris contrôles, les souris B- TGF-β1^–/– ont développé plus tôt l'apparition de l'EAE. L’accélération des signes cliniques chez les souris B-TGF-β1^–/– était associée à une capacité accrue des lymphocytes CD4⁺ de type Th17 à produire le facteur de stimulation des colonies de granulocytes et de macrophages (GM-CSF), une cytokine pro-inflammatoire puissante qui exerce une fonction cruciale dans l’initiation des réponses auto-immunes. Malgré le rôle important du TGF-β1 dans la génération des Tregs, aucune différence dans les fréquences des Tregs n’a été observée. En revanche, nous avons observé que les souris B- TGF-β1^–/– contenaient une fréquence plus élevée de cellules dendritiques (DCs) myéloïdes classiques, ainsi qu'une augmentation des niveaux d'expression de surface des molécules CMH de classe II et CD86 par ces

cellules. Nos données suggèrent que la production de TGF-β1 par les lymphocytes B limite l‘induction de l’EAE, du moins en partie, en affectant les fonctions immunogènes des DCs myéloïdes. Collectivement, ces données montrent que la production de TGF-β1 par les cellules B est un mécanisme important dans la régulation des maladies auto-immunes à médiation cellulaire T telles qu’on l’observe dans le modèle animal de la SEP.

L’activation des lymphocytes B humains régule négativement la production de TGF-β1

L'une des premières études illustrant une fonction diminuée des cellules B régulatrices chez les patients présentant une maladie auto-immune a été réalisée dans le cadre de la SEP. La diminution de la production d’IL -10 a été observée lors de l'activation des cellules B par le CD40, de même que par le récepteur de lymphocyte B (BCR) en conjonction avec le CD40, ou via le récepteur de type Toll-9 (TLR9), ce qui indique une altération générale des fonctions des cellules B dans la SEP plutôt qu'un défaut dans certaines voies de signalisation activatrices. Semblable à l'action des cellules B produisant l’IL-10, les résultats obtenus dans le cadre de cette thèse ont démontré la capacité des cellules B produisant le TGF-β1 à réprimer la phase d'initiation de l’EAE, ce qui les rend potentiellement important dans le maintien de la tolérance immunitaire périphérique dans les maladie auto-immunes, telle que la SEP. L’étude des effets de l'activation des lymphocytes B humains sur la production de TGF-β1 peut fournir une meilleure compréhension de la régulation des fonctions régulatrices des lymphocytes B dans les manifestations auto-immunes.

Dans cette optique, nous avons mesuré la production du TGF-β1 et de l'IL-10, deux cytokines régulatrices, ainsi que celle des deux cytokines pro- inflammatoires, l’IL-6 et facteur de nécrose tumorale (TNF), par des cellules B humaines circulantes dans des conditions basales et après stimulation combinée du récepteur des cellules B (BCR) et du récepteur de type Toll (TLR)9. Nous avons observés que les cellules B naïves (CD27^–) et mémoires (CD27⁺) exprimaient spontanément le TGF-β1. En revanche, nous avons découvert que la production d'IL-10 était négligeable dans les cellules B dans des conditions de culture basales. De même, les cellules B au repos ne produisaient pas les cytokines pro-inflammatoires IL-6 et TNF.

Remarquablement, les cellules B stimulées ont montré une diminution des capacités de production de TGF-β1. Nous avons également noté qu’une fois activée, une faible, mais significative, fraction des cellules B mémoires, mais pas naïves, exprimait l’IL-10. Enfin, la stimulation a fortement induit la production d'IL-6 et de TNF par les cellules B. Ces résultats indiquent que l'activation des cellules B pourrait contribuer à l’occurrence d’anomalies immunologiques, souvent observées dans les maladies auto-immunes telle que la SEP, en limitant les la production de TGF-β1 par les cellules B tout en favorisant globalement leur fonctions B pro-inflammatoires.

Conclusions

Dans ce travail de thèse, nous avons en premier lieu démontré l’importance in vivo des cellules B produisant le TGF-β1 dans la régulation des réponses auto-immunes pro-inflammatoires de l’EAE, un modèle animal de la SEP. Dans un deuxième temps, nous avons identifié que l’activation in vitro des cellules B humaines, au travers un mécanisme hautement relevant

des maladies auto-immunes, diminuait leur capacité à produire le TGF-β1. Nos résultats contribuent à fournir une compréhension supplémentaire des mécanismes par lesquels les lymphocytes B participent à la progression et/ou régulation des réponses auto-immunes dans la SEP, et pourraient également permettent à l’élaboration de nouvelles thérapies visant à cibler plus spécifiquement les lymphocytes B pro-inflammatoires. Finalement, nos observations dans l’EAE, en tant que modèle prototypique des maladies auto- immunes induites par des lymphocytes T CD4⁺, laissent suggérer que l’effet régulateur des cellules B produisant le TGF-β1 serait un mécanisme général dans le contrôle de l’auto-immunité.

Abstract

Multiple sclerosis (MS) is a chronic inflammatory, demyelinating and neurodegenerative disorder of the central nervous system (CNS) that affects approximately 2.5 million people worldwide, for which there is currently no cure. T lymphocytes were traditionally considered to be the main actors of CNS inflammation. Over the past few years the clinical efficacy of B cell depleting therapies (BCDT) for MS has shifted research focus towards the role of B lymphocytes. B cells may contribute to MS activity though different mechanisms, including production of autoantibodies that target components of the CNS, presentation of autoantigens and co-stimulatory signals to activate T cells, and production of a range of proinflammatory cytokines that can amplify immune responses. The clinical benefit of BCDT in MS patients has been suggested to be related to the elimination of proinflammatory B cells rather than targeting potentially relevant humoral B cell responses.

In MS, B cells may additionally play a regulatory role through production of regulatory cytokines. Of interest, an increased proinflammatory monocytic phenotype has been described in some patients with MS and in its animal model, experimental autoimmune encephalomyelitis (EAE), after BCDT, possibly related to the depletion of regulatory B cells. Among other mechanisms, production of interleukin (IL)-10 by B cells has been found to limit the extent of inflammation in MS and EAE. B cells have also been shown to produce other regulatory cytokines such as IL-35 and TGF-β1. While IL-35- producing B cells have been reported to potently supress EAE, so far no studies have evaluated the importance of B cell-produced TGF-β1 in regulating autoimmune neuroinflammation. As current BCDT not only deplete

proinflammatory B cells, but also regulatory B cells, additional mouse studies are needed to identify the role of different B cell subsets in the regulation and progression of CNS autoimmunity. To avoid potential negative consequences of BCDT further development of B cell-directed therapies for treating MS patients are needed to spare regulatory B cell functions.

Our studies were aimed at evaluating the role of B cell-produced TGF- β1 in EAE. To address our research objectives, we generated transgenic mice deficient for TGF-β1 expression in B cells (B-TGF-β1^−/−) and tested them in EAE induced by immunisation with recombinant mouse myelin oligodendrocyte glycoprotein (rmMOG), a mouse model in which B cells are considered to contribute to the development of the disease. In this EAE model we found that B-TGF-β1^−/− mice had earlier disease onset compared to control mice as well as increased CNS inflammation and demyelination during the acute phase. Accelerated disease onset in B-TGF-β1^−/− mice was associated with increased proinflammatory Th1/Th17 cell responses in the CNS and enhanced immunogenic profile and frequencies of peripheral myeloid dendritic cells, potent antigen-presenting cells. Together these data suggest that TGF-β1-producing B cells restrain disease initiation by affecting the priming of encephalitogenic T cells. To confirm this hypothesis, we additionally showed that B cell-specific deletion of TGF-β1 did not modify the development of the clinical signs of adoptive transfer EAE, a model that bypasses the need of peripheral priming of T cells.

Finally, we evaluated TGF-β1 production by human peripheral blood B cells under both steady-state and inflammatory conditions and found that resting, but not activated, B cells expressed TGF-β1. These findings provide

support for the observations that B cells appear to be abnormally polarised toward a more proinflammatory phenotype in MS.

Together, our findings have established that B cell-derived TGF-β1 production plays a regulatory role in autoimmune neuroinflammation. These results further stress the importance of designing novel BCDT as elimination of regulatory B cell subpopulations, such as TGF-β1-producing B cells, in MS may result in exacerbation of disease activity.

General Introduction

2 Introduction

2.1 Experimental Autoimmune Encephalomyelitis

2.1.1 Experimental autoimmune encephalomyelitis as a model for T cell-mediated autoimmunity

Classical experimental autoimmune encephalomyelitis (EAE) is a predominantly CD4⁺ T cell-mediated animal model of autoimmune neuroinflammation ¹. Although EAE is commonly employed as a model of MS, due to similarities in hallmark clinical features, discrepancies exist between EAE and MS, especially when taking into account that EAE symptoms and disease course can depend on the animal, genetic strain, immunisation method, and autoantigen used. Additionally, even though different EAE models can mimic individual facets of MS there is no animal model capable of recapitulating all pathological features of MS ^2,3. Despite these limitations, EAE has been valuable in understanding the fundamental aspects of the MS pathology and has been used in development of some of the currently used MS therapies including glatiramer acetate and natalizumab ^4,5. It is of interest to note that although not all MS therapies were developed using EAE, most, if not all, currently administered MS therapies are effective in EAE, indicating its suitability as a model of MS ^2,6.

In its classic form, EAE consists of an induction phase and an effector phase. During the induction phase of EAE, adjuvant-induced inflammation leads to activation of separate populations of professional antigen-presenting cells (APCs) such as dendritic cells (DCs), macrophages and B cells ⁷. These varying APC subsets aggregate in secondary lymphoid organs and present

myelin antigens, in the context of MHC class II, to myelin-specific CD4⁺ T cells leading to activation, differentiation, and expansion of myelin-specific CD4⁺ T cells. In the effector phase, peripherally activated myelin-specific CD4⁺ T cells migrate into the central nervous system (CNS) where, after reactivation by CNS-associated or local APCs (DCs, macrophages and microglia) presenting endogenous myelin antigens, they trigger myeloid cell activation by producing proinflammatory mediators such as the encephalitogenic T helper cell (Th)17 cytokine GM-CSF ⁸. The infiltration of inflammatory cells into the CNS leads to formation of inflammatory lesions and demyelination of the neuronal axons resulting in ascending paralysis. The chronic phase of EAE follows the acute phase and is characterised by sustained/persistent inflammation and paralysis. While some EAE models follow a chronic course, some other follow a relapsing-remitting disease course ^1-3,9.

EAE, as a disease model, has important implications outside the context of MS and can, in its own right, be considered as a prototypic model of organ specific CD4⁺ T cell-mediated autoimmunity. EAE has been important in the discovery of a plethora of molecular and effector mechanisms that control CD4⁺ T cell activities including the importance of RORγT as a transcription factor in Th17 development ¹⁰, the functional importance of GM- CSF production by Th17 cells in autoimmunity ¹¹, and the proinflammatory effects of GM-CSF on myeloid cells ¹².

2.1.2 Peptide myelin oligodendrocyte glycoprotein (MOG)-induced EAE in C57BL/6 mice

There are a variety of animal models of EAE available. Each model can have distinct features, mirroring different aspects of the MS disease,

depending on the animal, genetic strain, induction method, and autoantigen used ². C57BL/6 mice are commonly used in EAE, given the relative ease of genomic modification and the increasing availability of immune modified strains. EAE induction by active immunisation with myelin oligodendrocyte glycoprotein (MOG)35-55 peptide, a major immunodominant encephalitogenic epitope for C57BL/6 mice, was first used by Mendel et al in 1995 and has since been considered as a potent and reproducible model of EAE ¹³. MOG35- 55-induced EAE predominantly elicits a CD4⁺ T cell-driven immunological response. Active EAE in C57BL/6 mice is generally induced by subcutaneous administration of MOG35-55 peptide emulsified in complete Freund’s adjuvant (CFA) into naïve animals along with injections of pertussis toxin (Ptx) on day 0 and 2 ¹⁴. CFA contains killed Mycobacterium Tuberculosis, and pathogen- associated molecular patterns (PAMPs) from these bacteria. PAMPS bind to pattern-recognition receptors (PRRs) such as toll-like receptors (TLRs) that are essential in connecting innate and adaptive immunity by activating APCs that are instrumental for the priming of antigen-specific adaptive immunity ¹⁵. Ptx has been suggested to promote immune cell infiltration into the CNS ¹⁶, to increase cytokine production by Th1 cells ¹⁷, and to enhance Th1/Th17 generation by inducing IL-1β production by DCs ¹⁸. MOG35-55-induced EAE has been instrumental in enhancing the understanding of MS pathophysiology. Mechanisms elucidated by this animal model include the redundancy of the proinflammatory cytokine IL-12 ^19-21 and the crucial role of Th17 cells in driving CNS autoimmunity ²². Although IL-17 is dispensable for disease development by Th17 cells ²³, its other major cytokine, GM-CSF, is essential for EAE development ¹¹. Additionally, APC production of the Th17

differentiation factor IL-23 in the periphery and CNS was shown to be indispensable in the development of active EAE ^24,25. Although MOG35-55- induced EAE is mainly considered as a CD4⁺ T-cell mediated organ specific autoimmune disease of the CNS, data indicate that besides CD4⁺ T cells other immune cells actively contribute to its development. Recently it was suggested that monocyte-derived macrophages that infiltrate into the CNS during EAE could directly cause demyelination ²⁶. Interestingly, while CD8⁺ T cells do not contribute to MOG35-55-induced EAE ²⁷, they can passively transfer the disease ²⁸. Finally, studies conducted in B cell-deficient (μMT) or B cell-depleted mice have shown that B cells play a role in the regulation of EAE ^29-31. Other data have found that IL-6 production by B cells plays a crucial role for MOG35-55-inducedEAE progression ³².

Altogether these observations show that, although MOG35-55-induced EAE is considered to be a prototypical model of CD4⁺ T-cell mediated autoimmunity it also requires the contribution of other cell types for disease regulation and development, and that this model has been instrumental in elucidation of various cellular functions in MS and other autoimmune conditions.

2.1.3 Recombinant mouse MOG protein-induced EAE in C57BL/6 mice The interest in developing methods that activate both T cells and B cells in EAE has increased considerably in the recent years with development of successful B cell depleting therapeutic (BCDT) options in MS, including Rituximab ³³, Ofatumumab ³⁴, and Ocrelizumab ³⁵. Weber et al (2010) showed that immunisation of mice with recombinant mouse MOG (rmMOG) protein, but not MOG35-55, led to B cell activation ³¹. When B cells were depleted in

rmMOG-induced EAE mice prior to or after disease onset with anti-CD20 antibodies, disease severity and susceptibility was reduced, indicating that activated B cells played a pathogenic role in this EAE model. Conversely, when B cells were depleted with anti-CD20 in MOG35-55-induced EAE mice prior to or after disease onset, disease course was exacerbated, likely due to the elimination of non-activated B cells with regulatory functions ³¹. Additionally it has been demonstrated that MOG-specific autoantibodies generated in rmMOG-induced EAE are not pathogenic, contrasting with the EAE enhancing pathogenic antibodies produced when animals are immunised with recombinant human MOG proteins ^31,36,37. These studies indicate that immunisation of C57BL/6 mice with rmMOG protein is an appropriate alternative EAE model to the classical MOG35-55-induced EAE model when evaluating the cellular functions of B cells in autoimmune neuroinflammation.

2.1.4 Adoptive transfer EAE model in C57BL6/J mice

EAE induction in C57BL/6 mice by immunisation with recombinant MOG proteins or encephalitogenic MOG peptides emulsified in CFA or other TLR ligands ¹⁵ is referred to as active EAE while passive EAE refers to the adoptive transfer of pre-activated MOG-specific CD4⁺ T cells into naïve mice.

Traditionally, in the adoptive transfer EAE (at-EAE) model, pre-activated MOG-specific CD4⁺ T cells can be isolated from a primed donor mouse, i.e. a mouse immunised with MOG antigen emulsified in CFA to elicit the development of encephalitogenic MOG-specific CD4⁺ T cells ³⁸. Alternatively, at-EAE can be induced by the injection of in vitro differentiated and polarised CD4⁺ T cells isolated from MOG-specific T cell receptor (TCR) transgenic mice (2D2) ³⁹. As pre-activated effector CD4⁺ T cells are readily

encephalogenic upon injection, their adoptive transfer bypasses the need for T cell priming in secondary lymphoid tissues. In at-EAE, adoptively transferred pre-activated MOG-specific effector CD4⁺ T cells migrate directly into the CNS causing formation of inflammatory and demyelination lesions resulting in clinical disease onset ⁹. The EAE disease course seen in the at-EAE model is similar to that of actively-induced EAE, even though the induction phase is bypassed and the requirement for T cell reactivation within the CNS is reduced. The benefit of the passive at-EAE model over the active EAE model is that the at-EAE model enables direct evaluation of the immune effect of specific transgenic modifications during the effector phase of EAE pathogenesis.

2.2 Transforming Growth Factor (TGF)- β s

2.2.1 TGF-β isoforms

TGF-βs are pleiotropic cytokines with multiple biological functions in various organs, both during embryonic development and adult life. Three different TGF-β isoforms (TGF-β1-3) have been identified in mammals. There is 71% sequence homology between TGF-β1 and TGF-β2, and 80% between TGF-β3 and TGF-β1, and TGF-β2 ⁴⁰. TGF-β1 is the predominant isoform and its conserved amino acid sequence homology in many different species is indicative of its importance in evolution. The sequence of human, porcine and simian mature TGF-β1 proteins are identical and rodent (mouse and rat) differ form human by only one amino acid ⁴¹. TGF-βs are secreted as inactive complexes containing the TGF-β, the latent associated polypeptide (LAP), and latent TGF-β binding protein (LTBP). The biological activity of TGF-β is

governed by dissociation from its latent complex. TGF-βs become activated in response to a variety of stimuli including retinoic acid, endotoxins, and pH changes ⁴² or by integrins ⁴³.

2.2.2 TGF-β receptors and signalling pathways

Three types of TGFβ receptors, namely types I, II, and III (TβRI, TβRII, and TβRIII) have been identified. The TβRIIIs (β-glycan and endoglin) function as regulators of ligand access to the signalling receptors. β-glycan interacts with TGF-β1, TGF-β2, and TGF-β3 with similar affinities, while endoglin interacts preferentially with TGF-β1 and TGF-β3 but not efficiently with TGF- β2. TGF-β signalling is initiated through a heteromeric receptor complex composed of TβRI and TβRII. Active TGF-β binds to TβRII, referred to as the binding domain, and recruits TβRI to form the receptor complex. Binding affinity of individual TGF-βs to the different receptors depends on the receptor subunit isoforms. The downstream signalling of the TGF-βs is mainly achieved through the Smad pathway, which stimulates either gene transcription initiation or inhibition. Smad-independent pathways include ERK1/ERK2, PI3K and AKT/PKB transduction cascades which are involved in mitogenicity, survival and growth (Fig. 1) ^42,44.

Figure 1. Active TGF-β binds first to the binding domain on TβRII dimerising with TβRI to initiate downstream signalling cascades. Activated downstream signalling components include Smad2/3, Smad4, MAPK, PI3 kinases.

Adapted from Travis and Sheppard (2014) ⁴³.

2.2.3 Importance of TGF-β isoforms in embryonic development

During mouse embryogenesis TGF-β1 expression is evident from E11 and high levels are observed in the foetal liver and mesenchyme ^45,46. Expression patterns of TGF-β1, TGF-β2 and TGF-β3 during embryonic development are distinct, both temporally and anatomically, suggestive of independent functions ^47,48. In human embryogenesis, the expression patterns of TGF-β1-3 were found to be similar as those observed in the mouse embryo

49.

The importance and non-redundancy of the different TGF-β isoforms during development has been most strikingly established in TGF-β isoform- specific deficient mice. TGF-β1 knock out (KO) mice were found to suffer from

wasting and multifocal inflammatory disease 3-5 weeks after birth leading to premature death. No gross developmental abnormalities were seen prior to wasting although multi-organ T cell infiltration was observed post mortem. The observation that TGF-β1 KO mice died within days after weaning indicate that TGF-β1 could have been provided via maternal milk and that after weaning, when access to the milk was removed, TGF-β1 KO animals could not survive

50,51

. TGF-β2 KO mice die just before, during or just after birth due to severe defects in vital organs such as heart, lungs and CNS ⁵². TGF-β3 KO mice have isolated cleft palates and die within 24 hours of birth ⁵³.

The severe phenotype elicited in the different TGF-β KO mice illustrates the central importance of endogenous TGF-βs in embryogenesis and development. Additionally the striking phenotypical differences seen between TGF-β1-, TGF-β2- and TGF-β3-deficient mice during development demonstrate that the different TGF-β isoforms likely have different in vivo functions and as such should be considered as separate biological entities.

2.2.4 TGF-β1 function in immunity

Data indicate that TGF-β1 plays a major role during embryogenesis

45,46

. TGF-β1 is expressed by multiple cell types and exerts various vital functions, both in the developing and adult tissues, including regulation of cell proliferation, migration, and apoptosis ^54-56. Accumulating evidence also indicates that TGF-β1 plays an integral role in regulating immune responses.

In the following chapter, only the functions of TGF-β1 in the immune system will be discussed.

TGF-β1 impacts the development of both lymphoid tissue DCs and non-lymphoid tissue DCs, including epidermal Langerhans cells (LCs) ⁵⁷.

Additionally, exposing DCs to TGF-β1 has been reported to prevent the maturation of DCs, maintaining them in an immature and tolerogenic phenotype characterised by reduced expression of MHC class II and co- stimulatory molecules when compared to control DCs ⁵⁸. The tolerogenic DC phenotype might in part result from the capacity of TGF-β1 to stimulate IDO expression in DCs ⁵⁹. TGF-β1 was shown to promote conventional DC (cDC) development and block plasmacytoid DC (pDC) development ⁵⁷. TGF-β1 was further found to inhibit the activation of macrophages, central cells of the innate immune system, in response to LPS via the Smad3 pathway ⁶⁰. TGF- β1 was additionally found to inhibit macrophage activation via TLR engagement by inducing degradation of the MyD88 adaptor molecule, resulting in reduced activation and cytokine expression by macrophage in vitro ⁶¹.

Early studies showed that TGF-β1 could limit T cell proliferation ⁶² via inhibition of IL-2-induced proliferation as well as inhibition of other important proliferative transcription factors such as c-myc ⁶³. In the early 2000s it was demonstrated that TGF-β could convert naive CD4⁺CD25⁻ T cells into CD4⁺CD25⁺ anergic/suppressor T cells via induction of Foxp3 expression ⁶⁴. In adoptive transfer experiments, TGF-β-converted CD4⁺CD25⁺ T cells further blocked the activation of antigen-induced naïve CD4⁺ T cells in vivo, consistent with the functional features of previously described Tregs ^65,66. Additionally, TGF-β1-induced Tregs were shown to differ from IL-10-induced Tregs ⁶⁴. Another study aimed at evaluating the frequency of Treg populations in young (8-10 days) TGF-β1 KO mice found that peripheral CD4⁺CD25⁺ Tregs, but not thymic natural CD4⁺CD25⁺ Tregs (nTregs), were reduced in

frequency. These data demonstrated that TGF-β1 maintains Foxp3 expression in peripheral Tregs ⁶⁷. Various Treg subpopulations, which are developmentally, phenotypically or functionally different, including Th3 cells, have been shown to exert their regulatory functions via TGF-β secretion, or via cell-surface expression of LAP ⁶⁸. In particular, CD4⁺CD25⁺LAP⁺ cells have been shown to be potent immunoprotectors in EAE through membrane- bound and secreted TGF-β. In this regard, while CD4⁺ T cells do not express furin, a TGF-β1-converting enzyme, DCs have been suggested to be important for CD4⁺ T cells release of active TGF-β1 ⁶⁹. Bettelli and colleagues (2006) confirmed the importance of TGF-β1 in converting CD4⁺CD25^-FoxP3^- T cells into CD4⁺CD25⁺FoxP3⁺ Tregs and further demonstrated that the presence of the proinflammatory cytokines IL-6 and TGF-β1 induces the differentiation of pathogenic IL-17-producing Th17 cells rather than FoxP3- expressing Tregs from naïve T cells, indicating a reciprocal developmental pathway for the generation of Th17 and Tregs that may depend on the state of innate immune system ⁷⁰. Importantly, additional studies later showed that when activated myelin-reactive Th17 cells, differentiated with a combination of IL-6 and TGF-β1 or with a combination of IL-6, IL-23 and TGF-β1, were adoptively transferred into naïve host, the recipient mice remained healthy ⁷¹. In contrast, T cells differentiated with IL-23 alone were highly effective at inducing EAE after adoptive transfer ⁷². Further studies have demonstrated that in vitro differentiation of Th17 cells with IL-23 in the absence of TGF-β1 allowed the generation of highly encephalitogenic T cells. Interestingly, Th17 cells differentiated with IL-23 were later shown to express GM-CSF, a

cytokine that has an essential role in the encephalitogenic potential of Th17 cells ⁷³.

Th2 cells, another set of T helper cells, have been principally implicated in allergic responses. Th2 cells mediate their function by producing various cytokines such as IL-4, IL-5 and IL-13. The generation of these T cells is under the control of the GATA-3 transcription factor ⁷⁴. Interestingly, TGF- β1 has been shown to re-programme the differentiation of Th2 cells by erasing their characteristic profile and switching them towards producing IL-9 or, in combination with IL-4, to operate the differentiation of Th9 cells directly

75. Th1 differentiation is also affected by TGF-β1 as data indicate that TGF-β1 inhibits the expression of T-bet, the main Th1 transcription factor, by naïve T cells and the subsequent Th1 development ⁷⁶. The importance of TGF-β1 in regulating immune responses was further underlined by Li et al (2007) with the development of transgenic mice with a T cell-specific deletion of the TGF- β1 gene ⁷⁷. While these transgenic mice developed normally until around 4 months of age, they developed lethal immunopathology in multiple organs associated with enhanced T cell proliferation and activation, recapitulating what was seen in TGF-β1 KO mice ⁵⁰. Additionally, T cell-specific TGF-β1- deficient mice showed increased frequencies of nTregs, but not induced Tregs, in peripheral lymphoid organs ⁷⁷.

Using transgenic mice lacking TβRII selectively in B cells, rendering them unable to respond to all TGF-βs, it has been shown that the absence of TβRII in B cells resulted in decreased life span of conventional splenic B cells, expansion of peritoneal B1 cells, B cell hyperplasia in Peyer's patches and elevated serum immunoglobulin, and substantial IgG3 responses to a

normally weak immunogen, indicating that TβRII signalling is needed to control B cell hyper-responsiveness ⁷⁸. Using chimeras in which B cells, but not T cells, lacked TGF-β1 expression, B cell-produced TGF-β1 has been shown to play a major role in controlling B cell biology ⁷⁹. In particular, absence of B cell-derived TGF-β1 in mice was associated with decrease and increase of serum IgA and IgG1 levels, respectively, as well as boosted IgG3 production.

Altogether, these observations clearly indicate that TGF-β1 is central in the maintenance and stability of immune responses, especially as an essential immune regulatory factor. The multiple roles played by TGF-β1, as reported in numerous studies, stress that more attention should be devoted to the specific characterisation of the regulatory functions of this cytokine in immune disease models.

2.2.5 Role of TGF-β1 in EAE

Early studies showed that exogenous administration of TGF-β1 in EAE delayed disease induction and increased remission ^80,81. Likewise, EAE mice treated with neutralising anti-TGF-β1 antibodies were found to have a more severe disease course with higher incidence compared to control mice ⁸². Altogether, these findings established a central role for TGF-β1 in EAE regulation, further consolidated with the use of specific transgenic mice deficient for TGF-β1 or TβRII. Using a transgenic mouse model of targeted functional inactivation of TβRII signalling in CD11c⁺ cells (CD11c^dnR) mice), TGF-β was shown to limit severity of MOG35-55-induced EAE and Th1/Th17 responses via actions on DCs ⁸³. Moreover, CD11c^dnR mice crossed with 2D2 MOG-specific TCR transgenic mice developed spontaneous EAE-like

disease. Although these findings did not establish the identity of the TGF-β isoform implicated in these effects, these data nevertheless stress the role of TGF-βs in the regulation of CNS autoimmunity.

Interestingly, data indicate that mice in which CD4⁺ T cells cannot produce TGF-β1 are resistant to EAE ⁷⁷, recapitulating the results obtained by mice lacking TβRII expression by CD4⁺ T cells ⁸⁴. Mice with a T cell-specific deletion of TGF-β1 or mice whose T cells cannot respond to TGF-β signalling were both shown to lack Th17 cells, while no effects were seen on Th1 cells.

Remarkably, Bettelli et al (2006) showed that the adoptive transfer of MOG autoreactive CD4⁺ T cells, which were obtained from transgenic 2D2 MOG- specific TCR mice genetically modified in a way that TGF-β1 production was placed under the IL-2 receptor (2D2 x TgTGF-β mice) to upregulate TGF-β1 production, suppressed ongoing MOG35-55-induced EAE in recipient mice.

Conversely, immunisation of 2D2 x TgTGF-β mice with MOG35-55 emulsified in CFA resulted in a more severe disease than normal 2D2 mice immunised in the same way. These results suggest that activation of T cells in a highly inflammatory environment, characterised by high levels of IL-6, such as the one required for development of active EAE, and in the presence of elevated levels of TGF-β1, predominantly drives robust Th17 polarisation over Treg differentiation ⁷⁰.

2.3 B cells

2.3.1 Multifactorial B cells in autoimmune neuroinflammation

B cells have multiple functions during immune responses as they can serve as antigen-specific antibody secreting cells and antigen presenting

cells, as well as cytokine producing effector cells and finally as important players in the formation of tertiary lymphoid structures (Fig. 2).

Figure 2: B cells have multiple functions including antibody production, antigen presentation to T cells and immune modulation via the release of pro- or anti-inflammatory cytokines. Adapted from Li et al (2015) ⁸⁵.

Plasmablasts and plasma cells, the “professional” antibody secreting B cells, mediate humoral immune responses. While conventional B cells do not generate antibodies unless activated, antibody-forming cells can be readily detected prior to encountering pathogens. Production of antibodies against self-antigens are frequently observed in autoimmune diseases such as MS ⁸⁶ and may in some cases initiate or exacerbate disease pathology, such as in neuromylitis optica ^87,88. Besides their role in antibody production, B cells can be competent APCs for priming CD4⁺ T cells ^36,89. In contrast to resting

antigen-presenting B cells that promote T cell tolerance, activated B cells are effective accessory cells for T cell responses ^90,91. Finally, B cells can produce pro- and anti-inflammatory cytokines under innate or adaptive stimulation.

Cytokine production by B cells can regulate different important immune mechanisms, such as remodelling of immune tissue, differentiation of T cells or recruitment of DCs ⁹².

Among other proinflammatory mediators produced by B cells, the production of IL-6, a cytokine necessary for the development of EAE ³², by effector B cells was shown to be important for the development of EAE ⁹³. Barr et al (2012) elegantly demonstrated that B cells from MOG35-55-induced EAE mice produced elevated IL-6 levels compared with B cells from naïve mice ³². Using mixed bone marrow chimera mice, they further demonstrated that mice with a B cell-specific IL-6 deficiency showed less severe disease than mice with WT B cells. Finally, B cell depletion ameliorated EAE only in mice containing IL-6-producing B cells. Their findings further showed that IL-6 production by B cells was essential to the propagation of Th17 cells to increase disease severity. Of note, although eliminating IL-6 production by B cells ameliorated the disease course it did not prevent EAE development, indicating that although IL-6-producing B cells are pathogenic in EAE, they are non-essential for disease development ³². Recently, proinflammatory GM- CSF-producing B cells have also been shown to be associated with MS activity ⁹⁴.

2.3.2 Regulatory B cells

Accumulating evidence indicate that B cells exert regulatory functions during EAE. Seminal EAE studies by Wolf et al (1996) ⁹⁵ have shown that,

unlike WT mice, B cell-deficient (μMT) mice did not recover from MOG35-55- induced EAE. These data were the first to suggest a role for B cells in the modulation of EAE through immune deviation from Th1 to Th2 cytokines. It is interesting to note that the term ‘regulatory B cells (or Bregs)', which defines B cell subpopulations with regulatory properties, was only introduced for the first time 10 years later by Mizoguchi and Bhan ⁹⁶. In their review article the authors summarised a significant number of studies published over a period of 10 years that reported a regulatory role of B cells, mainly associated with the contribution of the regulatory cytokines IL-10 and TGF-β. More recently, IL-35 produced by B cells was also shown to function as central regulatory cytokine in EAE and immunity ⁹⁷.

Although the production of anti-inflammatory and immunosuppressive cytokines are often central in the regulatory functions of B cells, other mechanisms including IgM, programmed death-ligand 1 (PD-L1) and glucocorticoid-induced TNF receptor family-related protein (GITR) have also been documented to be support regulatory Breg functions ^98,99. The following chapters will specifically focus on cytokine-producing regulatory B cells.

2.3.3 Regulatory IL-10-producing B cells

Early studies from the mid 90s showed that activated B cells could secrete IL-10. In particular, B cell lymphomas and LPS-activated primary murine B cells were found to express multiple cytokines, including IL-6 and TNF as well as IL-10 ¹⁰⁰. With the demonstration of the potent protective function of IL-10 in CNS autoimmunity ¹⁰¹ the relative contribution of different immune cellular subsets producing IL-10 became of increasing interest.

With the development of mixed bone marrow chimera mice that specifically lacked IL-10 production by B cells research into the suppressive functions of B cells gained momentum. Fillatreau and colleagues (2002) illustrated that the lack of EAE remission seen in B cell-deficient μMT mice was recapitulated in IL10^-/- B cell chimera mice ²⁹. They further showed that adoptive transfer of IL-10-producing B cells isolated from WT mice recovering from EAE could rescue recovery of EAE in IL-10^-/- B cell chimera mice. Finally, the authors demonstrated that CD40^-/- B cell chimera showed the same lack of remission from EAE as seen in IL-10^-/- B cell mice, indicating that activation of B cells via CD40 was a requisite for the production of IL-10 by B cells. While these data taken together have demonstrated that IL-10 production by B cells has regulatory functions in EAE, they did not established whether all B cells, or a discrete B cell subset, were the direct targets of CD40-induced IL-10 production.

In 2002 it was demonstrated that chronic intestinal inflammation generates a subset of mesenteric lymph node (mLN) B cells producing IL-10 characterised by elevated expression of CD1d and that their transfer suppressed chronic intestinal inflammation via the downregulation of IL-1β production and STAT3 activation ¹⁰². Later studies established that in addition to CD1d^hi mLN B cells, phenotypically distinct B cell subsets could exert regulatory functions in various autoimmune disease models ¹⁰³. In particular, a subset of splenic CD19⁺CD1d^hiCD5⁺ IL-10 competent B cell, dubbed B10 cells, have been shown to negatively regulate inflammation and autoimmunity in diverse mouse models of disease ¹⁰⁴. While B10 cells are numerically rare

in both normal and autoimmune mice, this subset could make up around 10- 15% of the CD19⁺CD1d^hiCD5⁺ B cells following in vitro stimulation ¹⁰³.

In one study, aimed at evaluating the effects of the B cell-depleting anti-CD20 antibody Rituximab, a potent therapeutic option in MS ³³, in MOG35- 55-induced EAE it was demonstrated that the timing of B cell depletion is crucial in influencing the EAE disease course ³⁰. This study reported that when B cells were depleted prior (day -7) to EAE induction disease was exacerbated while it was ameliorated when B cells were depleted after onset (day 14). Increased EAE severity was associated with enhanced CNS inflammation and demyelination as well as augmented frequencies and responses of Th1 and Th17 cells. Opposite findings were found in mice with reduced EAE disease severity. In an elegant series of experiments the researchers showed that B10 cells were responsible for the outcomes of B cell depletion on disease progression. In brief, splenic CD1d^hiCD5⁺ B10 cells isolated from CD20^-/- mice (resistant to anti-CD20 treatment) were adoptively transferred into WT mice that were depleted of B cells prior to EAE induction.

As controls, WT mice received non-B10 B cells (non-CD1d^hiCD5⁺) purified from CD20^-/- mice. The results showed that adoptive transfer of B10 cells, but not non-B10 cells, normalised the EAE course in B cell-depleted mice.

Remarkably, adoptively transferred CD1d^hiCD5⁺ B cells isolated from IL-10- and CD20-deficient mice did not normalised EAE severity, further demonstrating that B10 cells negatively regulated EAE initiation via IL-10 production ³⁰. In another study from the same group using a similar study design as above it was found that preferential B10 cell depletion with anti- CD22 mAb treatment prior to EAE induction by active immunisation with

MOG35-55 exacerbated disease severity, while their depletion during the acute phase had not effect. Remarkably, immunisation of mice with MOG35-55 led to the rapid expansion of B10 cells within the spleen, but not CNS, which paralleled B10 cell regulation of disease initiation. As expected, adoptively transferred MOG35-55-sensitised B10 cells reduced EAE initiation, while they did not suppress ongoing EAE disease. Interestingly, depletion of Tregs during disease progression, but not initiation, decreased EAE severity. In contrast to B10 cells, Treg were found to expand within the CNS during disease progression, which paralleled their regulatory role of late-phase disease. These findings showed that while B10 cells are important regulators during EAE initiation, Tregs inhibit late-phase disease. Mechanistically, it was showed that B10 cells in MOG35-55-induced EAE were limiting T cell proliferation by restraining the APC functions of DCs ¹⁰⁵. Interestingly, another study from the same group further showed that inhibition of MOG35-55-induced EAE initiation by B10 cells require their maturation into functional IL-10- producer effector cells by IL-21 and CD40-dependent cognate interactions with T cells,¹⁰⁶.

While accumulating evidence indicate that splenic B10 cells play a major role in regulating EAE immunopathogenesis, other data point to a predominant role of plasmablasts as IL-10-producing regulatory B cells during CNS autoimmunity ¹⁰⁷. Plasmablasts are CD138⁺ cells that develop from activated B cells and produce IL-10. By using Venus reporter mice to track IL- 10 expression by B cells, it was shown that plasma blasts in the draining lymph nodes (dLNs), but not in the spleen, predominantly produce IL-10 during MOG35-55-induced EAE and that these plasma cells were generated

only during EAE inflammation. Moreover, mice that lack plasma cells, due to a B cell-specific deletion of Blimp1 or IRF1, showed an exacerbated EAE disease course and failed to recover from disease, paralleling the effects observed in mice that specifically lacked IL-10 production in B cells ²⁹. The increased disease severity observed in Blimp1-deficient mice was characterised by augmented numbers of Th1/17 cells within the CNS as well as increased Th1/17 responses in the dLNs. Mechanistically, the regulatory effects of IL-10-producing plasmablasts on EAE inflammation was mediated, at least in part, via inhibition of DC functions leading to reduction in autoreactive T cell differentiation ¹⁰⁷.

Altogether, these studies show that IL-10 production by B cells plays a major regulatory role in CNS autoimmunity and that various subpopulations of IL-10-expressing B cells with regulatory functions are likely controlling autoimmune neuroinflammation.

2.3.4 Regulatory IL-35-producing B cells

In contrast to the well documented importance of regulatory IL-10- producing B cells, B cell-derived IL-35 production was only recently reported to play a central role in controlling immune responses ⁹⁷. Data indicate that B cell co-stimulation with TLR4 and CD40 did not only induce IL-10 production but also IL-35, an established immunomodulatory cytokine ¹⁰⁸. Mice that specifically lacked IL-35 production in B cells lost their ability to recover from

MOG35-55-induced EAE and showed increased MOG-reactive Th1/17

responses compared to control mice. Remarkably, higher expression levels of activation markers (CD44, CD69) by B cells were also found, suggestive of IL- 35 as being a regulator of the APC capacity of B cells. Accordingly, B cells

from mice in which only B cells did not express IL-35 were shown to be more potent APCs than control B cells, stimulating higher proliferation and differentiation of MOG-specific Th17 cells. Of specific importance, while CD138⁺ plasma cells were the major source of B cell-derived IL-35 and IL-10 during MOG35-55-induced EAE, distinct sets of plasma cells expressed IL-35 and IL-10 ⁹⁷. Altogether, these data indicate that not only IL-10- but also IL-35- producing plasma cells play an important regulatory role during the resolution phase of EAE.

2.3.5 TGF-β1 expression by B cells

TGF-β1 production by B cells has been suggested to exert regulatory functions in various experimental settings. In one study evaluating the interaction between activated B cells and CD8⁺ T cells it was reported that B cells activated in a “T cell-dependent manner” with anti-CD40 and IgG induced significantly higher levels of proliferation, cytokine secretion, and cytotoxic ability of CD8⁺ T cells compared to when B cells were activated with LPS in a “T cell-independent manner” i.e. suggesting that LPS-activated B cells induced CD8⁺ T cell hypo-responsiveness ¹⁰⁹. Remarkably, this study showed that LPS-activated B cells had higher cell-surface TGF-β1 expression and secreted more TGF-β1 than B cells stimulated with anti-CD40 and IgG and that hypo-responsiveness of CD8⁺ T cells, stimulated with LPS, could be rescued by neutralising TGF-β1.

TGF-β1 production by B cells has also been studied in multiple experimental animal models, including in an ovalbumin (OVA)-induced model of allergic airway disease (AAD) ¹¹⁰. AAD is induced in mice initially sensitised to OVA by acute and repeated respiratory exposure to OVA. Continuous OVA

exposure for up to 42 days results in the resolution of the pulmonary AAD response, referred to as local inhalation tolerance (LIT). Data indicate that accumulation of Tregs in the dLNs of the lungs correlates with the debut of LIT, supporting a central function for Treg cells in AAD resolution ¹¹¹. One study showed that adoptive transfer of LIT B cells from hilar LNs (hLNs) prior to AAD induction in naïve mice attenuated lung inflammation and that LIT B cells were conferring protection in an antigen specific manner. Interestingly, LIT hLN B cells were found to produce TGF-β1 and to convert naive CD4⁺ T cells into Tregs ¹¹⁰. It was further illustrated in AAD that LIT hLN B cell reside in a CD5⁺ TGF-β1-producing subpopulation and co-localise with Tregs ¹¹². In addition to contributing to the development of tolerance in AAD, a Th2-driven inflammatory reaction against harmless inhaled antigens, other data have shown that B cells expressing TGF-β1 could also effectively down-regulate Th1 immunity to β cell autoantigens and inhibit diabetes progression in prediabetic NOD mice ¹¹³. Mechanistically, the regulatory actions of B cells in prediabetic NOD mice appeared to predominantly take place in the peripheral lymph tissues rather than in the pancreatic islets, and to be mediated through the down-regulation of Ag-presenting activity of APCs ¹¹³. Together, these studies suggest that several subpopulations of TGF-β1-producing Bregs may exist, and that their development may depend on the nature of the stimuli and the anatomical sites of Ag presentation.

It is also worth noting that numerous in vitro assays or experimental models utilising adoptively transferred have further established that B cell subpopulations expressing TGF-β (with no indication of the actual isoform) can control Treg induction, immune tolerance promotion, and/or innate and

adaptive immune response suppression 110,114-123

. Although these studies did not determine which TGF-β isoform(s) were involved, they altogether support a role for TGF-β in the regulatory capacity of B cells. Despite these interesting observations reported over the course of the last 15 years, no studies to date have employed transgenic models that specifically eliminate isoform- specific TGF-β production by B cells, a crucial avenue for specifically delineating the importance of TGF-β-producing B cells in autoimmunity or immunity in vivo.

2.3.6 Human Bregs

Evidence indicates that human B cells can also produce the regulatory cytokines IL-10 and TGF-β. Similar to IL-10-producing B cells, TGFβ- producing CD19⁺CD5⁺ B cells have been shown to regulate cow’s milk allergic responses in human subjects ^118,124. Stimulation of human B cells with combined CpG, a TLR9 ligand, and BCR engagement has been shown to stimulate IL-10 production by B cells ¹²⁵. While IL-10-producing B cells did not belong to a defined subpopulation, these cells were found to be enriched in the memory (CD27⁺) and the transitional (CD38^hi) B cell compartments and to inhibit proliferation of CD4⁺ T cells. Other data have shown that CD25⁺CD27⁺CD86⁺CD1d⁺IL-10⁺TGF-β⁺ Bregs could directly reduce the proliferation of autologous stimulated CD4⁺ T cells ¹¹⁵. These cells were also shown to increase FoxP3 and CTLA-4 expression in co-culture with Tregs, via a mechanism that was dependent on a direct contact between Bregs and Tregs as well as TGF-β, but not IL-10. Other data have reported that activated B cells could also regulate CD4⁺ T cell proliferation through production of TGF-β and IDO. Interestingly, CTLA-4 was shown to induce B cells to produce IDO and to become effective Bregs ¹²⁶.

Data indicate that B cells isolated from MS patients produce less IL-10 in response to stimulation when compared to B cells purified from healthy controls ¹²⁷. Remarkably, this study found that stimulation of CD27⁻ naïve B cells, but not CD27⁺ memory B cells, by CD40 engagement stimulated IL-10 production. Naïve or memory B cells stimulate in this way did not produce TNF. In contrast, memory, but not naïve, B cells stimulated sequentially by BCR engagement followed by CD40-mediated signalling, produced TNF, but not IL-10, suggesting that the reciprocal regulation of human B cell effector cytokines is context dependent. A different study showed that the frequencies of circulating B10 cells in patients with lupus, rheumatoid arthritis, Sjögren syndrome, autoimmune skin disease, and MS were not significantly different from those observed in healthy controls ¹²⁸.

Altogether, these studies indicate that human B cells have the capacity to produce regulatory cytokines and to exert regulatory functions in vitro and that their defect or deficiency could contribute to the development of excessive immune responses.

3 Results

3.1 Chapter I

B cell-derived transforming growth factor- β1 expression limits the induction phase of autoimmune neuroinflammation

Kristbjörg Bjarnadóttir, Mahdia Benkhoucha, Doron Merkler Martin S.

Weber, Natalie L. Payne, Claude C.A. Bernard, Nicolas Molnarfi and Patrice H. Lalive

Published in Scientific Reports. 2016 October 6: 34594

Introduction: Regulatory functions of B cells are of increasing interest with the success of B cell depleting therapies in autoimmune diseases such as MS

33-35

. By using a mouse strain in which B cells cannot produce TGF-β1 we showed that TGF-β1 production by B cells has a regulatory role in the initiation phase of EAE.

Objectives: Evaluating the regulatory role of TGF-β1 production by B cells in EAE, an animal model of MS and a prototypic T cell-mediated, organ specific autoimmune disease model.

My personal contribution as a collaborator to this manuscript: I performed the experiments and analysed the data. The research design and paper writing were principally done by myself and Dr Nicolas Molnarfi with conceptual input from co-authors.