The Role of CD1d-Mediated Lipid Presentation by Regulatory B Cells

in Invariant Natural Killer T Cell Suppression of Autoimmunity

by Kristine Oleinika

Supervisor

Professor Claudia Mauri UCL

Immunology

I, Kristine Oleinika, confirm that the work presented in this thesis is my own. Where information has been derived from other sources, I confirm that this has been indicated in the thesis.

Kristine Oleinika

Acknowledgements

To Professor Claudia Mauri, my supervisor and mentor

Throughout my PhD years I have so admired your energy and passion, which has made me fall in love with science over and over again. Thank you for looking after me, your guidance has been invaluable in my journey in both science and life. I am incredibly grateful for being part of the wonderful team that you have gathered around you, scientific debate that you have fostered in the group, and for encouraging the friendships within it.

To all my colleagues

You have made this such an adventure, sharing my enthusiasm for science, politics and footwear, with you I have always felt like a member of an incredibly, almost insultingly, over-educated and over-loving family. And I apologise for the many times you have had to endure listening to my idea for “PhD: the musical” and late night immunology aphorisms, or being on the receiving end of the impromptu enforced hugs.

Lizzy, Diego and Paul – thank you for making me think about immunology, socialism, feminism, literature and art (..) ad infinitum. And for your helpful comments on the thesis. Madhvi, overachieving mama bear, thanks for being a fantastic listener. Diana, my Romanian sister, thank you for all your help and always being so overwhelmingly positive. Marsilio, thank you for the advice, particularly when walking together to King’s X after long hours in the lab; and for providing so much positivity with statements about the “package” or stories about your boys giving you dad achievement stickers. Hannah, you discussing Cinnamon and his birthday cakes during writing kept me very happy. Scott, you have provided much banter and super energy. Laura, the lab cheerleader, thank you for always humming along and all valuable advice that you provided. Ale, thanks for keeping me company with conversations about signalling at 2am doing experiments. Professor Isenberg, your music and all our warm interactions have made the fourth floor more sunny, and time with you in the clinic re- iterated why it is important to do science. And particularly Chris(-py) – I could not have been able to survive, if not for your level-headedness, when I have been losing my mind, for the countless occasions you have saved me, all the fun we have had together and I want to simply thank you for being such a great person.

To friends

My Triumfeminate of SuperFemales – Inga, Ksenia, Marta (in alphabetical order), you have been my safe harbour and ideas-incubator, and I colossally appreciate just how badass you are. Mahdi, my little panda, you have always been there for me, with solutions to my science problems, when I am experiencing “technical difficulties”. I can only aspire to be as knowledgeable, hard working, and overwhelmingly kind.

Dion, my most inspiring and supportive partner, you have loved me, entertained my eccentric needs, and taught me about social interactions during these years. You have been the sunshine to many of my days, and thank you for mirroring my uncontainable excitement about experiments done in 1950s-1970s, watching all those documentaries with me, and taking me to lectures, museums and abroad. You are the best human I have ever met and I have so much fun with you.

The Family

Grandmothers and grandfather, Elžuk! Mani labie, mīļie, Jums visvairāk veltīts šis darbs, paldies par gēniem, grāmatām, padomu un atbalstu it visā.

And finally, my incredible superhuman mother, the original feminist, you have been the greatest role model, my inspiration in hard work, perseverance, love for others, pursuit of knowledge and from time to time enjoying the fruits of ones labour. You have the biggest heart and I love you more than you will ever know.

Abstract

Regulatory B cells (Bregs) express high levels of CD1d that presents lipid antigens to invariant natural killer T (iNKT) cells. The role of CD1d in Breg biology and the specific contribution to iNKT cell function and in suppressing inflammation remains unknown. Combining chimeric mice, cell depletion and adoptive transfer strategies, we show that CD1d lipid presentation by B cells to iNKT cells is critical for the induction of iNKT cells that down-regulate Th1 and Th17 adaptive immune responses and arthritis, whilst dispensable for Breg development. Mice lacking CD1d- expressing B cells developed exacerbated arthritis compared to wild-type mice and failed to respond to α-GalCer treatment. Absence of lipid presentation by B cells led to altered activation of iNKT cells, with disruption of regulatory pathways including those involved in and cytokine responses. Thus, we have identified a novel mechanism by which Bregs via CD1d, in an IL-10 independent manner, control and restrain excessive inflammation.

Table of contents

Table of figures ...... 9

List of abbreviations ...... 12

Chapter I: Introduction ...... 16 B cells ...... 16 B cell ontogeny in the bone marrow and central tolerance ...... 16 Generation of B cell antigen receptors ...... 17 B cell tolerance checkpoints ...... 18 Microenvironmental cues in B cell development ...... 19 A role for the BCR during B cell development, maturation and survival ...... 20 B cell heterogeneity and functional maturation in the periphery ...... 23 B-1 cells ...... 23 B-2 cells ...... 23 Characteristics of mature B cell subsets ...... 25 MZ versus Fo B cell fate decision ...... 28 Generation of humoral responses within GCs ...... 30 Molecules important in GC reaction ...... 31 B cell roles in addition to antibody production ...... 34 Regulatory B cells ...... 36 Historical perspective ...... 36 Breg Phenotype ...... 40 T2-MZP B cells ...... 40 B10 cells ...... 43 B-1 cells ...... 45 Other Breg phenotypes ...... 45 IL-10-independent mechanisms of Breg function ...... 48 Signals in Breg induction ...... 52 CD40 ...... 52 TLRs and inflammation ...... 53 BCR and antigen-specificity ...... 55 Bregs in humans ...... 57 iNKT cells ...... 59 iNKT cell ontogeny ...... 60 iNKT cell ligands ...... 63 α-GalCer and synthetic analogues ...... 63 Exogenous ligands ...... 64 Endogenous ligands ...... 65 CD1d and lipid presentation ...... 69 6 iNKT cell activation ...... 70 Role of different APCs in iNKT cell activation ...... 73 iNKT cell help for B cell antibody production ...... 74 iNKT cells in the regulation of inflammatory and autoimmune disease ...... 78 Type I diabetes mellitus (TIDM) ...... 78 RA ...... 79 SLE ...... 80 EAE ...... 81 Specialized iNKT cell subsets in immunomodulation ...... 81 Antigen-induced arthritis ...... 82

Chapter II: Materials and Methods ...... 84 Mice ...... 84 Induction of antigen-induced arthritis (AIA) and administration of α-GalCer ...... 84 B cell depletion ...... 84 CD11c+ cell depletion ...... 84 MZ B cell depletion ...... 85 In vivo IFN-γ neutralization ...... 85 Cell preparation ...... 85 Generation of mixed BM chimeric mice ...... 85 Adoptive transfer ...... 86 Flow cytometry and cell sorting ...... 86 Immunofluorescence ...... 86 Visualising CD1d and lipid raft co-localisation by ImageStream ...... 87 In vitro B cell culture ...... 87 In vitro iNKT cell suppression assay ...... 87 In vitro B cell suppression assay ...... 87 In vitro B cell and dendritic cell depletion ...... 88 RNA sequencing ...... 88 Bioinformatics analysis of RNA-seq data ...... 88 Statistical analysis ...... 88

Chapter III: Results I ...... 92 B cells mediate α-GalCer-iNKT cell dependent amelioration of arthritis ...... 93 B cell depletion abrogates the protective effect of α-GalCer on arthritis ...... 97 DCs are not required for α-GalCer-iNKT cell mediated suppression of arthritis 100 B cells and DCs play different roles in iNKT cell activation in vitro ...... 103

Chapter IV: Results II ...... 106

7 B cells induce iNKT cells that limit inflammation via a CD1d-dependent mechanism ...... 107 α-GalCer treatment leads to disruption of GCs in arthritic mice in a B cell CD1d- dependent manner ...... 113 B cell subset frequencies are not affected by α-GalCer treatment in B cell CD1d- sufficient mice ...... 117 IL-10 is dispensable for α-GalCer-mediated amelioration of arthritis ...... 119 iNKT cells, and not B cells, are the effectors of suppression in arthritis ...... 121

CHAPTER V: Results III ...... 124 iNKT cell transcriptional profile is altered in the absence of CD1d+ B cells ...... 125 B cell inability to present α-GalCer affects iNKT cell expression of involved in cytokine responses and metabolism ...... 127 B cell CD1d is involved in the induction of genes associated specifically with NKT1 phenotype and PLZF transcriptional control ...... 129 α-GalCer induced IFN-γ inhibits Th1 and Th17 cells in arthritis ...... 132

CHAPTER VI: Results IV ...... 138 T2-MZP Bregs via CD1d induce the differentiation of iNKT cells that limit inflammation ...... 139 α-GalCer mediated suppression does not require MZ B cells ...... 144

CHAPTER VII: Results V ...... 148 iNKT cells are dispensable for Breg development ...... 149

Chapter VIII: Discussion ...... 155

Bibliography ...... 164

8 Table of figures

Figure 1.1. B cell development in the BM…………………………………….…….....22 Figure 1.2. B cell development in the periphery…………………………………….....27 Figure 1.3. Signals in B cell maturation………………………………………………..29 Figure 1.4. Breg mechanisms…………………………………………………………..51 Figure 1.5. Synthetic iNKT cell ligands…………...………………………………...…68 Figure 1.6. iNKT cell activation modes.……...………………………….…………….72 Figure 1.7. iNKT cells provide cognate and non-cognate help for B cell antibody responses………………………………………………………………………………..77 Figure 2.1. Gating strategy and purity for iNKT cell, Fo, MZ and T2-MZP B cell transfer experiments……………………………………………………………………89 Figure 3.1. α-GalCer-iNKT cell-dependent amelioration of arthritis requires B cells………………………………………………………………………...…...………94 Figure 3.2. α-GalCer-driven iNKT cell expansion occurs in the , inguinal and mesenteric lymph nodes………………………………………………………………..95 Figure 3.3. Splenic, but not hepatic iNKT cells display reduced PLZF and IFN-γ induction in µMT mice following α-GalCer treatment……………………...…………96 Figure 3.4 B cell depletion abrogates the protective effect of α-GalCer on arthritis…..98 Figure 3.5. B cell depletion results in impaired induction of iNKT cell PLZF and IFN-γ in response to α-GalCer………………………………………………………………...99 Figure 3.6. Dendritic cells are dispensable in α-GalCer-mediated amelioration of arthritis……………………………………………..………………………………….101 Figure 3.7. CD11c+ DC depletion does not impair iNKT cell upregulation of PLZF and IFN-γ in response to α-GalCer…………………………………………………..…....102 Figure 3.8. B cells and dendritic cells play different roles in α-GalCer-iNKT cell activation in vitro………………………………………………………..…………….104 Figure 3.9. CD19+ B cell subsets display higher CD1d association with lipid rafts than CD11c+ DCs…………………………………………………………………...... 105 Figure 4.1. Chimeric mice lacking CD1d specifically on B cells develop exacerbated arthritis……………………………………………………….…………….………….109 Figure 4.2. B cell specific CD1d is critical for α-GalCer-iNKT cell-amelioration of arthritis……………………………………………………….…………….…..……...110 Figure 4.3. B cell specific expression of CD1d is required for iNKT cell upregulation of PLZF and IFN-γ in response to α-GalCer………………………………………….....111 Figure 4.4. α-GalCer induces reciprocal T-bet and GATA3 expression in iNKT cells from B-Cd1d-/- and B-WT mice………………………………………………..……..112 Figure 4.5. GC formation, Tfh and NKTfh cells are inhibited by α-GalCer treatment………………………………………..……………………………………..114 Figure 4.6. iNKT cells localise in the T cell area in the spleens of arthritic B-WT mice treated with α-GalCer……………………………..…….………………….…………116 Figure 4.7. Effect of α-GalCer treatment on splenic B cell populations ………..…....118 9 Figure 4.8. α-GalCer-mediated suppression of arthritis is independent of IL-10….....120 Figure 4.9. Anti-CD40-stimulated B cells do not suppress arthritis in the acute phase upon adoptive transfer to iNKT cell deficient mice.……………………………….....122 Figure 4.10. iNKT cells from donor mice treated with α-GalCer ameliorate arthritis upon transfer to B cell deficient and WT mice…………………………………..……123 Figure 5.1. Expression of immune-related genes in mice lacking CD1d specifically on B cells following α-GalCer treatment……………………………………….…………..126 Figure 5.2. B cell inability to present α-GalCer affects iNKT cell expression of genes involved in cytokine responses and metabolism……………………………...………128 Figure 5.3. Comparison of splenic iNKT cell expression to NKT1, NKT2 and NKT17 published data sets………………………………………………..…………..130 Figure 5.4. B cell CD1d controls the induction of transcriptional programs associated with NKT1 cells and PLZF regulation………………………………………………..131 Figure 5.5. α-GalCer amelioration of arthritis is partially dependent on IFN-γ production by iNKT cells……………………………………………………………..134 Figure 5.6. B cells are required for α-GalCer-driven and IFN-γ-mediated suppression of T-bet and ROR-γt expression in CD4+ T cells………………………………………..135 Figure 5.7. Following α-GalCer treatment in vivo iNKT cells from WT, but not B cell deficient mice, are suppressive in vitro………………………………………..…..….136 Figure 5.8. Exogenous IFN-γ regulates expression of T-bet and ROR-γt in CD4+ T cells…………………………………………………………………………………....137 Figure 6.1. T2-MZP B cells express high levels of CD1d and suppress arthritis upon adoptive transfer………………………………………………………………………140 Figure 6.2. T2-MZP Bregs mediate suppression via iNKT cells……………………..141 Figure 6.3. T2-MZP Bregs from CD1d-sufficient Jα18-/- mice suppress arthritis upon adoptive transfer…………………………………………………………………..…..142 Figure 6.4. T2-MZP Bregs from CD1d-deficient mice fail to suppress arthritis upon adoptive transfer……………………………………………………………..………..143 Figure 6.5. Anti-CD11a and anti-CD49d antibody treatment depletes MZ, but not T2- MZP B cells…………………………………………………………………………...145 Figure 6.6. MZ B cells are dispensable for α-GalCer-mediated amelioration of arthritis………………………………………………………………………………...146 Figure 6.7. MZ B cells are dispensable for α-GalCer-mediated induction of PLZF and IFN-γ…………………………………………………………………………..………147 Figure 7.1. α-GalCer does not induce B cell IL-10…………….……………………..150 Figure 7.2. Jα18-/- mice develop exacerbated arthritis associated with elevated Th1 and Th17 responses………………………………………………………………………..151 Figure 7.3. Cd1d-/- mice develop exacerbated arthritis associated with elevated Th1 and Th17 responses………………………………………………………………………..152 Figure 7.4. Breg development is not impaired in iNKT cell deficient mice……….…153 Figure 7.5. Breg in vitro suppressive capacity is not impaired in iNKT cell deficient mice………………………………………………………………...…………………154

10 Figure 8.1 Proposed mechanism of Breg modulation of inflammation in a CD1d- and iNKT cell-dependent manner…………………………………...………………….…163

11 List of abbreviations

AHR airway hyperresponsiveness AIA antigen-induced arthritis AID activation-induced cytidine deaminase APC antigen-presenting cell APRIL a proliferation-inducing ligand BAFF B cell-activating factor BAFFR B cell-activating factor receptor BCL-6 B cell lymphoma-6 BCMA B cell maturation antigen BCR B cell receptor Be B effector BGG bovine gamma globulin BLIMP-1 B lymphocyte-induced maturation protein 1 BM bone marrow BrdU 5-bromo-2'-deoxyuridine Breg regulatory B cell BTB-ZF broad-complex, tramtrack, and bric-a-brac zinc finger BTK Bruton’s tyrosine CCL CC chemokine ligand CCR CC chemokine receptor CD cluster of differentiation CGT ceramide galactosyl CHS contact hypersensitivity CIA collagen-induced arthritis CII type II collagen CLP common lymphoid progenitor CSR class-switch recombination CTB cholera toxin B CXCL CXC chemokine ligand CXCR CXC chemokine receptor D diversity in reference to gene rearrangement DC dendritic cell DL1 delta-like 1 DNFB 2,4-dinitrofluorobenzene DTH delayed type hypersensitivity 12 DTR diphtheria toxin receptor EAE experimental autoimmune encephalomyelitis EAU experimental autoimmune uveitis FDC follicular dendritic cell Fo follicular GALT gut associated lymphoid tissue GC germinal centre GCS glucosylceramide synthase GITRL glucocorticoid-induced TNF ligand GM-CSF granulocyte-macrophage colony-stimulating factor GPI 6--6-phosphate isomerase H heavy in reference to Ig chains HEL hen egg lysozyme HPLC high performance liquid chromatography HSC hematopoietic stem cell i.p. intraperitoneal i.v. intravenous IBD inflammatory bowel disease ICOS inducible T cell co-stimulator IFA incomplete Freund’s adjuvant Ig immunoglobulin IL interleukin iNKT invariant natural killer T ITAM immunoreceptor tyrosine-based activation motif ITK IL-2-inducible T cell kinase J joining in reference to gene rearrangement L light in reference to Ig chains LPS lipopolysaccharide mAb monoclonal antibody MAIT mucosal-associated invariant T MAML1 mastermind-like protein 1 MBP myelin basic protein MCMV murine cytomegalovirus MFI median fluorescence intensity MHC major histocompatibility complex MINT MSX2-interacting protein MLN mesenteric lymph node 13 MOG myelin oligodendrocyte glycoprotein MS multiple sclerosis MyD88 myeloid differentiation primary-response gene 88 MZ marginal zone NF-κB nuclear factor κB NOD non-obese diabetic OVA ovalbumin pAb polyclonal antibody PALS periarteriolar lymphoid sheath Pax5 paired box 5 PBMC peripheral blood mononuclear cell PD-1 programmed death-1 PD-L1 programmed death ligand-1 PI3K phosphoinositide 3-kinase PLZF promyelocytic leukaemia zinc finger PMA phorbol 12-myristate 13-acetate Prdm1 positive-regulatory-domain containing 1 PRR pattern recognition receptor RA rheumatoid arthritis RAG-1/-2 recombination activating genes-1/-2 RBP-Jκ recombination signal binding protein for Igκ J region RT room temperature S1P sphingosine 1-phosphate SAP SLAM-associated protein SHM somatic hypermutation SLC surrogate light chain SLE systemic lupus erythematosus STAT signal transducer and activator of transcriptions T2-MZP transitional 2-marginal zone precursor TACI transmembrane activator and calcium modulator and cyclophilin-ligand interactor TCR T cell receptor TD T-dependent TdT terminal deoxynucleotidyl transferase Tfh T follicular helper TGF-β transforming growth factor-β Th T helper TI T-independent 14 TIDM type I diabetes mellitus TIM T cell Ig and mucin domain TLR Toll-like receptor TNF tumour necrosis factor UDP uridine diphosphate V variable in reference to gene rearrangement WT wild-type Xid X-linked immunodeficiency α-GalCer α-galactosylceramide

15 Chapter I: Introduction

B cells In 1965 Cooper et al. reported the existence of a two-way division of the lymphoid immune system, which is recognised as a fundamental milestone of the B cell field (1). They determined that B cells developing in the bursa of Fabricius in birds gave rise to plasma cells and antibody responses, while T cells with thymic origin were required for cell-mediated immunity in delayed hypersensitivity reactions. Bone marrow development for B cells in humans and mice was described later (2).

B cells can be defined as lymphocytes that express an immunoglobulin (Ig) receptor, which recognises specific antigenic epitopes in their native conformation (as opposed to T cells that recognise processed peptides from protein antigens in the context of major histocompatibility complex (MHC) molecules). Furthermore, unlike T cells, B cells recognize a vast variety of antigens beyond proteins.

The surface Ig, together with the signal transduction molecules Ig-α (cluster of differentiation (CD) 79a) and Ig-β (CD79b), make up the B cell receptor (BCR) (3). Antibodies are the secreted form of the surface Ig. The Ig is composed of two identical heavy (H) chains and two identical light (L) chains (4). Both H and L chains have constant and variable regions. It is the combination of the variable regions of an H and an L chain that generates the antigen-binding site and determines the specificity for antigen (4). The constant region of the H chain can take five major distinguishable isoforms, which also determine some of the effector functions of the antibodies (4), as discussed below.

The primary role of B cells is to generate protective humoral responses, nevertheless it is now recognised that they play a broader role in the maintenance of immunological homeostasis, including in antigen presentation, cytokine production, and suppression of immune responses. In order to address the role of B cells in immune regulation, which is the focus of this thesis, B cell development, maturation, as well as phenotypic and functional diversity must first be appreciated.

B cell ontogeny in the bone marrow and central tolerance B cell development begins in the foetal liver and continues in the bone marrow (BM) in both mice and humans. B cell development from hematopoietic stem cell (HSC) progenitors is a highly regulated process, during which multiple checkpoints act

16 to ensure that largely non-autoreactive B cells expressing functional BCRs have access to the periphery (5).

Generation of B cell antigen receptors An absolute prerequisite in B cell development is the functional rearrangement of the Ig loci, and this characterizes early B cell development. B cells progress via pre- pro-, early and late pro-, large and small pre-B cell developmental stages that are defined by the rearrangement of Ig genes and the expression of cell surface Ig, and finally form immature B cells expressing surface IgM (6), as shown in Figure 1.1A. Ig H chain genes (variable (V), diversity (D) and joining (J) gene segments) are rearranged at the pro-B cell stage (7), while L chain genes (V and J gene segments) are rearranged at the late pre-B cell stage, following successful expression of the pre-BCR (8). This process of somatic recombination is unique to mammalian genes.

B cell gene rearrangement is initiated through the activation of recombination activating gene (RAG)-1 and RAG-2 products (9, 10). These , as part of multi- subunit complexes, introduce DNA strand breakage at specific signal sequences that flank Ig gene segments. The rejoining of these segments marks the end of the recombination event. The expression of RAG-1/2 is dynamically and tightly regulated throughout B cell progression through the developmental stages in the BM. Repertoire diversity is generated by the random rearrangement of the many V, D, J gene segments that form the variable regions of the Ig, summarized for humans in Figure 1.1B. Further combinatorial diversity is achieved through the many possible H and L chain variable region pairings to form the antigen-binding site (8). deletion, addition of palindromic sequence as well as of non-templated nucleotides by the specialized DNA terminal deoxynucleotidyl transferase (TdT) at the junctions between gene segments also contribute to diversity generation (11). Rearranged variable region gene segments are joined to the gene segments encoding the constant regions for the production of complete Ig H and L chain genes.

There are three main steps to every developmental stage starting from pro-B cells – assembly, expression and signalling via BCR. The pro-B cells first make the D-J rearrangement, followed by V-DJ rearrangement (8). Pre-B cells are defined by their expression of the pre-BCR – H chain joined with the invariant surrogate L chain (SLC) (6, 12, 13). SLC is a heterodimer composed of λ5 and VpreB. Ig-α and Ig-β also form a part of this pre-BCR complex. These molecules have immunoreceptor tyrosine-based activation motifs (ITAMs) that become phosphorylated upon BCR engagement and are

17 thus essential for signal transduction through the BCR (14). Signalling through the pre- BCR halts H chain rearrangement, and B cells undergo a proliferative burst, generating clones of pre-B cells carrying the same H chain (15). This is called the large pre-B cell stage since these cells are proliferating.

Following proliferation and internalisation of the pre-BCR, RAG-1/2 are re- expressed and V-J recombination commences at the L chain in small pre-B cells (first at the κ, then λ loci, if initially unsuccessful) (14). Once a functional L chain is synthesized, it is expressed on the surface of the B cells together with the H chain, marking the development of an immature B cell (14).

Somatic recombination can be productive or non-productive – due to generation of an out-of-frame V(D)J exon (16). Failure to make productive gene rearrangements and express Ig on the surface at appropriate times during development results in cell death (17). If human pro-B cells generate a non-productive H chain gene they can attempt to rearrange it from the other allele, while B cells have four attempts to rearrange the L chain – maternal and paternal κ and λ loci (14). Somatic recombination also leads to allelic exclusion. Allelic exclusion refers to the phenomenon of expression of only one of two alleles, despite the potential to express both, and this ensures the monospecificity of a B cell (14, 16).

B cell tolerance checkpoints Negative selection mechanisms operate in the BM, with an estimated 10-20% of newly formed B cells emerging from the BM into blood (18). Central tolerance refers to the negative selection processes that occur in the BM to ensure that the emerging B cell pool is largely non-autoreactive (18). In studies of B cell repertoire, it has been reported that up to 50-75% of immature B cells in the BM react with self-antigens (19, 20). This is in stark contrast to 20-40% peripheral B cells displaying self-reactivity.

Following BCR expression, immature B cells are tested for their reactivity against self-antigens (18). If a B cell is self-reactive, it can undergo rescue through receptor editing, or be controlled via clonal deletion or anergy. Receptor editing is the primary mode of ensuring non-autoreactivity (21). The rearrangement of V-J genes encoding the L chain was shown first in 1993 following the description of the other central tolerance mechanisms (22-24). Albeit less common, receptor editing can also occur at the H chain loci; this is known as H chain replacement (25). H chain replacement is estimated to occur in 5-10% of developing B cells (26). Clonal deletion occurs if receptor editing has been unsuccessful and a B cell binds a multivalent ligand 18 with high avidity, such as a highly expressed membrane-bound protein (27-29). These cells die by apoptosis (29). Anergy is a state of unresponsiveness that is caused by the recognition of monovalent, for instance soluble self-antigens (30). Anergic cells express IgD, but only low levels of IgM and have short-lifespans (with half-lives of 3-5 days), compared to their non-anergic counterparts in the periphery (31, 32). They also fail to respond to antigen and this is due to the recruitment of inhibitory signalling scaffolds to the BCR. Nevertheless, following these steps some self-reactivity remains, as some self- reactive B cells escape to the periphery (20, 33, 34). Peripheral tolerance mechanisms further act to purge the immature B cell pool from autoreactivity. It is estimated that in humans, tolerance mechanisms acting on immature B cells, reduce self-reactivity from 40 to 20% (20). It is also shown in transgenic mouse models that, when self-antigen expression is limited to periphery, negative selection events act primarily on immature B cells (35, 36).

Microenvironmental cues in B cell development Stromal cells play an important role in generating the bone marrow microenvironmental niches that critically support B cell development (37). CXC chemokine ligand (CXCL) 12 was the earliest factor reported to be required for pre-pro- B cell development (38). Through interacting with its ligand CXC chemokine receptor (CXCR) 4 it facilitates HSC homing to the BM as well as their retention (38-40). Early pre-pro-B cells expressing CXCR4 localise some distance away from interleukin (IL)-7 producing stromal cells (41). IL-7 is a critical signal for B cell maturation, proliferation and survival (42). Lack of IL-7 or IL-7Rα leads to severely reduced numbers of B cells following the pre-pro-B cell stage in the murine BM, suggesting that IL-7 plays a crucial role at the pro-B cell stage in mice (43, 44). Furthermore, IL-7 maintains the common lymphoid progenitor (CLP) and pre-pro-B cell potential to differentiate into B cells (45). The spatial dissociation of CXCL12 and IL-7 provision by stromal cells, is an example of specialised BM niches for the different B cell developmental stages. However, patients with primary immunodeficiency associated with mutations in the IL7R gene have been reported to have normal circulating B cell numbers (46). Therefore, in humans, the role of IL-7 in B cell development is not well understood. FLT3 ligand (FLT3L) is also essential for the development of pre-pro-B cells through regulation of CLPs (47, 48), nevertheless its exact function remains to be elucidated. The delivery of microenvironmental cues requires complex signal integration as well as coordinated cellular interactions to support immature B cell development.

19 A role for the BCR during B cell development, maturation and survival Many findings have led to consider the important role of the BCR signal in the “positive selection” of B cells (49). “Positive selection” refers to the process by which nascent B cells test the generation of a functional BCR and undergo competition for limited survival factors (49). Continued selection and maturation requires: (1) functional BCR expression and signalling as well as (2) competition for limited resources, including optimal signals through the BCR.

The requirement for the BCR expression in the survival of all B cells was shown by conditional deletion of BCR (50). This leads to the rapid loss of all B cells, including peripheral naïve B cells (50). In particular, pre-B cells that lack expression of a functional BCR, undergo developmental arrest and are unable to exit the BM (51-53). Targeting of Ig-α/Ig-β complexes to the B cell surface can rescue B cell development in the absence of IgH and IgL chains (54). Indeed, it has been shown that pre-BCR can signal immediately after assembly in the absence of any known ligand (55). This is known as “tonic signalling” (55). In the absence of basal signalling, B cells revert to pro- and pre-B cell developmental stages with associated reactivation of V(D)J recombination (56-58).

The evidence for “positive selection” based on BCR expression and role for BCR signalling in shaping the mature naïve B cell repertoire comes from studies assessing the BCR repertoire in the BM and periphery. This has revealed enrichment for selected V gene usage as well as a bias towards specific H/L chain pairings, suggesting selection events based either on the enhanced structural stability of the BCR or ligand binding (59, 60). Furthermore, under conditions of low antigen abundance or when BCR signalling is reduced, self-antigen-recognizing B cells have an advantage over cells with non-self BCR specificities (61, 62). Specifically, low amounts of membrane- bound antigens induce the selection and accumulation of B cells bearing self-reactive BCRs (61). These observations could be explained by requirement of BCR signalling to reach a threshold level, either by expression of sufficient levels of BCR on the cell surface, or by low affinity binding to undefined endogenous ligands. Further studies have shown the importance of BCR signal threshold in the continued differentiation of immature B cells (63-65). Reduced expression of BCRs is correlated with impaired transitional B cell development (65). Cyster et al. crossed mice deficient in the CD45 tyrosine phosphatase, which positively regulates signal transduction via BCR, to mice carrying Ig transgenes with specificity for hen egg lysozyme (HEL)(62). In these mice,

20 the authors observed significant impairment in mature B cell generation, while immature B cell development was normal (66). Mature B cell development could be rescued either through BCR stimulation or upon transfer into irradiated lymphopenic hosts that express the relevant self-antigen. Thus ligand binding by the BCR can compensate for the reduction in signal transduction resulting from CD45 deficiency, again supporting a requirement for overcoming a threshold of BCR signalling in B cell development.

Another crucial microenvironmental signal in B cell development is B cell activating factor (BAFF). BAFF is a trimeric member of the tumour necrosis factor (TNF) family, it signals through BAFF receptor (BAFFR) (as well as B cell maturation antigen (BCMA), and transmembrane activator and calcium modulator and cyclophilin- ligand interactor (TACI) on more mature B cell subsets) (67). This pro-survival cytokine is necessary for B cell survival beyond the late transitional stage (68, 69), discussed below. BAFF signalling can activate both non-canonical and canonical nuclear factor-κB (NF-κB) pathways (70). The non-canonical NF-κB pathway is critical for B cell survival, as evidenced by severe impairment in B cell survival in mice with deficiencies in this pathway (71-73). BAFF is present in high concentrations in lymphoid follicles and through interactions with its receptor BAFFR it acts as a survival signal for B cells (70). B cell survival is promoted through the integration of BAFFR and tonic BCR signals. BCR signals increase the level of NF-κB2 (p52) precursor p100 and p100 is a substrate for IκB kinase-α (IKKα), which is activated downstream of BAFFR (74). Of note, it has been reported that on immature B cells BAFFR expression correlates with that of BCR as well as with tonic signalling (75).

21

A Pre-pro-B cell Early pro-B cell Late pro-B cell Large pre-B cell Small pre-B cell Immature B cell VpreB pre-BCR BCR λ5 Ig-α/Ig-β

Recombination - DH-JH VH-DHJH - VL-JL - RAG/TdT expression - + + - + + Surface phenotype B220+ B220+ B220+ B220+ B220+ B220+ KIT- KIT+ KIT+ KIT- KIT- KIT- CD19- CD19+ CD19+ CD19+ CD19+ CD19+ FLT3+ FLT3- FLT3- FLT3- FLT3- FLT3- CD24-/lo CD24+ CD24+ CD24+ CD24+ CD24+ CD43+ CD43+ CD43+ CD43- CD43- CD43- IgM- IgM- IgM- IgM- IgM- IgM+ B H κλ V 38-46 34-38 29-33 D 23 - - J 6 5 4-5

C 9 1 4-5 Figure 1.1. B cell development in the BM. (A) Earliest identifiable B cell-committed progenitors are pre-pro-B cells, but these have also been reported to maintain alternative lineage differentiation potential (37). Early pro-B cells initiate IgH rearrangement. Following productive VHDHJH rearrangement, recombination is halted and pre-BCR expression marks the differentiation of large pre-B cells. The pre-BCR consists of the rearranged H chain and the invariant L chain, composed of VpreB and λ5, as well as Ig- α and Ig-β signal transduction molecules. These cells proliferate and differentiate into small pre-B cells. Small pre-B cells rearrange the L chain loci. Upon successful L chain rearrangement the BCR is expressed on nascent immature B cells. (B) The number of functional gene segments in the human Ig loci. The numbers are derived from cloning and sequencing in one individual.

22 B cell heterogeneity and functional maturation in the periphery Most B cell functional maturation occurs in secondary lymphoid tissues, including the lymph nodes and the spleen. B cells are divided into two main ontogenically different populations: B-1 and B-2 cells (76, 77).

B-1 cells B-1 cell arise primarily in the foetal liver and are generally self-renewing in adult life (78). B-1 cells are present abundantly in the peritoneal and pleural cavities, but are also present in the spleen (79). B-1 cells are polyreactive with restricted specificity, they are predominantly IgM-expressing (little class-switching to other antibody isotypes) and undergo very little affinity maturation, which enhances specificity of the BCR for a particular antigen (79). B-1 cells are phenotypically identified as CD19hiCD23-CD43+IgMhiIgDvariable (79), and can be further divided into two functionally distinct subsets, namely B-1a (CD5+) and B-1b (CD5-) B cells (80). B- 1a cells are proposed to be the main source of natural antibodies, while B-1b cells provide protection against polysaccharide antigens (80). B-1 cell responses are predominantly T-independent (TI), unlike B-2 cell responses which are largely T- dependent (TD) (81). In TD responses the B cell requires CD4+ T cell help for antibody production, while in TI responses the B cell can generate an antibody response without CD4+ T cell help (82). TI responses are further subdivided into type I and II, which are respectively elicited by mitogens and multivalent ligands. The existence, function, and subpopulations of human B-1 cells are still a matter of debate (83).

B-2 cells B-2 cells form the adaptive arm of the humoral immune response, going on to generate high affinity, class switched antibody secreting cells, and memory B cells, following cognate interactions with helper T cells. B-2 B cells exit the BM as immature B cells that home to the spleen, entering via terminal branches of central arterioles in the red pulp, from which they migrate to the outer zone of the periarteriolar lymphoid sheath (PALS) (84). In the spleen B cells progress through distinct transitional stages that result in the development of mature naïve follicular (Fo) and marginal zone (MZ) B cells (70). Of note, about a quarter of transitional B cells complete maturation to Fo B cells in the BM (85, 86). Most mature B cells are recirculating Fo B cells that are found within follicles in secondary lymphoid organs (70). MZ B cells are found in the splenic MZ and in mice are sessile, while in humans have capacity to recirculate (70).

23 There is a lack of consensus about the number, phenotype and developmental relationships between transitional B cell stages that bridge immature B cells leaving the bone marrow and mature Fo and MZ B-2 cells in the periphery (87). Similarly the events that shape transitional B cell pool are still disputed. Alternative schemes of B cell maturation are discussed below, and while they use similar nomenclature, it remains unclear to what extent the different populations of transitional B cells are functionally overlapping.

Loder and colleagues first proposed the existence of T1 and T2 B cells (88). T1 cells are phenotypically identified as CD21loCD24hiIgMhiIgD-/lo, and T2 cells as CD21int/hiCD24hiIgMhiIgDhi B cells in the murine spleen. They reported that, while T1 B cells undergo apoptosis upon BCR engagement, T2 B cells proliferate in response BCR cross-linking. This led the authors to propose that T1 cells are the key targets for negative selection. The cells that survive, differentiate into T2 cells and undergo BCR- signalling mediated positive selection to develop into either mature Fo or MZ B cells. T2 cells were also suggested to undergo a proliferative burst before entering the mature B cell pool (88).

Allman et al. proposed a second model by dividing transitional B cells into three non-proliferative/non-cycling transitional stages (T1-T3) (89, 90). In this study transitional B cells were designated as T1 (AA4+CD23−IgMhigh), T2 (AA4+CD23+IgMhigh), and T3 (AA4+CD23+IgMlow). While displaying high turnover rates, all subsets failed to proliferate in response to IgM cross-linking in vitro. Of interest, in this study they took advantage of an in vivo 5-bromo-2'-deoxyuridine (BrdU)-labelling tracking system and showed measurable cell loss at both the T1-T2 and T2-T3 transitions, suggesting the existence of multiple selection points within the peripheral immature B cell pool. Unlike the scheme reported by Loder et al., in this model it is the T3 B cell subset that gives rise to mature B cells. From T1 to T3 stages the cells display progressive increase in ability to respond to stimulation via the TNF receptor family member CD40 and Toll-like receptors (TLRs) (89). This model was later challenged by a report suggesting that T3 cells have properties of anergic rather than transitional B cells, including the expression of low levels of IgM following recognition of self-antigen (91).

Subsequent studies have tried to resolve the discrepancies between the two models. The general consensus almost 20 years later is that the T2 population proposed by Loder et al. (88) is heterogeneous both phenotypically and functionally, and that an

24 additional stage of transitional differentiation exists. For example, following the observation that NF-κB p50 or Notch2 deficiency led to the absence of not only MZ B cells, but also CD21hiCD24hi splenic B cells (92, 93), Pillai et al. suggested that T2 cells with CD21hiCD24hi phenotype are primarily MZ B cell precursors (49). These cells are now termed transitional 2-marginal zone precursor (T2-MZP) B cells. The finding that T2-MZP (CD21hiCD24hi T2) B cells are a separate population to CD21intCD24hi T2 is supported by other studies. It has been shown that T2-MZP B cell generation is delayed in comparison to Fo B cells in sublethally irradiated mice, thus they are not precursors capable of giving rise to Fo B cells (90). Furthermore, T2-MZP B cells have a heightened proliferative response to lipopolysaccharide (LPS), which makes them functionally more similar to MZ B cells than Fo B cells, discussed later in this section. More recently, Rawlings and colleagues characterized a bipotent CD21intCD24hi T2 B cell population (94). They show that these cells give rise to both MZ and Fo B cells, upon encounter with different stimuli, and display augmented responses to a range of microenvironmental cues, compared to T1 cells. Unlike T2-MZP B cells, they are not long-lived and display minimal Notch2 activation. Addressing the controversy about the ability of transitional B cells to proliferate, the authors show that a significant fraction of these T2 B cells are indeed cycling. This suggests that these cells include both precursors to cycling T2-MZP and non-cycling Fo B cells. This ability to proliferate is nevertheless not a prerequisite for B cell maturation. It is also shown that it is at the T2 stage that BCR-mediated selection occurs. Of note, transitional B cells have been identified in human peripheral blood (95-97). These immature cells are phenotypically CD19+CD38hiCD24hi (IgMhiIgDhiCD5+CD10+CD20+CD27-CD1dhi).

The model that is used in reference to murine transitional and mature B cells and their phenotypes is shown in Figure 1.2A. This model will be used throughout this thesis. In adult C57BL/6 mice these populations of splenic CD19+ cells can be distinguished based on the expression of CD21, CD23, and CD24 (Fig. 1.2B).

Characteristics of mature B cell subsets Most B-2 cells are non-self renewing Fo B cells, which home to B cell follicles in secondary lymphoid organs (70). These B cell follicles are adjacent to T cell areas, which facilitate B cell interaction with T cells at the border area. Fo B cells are therefore optimally positioned to respond to TD antigens. As shown in Fig. 1.2A, Fo B cells are CD21intCD23+CD24loIgDhiIgMlo. In response to antigen in the periphery Fo B cells can differentiate into plasmablasts, plasma cells, and memory B cells (70).

25 Like Fo B cells MZ B cells arise as antigen naïve cells, however, they display some characteristics of activated B cells including increased expression of MHC class II, CD80 and CD86 (98, 99). Unlike naïve Fo B cells, MZ B cells are also potent stimulators of naïve T cell responses (100). In mice MZ B cells reside in the MZ, which is the border area between the marginal sinus and the red pulp (101). This is where the arterial blood supply terminates and it enables MZ B cells to act as the first line of defence in protection from blood-borne pathogens (102). Of note, MZ B cells are self- renewing (70). MZ B cells share a number of similarities with B-1 cells: they can produce TI antibody (TI-II antibody) (103); they can produce antibody spontaneously in the absence of infection or immunization; as with B-1 cells the antibody they produce tends to be low affinity with broad specificity (101). Immunohistological studies have shown that the human MZ is morphologically different to rodent (104) and therefore the phenotypic and functional properties of MZ B cells in humans are less understood (101). In humans, the splenic MZ contains B cells that are phenotypically IgMhiIgDlowCD1c+CD21hiCD23-CD27+ (101). In addition to the spleen, they are also found in the subcapsular sinus in the LNs, tonsilar epithelium and subepithelial dome of Peyer’s patches (101). While like murine MZ B cells they act like front-line sentinels and generate TI responses, there is also evidence for their participation in TD responses (101). Furthermore, their expression of CD27 (a B cell memory marker), together with the observations of some V(D)J mutation and germinal centre (GC) footprint, suggests that human MZ B-like cells have some properties of memory B cells (101).

26

A

T2-MZP MZ

AA4.1-IgDhiIgMhi AA4.1-IgDloIgMhi CD21hiCD23+CD24hi CD21hiCD23-CD24hi CD1dhi CD1dhi T1 T2

AA4.1+IgD-IgM+ AA4.1+IgDhiIgMhi CD21loCD23-CD24hi CD21intCD23+CD24hi

Fo

AA4.1-IgDhiIgMlo CD21intCD23+CD24int B Gated on Gated on CD19+ cells CD19+CD21hiCD24hi

Fo T2-MZP CD21

CD24 CD23

T1 T2 MZ

Figure 1.2. B cell development in the periphery. (A) B cell maturation into Fo and MZ B cells occurs via transitional stages with the indicated phenotypes. (B) Splenic B cell developmental stages in mice as identified by flow cytometry staining for CD21, CD23 and CD24 in CD19+ B cells.

27 MZ versus Fo B cell fate decision The fate decision governing the differentiation of late transitional cells into Fo or MZ B cells is determined by the integration of signals from the BCR, BAFFR, canonical NF-κB pathway and Notch2 (70). In general, weak BCR signals lead to MZ B cell development, while strong signals favour commitment to Fo B cell fate (70). That BCR signal strength is a crucial determinant of Fo/MZ fate decision is demonstrated by studies where components of the BCR signalling cascade, as well as positive and negative modulators of this cascade, are absent. Bruton’s (BTK) is activated downstream of the BCR, and X-linked immunodeficiency (Xid) is caused by mutations in the BTK gene (105). Mice carrying the Xid mutation have impaired ability to transduce BCR signals and have substantially reduced Fo B cell compartment (105). Similarly, the absence of CD45, a component of the BCR signalling complex which amplifies BCR signals, leads to the loss of Fo B cells (106). Evidence for weak BCR signalling promoting MZ development was provided from Aiolos-deficient mice in which BCR signalling is increased. These mice had increased mature Fo B lymphocyte population, and an associated absence of MZ as well as T2-MZP B cells (107). Similarly, mice deficient in the inhibitory receptor CD22, which attenuates BCR signalling through recruitment of the tyrosine phosphatase SHP1, have a reduced MZ B cell population (108).

The MZ B cell fate decision also involves Notch2 signalling, as conditional deletion of Notch2 results in a defect in both MZ and their precursor T2-MZP B cell development (92). Intracellular Notch2, recombination signal binding protein for Igκ J region (RBP-Jκ) and Mastermind proteins form a complex that activates target genes (70). Mice with B cell specific deletion of Notch2 or RBP-Jκ, as well as the Notch ligand delta-like 1 (DL1) and mastermind-like protein 1 (MAML1) deficiency have impaired MZ B cell development (92, 109-112). On the contrary, increased differentiation of MZ B cells is observed in the absence of MSX2-interacting nuclear target protein (MINT) that negatively regulates Notch signalling (113).

As described above, BAFFR signalling can also activate the canonical NF-κB pathway (114). MZ B cell development requires BAFF signalling, but BAFF apparently plays a different role to that in B cell survival in the follicles (69, 115, 116). MZ B cell development is impaired in mice lacking NF-κB1 (also known as p50), and partially inhibited in mice deficient in REL (c-Rel) or RELA (p65) (93).

28 Weak BCR signals Canonical NF-κB signals Weak BCR signals Notch2 signals Canonical NF-κB signals Notch2 signals Tonic BCR signals BAFF survival signals T2-MZP MZ Non-canonical NF-κB signals AA4.1-IgDhiIgMhi AA4.1-IgDloIgMhi CD21hiCD23+CD24hi CD21hiCD23-CD24hi CD1dhi CD1dhi T1 T2

AA4.1+IgD-IgM+ AA4.1+IgDhiIgMhi CD21loCD23-CD24hi CD21intCD23+CD24hi

Strong BCR signals Fo Canonical NF-κB signals AA4.1-IgDhiIgMlo CD21intCD23+CD24int

Figure 1.3. Signals in B cell maturation. Immature B cells continue their development in the periphery trough transitional B cell stages. T1 and T2 B cells mature into Fo or MZ B cells with the integration of signals from the BCR, BAFFR, canonical NF-κB pathway and Notch2.

29 Generation of humoral responses within GCs Unlike the specific receptors of the T cell compartment, B cells have the capacity to modify their antigen receptors in the periphery to enhance protection against pathogens. These modifications occur within specialized areas of lymphoid organs, known as GCs (117). GCs are critical for the generation of the majority of memory and plasma cells and for the formation of high-affinity antibody responses to antigens. High affinity antibody responses depend on the processes of somatic hypermutation (SHM) of genes encoding the Ig variable region, and affinity maturation (competition between B cells for antigen binding to test the value of new mutations). Additionally the process of class-switch recombination (CSR) can alter the isotype of Ig, and thus function, of the antibody response depending on the nature of the pathogen (117).

During TD responses, GCs are formed by proliferating B cells in the follicles of lymphoid organs. When a mature naïve B cell encounters cognate antigen, it becomes partially activated, upregulates CCR7 expression, which enables it to migrate towards CCL19 and CCL21 chemokines of the T cell area (118). Here the B cell receives further activation signals from T cells. Crucial in this interaction is stimulation through CD40. CD40 is expressed by all B cells, and its ligand CD40L (CD154) is expressed on activated MHC class II-restricted CD4+ T helper (Th) cells. In turn B cells provide signals for further T cell activation (119). B cells present antigen in the context of MHC class II and express CD80/CD86, which interact with the T cell receptor (TCR) and CD28, respectively. Both MHC class II and CD80/86 are upregulated on B cells in a manner dependent on BCR recognition of antigen.

This interaction has two outcomes for B cells: (1) B cells differentiate into antibody-secreting plasma cells that localize in specialized sites, including the medullary cords of lymph nodes; (2) B cells mature into GC precursor B cells that colonize a follicle consisting of IgD+IgM+ B cells and non-hematopoietic follicular dendritic cells (FDCs). The latter outcome for B cells, results in the formation of a GC, where B cells proliferate, undergo SHM and affinity maturation. B cells of the GC reaction, which gain increased affinity for the cognate antigen outcompete their counterparts that bind antigen less avidly, for survival and proliferation signals. Finally, through mechanisms that are incompletely understood, some B cells differentiate into plasma cells, while others develop into memory B cells (120).

Histologically, the GC consists of dark and light zones, which are areas of proliferating B cells known as centroblasts and non-cycling cells, known as centrocytes.

30 The movement by centroblasts in the dark zone and centrocytes in the light zone is proposed to occur through CXCR4-CXCL12 and CXCR5-CXCL13 mediated signals, respectively (121).

Following seeding of the dark zone by activated B cells, B cells/centroblasts reinternalise their BCR and undergo SHM. SHM refers to the introduction of point mutations into the Ig V region genes by activation-induced cytidine deaminase (AID) (122-125), which catalyses the deamination of cytidines in the DNA (126-131). It is then through an error-prone repair process that somatic mutations are introduced. Point mutations accumulate at a rate of ~1/103 per cell division in V region genes (as opposed to ~1/1010 in the other DNA regions) (132). The introduction of random mutations in the V region genes may change the structure of the BCR so that it recognises cognate antigen with a higher, or lower, affinity than the BCR encoded by unmutated genes. The affinity of the new BCR is then tested in the light zone by competition for antigen held on FDCs. Cells with improved antigen binding are selected; this is the process of affinity maturation. Centrocytes give rise to memory B cells and plasma cells. CSR occurs at the centrocyte stage; it is an irreversible recombination event, which switches the class from IgM/IgD to other classes (IgG, IgA, IgE) with distinct effector functions. Like SHM, CSR requires AID (124, 125). The resulting Ig class is determined by integration of signals from the co-stimulatory molecules and cytokines.

The cellular interactions and molecular events in GCs, as well as in GC formation and involution are much more dynamic and less defined than the linear scheme described above proposes (133). For example, two-photon microscopy of murine lymph nodes shows that activated antigen-specific B cells can shuttle not only within, but also bi-directionally between dark and light zones (134, 135).

Molecules important in GC reaction An important role in the GC response is played by the nuclear phosphoprotein B cell lymphoma-6 (BCL-6), which is a transcriptional repressor. In B cells, BCL-6 is specifically expressed by GC centroblasts and centrocytes (136, 137) and, in the absence of BCL-6 in mice, GCs fail to form (138, 139). The proposed role of BCL-6 in GCs is to enable B cells undergoing SHM and CSR to withstand the genotoxic stress of DNA damage. It has also been suggested to suppress the expression of B cell activation factors that may enable their interaction with T cells prematurely. Finally, BCL-6 suppresses the expression of positive-regulatory-domain containing 1 (Prdm1). Prdm1 encodes B lymphocyte-induced maturation protein-1 (BLIMP-1), which is crucially

31 required for plasma cell development (140, 141). BCL-6 downregulation in centrocytes is mediated through BCR and CD40 (136, 142, 143).

The control of B cell proliferation and death is important in the GC response to ensure that (1) a large pool of Ig specificities that can undergo affinity maturation are generated and (2) B cells with non-functional gene rearrangements or non-binding Igs are eliminated promptly. Centroblasts are rapidly dividing and a crucial step in centroblast differentiation is the downregulation of cell cycle inhibitors and upregulation of genes involved in proliferation (144, 145). Of note, ensuring the maintenance of replicative potential, centroblasts upregulate the expression of (146). In addition to pro-proliferative program, B cells within GCs have an increased expression of pro-apoptotic molecules and reduced expression of anti- apoptotic molecules (144, 147). This allows their rapid elimination by cell death in case non-functional gene rearrangements or non-binding Igs are generated.

The important role for CD40 in GC formation is observed in both mouse and human, where CD40-/CD154-deficiency leads to lack of GC formation (148). Furthermore, anti-CD40L administration in mice leads to the disruption of already established GCs (149). CD40-CD154 interaction with GC T cells is important in the induction of CSR in B cells (117). Roles for inducible T cell co-stimulator (ICOS), TACI and BAFFR in CSR have been proposed (150-152).

There is considerable interest in understanding how the memory/plasma cell fate decision is made. BCR-driven signals are crucial in plasma cell development (153). However, how plasma and memory cell development is regulated outside of this requirement for BCR signals, is incompletely understood. It is known that inactivation of paired box 5 (Pax5), expressed in mature naïve, GC and memory B cells, as well as BCL-6 is required (154). Additionally, an important role for IRF4 in plasma cell generation through repression of the GC B cell transcriptional program has been suggested (142, 155, 156).

The factors that enable differentiation of memory cells are even less clear. A role for phosphorylated STAT5 (pSTAT5) through regulation of BCL-6 has been proposed (157), however memory B cells still develop in the absence of BCL-6 (158) and the ectopic expression of BCL-6 in human B cells inhibited memory B cell generation in vitro (159). The role for p-STAT5 in memory B cell differentiation therefore remains to be determined. As memory cells express Pax5, it is suggested that

32 memory cell development is brought about through the absence of Pax5 downregulation (117).

B cells receive help from specialized T cells, known as follicular helper T (Tfh) cells. Tfh cells express high levels of Bcl-6, CD40L, ICOS, SLAM-associated protein (SAP), CXCR5, CXCR4, IL-21, and programmed death-1 (PD-1) (160). The signals provided to B cells by Tfh cells include IL-4, IFN-γ, CD40L, IL-21 and BAFF. More recently invariant natural killer T (iNKT) cells with similar phenotype have been described (161).

33 B cell roles in addition to antibody production B cells are primarily recognized for their role in protection from pathogens through antibody secretion (15). Of interest, most successful vaccination strategies rely on eliciting a strong Ig response. B cells also play other roles in immune homeostasis and activation.

B lymphocytes can act as professional antigen-presenting cells (APCs) both to activate and tolerise T cells (119). There are several properties that enable the unique APC function of B cells – location, endocytic capacity and antigen processing (119).

Upon antigen encounter, B cells in LN follicles upregulate CCR7, which allows them to migrate towards the T cell zone chemokines CCL19 and CCL21 (162). B cells then form stable conjugates with antigen specific CD4+ T cells and thus are able to participate in the early phase of an immune response (163, 164).

BCR-mediated endocytosis allows B cells to concentrate small quantities of specific antigen and present it efficiently. Indeed, it has been shown that BCR affinity for antigen is proportional to B cell capacity to present it to CD4+ T cells and elicit their proliferation (165). Furthermore, because BCR-mediated uptake leads to blocking of regions of the antigen bound by the Ig, the determinants presented are different to those originally recognized by the B cell (166, 167). This leads to the diversification of the immune response (168). Antigen recognition by the Ig leads to BCR-antigen complex internalisation and guiding towards MHC class II-rich compartments (169-171). BCR- mediated uptake of antigen also changes the antigen processing machinery so that B cells preferentially present antigen internalized via the BCR (119). It also induces optimal conditions, such as acidification, within these compartments, which facilitate antigen processing and loading.

B lymphocytes need to become activated to become competent APCs. This is achieved through BCR signalling and CD40 engagement following specific antigen encounter (119). BCR signals induce B cell growth, proliferation, survival, and co- operate with CD40-derived signals to enhance expression of MHC class II and CD80/CD86, which reciprocally act to activate T cells (119). Therefore, optimal CD4+ T cell priming as well as expansion in response to antigens can be facilitated by B cells (172-174). B cell presentation of antigen can lead to T cell tolerance, if BCR-mediated activation does not occur (175, 176).

Finally, B cells release an array of cytokines that can polarize and maintain T cell responses (177). Harris et al. showed that B cells can be induced to differentiate 34 into distinct effector subsets – B effector (Be) 1 cells that produce IFN-γ and Be2 cells that produce IL-4 in vitro (177). These cells in turn, can participate in supporting the differentiation respectively of Th1 and Th2 cells from naïve CD4+ T cells. Confirming the potential of effector B cell subset differentiation in vivo, the authors went on to show that in C57BL/6 mice in response to infection with Toxoplasma gondii (Th1- mediated) and Heligmosomoides polygyrus (Th2-mediated), Be1 and Be2 cells arise.

As well as polarised Be1 and Be2 cells, IL-17-producing B cells have been reported to have a physiological role (178). It has been shown that B cells with a plasmablast phenotype are the source of innate IL-17 in response to Trypanosoma cruzi trans-sialidase (178). Using both murine and human primary B cells, Rawlings and colleagues demonstrated that exposure to parasite-derived trans-sialidase in vitro is sufficient to trigger modification of the cell surface mucin, CD45. Trans-sialidase- induced CD45 modifications, in a BTK-dependent manner, led to IL-17 induction; IL- 17 was then pivotal in neutrophil recruitment (178). B cell specific IL-17-deficiency in mice led to inability to control infection, increased liver damage and reduced survival compared to control animals (178).

More recently B cells have been recognized as potent modulators of immunity, with particular interest in regulatory B cells (Bregs) (179), these cells are discussed in the next section.

35 Regulatory B cells Historical perspective “Regulatory B cells” (Bregs) is a term given collectively to B cells with a number of different phenotypes that through an expanding array of mechanisms act to suppress pathogenic T cell responses and thus support tolerance.

The role of B cells in the homeostasis of T cell functions was first recognized in the 1970s. Studying delayed type hypersensitivity (DTH) reactions, Turk et al. had observed that the administration of cyclophosphamide (300 mg/kg) to guinea pigs, three days prior to epicutaneous sensitization with 2,4-dinitrofluorobenzene (DNFB), increased the intensity and persistence of contact skin reactions compared to control animals (180). They were also able to show that the effect of cyclophosphamide was reversed by the transfer of splenocytes and reproduced by splenectomy of animals; both were partial effects (180). The transfer of sera from pre-sensitized guinea pigs failed to transfer protection to cyclophosphamide treated animals, suggesting that protection was driven by cellular components (180). The histological examination of lymph nodes draining the site of sensitization revealed a depletion of B cells, identified as marked reduction of lymphocytes in follicles, GCs, cortico-medullary junction and medullary cords, as well as persistently reduced antibody responses triggered by passive cutaneous anaphylaxis. This suggested that the modulation of cell-mediated response to antigens required B cells (180).

Katz et al. then used an alternative DTH model, in which sensitization is induced by injecting ovalbumin (OVA) in incomplete Freund’s adjuvant (IFA) into the footpad, to demonstrate that adoptively transferred splenocytes or peritoneal exudate cells from sensitized donors suppressed skin reactions to OVA in recipients (181). If sensitized donor animals had been cyclophosphamide-pre-treated, skin reactions in recipient guinea pigs were more intense and indurated compared to no transfer control animals (181). The reaction was immunologically specific, as sensitization of donors with bovine gamma globulin (BGG) did not suppress skin reactions in recipients. Suppression was mediated by live cells, as heating them, abrogated their suppressive capacity (181). To demonstrate the direct role of B cells in the suppression of contact hypersensitivity (CHS), the authors showed that total splenocytes were able to transfer suppression to cyclophosphamide-treated recipients, when administered one hour prior to skin testing, but B cell-depleted splenocytes lost this ability (182). Nevertheless, Katz et al. failed to implicate antibodies in the mechanism of B cell suppression and did not

36 offer a mechanistic explanation for the protection that these cells provided (182). Concomitantly, experiments by Neta and Salvin also demonstrated the suppression of skin DTH reactions upon transfer of lymphocytes from pre-sensitized animals; and went on to suggest B cells as the mediators of suppression (183).

The early methods employed to study the role of B cells in the modulation of immunity relied on “crude” depletion strategies, such as the administration of anti-IgM in neonatal mice, with either B cell leakage or deleterious effects beyond the B cell compartment identified often as caveats in the interpretation of findings. The understanding of the role that B cells play in the modulation of immunity was therefore greatly facilitated by the generation of B cell deficient µMT mice, through the introduction of targeted gene disruption by nonsense mutation in the exon encoding IgM H chain transmembrane region (184).

The next important suggestion of an immune modulatory role of B cells came in 1996 from Janeway and colleagues in experimental autoimmune encephalomyelitis (EAE) – an animal model for multiple sclerosis (MS) (185). Previous studies had suggested a pathogenic role for B cells in EAE partly through antibody production (186-188), and also via antigen presentation to T cells (188). Using µMT mice, Wolf et al. investigated the role of B cells in EAE induced by acetyl-myelin basic protein (MBP) (1-11) immunization in B10.PL mice (185). They did not observe differences in the onset and overall severity of disease in B cell deficient compared to control animals, but in the absence of B cells, mice did not enter the recovery phase (185). Their interpretation of these results was that B cells acted as APCs and were skewing the T cell response towards a Th2 phenotype – “immune deviation” (185). Using a keyhole limpet haemocyanin (KLH) in CFA immunization model, Moulin et al. also showed, that in B cell deficient mice, dendritic cell (DC) production of IL-12 was enhanced, suggesting that B cells suppress DC activation and, as consequence, inhibit Th1 response (189). Examining the B cell role in oral tolerance, in 2001 Gonnella et al. showed that IL-4, IL-10 and transforming growth factor-β (TGF-β) production in gut associated lymphoid tissue (GALT) following low dose oral antigen administration was induced in wild-type (WT), but not µMT mice (190). Furthermore, in myelin oligodendrocyte glycoprotein (MOG)-induced EAE, B cells were required for the protection provided by oral administration of low doses of MOG antigen (190).

37 It was the publication of the following three initial reports, showing that B cells could exert suppressive effects during immune responses, that challenged the paradigm of B cells as the “perpetrators” of autoimmune pathology.

i) Using TCRα-/- mice, which develop spontaneous intestinal inflammation reminiscent of human ulcerative colitis, Mizoguchi et al. demonstrated that a subset of B cells induced in murine GALT, and characterized by high expression of CD1d, downmodulates inflammation through the production of IL-10 (191). They had previously shown the protective role for B cells in this model, as B cell and TCRα double deficient mice developed colitis at an earlier age and presented more severe disease than TCRα-/- single deficient mice (192, 193). TCRα-/- mice presented an increase in a population of CD1dhi B cells (CD21intCD23hiIgMintCD62Llow) in the mesenteric lymph node (MLN), but not in the spleen (191). Interestingly, these cells expanded during inflammation, as age-matched non-diseased mice had significantly lower expression of CD1d in MLN B cells. Generation of CD1d and TCRα double knockout mice (DKO) showed that, although initially these mice develop similar intestinal inflammation to the control group, at six months of age the DKO mice had exacerbated disease compared to single TCRα-/- mice. MLN B cell adoptive transfer experiments from the DKO and TCRα-/- mice into DKO mice with mild to moderate disease showed that, while B cells from TCRα-/- mice were able to down-regulate inflammation, those from DKO failed to exert suppression. MHC class II deficient B cells were also able to transfer protection, indicating that regulatory properties of B cells in this model did not require antigen presentation to CD4+ T cells. The authors observed that MLN B cells produced IL-10 in TCRα-/- mice upon onset of inflammation, when CD1dhi B cells emerged. Blockade of IL-10 and IL-10R in vivo using neutralizing antibodies abrogated B cell suppressive activity, therefore, B cell-mediated modulation occurred through IL-10. Furthermore, transfer of B cells from IL10-/-×TCRα-/- mice was not able to suppress intestinal inflammation. Addressing whether B cell-derived IL-10 was modulating inflammation through the inhibition of Th2 pathology showed that suppression of IL-4 production by CD4+ T cells did not occur upon protective B cell transfer, instead in the absence of IL-10 production by B cells, diseased mice were unable to downregulate IL-1β induction and STAT3 activation. Thus, Mizoguchi et al. had shown that upon intestinal inflammation a B cell subset is induced in GALT that produces IL-10 and that is characterized by high expression of CD1d; these B cells suppressed IL-1β and STAT3-associated signalling (191).

38 ii) Confirming observations by Janeway and colleagues (185), Fillatreau et al. showed that whereas WT mice with EAE enter remission phase by day 30, µMT mice present sustained inflammation and disease (194). Fillatreau et al. went on to demonstrate that splenocytes, isolated from µMT mice, produced higher levels of IFN-γ in response to autoantigen stimulation, compared to splenocytes from WT mice, suggesting that the mechanism that typically acts to constrain the Th1 response in WT mice may be missing from µMT mice (194). Mechanistically, they showed that B cells, isolated from the spleens of mice during the remission phase of disease, released higher amounts of IL-10 after stimulation with an autoantigen and an agonistic anti-CD40 monoclonal antibody (mAb; FGK-45), compared to B cells stimulated with autoantigen alone. More importantly, IL-10 production was increased in B cells from WT mice in the remission phase, compared to mice in the acute phase of EAE or mice immunized with CFA without MOG. In addition, they were also able to reproduce the effect of the antigen in this stimulation using anti-κ L chain antibodies, suggesting that B cell IL-10 production required BCR stimulation. Of note, the requirement of CD40 in IL-10 induction by B cells had been previously demonstrated (195). To finally prove that B cells producing IL-10 play a pivotal role in the modulation of EAE, the authors generated mixed BM chimeric mice with B cell-restricted deficiency in IL-10 or CD40. Both, the lack of B cells expressing IL-10 or B cells expressing CD40, prevented mice from entering the recovery phase of disease. Splenocytes from groups of chimeric mice that failed to enter remission from EAE had enhanced antigen-specific IFN-γ production, compared to splenocytes from groups that recovered from EAE. Finally, adoptive transfer of splenic B cells isolated from mice that had recovered from EAE, rescued disease remission in IL-10-/- B cell-chimeric mice. This therefore showed that B cells via provision of IL-10 can restrain Th1 responses, and that optimal IL-10 production requires signals through the BCR and CD40 ligation.

iii) At the same time Mauri et al. showed the role of B cells producing IL-10 in the suppression of chronic collagen-induced arthritis (CIA), an animal model of rheumatoid arthritis (RA) (196). This Th1-driven disease was induced by immunization of DBA/1-TCRβ-Tg mice with type II collagen (CII) in CFA (196). The authors were able to induce IL-10-producing B cells in splenocytes, isolated from arthritic mice, using antigen and agonistic anti-CD40 mAb and show that these cells were functionally suppressive in chronic disease upon adoptive transfer (196). B cells induced with this stimulation were able to prevent disease development in the recipient mice. The reduction of disease was mirrored by a decreased Th1 differentiation (196). That the 39 suppressive capacity relied on B cell ability to produce IL-10 was supported by two sets of experiments. Firstly, in vivo neutralization of IL-10 in B cell recipient mice led to abrogation of suppression conferred by B cells. Secondly, B cells isolated from spleens of IL-10 deficient animals, stimulated with collagen and anti-CD40, failed to suppress disease in recipient mice with CIA (196). Following the previously described findings, Carter et al. used mixed BM chimeric mice lacking IL-10 specifically in B cells (B-Il10- /-) to demonstrate the importance of endogenous IL-10 produced by B cells in the suppression of inflammation in arthritis. The authors showed that B-Il10-/- mice developed exacerbated antigen-induced arthritis (AIA) and CIA, compared to control mice, with increased frequencies of Th1 and Th17 cells, and reduced number of FoxP3+ regulatory T cells (Tregs) and expression of FoxP3 (197, 198). The role of B cells in the maintenance of Tregs had previously been suggested in experiments using µMT mice (199, 200). Upon in vitro stimulation with autoantigens, WT B cells formed more and longer-lasting contacts with CD4+CD25- T cells, as measured by the frequency of conjugates and duration of interactions, and were able to facilitate T cell conversion into FoxP3+ cells, in contrast to their IL-10-deficient counterparts. Overall, the report by Carter et al. was the first to demonstrate a role for B cell endogenous IL-10 production in reducing the severity of inflammation.

Since the original observations of IL-10 producing B cells, the field of Bregs has gained increasing interest and their role has become better defined. Thus, in addition, to “dampening” inflammation in the animal models of RA, MS and inflammatory bowel disease (IBD), Bregs have also been shown to play a role in inducing tolerance in transplantation (201) as well as suppressing anti-tumour immunity (202) and to negatively regulate immunity to bacterial and viral pathogens (203).

Breg Phenotype The term Breg is generally used for B cells that produce IL-10 and limit inflammation. The available evidence suggests that, unlike natural Tregs (nTregs), which constitute a separate lineage of T cells with a thymic origin, multiple Breg subsets can arise in the periphery in response to inflammation, and that they can originate from B cells at a number of different developmental stages.

T2-MZP B cells Building on their previous work, showing suppressive activity of anti-CD40- treated splenic B cells through the provision of IL-10 (196), Mauri and colleagues showed that IL-10 production was largely contained in splenic T2-MZP B cells

40 (identified as CD19+CD21hiCD23hiCD24hiIgMhiIgDhiCD1dhiAA4+) and that, while present in naïve mice, IL-10+ T2-MZP B cells were significantly expanded in mice in remission from CIA (204). Upon adoptive transfer, T2-MZP B cells were the only B cell subset able to both suppress development of disease as well as ameliorate established arthritis. Importantly, neither MZ nor Fo B cells suppressed disease upon adoptive transfer. T2-MZP transfer significantly reduced joint damage, CII-specific IgG response, and inhibited IFN-γ production by CD4+ T cells. They also showed that this effect requires IL-10 production as mice that received IL-10-deficient T2-MZPs developed severe disease.

Advancing the hypothesis of the T2-MZP B cells as the critical B cell subset in the regulation of immune responses, Blair et al. observed that in the MRL/lpr mice which spontaneously develop systemic lupus erythematosus (SLE)-like disease, there was an inverse correlation between T2-MZP B cell numbers and progression of disease measured as proteinuria (205). Furthermore, T2-MZP B cells isolated from MRL/lpr mice failed to suppress proteinuria or to increase survival upon transfer to mice with established lupus, suggesting functional impairment of T2-MZP cells in lupus-prone mice. As discussed above, CD40 signals have been shown to be important in the differentiation of Bregs (191, 194, 196) and agonistic anti-CD40 enhances their IL-10 production (194, 196). Blair et al. showed that the lack of IL-10 production in B cells from MRL/lpr mice could be restored upon in vitro stimulation with anti-CD40, but not upon anti-IgM activation. In fact, the latter appeared to suppress anti-CD40-induced IL- 10 production. Intracellular staining revealed that majority of CD40-activated IL-10- producing cells were also contained within T2-MZP B cells. These anti-CD40 treated T2-MZP cells were CD1dhi and expressed CD93, and were referred to as T2-like B cells (due to the in vitro induction). Of interest, unlike CpG and LPS stimulation, which induced both MZ and T2-MZP production of IL-10 (206), anti-CD40 effect appears to specifically induce the expansion of T2-MZP B cells producing IL-10 and to prevent spontaneous B cell maturation in culture.

Unlike MZ or Fo B cells following anti-CD40 stimulation, T2-like B cells were able to suppress IFN-γ and TNF-α production, as well as promote IL-10 production, by CD4+ T cells in in vitro cultures. Furthermore, only T2-like B cells significantly reduced mortality and lowered proteinuria in MRL/lpr mice upon in vivo transfer. This suppressive effect on disease was abrogated by treatment with anti-IL-10/anti-IL-10R antibodies. The authors were also able to recover the suppressive effect of endogenous

41 IL-10 producing T2-MZP B cells, by treating mice with four doses of anti-CD40 over a period of four weeks. The treatment regime enhanced survival, lowered proteinuria and anti-dsDNA IgG levels, as well as reduced skin disease. Administration of anti-CD40 resulted in the expansion of T2-MZP B cells producing IL-10 starting from two weeks following the beginning of treatment. This treatment did not seem to affect the secretion of IL-10, IFN-γ or TNF-α by CD40-expressing non-B cells.

The work by Blair et al. (205) was also amongst the first attempts to decipher where Bregs exert their suppressive effect. To do so, they stimulated T2-MZP B cells, isolated from transgenic MRL/lpr mice expressing a human copy of the CD20 gene (hCD20Tg), for 48 hours with anti-CD40, then transferred them into control MRL/lpr mice, and tracked their maturation and homing in vivo. Whereas, the majority of isotype-control stimulated T2-MZP B cells matured into Fo and MZ B cells, a small yet sizeable proportion of T2-like B cells retained their immature T2-MZP phenotype following transfer. Of interest, none of the transferred T2-like B cells homed to the kidneys, the primary site of inflammation in this disease, but appeared to home back to the spleen, suggesting systemic regulation by T2-like Bregs.

Fallon and colleagues showed that the infection with Schistosoma mansoni provides protection in both acute anaphylaxis and chronic allergen-induced airway hyperresponsiveness (AHR) models, and that this protection requires IL-10 and B cells (207, 208). They then followed this up by identifying the induction of IL-10-producing B cells in the spleen in response to the parasite and showed that these B cells modulated allergic airway inflammation in OVA-sensitized mice upon adoptive transfer (209). They characterized these cells as CD19+CD21hiCD23+IgD+IgMhiCD5+CD1dhi. While the authors reported that 82% of IL-10-producing B cells were CD1d-expressing, only a fraction of these were CD1dhi. Furthermore, they established the phenotypic similarity between these Bregs and T2-MZP described by Mauri and colleagues in the CIA model (204). The regulation by these B cells was IL-10 dependent as Il10-/- CD1dhi B cells were unable to mediate amelioration of AHR upon adoptive transfer. Anti-CD1d treatment also abrogated the protection provided by Schistosoma mansoni infection in OVA-induced airway inflammation. Furthermore, protection did not occur in CD1d-/- mice, but did in Jα18-/- iNKT cell deficient mice. The authors thus concluded that B cell mediated protection in their model was CD1d-dependent, but did not involve iNKT cells. The relevance of CD1d for B cell mediated protection was nevertheless not addressed mechanistically. Amu et al. showed that the transfer of post-infection

42 CD1dhiCD19+ B cells lead to a reduction of IL-4, IL-5, IL-13 and CCL11 (eotaxin) in the bronchoalveolar lavage fluid in the recipient mice. Upon transfer, Bregs induced the pulmonary, but not splenic, accumulation of CD4+CD25+FoxP3+ Tregs, independent of TGF-β (209).

Amu et al. also generated ex vivo Bregs using incubation with live worms, which were IL-10+CD1d+ B cells with comparable surface marker expression phenotype to those induced in vivo (209). Similar to T2-MZP identified by Mauri and colleagues (204), Bregs in this system arose or were enriched post initial challenge, and B cells from naïve mice had greatly diminished suppressive capacity. Both Breg subsets were also able to reverse established disease. However, the fact that helminth-induced Bregs provided protection in OVA-driven inflammation, suggests lack of antigen specificity in these B cells, in contrast to T2-MZP Bregs.

A number of other studies have also shown the role of T2-MZP Bregs in immune modulation. Schioppa et al. showed the role of T2-MZP B cells in TNF-α- mediated promotion of tumour development in a 7,12-dimethybenz-α-anthracene (DMBA)/ terephthalic acid (TPA) model of skin cancer (210). They showed that papilloma development required B cells and TNF-α; the antitumor immunity in TNF-α- deficient mice was associated with increased IFN-γ and CD8+ T cells in the skin. Reduced IL-10+ Bregs and increase in IFN-γ+CD8+ were also observed in the spleen, suggesting that Bregs are important in controlling CD8+ responses. Using a Helicobacter felis infection model, Sayi et al. showed the role of T2-MZP B cells in limiting gastric premalignant pathology (211). These Bregs acted by converting naïve T cells into IL-10-producing CD4+CD25+ T cells, in a CD40-CD40L and CD80-CD28- dependent manner. Nevertheless, B cell IL-10 production was redundant for protection, despite IL-10 production by T2-MZP B cells being increased in response to Helicobacter antigens. In a mouse model of transplant tolerance, T2-MZP B cell transfer from mice rendered tolerant to MHC class I mismatched skin grafts, prolonged graft survival in antigen-specific manner (212). This was associated with suppression of effector T cells, however, was independent of transferred T2-MZP ability to produce IL-10.

B10 cells CD19 is a molecule that acts to positively regulate B cell responses and defines signalling thresholds in B cells; mice expressing human CD19 (hCD19Tg) display hyperresponsiveness to transmembrane signals, enhanced proliferation in response to

43 mitogens and antibody production to TD antigens, as well as spontaneous autoantibody production with aging (213). Unexpectedly, similar to B cell deficient mice, Cd19-/- mice develop a severe non-remitting form of EAE (214) and exacerbated CHS reactions (215). The latter study also demonstrated the ability of splenic total and MZ B cells, but not peritoneal B-1a cells, to normalize CHS reactions in Cd19-/- mice upon adoptive transfer (215).

In 2008 Yanaba et al. identified CD1dhiCD5+ B cells as mediators of suppression in a model of CHS, using 4-ethoxymethylene-2-phenyl-oxazolin-5-one (oxazolone) sensitization. Yanaba et al. used Cd19-/- and human CD19 transgenic (hCD19Tg) mice, which develop exacerbated and ameliorated CHS compared to WT mice, respectively. B cell depletion in hCD19Tg mice resulted in restoration of responsiveness to oxazolone sensitization, establishing a protective role for B cells in this system. The depletion of B cells in WT mice using anti-CD20 antibodies also enhanced the response to oxazolone compared to control antibody treated WT mice. The depletion of B cells in Cd19-/- mice further enhanced CHS severity, albeit not significantly, which shows that residual regulatory B cell capacity remains in these mice. The authors also observed that IL-10 was decreased and increased in Cd19-/- and hCD19Tg and mice, respectively, compared to WT mice. IL-10 expression was inversely proportional to inflammatory T cell responses in these mice. They could, however, detect only very low levels of IL-10 in splenic B cells without stimulation, but this was enhanced by about 10-fold in response to five hour culture with LPS, phorbol 12-myristate 13-acetate (PMA) and ionomycin. The differences in strains were exacerbated by this treatment and observed in the spleen and peritoneal cavity, but not in blood, peripheral and MLNs, and Peyer’s patches.

In this study, most splenic IL-10-producing B cells were CD19hiCD24hiIgMhiCD1dhiCD5+ after stimulation, but also expressed IgD, some CD21, but not CD23. Thus, these splenic B cells share some features with T2-MZP and some with MZ B cells, and were designated “B10 cells” (216). They are able to protect mice from oxazolone-induced hypersensitivity upon adoptive transfer to Cd19-/- or B cell depleted mice. B10 cells were unable to mediate suppression if isolated from naïve or Il10-/- mice, or if non-CD1dhiCD5+ B cells were transferred. When the authors blocked IL-10 one hour prior to oxazolone sensitization, the protection from CHS observed in hCD19Tg, compared to WT mice, was abrogated. The authors did not offer any mechanistic explanation of how suppression is mediated.

44 Yanaba et al. were able to detect IL-10-producing B cells in the circulation, however, they did not observe an Il10 mRNA increase in B cells from draining lymph node (dLN). While there is a possibility that Bregs migrate in low numbers to sites of inflammation, these data suggest that Breg-mediated suppression occurs systemically, as is likely the case with T2-MZP B cells. Furthermore, like with T2-MZP B cells, inflammation appears to play a critical role in B10 induction. Yanaba et al. showed that inflammation increased IL-10 production by spleen B cells, but with the amount of IL- 10 produced being inversely related to inflammation. Antigen specificity was investigated by the transfer of splenic CD1dhiCD5+ B cells purified from the spleens of DNFB-sensitized mice, and these were again unable to mediate suppression upon oxazolone-challenge. These observations together emphasize the similarities between T2-MZP B cells and B10 cells, their mode of suppression, inflammation as stimulus for their induction and requirement for antigen-matching in induction and suppression. Yanaba et al. suggest that B-1a and B10 may be branches of the same lineage rather than related to MZ B cells, since only half of B10 were CD21hi.

B-1 cells As discussed previously, B-1 cells are developmentally different from B-2 cells and are self-renewing. Splenic, but particularly peritoneal B-1a cells have been shown to be an important source of IL-10 (217). In murine schistosomiasis, it was shown that peritoneal, but not splenic, B-1a cells expanded with infection (218). These cells produced copious amounts of IL-10 in vitro in response to the carbohydrate lacto-N- fucopentaose III, a Schistosoma mansoni egg antigen. The authors suggested that “immune deviation” in this case may lead to dominant Th2 response and thus prevent granulomatous pathology. Analysing Xid immunodeficiency on murine filarial infections, Fleischer et al. found that in the absence of B-1a cells mice were more susceptible to infection (219). Susceptibility to murine filariasis was associated with diminished Th2 cytokine production as well as reduced B cell-derived IL-10. CD5+(CD1d+) neonatal B cells have been shown to produce copious amounts of IL-10 in response to TLR9 triggering with CpG and to inhibit Th1 response by suppressing DC production of IL-12 (220).

Other Breg phenotypes It has been shown that MZ B cells produce copious amounts of IL-10 in response to activation with TLR ligands (206). Administration of apoptotic cells a month prior to the onset of CIA, protected mice from severe joint inflammation (221).

45 Suppression of disease was associated with increased IL-10 secretion from CII-specific T cells, and reduced anti-CII autoantibody production. Adoptive transfer of B cells from apoptotic cell-treated mice also provided protection from disease. This was IL-10- dependent as IL-10 neutralization in vivo abrogated the protective effect of apoptotic cells. The suppressive capacity and IL-10 production was attributed to MZ B cells. Brummel and Lenert found that stimulation of B cells, isolated from mice with lupus, with CpG A(D) led to their upregulation of IL-10, CD25 and CD86 (222). However, they also promoted secretion of IgM and IgG3. These cells were of MZ B cell phenotype, and were expanded in lupus compared to control mice.

A suppressive role for T cell Ig and mucin domain (TIM)-1-expressing B cells has been demonstrated. TIM-1 is expressed by most IL-10-producing B cells in all B cell populations, including transitional, MZ, Fo, as well as CD1dhiCD5+ B cells (223). An agonistic anti-TIM-1 antibody was able to enhance allograft tolerance, but only in the presence of B cells. It also induced IL-10 production and increased TIM-1+ Breg frequency in an IL-4-dependent manner. These TIM-1+ B cells were able to directly transfer suppression. Furthermore, a defect in Breg function was shown in TIM-1 mucin-domain mutant mice (224). These mice have a defect in IL-10 production by B cells and aged mice developed spontaneous autoimmunity, with increased T cell IFN-γ production and elevated serum levels of autoantibodies.

Ex vivo treatment of mouse splenocytes with granulocyte-macrophage colony- stimulating factor (GM-CSF) fused to IL-15 (a fusokine designated GIFT15) induced suppressive Bregs with a phenotype reminiscent of T2-MZP B cells (CD21+CD23+CD24+IgM+IgD+IL-10+), but in addition these cells expressed CD138 and lacked the expression of CD19 (225). GIFT15 Bregs were able to suppress EAE, and relied on IL-10 production and expression of MHC class II.

Madan et al. showed that B cell-produced IL-10 plays a non-redundant role during murine cytomegalovirus (MCMV) infection though restraining IFN-γ- producing MCMV-specific CD8+ T cells and inhibiting plasma cell expansion (203). Using an IL- 10 transcriptional reporter mouse, they also showed that in particular a population of IL- 10-producing CD138+ plasma cells arises in the spleen within 48 hours of systemic LPS challenge. More recently, IL-10 production by plasmablasts in the dLN during EAE induction has been reported. These plasmablasts inhibited DC generation of pathogenic T cells (226). Murine deficiency in Prdm1 and Irf4 genes, required for plasmablast generation, led to exacerbated EAE. Of note, suppression in this model was independent

46 of splenic B cells. This observation is in contrast to the majority of studies, where Bregs are shown to reside in the spleen. Furthermore, a role for splenic CD138+ plasma cells in the suppression of immune responses to Salmonella thyphimurium and EAE via IL- 10 and IL-35 has been demonstrated (227). It has nevertheless remained difficult to reconcile the pathogenic role that plasma cells play in antibody production with that of regulation. A study by Maseda et al. found that IL-10-producing B10 cells induce Prdm1 and Irf4 transcription leading to Ig-secreting B cell differentiation (203), suggesting IL-10-producing B cells may be an intermediate stage in plasma cell differentiation.

47 IL-10-independent mechanisms of Breg function Increased understanding of suppression mechanisms employed by immune cells, elucidation of networks of cellular interactions and more sophisticated tools to manipulate them has allowed IL-10-independent modes of immune modulation by B cells to be uncovered (Fig. 1.4). In 2010 Wilson et al. reported that helminth-induced MLN B cells were able to mediate suppression of MOG(35-55)-induced EAE and airway inflammation upon adoptive transfer. This suppression was maintained when IL- 10-deficient B cells were used (228). However, this study never identified the B cell- expressed mediator of suppression.

The first alternative mediators in B cell suppression identified were TGF-β and FasL (229). While not addressing regulation by endogenous B cells, the study by Tian et al. precedes the identification of IL-10-producing B cells as suppressors. The authors reported that LPS-stimulated B cells inhibited pathogenic Th1 cell-driven autoimmunity upon transfer to pre-diabetic non-obese diabetic (NOD) mice, which develop spontaneous diabetes with similarity to human autoimmune diabetes. This study nevertheless did not establish IL-10-independence. FasL-expressing B cell role in suppressing diabetogenic T cells upon adoptive transfer to NOD mice was also shown subsequently using CpG as stimulant (230). These cells shared phenotypic characteristics with pro-B cells and emerged within the bone marrow. The authors also showed that this suppression was preserved when blocking IL-10R on effector T cells.

The role of FasL-expressing B cells in suppressing male-to-female rejection in a skin graft model upon adoptive transfer has been demonstrated (231). Transfer of B cells lacking FasL (from gld mice) accelerated the rejection. FasL was also upregulated on B-1a cells in cockroach allergen-induced airway inflammation and acted to induce apoptosis of effector T cells (232). Subsequently the authors showed that in a transgenic model of CIA, where nearly all CD4+ T cells recognize a single peptide of CII, lower numbers of FasL+CD5+ B cells correlated with increased disease severity, associated with reduced T cell apoptosis (233).

B cell TGF-β production, as a mediator of suppression, was identified in studies of transplant tolerance. Anti-CD45RB mAb treatment induced murine allogeneic tolerance (234), with Tregs being the proposed facilitators of successful engraftment (235). Nevertheless, transplant tolerance was demonstrated to rely on B lymphocytes a year later by the same group (236). The anti-CD45RB mAb treatment induced ICAM-1 upregulation on B cells, and blocking this interaction resulted in abolition of B cell

48 suppressive effect (237). Transplantation tolerance upon anti-CD45R treatment was found to be IL-10-independent, in fact it seemed to play a counter-regulatory role, with chronic allograft vasculopathy and anti-graft antibodies reduced upon IL-10 neutralisation (238). The observations that both Bregs and Tregs play roles in graft tolerance were consolidated in a study by Lee et al. (239). The authors showed that upon adoptive transfer, Breg mediated suppression required Tregs for tolerance establishment and acted through increasing the number of these cells. Bregs expressed latency-associated peptide, a membrane-bound form of TGF-β1, and neutralization of TGF-β abrogated transferred Breg-mediated graft prolongation.

Glucocorticoid-induced TNF receptor ligand (GITRL) was identified to be important in B cell maintenance of the FoxP3+CD4+ T cell population (240). The authors observed a peripheral reduction of Tregs in the absence of B cells, achieved genetically or through antibody-mediated depletion. They showed that Treg numbers could be restored upon B cell transfer, and that B cell expression of GITRL, and not IL- 10 production was important for this effect.

The PD-1- programmed death ligand-1 (PD-L1) axis has also been implicated in the regulatory function of B cells. It was reported that oestrogen treatment in mice ameliorated EAE (241). This protection was abrogated in µMT mice and in WT mice associated with upregulation of PD-L1 on B cells and expansion of IL-10-producing CD1dhiCD5+ B cells (242). PD-L1-deficient mice failed to respond to oestrogen treatment. It was shown that PD-L1 leads to the up-regulation of PD-1 on FoxP3+CD4+ Tregs, potentiating their activity (243). Finally, B cell specific requirement for PD-L1 was shown (244). In 2015 Khan et al. published a study showing a role for PD-L1hi B cells in the regulation of humoral immunity through inhibition of PD-1+CXCR5+CD4+ Tfh cells (245). They also demonstrated the ability of the PD-L1hi B cells to suppress EAE upon transfer. Furthermore, as PD-1 ligation on human B cells inhibits their activation (246), it is therefore possible that PD-L1hi B cells could inhibit the activation of other B cells.

IL-35 was shown to be important in the negative regulation of immune responses both in Salmonella typhimurium infection and EAE (227). The main producers of IL-35 were identified to be IL-10- CD138+IgM+TACI+CXCR4+CD1dintTim1int plasma cells that expressed Blimp1. However, IL-35 as a more general mechanism of Breg-mediated suppression is disputed by an earlier study in AIA (247). In this study, mice deficient in the p35 subunit of IL-

49 35 developed milder disease compared to WT mice, through a mechanism that relied on negative regulation by IL-27.

Antibodies are overwhelmingly associated with enhanced immunity. Nevertheless, IgG molecules with catalytic activity are thought to be protective through the removal of toxic wastes. In sepsis, Lacroix-Desmazes et al. have showed that plasma IgG with serine protease-like activity was correlated with survival from sepsis (248). The IgG from three surviving patients hydrolysed factor VIII and from one also factor IX, suggesting that the protective activity may be via the inhibition of microvascular thrombosis. It was also shown than in renal transplanted patients the coagulation factor hydrolysis by circulating IgG was associated with reduced risk for chronic allograft nephropathy (249).

Granzyme-B expressing B cells have been found to be enriched in tumours (250). These cells were expanded by IL-21 and limited T cell proliferation through degradation of TCRζ chain. These cells were CD19+CD38+CD1d+IgM+CD147+ and expressed IL-10, CD25 and indolamine-2,3-dioxygenase.

Finally, B cells can also contribute to immune regulation by nucleotide catabolism. CD39 and CD73 ecto-enzymes, which together act to generate adenosine from extracellular nucleotides, have been shown to be expressed by B cells (251). In particular B-1 and B10 cells express high frequency of CD73. CD73+ B cell transfer is able to suppress disease in a dextran sulphate sodium-induced colitis model. Of interest, IL-10-deficient B cells have reduced expression of CD73, linking the two mechanisms of suppression.

50 Lipid

BCR CD1d IL-10 Granzyme B GITRL

FasL TGF-β Breg IL-35 CD73

PD-L1 CD39 Catalytic Ig G Figure 1.4. Breg mechanisms. The expanding array of B cell suppression mechanisms includes IL-10, CD1d, FasL, TGF-β, GITRL, PD-L1, FasL, CD39/CD73, granzyme B, IL-35 and catalytic IgG.

51 Signals in Breg induction IL-10 is the major effector of Breg suppression, and efforts to understand signals involved in Breg generation have focused on IL-10-producing subsets. Interestingly, the signals that have been discussed when describing the advances in the study of Bregs, including CD40, TLR and BCR ligation and inflammation, are all crucial components of a successful immune response. It is conceivable that more than one pathway can stimulate, including synergistically, Breg function. The induction of regulatory function in B cells thus likely depends on context and relies on the integration of multiple signals.

CD40 CD40 is expressed on APCs, including B cells and DCs, as well as on non- hematopoietic fibroblasts and endothelial cells (252). The interaction between CD40 and its ligand CD154 leads to activation of such cell functions as cytokine and chemokine production, as well as the upregulation of co-stimulatory molecules. The in vivo disruption of CD40-CD154 interaction on APCs and T cells, respectively, leads to diminished T cell priming and reduced differentiation of effector functions. In B cell biology, as previously described, CD40 plays a critical role in proliferation and class- switching, while in DCs CD40 ligation leads to superior induction of IL-12 compared to microbial molecules, including LPS, or TNF-α (253). Regarding the significance of CD40 signal dynamics in B cell activation, weaker signals through CD40 have been shown to drive plasma cell differentiation while during stronger stimulation, when ligand is abundant, it leads to the inhibition of antibody secretion in vitro (254).

The first report on the important role of CD40, expressed on B cells, in regulating autoimmune and inflammatory processes came from Mauri et al. (195). The authors demonstrated that administration of an agonistic anti-CD40 antibody, previously shown to induce protective immunity, had therapeutic effects in both chronic and self-limiting CIA in DBA/1-TCRβ and DBA/1 mice, respectively. Protection was associated with production of IL-10 by splenocytes and inhibition of Th1 responses. As discussed above, in both CIA and EAE, signals from CD40 were required for the induction of IL-10 by B cells (194, 196). CD40 stimulation also potentiates the suppressive capacity of B cells, including attenuation of lupus in MRL/lpr mice with in vivo anti-CD40 treatment or through transfer of anti-CD40-activated T2-like B cells (205). B cell specific lack of CD40 leads to mice developing non-remitting EAE (194). TCRα-/-×µMT mice develop a severe form of spontaneous intestinal inflammation

52 observed in TCRα-/- mice (192, 193), as mentioned above. The adoptive transfer of B cells led to the amelioration of colitis, but not if these cells were pre-incubated with anti-CD40 blocking antibodies (193).

TLRs and inflammation Although signals through the BCR and CD40 are critical for the induction of functionally regulatory B cells and their IL-10 production, these are not sufficient to induce IL-10 from naïve B cells. As inflammation commences in response to a pathogen or tissue injury, it also induces regulatory cascades to re-establish homeostasis and initiate wound healing and tissue repair; this is a critical aspect of immune tolerance. TLRs are pattern-recognition receptors (PRRs) that respond to signals generated by infection or tissue damage. While needed to initiate inflammation and mount an efficient immune response, certain TLRs have been observed to promote the suppression of immune responses, as described above. Of note, murine B cells express TLRs 1-9 (206). When stimulated with certain sets of agonists, TLRs promote B cell cytokine production, and TLR agonists are potent inducers of IL-10 from naïve B cells. Among these, LPS from gram-negative bacteria, peptidoglycan from gram-positive bacteria and CpG-containing oligonucleotides that mimic bacterial DNA, have been studied and shown to potently induce IL-10 production by B cells. Myeloid differentiation primary-response gene 88 (MyD88) is an adaptor molecule and principal mediator of signalling downstream of TLRs. While global MyD88-deficient mice are resistant to EAE, mice with a B cell specific deletion of TLR2 and TLR4 or MyD88 develop chronic EAE (255). Interestingly, in B cell MyD88-deficient mice the GC response, as well as specific antibody production, was maintained. MyD88 is also required to generate IL-10-producing plasma cell-like B cells in response to Salmonella typhimurium infection (256). LPS administration in vivo has been shown to modulate EAE, with significant delay in disease onset compared to control mice (257). The role of LPS in the in vitro induction of Bregs that mediate suppression of CHS (216) and pancreatic inflammation in NOD mice (229), has also been shown, as discussed above.

The role of TLR9 in the induction of Bregs has been more difficult to discern. Of TLR9 ligands type A(D) oligonucleotides strongly activate DCs and macrophages, while B(K) oligonucleotides activate B cells (258). CpG is a potent inducer of B cell production of IL-10 (206). Nevertheless, Lampropoulou et al. observed that B cell TLR9-deficient mice recovered from EAE similarly to WT mice (255). In lupus-prone MRL/lpr mice, opposing roles for DNA- and RNA-sensing TLR9 and TLR7,

53 respectively, have been proposed. Tlr9-/- MRL/lpr mice developed exacerbated inflammation, while disease was milder in Tlr7-/- mice (259). TLR9 polymorphisms causing decreased expression of TLR9 are associated with increased risk of SLE in humans (260). TLR9 agonist administration has also been reported to ameliorate disease in NOD mice (261), and inflammatory arthritis in K/BxN serum transfer model (262). In the latter study, however, the authors identified CD8+ DC crosstalk with natural killer (NK) cells that resulted in reduced neutrophil recruitment to the joint as the mechanism of suppression. B cell TLR2 activation by Helicobacter felis led to IL-10 production and was associated with reduced Helicobacter–induced gastric immunopathology (211). Finally, IL-10 production driven by TLR activation is optimal after CD40 engagement, suggesting that Breg activation may be more potent after their synergistic interaction with T cells and contact with pathogen-associated molecules (255, 263). A role for sterile inflammation in Breg induction was shown by Gray et al., who reported apoptotic cell induction of IL-10 and amelioration of CIA (221). Apoptotic cell-derived DNA complex stimulation of TLR9 to induce B cell IL-10 production was subsequently demonstrated (264).

It has been shown that inflammatory “danger” or “damage” signals, apart from those inducing TLR activation, can drive Breg development. Zhivaki et al. showed that respiratory syncytial virus (RSV), upon interaction with both BCR and CX3CR1, infects and induces neonate-specific Bregs that produce IL-10, which in turn downmodulates Th1 responses (265).

Suggesting a role for inflammatory environment in Breg potentiation, remission T2-MZP B cells were more potent in mediating suppression of CIA than their counterparts from naïve mice, but the effect was at least partially overcome when more T2-MZP B cells were transferred (204). Similarly, it has been shown that B cells, isolated on day 14 post-immunization to induce EAE, do not mediate suppression of EAE, unlike their counterparts isolated from mice in the recovery phase (194). IL-10 producing Bregs reported by Mizoguchi et al. also only expanded with inflammation, as age-matched non-diseased mice had significantly lower expression of CD1d in MLN B cells (191).

Further elucidation of inflammatory signals that can drive Breg differentiation was provided by Rosser et al. (266). The authors showed that microbiota-driven IL-1β and IL-6 signals are required for the induction of IL-10-producing T2-MZP Bregs in AIA. Similarly, Menon et al. demonstrated that human IL-10-producing Breg

54 differentiation is driven by IFN-α and CD40L signals from plasmacytoid DCs (pDCs); this Breg induction by pDCs was defective in patients with SLE and instead led to plasmablast differentiation (267).

Earlier, it was shown that IL-10 production could be induced by IL-21 (268). IL- 35 and IL-10 have also been proposed to induce B cell IL-10 in an autocrine manner (269). Administration of IL-33, an IL-1 family member, to WT mice resulted in the induction of IL-10 producing B cells and protection in IBD, while administration to Il10-/- mice led to exacerbated disease (270). Protection was provided by CD19+CD25+CD1dhiIgMhiCD5-CD23-Tim-1- B cells (270).

BCR and antigen-specificity Signals through the BCR may play a role in the induction of Bregs. In some of the studies described above, Breg suppression was dependent on “priming” and, in general, Bregs induced in response to one antigen were not “cross-suppressive” to another.

Mauri et al. showed that optimal IL-10 induction by B cells with anti-CD40 in the CIA model, required the presence of CII (196). B cells from the anti-CD40 stimulated splenocytes also lost ability to suppress clinical disease upon adoptive transfer to mice with CIA. Furthermore, they showed an approximately five-fold increase in IL-10 production if anti-κ L chain antibody was used instead of the antigen, again supporting a role for antigen recognition via BCR in Breg induction. Evans et al. similarly showed that in CIA, T2-MZP B cell suppression required the cells to be primed with the same antigen as the response they were suppressing (i.e. CII-primed T2-MZP B cells were unable to inhibit in vivo inflammatory response to OVA upon adoptive transfer) (204).

The role for antigen specificity in endogenous Breg function has also been demonstrated by Fillatreau et al. (194). The authors showed that recovery from EAE is dependent on the presence of relevant antigen-recognising B cells (not induced in MD4- transgenic model, in which all B cells recognize the same antigen). Furthermore, mice deficient in CD19, the receptor that potentiates BCR signalling, had exacerbated EAE compared to control animals, with reduced IL-10 production by B cells following MOG stimulation (214). Further evidence for BCR-derived signals in B cell-mediated suppression came from a study investigating the role of p110δ phosphoinositide 3- kinase (PI3K) subunit, which is critical in BCR downstream signalling (271, 272). B cell specific deficiency in this PI3K subunit resulted in spontaneous IBD (272). 55 Furthermore, p110δ inactivation resulted in reduced efficiency of conjugate formation with antigen-specific T cells (271).

56 Bregs in humans IL-10 producing B cells have been identified in humans, importantly in the context of autoimmunity, but also in infection and tumours (273). Breg defects, both numerical and functional, have been reported in autoimmune conditions, including, SLE (274, 275), RA (276), MS (277) and psoriasis (278).

Suppressive capacity of B cells was first observed in MS, where helminth- infected patients displayed increased frequency of IL-10-producing CD19+CD1dhi B cells and had better clinical outcome (279). The B cells from helminth-infected patients also suppressed the proliferation and IFN-γ production of MOG- and MBP-specific T cell lines in an IL-10 dependent manner, unlike those from Trypanosoma cruzi- and Paracoccidioides brasiliensis-infected subjects.

The existence of a human B cell subset with regulatory capacity analogous to T2-MZP identified in mice was reported by Blair et al. (274). These Bregs are characterized by CD19+CD24hiCD38hi (CD1dhiCD5hi) surface phenotype, enrichment for IL-10-expression following culture with anti-CD40, and display functional impairment in patients with SLE. Immature B cells regulate IFN-γ and TNF-α production by conventional T cells in vitro. The suppression is IL-10-, CD80- and CD86-dependent, but does not require TGF-β. The requirement for contact through CD80 and CD86 is reminiscent of experiments reported by Mizoguchi et al. (193) described above, and suggests synergy between IL-10 and CD80/86 signals in Breg- mediated suppression. Like murine counterparts isolated from MRL/lpr mice, immature B cells from SLE patients did not produce IL-10 in response to anti-CD40 stimulation, unlike those from healthy controls and were unable to suppress IFN-γ and TNF-α production by CD4+ T cells. Further elucidating suppressive mechanisms of CD19+CD24hiCD38hi B cells, immature B cells from healthy individuals were also shown to inhibit naïve CD4+ T cell differentiation towards Th1 and Th17 phenotypes and promote their differentiation into Tregs (276). Bregs from RA patients lost the ability to inhibit Th17 conversion and promote Treg induction. The authors of this study also reported the reduction of regulatory transitional CD19+CD24hiCD38hi B cells in active RA, compared to healthy individuals and patients with inactive disease, suggesting that Bregs may be unable to prevent the initiation of an autoreactive response, leading to autoimmunity. A rare population of B10 cells has also been described in human peripheral blood (280). These cells were phenotypically mostly CD19+CD24+CD27+ and regulated monocyte inflammatory cytokine production in an

57 IL-10-independent manner in vitro and were generally expanded in patients with autoimmune disease.

Human Bregs have now also been described in infection. Bregs with CD19+CD27hiCD38hi phenotype are increased in chronic HBV infection, and their frequency was found to correlate temporarily with flares (281). In HIV-1 infection, increase in IL-10-producing CD19+CD38+ (282) and TIM-1+CD19+ (283) B cells correlated with increased viral load. Unlike in autoimmunity, where Bregs suppress Th1/17 responses, in infection they inhibit CD8+ T cells. Finally, tumour-enriched granzyme B+CD19+CD38+CD1d+IgM+CD147+ B cells have been described (250) as discussed above.

58 iNKT cells NKT cells are a unique population of CD1d-restricted, lipid-reactive αβ T cells that bridge innate and adaptive immunity, and as such have been attributed with important immunomodulatory roles in both homeostasis and immune activation, including in autoimmunity (284), allergy (285), infection (286, 287), and cancer (288). The term ‘NKT cells’ was first used by Makino et al. to refer to a unique population of murine αβ T lymphocytes, enriched for Vα14 usage, co-expressing NK1.1 (289); the co-expression of the αβ TCR and NK1.1 among CD4-CD8- double negative thymocytes in adult murine bone marrow had already been shown earlier by Sykes (290). Unlike conventional αβ T cells that recognize peptides presented in the context of MHC molecules, NKT cells recognize lipids presented in the context of the non-polymorphic MHC class I-like molecule CD1d. It was only subsequent to identification of T cells with restricted αβ usage that NKT cell CD1d restriction was established (291). Characterizing NKT cells as CD1d restricted has helped to resolve the contention caused by classification based on the co-expression of an αβ TCR and NK cell markers. Many mouse strains, including BALB/c and 129, do not express NK1.1, and other T lymphocytes also possess NK cell characteristics, such as the MR1-restricted mucosal- associated invariant T (MAIT) cells (Vα19Jα33 in mice, Vα7.2Jα33 in humans), enriched in the lamina propria of the intestine (292), or γδ T cells. Of note, while γδ- TCR expressing cells are generally not CD1d-restricted, human γδ T cells that recognize lipid-loaded CD1d have been described (293). Furthermore, conventional T cells can upregulate NK markers, including NK1.1 (CD161) upon activation (294). The majority of NKT cells express a semi-invariant antigen receptor (295). It is composed of Vα14Jα18 paired with Vβ2, Vβ7, Vβ8.1, Vb8.2 and Vβ8.3 in mice. An analogous TCR repertoire has been described in human iNKT cells where Vα24-Jα18 is paired almost exclusively with Vβ11. These cells are therefore referred to as “invariant” NKT cells, also “classical” and “type I” NKT cells; approximately 80-90% of NKT cells in mice have this semi-invariant TCR (289). Of interest, this cell subset displays remarkable conservation in phenotype and function in mammals.

NKT cells with more diverse αβ TCRs are referred to as “type II” or “diverse” NKT cells; these represent a minority NKT cell population and are less studied than iNKT cells. This is in part due to the inability to readily identify these cells as discussed in more detail below. Although type II NKT cells express more diverse TCRs, recurrent usage of certain TCR genes is reported, including Vα3.2-Jα9 in mice (296, 297).

59 Furthermore, type II iNKT cell recognition of CD1d-bound lipids is different to that of iNKT cells (298, 299).

iNKT cells are thymically imprinted for rapid effector function in the periphery. They do not require co-stimulation upon TCR engagement for activation and produce copious amounts of cytokines following TCR engagement (300). This has been attributed to iNKT cell constitutive expression of cytokine mRNA (300). Through the rapid expression of different cytokines they have been shown to modulate the evolving innate and adaptive immune responses. Equipped with an antigen receptor generated through somatic recombination and poised for immediate effector function, iNKT cells were referred to as “hybrid” cells that bridge innate and adaptive immunity by Brennan et al. (295). Further support for iNKT cell “hybrid” nature comes from a report demonstrating the extensive sharing of transcriptional signature between iNKT cells and NK cells, as well as with conventional T cells (301). Taken together, these considerations have led to classification of these cells as innate lymphocytes, along with other cells including B-1 and MZ B cells, and γδ T cells.

iNKT cells in the mouse are found in the highest frequency in the liver, where they account for 20-50% of total T cells. However, they are also present in the thymus, bone marrow, peripheral blood and spleen, and at smaller frequencies in peripheral lymph nodes and other tissues. Of interest, recently it has been reported that they reside in the adipose tissue where they constitute a large proportion of T lymphocytes (291). In humans, the tissue distribution of iNKT cells is much less studied. iNKT cells account for around 0.1% of lymphocytes in peripheral blood (302-304) and they are less abundantly present in the liver than in mice (305, 306). Very little is known regarding the location of NKT cells in tissues mostly due to difficulty to access these sites. iNKT cell ontogeny iNKT cell development diverges from conventional T cell development at the CD4+CD8+ double positive (DP) cortical thymocyte stage; it relies on the stochastic generation of the semi-invariant TCR. That the generation of the iTCR is critical to iNKT cell development is evidenced by the complete absence of iNKT cells in mice that lack the subunits RAG-1/2 or Jα18 gene segment (307). Analogous to BCR generation in B cells, iNKT cell receptor is generated by V-J recombination of the α chain and V-D-J recombination of the β chain. While stochastic, there is greater probability of generating V-J recombinations using 5’ proximal segments that rely on

60 fewer excision events (308); therefore, the probability of generating iNKT cells bearing Vα14Jα18 TCRs is greater than that of generating T cells with upstream segment usage.

iNKT cells are selected in the thymus through interactions with CD4+CD8+ double positive cortical thymocytes. CD1d is critical in this process as CD1d deficient mice lack NKT cells (309). iNKT cell dependence on CD1d expression by hematopoietic lineage cells for selection was first described by Bendelac (310), and later confirmed by others (311). While the identity of the antigen recognized by iNKT cells in their development (in positive selection) is unclear, a study by Facciotti et al. has proposed a role for peroxisome-derived lipids in the selection process of iNKT cells (312).

Cortical thymocytes, and not cortical epithelial cells provide unique signals that play an important role in iNKT cell development. In this context, SLAM family proteins through their interaction with SLAM adaptor protein (SAP), encoded by Sh2d1a in mice and SH2D1A in humans, have been shown to drive iNKT cell development. SLAM family members are SLAM (SLAMF1/CD150), CD48 (BCM1), Ly9 (CD229/SLAMF3), 2B4 (CD244), CD84 (SLAMF5), SLAMF6 (Ly108/NTB-A), SLAMF7 (CRACC/CS1/CD319), SLAMF8 (SBBI42) and SLAMF9 (Cd2f10/CD84- H1), most of which form trans homotypic interactions. Evidence supporting the role of SLAM in iNKT cell development came from data showing defective iNKT cell development in mice lacking SLAM (SLAMF1/CD150) and SLAMF6 (Ly108/NTB-A) (313).

iNKT cell development is defective in patients with SAP-deficient X-linked lymphoproliferative disease (314-316) and in SAP-deficient mice (314-316). SAP has a Src homology 2 (SH2) domain and can be recruited to phosphorylated tyrosine residues of several SLAM family members. Upon SLAM binding, Src2 recruits Fyn. There is partial redundancy between the Src tyrosine Fyn and Lck (associated with CD4 and CD8 co-receptors). A similar impairment in iNKT cell development to that observed in SLAM protein-deficient mice, was also observed with Fyn or Lck deficiency (317, 318). Other signalling molecules, implicated in iNKT cell development are BTK and IL-2-inducible T cell kinase (ITK) downstream of Fyn and Lck, respectively. Activation of these leads to the of phospholipase C-γ, subsequent conversion of phostidylinositol-4,5-bisphosphate into diacylglycerol and inositol-1,4,5-triphosphate. These hydrolysis products then activate the NF-κB and calcineurin pathways, correspondingly. Studies have also suggested mechanisms by

61 which SLAM modulation of TCR signals may be achieved. DOCK2 is a downstream target of Fyn, and DOCK2 deficient mice have a marked reduction in iNKT cells in a cell-autonomous manner (319). Thus together these results show that Slamf1 and Slamf6 function via SAP and Fyn to induce critical signals for iNKT cell development.

As these signalling events are also present in the activation and survival of T cells, an important question is how the specific iNKT cell phenotype is achieved. Promyelocytic leukaemia zinc finger (PLZF), a broad-complex, tramtrack, and bric-a- brac zinc finger (BTB-ZF) transcription factor, encoded by Zbtb16, has been proposed to be the master regulator of the development of iNKT cell phenotype (320, 321). PLZF, while expressed by all iNKT cells, is not an iNKT cell specific factor; indeed, it is present in other nonconventional T cells, including human MAIT cells and a subset of γδ T cells as well as in ILC progenitors (322). PLZF is induced immediately after positive iNKT cell selection. Egr1 and Egr2 downstream of TCR signalling direct the expression of PLZF (323). The requirement for intra-thymic PLZF expression is evidenced by the fact that PLZF-deficient mice have very few iNKT cells (324). Moreover, the residual iNKT cells are defective with respect to development of effector functions, including cytokine production, activation marker expression, and trafficking.

T-bet has been also found to be a master regulator of iNKT cell development. iNKT cells in T-bet deficient mice failed to mature and were present in decreased numbers in the periphery. It was subsequently shown to be in control of many of the phenotypic attributes characteristic of iNKT cells, including inability to express NK1.1 and failure to produce IFN-γ (325, 326). Negative selection is thought to act on the iNKT cell compartment. In support of this, the overexpression of CD1d results in reduced iNKT cells numbers and the remaining iNKT cells are biased towards Vβ2 usage, which exhibits lower affinity for CD1d-α-galactosylceramide (α-GalCer) (327). Similarly, supporting negative selection acting on iNKT cells, perinatal in vivo α- GalCer treatment, results in significantly reduced frequencies of iNKT cells (328). Following selection events, iNKT cells undergo maturation that proceeds in stages from 0 to 3, as follows. iNKT cells that are first detectable are CD1d- tetramer+CD24hiCD69+CD44loNK1.1lo (stage 0) and these are thought to arise immediately following positive selection (329). These iNKT cells that are first detectable are CD4+ and mostly CD8- and can subsequently become DN; they are also a very rare and non-proliferating population. iNKT cells then downregulate CD24 and sequentially become CD24loCD44loNK1.1lo (stage 1), CD24loCD44hiNK1.1lo (stage 2)

62 and CD24loCD44hiNK1.1+ (stage 3) (330, 331). Several cell surface molecules, particularly DX5 (a CD49b epitope), have been used to study iNKT cell thymic development. Stetson et al. have described that the acquisition of DX5 expression between stages 2 and 3, coincides with the transcription of Il4 and Ifng (300). It has been reported that iNKT cells at stages 1 together with 2, and 3 have different ability to produce cytokines, with bias towards IL-4, IL-10 and IFN-γ production, respectively (330, 332). While stages 0,1 and 2 are believed to be thymus-dependent, most iNKT cells will exit to the periphery before completing stage 3 of their maturation (329, 331). iNKT cells acquire some activation markers and effector memory characteristics during their thymic development; unlike developing conventional T cells, iNKT cells also undergo rounds of division intrathymically. A report by Wei et al. suggests that the selection and maturation (proliferation and acquisition of effector phenotype, including cytokine production) of iNKT cells requires solely CD1d expression by cortical thymocytes and is largely independent of other antigen presenting cells (333). The transition to stage 3 is CD1d dependent, nevertheless iNKT cell survival and expansion in the periphery of lymphopenic mice (334) and upon adoptive transfer to CD1d deficient hosts (335) suggests independence of peripheral expression of CD1d.

Emigration of iNKT cells from the thymus has been demonstrated to depend on iNKT cell sphingosine 1-phosphate (S1P) receptor interactions with S1P, expressed on lymphocytes and endothelial cells (336). It has also been suggested that LTβR- expressing stromal cells, through a mechanism that is not completely understood, regulate iNKT cell emergence from the thymus (337). Of note, the latter is not a requirement for conventional T cell emigration (337). iNKT cell ligands iNKT cell ligands identified so far fall into two categories based on either a sphingosine or backbone, some of the important ligands for studying iNKT cell function are shown in Figure 1.5.

α-GalCer and synthetic analogues In 1997 Kawano et al. identified the first iNKT cell agonist glycosphingolipid KRN7000 with α-GalCer structure (the two terms are now used interchangeably) (338). α-GalCer has an α-linked carbohydrate moiety attached to sphingosine and chains. iNKT cell stimulation by α-GalCer-CD1d complexes leads to rapid and robust cytokine production by iNKT cells, among which IFN-γ and IL-4 are prominent. These cytokines then act to activate other cell types and have pleiotropic effects on the

63 resulting immune response. The antitumor potential of α-GalCer compounds was first recognized in B16 melanoma and EL4 lymphoma models (339). Since then a series of studies have also shown the immunomodulatory function of α-GalCer and its potent immunosuppressive activity in various models of autoimmune disease (284).

Although ligands of mammalian origin have largely failed to reproduce this potent iNKT cell-activating potential, a number of α-GalCer analogues have been synthesised. The structural modifications in these have been demonstrated to induce different iNKT cell functions. Among the most studied ligands are OCH (with a truncated sphingosine chain) and α-C-GalCer (with a carbon- rather than oxygen-based glycosidic linkage). OCH elicits a dominant Th2 (IL-4) response from iNKT cells upon in vivo challenge (340). α-C-GalCer stimulates murine iNKT cells and preferentially induces a Th1 (IFN-γ) response (341-343), as well as providing superior protection against malaria and melanoma metastases (341, 343). However, it was found to only weakly stimulate human iNKT cells in vitro. A truncated non-isosteric α-C-GalCer analogue was synthesised and found to induce cytokine production with the highest IFN-g:IL-4 and IFN-g:IL-13 ratios in an in vitro assay using human iNKT cells, compared to α-C-GalCer and α-GalCer (344). When the stimulatory activities to human iNKT cells of various C-glycoside analogues were investigated in in vitro screening assays, it was found that almost all ligands with E-alkene linker between carbohydrate and lipid moieties were stimulatory to iNKT cells and while still inferior to α-GalCer in ligand strength, could potentially be used to induce a Th1 response (345).

Importantly, iNKT cell semi-invariant TCR allows for the unambiguous detection of these cells using α-GalCer-analogue PBS-57-loaded tetrameric complexes of CD1d. Of note, type II iNKT cells do not recognize α-GalCer.

Exogenous ligands iNKT cells also recognize CD1d-binding microbial ligands and therefore have a role during microbial infections (286, 287). Examples of microbial ligands include Sphingomonas spp. α-galacturonosylceramides and α-glucuronosylceramides (346- 348), Borrelia burgdorferi α-galactosyldiacylglycerols (349), Streptococcus pneumoniae and group B Streptococcus α-glucosyldiacylglycerols (350), Helicobacter pylori glycolipids (351), and mycobacterial phosphatidylinositol mannosides (352). Apart from the latter, which are phospholipids, the rest have a sugar moiety in α-linkage to a ceramide or diacyglycerol. Sphingomonas is a gram-negative, LPS-free non-

64 pathogenic α-Proteobacteria the glycolipids from which were shown to directly activate iNKT cells and to control septic shock and bacterial clearance (346).

Endogenous ligands The identity of the endogenous glycolipid ligands that are naturally recognized by iNKT cells and trigger their activation has been the subject of extensive research. Nevertheless, despite continued efforts for over a decade, these remain poorly understood, with relatively few antigens that are recognized by iNKT cells defined that have withstood the “test” of advancing technology.

A caveat to endogenous ligand identification is that α-GalCer and the vast majority of other iNKT-stimulating ligands are α-anomeric with respect to the carbohydrate moiety. α-linkage is thought to be an important prerequisite for stimulatory capacity and in general is thought to represent a microbial signature, and α- GalCer to be a microbial mimetic (349), Brennan et al. have referred to it as the “motif for antigenicity” (295). Glycosylceramides in mammals are thought to only exist as β- anomers. The only two mammalian glycosylceramide synthases are glucosylceramide synthase (GCS) and ceramide galactosyl transferase (CGT)/galactosylceramide synthase, which are inverting glycosyltransferases. This means that they transfer α- glucose and α-galactose from uridine diphosphate (UDP)-sugar moieties onto ceramide in a β linkage. This has been at the basis of the assumption that mammals were not able to synthesise α-linked ceramides, and therefore that these ceramides were not part of the mammalian lipidome.

As the closest mammalian structural counterparts to α-GalCer, mammalian β- linked glucose- and galactose-based glycosphingolipids have been tested for their iNKT cell stimulating potential. It was found that synthetic β-GalCer with a non-mammalian fatty-acid tail could stimulate iNKT cells (353, 354). Nevertheless, murine deficiency in β-linked galactose lipids does not result in iNKT cell abnormalities (355). In regard to β-GlcCer, cell lines deficient in β-glucose-based lipids were unable to stimulate iNKT cells (355), conversely when injected into mice some β-GlcCer species potently stimulated iNKT cells (356). The disialoganglioside GD3, a β-glucosyl glycosphingolipid, has also been shown to stimulate some iNKT cells (357).

The first endogenous iNKT cell ligand to be reported was isoglobotrihexosylceramide (iGb3), a lysosomal glycosphingolipid of previously unknown function (358). Zhou et al. set out to investigate whether therefore lysosomal glycosphingolipids could be the endogenous ligands for iNKT cells (358). They 65 observed that mice with β-hexosaminidase b, the lysosomal glycosphingolipid degrading (conversion of iGb4 to iGb3), deficiency (Hexb-/-) had a severe reduction in iNKT cells (95% on average). They also observed that thymocytes from Hexb-/- mice were unable to stimulate an iNKT cell hybridoma. In an in vitro stimulation assay with various natural products of β-hexosaminidase b activity, they found that only iGb3 was able to activate iNKT cells. In fact iGb3 presented by bone marrow-derived dendritic cells activated a variety of iNKT cell hybridomas specifically, irrespective of Vβ usage. However, they were unable to show iNKT cell staining using iGb3-loaded tetramers. Authors had themselves identified several other caveats in their identification of iGb3 as an endogenous iNKT cell ligand. Biochemical evidence for iGb3 in mouse and human was lacking – iGb3 in CD1d-bound lipids had not been shown, and neither had the presence of iGb3 in mouse or human tissue.

Doubts as to whether iGb3 was a physiological ligand for iNKT cells were raised by studies showing reduced iNKT cell numbers in other models of glycosphingolipid storage disorders that did not affect the formation of iGb3, suggesting a defect in lysosomal function (affecting CD1d trafficking). In 2007 two groups published their studies. Speak et al. used highly sensitive high performance liquid chromatography (HPLC) and only detected iGb3 expression in murine dorsal root ganglion, but not in any other mouse or human tissue examined, including in DCs or in the thymus (359). Porubsky et al. generated mice deficient in iGb3 synthase (iGb3S-/-) and showed that these had normal numbers of iNKT cells in the thymus, liver and the spleen with comparable TCR Vβ usage to controls (360). Thus, while iGb3 is capable of stimulating iNKT cells in vitro, the in vivo role of this ligand has been refuted.

Elution of CD1d-associated lipids has identified glycosphingolipids, glycerol- based phospholipids and lysophospholipids as potential iNKT cell ligands (361-364). Nevertheless, most of these failed to stimulate iNKT cells. More recently, Facciotti et al. proposed that ether-bonded phospholipid antigens – plasmalogen lysophosphatidylethanolamine and ether lysophosphatidic acid – have a role in iNKT cell development (312). Indeed, they demonstrated that mice unable to generate plasmalogens have impaired iNKT cell development.

However, that mammals are unable to synthetize α-linked glycolipids has recently been contested in a report by Kain et al., where the authors show that small amounts of α-glycosylceramides may be produced, that have been overlooked due to the relatively low sensitivity of lipid analytical methods (365). They show that that C24:1

66 α-GalCer is an endogenous iNKT cell ligand. They utilized two new approaches. First, Kain et al. developed a new probing method for commercially available β- glucosylceramide and β-galactosylceramide structure and terminal sugar linkage using enzymes responsible for their digestion in cells. β-glucocerebrosidase (a catabolic enzyme of defined specificity) digestion of commercially available β-anomers did not always lead to the loss of their iNKT activating potential. They were also able to purify this stimulatory capacity with lectins specific for α-linked glucose. Second, they used the specificity of antibodies to detect subtle differences in chemical structure. Kain et al. observed that the stimulatory capacity of thymocytes, DCs or RBL-CD1d cells in culture with NKT cells or hybridoma cells was lost following blockade of CD1d. Furthermore, this autoreactivity was blocked with L363 antibody specific for CD1d-α- GalCer complexes. They then purified the ligand in RBL-CD1d cells using sodium 3- [(2-methyl-2-undecyl-1,3-dioxolan-4-yl)methoxy]-1-propanesulfonate, and by multiple reaction monitoring mass spectrometry determined its weight at 809.67kDa – a C24:1 monoglycosylceramide. To determine the isomerisation, they radiolabeled the cells using C14-UDP-galactose followed two days later with immunoprecipitation using either L363 antibody or polyclonal antibodies (pAbs) specific for α-GalCer they had generated. When these lipid extractions were borate-impregnated to separate α and β- anomers, and subjected and high performance thin layer chromatography, it showed that the radiolabelled material was α-anomeric, therefore C24:1 α-GalCer. Rather than disagreeing with the current paradigm of mammalian lipidome, the authors postulate that mammals are capable of producing α-linked ceramides through other means. They propose that this could occur through a number of mechanisms. First, under specific physicochemical conditions such as in lysozymes passive α-anomerization occurs, but the efforts to induce this in low pH conditions have failed. Second, ceramide glucosyl and galactosyl are unfaithful and produce small amounts/quantities of α- anomers. There is no evidence to substantiate that this could be occurring. Third, as yet unidentified enzyme catalyses the conversion from β- to α-anomer. Finally, it is possible that enzymes involved in the removal of the α-linked glycan could also transfer it to a new ceramide (366).

67 α-GalCer

OCH

α-C-GalCer

Figure 1.5. Synthetic iNKT cell ligands. α-GalCer was the first iNKT cell ligand identified and since then analogues capable of eliciting and skewing iNKT cell responses have been synthesised, including OCH and α-C-GalCer.

68 CD1d and lipid presentation The interaction between the iNKT cell TCR and antigen-loaded CD1d is central to the activation of iNKT cells. To better understand iNKT cell function in steady-state and in disease, as well as to harness iNKT cell therapeutic potential it is necessary to understand how they recognize self, microbial and synthetic lipid antigen in the context of CD1d specifically. This includes understanding the factors that govern the antigenicity of a given ligand, and how monomorphic CD1d facilitates the binding of an array of self-, microbial- and synthetic ligands. The type of response elicited by the ligands may explain the pro- and anti-inflammatory roles that iNKT cells play in immune responses.

CD1d is a transmembrane protein and, like MHC class I molecules, it is non- covalently associated with β2-microglobulin. CD1d is expressed in both mice and humans, while humans also express CD1a, CD1b, CD1c and CD1e (295). It is composed of intracellular, transmembrane and three extracellular domains designated α1, α2 and α3 (367). α1 and α2 helices make up the lipid antigen-binding surface of CD1d.

CD1d recycles between the cell surface and the endolysosomal compartment, where endogenous and exogenous lipid loading occurs (368). Lipid transfer proteins play a role in this process. It had been reported that lysosomal trafficking of CD1d molecules (369) and lysosomal proteases (370) are required for endogenous lipid presentation. Saposins, also referred to sphingolipid activating proteins, were shown to play a role in lipid presentation since mice deficient in prosaposin lack iNKT cells (371). In addition saposins are required to bind α-GalCer to CD1d for iNKT cell activation (372).

The crystal structure of synthetic α-GalCer-CD1d complexes reveals that hydrophobic sphingosine and acyl chains are buried within pockets formed by α1 and α2 helices, while the galactose head group is exposed to the aqueous environment (373, 374). Of interest, the crystal structure of iTCR-lipid-CD1d complexes has revealed a different TCR binding mode, compared to TCR-peptide-MHC complexes (375). iTCR positioning over lipid-CD1d complexes appears to be more parallel, compared to the diagonal mode of conventional TCR engaging peptide-MHC. In iTCR binding, only the α-chain is positioned over the head group of the lipid, in contrast to both chains positioned to recognize the peptide in conventional TCR binding of peptide-MHC. The latter observation may explain the conservation of the iTCR α-chain. The lipid antigen

69 is important as it stabilises the interactions between the iTCR and CD1d. Nevertheless, most of the interaction that supports iTCR binding of CD1d occurs between TCR α- and β-chains and CD1d. Mutational analysis of iNKT cell TCR “hotspots” revealed comparable modes of TCR-mediated recognition for multiple different α-linked lipid antigens (376). Crystal structures of iTCR in complex with CD1d and Streptococcus pneumoniae- and Borrelia burgdorferi-derived α-linked glycerol-based lipids, revealed strikingly similar binding to iTCR-α-GalCer-CD1d (377, 378). Of interest, similar trimolecular complexes appear to be formed, even when β-linked lipids are studied (379, 380). The recognition of β-linked lipids by the iTCR results from the β-linked head group being flattened into a similar conformation to that naturally assumed by α- anomeric counterparts. Taken together these data suggest that iTCR utilises one binding strategy that allows the recognition of a variety of lipid antigens. iNKT cell activation iNKT cell activation has been proposed to occur through two distinct pathways. Broadly, iNKT cells can be triggered through the direct TCR recognition of exogenous ligands in the context of CD1d, or by a combination of innate cytokines and CD1d- loaded endogenous ligands (Fig. 1.6).

iTCR-mediated foreign antigen recognition is not required for iNKT cell activation. Endogenous lipid-CD1d complexes are sufficient to activate iNKT cells, when PRR-induced cytokines are present. The best-studied cytokine in iNKT cell activation is IL-12. Potent iNKT cell IFN-γ production during in vivo Salmonella typhimurium infection can be induced by the combination of (1) low level endogenous lipid-CD1d complex stimulation of the iTCR and (2) DC-derived IL-12 that is driven by pathogen-recognition via TLRs (381). The ability of Salmonella typhimurium to elicit iNKT cell activation through effect on DCs was also confirmed in a subsequent study (346). In a later study of murine responses to Escherichia coli LPS, Nagarajan et al. reported a role for both IL-12 and IL-18 in the IFN-γ production by iNKT cells, and surprisingly showed that such activation could occur in the absence of TCR signals altogether (382). In support of TCR-independent activation, it has also been reported that TLR9-activated DCs can drive iNKT cell IFN-γ in a CD1d-independent, IL-12- dependent manner in vitro (383). Furthermore, during MCMV infection, IFN-γ by iNKT cells was primarily driven by PRR signals and cytokines, and not iTCR stimulation by CD1d (383, 384).

70 Consolidating these findings, it is thus considered that pathogens expressing strongly activating ligands primarily stimulate iNKT cells in a cognate, while those expressing weakly stimulating ligands or not expressing iNKT cell ligands altogether, activate them in an innate, cytokine driven manner. Of note, Brigl et al. have subsequently shown that in iNKT cell IFN-γ production during microbial infection dominant role is played by innate TLR-driven IL-12, and endogenous ligand-derived signals (385). Together these studies have shown how iNKT cell responses to pathogens that do not contain TCR activating lipids can be elicited, expanding iNKT cell relevance in immunity.

There have been a number of studies that have explored where iNKT cells are activated and exert their effector function (386-388). King et al. attempted to elucidate the spatiotemporal nature of each mode of iNKT cell activation using novel in situ tracking of iNKT cells with loaded CD1d-tetramers and cytokine reporter mice in fresh splenic tissue by confocal imaging (386). They used cytokine-induced activation, glycolipid immunization and systemic infection with Streptococcus pneumoniae. Under steady-state conditions iNKT cells are distributed throughout the spleen, but accumulate in the MZ and bridging channels following α-GalCer administration or systemic infection. Upon activation by IL-12 and IL-18, King et al. showed that iNKT cells remain widely and evenly distributed throughout the spleen. In addition, following α- GalCer administration iNKT cells induced CD11c+ DCs to upregulate CCR7 and relocate to the T cell zone (386). Interestingly, despite localized cytokine production, non-cognate (CD1d-independent) iNKT cell-dependent mechanisms were sufficient to induce STAT signalling in the total B cell compartment and DC licensing across the spleen. This suggests that α-GalCer administration may affect lymphocytic responses by profoundly changing the landscape of secondary lymphoid organs.

71

Cytokine receptor

Weak cytokine signal iNKT cell

Cytokine PRR activation Pathogen iTCR PRR ligand PRR Strong TCR signal CD1d High-affinity microbial lipid

DC

Cytokine receptor

Strong cytokine signal iNKT cell

Cytokine PRR activation Pathogen iTCR PRR ligand Weak TCR signal CD1d

Low-affinity self or microbial lipid DC

Figure 1.6. iNKT cell activation modes. The two important signals in iNKT cell activation are (1) iTCR engagement by lipid-CD1d complexes, and (2) cytokine receptor stimulation. Pathogen-derived antigens can contain iNKT cell ligands that induce strong TCR signals (top). This mode of activation does not rely on APC-derived cytokines, induced by pathogen recognition through PRRs. Alternatively, pathogens that lack lipid antigens that strongly stimulate iNKT cells, can activate iNKT cells indirectly via cytokines (bottom). In most cases, the absence of lipid antigens derived from the pathogen still relies on iTCR engagement. This can be provided by endogenous CD1d-bound ligands. 72 Role of different APCs in iNKT cell activation As discussed above, the lipid ligands presented in the context of CD1d, cytokines and more broadly inflammatory signals are all needed in directing and conditioning iNKT cell responses. CD1d is expressed by various cells of the hematopoietic lineage, including B cells, DCs, and macrophages. However, it is also found on tissue cells, such as epithelial cells lining the gastrointestinal tract (389). Indeed the ability of iNKT cells to fulfil pro- or anti-inflammatory roles in immunity could be dictated by the instructions provided by the specific CD1d-lipid presenting cells.

The murine spleen has been of particular interest in the study of iNKT cell activation and cellular interactions. This is driven in part by a substantial reservoir of iNKT cells in the murine spleen, and part by the important role of this organ in immune surveillance. MZ, where antigen and cells are first encountered in the spleen, acts as a “bottleneck” for access to lymphoid areas. MZ zone contains innate immune cells, including macrophages (metallophilic CD169+ and SIGN-R1+MARCO-1+) and MZ B cells, which enable antigen trapping/degradation and early production of antibodies, respectively. MZ B cells express particularly high levels of CD1d, which has been reported previously (390). MZ B cells were nevertheless found by Bialecki et al. to play an accessory role in iNKT cell activation (391). They showed that in co-culture with splenic and hepatic iNKT cells, α-GalCer-sensitized MZ B cells were incapable of promoting iNKT cell IFN-γ or IL-4 production, while they maintained their ability to stimulate IL-2 production form iNKT cell hybridomas. The authors explain the ability to activate hybridomas, but not primary iNKT cells by the fact that hybridoma activation can be driven by TCR-mediated signals alone, without co-stimulation. When unstimulated DCs were added to the α-GalCer-pulsed MZB cell cultures with iNKT cells, ability of MZ B cells regained potential to stimulate iNKT cell IFN-γ and IL-4. The authors also demonstrated that this was 50-70% dependent on DC-expression of CD1d, suggesting either cross-presentation of α-GalCer or self-ligand involvement. Indeed addition of α-GalCer-loaded MZ B cells to DC-iNKT cell cultures amplified iNKT cell IFN-γ and IL-4 production. When exposed to various concentrations of α- GalCer, MZ B cell-depleted splenocytes were less able to stimulate the production of cytokines by iNKT cells than undepleted splenocytes in culture. Furthermore, α-GalCer- loaded MZ B cells were able to stimulate iNKT cells upon in vivo transfer. The authors’ conclusion was therefore that B cells and DCs cooperate to elicit optimal iNKT cell activation. However, it has been shown that MZ B cells in a CD1-dependent manner, at 73 least in vitro, potently activate iNKT cells (392, 393). Shin et al. show that B cells can induce a population of iNKT cells with an increased capacity to produce IFN-γ and decreased IL-4 which consequently drives a Th1 response (394). In this study the B cell role in skewing of the response was dependent on the expression of CD1d (394). B cell role in iNKT cell homeostasis and function has been shown in humans, and appears to be aberrant in disease (275). Reciprocally, iNKT cells can provide both direct and indirect help for B cell antibody production; this is discussed in the next section.

DCs are the major population attributed with initiating iNKT cell responses to α- GalCer (295). Steinman and colleagues have reported the ability of α-GalCer-loaded DCs to induce potent and sustained iNKT cell IFN-γ, compared to low level IL-4 and no IFN-γ induction by non-DC leukocytes (395). Furthermore, this increased IFN-γ production was associated with protection from B16 melanoma metastases. Nevertheless, the aim for the study was to compare the ability of DC-α-GalCer presentation to free α-GalCer to activate iNKT cells.

Bezbradica et al. reported that in B cell deficient mice iNKT cells show an enhanced ability to produce IFN-γ in response to α-GalCer. In contrast, in α-GalCer- administered DC-depleted mice both the frequency of iNKT cells expressing IFN-γ and the level of IFN-γ in the serum were profoundly reduced (396). More recently it has been shown that activation of iNKT cells by synthetic analogues of α-GalCer is impaired in Batf3 deficient mice (that lack a CD8α+ DC population)(397). King et al. observed that clodronate treatment of mice depleted approximately 1/3 of dendritic cells. Upon this treatment iNKT cells were unable to produce IFN-γ and less able to produce IL-4 in response to α-GalCer (386). The authors did not offer an explanation as to why residual cells were unable to, at least in part, activate iNKT cells. This loss of iNKT cell IFN-γ production upon DC depletion was nevertheless restored upon exogenous administration of IL-12 in combination with IL-18.

It is possible that while under steady-state conditions DCs are the primary lipid- presenting cell, B cell role increases when co-stimulatory signals are provided through a systemic inflammatory response. iNKT cell help for B cell antibody production The generation of high-affinity, class-switched antibodies relies critically on help from peptide-specific CD4+ T cells. In the past decade it has become clear that also iNKT cells can provide help to B cells for Ig responses.

74 iNKT cell ability to enhance humoral immunity was first reported by Galli et al. (398). They showed that human iNKT cells induce antibody production by autologous naive and memory B cells in vitro similarly to conventional CD4+ T cells. The induction of antibody production was CD1d-dependent and occurred even in the absence of α- GalCer, suggesting involvement of endogenous ligands. Since then, studies have identified that iNKT cells provide both cognate (innate) and non-cognate (adaptive) help for B cells (Fig. 1.7). Depending on the type of iNKT cell help provided, the outcomes for B cells and ensuing antibody responses are different.

iNKT cells that provide help to B cells acquire a phenotype similar to Tfh cells and are referred to as iNKTfh cells, as alluded to in the section “Generation of humoral responses within GCs”. These iNKT cells are CXCR5+PD-1+ICOS+Bcl-6+ and produce IL-4, IFN- γ, IL-21, and BAFF (161, 399).

It has been shown in a number of murine models that iNKT cells can provide CD1d-mediated cognate help for B cell antibody production in vivo (392, 393, 400). BCR-mediated uptake of CD1d-restricted antigens as well as CD1d-expression by B cells is required for iNKT cell cognate help (392, 400). Provision of cognate help also depends on the engagement of CD40 and CD80/86 on B cells by iNKT cells, as well as on iNKT cell release of IL-21 and IFN-γ, but not IL-4 (393, 401). The persistence of antibody titres and plasma cell survival is also shown to depend on iNKT cell production of BAFF and a proliferation-inducing ligand (APRIL) (402).

Non-cognate help refers to CD1d-lipid presentation by DCs to iNKT cells that results in DC licensing for enhanced antigen-presenting capacity (403). These DCs are then able to activate conventional T cells to acquire a Tfh phenotype for provision of help to B cells. In this case, antigen is presented by B cells to Tfh cells in the context of MHC class II.

The two modes of help generate different outcomes (404). Upon receiving cognate help, iNKT cell stimulated B cells produce high quantities of IgM as well as early class-switched antibodies. This robust primary response is associated with the formation of plasmablast foci and short-lived GCs, and lacks SHM/affinity maturation. This is a transient response that does not generate memory. Non-cognate help leads to GC formation, plasma cells and memory B cell induction, affinity maturation and class- switching. In addition to enhancing antibody responses, iNKT cells also act to curtail them, which may be important in the context of inflammatory and autoimmune disease. It has 75 been shown that in IL-18-driven sterile inflammation, associated with self-reactive antibody production, neutrophil accumulation leads to the CD1d-dependent induction of FasL on iNKT cells (405). FasL then acts to inhibit autoantibody producing CD1d+ B cells.

76 Direct Antigen Robust primary BCR antibody production No memory CD80/86 CD28 Lipid

B cell CD1d iTCR iNKT cell CD40 CD40L IFN-γ

IL-21

Indirect Antigen BCR

Isotype switching IFN-γ Affinity maturation BAFF Memory BAFF IL-21 APRIL IL-21 APRIL B cell CD40 MHC II

CD40L ICOSL

ICOS TCR iNKT cell CD28 CD80/86 T cell TCR iTCR CD40L CD1d CD40 MHC II DC

IL-12

IL-12

Figure 1.7. iNKT cells provide cognate and non-cognate help for B cell antibody responses. Upon recognition of CD1d-lipid complexes on B cells iNKT cells provide co-stimulation including via CD40-CD40L and CD80/86-CD28, as well as cytokines, such as IFN-γ and IL-21, that lead to antibody production (top). This cognate help leads to robust antibody production, but does not generate memory. Alternatively, iNKT cells through CD1d licence DCs to be more effective APCs and activate T cells (bottom). These T cells then interact with B cells to generate antibody production. This non- cognate iNKT cell help leads to memory induction.

77 iNKT cells in the regulation of inflammatory and autoimmune disease iNKT cells are important in the protection from pathogens and enhancement of anti-tumour immunity (295). The investigation of the role that iNKT cells play in the modulation of immune responses was prompted by recognition of their ability to produce regulatory cytokines (284). iNKT cell numerical deficiencies and functional defects have been reported in animal models of spontaneous autoimmunity (406, 407) and in patients with SLE (275, 408, 409), RA (408-410), MS (411) and IBD (409). Although mice deficient in iNKT cells, namely Jα18-/- and Cd1d-/- fail to develop spontaneous autoimmunity, aged Jα18-/- mice develop lupus-like nephritis (412). iNKT cell role and the mechanism by which they may function in various disease models and in human disease has generated a substantial amount of research, which is discussed below. Finally, iNKT cell capacity to both positively and negatively regulate immune responses has more recently been attributed to the existence of specialized subsets; the subsets ascribed with regulatory properties will also be discussed.

Type I diabetes mellitus (TIDM) In NOD mice, iNKT cell numbers inversely correlate with disease incidence (407). NOD mice crossed with Cd1d-/- mice develop an exacerbated disease compared to control NOD mice (413, 414). Experimental strategies, which increase the number of iNKT cells in various experimental models, demonstrate an improved level of protection against the development of autoimmunity, including diabetes in NOD mice. Adoptive transfer of syngeneic DN thymocytes enriched for iNKT cells prevent diabetes in NOD mice (415). Vα14Tg NOD mice, which have increased numbers of iNKT cells, are protected from developing diabetes compared to the non-transgene NOD control mice. Protection from diabetes is observed upon treatment of NOD mice with the prototypical iNKT cell agonist, α-GalCer (407, 414, 416-418), as well as with OCH (419) and the α-GalCer analogue C20:2 (420). Unlike in mice, where the role of iNKT cells in the protection from diabetes is established, translation in humans has proven to be more complex. Two original studies showed deficiency in iNKT cell number or function in patients with TIDM (421, 422), however these findings were not reproduced in other cohorts of patients (304, 423-427). One potential explanation for the discrepancies observed in the human studies could be that iNKT cell defect is observed in peripheral lymphoid tissues and not in the blood of NOD mice compared to non-autoimmune strains (428). Indeed, an iNKT cell functional defect in cytokine secretion has been reported in the pancreatic lymph nodes of TIDM individuals (429).

78 It has been reported that diabetes in humans and in NOD mice is Th1 driven, with associated inability to mount a Th2 response (430-436). Initial data suggested that the production of IL-4 by iNKT cell was pivotal in skewing pathogenic Th1 towards a protective Th2 response. This hypothesis is supported by increased IL-4 production in Vα14Tg mice (437) and correlation of increased IL-4 levels in the periphery with protection from diabetes observed in NOD Vα14Tg (438).

Transfer of diabetogenic BDC2.5 CD4+ T cells, isolated from BDC2.5 NOD mice, into recipient NOD mice induces diabetes (439, 440). In this model, iNKT cells provide protection as Vα14Tg NOD mice present reduced disease incidence compared to control NOD mice lacking iNKT cells. In addition, when BDC2.5 CD4+ T cells are co-transferred with CD5+ T cells from Vα14Tg NOD mice, the iNKT cell deficient mice are protected against diabetes. However, the protection in Vα14Tg mice is associated with pathogenic T cell anergy rather than differentiation of a Th2 response. CD4+ T cell production of IFN-γ, IL-2, IL-4 and IL-10, as well as their proliferation are inhibited in Vα14Tg compared to control iNKT cell deficient mice (440). More in depth analysis of the suppressive effect of iNKT cells shows that that protection provided by iNKT cells does not require functional IL-4, IL-10, IL-13 or TGF-β, but is dependent on cell-to-cell contact (441). Elucidation of how iNKT cells may regulate diabetes induction was provided in a co-transfer model. When diabetogenic BDC2.5 CD4+ T cells are co-transferred with normal NOD splenocytes from young mice, the development of diabetes is prevented (442). The absence of disease is dependent on iNKT cell production of IFN-γ, as iNKT cells from IFN-γ-deficient mice were unable to mediate regulation. Of note, IFN-γ production by iNKT cells is impaired in NOD mice (443). In addition, α-GalCer-iNKT cell-mediated protection from diabetes in NOD mice requires the presence of FoxP3+CD4+ Treg (444). However, whether or not Tregs are implicated in the iNKT cells mediated suppression remains to be fully ascertained as the activation of iNKT cells does not alter the frequency, phenotype or regulatory function of Tregs either in vitro or in vivo (417, 444).

RA Human studies in RA patients consistently report that iNKT cell numbers in the peripheral blood of RA patients are reduced compared to healthy individuals (408-410). Nevertheless, experiments in animal models have produced inconsistent results. The role of iNKT cells in arthritogenic inflammation fluctuates between pathogenic and protective, depending on the experimental system and mouse strain used. In CIA,

79 disease is less severe in iNKT cell deficient Jα18-/- mice and ameliorated upon CD1d- blockade (445). It was also reported that in two antibody-induced arthritis models (K/B×N serum transfer and anti-CII antibody-induced) disease was less severe in Jα18-/- and Cd1d-/- compared to WT mice. Nevertheless, administration of α-GalCer or OCH has been shown to inhibit CIA (446-448). Interestingly, in one of the studies, α-GalCer administration was shown to be associated with increased frequency of IL-10+CD4+ cells in the lymph node, and α-GalCer effects were abrogated by anti-IL-10 receptor antibody, implicating IL-10 in suppression mediated by α-GalCer (448). More recently, it has been published that iNKT cells play a protective role in arthritis as in Cd1d-/- mice both CIA and AIA are more severe (449). In the latter study it was proposed that there is a skew towards arthritogenic Th1 type responses in the absence of iNKT cells as authors observed a specific increase in IFN-γ production by T cells.

SLE Similarly to experimental arthritis, investigation of iNKT cell role in lupus has not always provided consistent results. Thus, lupus is ameliorated in (NZB×NZW)F1 mice by anti-CD1d treatment (450, 451) and disease could be transferred by NK1.1+CD3int T cells from diseased to healthy (NZB×NZW)F1 mice (452). α-GalCer treatment exacerbates nephritis and reduces survival (450, 452), but ameliorates dermatitis in MRL/lpr mice (453). Taking into account that MRL/lpr, C3H/gld and (NZB×NZW)F1 have reduced iNKT cell numbers (453-455), it is suggested that in murine lupus iNKT cells are protective before disease onset and pathogenic following initiation. It has been demonstrated that iNKT cells can play a protective role in limiting autoantibody response in a model, in which autoantibody production was induced by repeated injections with syngeneic apoptotic cells, which leads to an acute lupus-like disease (456). The absence of iNKT cells or disruption of their interaction with CD1d- expressing B cells leads to increased autoreactive B cell activation and exacerbation of disease. This suppression is mediated before B cell entry into GCs. Research from our laboratory has confirmed observations of reduced iNKT cell numbers in the peripheral blood of patients with SLE (275, 408, 409). Furthermore, iNKT cell homeostasis and activation appear to be regulated by CD1d-expressing immature B cells. In patients with active disease where there is a recycling defect of CD1d specifically on B cells, iNKT cells fail to produce IFN-γ upon activation (275). In patients responding to B cell depletion therapy, iNKT cell homeostasis and activation are normalized upon B cell repopulation; immature CD1dhi B cells correlate with increased iNKT cell frequency in patients responding to therapy. 80 EAE In Vα14Tg NOD mice that are protected form EAE compared to the non- transgene NOD controls, protection has been shown to be independent of iNKT cell production of IL-4 (457). IL-4-deficient Vα14Tg NOD mice are still protected from EAE. Of note, IFN-γ+granzyme B+ iNKT cells infiltrate the CNS in this model (458). Protection in EAE appears to be mediated through inhibition of Th1 and Th17 responses (458). Through apparently similar mechanisms employed by iNKT cells in suppression of autoimmunity in NOD mice, iNKT cells display immunomodulatory role in experimental autoimmune uveitis (EAU). α-GalCer-iNKT cell mediated protection was associated with reduced Th1/Th17 responses and was abolished by the systemic IFN-γ neutralization (459). Thus further suggesting that iNKT cells producing IFN-γ play a protective role in autoimmunity.

Specialized iNKT cell subsets in immunomodulation More recently, specialized iNKT cell subsets that are regulatory in autoimmunity or inflammation have been identified. IL-10 producing NKT10 cells have been shown to be induced by repeated administration of α-GalCer (460). These cells impaired anti-tumour immunity and ameliorated EAE in an IL-10-dependent manner. iNKT cell activation with α-GalCer leads to amelioration of disease. This is associated with the emergence of a FoxP3+ iNKT cell population in the cervical lymph nodes, which was critically dependent on TGF-β (461). These cells had similar PLZF mRNA levels to their non-FoxP3+ iNKT cell counterparts. Adipose tissue E4BP4+PLZF- iNKT cells produce IL-10 and IL-2 modulating obesity-associated immune activation (462). These iNKT cells suppress pro-inflammatory phenotype in macrophages and control the proliferation and suppressive function of Tregs.

81 Antigen-induced arthritis Antigen-induced arthritis (AIA) is T cell-driven delayed type hypersensitivity disease model that aims to mimic the pathology observed in RA (463). It reproduces multiple characteristics of RA, including immune complex deposition in the cartilage, cartilage and bone erosion, as well as immune cell infiltration of the synovial tissue (463).

The model was developed in rabbits by Dumonde and Glynn in 1962 and relied upon their assumption that repeated antigen injections were required to initiate pathogenesis (464). AIA is dependent on local antigen administration in a pre- immunized animal (463, 464). Among antigens, OVA, bovine serum albumin (BSA) and fibrin have been exploited (463, 464). Antigen retention in the avascular collagenous structures of the joint occurs either through charge- or antibody-mediated mechanisms (465, 466). While antibody-mediated trapping occurs in rabbits, it does not appear to significantly contribute to antigen retention in mice (463). Nevertheless, antigen retention can be achieved through the use of cationic antigens (467, 468). As such, methylated BSA (mBSA) is used as the antigen in mice due to its cationicity (463).

The murine model of AIA was first described in 1977 (469). Arthritis is initiated by the injection of antigen directly into the joint of a mouse pre-immunized with complete Freund’s adjuvant (CFA) (463). The disease is unilateral, which enables the assessment of disease severity clinically and immunologically through the comparison of contra-lateral joints in the same animal (463). Similarly, the disease incidence is 100% and the onset is defined – in mice swelling of the joint can be detected within the first 24 hours after local administration of antigen. These features are in contrast to classical arthritis models, such as the most commonly used collagen-induced arthritis (polyarthritis with disease onset 21-28 days following immunization and disease incidence 80-100% in DBA/1 mice) (470).

The dominant role for T cells using adoptive transfer experiments in athymic nude mice was established upon description of AIA in mice (469). A T cell role in driving pathogenesis of AIA was further shown in experiments in which lymphocytes from pre-immunized animals were able to transfer disease to susceptible C57BL/6 mice (471). This disease-inducing capacity was enhanced in B cell-depleted lymphocytes and diminished upon anti-Thy-1 treatment of lymphocytes (471). It was found that while oligogenic, the high responder state to antigen and susceptibility to AIA was primarily

82 determined by T cells (472). That the T cells driving arthritis were CD4+ was demonstrated by Petrow et al. who showed that in SCID mice in which transfer of lymphocytes is able to transfer disease, CD4+ T cell-, but not CD8+ T cell-depletion abrogated this effect (473).

While chronicity was not induced in our model, it is driven by antigen retention and the destruction of the cartilage is greater with high antibody titres (463). Flares are also T cell-driven and mBSA specific (474-476) and can be induced through repeated administration of low dose antigen either locally or systemically (463). Bordetella pertussis boosting of the immune response can enhance the generation of antibodies, and thus cartilage erosion (463); this has been reported to be due to IL-23-driven enhancement of a Th1 response. In exacerbations during flares, IL-1 drives cartilage pathology (477). T cell-driven IL-17 has been shown to regulate the pathogenic IL-1 and TNF-α levels in the synovial washouts during flares and the authors were able to ameliorate flare pathology by IL-17 neutralization (478).

To summarize the experimental procedure, mice are sedated under isoflurane and mBSA emulsified in CFA is administered subcutaneously. The injection of mBSA is made under the patellar ligament one week later (463). The control knee is injected with equivalent volume of PBS without the antigen. Symmetric butterfly-like swelling develops around the mBSA-injected joint. In our experiments, peak swelling occurs on days 1-3 following intra-articular injection, and mice enter remission between days 5 and 7. Matrix proteoglycans have been reported to be severely depleted at day 7 following disease induction (463).

83 Chapter II: Materials and Methods

Mice C57BL/6 mice were from Harlan Laboratories (now Envigo Laboratories). µMT (184), Cd1d-/- (479), Il10-/- (480) mice were from The Jackson Laboratory. CD11c-DTR (481) mice were provided by C. Bennett, University College London, UK. Jα18-/- (307) mice were provided by A. Karadimitris, Imperial College London, UK. Il10ra-/- mice were provided by W. Muller, University of Manchester, UK. IL-10 transcriptional reporter (Vert-X) mice were provided by C. Karp, Bill & Melinda Gates Foundation (203). All mice were on C57BL/6 background. Mice were age- and sex-matched, and used at 6-12 weeks of age. Mice were bred and maintained at the animal facility at University College London. All experiments were approved by the Animal Welfare and Ethical Review Body of University College London and authorized by the United Kingdom Home Office.

Induction of antigen-induced arthritis (AIA) and administration of α-GalCer AIA was induced as previously described (197, 247). Briefly, mice were immunized subcutaneously at the tail base with 200µg methylated bovine serum albumin (mBSA, Sigma-Aldrich) in 200µl CFA and phosphate-buffered saline (PBS, Sigma-Aldrich). 3mg/ml CFA was prepared by mixing of Mycobacterium tuberculosis (Difco) in IFA (Sigma-Aldrich). After 7 days, mice received an intra-articular injection of 200µg mBSA in 10µl PBS in the right knee or PBS alone as a control in the left knee. Joint size was measured using POCO 2T calipers (Kroeplin Längenmesstechnik), and swelling was calculated as percentage increase in size between antigen- and control- injected knee. When treated with α-GalCer (Enzo Life Sciences), mice received an intraperitoneal (i.p.) injection of 2µg α-GalCer in 200µl 0.5% DMSO (Sigma-Aldrich) in PBS at the time of the intra-articular injection, unless otherwise indicated. As a control, a separate group of mice were treated with 200µl 0.5% DMSO in PBS.

B cell depletion For B cell depletion, mice were treated intraperitoneally with 200µg anti-CD20 (5D2, Genentech) in 200µl PBS, or isotype control antibody in PBS, on days -21, -14 and -7 prior to intra-articular injection to induce AIA.

CD11c+ cell depletion CD11c-DTR mice were treated intraperitoneally with 100ng diphtheria toxin from Corynebacterium diphtheriae (Sigma-Aldrich) in 200µl PBS, or PBS alone as

84 control, as previously described (482), at the time of intra-articular injection to induce AIA, and 8 hours prior to α-GalCer or vehicle administration.

MZ B cell depletion For MZ B cell depletion, mice were treated intraperitoneally with 100µg of each anti-CD11a (M17/4, eBioscience) and anti-CD49d (cR1-2, BioLegend), or isotype control antibodies, as previously described (483, 484), 16 hours prior to intra-articular injection to induce AIA.

In vivo IFN-γ neutralization For IFN-γ neutralization, mice were treated intraperitoneally with 200µg of anti- IFN-γ (XMG1.2, BioXCell) in 200µl PBS, or isotype control antibody in PBS 2 days prior to intra-articular injection, and at day 0 of and 2 days post intra-articular injection with mBSA as previously described (459).

Cell preparation Spleens and were collected in complete RPMI-1640 with L-glutamine and

NaHCO3 (cRPMI, Sigma-Aldrich) with 1:10 FBS (Biosera), 1:100 Penicillin (10’000 units/ml)/Streptomycin (10mg/ml) (Sigma-Aldrich) and 1:1000 50mM 2- Mercaptoethanol (Gibco, Invitrogen). Single cell suspensions were obtained by dissociating tissue through 70µM cell strainers (BD Biosciences), and erythrocytes in spleens lysed using Red Blood Cell (RBC) Lysis buffer (Sigma-Aldrich). For the isolation of liver mononuclear cells, Percoll (Amersham Biosciences) 30%/70% gradient was used. Briefly, following extensive washes, liver homogenates were resuspended in 30% Percoll in RPMI-1640 and layered on 70% Percoll in PBS. Following centrifugation, the pellet was recovered. Following washes, RBCs were lysed.

Generation of mixed BM chimeric mice Mixed BM chimeric mice were generated as previously described (194, 197, 266). Briefly, recipient WT mice received 8Gy γ-irradiation via a caesium source. Five hours following irradiation, recipients were administered 107 donor BM cells. To restrict genetic deficiency of CD1d or IL-10 to B cells, lethally irradiated WT mice were reconstituted with a mixture consisting of 80% BM from µMT mice and 20% BM from Cd1d-/- or Il10-/- mice. Control mice received 80% BM from µMT mice and 20% BM from WT mice. The efficiency of irradiation of WT mice was confirmed by the absence of B cells in mice reconstituted with 100% µMT BM. Chimeric mice were left to reconstitute for at least 8 weeks prior to use in AIA experiments.

85 Adoptive transfer T2-MZP B cells were isolated from the spleens of Jα18-/-, Cd1d-/- or WT mice on day 7 post-disease onset by FACS (Figure 2.1). iNKT cells were isolated from the spleens of WT mice on day 3 following α-GalCer treatment by FACS (Figure 2.1). 2×106 T2-MZP B cells or iNKT cells were then transferred intravenously (i.v.) to Cd1d- /- and WT, or µMT and WT mice respectively, immediately following induction of arthritis. The control group received a Hanks’ Balanced Salt Solution (Sigma-Aldrich) injection.

Flow cytometry and cell sorting Single-cell suspensions were incubated with FcR blocking reagent for mouse (Miltenyi Biotec) prior to staining with specific antibodies (Table 1) and CD1d tetramers. CD1d tetramers were conjugated to PE, AF488, APC and AF647 and loaded with PBS57 or left unloaded, and were obtained from the National Institutes of Health Tetramer Core Facility. Tetramer staining was carried out prior to surface antibody staining. All staining was performed at 4°C for 30 minutes as described previously (197, 204, 247, 266, 274). For transcription factor and Ki-67 staining, cells were fixed and permeabilized using Foxp3/Transcription Factor Staining Buffer Set (eBioscience) according to manufacturer’s instructions. For the detection of IFN-γ and IL-17, cells were stimulated for 5 hours in cRPMI with 50ng/ml PMA and 250ng/ml ionomycin (both Sigma-Aldrich). For the detection of IL-10, cells were stimulated for 5 hours in cRPMI with 100ng/ml PMA and 1000ng/ml ionomycin. Protein Transport Inhibitor (either containing brefeldin A or monensin, both from BD Biosciences) was added during stimulations according to manufacturer’s instructions. Cells were then stained for surface markers, fixed and permeabilized using Intracellular Fixation & Permeabilization Buffer Set (eBioscience) according to manufacturer’s instructions. Dead cells were excluded with LIVE/DEAD Fixable Dead Cell Stains or DAPI (both Invitrogen). Flow cytometry was performed on LSRFortessa, LSRFortessa X-20, LSRII and cell sorting on FACSAria cell sorter (all BD Biosciences). Data were analysed using FlowJo (Treestar).

Immunofluorescence Spleens were harvested and embedded in O.C.T compound (Tissue-Tek) and snap-frozen in liquid for cryosectioning. Tissue was cut into 6µm sections. Sections were then fixed with 100% ethanol for 5-10 min (4°C), followed by rehydration in PBS for 5 min (4°C). The sections were blocked with 10% normal goat serum for 20 min at room temperature (RT), then rinsed with PBS. The tissues were 86 incubated with primary antibodies for 1 hour at RT and then washed 3X in PBS (2 min each wash). When secondary antibodies were used, the tissue was further incubated with the secondary antibody for 1 hour at RT, again followed by 3 washes in PBS (2 min each wash). The slides were finally mounted in Prolong Antifade Mountant with DAPI (Thermo Fisher), imaged on a Leica SPE confocal microscope using LAS X software and analysed using Fiji (ImageJ).

Visualising CD1d and lipid raft co-localisation by ImageStream Splenocytes were stained ex vivo as described for flow cytometry. Following this, cells were stained with cholera toxin B subunit (CTB)-FITC (Sigma-Aldrich). Cells were then washed in PBS, and incubated with goat anti-CTB pAb (Merck) in the dark for 20 min (4°C), before transferring to an incubator (37°C) for 20 min to induce clustering of CTB. Following washes and fixing, cells were imaged on ImageStreamX (Merck). Analysis was performed using IDEAS (Merck).

In vitro B cell culture Splenic B cells were negatively purified (>95%) by magnetic separation using CD43 (Ly-48) microbeads (Miltenyi Biotec) and cultured for 72 hours with 10µg/ml mBSA, 10µg/ml anti-CD40 (FGK4.5, BioXCell) or mBSA with anti-CD40. Supernatants from cell cultures were harvested and IL-10 was measured using mouse IL-10 DuoSet ELISA (R&D Systems) according to manufacturer's instructions.

In vitro iNKT cell suppression assay Splenic iNKT cells from arthritic µMT and WT mice were sorted by FACS 16 hours following α-GalCer treatment. iNKT cells were then co-cultured 1:1 with CD4+CD25- T cells, isolated from the spleens of WT mice by magnetic separation (Miltenyi Biotec), for 72 hours in the presence of 0.5µg/ml plate-bound anti-CD3 (145- 2C11, BD Biosciences). PMA, ionomycin and Protein Transport Inhibitor were added for the last 6 hours of culture. Following stimulation cells were analysed for the expression of IFN-γ by flow cytometry.

In vitro B cell suppression assay B cells were isolated from the spleens of WT, Jα18-/- and Cd1d-/- mice on day 7 post AIA induction followed by stimulation with 10µg/ml anti-CD40 in cRPMI for 5 hours. B cells were then co-cultured 1:1 with CD4+CD25- T cells, isolated from the spleens of WT mice by magnetic separation (Miltenyi Biotec), for 72 hours in the presence of 0.5µg/ml plate-bound anti-CD3 (145-2C11, BD Biosciences). PMA, ionomycin and Protein Transport Inhibitor were added for the last 6 hours of culture.

87 Following stimulation cells were analysed for the expression of TNF-α by flow cytometry.

In vitro B cell and dendritic cell depletion Splenocytes were negatively depleted of B cells (CD19) or dendritic cells (CD11c) by FACS and stimulated with IL-2 (80U/ml) in the presence of 100ng/ml α- GalCer for 48 hours.

RNA sequencing Splenic iNKT cells from B-Cd1d-/- and B-WT mice that received α-GalCer or vehicle as control were sorted by FACS. Sorted cells were >95% pure. Total RNA was isolated using PicoPure RNA Isolation Kit (Applied Biosystems) according to manufacturer’s instructions. cDNA was synthesized and amplified using SMARTer v4 Ultra Low Input Kit (Clontech). 200pg of amplified cDNA was then converted to sequencing library using the Nextera XT Kit (Illumina, 12 cycles of PCR). Samples were individually indexed during library preparation, which allowed pooling and sequencing on the Illumina NextSeq 500 platform (43bp paired end). An average of 18M read pairs was obtained for each sample. Reads were adapter trimmed (FASTX- Toolkit) and aligned to the Mus musculus genome (mm10) using Tophat v2.0.7.

Bioinformatics analysis of RNA-seq data Differential expression analysis was performed using the edgeR algorithm (485) under default settings. Genes with false discovery rate (FDR) values <0.05 were considered to be differentially expressed. Significantly over-represented pathways were identified using Signaling Pathway Impact Analysis (SPIA) (486). Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathways database (487) was used as reference, whilst full mouse genome was used as background for enrichment.

Statistical analysis All data are expressed as mean±s.e.m. Replicates were biological replicates. Clinical scoring was performed blinded. Mice were assigned to treatment groups at random for all mouse studies and, where possible, mixed among cages. Groups were compared with two-tailed Student’s t-test (2 groups), one-way analysis of variance (ANOVA) (≥3 groups) or two-way ANOVA (clinical data and other instances, when two factors were compared between groups simultaneously) followed by Bonferroni post-test with Prism (GraphPad). Data distribution, and variance between groups were considered before statistical comparison. Data followed Gaussian distribution with similar variance between groups that were compared. Careful consideration was given

88 to calculate the group sizes needed to give statistically significant results, while using the minimum number of mice possible. Experimental group sizes were decided after consultation with a statistician at UCL. For in vivo studies, power calculations were based on data showing mean maximum WT arthritic knee swelling of 2.00mm with a standard deviation of 0.39mm, and an expected test group arthritic knee swelling of ±0.6mm compared to control group depending on the experiment. Group sizes of 5 or above were sufficient to reach a statistical power of at least 80%. Further, experiments were performed independently in triplicate to control for experimental variation.

A Pre-sort Sorted Gated on live CD19- cells Gated on live CD19- cells CD1d-tetramer CD1d-tetramer

CD3 CD3 B Pre-sort Gated on live CD19+ Gated on live CD19+ cells CD21hiCD24hi cells

T2-MZP and MZ MZ T2-MZP

Fo CD21

CD24 CD23 Sorted Fo MZ T2-MZP Gated on live CD19+ Gated on live CD19+ Gated on live CD19+ cells Gated on live CD19+ cells CD21hiCD24hi cells Gated on live CD19+ cells CD21hiCD24hi cells

T2-MZP and MZ T2-MZP and MZ MZ T2-MZP T2-MZP and MZ MZ T2-MZP

Fo Fo Fo CD21 CD24 CD24 CD23 CD24 CD23 Figure 2.1. Gating strategy and purity for iNKT cell, Fo, MZ and T2-MZP B cell transfer experiments. (A) Representative flow cytometry plots showing iNKT cells prior to flow cytometry sorting (left) and purity following sorting (right). (B) Representative flow cytometry plots showing B cell subsets prior to flow cytometry sorting (top) and purity of indicated subsets following sorting (bottom).

89 Antibody Clone Source CD3-BUV395 145-2C11 BD Biosciences CD3-PerCPCy5.5 17a2 BioLegend CD4-BV711 RM4-5 BioLegend CD8a-PECy7 53-6.7 BioLegend NK1.1-BV421 PK136 BioLegend NK1.1-PE PK136 BD Biosciences CD19-PECy7 1D3 BD Biosciences CD19-PE 1D3 BD Biosciences CD19-BV785 6D5 BioLegend CD21-FITC 7G6 BD Biosciences CD21-APC 7G6 BD Biosciences CD21-PerCPCy5.5 7E9 BioLegend CD23-PE B3B4 BD Biosciences CD23-BV605 B3B4 BD Biosciences

CD23-PECy7 B3B4 BioLegend CD24-PerCPCy5.5 M1/69 BD Biosciences CD24-APC M1/69 BD Biosciences CD24-BV421 M1/69 BioLegend CD1d-PE 1B1 BD Biosciences CD1d-PerCPCy5.5 1B1 BioLegend CD1d-BV510 1B1 BD Biosciences CD5-AF647 53-7.3 BioLegend CD5-PE 53-7.3 BD Biosciences IgD-PerCPCy5.5 11-26c.2a BioLegend IgD-BV510 11-26c.2a BioLegend IgD-BV711 11-26c.2a BioLegend IgM-APCCy7 RMM-1 BioLegend IgM-FITC RMM-1 BioLegend B220-BUV395 RA3-6B2 BD Biosciences CD11b-APC M1/70 BioLegend CD11b-APCCy7 M1/70 BioLegend CD11b-BV711 M1/70 BioLegend CD11c-PECy7 N418 BioLegend CD11c-FITC N418 BioLegend

90 I-A/I-E-PerCPCy5.5 M5/114.15.2 BioLegend F4/80-BV605 BM8 BioLegend F4/80-APC BM8 eBioscience Ly6C-BV785 HK1.4 BioLegend Ly6G-BV421 1A8 BioLegend Ly6G-BV605 1A8 BioLegend Siglec F-PE E50-2440 BD Biosciences CD25-BV510 PC61 BioLegend CD25-FITC PC61 BioLegend CD25-PerCPCy5.5 PC61 BioLegend CD69-PECy7 H1.2F3 BioLegend CXCR5-PerCPCy5.5 L138D7 BioLegend PLZF-PE Mags.21F7 eBioscience PLZF-PECy7 9E12 BioLegend PLZF-PerCPCy5.5 9E12 BioLegend Ki-67-PE B56 BD Biosciences Ki-67-PECy7 B56 BD Biosciences T-bet-PE eBio4B10 eBioscience T-bet-eFluor660 eBio4B10 eBioscience GATA3-AF647 16E10A23 BioLegend IL-10-PE JES5-16E3 BioLegend IFN-γ-APC XMG1.2 BD Biosciences IFN- γ -FITC XMG1.2 BD Biosciences IL-17A-PE TC11-18H10.1 BioLegend IL-17-AF647 eBio17B7 eBioscience TNF-α-APC MP6-XT22 eBioscience TNF-α-PE MP6-XT22 BD Biosciences Table 1. Antibodies used for flow cytometry and cell sorting.

91 Chapter III: Results I

As discussed in the introduction, the administration of the prototypical iNKT cell agonist α-GalCer has been shown to induce suppression of immune responses in various models of inflammatory and autoimmune diseases (284, 407, 488). Nevertheless, the contribution of different CD1d-expressing cells to α-GalCer-driven modulation of immunity has remained relatively unexplored.

A numerical deficiency in iNKT cells is observed in various autoimmune conditions (275, 408-411). The role of CD1d expression on B cells for the maintenance of iNKT cells as well their functional characteristics in human peripheral blood has previously been demonstrated (275). It was shown that the interaction between B cells and iNKT cells was perturbed in patients with SLE due to a defect in CD1d recycling on B cells. CD1d expression on B cells was normalized in patients responding to anti- CD20 treatment, and iNKT cell responses reversed to healthy control levels. This correlated B-iNKT cell interaction to suppression of autoimmune responses.

In this chapter we confirm that α-GalCer is suppressive in AIA and explore the specific requirement for B cells and CD11c+ DCs for this effect. We show that B cells are responsible for the immunomodulatory effect following in vivo α-GalCer administration. This is associated with the ability of B cells to regulate the expression of the master regulator of iNKT cell development and function, PLZF, as well as IFN-γ. We also demonstrate that DCs are instead required for iNKT cell expansion in vivo. We show the different roles of B cells and DCs in iNKT cell activation in response to α- GalCer in vitro. While depletion of B cells from splenocytes resulted in lower PLZF and IFN-γ expression, DCs in turn had superior ability to induce iNKT cell expression of CD25 and CD69 activation markers.

92 B cells mediate α-GalCer-iNKT cell dependent amelioration of arthritis To address the role that B cells play in α-GalCer-iNKT cell-driven amelioration of arthritis, B cell deficient (µMT) and WT mice were immunized with methylated bovine serum albumin (mBSA) in CFA, followed by an intra-articular injection of mBSA at day 7 in the right knee or PBS alone as a control in the left knee. Mice were treated with α-GalCer (to promote interactions with CD1d-expressing cells) at the time of intra-articular injection of mBSA. α-GalCer treatment suppressed the development of arthritis in WT whilst it failed to inhibit disease in µMT mice (Fig. 3.1A). In response to α-GalCer we observed an equal increase in the frequencies and absolute numbers of iNKT cells in the spleen (Fig. 3.1B). Analysis by flow cytometry revealed that splenic iNKT cells from both α-GalCer treated µMT and WT mice expressed comparable levels of Ki-67, suggesting that α-GalCer-driven proliferation of iNKT cells was normal in the absence of B cells (Fig. 3.1C).

iNKT cell expansion in response to α-GalCer also occurred in the liver, the mBSA-injected joint-draining inguinal LN as well MLN of µMT and WT mice (Fig. 3.2, A,B, and C, respectively). Comparing the α-GalCer-treated groups, the frequency of iNKT cells was significantly lower in the MLN of µMT mice (Fig. 3.2C). Albeit not significantly, in the vehicle-treated µMT mice the frequency of iNKT cells in the MLN was also lower than in the vehicle-treated WT mice. Therefore, it is likely that B cells play a role in the iNKT cell maintenance in steady-state rather than in iNKT cell expansion in response to α-GalCer. It can thus be concluded that iNKT cell expansion in response to α-GalCer is normal in the absence of B cells.

It has been previously shown that the BTB-ZF transcriptional regulator PLZF, is responsible for the innate-like effector phenotype of iNKT cells and rapid cytokine production, including IFN-γ (320, 321), which has also been associated with iNKT cell suppressive function (442, 458, 459, 489, 490). We observed that in the absence of B cells, splenic iNKT cell ability to up-regulate the expression of PLZF and IFN-γ in response to α-GalCer was significantly impaired compared to WT mice (Fig. 3.3, A and B). This is in contrast to the liver where α-GalCer-induced iNKT cells expressed comparable levels of PLZF and IFN-γ in µMT and WT mice (Fig. 3.3, C and D). We could not detect FoxP3+ or E4BP4+iNKT cells, previously ascribed with regulatory function in the lymph nodes and adipose tissue, respectively (462, 491), in the spleens of mice treated with α-GalCer. Of note, iNKT cell activation leads to the downregulation of the iTCR (492), and thus the inability to detect and interrogate the

93 entire iNKT cell population at early time-points constitutes an important limitation in our assays.

A WT - vehicle µMT - vehicle 50 WT - -GalCer 50 µMT - -GalCer

40 40

30 30

20 20 **** **** 10 10 Increase in knee swelling Increase in knee swelling comparedcontrol to knee (%) comparedcontrol to knee (%) 0 0 0123 0123 Day post disease onset Day post disease onset

- Gated on live CD19- cells Gated on live CD19 BC+ int Vehicle α-GalCer 6 8 CD1d-tetramer CD3 cells 8000 **

** ) *** *** 6 6 6000 4 10 ** *** 1.28 3.62 μMT - α-GalCer

WT 4 4000 μMT - vehicle

2 Ki-67(MFI) iNKT cells (%) cells iNKT WT - α-GalCer 2000 iNKT cells ( cells iNKT 2

0 0 WT - vehicle 0 α-GalCer -+-+α-GalCer -+-+ α-GalCer -+-+ WT μMT WT μMT Ki-67 WT μMT

1.35 4.05 MT μ CD1d-tetramer

CD3 Figure 3.1. α-GalCer-iNKT cell-dependent amelioration of arthritis requires B cells. (A) Mean clinical score of wild-type (WT) mice (left) and µMT mice (right) that received α-GalCer or vehicle alone following induction of arthritis. Y axis shows percentage swelling in antigen-injected knee compared to control knee (µMT±α-GalCer n=14 per group, WT±α-GalCer n=20 per group, one of three experiments is shown). (B) Representative flow cytometry plots and bar charts showing the frequency and number of splenic iNKT cells in µMT and WT mice that received α-GalCer or vehicle alone (n=5 per group, one of three experiments is shown). (C) Representative histograms and bar chart showing median fluorescence intensity (MFI) of Ki-67 in splenic iNKT cells from α-GalCer- or vehicle-treated µMT and WT mice (n=3 per group, one of three experiments is shown). B, C analysed on day 3 post-disease onset. Data are mean±s.e.m. **P<0.01, ***P<0.001, ****P<0.0001 (A two-way analysis of variance (ANOVA), B, C one-way ANOVA).

94 Liver A Gated on live CD19- cells Vehicle α-GalCer 40 4 ** ****

) ****

6 **** 30 3

19.8 35.0 10

WT 20 2

iNKT cells (%) cells iNKT 1

10 ( cells iNKT

0 0 α-GalCer -+-+α-GalCer -+-+ WT μMT WT μMT

15.7 36.6 MT μ CD1d-tetramer

CD3

Inguinal lymph node B Gated on live CD19- cells Vehicle α-GalCer 0.8 0.008 ** ) 0.6 * 6 0.006

0.24 0.68 10

WT 0.4 0.004 iNKT cells (%) cells iNKT

0.2 ( cells iNKT 0.002

0.0 0.000 α-GalCer -+-+α-GalCer -+-+ WT μMT WT μMT

0.32 0.51 MT μ CD1d-tetramer

CD3

Mesenteric lymph node C Gated on live CD19- cells Vehicle α-GalCer 1.0 0.05 **** ** ** )

0.8 6 0.04

0.28 0.77 10 0.6 0.03 *

WT * 0.4 0.02 iNKT cells (%) cells iNKT 0.2 ( cells iNKT 0.01

0.0 0.00 α-GalCer -+-+α-GalCer -+-+ WT μMT WT μMT

0.14 0.41 MT μ CD1d-tetramer CD3 Figure 3.2. α-GalCer-driven iNKT cell expansion occurs in the liver, inguinal and mesenteric lymph nodes. Representative flow cytometry plots and bar charts showing the frequency and number of iNKT cells in the liver (A), inguinal (B) and mesenteric (C) lymph nodes of µMT and WT mice that received α-GalCer or vehicle alone (n=4 per group, one of two experiments is shown). Analysed on day 3 post-disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01, ****P<0.0001 (one-way ANOVA).

95 Spleen A B Gated on live CD19-CD1d-tetramer+CD3int cells Spleen Vehicle α-GalCer Gated on live CD19- 80 + int CD1d-tetramer CD3 cells 8000 18.0 60.7 **** *** *** ** 60 6000 **** 40 μMT - α-GalCer WT

* iNKT cells (%) 4000 + μMT - vehicle 20 PLZF (MFI) PLZF IFN- WT - α-GalCer 2000 0 α-GalCer -+-+ WT - vehicle 0 WT μMT α-GalCer -+-+ 9.76 43.1 PLZF WT μMT MT μ γ IFN-

CD3

Liver Gated on live CD19-CD1d-tetramer+CD3int cells C Liver D Vehicle α-GalCer 80 Gated on live CD19- + int 3.80 28.0 CD1d-tetramer CD3 cells 8000 60

6000 **** 40 μMT - α-GalCer WT **** iNKT cells (%) 4000 + μMT - vehicle 20 IFN- PLZF (MFI) PLZF WT - α-GalCer 2000 0 α-GalCer -+-+ WT - vehicle 0 WT μMT α-GalCer -+-+ 1.87 33.2 PLZF WT μMT MT μ γ IFN-

CD3 Figure 3.3. Splenic, but not hepatic iNKT cells display reduced PLZF and IFN-γ induction in µMT mice following α-GalCer treatment. (A) Representative histograms and bar chart showing MFI of PLZF in splenic iNKT cells from α-GalCer- or vehicle-treated µMT and WT mice (n=3 per group, one of three experiments is shown). (B) Representative flow cytometry plots and bar chart showing the frequency of IFN-γ+ splenic iNKT cells from α-GalCer- or vehicle-treated µMT and WT mice (n=4 per group, one of three experiments is shown). (C) Representative histograms and bar chart showing MFI of PLZF in hepatic iNKT cells from α-GalCer- or vehicle- treated µMT and WT mice (n=3 per group, one of two experiments is shown). (D) Representative flow cytometry plots and bar chart showing the frequency of IFN-γ+ hepatic iNKT cells from α-GalCer- or vehicle-treated µMT and WT mice (n=4 per group, one of two experiments is shown). Analysed at 16h post-disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (one-way ANOVA).

96 B cell depletion abrogates the protective effect of α-GalCer on arthritis µMT mice lack B cells from birth, which could impair the development of iNKT cells. In particular, multiple B cell-extrinsic immune defects in µMT mice have been reported, including aberrant T cell and DC function (190, 199, 493-495). To explore the role of B cells in lipid presentation in an environment where iNKT cells have developed in the presence of B cells, adult WT mice were treated either with a depleting anti- CD20 antibody or with an isotype control antibody. Mice were then immunized and treated with α-GalCer. The administration of anti-CD20 antibody depleted over 98.9% of B cells in the spleens of mice (Fig. 3.4A). Of note, anti-CD20 treatment fails to deplete plasma cells, which lack CD20 expression (496, 497) and peritoneal cavity B cells, including B-1 cells (498). The peritoneum has been proposed to constitute a protective niche during anti-CD20 immunotherapy. In agreement with the results observed in µMT mice, α-GalCer treatment ameliorated disease in control but not in B cell-depleted mice (Fig. 3.4B). The frequencies and absolute numbers of iNKT cells were similar between B cell-depleted and control mice, with or without α-GalCer treatment (Fig. 3.4C). A significant increase in the expression of PLZF and IFN-γ in iNKT cells, following treatment with α-GalCer, was observed in control mice, but their upregulation was significantly lower in B cell-depleted mice (Fig. 3.5, A and B).

97

AB 40 30 control - vehicle B cell depleted - vehicle )

6 **** control - -GalCer B cell depleted - -GalCer s.c. mBSA-CFA 10

(%) 30 *** 60 60 20 20 B cells +

B cells ( 40 40 + 10 Day -21Day -14 Day -7 Day 0 10 CD19

CD19 * 20 20 i.a. mBSA-PBS or PBS 0 0 i.p. anti-CD20 or isotype control mAbs i.p. αGalCer or vehicle Increase in knee swelling Increase in knee swelling comparedcontrol to knee (%) comparedcontrol to knee (%) 0 0 iso controlanti-CD20 iso controlanti-CD20 0123 0123 C Day post disease onset Day post disease onset Gated on live CD19- cells Vehicle α-GalCer 5 4 1.22 3.15 ** ** ) 4 6 3 *

10 * 3

control 2 2 iNKT cells (%) cells iNKT

iNKT cells ( cells iNKT 1 1

0 0 α-GalCer -+-+α-GalCer -+-+ control B cell control B cell 1.25 3.55 depleted depleted B cell depleted CD1d-tetramer

CD3 Figure 3.4 B cell depletion abrogates the protective effect of α-GalCer on arthritis. (A) Left, schematic showing experimental design for B cell depletion. Right, bar charts showing the frequency and number of splenic CD19+ B cells in B cell depleted and isotype-control-treated WT mice prior to α-GalCer or vehicle administration (n=5 per group, one of three experiments is shown). (B) Mean clinical score of isotype-control- treated WT mice (left) and B cell depleted WT mice (right) that received α-GalCer or vehicle alone following induction of arthritis. Y axis shows percentage swelling in antigen-injected knee compared to control knee (n=5 per group, one of two experiments is shown). (C) Representative flow cytometry plots and bar charts showing the frequency and number of splenic iNKT cells in B cell depleted and isotype-control- treated WT mice that received α-GalCer or vehicle alone (n=5 per group, one of two experiments is shown). C analysed on day 3 post-disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (A Student’s t-test, B two-way ANOVA, C one-way ANOVA).

98

A B Gated on live CD19- Gated on live CD19-CD1d-tetramer+CD3int cells + int Vehicle α-GalCer CD1d-tetramer CD3 cells 500 50 **** **** ** 15.6 48.3 400 ** 40

B cell depleted - α-GalCer 300 30

(MFI) ***

B cell depleted - vehicle (%) cells iNKT

control 20

200 + PLZF control - α-GalCer 10

100 IFN-

control - vehicle 0 0 α-GalCer -+-+ α-GalCer -+-+ control B cell PLZF control B cell depleted depleted 13.6 26.6 B cell depleted γ IFN- CD1d-tetramer Figure 3.5. B cell depletion results in impaired induction of iNKT cell PLZF and IFN-γ in response to α-GalCer. (A) Representative histograms and bar chart showing MFI of PLZF in splenic iNKT cells in B cell depleted and isotype-control-treated WT mice that received α-GalCer or vehicle alone (n=3 per group, one of two experiments is shown). (B) Representative flow cytometry plots and bar chart showing the frequency of IFN-γ+ splenic iNKT cells in B cell depleted and isotype-control-treated WT mice that received α-GalCer or vehicle alone (n=4 per group, one of two experiments is shown). Analysed at 16h post-disease onset. Data are mean±s.e.m. **P<0.01, ***P<0.001, ****P<0.0001 (one-way ANOVA).

99 DCs are not required for α-GalCer-iNKT cell mediated suppression of arthritis As CD11c+ DCs have been shown to be the major CD1d-expressing cell population in iNKT cell activation in vivo (295), we next wanted to explore the role of DCs in the α-GalCer-mediated suppression of arthritis. We selectively depleted DCs to determine what effect they were having on iNKT cells in AIA. Diphtheria toxin was administered to mice that express the diphtheria toxin receptor (DTR) under the control of the Cd11c promoter (482).

Due to the central role that DCs play in the initiation of arthritis through the priming of pathogenic T cells, we had to change the protocol for α-GalCer administration. α-GalCer, was, in this instance, administered 8 hours after intra-articular injection of mBSA and intraperitoneal administration of diphtheria toxin. We confirmed CD11c+ DC depletion at the time of α-GalCer administration (Fig. 3.6A). α-GalCer mediated suppression of arthritis in CD11c+ cell-depleted mice was equivalent to control mice (Fig. 3.6B). Crucially, we did not observe significant iNKT cell expansion in response to α-GalCer in mice lacking CD11c+ DCs (Fig. 3.6C). iNKT cell upregulation of PLZF or IFN-γ, in response to α-GalCer was comparable in CD11c+ cell-depleted mice and controls (Fig. 3.7, A and B).

100 B A control - vehicle CD11c+ cell depleted - vehicle + i.p. α-GalCer or vehicle control - -GalCer CD11c cell depleted - -GalCer 1.5 1.5 60 60 s.c. mBSA-CFA ** ) 6 10 1.0 1.0 **** 40 40 cells (%)

+ ** cells ( + ** Day -7 Day 0 0.5 0.5 20 20 * CD11c CD11c i.a. mBSA-PBS or PBS Day 1 Increase in knee swelling Increase in knee swelling α-GalCer α-GalCer comparedcontrol to knee (%) i.p. DT or vehicle 0.0 0.0 comparedcontrol to knee (%) 0 0 e 01234 01234 DT DT Day post disease onset Day post disease onset vehicle vehicl C Gated on live CD19- cells Vehicle α-GalCer 8 **** 15 *** ***

) ** 6 6 0.93 4.50 10 10

4 control 5 iNKT cells (%) cells iNKT 2 iNKT cells ( cells iNKT

0 0 α-GalCer -+-+α-GalCer -+-+ control CD11c+ cell control CD11c+ cell depleted depleted 0.36 1.48 cell depleted + CD11c CD1d-tetramer

CD3 Figure 3.6. Dendritic cells are dispensable in α-GalCer-mediated amelioration of arthritis. (A) Left, schematic showing experimental design. Right, bar charts showing the frequency and number of splenic CD11c+ cells in CD11c+ cell depleted (DT administered) and control mice prior to α-GalCer or vehicle administration (n=3 per group, one of two experiments is shown). (B) Mean clinical score of CD11c+ cell depleted mice (right) and control mice (left) that received α-GalCer or vehicle alone, 8 hours following induction of arthritis. Y axis shows percentage swelling in antigen- injected knee compared to control knee, arrow indicates time of α-GalCer administration (n=5 per group, one of two experiments is shown). (C) Representative flow cytometry plots and bar charts showing the frequency and number of splenic iNKT cells in CD11c+ cell depleted and control mice that received α-GalCer or vehicle alone (n=5 per group, one of two experiments is shown). C analysed on day 3 post-disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (A Student’s t-test, B two-way ANOVA, C one-way ANOVA).

101

A B Gated on live CD19- Gated on live CD19-CD1d-tetramer+CD3int cells CD1d-tetramer+CD3int cells Vehicle α-GalCer 4000 60 * 17.3 48.3 **** **** ** 3000 CD11c+ cell depleted - α-GalCer 40 2000 CD11c+ cell depleted - vehicle control iNKT cells (%) cells iNKT + PLZF (MFI) PLZF 20 control - α-GalCer 1000 IFN- control - vehicle 0 0 α-GalCer -+-+ α-GalCer -+-+ PLZF control CD11c+ cell control CD11c+ cell depleted 14.2 43.7 depleted cell depleted + γ CD11c IFN-

CD1d-tetramer Figure 3.7. CD11c+ DC depletion does not impair iNKT cell upregulation of PLZF and IFN-γ in response to α-GalCer. (A) Representative histograms and bar chart showing MFI of PLZF in splenic iNKT cells from α-GalCer- or vehicle-treated CD11c+ cell depleted and control mice (n=3 per group, one of two experiments is shown). (B) Representative flow cytometry plots and bar chart showing the frequency of IFN-γ+ splenic iNKT cells from α-GalCer- or vehicle-treated CD11c+ cell depleted and control mice (n=3 per group, one of two experiments is shown). Analysed 16h post-disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01, ****P<0.0001 (one-way ANOVA).

102 B cells and DCs play different roles in iNKT cell activation in vitro To gain more understanding of the differential roles of B cells and DCs in presenting α-GalCer in mice with arthritis, we depleted B cells or DCs from splenocytes, isolated from arthritic mice, before stimulating them in vitro with α-GalCer and IL-2. Similarly to the in vivo results, only depletion of B cells led to a reduced PLZF and IFN-γ expression by iNKT cells, compared to both control and DC-depleted splenocytes. Depletion of DCs led to decreased expression of CD25 and CD69 by iNKT cells (Fig. 3.8). Thus, these data show that whereas B cells are responsible for driving the expression of PLZF and IFN-γ in iNKT cells, CD11c+ DCs are required for iNKT cell expansion and activation.

It has been previously shown that CD1d exclusion from lipid rafts leads to the induction of a type 2 response by iNKT cells, suggesting that lipid raft-associated CD1d may drive a type 1 response, with associated IFN-γ production (499). We therefore wanted to explore whether this could in part explain the ability for B cells and DCs to stimulate different iNKT cell response. We found that indeed while T2-MZP, MZ and Fo B cell CD1d was co-localizing with lipid rafts, in CD11c+ cells it was largely excluded (Fig. 3.9).

103 5000 40 5000 5000 * * ** 4000 ** ** 4000 4000 30 * * 3000 3000 3000 20

2000 iNKT cells (%) 2000 2000 + PLZF (MFI) PLZF CD25 (MFI) CD25 CD69 (MFI) CD69 10 1000 1000 1000 IFN-

0 0 0 0 Depletetion - B cell DC Depletetion - B cell DC Depletetion - B cell DC Depletetion - B cell DC Figure 3.8. B cells and dendritic cells play different roles in α-GalCer-iNKT cell activation in vitro. Bar charts showing the expression of PLZF, IFN-γ, CD25 and CD69 in iNKT cells from B cell-depleted, dendritic cell-depleted, and undepleted control splenocytes following in vitro treatment with α-GalCer and IL-2 (n=3 per group, one of two experiments is shown). Data are mean±s.e.m. *P<0.05, **P<0.01 (one-way ANOVA).

104 Brightfield CD1d CTB CD11c CD19 CD21 CD23 CD1d/CTB B cells T2-MZP cells MZ B cells Fo B + cells CD11c 1

0.98 1.10 0.80 0.63

0123 CD1d/CTB co-localization similarity feature analysis score Figure 3.9. CD19+ B cell subsets display higher CD1d association with lipid rafts than CD11c+ DCs. Representative images (top) and summary histograms (bottom) showing co-localization of CD1d (red) and lipid rafts by CTB (green) binding in T2- MZP B, MZ, and Fo B and CD11c+ cells in the spleens of WT mice, isolated from arthritic mice (generated using ImageStreamX; one of two experiments is shown). Analysed on day 3 post-disease onset.

105 Chapter IV: Results II

We have so far shown the importance of B cells in the α-GalCer-iNKT cell- mediated amelioration of arthritis. As discussed in the introduction, iNKT cell activation can occur in both CD1d-dependent and -independent manners. In this chapter, I address the mechanism of B cell interaction with iNKT cells.

To investigate whether CD1d on B cells was required for the suppressive effect of α-GalCer and the changes we have so far observed on iNKT cell phenotype, we generated mixed BM chimeric mice with B cell-restricted CD1d deficiency. We show the requirement for B cell CD1d expression in suppression of arthritis, as mice specifically lacking CD1d on B cells develop exacerbated arthritis to control group and fail to respond to α-GalCer treatment. B cells via CD1d induce the differentiation of PLZFhiIFN-γ+ iNKT cells. We also show that α-GalCer-mediated suppression of arthritis is independent of B cell IL-10 expression.

We demonstrate that uniquely in control chimeric mice α-GalCer administration leads to iNKT cell localisation in the T cell area where they are in contact with IgDhi B cells, suggesting that this is where iNKT-B interaction may occur in vivo. Furthermore, we show that CD1d+ B cells control GC formation. Mice lacking CD1d+ B cells have an increased number of GCs. When we treated WT mice with α-GalCer there was a profound decrease in the number of GCs, and this reduction was not observed in the absence of B cells expressing CD1d.

Finally, we address the directionality of iNKT-B cell interaction. We show that while B cells following in vitro culture with anti-CD40 are able to transfer protection to arthritic WT mice, as previously demonstrated (204), they lost this ability upon transfer to iNKT cell deficient mice. α-GalCer-activated iNKT cells from WT mice were able to suppress arthritis in both µMT and WT mice, suggesting that B cells empowered iNKT cell suppressive capacity, and not the other way around.

Note that immunofluorescence staining was performed in collaboration with Miss Diana E. Matei, and iNKT cell transfer experiments with Dr Anneleen Bosma. When experiments were performed in collaboration with Diana E. Matei or Anneleen Bosma, this is indicated in the figure legend.

106 B cells induce iNKT cells that limit inflammation via a CD1d-dependent mechanism To ascertain whether B cells interact with iNKT cells directly via CD1d, we generated mixed BM chimeric mice lacking CD1d on B cells (B-Cd1d-/-) and a group of control chimeras (B-WT) (Fig. 4.1A). B-Cd1d-/- mice developed exacerbated disease compared to B-WT mice, suggesting a role for endogenous lipid presentation by B cells in controlling the severity of arthritis (Fig. 4.1B).

To assess the importance of B cells expressing CD1d in the differentiation of iNKT cells that limit disease, we treated B-Cd1d-/- and B-WT mice with α-GalCer. Whilst α-GalCer suppressed disease development in B-WT mice compared to vehicle- treated B-WT mice, B-Cd1d-/- mice failed to respond to the treatment (Fig. 4.2A).

AIA is a T cell driven pathology, as originally demonstrated by the inability of T cell-depleted lymphocytes to induce arthritis upon transfer to C57BL/6 mice, compared to enhanced disease-inducing capacity when B cells were depleted under the same conditions (471). Furthermore, the production of pro-inflammatory IFN-γ and IL-17 by CD4+ T cells as well as the enhanced differentiation of Th1 and Th17 cells has been associated with exacerbation of AIA (knee swelling, accumulation of inflammatory cells in the affected joints, cartilage damage, bone erosion) in multiple reports published by our group (197, 247, 266). Th1 and Th17 cells are present in increased frequencies in the synovial infiltrate of mice with more severe disease and correlate with increased frequencies of IFN-γ+ and IL-17+ CD4+ cells in the both the spleen and the dLN (197, 247, 266). Based on these findings, we therefore assayed IFN-γ and IL-17 expressing CD4+ T cells in the spleen as an immunological readout for the peripheral inflammatory milieu. This also allowed us to confirm the clinical disease scores obtained from measuring joint swelling.

α-GalCer-induced inhibition of disease in B-WT mice was accompanied by a reduction in the frequency and absolute number of IFN-γ+ and IL-17+ CD4+ T cells (Fig. 4.2B). iNKT cell expansion and proliferation in response to α-GalCer were comparable in B-Cd1d-/- and B-WT mice (Fig. 4.2, C and D). However, upon α-GalCer treatment, we observed a significant decrease in PLZF expression by iNKT cells and a decrease in the frequency of IFN-γ+PLZF+ iNKT cells in B-Cd1d-/- compared to B-WT mice (Fig. 4.3, A and B). This allowed us to establish that iNKT cells expressing PLZF were also producing IFN-γ, indeed in B-WT mice these phenotypes were overlapping.

107 We had observed that CD11c+ cells were required for iNKT cell expansion in response to α-GalCer (Fig. 3.6D) and that B cells via CD1d induced iNKT cell upregulation of PLZF and IFN-γ.

It has been shown that iNKT cells can respond via the activation of different transcription factor programs. iNKT cell subsets analogous to Th1, Th2 and Th17 have been described, which are T-bet+PLZFlo, T-bet-PLZFhiGATA3+, and T-bet- PLZFintROR-γt+, respectively (500). High PLZF expression is reported to be a characteristic of iNKT2 cells, but there are two main caveats. First, these studies were conducted under steady-state conditions, but we observe the induction of PLZF upon α- GalCer treatment in arthritic animals. Second, each of the subsets produces a characteristic cytokine, with IFN-γ, IL-4 and IL-17 production defining respectively iNKT1, iNKT2 and iNKT17 cells (500).

To attempt to reconcile our observations with previously published data, we investigated the expression of T-bet and GATA3 in iNKT cells. We report a decrease in the MFI value of T-bet and an increase in GATA3 expression in iNKT cells from B- Cd1d-/- compared to B-WT mice in response to α-GalCer (Fig. 4.4, A and B). Therefore, it is likely that upon α-GalCer administration during an inflammatory response a unique iNKT1-like subset is induced. Collectively, these data show that CD1d expression by B cells is critical for the induction of PLZF+IFN-γ+ iNKT cells that restrain arthritogenic responses.

108 ABB-WT Donors B-Cd1d-/- 50 20% 80% μMT 20% WT 80% μMT -/- -/- Cd1d or Il10 40 * **

30

Recipients 800 cGy 800 cGy 20

WT B cell mouse Cd1d-/- or Il10-/- B cell mouse 10 -/- -/- B cells:WT B cells:Cd1d or Il10 Increase in knee swelling

Other cells: WT Other cells: WT comparedcontrol to knee (%) 0 0123 Arthritis induction Day post disease onset Figure 4.1. Chimeric mice lacking CD1d specifically on B cells develop exacerbated arthritis. (A) Schematic showing experimental design for the generation of B-Cd1d-/-, B-Il10-/- and B-WT mixed BM chimeric mice. (B) Mean clinical score of B-Cd1d-/- and B-WT chimeric mice following induction of arthritis. Y axis shows percentage swelling in antigen-injected knee compared to control knee (B-Cd1d-/- n=5, B-WT n=5, one of three experiments is shown). *P<0.05, **P<0.01 (B two-way ANOVA).

109

Gated on live CD4+ cells ABB-WT - vehicle B-Cd1d-/- - vehicle Vehicle α-GalCer 5 * 1.5 B-WT - α-GalCer B-Cd1d-/- - α-GalCer 0.50 0.19 0.40 0.05 50 50 4 * 1.0 40 40 3 T cells (%) T cells (%) + +

30 30 B-WT 2 CD4 CD4 + + 0.5 20 20 1 IL-17 *** IFN- 10 10 2.76 1.48 0 0.0 α-GalCer -+-+α-GalCer -+-+ Increase in knee swelling Increase in knee swelling B-WT B-Cd1d-/- B-WT B-Cd1d-/- comparedcontrol to knee (%) comparedcontrol to knee (%) 0 0 2.0 0.5 ) 0.62 0.14 0.66 0.32 ) 6 0123 0123 6 * 10 Day post disease onset Day post disease onset 10 0.4 -/- 1.5 0.3 Cd1d T cells ( 1.0 T cells ( B- + + 0.2 CD4 CD4 + 0.5 + 0.1 IL-17 IL-17 3.32 3.87 IFN- 0.0 0.0 IFN-γ α-GalCer -+-+α-GalCer -+-+ B-WT B-Cd1d-/- B-WT B-Cd1d-/-

C Gated on live CD19- cells D Gated on live CD19- Vehicle α-GalCer + int 15 4 CD1d-tetramer CD3 cells 30000 ** ** ) ** 6 3 * 10 10 -/- 20000 0.61 6.01 B-Cd1d - α-GalCer

B-WT 2 * B-Cd1d-/- - vehicle

5 Ki-67(MFI) 10000 iNKT cells (%) cells iNKT

iNKT cells ( cells iNKT 1 B-WT - α-GalCer

0 0 B-WT - vehicle 0 α-GalCer -+-+α-GalCer -+-+ α-GalCer -+-+ B-WT B-Cd1d-/- B-WT B-Cd1d-/- Ki-67 B-WT B-Cd1d-/- -/- 0.64 6.11 Cd1d B- CD1d-tetramer CD3 Figure 4.2. B cell specific CD1d is critical for α-GalCer-iNKT cell-amelioration of arthritis. (A) Mean clinical score of B-WT chimeric mice (left) and B-Cd1d-/- chimeric mice (right) that received α-GalCer or vehicle alone following induction of arthritis. Y axis shows percentage swelling in antigen-injected knee compared to control knee (B- Cd1d-/-±α-GalCer n=5 per group, B-WT±α-GalCer n=4 per group, one of three experiments is shown). (B) Representative flow cytometry plots and bar charts showing the frequency (top) and number (bottom) of splenic IFN-γ+ and IL-17+ CD4+ T cells in B-Cd1d-/- and B-WT chimeric mice that received α-GalCer or vehicle alone (n=3 per group, one of two experiments is shown). (C) Representative flow cytometry plots and bar charts showing the frequency and number of splenic iNKT cells in B-Cd1d-/- and B- WT mice that received α-GalCer or vehicle alone (n=3 per group, one of two experiments is shown). (D) Representative histograms and bar chart showing MFI of Ki-67 in splenic iNKT cells from α-GalCer- or vehicle-treated B-Cd1d-/- and B-WT mice (n=3 per group, one of two experiments is shown). B analysed on day 7 post- disease onset, C, D analysed on day 3 post-disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01, ***P<0.001 (A two-way ANOVA, B-D one-way ANOVA).

110

A B Gated on live Gated on live CD19-CD1d-tetramer+CD3int cells CD1d-tetramer+CD3int cells 1500 Vehicle α-GalCer 25 6.82 27.0 ** * 20 B-Cd1d-/- - α-GalCer ** 1000 ** 15 B-Cd1d-/- - vehicle iNKT cells (%) B-WT + 10

B-WT - α-GalCer (MFI) PLZF 500 PLZF + 5 B-WT - vehicle

0 IFN- 0 α-GalCer -+-+ α-GalCer -+-+ PLZF B-WT B-Cd1d-/- B-WT B-Cd1d-/- 5.19 9.56 -/- Cd1d B- γ IFN-

PLZF Figure 4.3. B cell specific expression of CD1d is required for iNKT cell upregulation of PLZF and IFN-γ in response to α-GalCer. (A) Representative histograms and bar chart showing MFI of PLZF in splenic iNKT cells from α-GalCer- or vehicle-treated B-Cd1d-/- and B-WT mice (n=3 per group, one of two experiments is shown). (B) Representative flow cytometry plots and bar chart showing the frequency of IFN-γ+PLZF+ splenic iNKT cells from α-GalCer- or vehicle-treated B-Cd1d-/- and B- WT mice (n=4 per group, one of two experiments is shown). Analysed at 16h post- disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01 (one-way ANOVA).

111 A Gated on live B Gated on live + int + int CD1d-tetramer CD3 cells 3000 *** CD1d-tetramer CD3 cells 500 *** * 400

-/- 2000 -/- B-Cd1d - α-GalCer B-Cd1d - α-GalCer 300

-/- B-Cd1d - vehicle B-Cd1d-/- - vehicle 200

T-bet (MFI) 1000 GATA3 (MFI) B-WT - α-GalCer B-WT - α-GalCer 100

B-WT - vehicle 0 B-WT - vehicle 0 α-GalCer -+-+ α-GalCer -+-+ T-bet B-WT B-Cd1d-/- GATA3 B-WT B-Cd1d-/- Figure 4.4. α-GalCer induces reciprocal T-bet and GATA3 expression in iNKT cells from B-Cd1d-/- and B-WT mice. Representative histograms and bar chart showing MFI of T-bet (A) and GATA3 (B) in splenic iNKT cells from α-GalCer- or vehicle-treated B-Cd1d-/- and B-WT mice (n=3 per group, one of two experiments is shown). Analysed at 16h post-disease onset. Data are mean±s.e.m. *P<0.05, ***P<0.001 (one-way ANOVA).

112 α-GalCer treatment leads to disruption of GCs in arthritic mice in a B cell CD1d- dependent manner It has been previously suggested that iNKT cells regulate the GC entry of B cells and that this effect is CD1d+ B cell dependent (456). We wanted to explore the GC response in our model. Antibodies against GL7, B220 and CD3 were used to delineate GCs, B and T cell areas, respectively. Analysis of GCs in chimeric mice revealed that amelioration of disease by α-GalCer in B-WT mice was accompanied by a decrease in the number of GCs compared to untreated group (Fig. 4.5A). This reduction was controlled by CD1d+ B cells, since α-GalCer-driven decreased in GCs was not observed in arthritic mice lacking CD1d+ B cells. We also observed that in WT mice, α-GalCer treatment led to a reduction in Tfh cells compared to the vehicle treated group, whereas iNKTfh cells were not significantly changed between the two groups (Fig. 4.5B).

113 A α B-WT - vehicle B-WT - -GalCer 15

10 ** ***

5 GCs (per 10 follicles)

0 B-Cd1d-/- - vehicle B-Cd1d-/- - α-GalCer α-GalCer -+-+ B-WT B-Cd1d-/-

GL7 BB220220 CD3 ) 4 B 1.5 20 1.5 ) 8 4 10 * * 10 15 6 1.0 1.0 T cells (%) + T cells ( + 10 4 iNKT cells (%) + CD4 iNKT cells ( + CD4 +

0.5 + 0.5 5 2 CXCR5 CXCR5 CXCR5 0.0 CXCR5 0 0.0 0 α-GalCer - + α-GalCer - + α-GalCer - + α-GalCer - + Figure 4.5. GC formation, Tfh and NKTfh cells are inhibited by α-GalCer treatment. (A) Representative immunofluorescence and bar chart showing the number of splenic GCs, defined as discrete GL7+ areas within B220+ follicles, in B-Cd1d-/- and B-WT chimeric mice that received α-GalCer or vehicle alone. Bar, 100 µM (n=3 per group, one of two experiments is shown). (B) Bar charts showing the frequency and number of splenic CXCR5+CD4+ T cells and CXCR5+ iNKT cells in WT mice that received α-GalCer or vehicle alone (n=3, one of two experiments is shown). Analysed on day 3 post-disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01, ***P<0.001 (A one-way ANOVA, B Student’s t-test). Immunofluorescence staining performed in collaboration with Diana E. Matei.

114 iNKT cells localise with IgDhi B cells in the T cell zone in arthritic B cell CD1d- sufficient mice

Where iNKT cells localize in the spleen or in other tissues has been addressed by a number of studies (386-388). However, in our model we not only introduce inflammation, but also administer a strong iNKT cell ligand. iNKT cell migration in response to α-GalCer during an inflammatory response has not been previously studied. We found that following α-GalCer treatment, iNKT cells were found in close proximity to IgDhi B cells in the T cell area of the spleen in B-WT mice, but not in B-Cd1d-/- mice (Fig. 4.6).

115

B-WT - vehicle B-WT - α-GalCer

B-Cd1d-/- - vehicle B-Cd1d-/- - α-GalCer

CD3 NK1.1NK1.1 IgDIgD Figure 4.6. iNKT cells localise in the T cell area in the spleens of arthritic B-WT mice treated with α-GalCer. Representative immunofluorescence showing the localisation of splenic NK1.1+CD3+ T cells and IgD+ B cells in B-Cd1d-/- and B-WT chimeric mice that received α-GalCer or vehicle alone; arrowheads indicate selected iNKT cells. Bar, 100 µM (n=3 per group, one of two experiments is shown). Analysed on day 3 post-disease onset. Immunofluorescence staining performed in collaboration with Diana E. Matei.

116 B cell subset frequencies are not affected by α-GalCer treatment in B cell CD1d- sufficient mice In 2010 Wingender et al. published a report showing the in vivo killing of MZ B cells and to a lesser extent CD11c+ and CD11b+ cells in a CD1d-dependent manner (501). MZ B cells are enriched for autoreactivity and associated with autoantibody production. An explanation for α-GalCer mediated suppression could therefore be iNKT cell-dependent killing of pathogenic CD1d-expressing MZ B cells. Investigating the various B cell populations in our model, we did not observe a reduction in the frequency in any of the subsets of B-WT mice with α-GalCer treatment (Fig.4.7A). Furthermore, we observed significant increases in the number of cells of each B cell subset in WT mice upon α-GalCer administration. We show that in B-Cd1d-/- mice, MZ B cell frequency is significantly reduced upon α-GalCer treatment. Importantly, there was no difference in CD1d MFI between vehicle and α-GalCer treated B-WT mice (Fig. 4.7B).

While this experiment does not directly address whether iNKT cell killing of B cells in a CD1d-dependent manner occurs in our model, if it does, there appears to be a mechanism that allows for the rapid repopulation of the niche with cells of equivalent CD1d expression.

117 A B-WT B-Cd1d-/- Vehicle α-GalCer Vehicle α-GalCer

cells 14.3 14.4 15.1 10.7 +

55.8 56.0 56.9 63.2

Gated on live CD19 13.1 15.2 13.2 12.4

CD24

+ cells hi CD24

hi 33.6 60.5 34.6 59.5 37.0 57.9 20.8 74.9 CD21 Gated on live CD19 CD21

CD23

20 15 8 80 ** * **** ** 15 6 **** 60 10

10 4 40 5 T1 B cells (%) Fo B cells (%) 5 cellsB (%) MZ 2 20 T2-MZP B cells (%)

0 0 0 0 α-GalCer -+-+α-GalCer -+-+α-GalCer -+-+α-GalCer -+-+ B-WT B-Cd1d-/- B-WT B-Cd1d-/- B-WT B-Cd1d-/- B-WT B-Cd1d-/- 2.0 **** 1.5 0.5 15 **** ** *** ) 6 ** *** ) ) ) 6 10

6 0.4 6 1.5 **** 10 10 10 ******** 1.0 10 **** 0.3 ** 1.0 ** 0.2 *** 0.5 5 **** T1 B cells ( 0.5 Fo B cells ( MZ B cells B ( MZ 0.1 T2-MZP B cells ( 0.0 0.0 0.0 0 α-GalCer -+-+α-GalCer -+-+α-GalCer -+-+α-GalCer -+-+ -/- B-WT B-Cd1d-/- B-WT B-Cd1d-/- B-WT B-Cd1d-/- B-WT B-Cd1d

B Gated on live CD19+ cells 1500 ****** 1000 **** B-Cd1d-/--α-GalCer 500 B-Cd1d-/--vehcile 0

B-WT-α-GalCer (MFI) CD1d -500 B-WT-vehcile -1000 α-GalCer -+-+ CD1d B-WT B-Cd1d-/- Figure 4.7. Effect of α-GalCer treatment on splenic B cell populations. (A) Representative flow cytometry plots (top) and bar charts (bottom) showing the frequency and number of splenic B cell subsets in B-Cd1d-/- and B-WT mice that received α-GalCer or vehicle alone (n=3 per group, one of two experiments is shown). (B) Representative histograms and bar chart showing MFI of CD1d in splenic B cells from α-GalCer- or vehicle-treated B-Cd1d-/- and B-WT mice (n=3 per group, one of two experiments is shown). Analysed at 24h post-disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (one-way ANOVA).

118 IL-10 is dispensable for α-GalCer-mediated amelioration of arthritis As Bregs are known to suppress via the production of IL-10, we assessed whether, in addition to B cell expression of CD1d, IL-10 was necessary for α-GalCer mediated suppression of arthritis. We generated mixed BM chimeric mice lacking IL- 10-producing B cells (B-Il10-/-) and WT (B-WT) chimeric mice (Fig. 4.1A). α-GalCer ameliorated disease in both B-Il10-/- and B-WT mice (Fig. 4.8A). We also found that α- GalCer-treatment ameliorated arthritis in global Il10ra-/- mice, just as in WT mice (Fig. 4.8B). These data demonstrate that B cell IL-10 is not required for α-GalCer mediated suppression of arthritis, and also that iNKT cell IL-10 is not critical for suppression observed with α-GalCer treatment.

119

B-WT - vehicle B-Il10-/- - vehicle A B-WT - -GalCer B-Il10-/- - -GalCer 60 60

40 40

* 20 ** 20 **** *** **** Increase in knee swelling Increase in knee swelling comparedcontrol to knee (%) 0 comparedcontrol to knee (%) 0 0123 0123 Day post disease onset Day post disease onset

B WT - vehicle Il10ra-/- - vehicle -/- - 60 WT - -GalCer 60 Il10ra -GalCer

40 40

**** **** **** 20 20 Increase in knee swelling Increase in knee swelling comparedcontrol to knee (%) comparedcontrol to knee (%) 0 0 0123 0123 Day post disease onset Day post disease onset Figure 4.8. α-GalCer-mediated suppression of arthritis is independent of IL-10. (A) Mean clinical score of B-Il10-/- chimeric mice (right) and B-WT chimeric mice (left) that received α-GalCer or vehicle alone following induction of arthritis; y axis shows percentage swelling in antigen-injected knee compared to control knee (B-Il10-/-±α- GalCer n=5 per group, B-WT±α-GalCer n=6 per group, one of two experiments is shown). (B) Mean clinical score of Il10ra-/- mice (right) and WT mice (left) that received α-GalCer or vehicle alone following induction of arthritis. Y axis shows percentage swelling in antigen-injected knee compared to control knee (Il10ra-/- n=6, WT n=8, one of two experiments is shown). Data are mean±s.e.m. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (two-way ANOVA).

120 iNKT cells, and not B cells, are the effectors of suppression in arthritis Having shown that CD1d-mediated interaction between B cells and iNKT cells helps suppress inflammatory responses, we next wanted to explore which of the two cell types, iNKT cells or B cells, was the effector of suppression.

To address this, we first stimulated B cells isolated from the spleens of WT mice in remission from arthritis with anti-CD40 in vitro. As discussed, it has been previously shown that this treatment expands Bregs and results in the induction of B cells capable of mediating suppression upon adoptive transfer (179, 196). We transferred these B cells into WT mice, as well as Jα18-/- and Cd1d-/- mice that lack iNKT cells and both type I and II NKT cells, respectively. Anti-CD40 stimulated B cells were suppressive upon transfer to WT mice, but failed to mediate suppression in iNKT cell deficient mice (Figure 4.9). This suggests that iNKT cells are required for B cells to mediate suppression of arthritis following adoptive transfer.

Secondly, we transferred iNKT cells isolated from α-GalCer- or vehicle-treated WT mice into WT mice. While iNKT cells from vehicle-treated mice suppressed arthritis in WT mice, the iNKT cell capacity to ameliorate disease was greatly enhanced, if the donor mice had received α-GalCer (Figure 4.10). Furthermore, α- GalCer-potentiated iNKT cells were also able to suppress disease in B cell deficient µMT mice.

Collectively these data show that while B cell suppression of disease in the acute phase requires the presence of iNKT cells, iNKT cells maintain their suppressive potential in B cell deficient environment.

121 Arthritis induction 60 **** **** *** WT * **** J 18-/- Splenic Adoptive 40 ** Cd1d-/- B cell transfer WT *** separation 20

WT anti-CD40 Jα18-/- 0 In remission stimulation 123

from Suppresionknee of swelling -20 arthritis

comparednocontroltransfer to (%) Day post disease onset Cd1d−/− Figure 4.9. Anti-CD40-stimulated B cells do not suppress arthritis in the acute phase upon adoptive transfer to iNKT cell deficient mice. Left, schematic showing experimental design. Right, bar chart showing suppression of knee swelling in Jα18-/-, Cd1d-/- and WT mice that received anti-CD40 stimulated B cells, isolated from the spleens of WT mice on day 7 post-disease onset, compared to Jα18-/-, Cd1d-/- and WT mice that received no transfer (n=3 per group, one of two experiments is shown). Data are mean±s.e.m. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (two-way ANOVA).

122 WT - no transfer * μMT - no transfer Arthritis WT - WT iNKT cell transfer (vehicle) • μMT - WT iNKT cell transfer (α-GalCer) vehicle Adoptiveinduction WT - WT iNKT cell transfer (α-GalCer) transfer 80 80

60 60 WT WT iNKT cell sort 40 40 α-GalCer • WT 20 **** 20 • • *** *** *** Increase in knee swelling Increase in knee swelling comparedcontrol to knee (%) WT comparedcontrol to knee (%) 0 0 0123 0123 μMT Day post disease onset Day post disease onset Figure 4.10. iNKT cells from donor mice treated with α-GalCer ameliorate arthritis upon transfer to B cell deficient and WT mice. Left, schematic showing experimental design. Middle, mean clinical score of WT mice that received iNKT cells isolated from the spleens of α-GalCer- or vehicle-treated WT mice on day 3 post- disease onset, or WT mice that received no transfer following induction of arthritis. Y axis shows percentage swelling in antigen-injected knee compared to control knee. Right, mean clinical score of µMT mice that received iNKT cells isolated from the spleens of α-GalCer-treated WT mice on day 3 post-disease onset, or µMT mice that received no transfer following induction of arthritis. Y axis shows percentage swelling in antigen-injected knee compared to control knee (n=4 per group, one of two experiments is shown). ŸP<0.05, ***P<0.001, ****P<0.0001 (two-way ANOVA). Experiment performed in collaboration with Anneleen Bosma.

123 CHAPTER V: Results III

So far we have shown that signals provided by B cells to iNKT cells lead to the induction of iNKT cells, characterized as PLZF+IFN-γ+. This B-iNKT cell interaction through CD1d is important in the suppression of arthritis, and iNKT cells are the effectors of this suppression. In this chapter we explore the characteristics of iNKT cells that enable their immunomodulatory function during an inflammatory response.

Through RNA-seq analysis we show that CD1d+ B cell lipid presentation controls iNKT cell expression of genes related to metabolism and effector function, including cytokine response. We show that the transcriptional program of PLZF is controlled by lipid presenting CD1d+ B cells, which is important as PLZF has a central role in establishing iNKT cell phenotype and function during differentiation and their maintenance in the periphery (320, 321).

We also provide mechanistic insight into how the balance in IFN-γ production by iNKT cells can differentially affect pro-inflammatory response by conventional CD4+ T cells. Indeed, we demonstrate that IFN-γ production by B cell-induced iNKT cells is partially responsible for the suppression of adaptive immune responses (measured as Th1 and Th17 response inhibition) in arthritic mice and amelioration of arthritis in response to α-GalCer.

Note that analysis of RNA-seq data was performed in collaboration with Dr Ignat Drozdov. When analysis was done in collaboration with Ignat Drozdov, this is indicated in the figure legend.

124 iNKT cell transcriptional profile is altered in the absence of CD1d+ B cells To gain insight into the potential mechanism underlying the suppressive capacity of iNKT cells conferred by B cell lipid presentation, we profiled the transcriptomes of iNKT cells sorted from α-GalCer- or vehicle-treated B-Cd1d-/- and B- WT chimeric mice using RNA-seq. iNKT cell purity was >95% (see Materials and Methods Fig. 2.1A). α-GalCer had similar effects on the majority of in iNKT cells from both B-Cd1d-/- and B-WT mice (Fig. 5.1A). However, 3826 genes were significantly differentially expressed in α-GalCer-treated B-Cd1d-/- and B-WT mice (p<0.05). Amongst these were those encoding immune-related genes essential for T and iNKT cell effector function (Fig. 5.1B). Of interest, Zbtb16, which encodes PLZF, was expressed at significantly higher levels in iNKT cells from control B-WT than from B-Cd1d-/- chimeric mice following α-GalCer treatment, corroborating the results shown in Fig. 4.3.

To understand aspects of iNKT cell function regulated by B cells presenting lipid antigens, we enriched differentially expressed genes from α-GalCer-treated B- Cd1d-/- versus B-WT mice for over-represented pathways. This analysis revealed gene sets encoding products involved in cell cycle, NK cell-mediated cytotoxicity as well as cytokine-cytokine receptor interaction (Fig. 5.1C).

125 AB 101 n=3094

100

-vehicle -1 -/- 10 p=0.05 510 10-2 Hif1a Cd1d Myc 10-3 Klrd1 Cxcr2 Irf4 10-4 Il2rb Tnfrsf4 P value Il18rap Fold change Klrb1f Zbtb16 Gzmb 10-5 Ccr2 Itga1 -GalCer vs B- Il18r1 Ccr7 α -5 0 Klrb1b Zbtb32 - -6 Il7r -/- 10 Il17rb Icosl Cxcr6 10-7 Ccr9 Cd1d Degree of change Klrb1c B- -10 n=2703 10-8 -10 -5 0 5 10 -5 0 5 Fold change log fold change B-WT-α-GalCer vs B-WT-vehicle C Chemokine signaling pathway Cell cycle NOD-like receptor signaling pathway Calcium signaling pathway MAPK signaling pathway Antigen processing and presentation ECM-receptor interaction NF-κB signaling pathway Phosphatidylinositol signaling system Toll-like receptor signaling pathway T cell receptor signaling pathway RNA transport Natural killer cell mediated cytotoxicity JAK-STAT signaling pathway Cytokine-cytokine receptor interaction -15 -10 -5 0 Pathway activity Figure 5.1. Expression of immune-related genes in mice lacking CD1d specifically on B cells following α-GalCer treatment. (A) Scatter plot showing fold change relationships between common expressed genes (n=5797) in iNKT cells from B-Cd1d-/- and B-WT chimeric mice that received α-GalCer or vehicle alone following induction of arthritis (data sets from RNA-seq). Colour gradient indicates the degree of difference in fold change values between iNKT cells in B-Cd1d-/- and B-WT mice following α- GalCer treatment. (B) Volcano plot showing fold changes of differentially expressed genes between iNKT cells from B-Cd1d-/- and B-WT chimeric mice following α-GalCer treatment, plotted against P values. (C) Pathway enrichment analysis of pathways (right) showing enrichment among genes expressed differentially in iNKT cells from B-Cd1d-/- and B-WT mice following α-GalCer treatment. Presented as net pathway perturbation in iNKT cells from B-Cd1d-/- mice compared to B-WT mice. Negative values mean inhibited in iNKT cells from B-Cd1d-/-, positive values mean activated in iNKT cells from B-Cd1d. Analysed at 16 h post-disease onset. Analysis performed in collaboration with Ignat Drozdov.

126 B cell inability to present α-GalCer affects iNKT cell expression of genes involved in cytokine responses and metabolism A substantial increase in the bioenergetic demands over the resting state is a defining feature of T cell activation (502, 503). Analysis of metabolic pathways showed that in α-GalCer-treated B-Cd1d-/- mice, iNKT cells were highly anabolic and presented an increase in genes encoding enzymes involved in and a reduction in those involved in fatty acid oxidation (FAO) compared to B-WT mice with α-GalCer treatment (Fig. 5.2A). In particular, crucial mediators of the metabolic switch in effector T cells, namely Irf4, Myc and Hif1a (502, 503), were significantly increased in B-Cd1d- /- compared to B-WT mice following α-GalCer treatment (Fig. 5.1B).

iNKT cells have been assigned, similarly to Th cells, into subsets according to their transcription factor expression and cytokine profile (504, 505). To further elucidate whether lipid presentation by B cells selectively skewed the response of iNKT cells towards a specific effector subset we compared the expression of cytokine genes by iNKT cells from B-Cd1d-/- and from control B-WT mice with or without α-GalCer treatment. The expression of most cytokines was enhanced by the absence of CD1d- presenting B cells in α-GalCer treated mice, suggesting that B cells have a different role in iNKT cell activation compared to other antigen-presenting cells. For instance, we found an increase of Il21, Il9 and Il10 in iNKT cells from B-Cd1d-/- compared to B-WT mice following α-GalCer treatment (Fig. 5.2B). We also found an increase of Il17a, however we could no detect any iNKT cell IL-17 by intracellular staining or ELISA.

127 A Glycolysis B Cytokines Vehicle α-GalCer Vehicle α-GalCer B-WTB-Cd1d-/- B-WT B-Cd1d-/- B-WTB-Cd1d-/- B-WT B-Cd1d-/- Aldoc Ebi3 Pgam2 Il15 Bpgm Il18 Pfkm Il1b Pgm1 Ltb Hk1 Tgfb3 Galm Il17c Pklr Il23a Eno2 Il12a Adpgk Il12b Pik3ca Il22 Hk3 Il17d Pgm2 Il9 Mtor Il27 Eno3 Il10 Hif1a Il17a Slc2a1 Il21 Eno1 Ifng Gpi1 Il13 Pfkl Il2 Hk2 Lta Ldha Il5 Pgam1 Tgfb2 Gapdh Tgfb1 Pgk1 Il4 Pgm3 Tnf Aldoa Row min Row max Myc Expression (Relative) Pfkp Pkm Tpi1 Fatty acid oxidation Vehicle α-GalCer B-WTB-Cd1d-/- B-WT B-Cd1d-/- Hadh Cyp2e1 Acox1 Acad1 Cpt2 Ucp2 Cpt1a Acss1 Ppara Acadm

Row min Row max Expression (Relative) Figure 5.2. B cell inability to present α-GalCer affects iNKT cell expression of genes involved in cytokine responses and metabolism. (A) Heatmaps showing expression of genes involved in glycolysis (top) and fatty acid oxidation (bottom) by iNKT cells from B-Cd1d-/- and B-WT mice that received α-GalCer or vehicle alone. (B) Heatmap showing expression of cytokine genes by iNKT cells from B-Cd1d-/- and B- WT mice that received α-GalCer or vehicle alone. Analysed at 16h post-disease onset.

128 B cell CD1d is involved in the induction of genes associated specifically with NKT1 phenotype and PLZF transcriptional control The transcriptional profiles of iNKT cells isolated from B-Cd1d-/- and B-WT mice treated with α-GalCer or vehicle control did not overlap with those previously assigned to sorted thymic NKT1, NKT2 and NKT17 cells (Fig. 5.3) (505). However, at transcriptome-wide level changes have to be extensive to establish similarity. In addition, in our experiments, iNKT cells were obtained from the spleens of mice during acute inflammation, rather than from the thymus at steady state as previously reported (505).

Therefore, we compared the transcripts that were identified as most significantly enriched in NKT1, NKT2, and NKT17 cells relative to the other subsets. Among significant genes differentially expressed, there was an overall skew away from NKT1 and towards NKT2 phenotype in B-Cd1d-/- mice compared to B-WT mice after α- GalCer treatment (Fig. 5.4A). This change in transcriptional profile was consistent with the results reported in Fig. 4.4 showing a lack of upregulation of T-bet, and an upregulation of GATA3 expression in B-Cd1d-/- mice compared to B-WT mice following α-GalCer treatment. A low number of genes, previously associated with NKT17 cells were found to be significantly expressed in iNKT cells from B-Cd1d-/- mice compared to B-WT mice following α-GalCer treatment.

We have consistently shown that upon α-GalCer treatment the expression of PLZF by iNKT cells from µMT, B cell-depleted or B-Cd1d-/- mice is reduced compared to the control groups. Therefore, we compared our α-GalCer treated data sets to 99 genes previously identified as PLZF targets and found that 39 of these transcripts were significantly differentially expressed, with 36 down-regulated in the absence of B cells expressing CD1d. This is consistent with an induction of an iNKT cell PLZF signature supported by B cell CD1d. Of interest, Bach2, a known repressor of effector differentiation in naïve T cells (506), and previously shown to be bound to and repressed by PLZF, was significantly upregulated in iNKT cells of B-Cd1d-/- mice compared to B-WT mice after α-GalCer administration (Fig. 5.5B). Thus, iNKT cell suppression of arthritis is dependent on B cell-driven calibration of iNKT cell cytokine response, as well pathways in iNKT cells associated with effector function and metabolism.

129 B-Cd1d-/- NKT2 α-GalCer

25000 B-WT α-GalCer

NKT1 0 B-Cd1d-/- vehicle PC2 -25000 B-WT vehicle NKT17 -50000

-100000 -50000 0 50000 100000 PC1 Figure 5.3. Comparison of splenic iNKT cell gene expression to NKT1, NKT2 and NKT17 published data sets. Principal component analysis of transcripts in iNKT cells from B-Cd1d-/- and B-WT chimeric mice that received α-GalCer or vehicle alone and published data sets for NKT1, NKT2 and NKT17 cells across experimental replicates. Analysed at 16h post-disease onset. Analysis performed in collaboration with Ignat Drozdov.

130 AB NKT1 PLZF target genes 16 16

12 12

-log (P value) -log (P 8 value) -log (P 8

4 4

-2 0 2 -2 0 2 Fold change Fold change B-Cd1d-/--α-GalCer vs B-WT-α-GalCer B-Cd1d-/--α-GalCer vs B-WT-α-GalCer

NKT2 16

12 -log (P value) -log (P 8

4

-2 0 2 Fold change B-Cd1d-/--α-GalCer vs B-WT-α-GalCer

NKT17 16

12 -log (P value) -log (P 8

4

-2 0 2 Fold change B-Cd1d-/--α-GalCer vs B-WT-α-GalCer Figure 5.4. B cell CD1d controls the induction of transcriptional programs associated with NKT1 cells and PLZF regulation. (A) Volcano plots showing fold changes of differentially expressed genes associated with NKT1 (top), NKT2 (middle), NKT17 (bottom) phenotypes, and (B) PLZF transcriptional regulation, between iNKT cells from B-Cd1d-/- and B-WT mice following α-GalCer treatment, plotted against P values. Analysed at 16 post-disease onset. Analysis performed in collaboration with Ignat Drozdov.

131 α-GalCer induced IFN-γ inhibits Th1 and Th17 cells in arthritis Through their rapid production of cytokines, iNKT cells can regulate other cells and act as a bridge between innate and adaptive immunity. We had observed inhibition of Th1 and Th17 responses associated with increased IFN-γ production by iNKT cells, and it has been previously demonstrated that IFN-γ can be protective in autoimmunity, including in AIA (507). Thus, we investigated whether α-GalCer-CD1d-B cell dependent protection is mediated by the iNKT cell IFN-γ production. We neutralized IFN-γ in vivo as previously published (459), prior to α-GalCer or vehicle treatment, aiming to block α-GalCer-driven iNKT cell IFN-γ. Control mice treated with α-GalCer, which were protected from disease (Fig. 5.5A), displayed significantly lower adaptive IFN-γ and IL-17 production compared to control mice that received vehicle (Fig. 5.5B). In contrast, neutralization of IFN-γ reduced the suppressive effect of α-GalCer on clinical disease and prevented α-GalCer treatment from reducing the frequency and number of Th1 and Th17 cells. This supports the notion that innate production of IFN-γ elicited by α-GalCer has a role in the protection from arthritis and inhibition of adaptive responses.

To understand whether iNKT cell IFN-γ, induced by CD1d+ B cells, plays a role in suppressing Th1 and Th17 cells, we used an in vitro suppression assay in which splenocytes, isolated from arthritic µMT or WT mice, were challenged with anti-CD3 (to activate CD4+ T cells) and α-GalCer in the presence of anti-IFN-γ. We observed significant down-regulation in the expression of T-bet and ROR-γt (transcription factors for Th1 and Th17 cells respectively) in CD4+ T cells from WT mice, but not in CD4+ T cells from µMT mice after α-GalCer stimulation (Fig. 5.6). Neutralization of IFN-γ abolished the α-GalCer-driven suppression of T-bet and ROR-γt in WT CD4+ T cells. No significant effect on GATA3 expression in CD4+ T cells from WT or µMT mice by α-GalCer was observed. To ascertain whether iNKT cells can directly inhibit pathogenic T cells, iNKT cells were isolated from either µMT or from WT arthritic mice treated with α-GalCer and cultured with CD4+ T cells isolated from arthritic WT mice. WT iNKT cells significantly inhibited the differentiation of WT IFN-γ+ CD4+ T cells, whereas iNKT cells from µMT mice failed to show any suppressive capacity (Fig. 5.7).

Given that blocking IFN-γ, both in vivo and in vitro, ablated the ability of α- GalCer stimulated iNKT cells to suppress Th1 and Th17 responses, we next confirmed whether IFN-γ could directly suppress T-bet and ROR-γt in vitro. We stimulated CD4+ T cells isolated from spleens of mice with arthritis with anti-CD3 and increasing doses

132 of IFN-γ. Both the expression of T-bet and ROR-γt was significantly reduced as the concentrations of IFN-γ were increased (Fig. 5.8). Taken together these results support the notion that B cell CD1d-induced innate production of IFN-γ by iNKT cells has a role in the suppression of inflammation and inhibition of adaptive Th1/Th17 responses.

133

control - vehicle A control - -GalCer anti-IFN- - vehicle control anti-IFN- - -GalCer anti-IFN-γ 60 100 ** 80 *** 40 **** 60

40 20 **** **** **** 20 Increase in knee swelling Suppressionknee of swelling comparedcontrol to knee (%) 0 0 0234 comparedvehicle to treated(%) 234 Day post disease onset Day post disease onset B 4

2.5 1.5 ) 2.0 ) 6 *** 6 10 2.0 10 * 3 * 1.5 ** * 1.0 1.5 * T cells (%) T cells (%) + + T cells ( 2 T cells ( 1.0 + + 1.0 CD4 CD4 +

+ 0.5 CD4 CD4 + 1 + 0.5 0.5 IL-17 IFN- IL-17 IFN- 0.0 0.0 0 0.0 α-GalCer -+-+α-GalCer -+-+α-GalCer -+-+α-GalCer -+-+ control anti-IFN-γ control anti-IFN-γ control anti-IFN-γ control anti-IFN-γ Figure 5.5. α-GalCer amelioration of arthritis is partially dependent on IFN-γ production by iNKT cells. (A) Left, mean clinical score of anti-IFN-γ-treated and isotype-control-treated WT mice that received α-GalCer or vehicle alone following induction of arthritis. Y axis shows percentage swelling in antigen-injected knee compared to control knee. Right, bar chart showing suppression of knee swelling in anti-IFN-γ-treated and isotype-control-treated WT mice that received α-GalCer, compared to anti-IFN-γ-treated and isotype-control-treated WT mice that received vehicle alone (anti-IFN-γ-treated WT±α-GalCer n=5 per group, isotype-control-treated WT±α-GalCer n=4 per group, one of two experiments is shown). (B) Bar charts showing the frequency and number of splenic IFN-γ+ and IL-17+ CD4+ T cells in anti- IFN-γ-treated and isotype-control-treated WT mice that received α-GalCer or vehicle alone (n=3 per group, one of two experiments is shown). B analysed on day 3 post- disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01, ***P<0.001 (A two-way ANOVA, B one-way ANOVA).

134

WT µMT * 1.5 2.0 **** 0.4 *

* 1.5 0.3 1.0 T cells (%) T cells (%) + + T cells (%) + 1.0 0.2 CD4 CD4 + +

CD4 * t

+ 0.5 0.5 0.1 T-bet ROR- GATA3 0.0 0.0 0.0 anti-CD3 + + + + + + anti-CD3 + + + + + + anti-CD3 + + + + + + α-GalCer - + + - + + α-GalCer - + + - + + α-GalCer - + + - + + anti-IFN-γ - - + - - + anti-IFN-γ - - + - - + anti-IFN-γ - - + - - + Figure 5.6. B cells are required for α-GalCer-driven and IFN-γ-mediated suppression of T-bet and ROR-γt expression in CD4+ T cells. Bar charts showing the frequency of T-bet+, ROR-γt+, and GATA3+ CD4+ T cells in µMT and WT splenocyte culture with α-GalCer in the presence or absence of anti-IFN-γ blocking antibodies (µMT n=3, WT n=6, one of two experiments is shown). Data are mean±s.e.m. *P<0.05, ****P<0.0001 (two-way ANOVA).

135 3 α-GalCer in vivo * **** *** 2 T cells (%) WT iNKT + cell sort CD4

+ 1 IFN- Suppression assay μMT 0 with CD4+CD25- T cells T cells +++ in vitro WT iNKT cells - + - μMT iNKT cells - - + Figure 5.7. Following α-GalCer treatment in vivo iNKT cells from WT, but not B cell deficient mice, are suppressive in vitro. Left, schematic showing experimental design. Right, bar chart showing the frequency of IFN-γ+CD4+ T cells following co- culture with in vivo α-GalCer stimulated iNKT cells isolated from the spleens of µMT and WT mice (n=4 per condition, one of two experiments is shown). Data are mean±s.e.m. *P<0.05, ***P<0.001, ****P<0.0001 (one-way ANOVA).

136 400 2000 ** ** 1500 300

1000 t (MFI) 200 T-bet (MFI) 500 ROR- 100

0 0 0 0 10 50 10 50 100 500 100 500 IFN- (IU/ml) IFN- (IU/ml) Figure 5.8. Exogenous IFN-γ regulates expression of T-bet and ROR-γt in CD4+ T cells. Bar charts showing MFI of T-bet (left) and ROR-γt (right) in CD4+ T cells, isolated from the spleens of WT mice, following stimulation with increasing concentrations of IFN-γ (n=3 per condition, one of two experiments is shown). Data are mean±s.e.m. **P<0.01 (one-way ANOVA).

137 CHAPTER VI: Results IV

In the previous chapter I addressed how B cell CD1d-instructed iNKT cells may regulate the inflammatory response in arthritis. As we had observed the critical role for B cells in the α-GalCer-iNKT amelioration of arthritis, we next wanted to address whether a particular B cell subset was responsible for this effect.

We performed a series of adoptive transfer and selective depletion experiments showing that the amongst B cells that express high levels of CD1d, T2-MZP, previously ascribed with regulatory function, and not MZ B cells are responsible for the differentiation of this novel iNKT cell subset that protects from arthritis.

138 T2-MZP Bregs via CD1d induce the differentiation of iNKT cells that limit inflammation Among splenic APC subsets T2-MZP and MZ B cells express the highest levels of CD1d (Fig. 6.1A). Furthermore, T2-MZP B cell expression of the marker can be as high as that of MZ B cells (Fig.6.1B). Indeed CD1dhi B cells are composed of equal frequencies of T2-MZP and MZ B cells. However, only T2-MZP, and not MZ or Fo B cells, isolated from mice in remission from arthritis, conferred protection upon adoptive transfer to arthritic WT mice (Fig. 6.1C). These results are consistent with previously published reports (204, 205, 266).

Based on our findings showing that B cells are important in mediating suppression conferred by α-GalCer, we hypothesized that T2-MZP B cells, induced the differentiation of iNKT cells that ameliorate disease through suppression of Th1 and Th17 responses.

To address this hypothesis we adoptively transferred T2-MZP B cells (see Materials and Methods Fig. 2.1B), isolated from the spleens of WT mice in remission from arthritis, into WT mice and Cd1d-/- mice that lack both iNKT cells and type II NKT cells (307, 479). As previously shown, T2-MZP Bregs suppressed arthritis and reduced the frequency and absolute number of IFN-γ+ and IL-17+ CD4+ T cells in WT mice, however WT T2-MZP Bregs were unable to convey the same suppressive effect in iNKT cell deficient mice (Fig. 6.2, A and B).

To confirm that CD1d plays a functional role in T2-MZP Breg suppressive function and in the differentiation of disease-ameliorating iNKT cells, we isolated T2- MZP Bregs from the spleens of WT, Jα18-/- and Cd1d-/- mice in remission from arthritis and transferred them into WT mice. T2-MZP Bregs from WT and Jα18-/- mice (T2- MZP Bregs express CD1d in both strains) suppressed arthritis and significantly inhibited IFN-γ+CD4+ T cells in WT recipient mice (Fig. 6.3, A and B). A trend in the reduction of IL-17+CD4+ T cells was also observed. In contrast, T2-MZP Bregs from Cd1d-/- mice failed to inhibit arthritis or to limit Th1/Th17 responses in the recipient WT mice (Fig. 6.4, A and B). Together these data demonstrate that suppression of arthritis by T2-MZP Bregs requires both iNKT cells and the expression of CD1d.

139 A Gated on live cells 8000 eosinophils neutrophils 6000 macrophages resident monocytes inflammatory monocytes 4000 pDC cDC CD11b-CD11c+

cDC CD11b+CD11c+ (MFI) CD1d 2000 Fo MZ T2-MZP 0 CD1d MZ Fo -MZP + cDC+ cDC pDC 2 c T onocytes neutrophilseosinophils t m + CD11c- CD11 macrophages

residen CD11bCD11b inflammatory monocytes B Gated on live CD19+ Gated on indicated Gated on live CD19+ cells CD21hiCD24hicells B cell subsets in CD19+ cells 100 CD1dhi ** T2-MZP and MZ MZ T2-MZP CD1dlo**** 80 ****

cells(%) +

MZ T2-MZP lo CD19 Fo 60 **** **** **** /CD1d

hi **** 40 /CD1d + Frequancy among 20 Fo * CD1d CD21

CD19 0

Fo CD24 CD23 CD23 MZ T2-MZP

C 100 ••••°°°° No transfer 80 T2-MZP * MZ • 60 •• Fo ° 40 * * 20 Increase in knee swelling

comparedcontrol to knee (%) 0 0123 Day post disease onset Figure 6.1. T2-MZP B cells express high levels of CD1d and suppress arthritis upon adoptive transfer. (A) Representative histograms and bar chart showing MFI of CD1d in APCs the spleens of arthritic WT mice (n=3 per group, one of four experiments is shown). (B) Representative flow cytometry plots showing the expression of CD1d in Fo, MZ and T2-MZP B cells in the spleens of arthritic WT mice, and bar chart showing the frequency of Fo, MZ and T2-MZP B cells in CD1dhiCD19+, CD1dloCD19+ or total CD19+ B cells in the spleens of WT mice (n=3, one of five experiments is shown). (C) Mean clinical score of WT mice that received T2-MZP, MZ or Fo B cells, isolated from the spleen of WT mice on day 7 post-disease onset, or WT mice that received no transfer following induction of arthritis. Y axis shows percentage swelling in antigen-injected knee compared to control knee (n=4 per group, one of two experiments is shown). A, B analysed on day 3 post-disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01, ****P<0.0001, °°P<0.01, °°°°P<0.0001, °°°°P<0.0001 (B, C two-way ANOVA).

140 A WT 80 Cd1d-/- Adoptive **** transfer 60 ** WT ** 40 T2-MZP WT 20 In remission B cell sort Cd1d-/- from 0 arthritis Arthritis

induction comparedno(%)transfer to 123 Suppresionknee of swelling -20 Day post disease onset B 3 2.5 4 3 ) ) 6 * 6 * 10 2.0 10 3 ** 2 2 1.5 T cells (%) T cells (%) + + T cells ( 2 T cells ( + 1.0 + CD4 CD4

+ 1 1 + CD4 CD4 + 1 + 0.5 IL-17 IFN- IFN- IL-17 0 0.0 0 0 WT T2-MZP -+-+WT T2-MZP -+-+WT T2-MZP -+-+WT T2-MZP -+-+ WT Cd1d-/- WT Cd1d-/- WT Cd1d-/- WT Cd1d-/- Figure 6.2. T2-MZP Bregs mediate suppression via iNKT cells. (A) Left, schematic showing experimental design. Right, bar chart showing suppression of knee swelling in Cd1d-/- and WT mice that received T2-MZP B cells, isolated from the spleens of WT mice on day 7 post-disease onset, compared to Cd1d-/- and WT mice that received no transfer (Cd1d-/- n=11, WT n=16, one of two experiments is shown). (B) Bar charts showing the frequency and number of splenic IFN-γ+ and IL-17+ CD4+ T cells in Cd1d- /- and WT mice that received T2-MZP B cells, isolated from the spleens of WT mice on day 7 post-disease onset, or Cd1d-/- and WT mice that received no transfer (n=3 per group, one of two experiments is shown). B analysed on day 3 post-disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01, ****P<0.0001 (A two-way ANOVA, B one-way ANOVA).

141 WT - no transfer A WT - WT T2-MZP B cells * Adoptive WT - J 18-/- T2-MZP B cells • transfer 80

WT WT 60

40 ••• •• T2-MZP -/- *** Jα18 B cell sort WT 20 In remission Arthritis ** *

from induction Increase in knee swelling

comparedcontrol to knee (%) 0 arthritis 01234 Day post disease onset B

8 2.0 8 2.5 ) ** ) 6 ** 6 10 10 2.0 * 6 1.5 6 1.5 T cells (%) T cells (%) + + T cells ( 4 1.0 4 T cells ( + + 1.0 CD4 CD4 + + CD4 CD4 + 2 0.5 2 + 0.5 IL-17 IFN- IL-17 IFN- 0 0.0 0 0.0 Recipient WT WT WT Recipient WT WT WT Recipient WT WT WT Recipient WT WT WT Donor -WTJα18-/- Donor -WTJα18-/- Donor -WTJα18-/- Donor -WTJα18-/- T2-MZP T2-MZP T2-MZP T2-MZP Figure 6.3. T2-MZP Bregs from CD1d-sufficient Jα18-/- mice suppress arthritis upon adoptive transfer. (A) Left, schematic showing experimental design. Right, mean clinical score of WT mice that received T2-MZP B cells, isolated from the spleens of Jα18-/- or WT mice on day 7 post-disease onset, or WT mice that received no transfer following induction of arthritis. Y axis shows percentage swelling in antigen-injected knee compared to control knee (Jα18-/- T2-MZP transfer n=4, WT T2-MZP transfer n=3, no transfer n=4, one of two experiments is shown). (B) Bar charts showing the frequency and number of splenic IFN-γ+ and IL-17+ CD4+ T cells in WT mice that received T2-MZP B cells, isolated from the spleens of Jα18-/- or WT mice on day 7 post-disease onset, or WT mice that received no transfer (n=4 per group, one of two experiments is shown). B analysed on day 3 post-disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01, ***P<0.001, ŸŸP<0.01, ŸŸŸP<0.001 (A two-way ANOVA, B one- way ANOVA).

142 WT - no transfer A WT - WT T2-MZP B cells Adoptive WT - Cd1d -/- T2-MZP B cells transfer 80

WT WT 60

40

T2-MZP **** -/- Cd1d B cell sort WT 20 **** In remission Arthritis **** from induction Increase in knee swelling

comparedcontrol to knee (%) 0 arthritis 0123 Day post disease onset B

6 1.0 4 1.5 ) ) 6 ** * 6 10 0.8 10 3 4 1.0 * 0.6 T cells (%) T cells (%) + + T cells ( 2 T cells ( + + 0.4 CD4 CD4 +

2 + 0.5 CD4 CD4 + 1 + 0.2 IL-17 IFN- IL-17 IFN- 0 0.0 0 0.0 Recipient WT WT WT Recipient WT WT WT Recipient WT WT WT Recipient WT WT WT Donor -WTCd1d-/- Donor -WTCd1d-/- Donor -WTCd1d-/- Donor -WTCd1d-/- T2-MZP T2-MZP T2-MZP T2-MZP Figure 6.4. T2-MZP Bregs from CD1d-deficient mice fail to suppress arthritis upon adoptive transfer. (A) Left, schematic showing experimental design. Right, mean clinical score of WT mice that received T2-MZP B cells, isolated from the spleens of Cd1d-/- or WT mice on day 7 post-disease onset, or WT mice that received no transfer following induction of arthritis. Y axis shows percentage swelling in antigen- injected knee compared to control knee (n=6 per group, one of two experiments is shown). (B) Bar charts showing the frequency and number of splenic IFN-γ+ and IL-17+ CD4+ T cells in WT mice that received T2-MZP B cells, isolated from the spleens of Cd1d-/- or WT mice on day 7 post-disease onset, or WT mice that received no transfer (n=4 per group, one of two experiments is shown). B analysed on day 3 post-disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01, ****P<0.0001 (A two-way ANOVA, B one-way ANOVA).

143 α-GalCer mediated suppression does not require MZ B cells In addition to T2-MZP, MZ B cells also express high levels of CD1d (Fig. 6.1, A and B) and at least in vitro have been shown to exert suppression (221). We have previously reported and demonstrated in Fig. 6.1C that adoptively transferred MZ B cells are unable to suppress arthritis in the recipient mice (204), which is due to their inability to re-populate the spleen (P.A. Blair, unpublished results). To circumvent this issue and to ascertain the role that MZ B cells play in the differentiation of anti- arthritogenic iNKT cells in vivo we depleted MZ B cells in WT mice, by administering anti-CD11a and anti-CD49d antibodies (483, 484) prior to the induction of arthritis and treatment with α-GalCer. Isotype-treated mice were used as controls. No changes in the number of T2-MZP B cells were observed in MZ B cell-depleted compared to control mice (Fig. 6.5). α-GalCer significantly reduced disease both in the control group and in MZ B cell depleted mice (Fig. 6.6A). The reduction in disease severity was accompanied by iNKT cell expansion and increased expression of Ki-67 (Fig. 6.6, B and C). The induction of PLZF and IFN-γ expression in iNKT cells from both MZ B cell-depleted and control mice was also comparable (Fig. 6.7, A and B). These results, together with those shown in Fig. 4.6, demonstrating that during inflammation upon α- GalCer treatment iNKT cells relocate to the T cell area, suggest that MZ B cells are not required for iNKT cell mediated suppression of arthritis.

144 Isotype control anti-CD11a Iso control mAbs anti-CD49d anti-CD11a+anti-CD49d WT WT 1.0 5

* ) 6 ) 6 16h 0.8 10 4 10 0.6 3

0.4 2 B cellsB (

MZ B cell MZ 0.2 1

WT T2-MZP B cells ( depleted WT Arthritis induction 0.0 0 Figure 6.5. Anti-CD11a and anti-CD49d antibody treatment depletes MZ, but not T2-MZP B cells. Left, schematic showing experimental design. Right, bar charts showing the number of splenic MZ and T2-MZP B cells in anti-CD11a- and anti- CD49d-treated mice and isotype-control-treated mice prior to α-GalCer or vehicle administration (n=6 per group, one of two experiments is shown). Data are mean±s.e.m. *P<0.05 (Student’s t-test).

145 A control - vehicle MZ B cell depleted - vehicle control - -GalCer MZ B cell depleted - -GalCer 60 60

40 40

* *** **** * 20 **** 20 * Increase in knee swelling Increase in knee swelling comparedcontrol to knee (%) comparedcontrol to knee (%) 0 0 0123 0123 Day post disease onset Day post disease onset B Gated on live CD19- cells Vehicle α-GalCer 15 15 * ** ) 0.71 9.92 6

10 10 10 control 5 5 iNKT cells (%) cells iNKT iNKT cells ( cells iNKT

0 0 α-GalCer -+-+α-GalCer -+-+ control MZ B cell control MZ B cell depleted depleted 1.20 5.41 MZ B cell depleted CD1d-tetramer

CD3

C Gated on live CD19- 80000 CD1d-tetramer+CD3int cells 60000 *

MZ B cell depleted - α-GalCer 40000 * Ki-67(MFI) MZ B cell depleted - vehicle 20000

control - α-GalCer 0 control - vehicle α-GalCer -+-+ control MZ B cell Ki-67 depleted Figure 6.6. MZ B cells are dispensable for α-GalCer-mediated amelioration of arthritis. (A) Mean clinical score of control mice (left) or MZ B cell-depleted mice (right) that received α-GalCer or vehicle alone following induction of arthritis. Y axis shows percentage swelling in antigen-injected knee compared to control knee (α- GalCer-treated n=5 per group, vehicle-treated n=6 per group, one of two experiments is shown). (B) Representative flow cytometry plots and bar charts showing the frequency and number of splenic iNKT cells in MZ B cell-depleted mice and control mice that received α-GalCer or vehicle alone (n=3 per group, one of two experiments is shown). (C) Representative histograms and bar chart showing MFI of Ki-67 in splenic iNKT cells from α-GalCer- or vehicle-treated MZ B cell-depleted mice and control mice (n=3 per group, one of two experiments is shown). B, C analysed on day 3 post-disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (A two- way ANOVA, B, C one-way ANOVA).

146 Gated on live CD19- + int A CD1d-tetramer CD3 cells 6000 **

MZ B cell depleted - α-GalCer 4000 *

MZ B cell depleted - vehicle

PLZF (MFI) PLZF 2000 control - α-GalCer

control - vehicle 0 α-GalCer -+-+ PLZF control MZ B cell depleted

B Gated on live CD19-CD1d-tetramer+CD3int cells Vehicle α-GalCer 60 *** *** 22.6 42.1

40 iNKT cells (%) control + 20 IFN-

0 α-GalCer -+-+ control MZ B cell 26.3 46.7 depleted MZ B cell depleted γ IFN-

CD1d-tetramer Figure 6.7. MZ B cells are dispensable for α-GalCer-mediated induction of PLZF and IFN-γ. (A) Representative histograms and bar chart showing MFI of PLZF in splenic iNKT cells from α-GalCer- or vehicle-treated MZ B cell-depleted mice and control mice (n=3 per group, one of two experiments is shown). (B) Representative flow cytometry plots and bar chart showing the frequency of IFN-γ+ splenic iNKT cells from α-GalCer- or vehicle-treated MZ B cell-depleted mice and control mice (n=4 per group, one of two experiments is shown). Analysed at 16h post-disease onset. Data are mean±s.e.m. *P<0.05, **P<0.01, ***P<0.001 (one-way ANOVA).

147 CHAPTER VII: Results V

In the previous chapter we have shown that of CD1dhi splenic B cells T2-MZP and not MZ B cells are required for α-GalCer-mediated amelioration of arthritis.

Having asserted the role of B cells in directing iNKT cell phenotype we finally wanted to explore whether a feedback loop existed whereby iNKT cell provided signals lead to the empowerment of Breg suppressive capacity.

We show that the development of Bregs and their CD1d-independent suppressive capacity is preserved in iNKT cell-deficient mice.

Note that assessment of B cell in vitro suppressive capacity was performed by Dr Anneleen Bosma. When experiments were performed by Anneleen Bosma, this is indicated in the figure legend.

148 iNKT cells are dispensable for Breg development Recently, it has been shown that iNKT cells can induce the expansion of IL-10- producing Bregs in a cognate manner (508), a finding that could also explain the inability of α-GalCer treatment to suppress arthritis in the absence of B cells. We therefore wanted to explore whether a feedback loop exists, whereby iNKT cells support the differentiation of IL-10-producing Bregs. We did not observe an enhancement of B cell IL-10 expression following administration of α-GalCer to IL-10- eGFP reporter mice at time of arthritis induction (Fig. 7.1).

We then induced arthritis in iNKT deficient Jα18-/- and Cd1d-/- mice to further explore whether iNKT cells had a role in Breg homeostasis. While Jα18-/- mice only lack iNKT cells (due to the inability to rearrange the iTCR) (307), Cd1d-/- mice lack both iNKT and type II NKT cells due to lack of thymic selection (479). Confirming that iNKT cells play an immunomodulatory role in the context of arthritis, both Jα18-/- and Cd1d-/- mice developed severe disease that was associated with exacerbated Th1 and Th17 responses both in the spleen and dLN (Fig. 7.2 and 7.3).

Confirming previously published observations, B cell development was comparable in iNKT cell deficient and WT mice (Fig. 7.4A) (456). We could not find any differences in the frequency of IL-10+ B cells and the amount of IL-10 produced by B cells in response to different stimuli in Jα18-/-, Cd1d-/- and WT mice (Fig. 7.4, B and C).

To further understand whether Bregs differentiating in iNKT cell deficient mice were able to suppress conventional T cells as efficiently as their WT counterparts, B cells were isolated from spleens of Jα18-/- and Cd1d-/- mice, as well as WT mice, and were stimulated in vitro with anti-CD40. We have previously shown that this protocol induces the differentiation of Bregs that can suppress autoimmune diseases (179, 196). Bregs isolated from iNKT cell deficient mice retained their ability to inhibit TNF-α- producing T cells in vitro (Fig. 7.5). Collectively these results demonstrate that iNKT cells are not required for the differentiation of functional Bregs.

149

Vehicle

4 ) 1.5 6 -GalCer 10 3 1.0 B cellsB ( + 2 +

0.5 1 IL-10-eGFP B cells (%) 0 IL-10-eGFP 0.0 CD1dhi CD1dlo CD1dhi CD1dlo Figure 7.1. α-GalCer does not induce B cell IL-10. Bar charts showing the frequency and number of splenic IL-10-eGFP+CD19+ B cells from IL-10-eGFP mice that received α-GalCer or vehicle alone following induction of arthritis (α-GalCer-treated group n=4, vehicle-treated group n=3, one of two experiments is shown). Analysed on day 7 post- disease onset. Data are mean±s.e.m. (two-way ANOVA).

150 Gated on live CD4+ cells -/- ) A B WT Jα18 ) 6 5 1.0 6 1.0 0.15 0.19 0.022 0.78 0.096 * WT 10 10 * -/- 4 0.8 0.8 Jα18 WT 0.10 3 0.6 0.6 T cells (%) -/- T cells (%) + +

α T cells (

J 18 T cells ( + 60 + 2 0.4 0.4 CD4 CD4 0.05 + + CD4 CD4 Spleen + + ** Spleen 1 0.2 0.2 IL-17 40 4.07 3.97 IFN- IL-17 ** **** 0 0.0 IFN- 0.0 0.00 ) *** ) 4

** 4 0.098 0.14 0.44 0.013 0.8 0.5 1.5 0.6 10 20 * 10 ** 0.4 0.6 1.0 * 0.4

Increase in knee swelling 0.3 T cells (%) T cells (%) + + T cells ( T cells ( comparedcontrol to knee (%) 0 0.4 + + dLN

01234567 dLN 0.2 CD4 CD4 0.5 0.2 + + CD4 CD4 + 0.2 + Day post disease onset 0.1 IL-17 IFN-

IL-17 0.58 0.60 IL-17 0.0 0.0 IFN- 0.0 0.0 IFN-γ Figure 7.2. Jα18-/- mice develop exacerbated arthritis associated with elevated Th1 and Th17 responses. (A) Mean clinical score of Jα18-/- and WT mice following induction of arthritis; y axis shows percentage swelling in antigen-injected knee compared to control knee (Jα18-/- n=16, WT n=23). (B) Representative flow cytometry plots (left) and bar charts (right) showing the frequency and number of IFN-γ+ and IL- 17+ CD4+ T cells in the spleen (top) and inguinal lymph node (bottom) of Jα18-/- and WT mice (n=3 per group, one of three experiments is shown). B analysed on day 7 post- disease onset. Data are mean±s.e.m. *P<0.05, **P< 0.01, ***P<0.001, ****P<0.0001 (A two-way ANOVA, B Student’s t-test).

151 Gated on live CD4+ cells -/- ) WT Cd1d ) 6 A B 6 6 1.02 0.19 1.02 0.27 1.5 1.5 0.4 WT 10 10 Cd1d-/- WT * 0.3 * Cd1d-/- 4 1.0 1.0

80 T cells (%) T cells (%) + + T cells ( T cells ( 0.2 + + CD4 ** 2 CD4 0.5 0.5 + 60 + Spleen CD4 CD4 Spleen + + 0.1 IFN- * IL-17 IL-17 40 *** 3.06 4.18 0 0.0 IFN- 0.0 0.0 ) ** ) 4 * * 0.77 0.035 0.88 0.11 2.5 *** 1.5 4 2.0 *** 1.0 10 20 10 * 2.0 0.8 1.5 Increase in knee swelling 1.0 comparedcontrol to knee (%) 0 1.5 0.6 T cells (%) T cells (%) + + T cells ( 01234567 T cells ( dLN 1.0 + + dLN Day post disease onset 1.0 0.4 CD4 CD4 0.5 + + CD4 CD4 + + 0.5 0.5 0.2 IL-17 IFN- IL-17 1.05 2.10 IL-17 0.0 0.0 IFN- 0.0 0.0 IFN-γ Figure 7.3. Cd1d-/- mice develop exacerbated arthritis associated with elevated Th1 and Th17 responses. (A) Mean clinical score of Cd1d-/- and WT mice following induction of arthritis; y axis shows percentage swelling in antigen-injected knee compared to control knee (Cd1d-/- n=22, WT n=24). (B) Representative flow cytometry plots (left) and bar charts (right) showing the frequency and number of IFN-γ+ and IL- 17+ CD4+ T cells in the spleen (top) and inguinal lymph node (bottom) of Cd1d-/- and WT mice (n=3 per group, one of three experiments is shown). B analysed on day 7 post- disease onset. Data are mean±s.e.m. *P<0.05, **P< 0.01, ***P<0.001 (A two-way ANOVA, B Student’s t-test).

152

A WT B WT C WT ) 6 -/- -/- -/- 80 25 J 18 4 ) 0.8 J 18 100 J 18 6 10 70 -/- -/- -/- 20 Cd1d 10 Cd1d Cd1d B cells (%) 60 80 + 3 0.6 50 15 40 60 B cells (%)

30 10 + B cells ( 10 2 + 0.4 8 40

4 CD19 IL-10 (pg/ml) +

6 CD19

1 + 0.2 4 2 20

2 IL-10 0 0 0 IL-10 0.0 0 Number among live cells ( cells live among Number Z

Frequency among CD19 MZ Fo M Fo mBSA mBSA+ T2-MZP T2-MZP anti-CD40 anti-CD40 Figure 7.4. Breg development is not impaired in iNKT cell deficient mice. (A) Bar charts showing the frequency and number of splenic T2-MZP, MZ and Fo B cells in Jα18-/-, Cd1d-/- and WT mice following arthritis (Jα18-/- n=6, Cd1d-/-n=7 and WT n=10, one of two experiments is shown). (B) Bar charts showing the frequency and number of splenic IL-10+CD19+ B cells in Jα18-/-, Cd1d-/- and WT mice following arthritis (Jα18-/- n=5, Cd1d-/-n=8 and WT n=9, one of two experiments is shown). (C) Bar chart showing IL-10 production by B cells isolated from the spleens of arthritic Jα18-/-, Cd1d-/- and WT mice following stimulation with methylated BSA (mBSA), anti-CD40 or mBSA+anti-CD40 as measured by ELISA (n=9 per group, one of three experiments is shown). Analysed on day 7 post-disease onset. Data are mean±s.e.m. (A, C two-way ANOVA, B one-way ANOVA).

153 Gated on live CD4+ cells CD4+CD25- T cells CD4+CD25- T cells CD4+CD25- T cells CD4+CD25- T cells α -/- -/- alone WT B cells J 18 B cells Cd1d B cells 80 0.60 51.7 27.2 28.5 24.8 *** *** 60 **** T cells (%) + 40 CD4 + 20 α TNF-

FMO TNF- 0 CD4+CD25- T cells + + + + CD4 WT B cells - + - - Jα18-/- B cells - - + - Cd1d-/- B cells - - - + Figure 7.5. Breg in vitro suppressive capacity is not impaired in iNKT cell deficient mice. Representative flow cytometry plots and bar chart showing the frequency of TNF- α+CD4+ T cells following co-culture with anti-CD40 stimulated B cells isolated from the spleens of arthritic Jα18-/-, Cd1d-/- and WT mice (n=3 per group, one of two experiments is shown). Analysed on 7 post-disease onset. Data are mean±s.e.m. ***P<0.001, ****P<0.0001 (one-way ANOVA). Experiment performed in collaboration with Anneleen Bosma.

154 Chapter VIII: Discussion

Bregs help to restrain immune-driven diseases by inhibiting pathogenic T cell responses and thus keeping the immune system in check. Most Bregs described in mice and humans have been identified by their IL-10 production and/or by the expression of high levels of CD1d. Indeed, expression of CD1d is a hallmark of several identified Breg subsets including B10, T2-MZP and MZ B cells (179, 509). To date, however, whether CD1d plays a functional role in Breg suppressive function remains unknown. Here, we show that the CD1d-dependent interaction with iNKT cells is dispensable for Breg development, but critical for their regulatory function by supporting the differentiation of iNKT cells with suppressive capacity.

For the first time, we show that at an early stage of an autoimmune response T2- MZP B cells, previously ascribed with regulatory function (and not any of the other B cell subsets expressing CD1d), present lipid antigen via CD1d and induce the differentiation of iNKT cells, which partially via the release of IFN-γ exert suppressive function. iNKT cells, primed by Bregs, in turn, downregulate Th1 and Th17 adaptive immune responses that contribute to the development of arthritis (Fig. 8.1). Therefore, we have identified a novel mechanism by which Bregs, via CD1d, bridge innate and adaptive immune responses in an IL-10-independent manner, and control and restrain excessive inflammation.

B cells perform a diverse array of immunological functions, including presentation of both peptide and lipid antigens. Whereas the upregulation of major histocompatibility complex class II molecules upon B cell activation promotes the activation of CD4+ Th cells, CD1d-dependent presentation of α-GalCer by B cells to iNKT cells generates an activation signal that drives antibody production (392, 393, 398, 400, 510).

It is important to note that in the iNKT cell field, the current “dogma” states that the effector function of iNKT cells is almost exclusively driven by lipid presentation by DCs (295). In addition, the majority of the research has focussed on the role that iNKT cells have in providing help to B cells and promoting their differentiation into plasma cells (392, 393, 398, 400, 510). Whereas very little is known in regards to the function that lipid presentation by B cells has on iNKT cell biology. Emerging evidence also shows that optimal CD1d expression on B cells is essential for the generation of iNKT cell responses. For example, the loss of CD1d expression on B cells, following Epstein-

155 Barr virus infection, results in severe attenuation of iNKT cell effector functions (511). Published data by our laboratory and confirmed by other groups (275, 512) also suggested that aberrant CD1d expression by B cells is associated with iNKT cell functional defect observed in SLE patients, and that B cell specific expression of CD1d is important for the in vitro iNKT cell activation. Together, these findings suggest that crosstalk between CD1d-expressing B cells and iNKT cells may be important in the maintenance of iNKT cell homeostasis and for regulation of suppression (275).

By taking advantage of a combination of chimeric mice and adoptive transfer strategies we have demonstrated that CD1d+ Bregs inhibit inflammation by directly interacting with iNKT cells, thus underscoring a new mechanism for Breg suppression. It is becoming increasingly clear that Bregs can suppress in an IL-10-independent manner, including via the production of IL-35, TGF-β, adenosine or expression of PD- L1 (245, 513, 514). In contrast to the induction of CD4+ Tregs by Bregs (197), which is impaired in the absence of IL-10+ B cells, here we have shown that the differentiation of iNKT cells that limit inflammation in arthritis is CD1d-dependent, but IL-10- independent. Indeed, arthritis in B-Il10-/- or in global Il10ra-/- mice was ameliorated by α-GalCer treatment.

To date, several mechanisms for iNKT cell-mediated regulation of pathogenic Th cell responses that drive autoimmunity have been suggested (284, 407, 488). For example, expression of IL-4, IL-10 and IFN-γ have all been shown to mediate iNKT cell-dependent inhibition of autoimmunity (284). We consistently found that iNKT cell IFN-γ upregulation was impaired in response to α-GalCer treatment in B cell deficient or B-Cd1d-/- mice compared to control mice. The protective role of iNKT cell IFN-γ in arthritis was supported by the experiments where the neutralization IFN-γ reversed to some extent the protective effect of α-GalCer and restored the subsequent pro- inflammatory cytokine production. Grajewski et al. originally showed that iNKT cells producing IFN-γ ameliorate uveitis by inhibiting the adaptive Th1 and Th17 responses, suggesting that innate, iNKT cell produced, IFN-γ is critical to initiate regulatory circuits that inhibit pathogenic Th cells in uveitis (459). In EAE and TIDM, iNKT cells producing IFN-γ protect mice from developing disease by dampening CD4+ T cell activation, including their production of IFN-γ and IL-17 (442, 489, 490).

Of interest, at least in vitro iNKT cells isolated from SLE patients produce less IFN-γ compared to healthy individuals (275). Furthermore, iNKT cell production of IFN-γ after healthy peripheral blood mononuclear cell (PBMC) culture with α-GalCer

156 was ablated upon depletion of B cells (275). Therefore, we propose that early (innate) production of IFN-γ by iNKT cells is protective, but late (adaptive) production of IFN-γ (together with IL-17) by CD4+ T cells is associated with increased inflammation and pathology. In agreement with our data, several reports using different experimental models of RA, including CIA and glucose-6-phosphate isomerase (GPI)-induced arthritis, have shown that α-GalCer ameliorates arthritis via inhibition of pro-inflammatory cytokines produced by CD4+ T cells (447, 515). Coppieters et al. also show that high levels of

IFN-γ accumulate in the serum of mice treated with α-GalCer and that neutralization of

IFN-γ at the time of α-GalCer administration during disease onset reduced the efficacy of the treatment, which again supports our data showing that α-GalCer protection is mediated by iNKT cell production of IFN-γ (447). In contrast to our results, however, it has been reported that the protective role of α-GalCer in CIA was also mediated by an upregulation of IL-10 by T cells, and that administration of an anti-IL-10R mAb abrogated the protective effect. In our model suppression of arthritis following α-GalCer was IL-10-independent as there was equivalent suppression of disease in Il10ra-/- and

WT mice treated with α-GalCer. It is interesting to note that the same authors reported different effects of α-GalCer treatment on cytokine production between DBA/1 and

C57BL/6 mice with CIA; higher levels of serum IFN-γ were observed in C57BL/6 mice compared to DBA/1 mice. This suggests that α-GalCer may mediate suppression differentially accordingly to the disease model (CIA versus AIA, chronic versus acute) and to the genetic background of the mice (C57BL/6 versus DBA/1) (448). Importantly, when applying our findings, showing an important role for CD1d-mediated B-iNKT cell interaction in the suppression of acute inflammatory responses to chronic autoimmune conditions, limitations of the current study, including the use of a self-limiting disease model, should be considered.

157 The induction of different populations of regulatory iNKT cells, such as lymph node FoxP3+ iNKT cells (491), adipose tissue-enriched FoxP3-IL-10+ iNKT cells (460), or adipose tissue-resident FoxP3-PLZF-E4BP4+ (462) iNKT cells, has also been associated with iNKT cell anti-inflammatory effects. Our data suggests that Breg- induced iNKT cells are FoxP3-PLZF+E4BP4-. The different phenotype may reflect the tissue from which they were isolated, the particular inflammation model, or the stimuli received from Bregs.

The comparative analysis of gene expression in iNKT cells from B-Cd1d-/- and B-WT mice showed changes in several regulatory pathways. In particular, in the absence of CD1d-presenting B cells, we observed a downregulation of many genes that are enriched in NKT1 cells and the upregulation of those associated with NKT2 phenotype (505). The expression of PLZF has been linked with the acquisition and regulation of multiple facets of iNKT cell phenotype, development, homeostasis and function (320, 321). Here we consistently found PLZF expressed at lower levels when B cells were not able to present α-GalCer. Considering the central role of PLZF in establishing iNKT cell function during differentiation and its maintenance in the periphery, it is tempting to speculate that T2-MZP Bregs, via the expression of CD1d, and possibly other factors, may play a role in PLZF expression or in the stabilization of PLZF locus. More in-depth experiments including the analysis of proximal conserved non-coding sequences in the PLZF locus should be performed in the future to assess this question in more detail. Taken together with the results showing alteration in genes involved in immunometabolism in iNKT cells from B-Cd1d-/- compared to B-WT mice, our data suggest that Bregs expressing CD1d coordinate the induction of iNKT cell effector program, and that Breg-induced iNKT cells are likely to suppress inflammation via multiple mechanisms.

Recently, work carried out by Vomhof-DeKrey et al. (508) showed that 4- hydroxy-3-nitrophenyl-α-GalCer administration does not induce humoral memory, but instead drives the expansion of IL-10+ Bregs. This suggests that iNKT-B cell interactions lead to a feedback loop whereby iNKT cells provide pivotal signals for Breg differentiation in vivo. In contrast, we could not see the induction of IL-10+ B cells upon α-GalCer treatment. We also show that Breg development and in vitro suppressive function on conventional T cells is unaltered in mice lacking iNKT cells. It was also demonstrated previously that Schistosoma mansoni-induced Bregs suppress allergic airway inflammation in mice in a CD1d-dependent, iNKT cell-independent manner

158 (209). We however, observed the Breg effect required iNKT cells. iNKT cells isolated from α-GalCer treated AIA immunized WT mice transferred disease suppression to µMT mice and Bregs failed to suppress disease in iNKT cell deficient mice. Therefore, at least in our experimental system, there is a “long-lasting” effect of B cell-activated iNKT cells that persists following transfer of iNKT cells to a B cell deficient host. Previous work has shown that in an in vivo killing assay α-GalCer loaded, CD1d- expressing cells, in particular MZ B cells and to certain extent CD11c+ and CD11b+ cells are killed (501). In the context of arthritis we could not recapitulate this finding. In contrast we found that splenic B cell subset frequency was not reduced and absolute numbers of each subset were increased in B-WT mice following α-GalCer treatment. In B-Cd1d-/- mice upon α-GalCer treatment MZ B cell frequency was reduced.

Although MZ B cells as a subset express higher levels of CD1d than T2-MZP B cells, examination of their respective roles in the regulation of inflammation via CD1d- dependent interaction with iNKT cells revealed that only T2-MZP Bregs were required to induce iNKT cells with suppressive function. We observed that, upon α-GalCer administration to arthritic mice, iNKT cells localized in the T cell area, but only in B- WT and not B-Cd1d-/- mice. This could explain the requirement for CD1d+ T2-MZP B cells that have access to T cell area in the regulation of iNKT cell phenotype, over MZ B cells that are located in the MZ. Nevertheless, King et al. published a report in which they tried to elucidate the spatiotemporal nature of various modes of iNKT cell activation, using a combination of IL-12 and IL-18, α-GalCer administration or Streptococcus pneumoniae infection (386). Their observations were that both α-GalCer administration and systemic infection led to the redistribution of iNKT cells to the MZ, while IL-12 and IL-18 did not affect iNKT cell location compared to steady-state. The discrepancy with our results, in regards to iNKT cell localization following α-GalCer treatment, may arise from the different route of administration, i.p. versus i.v., and dose of α-GalCer, 2µg versus 0.5µg. Furthermore, they administered α-GalCer in naïve mice, while we used the glycolipid to treat arthritic mice. The cytokines elicited under our experimental conditions together with antigen stimulation are likely to provide unique instructions for iNKT cell migration to those investigated by King et al. (386). It may also be that by the time we assayed iNKT cell location on day three following α-GalCer administration, inflammation had already resolved and instead we observed the reversal to homeostasis-like iNKT cell location in α-GalCer-treated B-WT mice, including them residing in the T cell area. In support of this, even in arthritic vehicle-treated control mice iNKT cells appear to localise in the MZ and splenic parenchyma rather than across 159 the white pulp as observed by King et al. (386). Thus taken together this suggests that migration of iNKT cells is finely tuned by the microenvironmental input that they received during activation.

Using a model of syngeneic apoptotic cell injection to trigger autoantibody production, Wermeling et al. (456) have shown a role for iNKT restriction of autoreactive CD1-expressing B cells. Upon transfer of Cd1d-/- or WT B cells to Cd19-/- mice, which lack MZ B cells and do not mount an antibody response in this model, Cd1d-/- B cells localized in the GCs. The mice that received Cd1d-/- B cells also had elevated antibody titres, compared to the group that received WT B cells. The authors confirmed these observations in 50/50 mixed BM chimeric mice with CD1d-deficient and CD1d-sufficient B cells. CD1d-deficient B cells were more likely to acquire a GC phenotype upon apoptotic cell injection. These results are in agreement with our data showing that in the absence of CD1d-expressing B cells GCs were unaffected by α- GalCer treatment unlike in B-WT control mice, in which we observed a decrease in the number of GCs. However, the present study cannot exclude that the lack of GCs in α- GalCer-treated B-WT mice is due to reduced inflammation, and thus GCs are not formed.

Previously it has been shown that MZ B cells can enhance the activation of iNKT cells in the presence of DCs (391). In addition, a pivotal role for DCs in the activation of iNKT cells, as determined by downstream cytokine responses, was shown in diphtheria toxin-treated human DTR transgenic animals (396). We also found that in the context of arthritis, DCs and MZ B cells were required for the α-GalCer-mediated activation and expansion of iNKT cells. In addition, our data demonstrate that MZ B cells and DCs are not essential in the differentiation of iNKT cells that limit inflammation. Instead, B cells via CD1d induce iNKT cell upregulation of PLZF and IFN-γ. Therefore, we propose a new mechanism by which during an ongoing immune response the role of regulatory T2-MZP B cells is distinct from the function of MZ B cells and DCs with respect to iNKT cells. As discussed in the introduction, iNKT cells can be potently activated by CD1d-lipid presentation in combination with cytokines (295). An explanation to our results could be that some of the co-stimulatory signals to iNKT cells may have been provided in vivo by other cells, and therefore that in our model of inflammation, requirement for DC-mediated stimulation is overridden by signals from inflammatory cytokines. In support of this hypothesis, is the study by Bialecki et al., who reported reduced capacity of α-GalCer to stimulate iNKT cell

160 cytokine production in vitro when MZ B cells were depleted (391). They also observed that MZ B cells were not able to stimulate iNKT cells, but were able to stimulate iNKT cell hybridomas. The authors explain the ability to activate hybridomas, but not primary iNKT cells by the fact that hybridoma activation can be driven by TCR-mediated signals alone, without co-stimulation (of note, the authors excluded IL-12 as blocking it did not abrogate effect). Similarly in support of our observations, King et al. observed that clodronate treatment of mice depleted approximately 1/3 of dendritic cells, but upon this treatment iNKT cells completely lost ability to induce IFN-γ and significantly reduced capacity to produce IL-4 in response to α-GalCer (386). The loss of iNKT cell IFN-γ production upon DC depletion was restored upon exogenous administration of IL-12 in combination with IL-18. It is therefore possible that, in our model, the inflammation, induced upon immunization, overrides the requirement for DCs in iNKT cell activation and explains why we did not see impaired IFN-γ induction upon in vivo or in vitro DC depletion. Thus, we propose that the inflammatory environment promotes B cells to efficiently present α-GalCer, even in the absence of DCs. We have also shown that there were 50% less IFN-γ + iNKT cells in the liver compared to the spleen after in vivo α-GalCer treatment of arthritic mice. T2-MZP B cells do not reside in the liver (K. Oleinika, unpublished observations) and in this organ the majority of B cells are Fo which express low levels of CD1d, providing indirect evidence that T2-MZP Bregs are important for the differentiation of suppressive IFN-γ+ iNKT cells. One possible explanation of how MZ B cells contribute differentially to iNKT cell driven disease outcome in arthritis is the location, as discussed.

What remains to be answered in a lot of mechanistic Breg studies is how exactly Bregs suppress Th1/Th2/Th17 responses i.e. whether the effect is direct. It has been shown in vitro that Bregs can directly induce Tregs as well as suppress Th1/Th17 responses (274, 276). Here we have shown that in addition Bregs can suppress Th1/Th17 responses via their effect on iNKT cell function.

In conclusion, we have described a novel T2-MZP Breg regulatory mechanism that acts as a rheostat in the differentiation of iNKT cells with suppressive capacity. We propose that Breg CD1d-mediated suppression may allow more immediate suppression and control of inflammation, while IL-10 response is induced. Indeed, cytoplasmic IL- 10 is not a feature of resting B cells (216), and in vivo Breg IL-10 induction, as discussed, requires signals such as CD40 or TLR ligation, and inflammation. Similarly in vitro induction of IL-10 requires strong signals or longer cultures (179, 263).

161 In an era where B cells constitute a primary therapeutic target in several autoimmune diseases and in cancer, these data add a further degree of complexity that needs to be considered when designing new therapies aimed to eliminate the entirety of B cells.

162 BCR

Lipid Th1 T-bet IFN-γ iNKT Breg CD1d iTCR PLZF Th17 Transitional 2-marginal zone precursor (T2-MZP) B cell phenotype

Figure 8.1. Proposed mechanism of Breg modulation of inflammation in a CD1d- and iNKT cell-dependent manner. Lipid presentation by CD1d-expressing B cells with a T2-MZP phenotype induces PLZF+IFN-γ+ (T-bet+) iNKT cells. These iNKT cells in turn regulate Th1 and Th17 response partially via the production of IFN-γ.

163 Bibliography

1. Cooper MD, Peterson RD, Good RA. Delineation of the Thymic and Bursal Lymphoid Systems in the Chicken. Nature. 1965;205:143-6. 2. Cooper MD. The early history of B cells. Nature reviews Immunology. 2015;15(3):191-7. 3. Dal Porto JM, Gauld SB, Merrell KT, Mills D, Pugh-Bernard AE, Cambier J. B cell antigen receptor signaling 101. Molecular immunology. 2004;41(6-7):599-613. 4. Schroeder HW, Jr., Cavacini L. Structure and function of immunoglobulins. The Journal of allergy and clinical immunology. 2010;125(2 Suppl 2):S41-52. 5. Pelanda R, Torres RM. Central B-cell tolerance: where selection begins. Cold Spring Harbor perspectives in biology. 2012;4(4):a007146. 6. Hardy RR, Carmack CE, Shinton SA, Kemp JD, Hayakawa K. Resolution and characterization of pro-B and pre-pro-B cell stages in normal mouse bone marrow. The Journal of experimental medicine. 1991;173(5):1213-25. 7. Brack C, Hirama M, Lenhard-Schuller R, Tonegawa S. A complete immunoglobulin gene is created by somatic recombination. Cell. 1978;15(1):1-14. 8. Fugmann SD, Lee AI, Shockett PE, Villey IJ, Schatz DG. The RAG proteins and V(D)J recombination: complexes, ends, and transposition. Annual review of immunology. 2000;18:495-527. 9. Oettinger MA, Schatz DG, Gorka C, Baltimore D. RAG-1 and RAG-2, adjacent genes that synergistically activate V(D)J recombination. Science. 1990;248(4962):1517- 23. 10. Schatz DG, Oettinger MA, Baltimore D. The V(D)J recombination activating gene, RAG-1. Cell. 1989;59(6):1035-48. 11. Desiderio SV, Yancopoulos GD, Paskind M, Thomas E, Boss MA, Landau N, et al. Insertion of N Regions into Heavy-Chain Genes Is Correlated with Expression of Terminal Deoxytransferase in B-Cells. Nature. 1984;311(5988):752-5. 12. Sakaguchi N, Melchers F. Lambda-5, a New Light-Chain-Related Locus Selectively Expressed in Pre-Lymphocytes-B. Nature. 1986;324(6097):579-82. 13. Melchers F. The pre-B-cell receptor: selector of fitting immunoglobulin heavy chains for the B-cell repertoire. Nature reviews Immunology. 2005;5(7):578-84. 14. Nemazee D. Mechanisms of central tolerance for B cells. Nature reviews Immunology. 2017;17(5):281-94. 15. LeBien TW, Tedder TF. B lymphocytes: how they develop and function. Blood. 2008;112(5):1570-80. 16. Vettermann C, Schlissel MS. Allelic exclusion of immunoglobulin genes: models and mechanisms. Immunological reviews. 2010;237(1):22-42. 17. Melchers F. Checkpoints that control B cell development. The Journal of clinical investigation. 2015;125(6):2203-10. 18. Rolink AG, Andersson J, Melchers F. Characterization of immature B cells by a novel monoclonal antibody, by turnover and by mitogen reactivity. European journal of immunology. 1998;28(11):3738-48. 19. Grandien A, Fucs R, Nobrega A, Andersson J, Coutinho A. Negative selection of multireactive B cell clones in normal adult mice. European journal of immunology. 1994;24(6):1345-52. 20. Wardemann H, Yurasov S, Schaefer A, Young JW, Meffre E, Nussenzweig MC. Predominant autoantibody production by early human B cell precursors. Science. 2003;301(5638):1374-7. 21. Halverson R, Torres RM, Pelanda R. Receptor editing is the main mechanism of B cell tolerance toward membrane antigens. Nature immunology. 2004;5(6):645-50.

164 22. Gay D, Saunders T, Camper S, Weigert M. Receptor editing: an approach by autoreactive B cells to escape tolerance. The Journal of experimental medicine. 1993;177(4):999-1008. 23. Radic MZ, Erikson J, Litwin S, Weigert M. B lymphocytes may escape tolerance by revising their antigen receptors. The Journal of experimental medicine. 1993;177(4):1165-73. 24. Tiegs SL, Russell DM, Nemazee D. Receptor editing in self-reactive bone marrow B cells. The Journal of experimental medicine. 1993;177(4):1009-20. 25. Chen C, Nagy Z, Prak EL, Weigert M. Immunoglobulin heavy chain gene replacement: a mechanism of receptor editing. Immunity. 1995;3(6):747-55. 26. Zhang Z, Zemlin M, Wang YH, Munfus D, Huye LE, Findley HW, et al. Contribution of Vh gene replacement to the primary B cell repertoire. Immunity. 2003;19(1):21-31. 27. Nemazee DA, Burki K. Clonal deletion of B lymphocytes in a transgenic mouse bearing anti-MHC class I antibody genes. Nature. 1989;337(6207):562-6. 28. Hartley SB, Crosbie J, Brink R, Kantor AB, Basten A, Goodnow CC. Elimination from peripheral lymphoid tissues of self-reactive B lymphocytes recognizing membrane-bound antigens. Nature. 1991;353(6346):765-9. 29. Hartley SB, Cooke MP, Fulcher DA, Harris AW, Cory S, Basten A, et al. Elimination of self-reactive B lymphocytes proceeds in two stages: arrested development and cell death. Cell. 1993;72(3):325-35. 30. Goodnow CC, Crosbie J, Adelstein S, Lavoie TB, Smith-Gill SJ, Brink RA, et al. Altered immunoglobulin expression and functional silencing of self-reactive B lymphocytes in transgenic mice. Nature. 1988;334(6184):676-82. 31. Fulcher DA, Basten A. Reduced life span of anergic self-reactive B cells in a double-transgenic model. The Journal of experimental medicine. 1994;179(1):125-34. 32. Cyster JG, Hartley SB, Goodnow CC. Competition for follicular niches excludes self-reactive cells from the recirculating B-cell repertoire. Nature. 1994;371(6496):389- 95. 33. Hayakawa K, Asano M, Shinton SA, Gui M, Wen LJ, Dashoff J, et al. Positive selection of anti-thy-1 autoreactive B-1 cells and natural serum autoantibody production independent from bone marrow B cell development. The Journal of experimental medicine. 2003;197(1):87-99. 34. Wen L, Brill-Dashoff J, Shinton SA, Asano M, Hardy RR, Hayakawa K. Evidence of marginal-zone B cell-positive selection in spleen. Immunity. 2005;23(3):297-308. 35. Kench JA, Russell DM, Nemazee D. Efficient peripheral clonal elimination of B lymphocytes in MRL/lpr mice bearing autoantibody transgenes. The Journal of experimental medicine. 1998;188(5):909-17. 36. Russell DM, Dembic Z, Morahan G, Miller JF, Burki K, Nemazee D. Peripheral deletion of self-reactive B cells. Nature. 1991;354(6351):308-11. 37. Nagasawa T. Microenvironmental niches in the bone marrow required for B-cell development. Nature reviews Immunology. 2006;6(2):107-16. 38. Egawa T, Kawabata K, Kawamoto H, Amada K, Okamoto R, Fujii N, et al. The earliest stages of B cell development require a chemokine stromal cell-derived factor/pre-B cell growth-stimulating factor. Immunity. 2001;15(2):323-34. 39. Ma Q, Jones D, Springer TA. The chemokine receptor CXCR4 is required for the retention of B lineage and granulocytic precursors within the bone marrow microenvironment. Immunity. 1999;10(4):463-71. 40. Nie Y, Waite J, Brewer F, Sunshine MJ, Littman DR, Zou YR. The role of CXCR4 in maintaining peripheral B cell compartments and humoral immunity. The Journal of experimental medicine. 2004;200(9):1145-56.

165 41. Tokoyoda K, Egawa T, Sugiyama T, Choi BI, Nagasawa T. Cellular niches controlling B lymphocyte behavior within bone marrow during development. Immunity. 2004;20(6):707-18. 42. Milne CD, Paige CJ. IL-7: a key regulator of B lymphopoiesis. Seminars in immunology. 2006;18(1):20-30. 43. Peschon JJ, Morrissey PJ, Grabstein KH, Ramsdell FJ, Maraskovsky E, Gliniak BC, et al. Early lymphocyte expansion is severely impaired in interleukin 7 receptor- deficient mice. The Journal of experimental medicine. 1994;180(5):1955-60. 44. von Freeden-Jeffry U, Vieira P, Lucian LA, McNeil T, Burdach SE, Murray R. Lymphopenia in interleukin (IL)-7 gene-deleted mice identifies IL-7 as a nonredundant cytokine. The Journal of experimental medicine. 1995;181(4):1519-26. 45. Dias S, Silva H, Jr., Cumano A, Vieira P. Interleukin-7 is necessary to maintain the B cell potential in common lymphoid progenitors. The Journal of experimental medicine. 2005;201(6):971-9. 46. Buckley RH. Molecular defects in human severe combined immunodeficiency and approaches to immune reconstitution. Annual review of immunology. 2004;22:625- 55. 47. Sitnicka E, Brakebusch C, Martensson IL, Svensson M, Agace WW, Sigvardsson M, et al. Complementary signaling through flt3 and interleukin-7 receptor alpha is indispensable for fetal and adult B cell genesis. The Journal of experimental medicine. 2003;198(10):1495-506. 48. Sitnicka E, Bryder D, Theilgaard-Monch K, Buza-Vidas N, Adolfsson J, Jacobsen SE. Key role of flt3 ligand in regulation of the common lymphoid progenitor but not in maintenance of the hematopoietic stem cell pool. Immunity. 2002;17(4):463- 72. 49. Pillai S, Cariappa A, Moran ST. Positive selection and lineage commitment during peripheral B-lymphocyte development. Immunological reviews. 2004;197:206- 18. 50. Lam KP, Kuhn R, Rajewsky K. In vivo ablation of surface immunoglobulin on mature B cells by inducible gene targeting results in rapid cell death. Cell. 1997;90(6):1073-83. 51. Reichman-Fried M, Hardy RR, Bosma MJ. Development of B-lineage cells in the bone marrow of scid/scid mice following the introduction of functionally rearranged immunoglobulin transgenes. Proceedings of the National Academy of Sciences of the United States of America. 1990;87(7):2730-4. 52. Spanopoulou E, Roman CA, Corcoran LM, Schlissel MS, Silver DP, Nemazee D, et al. Functional immunoglobulin transgenes guide ordered B-cell differentiation in Rag-1-deficient mice. Genes & development. 1994;8(9):1030-42. 53. Young F, Ardman B, Shinkai Y, Lansford R, Blackwell TK, Mendelsohn M, et al. Influence of immunoglobulin heavy- and light-chain expression on B-cell differentiation. Genes & development. 1994;8(9):1043-57. 54. Bannish G, Fuentes-Panana EM, Cambier JC, Pear WS, Monroe JG. Ligand- independent signaling functions for the B lymphocyte antigen receptor and their role in positive selection during B lymphopoiesis. The Journal of experimental medicine. 2001;194(11):1583-96. 55. Monroe JG. ITAM-mediated tonic signalling through pre-BCR and BCR complexes. Nature reviews Immunology. 2006;6(4):283-94. 56. Tze LE, Schram BR, Lam KP, Hogquist KA, Hippen KL, Liu J, et al. Basal immunoglobulin signaling actively maintains developmental stage in immature B cells. PLoS biology. 2005;3(3):e82.

166 57. Verkoczy L, Duong B, Skog P, Ait-Azzouzene D, Puri K, Vela JL, et al. Basal B cell receptor-directed phosphatidylinositol 3-kinase signaling turns off RAGs and promotes B cell-positive selection. Journal of immunology. 2007;178(10):6332-41. 58. Schram BR, Tze LE, Ramsey LB, Liu J, Najera L, Vegoe AL, et al. B cell receptor basal signaling regulates antigen-induced Ig light chain rearrangements. Journal of immunology. 2008;180(7):4728-41. 59. Levine MH, Haberman AM, Sant'Angelo DB, Hannum LG, Cancro MP, Janeway CA, Jr., et al. A B-cell receptor-specific selection step governs immature to mature B cell differentiation. Proceedings of the National Academy of Sciences of the United States of America. 2000;97(6):2743-8. 60. Gu H, Tarlinton D, Muller W, Rajewsky K, Forster I. Most peripheral B cells in mice are ligand selected. The Journal of experimental medicine. 1991;173(6):1357-71. 61. Gaudin E, Hao Y, Rosado MM, Chaby R, Girard R, Freitas AA. Positive selection of B cells expressing low densities of self-reactive BCRs. The Journal of experimental medicine. 2004;199(6):843-53. 62. Cyster JG, Healy JI, Kishihara K, Mak TW, Thomas ML, Goodnow CC. Regulation of B-lymphocyte negative and positive selection by tyrosine phosphatase CD45. Nature. 1996;381(6580):325-8. 63. Heltemes LM, Manser T. Level of B cell antigen receptor surface expression influences both positive and negative selection of B cells during primary development. Journal of immunology. 2002;169(3):1283-92. 64. Wang LD, Lopes J, Cooper AB, Dang-Lawson M, Matsuuchi L, Clark MR. Selection of B lymphocytes in the periphery is determined by the functional capacity of the B cell antigen receptor. Proceedings of the National Academy of Sciences of the United States of America. 2004;101(4):1027-32. 65. Rowland SL, DePersis CL, Torres RM, Pelanda R. Ras activation of Erk restores impaired tonic BCR signaling and rescues immature B cell differentiation. The Journal of experimental medicine. 2010;207(3):607-21. 66. Wang H, Clarke SH. Evidence for a ligand-mediated positive selection signal in differentiation to a mature B cell. Journal of immunology. 2003;171(12):6381-8. 67. Mackay F, Schneider P. Cracking the BAFF code. Nature reviews Immunology. 2009;9(7):491-502. 68. Smith SH, Cancro MP. Cutting edge: B cell receptor signals regulate BLyS receptor levels in mature B cells and their immediate progenitors. Journal of immunology. 2003;170(12):5820-3. 69. Schiemann B, Gommerman JL, Vora K, Cachero TG, Shulga-Morskaya S, Dobles M, et al. An essential role for BAFF in the normal development of B cells through a BCMA-independent pathway. Science. 2001;293(5537):2111-4. 70. Pillai S, Cariappa A. The follicular versus marginal zone B lymphocyte cell fate decision. Nature reviews Immunology. 2009;9(11):767-77. 71. Kaisho T, Takeda K, Tsujimura T, Kawai T, Nomura F, Terada N, et al. IkappaB kinase alpha is essential for mature B cell development and function. The Journal of experimental medicine. 2001;193(4):417-26. 72. Senftleben U, Cao Y, Xiao G, Greten FR, Krahn G, Bonizzi G, et al. Activation by IKKalpha of a second, evolutionary conserved, NF-kappa B signaling pathway. Science. 2001;293(5534):1495-9. 73. Yamada T, Mitani T, Yorita K, Uchida D, Matsushima A, Iwamasa K, et al. Abnormal immune function of hemopoietic cells from alymphoplasia (aly) mice, a natural strain with mutant NF-kappa B-inducing kinase. Journal of immunology. 2000;165(2):804-12.

167 74. Stadanlick JE, Kaileh M, Karnell FG, Scholz JL, Miller JP, Quinn WJ, 3rd, et al. Tonic B cell antigen receptor signals supply an NF-kappaB substrate for prosurvival BLyS signaling. Nature immunology. 2008;9(12):1379-87. 75. Rowland SL, Leahy KF, Halverson R, Torres RM, Pelanda R. BAFF receptor signaling aids the differentiation of immature B cells into transitional B cells following tonic BCR signaling. Journal of immunology. 2010;185(8):4570-81. 76. Hayakawa K, Hardy RR. Development and function of B-1 cells. Current opinion in immunology. 2000;12(3):346-53. 77. Martin F, Kearney JF. B1 cells: similarities and differences with other B cell subsets. Current opinion in immunology. 2001;13(2):195-201. 78. Montecino-Rodriguez E, Dorshkind K. Formation of B-1 B cells from neonatal B-1 transitional cells exhibits NF-kappaB redundancy. Journal of immunology. 2011;187(11):5712-9. 79. Baumgarth N. The double life of a B-1 cell: self-reactivity selects for protective effector functions. Nature reviews Immunology. 2011;11(1):34-46. 80. Haas KM, Poe JC, Steeber DA, Tedder TF. B-1a and B-1b cells exhibit distinct developmental requirements and have unique functional roles in innate and adaptive immunity to S. pneumoniae. Immunity. 2005;23(1):7-18. 81. Berland R, Wortis HH. Origins and functions of B-1 cells with notes on the role of CD5. Annual review of immunology. 2002;20:253-300. 82. Mond JJ, Lees A, Snapper CM. T cell-independent antigens type 2. Annual review of immunology. 1995;13:655-92. 83. Tangye SG. To B1 or not to B1: that really is still the question! Blood. 2013;121(26):5109-10. 84. Hardy RR, Hayakawa K. B cell development pathways. Annual review of immunology. 2001;19:595-621. 85. Cariappa A, Chase C, Liu H, Russell P, Pillai S. Naive recirculating B cells mature simultaneously in the spleen and bone marrow. Blood. 2007;109(6):2339-45. 86. Lindsley RC, Thomas M, Srivastava B, Allman D. Generation of peripheral B cells occurs via two spatially and temporally distinct pathways. Blood. 2007;109(6):2521-8. 87. Srivastava B, Lindsley RC, Nikbakht N, Allman D. Models for peripheral B cell development and homeostasis. Seminars in immunology. 2005;17(3):175-82. 88. Loder F, Mutschler B, Ray RJ, Paige CJ, Sideras P, Torres R, et al. B cell development in the spleen takes place in discrete steps and is determined by the quality of B cell receptor-derived signals. The Journal of experimental medicine. 1999;190(1):75-89. 89. Allman D, Lindsley RC, DeMuth W, Rudd K, Shinton SA, Hardy RR. Resolution of three nonproliferative immature splenic B cell subsets reveals multiple selection points during peripheral B cell maturation. Journal of immunology. 2001;167(12):6834-40. 90. Srivastava B, Quinn WJ, 3rd, Hazard K, Erikson J, Allman D. Characterization of marginal zone B cell precursors. The Journal of experimental medicine. 2005;202(9):1225-34. 91. Merrell KT, Benschop RJ, Gauld SB, Aviszus K, Decote-Ricardo D, Wysocki LJ, et al. Identification of anergic B cells within a wild-type repertoire. Immunity. 2006;25(6):953-62. 92. Saito T, Chiba S, Ichikawa M, Kunisato A, Asai T, Shimizu K, et al. Notch2 is preferentially expressed in mature B cells and indispensable for marginal zone B lineage development. Immunity. 2003;18(5):675-85.

168 93. Cariappa A, Liou HC, Horwitz BH, Pillai S. Nuclear factor kappa B is required for the development of marginal zone B lymphocytes. The Journal of experimental medicine. 2000;192(8):1175-82. 94. Meyer-Bahlburg A, Andrews SF, Yu KO, Porcelli SA, Rawlings DJ. Characterization of a late transitional B cell population highly sensitive to BAFF- mediated homeostatic proliferation. The Journal of experimental medicine. 2008;205(1):155-68. 95. Carsetti R, Rosado MM, Wardmann H. Peripheral development of B cells in mouse and man. Immunological reviews. 2004;197:179-91. 96. Sims GP, Ettinger R, Shirota Y, Yarboro CH, Illei GG, Lipsky PE. Identification and characterization of circulating human transitional B cells. Blood. 2005;105(11):4390-8. 97. Palanichamy A, Barnard J, Zheng B, Owen T, Quach T, Wei C, et al. Novel human transitional B cell populations revealed by B cell depletion therapy. Journal of immunology. 2009;182(10):5982-93. 98. Martin F, Kearney JF. Marginal-zone B cells. Nature reviews Immunology. 2002;2(5):323-35. 99. Pillai S, Cariappa A, Moran ST. Marginal zone B cells. Annual review of immunology. 2005;23:161-96. 100. Attanavanich K, Kearney JF. Marginal zone, but not follicular B cells, are potent activators of naive CD4 T cells. Journal of immunology. 2004;172(2):803-11. 101. Cerutti A, Cols M, Puga I. Marginal zone B cells: virtues of innate-like antibody-producing lymphocytes. Nature reviews Immunology. 2013;13(2):118-32. 102. Balazs M, Martin F, Zhou T, Kearney J. Blood dendritic cells interact with splenic marginal zone B cells to initiate T-independent immune responses. Immunity. 2002;17(3):341-52. 103. MacLennan IC, Liu YJ. Marginal zone B cells respond both to polysaccharide antigens and protein antigens. Research in immunology. 1991;142(4):346-51. 104. Steiniger B, Timphus EM, Barth PJ. The splenic marginal zone in humans and rodents: an enigmatic compartment and its inhabitants. Histochemistry and cell biology. 2006;126(6):641-8. 105. Hardy RR, Hayakawa K, Parks DR, Herzenberg LA. Demonstration of B-Cell Maturation in X-Linked Immunodeficient Mice by Simultaneous 3-Color Immunofluorescence. Nature. 1983;306(5940):270-2. 106. Pani G, Siminovitch KA, Paige CJ. The motheaten mutation rescues B cell signaling and development in CD45-deficient mice. Journal of Experimental Medicine. 1997;186(4):581-8. 107. Cariappa A, Tang M, Parng C, Nebelitskiy E, Carroll M, Georgopoulos K, et al. The follicular versus marginal zone B lymphocyte cell fate decision is regulated by Aiolos, Btk, and CD21. Immunity. 2001;14(5):603-15. 108. Samardzic T, Marinkovic D, Danzer CP, Gerlach J, Nitschke L, Wirth T. Reduction of marginal zone B cells in CD22-deficient mice. European journal of immunology. 2002;32(2):561-7. 109. Tanigaki K, Han H, Yamamoto N, Tashiro K, Ikegawa M, Kuroda K, et al. Notch-RBP-J signaling is involved in cell fate determination of marginal zone B cells. Nature immunology. 2002;3(5):443-50. 110. Oyama T, Harigaya K, Muradil A, Hozumi K, Habu S, Oguro H, et al. Mastermind-1 is required for Notch signal-dependent steps in lymphocyte development in vivo. Proceedings of the National Academy of Sciences of the United States of America. 2007;104(23):9764-9.

169 111. Wu L, Maillard I, Nakamura M, Pear WS, Griffin JD. The transcriptional coactivator Maml1 is required for Notch2-mediated marginal zone B-cell development. Blood. 2007;110(10):3618-23. 112. Hozumi K, Negishi N, Suzuki D, Abe N, Sotomaru Y, Tamaoki N, et al. Delta- like 1 is necessary for the generation of marginal zone B cells but not T cells in vivo. Nature immunology. 2004;5(6):638-44. 113. Kuroda K, Han H, Tani S, Tanigaki K, Tun T, Furukawa T, et al. Regulation of marginal zone B cell development by MINT, a suppressor of Notch/RBP-J signaling pathway. Immunity. 2003;18(2):301-12. 114. Siebenlist U, Brown K, Claudio E. Control of lymphocyte development by nuclear factor-kappaB. Nature reviews Immunology. 2005;5(6):435-45. 115. Batten M, Groom J, Cachero TG, Qian F, Schneider P, Tschopp J, et al. BAFF mediates survival of peripheral immature B lymphocytes. The Journal of experimental medicine. 2000;192(10):1453-66. 116. Otipoby KL, Sasaki Y, Schmidt-Supprian M, Patke A, Gareus R, Pasparakis M, et al. BAFF activates Akt and Erk through BAFF-R in an IKK1-dependent manner in primary mouse B cells. Proceedings of the National Academy of Sciences of the United States of America. 2008;105(34):12435-8. 117. Klein U, Dalla-Favera R. Germinal centres: role in B-cell physiology and malignancy. Nature reviews Immunology. 2008;8(1):22-33. 118. Reif K, Ekland EH, Ohl L, Nakano H, Lipp M, Forster R, et al. Balanced responsiveness to chemoattractants from adjacent zones determines B-cell position. Nature. 2002;416(6876):94-9. 119. Rodriguez-Pinto D. B cells as antigen presenting cells. Cellular immunology. 2005;238(2):67-75. 120. Shlomchik MJ, Weisel F. Germinal center selection and the development of memory B and plasma cells. Immunological reviews. 2012;247(1):52-63. 121. Allen CD, Ansel KM, Low C, Lesley R, Tamamura H, Fujii N, et al. Germinal center dark and light zone organization is mediated by CXCR4 and CXCR5. Nature immunology. 2004;5(9):943-52. 122. Jacob J, Kelsoe G, Rajewsky K, Weiss U. Intraclonal generation of antibody mutants in germinal centres. Nature. 1991;354(6352):389-92. 123. Berek C, Berger A, Apel M. Maturation of the immune response in germinal centers. Cell. 1991;67(6):1121-9. 124. Revy P, Muto T, Levy Y, Geissmann F, Plebani A, Sanal O, et al. Activation- induced cytidine deaminase (AID) deficiency causes the autosomal recessive form of the Hyper-IgM syndrome (HIGM2). Cell. 2000;102(5):565-75. 125. Muramatsu M, Kinoshita K, Fagarasan S, Yamada S, Shinkai Y, Honjo T. Class switch recombination and hypermutation require activation-induced cytidine deaminase (AID), a potential RNA editing enzyme. Cell. 2000;102(5):553-63. 126. Petersen-Mahrt SK, Harris RS, Neuberger MS. AID mutates E. coli suggesting a DNA deamination mechanism for antibody diversification. Nature. 2002;418(6893):99- 103. 127. Bransteitter R, Pham P, Scharff MD, Goodman MF. Activation-induced cytidine deaminase deaminates deoxycytidine on single-stranded DNA but requires the action of RNase. Proceedings of the National Academy of Sciences of the United States of America. 2003;100(7):4102-7. 128. Chaudhuri J, Tian M, Khuong C, Chua K, Pinaud E, Alt FW. Transcription- targeted DNA deamination by the AID antibody diversification enzyme. Nature. 2003;422(6933):726-30. 129. Di Noia J, Neuberger MS. Altering the pathway of immunoglobulin hypermutation by inhibiting uracil-DNA glycosylase. Nature. 2002;419(6902):43-8.

170 130. Dickerson SK, Market E, Besmer E, Papavasiliou FN. AID mediates hypermutation by deaminating single stranded DNA. The Journal of experimental medicine. 2003;197(10):1291-6. 131. Ramiro AR, Stavropoulos P, Jankovic M, Nussenzweig MC. Transcription enhances AID-mediated cytidine deamination by exposing single-stranded DNA on the nontemplate strand. Nature immunology. 2003;4(5):452-6. 132. Wagner SD, Neuberger MS. Somatic hypermutation of immunoglobulin genes. Annual review of immunology. 1996;14:441-57. 133. Camacho SA, Kosco-Vilbois MH, Berek C. The dynamic structure of the germinal center. Immunology today. 1998;19(11):511-4. 134. Allen CD, Okada T, Tang HL, Cyster JG. Imaging of germinal center selection events during affinity maturation. Science. 2007;315(5811):528-31. 135. Schwickert TA, Lindquist RL, Shakhar G, Livshits G, Skokos D, Kosco-Vilbois MH, et al. In vivo imaging of germinal centres reveals a dynamic open structure. Nature. 2007;446(7131):83-7. 136. Allman D, Jain A, Dent A, Maile RR, Selvaggi T, Kehry MR, et al. BCL-6 expression during B-cell activation. Blood. 1996;87(12):5257-68. 137. Cattoretti G, Chang CC, Cechova K, Zhang J, Ye BH, Falini B, et al. BCL-6 protein is expressed in germinal-center B cells. Blood. 1995;86(1):45-53. 138. Dent AL, Shaffer AL, Yu X, Allman D, Staudt LM. Control of inflammation, cytokine expression, and germinal center formation by BCL-6. Science. 1997;276(5312):589-92. 139. Ye BH, Cattoretti G, Shen Q, Zhang J, Hawe N, de Waard R, et al. The BCL-6 proto-oncogene controls germinal-centre formation and Th2-type inflammation. Nature genetics. 1997;16(2):161-70. 140. Shapiro-Shelef M, Lin KI, McHeyzer-Williams LJ, Liao J, McHeyzer-Williams MG, Calame K. Blimp-1 is required for the formation of immunoglobulin secreting plasma cells and pre-plasma memory B cells. Immunity. 2003;19(4):607-20. 141. Turner CA, Mack DH, Davis MM. Blimp-1, a Novel Zinc Finger-Containing Protein That Can Drive the Maturation of B-Lymphocytes into Immunoglobulin- Secreting Cells. Cell. 1994;77(2):297-306. 142. Saito M, Gao J, Basso K, Kitagawa Y, Smith PM, Bhagat G, et al. A signaling pathway mediating downregulation of BCL6 in germinal center B cells is blocked by BCL6 gene alterations in B cell lymphoma. Cancer cell. 2007;12(3):280-92. 143. Niu H, Ye BH, Dalla-Favera R. Antigen receptor signaling induces MAP kinase-mediated phosphorylation and degradation of the BCL-6 transcription factor. Genes & development. 1998;12(13):1953-61. 144. Klein U, Tu Y, Stolovitzky GA, Keller JL, Haddad J, Jr., Miljkovic V, et al. Transcriptional analysis of the B cell germinal center reaction. Proceedings of the National Academy of Sciences of the United States of America. 2003;100(5):2639-44. 145. Shaffer AL, Rosenwald A, Hurt EM, Giltnane JM, Lam LT, Pickeral OK, et al. Signatures of the immune response. Immunity. 2001;15(3):375-85. 146. Hu BT, Lee SC, Marin E, Ryan DH, Insel RA. Telomerase is up-regulated in human germinal center B cells in vivo and can be re-expressed in memory B cells activated in vitro. Journal of immunology. 1997;159(3):1068-71. 147. Liu YJ, Mason DY, Johnson GD, Abbot S, Gregory CD, Hardie DL, et al. Germinal center cells express bcl-2 protein after activation by signals which prevent their entry into apoptosis. European journal of immunology. 1991;21(8):1905-10. 148. de Villartay JP, Fischer A, Durandy A. The mechanisms of immune diversification and their disorders. Nature reviews Immunology. 2003;3(12):962-72.

171 149. Han S, Hathcock K, Zheng B, Kepler TB, Hodes R, Kelsoe G. Cellular interaction in germinal centers. Roles of CD40 ligand and B7-2 in established germinal centers. Journal of immunology. 1995;155(2):556-67. 150. Tafuri A, Shahinian A, Bladt F, Yoshinaga SK, Jordana M, Wakeham A, et al. ICOS is essential for effective T-helper-cell responses. Nature. 2001;409(6816):105-9. 151. McAdam AJ, Greenwald RJ, Levin MA, Chernova T, Malenkovich N, Ling V, et al. ICOS is critical for CD40-mediated antibody class switching. Nature. 2001;409(6816):102-5. 152. Castigli E, Wilson SA, Scott S, Dedeoglu F, Xu S, Lam KP, et al. TACI and BAFF-R mediate isotype switching in B cells. The Journal of experimental medicine. 2005;201(1):35-9. 153. Phan TG, Paus D, Chan TD, Turner ML, Nutt SL, Basten A, et al. High affinity germinal center B cells are actively selected into the plasma cell compartment. The Journal of experimental medicine. 2006;203(11):2419-24. 154. Cobaleda C, Schebesta A, Delogu A, Busslinger M. Pax5: the guardian of B cell identity and function. Nature immunology. 2007;8(5):463-70. 155. Klein U, Casola S, Cattoretti G, Shen Q, Lia M, Mo T, et al. Transcription factor IRF4 controls plasma cell differentiation and class-switch recombination. Nature immunology. 2006;7(7):773-82. 156. Sciammas R, Shaffer AL, Schatz JH, Zhao H, Staudt LM, Singh H. Graded expression of interferon regulatory factor-4 coordinates isotype switching with plasma cell differentiation. Immunity. 2006;25(2):225-36. 157. Scheeren FA, Naspetti M, Diehl S, Schotte R, Nagasawa M, Wijnands E, et al. STAT5 regulates the self-renewal capacity and differentiation of human memory B cells and controls Bcl-6 expression. Nature immunology. 2005;6(3):303-13. 158. Toyama H, Okada S, Hatano M, Takahashi Y, Takeda N, Ichii H, et al. Memory B cells without somatic hypermutation are generated from Bcl6-deficient B cells. Immunity. 2002;17(3):329-39. 159. Kuo TC, Shaffer AL, Haddad J, Jr., Choi YS, Staudt LM, Calame K. Repression of BCL-6 is required for the formation of human memory B cells in vitro. The Journal of experimental medicine. 2007;204(4):819-30. 160. Crotty S. Follicular helper CD4 T cells (TFH). Annual review of immunology. 2011;29:621-63. 161. Chang PP, Barral P, Fitch J, Pratama A, Ma CS, Kallies A, et al. Identification of Bcl-6-dependent follicular helper NKT cells that provide cognate help for B cell responses. Nature immunology. 2012;13(1):35-U56. 162. Okada T, Miller MJ, Parker I, Krummel MF, Neighbors M, Hartley SB, et al. Antigen-engaged B cells undergo chemotaxis toward the T zone and form motile conjugates with helper T cells. PLoS biology. 2005;3(6):e150. 163. Garside P, Ingulli E, Merica RR, Johnson JG, Noelle RJ, Jenkins MK. Visualization of specific B and T lymphocyte interactions in the lymph node. Science. 1998;281(5373):96-9. 164. Bajenoff M, Granjeaud S, Guerder S. The strategy of T cell antigen-presenting cell encounter in antigen-draining lymph nodes revealed by imaging of initial T cell activation. The Journal of experimental medicine. 2003;198(5):715-24. 165. Batista FD, Neuberger MS. Affinity dependence of the B cell response to antigen: a threshold, a ceiling, and the importance of off-rate. Immunity. 1998;8(6):751- 9. 166. Davidson HW, Watts C. Epitope-directed processing of specific antigen by B lymphocytes. The Journal of cell biology. 1989;109(1):85-92. 167. Watts C, Lanzavecchia A. Suppressive effect of antibody on processing of T cell epitopes. The Journal of experimental medicine. 1993;178(4):1459-63.

172 168. Liang B, Mamula MJ. Molecular mimicry and the role of B lymphocytes in the processing of autoantigens. Cellular and molecular life sciences : CMLS. 2000;57(4):561-8. 169. Lankar D, Vincent-Schneider H, Briken V, Yokozeki T, Raposo G, Bonnerot C. Dynamics of major histocompatibility complex class II compartments during B cell receptor-mediated cell activation. The Journal of experimental medicine. 2002;195(4):461-72. 170. Granboulan M, Lankar D, Raposo G, Bonnerot C, Hivroz C. Phosphoinositide 3-kinase activation by Igbeta controls de novo formation of an antigen-processing compartment. The Journal of biological chemistry. 2003;278(6):4331-8. 171. Siemasko K, Eisfelder BJ, Williamson E, Kabak S, Clark MR. Cutting edge: signals from the B lymphocyte antigen receptor regulate MHC class II containing late endosomes. Journal of immunology. 1998;160(11):5203-8. 172. Bouaziz JD, Yanaba K, Venturi GM, Wang Y, Tisch RM, Poe JC, et al. Therapeutic B cell depletion impairs adaptive and autoreactive CD4+ T cell activation in mice. Proceedings of the National Academy of Sciences of the United States of America. 2007;104(52):20878-83. 173. Constant S, Schweitzer N, West J, Ranney P, Bottomly K. B lymphocytes can be competent antigen-presenting cells for priming CD4+ T cells to protein antigens in vivo. Journal of immunology. 1995;155(8):3734-41. 174. Linton PJ, Bautista B, Biederman E, Bradley ES, Harbertson J, Kondrack RM, et al. Costimulation via OX40L expressed by B cells is sufficient to determine the extent of primary CD4 cell expansion and Th2 cytokine secretion in vivo. The Journal of experimental medicine. 2003;197(7):875-83. 175. Eynon EE, Parker DC. Small B cells as antigen-presenting cells in the induction of tolerance to soluble protein antigens. The Journal of experimental medicine. 1992;175(1):131-8. 176. Fuchs EJ, Matzinger P. B cells turn off virgin but not memory T cells. Science. 1992;258(5085):1156-9. 177. Harris DP, Haynes L, Sayles PC, Duso DK, Eaton SM, Lepak NM, et al. Reciprocal regulation of polarized cytokine production by effector B and T cells. Nature immunology. 2000;1(6):475-82. 178. Bermejo DA, Jackson SW, Gorosito-Serran M, Acosta-Rodriguez EV, Amezcua-Vesely MC, Sather BD, et al. Trypanosoma cruzi trans-sialidase initiates a program independent of the transcription factors RORgammat and Ahr that leads to IL- 17 production by activated B cells. Nature immunology. 2013;14(5):514-22. 179. Mauri C, Bosma A. Immune regulatory function of B cells. Annual review of immunology. 2012;30:221-41. 180. Turk JL, Parker D, Poulter LW. Functional aspects of the selective depletion of lymphoid tissue by cyclophosphamide. Immunology. 1972;23(4):493-501. 181. Katz SI, Parker D, Sommer G, Turk JL. Suppressor cells in normal immunisation as a basic homeostatic phenomenon. Nature. 1974;248(449):612-4. 182. Katz SI, Parker D, Turk JL. B-cell suppression of delayed hypersensitivity reactions. Nature. 1974;251(5475):550-1. 183. Neta R, Salvin SB. Specific suppression of delayed hypersensitivity: the possible presence of a suppressor B cell in the regulation of delayed hypersensitivity. Journal of immunology. 1974;113(6):1716-25. 184. Kitamura D, Roes J, Kuhn R, Rajewsky K. A B cell-deficient mouse by targeted disruption of the membrane exon of the immunoglobulin mu chain gene. Nature. 1991;350(6317):423-6.

173 185. Wolf SD, Dittel BN, Hardardottir F, Janeway CA, Jr. Experimental autoimmune encephalomyelitis induction in genetically B cell-deficient mice. The Journal of experimental medicine. 1996;184(6):2271-8. 186. Willenborg DO, Prowse SJ. Immunoglobulin-deficient rats fail to develop experimental allergic encephalomyelitis. Journal of neuroimmunology. 1983;5(2):99- 109. 187. Willenborg DO, Sjollema P, Danta G. Immunoglobulin deficient rats as donors and recipients of effector cells of allergic encephalomyelitis. Journal of neuroimmunology. 1986;11(2):93-103. 188. Myers KJ, Sprent J, Dougherty JP, Ron Y. Synergy between encephalitogenic T cells and myelin basic protein-specific antibodies in the induction of experimental autoimmune encephalomyelitis. Journal of neuroimmunology. 1992;41(1):1-8. 189. Moulin V, Andris F, Thielemans K, Maliszewski C, Urbain J, Moser M. B lymphocytes regulate dendritic cell (DC) function in vivo: increased interleukin 12 production by DCs from B cell-deficient mice results in T helper cell type 1 deviation. The Journal of experimental medicine. 2000;192(4):475-82. 190. Gonnella PA, Waldner HP, Weiner HL. B cell-deficient (mu MT) mice have alterations in the cytokine microenvironment of the gut-associated lymphoid tissue (GALT) and a defect in the low dose mechanism of oral tolerance. Journal of immunology. 2001;166(7):4456-64. 191. Mizoguchi A, Mizoguchi E, Takedatsu H, Blumberg RS, Bhan AK. Chronic intestinal inflammatory condition generates IL-10-producing regulatory B cell subset characterized by CD1d upregulation. Immunity. 2002;16(2):219-30. 192. Mizoguchi A, Mizoguchi E, Smith RN, Preffer FI, Bhan AK. Suppressive role of B cells in chronic colitis of T cell receptor alpha mutant mice. The Journal of experimental medicine. 1997;186(10):1749-56. 193. Mizoguchi E, Mizoguchi A, Preffer FI, Bhan AK. Regulatory role of mature B cells in a murine model of inflammatory bowel disease. International immunology. 2000;12(5):597-605. 194. Fillatreau S, Sweenie CH, McGeachy MJ, Gray D, Anderton SM. B cells regulate autoimmunity by provision of IL-10. Nature immunology. 2002;3(10):944-50. 195. Mauri C, Mars LT, Londei M. Therapeutic activity of agonistic monoclonal antibodies against CD40 in a chronic autoimmune inflammatory process. Nature medicine. 2000;6(6):673-9. 196. Mauri C, Gray D, Mushtaq N, Londei M. Prevention of arthritis by interleukin 10-producing B cells. The Journal of experimental medicine. 2003;197(4):489-501. 197. Carter NA, Vasconcellos R, Rosser EC, Tulone C, Munoz-Suano A, Kamanaka M, et al. Mice lacking endogenous IL-10-producing regulatory B cells develop exacerbated disease and present with an increased frequency of Th1/Th17 but a decrease in regulatory T cells. Journal of immunology. 2011;186(10):5569-79. 198. Carter NA, Rosser EC, Mauri C. Interleukin-10 produced by B cells is crucial for the suppression of Th17/Th1 responses, induction of T regulatory type 1 cells and reduction of collagen-induced arthritis. Arthritis research & therapy. 2012;14(1):R32. 199. Sun JB, Flach CF, Czerkinsky C, Holmgren J. B lymphocytes promote expansion of regulatory T cells in oral tolerance: powerful induction by antigen coupled to cholera toxin B subunit. Journal of immunology. 2008;181(12):8278-87. 200. Tadmor T, Zhang Y, Cho HM, Podack ER, Rosenblatt JD. The absence of B lymphocytes reduces the number and function of T-regulatory cells and enhances the anti-tumor response in a murine tumor model. Cancer immunology, immunotherapy : CII. 2011;60(5):609-19.

174 201. Chesneau M, Michel L, Degauque N, Brouard S. Regulatory B cells and tolerance in transplantation: from animal models to human. Frontiers in immunology. 2013;4:497. 202. Inoue S, Leitner WW, Golding B, Scott D. Inhibitory effects of B cells on antitumor immunity. Cancer research. 2006;66(15):7741-7. 203. Madan R, Demircik F, Surianarayanan S, Allen JL, Divanovic S, Trompette A, et al. Nonredundant roles for B cell-derived IL-10 in immune counter-regulation. Journal of immunology. 2009;183(4):2312-20. 204. Evans JG, Chavez-Rueda KA, Eddaoudi A, Meyer-Bahlburg A, Rawlings DJ, Ehrenstein MR, et al. Novel suppressive function of transitional 2 B cells in experimental arthritis. Journal of immunology. 2007;178(12):7868-78. 205. Blair PA, Chavez-Rueda KA, Evans JG, Shlomchik MJ, Eddaoudi A, Isenberg DA, et al. Selective targeting of B cells with agonistic anti-CD40 is an efficacious strategy for the generation of induced regulatory T2-like B cells and for the suppression of lupus in MRL/lpr mice. Journal of immunology. 2009;182(6):3492-502. 206. Barr TA, Brown S, Ryan G, Zhao J, Gray D. TLR-mediated stimulation of APC: Distinct cytokine responses of B cells and dendritic cells. European journal of immunology. 2007;37(11):3040-53. 207. Mangan NE, Fallon RE, Smith P, van Rooijen N, McKenzie AN, Fallon PG. Helminth infection protects mice from anaphylaxis via IL-10-producing B cells. Journal of immunology. 2004;173(10):6346-56. 208. Mangan NE, van Rooijen N, McKenzie AN, Fallon PG. Helminth-modified pulmonary immune response protects mice from allergen-induced airway hyperresponsiveness. Journal of immunology. 2006;176(1):138-47. 209. Amu S, Saunders SP, Kronenberg M, Mangan NE, Atzberger A, Fallon PG. Regulatory B cells prevent and reverse allergic airway inflammation via FoxP3-positive T regulatory cells in a murine model. The Journal of allergy and clinical immunology. 2010;125(5):1114-24 e8. 210. Schioppa T, Moore R, Thompson RG, Rosser EC, Kulbe H, Nedospasov S, et al. B regulatory cells and the tumor-promoting actions of TNF-alpha during squamous carcinogenesis. Proceedings of the National Academy of Sciences of the United States of America. 2011;108(26):10662-7. 211. Sayi A, Kohler E, Toller IM, Flavell RA, Muller W, Roers A, et al. TLR-2- activated B cells suppress Helicobacter-induced preneoplastic gastric immunopathology by inducing T regulatory-1 cells. Journal of immunology. 2011;186(2):878-90. 212. Moreau A, Blair PA, Chai JG, Ratnasothy K, Stolarczyk E, Alhabbab R, et al. Transitional-2 B cells acquire regulatory function during tolerance induction and contribute to allograft survival. European journal of immunology. 2015;45(3):843-53. 213. Inaoki M, Sato S, Weintraub BC, Goodnow CC, Tedder TF. CD19-regulated signaling thresholds control peripheral tolerance and autoantibody production in B lymphocytes. The Journal of experimental medicine. 1997;186(11):1923-31. 214. Matsushita T, Fujimoto M, Hasegawa M, Komura K, Takehara K, Tedder TF, et al. Inhibitory role of CD19 in the progression of experimental autoimmune encephalomyelitis by regulating cytokine response. The American journal of pathology. 2006;168(3):812-21. 215. Watanabe R, Fujimoto M, Ishiura N, Kuwano Y, Nakashima H, Yazawa N, et al. CD19 expression in B cells is important for suppression of contact hypersensitivity. The American journal of pathology. 2007;171(2):560-70. 216. Yanaba K, Bouaziz JD, Haas KM, Poe JC, Fujimoto M, Tedder TF. A regulatory B cell subset with a unique CD1dhiCD5+ phenotype controls T cell- dependent inflammatory responses. Immunity. 2008;28(5):639-50.

175 217. O'Garra A, Chang R, Go N, Hastings R, Haughton G, Howard M. Ly-1 B (B-1) cells are the main source of B cell-derived interleukin 10. European journal of immunology. 1992;22(3):711-7. 218. Velupillai P, Secor WE, Horauf AM, Harn DA. B-1 cell (CD5+B220+) outgrowth in murine schistosomiasis is genetically restricted and is largely due to activation by polylactosamine sugars. Journal of immunology. 1997;158(1):338-44. 219. Al-Qaoud KM, Fleischer B, Hoerauf A. The Xid defect imparts susceptibility to experimental murine filariosis--association with a lack of antibody and IL-10 production by B cells in response to phosphorylcholine. International immunology. 1998;10(1):17-25. 220. Sun CM, Deriaud E, Leclerc C, Lo-Man R. Upon TLR9 signaling, CD5+ B cells control the IL-12-dependent Th1-priming capacity of neonatal DCs. Immunity. 2005;22(4):467-77. 221. Gray M, Miles K, Salter D, Gray D, Savill J. Apoptotic cells protect mice from autoimmune inflammation by the induction of regulatory B cells. Proc Natl Acad Sci U S A. 2007;104(35):14080-5. 222. Brummel R, Lenert P. Activation of marginal zone B cells from lupus mice with type A(D) CpG-oligodeoxynucleotides. Journal of immunology. 2005;174(4):2429-34. 223. Ding Q, Yeung M, Camirand G, Zeng Q, Akiba H, Yagita H, et al. Regulatory B cells are identified by expression of TIM-1 and can be induced through TIM-1 ligation to promote tolerance in mice. The Journal of clinical investigation. 2011;121(9):3645- 56. 224. Xiao S, Brooks CR, Zhu C, Wu C, Sweere JM, Petecka S, et al. Defect in regulatory B-cell function and development of systemic autoimmunity in T-cell Ig mucin 1 (Tim-1) mucin domain-mutant mice. Proceedings of the National Academy of Sciences of the United States of America. 2012;109(30):12105-10. 225. Rafei M, Hsieh J, Zehntner S, Li M, Forner K, Birman E, et al. A granulocyte- macrophage colony-stimulating factor and interleukin-15 fusokine induces a regulatory B cell population with immune suppressive properties. Nature medicine. 2009;15(9):1038-45. 226. Matsumoto M, Baba A, Yokota T, Nishikawa H, Ohkawa Y, Kayama H, et al. Interleukin-10-producing plasmablasts exert regulatory function in autoimmune inflammation. Immunity. 2014;41(6):1040-51. 227. Shen P, Roch T, Lampropoulou V, O'Connor RA, Stervbo U, Hilgenberg E, et al. IL-35-producing B cells are critical regulators of immunity during autoimmune and infectious diseases. Nature. 2014;507(7492):366-70. 228. Wilson MS, Taylor MD, O'Gorman MT, Balic A, Barr TA, Filbey K, et al. Helminth-induced CD19+CD23hi B cells modulate experimental allergic and autoimmune inflammation. European journal of immunology. 2010;40(6):1682-96. 229. Tian J, Zekzer D, Hanssen L, Lu Y, Olcott A, Kaufman DL. Lipopolysaccharide-activated B cells down-regulate Th1 immunity and prevent autoimmune diabetes in nonobese diabetic mice. Journal of immunology. 2001;167(2):1081-9. 230. Montandon R, Korniotis S, Layseca-Espinosa E, Gras C, Megret J, Ezine S, et al. Innate pro-B-cell progenitors protect against type 1 diabetes by regulating autoimmune effector T cells. Proceedings of the National Academy of Sciences of the United States of America. 2013;110(24):E2199-208. 231. Minagawa R, Okano S, Tomita Y, Kishihara K, Yamada H, Nomoto K, et al. The critical role of Fas-Fas ligand interaction in donor-specific transfusion-induced tolerance to H-Y antigen. Transplantation. 2004;78(6):799-806.

176 232. Lundy SK, Berlin AA, Martens TF, Lukacs NW. Deficiency of regulatory B cells increases allergic airway inflammation. Inflammation research : official journal of the European Histamine Research Society [et al]. 2005;54(12):514-21. 233. Lundy SK, Fox DA. Reduced Fas ligand-expressing splenic CD5+ B lymphocytes in severe collagen-induced arthritis. Arthritis research & therapy. 2009;11(4):R128. 234. Zhang Z, Lazarovits A, Grant D, Garcia B, Stiller C, Zhong R. CD45RB monoclonal antibody induces tolerance in the mouse kidney graft, but fails to prevent small bowel graft rejection. Transplantation proceedings. 1996;28(5):2514. 235. Deng S, Moore DJ, Huang X, Mohiuddin M, Lee MKt, Velidedeoglu E, et al. Antibody-induced transplantation tolerance that is dependent on thymus-derived regulatory T cells. Journal of immunology. 2006;176(5):2799-807. 236. Deng S, Moore DJ, Huang X, Lian MM, Mohiuddin M, Velededeoglu E, et al. Cutting edge: transplant tolerance induced by anti-CD45RB requires B lymphocytes. Journal of immunology. 2007;178(10):6028-32. 237. Huang X, Moore DJ, Mohiuddin M, Lian MM, Kim JI, Sonawane S, et al. Inhibition of ICAM-1/LFA-1 interactions prevents B-cell-dependent anti-CD45RB- induced transplantation tolerance. Transplantation. 2008;85(5):675-80. 238. Zhao G, Moore DJ, Lee KM, Kim JI, Duff PE, O'Connor MR, et al. An unexpected counter-regulatory role of IL-10 in B-lymphocyte-mediated transplantation tolerance. American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons. 2010;10(4):796- 801. 239. Lee KM, Stott RT, Zhao G, SooHoo J, Xiong W, Lian MM, et al. TGF-beta- producing regulatory B cells induce regulatory T cells and promote transplantation tolerance. European journal of immunology. 2014;44(6):1728-36. 240. Ray A, Basu S, Williams CB, Salzman NH, Dittel BN. A novel IL-10- independent regulatory role for B cells in suppressing autoimmunity by maintenance of regulatory T cells via GITR ligand. Journal of immunology. 2012;188(7):3188-98. 241. Bebo BF, Jr., Fyfe-Johnson A, Adlard K, Beam AG, Vandenbark AA, Offner H. Low-dose estrogen therapy ameliorates experimental autoimmune encephalomyelitis in two different inbred mouse strains. Journal of immunology. 2001;166(3):2080-9. 242. Bodhankar S, Wang C, Vandenbark AA, Offner H. Estrogen-induced protection against experimental autoimmune encephalomyelitis is abrogated in the absence of B cells. European journal of immunology. 2011;41(4):1165-75. 243. Bodhankar S, Vandenbark AA, Offner H. Oestrogen treatment of experimental autoimmune encephalomyelitis requires 17beta-oestradiol-receptor-positive B cells that up-regulate PD-1 on CD4+ Foxp3+ regulatory T cells. Immunology. 2012;137(4):282- 93. 244. Bodhankar S, Galipeau D, Vandenbark AA, Offner H. PD-1 Interaction with PD-L1 but not PD-L2 on B-cells Mediates Protective Effects of Estrogen against EAE. Journal of clinical & cellular immunology. 2013;4(3):143. 245. Khan AR, Hams E, Floudas A, Sparwasser T, Weaver CT, Fallon PG. PD-L1hi B cells are critical regulators of humoral immunity. Nature communications. 2015;6:5997. 246. Thibult ML, Mamessier E, Gertner-Dardenne J, Pastor S, Just-Landi S, Xerri L, et al. PD-1 is a novel regulator of human B-cell activation. International immunology. 2013;25(2):129-37. 247. Vasconcellos R, Carter NA, Rosser EC, Mauri C. IL-12p35 subunit contributes to autoimmunity by limiting IL-27-driven regulatory responses. Journal of immunology. 2011;187(6):3402-12.

177 248. Lacroix-Desmazes S, Bayry J, Kaveri SV, Hayon-Sonsino D, Thorenoor N, Charpentier J, et al. High levels of catalytic antibodies correlate with favorable outcome in sepsis. Proceedings of the National Academy of Sciences of the United States of America. 2005;102(11):4109-13. 249. Wootla B, Nicoletti A, Patey N, Dimitrov JD, Legendre C, Christophe OD, et al. Hydrolysis of coagulation factors by circulating IgG is associated with a reduced risk for chronic allograft nephropathy in renal transplanted patients. Journal of immunology. 2008;180(12):8455-60. 250. Lindner S, Dahlke K, Sontheimer K, Hagn M, Kaltenmeier C, Barth TF, et al. Interleukin 21-induced granzyme B-expressing B cells infiltrate tumors and regulate T cells. Cancer research. 2013;73(8):2468-79. 251. Kaku H, Cheng KF, Al-Abed Y, Rothstein TL. A novel mechanism of B cell- mediated immune suppression through CD73 expression and adenosine production. Journal of immunology. 2014;193(12):5904-13. 252. Banchereau J, Bazan F, Blanchard D, Briere F, Galizzi JP, van Kooten C, et al. The CD40 antigen and its ligand. Annual review of immunology. 1994;12:881-922. 253. Cella M, Scheidegger D, Palmer-Lehmann K, Lane P, Lanzavecchia A, Alber G. Ligation of CD40 on dendritic cells triggers production of high levels of interleukin-12 and enhances T cell stimulatory capacity: T-T help via APC activation. The Journal of experimental medicine. 1996;184(2):747-52. 254. Miyashita T, McIlraith MJ, Grammer AC, Miura Y, Attrep JF, Shimaoka Y, et al. Bidirectional regulation of human B cell responses by CD40-CD40 ligand interactions. Journal of immunology. 1997;158(10):4620-33. 255. Lampropoulou V, Hoehlig K, Roch T, Neves P, Calderon Gomez E, Sweenie CH, et al. TLR-activated B cells suppress T cell-mediated autoimmunity. Journal of immunology. 2008;180(7):4763-73. 256. Neves P, Lampropoulou V, Calderon-Gomez E, Roch T, Stervbo U, Shen P, et al. Signaling via the MyD88 adaptor protein in B cells suppresses protective immunity during Salmonella typhimurium infection. Immunity. 2010;33(5):777-90. 257. Buenafe AC, Bourdette DN. Lipopolysaccharide pretreatment modulates the disease course in experimental autoimmune encephalomyelitis. Journal of neuroimmunology. 2007;182(1-2):32-40. 258. Krieg AM. CpG motifs in bacterial DNA and their immune effects. Annual review of immunology. 2002;20:709-60. 259. Christensen SR, Shupe J, Nickerson K, Kashgarian M, Flavell RA, Shlomchik MJ. Toll-like receptor 7 and TLR9 dictate autoantibody specificity and have opposing inflammatory and regulatory roles in a murine model of lupus. Immunity. 2006;25(3):417-28. 260. Tao K, Fujii M, Tsukumo S, Maekawa Y, Kishihara K, Kimoto Y, et al. Genetic variations of Toll-like receptor 9 predispose to systemic lupus erythematosus in Japanese population. Annals of the rheumatic diseases. 2007;66(7):905-9. 261. Quintana FJ, Rotem A, Carmi P, Cohen IR. Vaccination with empty plasmid DNA or CpG oligonucleotide inhibits diabetes in nonobese diabetic mice: modulation of spontaneous 60-kDa heat shock protein autoimmunity. Journal of immunology. 2000;165(11):6148-55. 262. Wu HJ, Sawaya H, Binstadt B, Brickelmaier M, Blasius A, Gorelik L, et al. Inflammatory arthritis can be reined in by CpG-induced DC-NK cell cross talk. The Journal of experimental medicine. 2007;204(8):1911-22. 263. Gray D, Gray M, Barr T. Innate responses of B cells. European journal of immunology. 2007;37(12):3304-10. 264. Miles K, Heaney J, Sibinska Z, Salter D, Savill J, Gray D, et al. A tolerogenic role for Toll-like receptor 9 is revealed by B-cell interaction with DNA complexes

178 expressed on apoptotic cells. Proceedings of the National Academy of Sciences of the United States of America. 2012;109(3):887-92. 265. Zhivaki D, Lemoine S, Lim A, Morva A, Vidalain PO, Schandene L, et al. Respiratory Syncytial Virus Infects Regulatory B Cells in Human Neonates via Chemokine Receptor CX3CR1 and Promotes Lung Disease Severity. Immunity. 2017;46(2):301-14. 266. Rosser EC, Oleinika K, Tonon S, Doyle R, Bosma A, Carter NA, et al. Regulatory B cells are induced by gut microbiota-driven interleukin-1beta and interleukin-6 production. Nature medicine. 2014;20(11):1334-9. 267. Menon M, Blair PA, Isenberg DA, Mauri C. A Regulatory Feedback between Plasmacytoid Dendritic Cells and Regulatory B Cells Is Aberrant in Systemic Lupus Erythematosus. Immunity. 2016;44(3):683-97. 268. Yoshizaki A, Miyagaki T, DiLillo DJ, Matsushita T, Horikawa M, Kountikov EI, et al. Regulatory B cells control T-cell autoimmunity through IL-21-dependent cognate interactions. Nature. 2012;491(7423):264-8. 269. Wang RX, Yu CR, Dambuza IM, Mahdi RM, Dolinska MB, Sergeev YV, et al. Interleukin-35 induces regulatory B cells that suppress autoimmune disease. Nature medicine. 2014;20(6):633-41. 270. Sattler S, Ling GS, Xu D, Hussaarts L, Romaine A, Zhao H, et al. IL-10- producing regulatory B cells induced by IL-33 (Breg(IL-33)) effectively attenuate mucosal inflammatory responses in the gut. Journal of autoimmunity. 2014;50:107-22. 271. Al-Alwan MM, Okkenhaug K, Vanhaesebroeck B, Hayflick JS, Marshall AJ. Requirement for phosphoinositide 3-kinase p110delta signaling in B cell antigen receptor-mediated antigen presentation. Journal of immunology. 2007;178(4):2328-35. 272. Okkenhaug K, Bilancio A, Farjot G, Priddle H, Sancho S, Peskett E, et al. Impaired B and T cell antigen receptor signaling in p110delta PI 3-kinase mutant mice. Science. 2002;297(5583):1031-4. 273. Mauri C, Menon M. Human regulatory B cells in health and disease: therapeutic potential. The Journal of clinical investigation. 2017;127(3):772-9. 274. Blair PA, Norena LY, Flores-Borja F, Rawlings DJ, Isenberg DA, Ehrenstein MR, et al. CD19(+)CD24(hi)CD38(hi) B cells exhibit regulatory capacity in healthy individuals but are functionally impaired in systemic Lupus Erythematosus patients. Immunity. 2010;32(1):129-40. 275. Bosma A, Abdel-Gadir A, Isenberg DA, Jury EC, Mauri C. Lipid-antigen presentation by CD1d(+) B cells is essential for the maintenance of invariant natural killer T cells. Immunity. 2012;36(3):477-90. 276. Flores-Borja F, Bosma A, Ng D, Reddy V, Ehrenstein MR, Isenberg DA, et al. CD19+CD24hiCD38hi B cells maintain regulatory T cells while limiting TH1 and TH17 differentiation. Science translational medicine. 2013;5(173):173ra23. 277. Knippenberg S, Peelen E, Smolders J, Thewissen M, Menheere P, Cohen Tervaert JW, et al. Reduction in IL-10 producing B cells (Breg) in multiple sclerosis is accompanied by a reduced naive/memory Breg ratio during a relapse but not in remission. Journal of neuroimmunology. 2011;239(1-2):80-6. 278. Hayashi M, Yanaba K, Umezawa Y, Yoshihara Y, Kikuchi S, Ishiuji Y, et al. IL-10-producing regulatory B cells are decreased in patients with psoriasis. Journal of dermatological science. 2016;81(2):93-100. 279. Correale J, Farez M, Razzitte G. Helminth infections associated with multiple sclerosis induce regulatory B cells. Annals of neurology. 2008;64(2):187-99. 280. Iwata Y, Matsushita T, Horikawa M, Dilillo DJ, Yanaba K, Venturi GM, et al. Characterization of a rare IL-10-competent B-cell subset in humans that parallels mouse regulatory B10 cells. Blood. 2011;117(2):530-41.

179 281. Das A, Ellis G, Pallant C, Lopes AR, Khanna P, Peppa D, et al. IL-10-producing regulatory B cells in the pathogenesis of chronic hepatitis B virus infection. Journal of immunology. 2012;189(8):3925-35. 282. Siewe B, Stapleton JT, Martinson J, Keshavarzian A, Kazmi N, Demarais PM, et al. Regulatory B cell frequency correlates with markers of HIV disease progression and attenuates anti-HIV CD8(+) T cell function in vitro. Journal of leukocyte biology. 2013;93(5):811-8. 283. Liu J, Zhan W, Kim CJ, Clayton K, Zhao H, Lee E, et al. IL-10-producing B cells are induced early in HIV-1 infection and suppress HIV-1-specific T cell responses. PloS one. 2014;9(2):e89236. 284. Novak J, Lehuen A. Mechanism of regulation of autoimmunity by iNKT cells. Cytokine. 2011;53(3):263-70. 285. Meyer EH, DeKruyff RH, Umetsu DT. iNKT cells in allergic disease. Current topics in microbiology and immunology. 2007;314:269-91. 286. Brigl M, Brenner MB. How invariant natural killer T cells respond to infection by recognizing microbial or endogenous lipid antigens. Seminars in immunology. 2010;22(2):79-86. 287. Tupin E, Kinjo Y, Kronenberg M. The unique role of natural killer T cells in the response to microorganisms. Nature reviews Microbiology. 2007;5(6):405-17. 288. Vivier E, Ugolini S, Blaise D, Chabannon C, Brossay L. Targeting natural killer cells and natural killer T cells in cancer. Nature reviews Immunology. 2012;12(4):239- 52. 289. Makino Y, Kanno R, Ito T, Higashino K, Taniguchi M. Predominant expression of invariant V alpha 14+ TCR alpha chain in NK1.1+ T cell populations. International immunology. 1995;7(7):1157-61. 290. Sykes M. Unusual T cell populations in adult murine bone marrow. Prevalence of CD3+CD4-CD8- and alpha beta TCR+NK1.1+ cells. Journal of immunology. 1990;145(10):3209-15. 291. Bendelac A, Savage PB, Teyton L. The biology of NKT cells. Annual review of immunology. 2007;25:297-336. 292. Treiner E, Duban L, Bahram S, Radosavljevic M, Wanner V, Tilloy F, et al. Selection of evolutionarily conserved mucosal-associated invariant T cells by MR1. Nature. 2003;422(6928):164-9. 293. Uldrich AP, Le Nours J, Pellicci DG, Gherardin NA, McPherson KG, Lim RT, et al. CD1d-lipid antigen recognition by the gammadelta TCR. Nature immunology. 2013;14(11):1137-45. 294. Godfrey DI, Kronenberg M. Going both ways: immune regulation via CD1d- dependent NKT cells. The Journal of clinical investigation. 2004;114(10):1379-88. 295. Brennan PJ, Brigl M, Brenner MB. Invariant natural killer T cells: an innate activation scheme linked to diverse effector functions. Nature reviews Immunology. 2013;13(2):101-17. 296. Park SH, Weiss A, Benlagha K, Kyin T, Teyton L, Bendelac A. The mouse CD1d-restricted repertoire is dominated by a few autoreactive T cell receptor families. The Journal of experimental medicine. 2001;193(8):893-904. 297. Arrenberg P, Halder R, Dai Y, Maricic I, Kumar V. Oligoclonality and innate- like features in the TCR repertoire of type II NKT cells reactive to a beta-linked self- glycolipid. Proceedings of the National Academy of Sciences of the United States of America. 2010;107(24):10984-9. 298. Girardi E, Maricic I, Wang J, Mac TT, Iyer P, Kumar V, et al. Type II natural killer T cells use features of both innate-like and conventional T cells to recognize sulfatide self antigens. Nature immunology. 2012;13(9):851-6.

180 299. Patel O, Pellicci DG, Gras S, Sandoval-Romero ML, Uldrich AP, Mallevaey T, et al. Recognition of CD1d-sulfatide mediated by a type II natural killer T cell antigen receptor. Nature immunology. 2012;13(9):857-63. 300. Stetson DB, Mohrs M, Reinhardt RL, Baron JL, Wang ZE, Gapin L, et al. Constitutive cytokine mRNAs mark natural killer (NK) and NK T cells poised for rapid effector function. The Journal of experimental medicine. 2003;198(7):1069-76. 301. Cohen NR, Brennan PJ, Shay T, Watts GF, Brigl M, Kang J, et al. Shared and distinct transcriptional programs underlie the hybrid nature of iNKT cells. Nature immunology. 2013;14(1):90-9. 302. Gumperz JE, Miyake S, Yamamura T, Brenner MB. Functionally distinct subsets of CD1d-restricted natural killer T cells revealed by CD1d tetramer staining. The Journal of experimental medicine. 2002;195(5):625-36. 303. Lee PT, Benlagha K, Teyton L, Bendelac A. Distinct functional lineages of human V(alpha)24 natural killer T cells. The Journal of experimental medicine. 2002;195(5):637-41. 304. Montoya CJ, Pollard D, Martinson J, Kumari K, Wasserfall C, Mulder CB, et al. Characterization of human invariant natural killer T subsets in health and disease using a novel invariant natural killer T cell-clonotypic monoclonal antibody, 6B11. Immunology. 2007;122(1):1-14. 305. Kita H, Naidenko OV, Kronenberg M, Ansari AA, Rogers P, He XS, et al. Quantitation and phenotypic analysis of natural killer T cells in primary biliary cirrhosis using a human CD1d tetramer. Gastroenterology. 2002;123(4):1031-43. 306. Kenna T, Golden-Mason L, Porcelli SA, Koezuka Y, Hegarty JE, O'Farrelly C, et al. NKT cells from normal and tumor-bearing human livers are phenotypically and functionally distinct from murine NKT cells. Journal of immunology. 2003;171(4):1775-9. 307. Cui J, Shin T, Kawano T, Sato H, Kondo E, Toura I, et al. Requirement for Valpha14 NKT cells in IL-12-mediated rejection of tumors. Science. 1997;278(5343):1623-6. 308. Guo J, Hawwari A, Li H, Sun Z, Mahanta SK, Littman DR, et al. Regulation of the TCRalpha repertoire by the survival window of CD4(+)CD8(+) thymocytes. Nature immunology. 2002;3(5):469-76. 309. Mendiratta SK, Martin WD, Hong S, Boesteanu A, Joyce S, Van Kaer L. CD1d1 mutant mice are deficient in natural T cells that promptly produce IL-4. Immunity. 1997;6(4):469-77. 310. Bendelac A. Positive selection of mouse NK1+ T cells by CD1-expressing cortical thymocytes. The Journal of experimental medicine. 1995;182(6):2091-6. 311. Coles MC, Raulet DH. NK1.1+ T cells in the liver arise in the thymus and are selected by interactions with class I molecules on CD4+CD8+ cells. Journal of immunology. 2000;164(5):2412-8. 312. Facciotti F, Ramanjaneyulu GS, Lepore M, Sansano S, Cavallari M, Kistowska M, et al. Peroxisome-derived lipids are self antigens that stimulate invariant natural killer T cells in the thymus. Nature immunology. 2012;13(5):474-80. 313. Griewank K, Borowski C, Rietdijk S, Wang N, Julien A, Wei DG, et al. Homotypic interactions mediated by Slamf1 and Slamf6 receptors control NKT cell lineage development. Immunity. 2007;27(5):751-62. 314. Chung B, Aoukaty A, Dutz J, Terhorst C, Tan R. Signaling lymphocytic activation molecule-associated protein controls NKT cell functions. Journal of immunology. 2005;174(6):3153-7. 315. Nichols KE, Hom J, Gong SY, Ganguly A, Ma CS, Cannons JL, et al. Regulation of NKT cell development by SAP, the protein defective in XLP. Nature medicine. 2005;11(3):340-5.

181 316. Pasquier B, Yin L, Fondaneche MC, Relouzat F, Bloch-Queyrat C, Lambert N, et al. Defective NKT cell development in mice and humans lacking the adapter SAP, the X-linked lymphoproliferative syndrome gene product. The Journal of experimental medicine. 2005;201(5):695-701. 317. Eberl G, Lowin-Kropf B, MacDonald HR. Cutting edge: NKT cell development is selectively impaired in Fyn- deficient mice. Journal of immunology. 1999;163(8):4091-4. 318. Gadue P, Morton N, Stein PL. The Src family tyrosine kinase Fyn regulates natural killer T cell development. The Journal of experimental medicine. 1999;190(8):1189-96. 319. Kunisaki Y, Tanaka Y, Sanui T, Inayoshi A, Noda M, Nakayama T, et al. DOCK2 is required in T cell precursors for development of Valpha14 NK T cells. Journal of immunology. 2006;176(8):4640-5. 320. Savage AK, Constantinides MG, Han J, Picard D, Martin E, Li B, et al. The transcription factor PLZF directs the effector program of the NKT cell lineage. Immunity. 2008;29(3):391-403. 321. Kovalovsky D, Uche OU, Eladad S, Hobbs RM, Yi W, Alonzo E, et al. The BTB-zinc finger transcriptional regulator PLZF controls the development of invariant natural killer T cell effector functions. Nature immunology. 2008;9(9):1055-64. 322. Constantinides MG, McDonald BD, Verhoef PA, Bendelac A. A committed precursor to innate lymphoid cells. Nature. 2014;508(7496):397-401. 323. Seiler MP, Mathew R, Liszewski MK, Spooner CJ, Barr K, Meng F, et al. Elevated and sustained expression of the transcription factors Egr1 and Egr2 controls NKT lineage differentiation in response to TCR signaling. Nature immunology. 2012;13(3):264-71. 324. Savage AK, Constantinides MG, Bendelac A. Promyelocytic leukemia zinc finger turns on the effector T cell program without requirement for agonist TCR signaling. Journal of immunology. 2011;186(10):5801-6. 325. Townsend MJ, Weinmann AS, Matsuda JL, Salomon R, Farnham PJ, Biron CA, et al. T-bet regulates the terminal maturation and homeostasis of NK and Valpha14i NKT cells. Immunity. 2004;20(4):477-94. 326. Matsuda JL, Zhang Q, Ndonye R, Richardson SK, Howell AR, Gapin L. T-bet concomitantly controls migration, survival, and effector functions during the development of Valpha14i NKT cells. Blood. 2006;107(7):2797-805. 327. Chun T, Page MJ, Gapin L, Matsuda JL, Xu H, Nguyen H, et al. CD1d- expressing dendritic cells but not thymic epithelial cells can mediate negative selection of NKT cells. The Journal of experimental medicine. 2003;197(7):907-18. 328. Pellicci DG, Uldrich AP, Kyparissoudis K, Crowe NY, Brooks AG, Hammond KJ, et al. Intrathymic NKT cell development is blocked by the presence of alpha- galactosylceramide. European journal of immunology. 2003;33(7):1816-23. 329. Benlagha K, Wei DG, Veiga J, Teyton L, Bendelac A. Characterization of the early stages of thymic NKT cell development. The Journal of experimental medicine. 2005;202(4):485-92. 330. Benlagha K, Kyin T, Beavis A, Teyton L, Bendelac A. A thymic precursor to the NK T cell lineage. Science. 2002;296(5567):553-5. 331. Pellicci DG, Hammond KJ, Uldrich AP, Baxter AG, Smyth MJ, Godfrey DI. A natural killer T (NKT) cell developmental pathway iInvolving a thymus-dependent NK1.1(-)CD4(+) CD1d-dependent precursor stage. The Journal of experimental medicine. 2002;195(7):835-44. 332. Loza MJ, Perussia B. Final steps of natural killer cell maturation: a model for type 1-type 2 differentiation? Nature immunology. 2001;2(10):917-24.

182 333. Wei DG, Lee H, Park SH, Beaudoin L, Teyton L, Lehuen A, et al. Expansion and long-range differentiation of the NKT cell lineage in mice expressing CD1d exclusively on cortical thymocytes. The Journal of experimental medicine. 2005;202(2):239-48. 334. Matsuda JL, Gapin L, Sidobre S, Kieper WC, Tan JT, Ceredig R, et al. Homeostasis of V alpha 14i NKT cells. Nature immunology. 2002;3(10):966-74. 335. Ranson T, Vosshenrich CA, Corcuff E, Richard O, Laloux V, Lehuen A, et al. IL-15 availability conditions homeostasis of peripheral natural killer T cells. Proceedings of the National Academy of Sciences of the United States of America. 2003;100(5):2663-8. 336. Allende ML, Zhou D, Kalkofen DN, Benhamed S, Tuymetova G, Borowski C, et al. S1P1 receptor expression regulates emergence of NKT cells in peripheral tissues. FASEB journal : official publication of the Federation of American Societies for Experimental Biology. 2008;22(1):307-15. 337. Franki AS, Van Beneden K, Dewint P, Hammond KJ, Lambrecht S, Leclercq G, et al. A unique lymphotoxin {alpha}beta-dependent pathway regulates thymic emigration of V{alpha}14 invariant natural killer T cells. Proceedings of the National Academy of Sciences of the United States of America. 2006;103(24):9160-5. 338. Kawano T, Cui J, Koezuka Y, Toura I, Kaneko Y, Motoki K, et al. CD1d- restricted and TCR-mediated activation of valpha14 NKT cells by glycosylceramides. Science. 1997;278(5343):1626-9. 339. Kobayashi E, Motoki K, Uchida T, Fukushima H, Koezuka Y. KRN7000, a novel immunomodulator, and its antitumor activities. Oncology research. 1995;7(10- 11):529-34. 340. Miyamoto K, Miyake S, Yamamura T. A synthetic glycolipid prevents autoimmune encephalomyelitis by inducing TH2 bias of natural killer T cells. Nature. 2001;413(6855):531-4. 341. Fujii S, Shimizu K, Hemmi H, Fukui M, Bonito AJ, Chen G, et al. Glycolipid alpha-C-galactosylceramide is a distinct inducer of dendritic cell function during innate and adaptive immune responses of mice. Proceedings of the National Academy of Sciences of the United States of America. 2006;103(30):11252-7. 342. Franck RW, Tsuji M. Alpha-c-galactosylceramides: synthesis and immunology. Accounts of chemical research. 2006;39(10):692-701. 343. Schmieg J, Yang G, Franck RW, Tsuji M. Superior protection against malaria and melanoma metastases by a C-glycoside analogue of the natural killer T cell ligand alpha-Galactosylceramide. The Journal of experimental medicine. 2003;198(11):1631- 41. 344. Lu X, Song L, Metelitsa LS, Bittman R. Synthesis and evaluation of an alpha-C- galactosylceramide analogue that induces Th1-biased responses in human natural killer T cells. Chembiochem : a European journal of chemical biology. 2006;7(11):1750-6. 345. Li X, Chen G, Garcia-Navarro R, Franck RW, Tsuji M. Identification of C- glycoside analogues that display a potent biological activity against murine and human invariant natural killer T cells. Immunology. 2009;127(2):216-25. 346. Mattner J, Debord KL, Ismail N, Goff RD, Cantu C, 3rd, Zhou D, et al. Exogenous and endogenous glycolipid antigens activate NKT cells during microbial infections. Nature. 2005;434(7032):525-9. 347. Sriram V, Du W, Gervay-Hague J, Brutkiewicz RR. Cell wall glycosphingolipids of Sphingomonas paucimobilis are CD1d-specific ligands for NKT cells. European journal of immunology. 2005;35(6):1692-701. 348. Kinjo Y, Wu D, Kim G, Xing GW, Poles MA, Ho DD, et al. Recognition of bacterial glycosphingolipids by natural killer T cells. Nature. 2005;434(7032):520-5.

183 349. Kinjo Y, Tupin E, Wu D, Fujio M, Garcia-Navarro R, Benhnia MR, et al. Natural killer T cells recognize diacylglycerol antigens from pathogenic bacteria. Nature immunology. 2006;7(9):978-86. 350. Kinjo Y, Illarionov P, Vela JL, Pei B, Girardi E, Li X, et al. Invariant natural killer T cells recognize glycolipids from pathogenic Gram-positive bacteria. Nature immunology. 2011;12(10):966-74. 351. Chang YJ, Kim HY, Albacker LA, Lee HH, Baumgarth N, Akira S, et al. Influenza infection in suckling mice expands an NKT cell subset that protects against airway hyperreactivity. The Journal of clinical investigation. 2011;121(1):57-69. 352. Fischer K, Scotet E, Niemeyer M, Koebernick H, Zerrahn J, Maillet S, et al. Mycobacterial phosphatidylinositol mannoside is a natural antigen for CD1d-restricted T cells. Proceedings of the National Academy of Sciences of the United States of America. 2004;101(29):10685-90. 353. Ortaldo JR, Young HA, Winkler-Pickett RT, Bere EW, Jr., Murphy WJ, Wiltrout RH. Dissociation of NKT stimulation, cytokine induction, and NK activation in vivo by the use of distinct TCR-binding ceramides. Journal of immunology. 2004;172(2):943-53. 354. Parekh VV, Singh AK, Wilson MT, Olivares-Villagomez D, Bezbradica JS, Inazawa H, et al. Quantitative and qualitative differences in the in vivo response of NKT cells to distinct alpha- and beta-anomeric glycolipids. Journal of immunology. 2004;173(6):3693-706. 355. Stanic AK, De Silva AD, Park JJ, Sriram V, Ichikawa S, Hirabyashi Y, et al. Defective presentation of the CD1d1-restricted natural Va14Ja18 NKT lymphocyte antigen caused by beta-D-glucosylceramide synthase deficiency. Proceedings of the National Academy of Sciences of the United States of America. 2003;100(4):1849-54. 356. Brennan PJ, Tatituri RV, Brigl M, Kim EY, Tuli A, Sanderson JP, et al. Invariant natural killer T cells recognize lipid self antigen induced by microbial danger signals. Nature immunology. 2011;12(12):1202-11. 357. Wu DY, Segal NH, Sidobre S, Kronenberg M, Chapman PB. Cross-presentation of disialoganglioside GD3 to natural killer T cells. The Journal of experimental medicine. 2003;198(1):173-81. 358. Zhou D, Mattner J, Cantu C, 3rd, Schrantz N, Yin N, Gao Y, et al. Lysosomal glycosphingolipid recognition by NKT cells. Science. 2004;306(5702):1786-9. 359. Speak AO, Salio M, Neville DC, Fontaine J, Priestman DA, Platt N, et al. Implications for invariant natural killer T cell ligands due to the restricted presence of isoglobotrihexosylceramide in mammals. Proceedings of the National Academy of Sciences of the United States of America. 2007;104(14):5971-6. 360. Porubsky S, Speak AO, Luckow B, Cerundolo V, Platt FM, Grone HJ. Normal development and function of invariant natural killer T cells in mice with isoglobotrihexosylceramide (iGb3) deficiency. Proceedings of the National Academy of Sciences of the United States of America. 2007;104(14):5977-82. 361. Cox D, Fox L, Tian R, Bardet W, Skaley M, Mojsilovic D, et al. Determination of cellular lipids bound to human CD1d molecules. PloS one. 2009;4(5):e5325. 362. Haig NA, Guan Z, Li D, McMichael A, Raetz CR, Xu XN. Identification of self- lipids presented by CD1c and CD1d proteins. The Journal of biological chemistry. 2011;286(43):37692-701. 363. Muindi K, Cernadas M, Watts GF, Royle L, Neville DC, Dwek RA, et al. Activation state and intracellular trafficking contribute to the repertoire of endogenous glycosphingolipids presented by CD1d [corrected]. Proceedings of the National Academy of Sciences of the United States of America. 2010;107(7):3052-7.

184 364. Yuan W, Kang SJ, Evans JE, Cresswell P. Natural lipid ligands associated with human CD1d targeted to different subcellular compartments. Journal of immunology. 2009;182(8):4784-91. 365. Kain L, Webb B, Anderson BL, Deng S, Holt M, Costanzo A, et al. The identification of the endogenous ligands of natural killer T cells reveals the presence of mammalian alpha-linked glycosylceramides. Immunity. 2014;41(4):543-54. 366. Kain L, Costanzo A, Webb B, Holt M, Bendelac A, Savage PB, et al. Endogenous ligands of natural killer T cells are alpha-linked glycosylceramides. Molecular immunology. 2015;68(2 Pt A):94-7. 367. Borg NA, Kjer-Nielsen L, McCluskey J, Rossjohn J. Structural insight into natural killer T cell receptor recognition of CD1d. Advances in experimental medicine and biology. 2007;598:20-34. 368. McEwen-Smith RM, Salio M, Cerundolo V. CD1d-dependent endogenous and exogenous lipid antigen presentation. Current opinion in immunology. 2015;34:116-25. 369. Chiu YH, Park SH, Benlagha K, Forestier C, Jayawardena-Wolf J, Savage PB, et al. Multiple defects in antigen presentation and T cell development by mice expressing cytoplasmic tail-truncated CD1d. Nature immunology. 2002;3(1):55-60. 370. Honey K, Benlagha K, Beers C, Forbush K, Teyton L, Kleijmeer MJ, et al. Thymocyte expression of cathepsin L is essential for NKT cell development. Nature immunology. 2002;3(11):1069-74. 371. Zhou D, Cantu C, 3rd, Sagiv Y, Schrantz N, Kulkarni AB, Qi X, et al. Editing of CD1d-bound lipid antigens by endosomal lipid transfer proteins. Science. 2004;303(5657):523-7. 372. Kang SJ, Cresswell P. Saposins facilitate CD1d-restricted presentation of an exogenous lipid antigen to T cells. Nature immunology. 2004;5(2):175-81. 373. Koch M, Stronge VS, Shepherd D, Gadola SD, Mathew B, Ritter G, et al. The crystal structure of human CD1d with and without alpha-galactosylceramide. Nature immunology. 2005;6(8):819-26. 374. Zajonc DM, Cantu C, 3rd, Mattner J, Zhou D, Savage PB, Bendelac A, et al. Structure and function of a potent agonist for the semi-invariant natural killer T cell receptor. Nature immunology. 2005;6(8):810-8. 375. Borg NA, Wun KS, Kjer-Nielsen L, Wilce MC, Pellicci DG, Koh R, et al. CD1d-lipid-antigen recognition by the semi-invariant NKT T-cell receptor. Nature. 2007;448(7149):44-9. 376. Scott-Browne JP, Matsuda JL, Mallevaey T, White J, Borg NA, McCluskey J, et al. Germline-encoded recognition of diverse glycolipids by natural killer T cells. Nature immunology. 2007;8(10):1105-13. 377. Li Y, Girardi E, Wang J, Yu ED, Painter GF, Kronenberg M, et al. The Valpha14 invariant natural killer T cell TCR forces microbial glycolipids and CD1d into a conserved binding mode. The Journal of experimental medicine. 2010;207(11):2383-93. 378. Girardi E, Yu ED, Li Y, Tarumoto N, Pei B, Wang J, et al. Unique interplay between sugar and lipid in determining the antigenic potency of bacterial antigens for NKT cells. PLoS biology. 2011;9(11):e1001189. 379. Pellicci DG, Clarke AJ, Patel O, Mallevaey T, Beddoe T, Le Nours J, et al. Recognition of beta-linked self glycolipids mediated by natural killer T cell antigen receptors. Nature immunology. 2011;12(9):827-33. 380. Yu ED, Girardi E, Wang J, Zajonc DM. Cutting edge: structural basis for the recognition of beta-linked glycolipid antigens by invariant NKT cells. Journal of immunology. 2011;187(5):2079-83.

185 381. Brigl M, Bry L, Kent SC, Gumperz JE, Brenner MB. Mechanism of CD1d- restricted natural killer T cell activation during microbial infection. Nature immunology. 2003;4(12):1230-7. 382. Nagarajan NA, Kronenberg M. Invariant NKT cells amplify the innate immune response to lipopolysaccharide. Journal of immunology. 2007;178(5):2706-13. 383. Tyznik AJ, Tupin E, Nagarajan NA, Her MJ, Benedict CA, Kronenberg M. Cutting edge: the mechanism of invariant NKT cell responses to viral danger signals. Journal of immunology. 2008;181(7):4452-6. 384. Wesley JD, Tessmer MS, Chaukos D, Brossay L. NK cell-like behavior of V alpha 14i NK T cells during MCMV infection. Plos Pathog. 2008;4(7). 385. Brigl M, Tatituri RV, Watts GF, Bhowruth V, Leadbetter EA, Barton N, et al. Innate and cytokine-driven signals, rather than microbial antigens, dominate in natural killer T cell activation during microbial infection. The Journal of experimental medicine. 2011;208(6):1163-77. 386. King IL, Amiel E, Tighe M, Mohrs K, Veerapen N, Besra G, et al. The mechanism of splenic invariant NKT cell activation dictates localization in vivo. Journal of immunology. 2013;191(2):572-82. 387. Barral P, Sanchez-Nino MD, van Rooijen N, Cerundolo V, Batista FD. The location of splenic NKT cells favours their rapid activation by blood-borne antigen. The EMBO journal. 2012;31(10):2378-90. 388. Lee YJ, Wang H, Starrett GJ, Phuong V, Jameson SC, Hogquist KA. Tissue- Specific Distribution of iNKT Cells Impacts Their Cytokine Response. Immunity. 2015;43(3):566-78. 389. Brossay L, Jullien D, Cardell S, Sydora BC, Burdin N, Modlin RL, et al. Mouse CD1 is mainly expressed on hemopoietic-derived cells. Journal of immunology. 1997;159(3):1216-24. 390. Roark JH, Park SH, Jayawardena J, Kavita U, Shannon M, Bendelac A. CD1.1 expression by mouse antigen-presenting cells and marginal zone B cells. Journal of immunology. 1998;160(7):3121-7. 391. Bialecki E, Paget C, Fontaine J, Capron M, Trottein F, Faveeuw C. Role of marginal zone B lymphocytes in invariant NKT cell activation. Journal of immunology. 2009;182(10):6105-13. 392. Barral P, Eckl-Dorna J, Harwood NE, De Santo C, Salio M, Illarionov P, et al. B cell receptor-mediated uptake of CD1d-restricted antigen augments antibody responses by recruiting invariant NKT cell help in vivo. Proceedings of the National Academy of Sciences of the United States of America. 2008;105(24):8345-50. 393. Leadbetter EA, Brigl M, Illarionov P, Cohen N, Luteran MC, Pillai S, et al. NK T cells provide lipid antigen-specific cognate help for B cells. Proceedings of the National Academy of Sciences of the United States of America. 2008;105(24):8339-44. 394. Shin JH, Park SH. B Cells Promote Th1- Skewed NKT Cell Response by CD1d- TCR Interaction. Immune network. 2013;13(5):218-21. 395. Fujii S, Shimizu K, Kronenberg M, Steinman RM. Prolonged IFN-gamma- producing NKT response induced with alpha-galactosylceramide-loaded DCs. Nature immunology. 2002;3(9):867-74. 396. Bezbradica JS, Stanic AK, Matsuki N, Bour-Jordan H, Bluestone JA, Thomas JW, et al. Distinct roles of dendritic cells and B cells in Va14Ja18 natural T cell activation in vivo. Journal of immunology. 2005;174(8):4696-705. 397. Arora P, Baena A, Yu KO, Saini NK, Kharkwal SS, Goldberg MF, et al. A single subset of dendritic cells controls the cytokine bias of natural killer T cell responses to diverse glycolipid antigens. Immunity. 2014;40(1):105-16.

186 398. Galli G, Nuti S, Tavarini S, Galli-Stampino L, De Lalla C, Casorati G, et al. CD1d-restricted help to B cells by human invariant natural killer T lymphocytes. The Journal of experimental medicine. 2003;197(8):1051-7. 399. Tonti E, Fedeli M, Napolitano A, Iannacone M, von Andrian UH, Guidotti LG, et al. Follicular helper NKT cells induce limited B cell responses and germinal center formation in the absence of CD4(+) T cell help. Journal of immunology. 2012;188(7):3217-22. 400. Lang GA, Devera TS, Lang ML. Requirement for CD1d expression by B cells to stimulate NKT cell-enhanced antibody production. Blood. 2008;111(4):2158-62. 401. King IL, Fortier A, Tighe M, Dibble J, Watts GFM, Veerapen N, et al. Invariant natural killer T cells direct B cell responses to cognate lipid antigen in an IL-21- dependent manner. Nature immunology. 2012;13(1):44-U64. 402. Shah HB, Joshi SK, Rampuria P, Devera TS, Lang GA, Stohl W, et al. BAFF- and APRIL-dependent maintenance of antibody titers after immunization with T- dependent antigen and CD1d-binding ligand. Journal of immunology. 2013;191(3):1154-63. 403. Liu K, Idoyaga J, Charalambous A, Fujii S, Bonito A, Mordoh J, et al. Innate NKT lymphocytes confer superior adaptive immunity via tumor-capturing dendritic cells. The Journal of experimental medicine. 2005;202(11):1507-16. 404. Vomhof-DeKrey EE, Yates J, Leadbetter EA. Invariant NKT cells provide innate and adaptive help for B cells. Current opinion in immunology. 2014;28:12-7. 405. Hagglof T, Sedimbi SK, Yates JL, Parsa R, Salas BH, Harris RA, et al. Neutrophils license iNKT cells to regulate self-reactive mouse B cell responses. Nature immunology. 2016;17(12):1407-14. 406. Gombert JM, Herbelin A, Tancrede-Bohin E, Dy M, Carnaud C, Bach JF. Early quantitative and functional deficiency of NK1+-like thymocytes in the NOD mouse. European journal of immunology. 1996;26(12):2989-98. 407. Naumov YN, Bahjat KS, Gausling R, Abraham R, Exley MA, Koezuka Y, et al. Activation of CD1d-restricted T cells protects NOD mice from developing diabetes by regulating dendritic cell subsets. Proceedings of the National Academy of Sciences of the United States of America. 2001;98(24):13838-43. 408. Kojo S, Adachi Y, Keino H, Taniguchi M, Sumida T. Dysfunction of T cell receptor AV24AJ18+, BV11+ double-negative regulatory natural killer T cells in autoimmune diseases. Arthritis and rheumatism. 2001;44(5):1127-38. 409. van der Vliet HJ, von Blomberg BM, Nishi N, Reijm M, Voskuyl AE, van Bodegraven AA, et al. Circulating V(alpha24+) Vbeta11+ NKT cell numbers are decreased in a wide variety of diseases that are characterized by autoreactive tissue damage. Clinical immunology. 2001;100(2):144-8. 410. Yanagihara Y, Shiozawa K, Takai M, Kyogoku M, Shiozawa S. Natural killer (NK) T cells are significantly decreased in the peripheral blood of patients with rheumatoid arthritis (RA). Clinical and experimental immunology. 1999;118(1):131-6. 411. Illes Z, Kondo T, Newcombe J, Oka N, Tabira T, Yamamura T. Differential expression of NK T cell V alpha 24J alpha Q invariant TCR chain in the lesions of multiple sclerosis and chronic inflammatory demyelinating polyneuropathy. Journal of immunology. 2000;164(8):4375-81. 412. Sireci G, Russo D, Dieli F, Porcelli SA, Taniguchi M, La Manna MP, et al. Immunoregulatory role of Jalpha281 T cells in aged mice developing lupus-like nephritis. European journal of immunology. 2007;37(2):425-33. 413. Wang B, Geng YB, Wang CR. CD1-restricted NK T cells protect nonobese diabetic mice from developing diabetes. The Journal of experimental medicine. 2001;194(3):313-20.

187 414. Shi FD, Flodstrom M, Balasa B, Kim SH, Van Gunst K, Strominger JL, et al. Germ line deletion of the CD1 locus exacerbates diabetes in the NOD mouse. Proceedings of the National Academy of Sciences of the United States of America. 2001;98(12):6777-82. 415. Baxter AG, Kinder SJ, Hammond KJ, Scollay R, Godfrey DI. Association between alphabetaTCR+CD4-CD8- T-cell deficiency and IDDM in NOD/Lt mice. Diabetes. 1997;46(4):572-82. 416. Hong S, Wilson MT, Serizawa I, Wu L, Singh N, Naidenko OV, et al. The natural killer T-cell ligand alpha-galactosylceramide prevents autoimmune diabetes in non-obese diabetic mice. Nature medicine. 2001;7(9):1052-6. 417. Sharif S, Arreaza GA, Zucker P, Mi QS, Sondhi J, Naidenko OV, et al. Activation of natural killer T cells by alpha-galactosylceramide treatment prevents the onset and recurrence of autoimmune Type 1 diabetes. Nature medicine. 2001;7(9):1057- 62. 418. Mi QS, Ly D, Zucker P, McGarry M, Delovitch TL. Interleukin-4 but not interleukin-10 protects against spontaneous and recurrent type 1 diabetes by activated CD1d-restricted invariant natural killer T-cells. Diabetes. 2004;53(5):1303-10. 419. Mizuno M, Masumura M, Tomi C, Chiba A, Oki S, Yamamura T, et al. Synthetic glycolipid OCH prevents insulitis and diabetes in NOD mice. Journal of autoimmunity. 2004;23(4):293-300. 420. Forestier C, Takaki T, Molano A, Im JS, Baine I, Jerud ES, et al. Improved outcomes in NOD mice treated with a novel Th2 cytokine-biasing NKT cell activator. Journal of immunology. 2007;178(3):1415-25. 421. Wilson SB, Kent SC, Patton KT, Orban T, Jackson RA, Exley M, et al. Extreme Th1 bias of invariant Valpha24JalphaQ T cells in type 1 diabetes. Nature. 1998;391(6663):177-81. 422. Kukreja A, Cost G, Marker J, Zhang C, Sun Z, Lin-Su K, et al. Multiple immuno-regulatory defects in type-1 diabetes. The Journal of clinical investigation. 2002;109(1):131-40. 423. Kis J, Engelmann P, Farkas K, Richman G, Eck S, Lolley J, et al. Reduced CD4+ subset and Th1 bias of the human iNKT cells in Type 1 diabetes mellitus. Journal of leukocyte biology. 2007;81(3):654-62. 424. Lee PT, Putnam A, Benlagha K, Teyton L, Gottlieb PA, Bendelac A. Testing the NKT cell hypothesis of human IDDM pathogenesis. The Journal of clinical investigation. 2002;110(6):793-800. 425. Michalek J, Vrabelova Z, Hrotekova Z, Kyr M, Pejchlova M, Kolouskova S, et al. Immune regulatory T cells in siblings of children suffering from type 1 diabetes mellitus. Scandinavian journal of immunology. 2006;64(5):531-5. 426. Oikawa Y, Shimada A, Yamada S, Motohashi Y, Nakagawa Y, Irie J, et al. High frequency of valpha24(+) vbeta11(+) T-cells observed in type 1 diabetes. Diabetes care. 2002;25(10):1818-23. 427. Roman-Gonzalez A, Moreno ME, Alfaro JM, Uribe F, Latorre-Sierra G, Rugeles MT, et al. Frequency and function of circulating invariant NKT cells in autoimmune diabetes mellitus and thyroid diseases in Colombian patients. Human immunology. 2009;70(4):262-8. 428. Berzins SP, Kyparissoudis K, Pellicci DG, Hammond KJ, Sidobre S, Baxter A, et al. Systemic NKT cell deficiency in NOD mice is not detected in peripheral blood: implications for human studies. Immunology and cell biology. 2004;82(3):247-52. 429. Kent SC, Chen Y, Clemmings SM, Viglietta V, Kenyon NS, Ricordi C, et al. Loss of IL-4 secretion from human type 1a diabetic pancreatic draining lymph node NKT cells. Journal of immunology. 2005;175(7):4458-64.

188 430. Atkinson MA, Eisenbarth GS. Type 1 diabetes: new perspectives on disease pathogenesis and treatment. Lancet. 2001;358(9277):221-9. 431. Bach JF. -dependent diabetes mellitus as an autoimmune disease. Endocrine reviews. 1994;15(4):516-42. 432. Tisch R, McDevitt H. Insulin-dependent diabetes mellitus. Cell. 1996;85(3):291- 7. 433. Adorini L, Gregori S, Harrison LC. Understanding autoimmune diabetes: insights from mouse models. Trends in molecular medicine. 2002;8(1):31-8. 434. Atkinson MA, Leiter EH. The NOD mouse model of type 1 diabetes: as good as it gets? Nature medicine. 1999;5(6):601-4. 435. Pilstrom B, Bjork L, Bohme J. Demonstration of a TH1 cytokine profile in the late phase of NOD insulitis. Cytokine. 1995;7(8):806-14. 436. von Herrath MG, Oldstone MB. Interferon-gamma is essential for destruction of beta cells and development of insulin-dependent diabetes mellitus. The Journal of experimental medicine. 1997;185(3):531-9. 437. Bendelac A, Hunziker RD, Lantz O. Increased interleukin 4 and immunoglobulin E production in transgenic mice overexpressing NK1 T cells. The Journal of experimental medicine. 1996;184(4):1285-93. 438. Lehuen A, Lantz O, Beaudoin L, Laloux V, Carnaud C, Bendelac A, et al. Overexpression of natural killer T cells protects Valpha14- Jalpha281 transgenic nonobese diabetic mice against diabetes. The Journal of experimental medicine. 1998;188(10):1831-9. 439. Katz JD, Wang B, Haskins K, Benoist C, Mathis D. Following a diabetogenic T cell from genesis through pathogenesis. Cell. 1993;74(6):1089-100. 440. Beaudoin L, Laloux V, Novak J, Lucas B, Lehuen A. NKT cells inhibit the onset of diabetes by impairing the development of pathogenic T cells specific for pancreatic beta cells. Immunity. 2002;17(6):725-36. 441. Novak J, Beaudoin L, Griseri T, Lehuen A. Inhibition of T cell differentiation into effectors by NKT cells requires cell contacts. Journal of immunology. 2005;174(4):1954-61. 442. Cain JA, Smith JA, Ondr JK, Wang B, Katz JD. NKT cells and IFN-gamma establish the regulatory environment for the control of diabetogenic T cells in the nonobese diabetic mouse. Journal of immunology. 2006;176(3):1645-54. 443. Falcone M, Yeung B, Tucker L, Rodriguez E, Sarvetnick N. A defect in interleukin 12-induced activation and interferon gamma secretion of peripheral natural killer T cells in nonobese diabetic mice suggests new pathogenic mechanisms for insulin-dependent diabetes mellitus. The Journal of experimental medicine. 1999;190(7):963-72. 444. Ly D, Mi QS, Hussain S, Delovitch TL. Protection from type 1 diabetes by invariant NK T cells requires the activity of CD4+CD25+ regulatory T cells. Journal of immunology. 2006;177(6):3695-704. 445. Chiba A, Kaieda S, Oki S, Yamamura T, Miyake S. The involvement of V(alpha)14 natural killer T cells in the pathogenesis of arthritis in murine models. Arthritis and rheumatism. 2005;52(6):1941-8. 446. Chiba A, Oki S, Miyamoto K, Hashimoto H, Yamamura T, Miyake S. Suppression of collagen-induced arthritis by natural killer T cell activation with OCH, a sphingosine-truncated analog of alpha-galactosylceramide. Arthritis and rheumatism. 2004;50(1):305-13. 447. Coppieters K, Van Beneden K, Jacques P, Dewint P, Vervloet A, Vander Cruyssen B, et al. A single early activation of invariant NK T cells confers long-term protection against collagen-induced arthritis in a ligand-specific manner. J Immunol. 2007;179(4):2300-9.

189 448. Miellot A, Zhu R, Diem S, Boissier MC, Herbelin A, Bessis N. Activation of invariant NK T cells protects against experimental rheumatoid arthritis by an IL-10- dependent pathway. European journal of immunology. 2005;35(12):3704-13. 449. Teige A, Bockermann R, Hasan M, Olofsson KE, Liu Y, Issazadeh-Navikas S. CD1d-dependent NKT cells play a protective role in acute and chronic arthritis models by ameliorating antigen-specific Th1 responses. Journal of immunology. 2010;185(1):345-56. 450. Zeng D, Liu Y, Sidobre S, Kronenberg M, Strober S. Activation of natural killer T cells in NZB/W mice induces Th1-type immune responses exacerbating lupus. The Journal of clinical investigation. 2003;112(8):1211-22. 451. Zeng D, Lee MK, Tung J, Brendolan A, Strober S. Cutting edge: a role for CD1 in the pathogenesis of lupus in NZB/NZW mice. Journal of immunology. 2000;164(10):5000-4. 452. Morshed SR, Mannoor K, Halder RC, Kawamura H, Bannai M, Sekikawa H, et al. Tissue-specific expansion of NKT and CD5+B cells at the onset of autoimmune disease in (NZBxNZW)F1 mice. European journal of immunology. 2002;32(9):2551- 61. 453. Yang JQ, Chun T, Liu H, Hong S, Bui H, Van Kaer L, et al. CD1d deficiency exacerbates inflammatory dermatitis in MRL-lpr/lpr mice. European journal of immunology. 2004;34(6):1723-32. 454. Mieza MA, Itoh T, Cui JQ, Makino Y, Kawano T, Tsuchida K, et al. Selective reduction of V alpha 14+ NK T cells associated with disease development in autoimmune-prone mice. Journal of immunology. 1996;156(10):4035-40. 455. Takeda K, Dennert G. The development of autoimmunity in C57BL/6 lpr mice correlates with the disappearance of natural killer type 1-positive cells: evidence for their suppressive action on bone marrow stem cell proliferation, B cell immunoglobulin secretion, and autoimmune symptoms. The Journal of experimental medicine. 1993;177(1):155-64. 456. Wermeling F, Lind SM, Jordo ED, Cardell SL, Karlsson MCI. Invariant NKT cells limit activation of autoreactive CD1d-positive B cells. Journal of Experimental Medicine. 2010;207(5):943-52. 457. Mars LT, Laloux V, Goude K, Desbois S, Saoudi A, Van Kaer L, et al. Cutting edge: V alpha 14-J alpha 281 NKT cells naturally regulate experimental autoimmune encephalomyelitis in nonobese diabetic mice. Journal of immunology. 2002;168(12):6007-11. 458. Mars LT, Gautron AS, Novak J, Beaudoin L, Diana J, Liblau RS, et al. Invariant NKT cells regulate experimental autoimmune encephalomyelitis and infiltrate the central nervous system in a CD1d-independent manner. Journal of immunology. 2008;181(4):2321-9. 459. Grajewski RS, Hansen AM, Agarwal RK, Kronenberg M, Sidobre S, Su SB, et al. Activation of invariant NKT cells ameliorates experimental ocular autoimmunity by a mechanism involving innate IFN-gamma production and dampening of the adaptive Th1 and Th17 responses. Journal of immunology. 2008;181(7):4791-7. 460. Sag D, Krause P, Hedrick CC, Kronenberg M, Wingender G. IL-10-producing NKT10 cells are a distinct regulatory invariant NKT cell subset. The Journal of clinical investigation. 2014;124(9):3725-40. 461. Monteiro M, Almeida CF, Caridade M, Ribot JC, Duarte J, Agua-Doce A, et al. Identification of regulatory Foxp3+ invariant NKT cells induced by TGF-beta. Journal of immunology. 2010;185(4):2157-63. 462. Lynch L, Michelet X, Zhang S, Brennan PJ, Moseman A, Lester C, et al. Regulatory iNKT cells lack expression of the transcription factor PLZF and control the

190 homeostasis of T(reg) cells and macrophages in adipose tissue. Nature immunology. 2015;16(1):85-95. 463. van den Berg WB, Joosten LA, van Lent PL. Murine antigen-induced arthritis. Methods in molecular medicine. 2007;136:243-53. 464. Dumonde DC, Glynn LE. The production of arthritis in rabbits by an immunological reaction to fibrin. British journal of experimental pathology. 1962;43:373-83. 465. Cooke TD, Hurd ER, Ziff M, Jasin HE. The pathogenesis of chronic inflammation in experimental antigen-induced arthritis. II. Preferential localization of antigen-antibody complexes to collagenous tissues. The Journal of experimental medicine. 1972;135(2):323-38. 466. van den Berg WB, van Beusekom HJ, van de Putte LB, Zwarts WA, van der Sluis M. Antigen handling in antigen-induced arthritis in mice: an autoradiographic and immunofluorescence study using whole joint sections. The American journal of pathology. 1982;108(1):9-16. 467. van den Berg WB, van de Putte LB, Zwarts WA, Joosten LA. Electrical charge of the antigen determines intraarticular antigen handling and chronicity of arthritis in mice. The Journal of clinical investigation. 1984;74(5):1850-9. 468. van den Berg WB, van de Putte LB. Electrical charge of the antigen determines its localization in the mouse knee joint. Deep penetration of cationic BSA in hyaline articular cartilage. The American journal of pathology. 1985;121(2):224-34. 469. Brackertz D, Mitchell GF, Mackay IR. Antigen-induced arthritis in mice. I. Induction of arthritis in various strains of mice. Arthritis and rheumatism. 1977;20(3):841-50. 470. Brand DD, Latham KA, Rosloniec EF. Collagen-induced arthritis. Nature protocols. 2007;2(5):1269-75. 471. Brackertz D, Mitchell GF, Vadas MA, Mackay IR. Studies on antigen-induced arthritis in mice. III. Cell and serum transfer experiments. Journal of immunology. 1977;118(5):1645-8. 472. Brackertz D, Mitchell GF, Vadas MA, Mackay IR, Miller JF. Studies on antigen-induced arthritis in mice. II. Immunologic correlates of arthritis susceptibility in mice. Journal of immunology. 1977;118(5):1639-44. 473. Petrow PK, Thoss K, Katenkamp D, Brauer R. Adoptive transfer of susceptibility to antigen-induced arthritis into severe combined immunodeficient (SCID) mice: role of CD4+ and CD8+ T cells. Immunological investigations. 1996;25(4):341-53. 474. Lens JW, van den Berg WB, van de Putte LB. Flare-up of antigen-induced arthritis in mice after challenge with intravenous antigen: studies on the characteristics of and mechanisms involved in the reaction. Clinical and experimental immunology. 1984;55(2):287-94. 475. Lens JW, van den Berg WB, van de Putte LB, Berden JH, Lems SP. Flare-up of antigen-induced arthritis in mice after challenge with intravenous antigen: effects of pre-treatment with cobra venom factor and anti-lymphocyte serum. Clinical and experimental immunology. 1984;57(3):520-8. 476. Lens JW, van den Berg WB, van de Putte LB, Zwarts WA. Flare of antigen- induced arthritis in mice after intravenous challenge. Kinetics of antigen in the circulation and localization of antigen in the arthritic and noninflamed joint. Arthritis and rheumatism. 1986;29(5):665-74. 477. van de Loo AA, Arntz OJ, Bakker AC, van Lent PL, Jacobs MJ, van den Berg WB. Role of interleukin 1 in antigen-induced exacerbations of murine arthritis. The American journal of pathology. 1995;146(1):239-49.

191 478. Koenders MI, Lubberts E, Oppers-Walgreen B, van den Bersselaar L, Helsen MM, Di Padova FE, et al. Blocking of interleukin-17 during reactivation of experimental arthritis prevents joint inflammation and bone erosion by decreasing RANKL and interleukin-1. The American journal of pathology. 2005;167(1):141-9. 479. Chen YH, Chiu NM, Mandal M, Wang N, Wang CR. Impaired NK1+ T cell development and early IL-4 production in CD1-deficient mice. Immunity. 1997;6(4):459-67. 480. Kuhn R, Lohler J, Rennick D, Rajewsky K, Muller W. Interleukin-10-Deficient Mice Develop Chronic Enterocolitis. Cell. 1993;75(2):263-74. 481. Jung S, Unutmaz D, Wong P, Sano GI, De los Santos K, Sparwasser T, et al. In vivo depletion of CD11c(+) dendritic cells abrogates priming of CD8(+) T cells by exogenous cell-associated antigens. Immunity. 2002;17(2):211-20. 482. Goold HD, Escors D, Conlan TJ, Chakraverty R, Bennett CL. Conventional Dendritic Cells Are Required for the Activation of Helper-Dependent CD8 T Cell Responses to a Model Antigen After Cutaneous Vaccination with Lentiviral Vectors. Journal of Immunology. 2011;186(8):4565-72. 483. Lu TT, Cyster JG. Integrin-mediated long-term B cell retention in the splenic marginal zone. Science. 2002;297(5580):409-12. 484. Belperron AA, Dailey CM, Booth CJ, Bockenstedt LK. Marginal zone B-cell depletion impairs murine host defense against Borrelia burgdorferi infection. Infect Immun. 2007;75(7):3354-60. 485. Robinson MD, McCarthy DJ, Smyth GK. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. 2010;26(1):139-40. 486. Tarca AL, Draghici S, Khatri P, Hassan SS, Mittal P, Kim JS, et al. A novel signaling pathway impact analysis. Bioinformatics. 2009;25(1):75-82. 487. Kanehisa M, Sato Y, Kawashima M, Furumichi M, Tanabe M. KEGG as a reference resource for gene and protein annotation. Nucleic Acids Res. 2016;44(D1):D457-62. 488. Van Kaer L. alpha-Galactosylceramide therapy for autoimmune diseases: prospects and obstacles. Nature reviews Immunology. 2005;5(1):31-42. 489. Furlan R, Bergami A, Cantarella D, Brambilla E, Taniguchi M, Dellabona P, et al. Activation of invariant NKT cells by alphaGalCer administration protects mice from MOG35-55-induced EAE: critical roles for administration route and IFN-gamma. European journal of immunology. 2003;33(7):1830-8. 490. Parekh VV, Wu L, Olivares-Villagomez D, Wilson KT, Van Kaer L. Activated invariant NKT cells control central nervous system autoimmunity in a mechanism that involves myeloid-derived suppressor cells. Journal of immunology. 2013;190(5):1948- 60. 491. Monteiro M, Almeida CF, Caridade M, Ribot JC, Duarte J, Agua-Doce A, et al. Identification of Regulatory Foxp3(+) Invariant NKT Cells Induced by TGF-beta. Journal of Immunology. 2010;185(4):2157-63. 492. Harada M, Seino K, Wakao H, Sakata S, Ishizuka Y, Ito T, et al. Down- regulation of the invariant Valpha14 antigen receptor in NKT cells upon activation. International immunology. 2004;16(2):241-7. 493. Angeli V, Ginhoux F, Llodra J, Quemeneur L, Frenette PS, Skobe M, et al. B cell-driven lymphangiogenesis in inflamed lymph nodes enhances dendritic cell mobilization. Immunity. 2006;24(2):203-15. 494. Goenka R, Parent MA, Elzer PH, Baldwin CL. B cell-deficient mice display markedly enhanced resistance to the intracellular bacterium Brucella abortus. The Journal of infectious diseases. 2011;203(8):1136-46.

192 495. Homann D, Tishon A, Berger DP, Weigle WO, von Herrath MG, Oldstone MB. Evidence for an underlying CD4 helper and CD8 T-cell defect in B-cell-deficient mice: failure to clear persistent virus infection after adoptive immunotherapy with virus- specific memory cells from muMT/muMT mice. Journal of virology. 1998;72(11):9208-16. 496. Uchida J, Hamaguchi Y, Oliver JA, Ravetch JV, Poe JC, Haas KM, et al. The innate mononuclear phagocyte network depletes B lymphocytes through Fc receptor- dependent mechanisms during anti-CD20 antibody immunotherapy. The Journal of experimental medicine. 2004;199(12):1659-69. 497. Uchida J, Lee Y, Hasegawa M, Liang Y, Bradney A, Oliver JA, et al. Mouse CD20 expression and function. International immunology. 2004;16(1):119-29. 498. Hamaguchi Y, Uchida J, Cain DW, Venturi GM, Poe JC, Haas KM, et al. The peritoneal cavity provides a protective niche for B1 and conventional B lymphocytes during anti-CD20 immunotherapy in mice. Journal of immunology. 2005;174(7):4389- 99. 499. Im JS, Arora P, Bricard G, Molano A, Venkataswamy MM, Baine I, et al. Kinetics and cellular site of glycolipid loading control the outcome of natural killer T cell activation. Immunity. 2009;30(6):888-98. 500. Constantinides MG, Bendelac A. Transcriptional regulation of the NKT cell lineage. Current opinion in immunology. 2013;25(2):161-7. 501. Wingender G, Krebs P, Beutler B, Kronenberg M. Antigen-specific cytotoxicity by invariant NKT cells in vivo is CD95/CD178-dependent and is correlated with antigenic potency. Journal of immunology. 2010;185(5):2721-9. 502. Pollizzi KN, Powell JD. Integrating canonical and metabolic signalling programmes in the regulation of T cell responses. Nature Reviews Immunology. 2014;14(7):435-46. 503. Man K, Kallies A. Synchronizing transcriptional control of T cell metabolism and function. Nature Reviews Immunology. 2015;15(9):574-84. 504. Lee YJ, Holzapfel KL, Zhu J, Jameson SC, Hogquist KA. Steady-state production of IL-4 modulates immunity in mouse strains and is determined by lineage diversity of iNKT cells. Nat Immunol. 2013;14(11):1146-54. 505. Engel I, Seumois G, Chavez L, Samaniego-Castruita D, White B, Chawla A, et al. Innate-like functions of natural killer T cell subsets result from highly divergent gene programs. Nature immunology. 2016;17(6):728-39. 506. Tsukumo S, Unno M, Muto A, Takeuchi A, Kometani K, Kurosaki T, et al. Bach2 maintains T cells in a naive state by suppressing effector memory-related genes. Proc Natl Acad Sci U S A. 2013;110(26):10735-40. 507. Williams AS, Richards PJ, Thomas E, Carty S, Nowell MA, Goodfellow RM, et al. Interferon-gamma protects against the development of structural damage in experimental arthritis by regulating polymorphonuclear neutrophil influx into diseased joints. Arthritis Rheum. 2007;56(7):2244-54. 508. Vomhof-DeKrey EE, Yates J, Hagglof T, Lanthier P, Amiel E, Veerapen N, et al. Cognate interaction with iNKT cells expands IL-10-producing B regulatory cells. Proc Natl Acad Sci U S A. 2015;112(40):12474-9. 509. Rosser EC, Mauri C. Regulatory B Cells: Origin, Phenotype, and Function. Immunity. 2015;42(4):607-12. 510. Galli G, Pittoni P, Tonti E, Malzone C, Uematsu Y, Tortoli M, et al. Invariant NKT cells sustain specific B cell responses and memory. Proceedings of the National Academy of Sciences of the United States of America. 2007;104(10):3984-9. 511. Chung BK, Tsai K, Allan LL, Zheng DJ, Nie JC, Biggs CM, et al. Innate immune control of EBV-infected B cells by invariant natural killer T cells. Blood. 2013;122(15):2600-8.

193 512. Tan AHM, Chong WPK, Ng SW, Basri N, Xu SL, Lam KP. Aberrant Presentation of Self-Lipids by Autoimmune B Cells Depletes Peripheral iNKT Cells. Cell Reports. 2014;9(1):24-31. 513. Shen P, Roch T, Lampropoulou V, O'Connor RA, Stervbo U, Hilgenberg E, et al. IL-35-producing B cells are critical regulators of immunity during autoimmune and infectious diseases. Nature. 2014;507(7492):366-70. 514. Ray A, Wang L, Dittel BN. IL-10-independent regulatory B-cell subsets and mechanisms of action. Int Immunol. 2015;27(10):531-6. 515. Horikoshi M, Goto D, Segawa S, Yoshiga Y, Iwanami K, Inoue A, et al. Activation of Invariant NKT cells with glycolipid ligand alpha-galactosylceramide ameliorates glucose-6-phosphate isomerase peptide-induced arthritis. PLoS One. 2012;7(12):e51215.

194