BACKGROUND Inducible costimulator (ICOS) and its ligand (ICOSL) are a pair of costimulatory molecules that co-localize in germinal centers (GC). This interaction is critical for the maturation of GC B cells to affinity-matured memory B cells and long-lived plasma cells. Both ICOS and ICOSL are implicated in systemic lupus erythematosus (SLE). It is known that ICOSL sheds from the cell membrane and that the soluble form of ICOSL (sICOSL) is elevated in SLE; though the function of sICOSL is poorly understood. While it is known that binding of ICOSL on antigen-presenting cells (APC) to ICOS on T cells leads to cell signaling resulting in T cell activation and differentiation, there is also some preliminary evidence that reverse signaling may also occur through ICOSL in APCs. The binding interaction between ICOS and sICOSL has not been fully characterized and is important to understand if either molecule is to be targeted therapeutically. The hypothesis evaluated in this study was that the ICOS: ICOSL interaction is a potent and critical mediator of proinflammatory signaling and immune activation that functions both via activated T cell-mediated forward signaling and APC-mediated reverse signaling mechanisms and that ectodomain shedding of ICOSL is a protective mechanism that leads to down-regulation of the proinflammatorysignaling cascade initiated by this interaction. The aim of this thesis is to characterize the binding interaction between ICOS and ICOSL and to provide a review of the literature and discuss future work that would enhance the biological understanding of this interaction and its role in lupus and other autoimmune diseases. METHODS The binding interaction between ICOS and ICOSL was characterized using both soluble proteins and cells with expressed recombinant proteins. Purified soluble ICOSL (sICOSL) was characterized using size-exclusion chromatography multiangle light scattering (SEC-MALS). Surface plasmon resonance (SPR) was used to measure the binding affinity between sICOSL and human ICOS fused to the fragment crystallizable (Fc) portion of an immunoglobulin molecule (hICOS.Fc). The binding interaction was further characterized to account for avidity between hICOS.Fc and sICOSL and between hICOS.Fc and ICOSL expressed recombinant on the cell surface using a solution-based binding method. RESULTS Expressed recombinant and purified sICOSL dimerized over time and with increasing temperatures. The sICOSL: hICOS.Fc interaction did not follow a typical 1:1 binding interaction. In-solution binding experiments resulted in a tighter equilibrium dissociation binding constant (KD) than the surface-based results obtained by SPR. The KD for hICOS.Fc binding to human ICOSL(hICOSL) expressed on cells agreed well with the KD for hICOS.Fc to the soluble protein, indicating that the in-solution binding measurement may measure binding avidity rather than affinity and that this may be the more physiologically relevant interaction. CONCLUSIONS I show in the experimental part of this study that the interaction between ICOS and ICOSL is quite potent and that much of the binding strength is due to avidity, or the combined strength of multiple parts of the molecules interacting with one another, rather than the affinity alone. As this interaction is implicated in SLE pathogenesis, it would be useful to develop a clearer understanding of the most relevant physiological form of these molecules (soluble or transmembrane) and of the biological signaling events that are initiated via this interaction in order to determine whether targeting ICOS or ICOSL may be therapeutically viable approaches.

The proper regulation of inducible costimulator (ICOS) and its ligand (ICOSL) have been shown to be essential for maintaining immune homeostasis. Loss of either protein results in defective humoral immunity, and overexpression of ICOS results in aberrant antibody production resembling lupus. How ICOSL is regulated in response to ICOS interaction is still unclear. We demonstrate that ADAM10 is the primary physiological sheddase of ICOSL in both mouse and human. Using an in vivo system in which ADAM10 is deleted only on B cells (ADAM10B-/-), elevated levels of ICOSL were seen. This increase is also seen when ADAM10 is deleted from human B cell lines. Identification of the primary sheddase has allowed the characterization of a novel mechanism of ICOS regulation. In wildtype (WT) mice, interaction of ICOSL/ICOS results in ADAM10 induced shedding of ICOSL on B cells and moderate ICOS internalization on T cells. When this shedding is blocked, excessive ICOS internalization occurs. This results in severe defects in T follicular helper (TFH) development and Th2 polarization, seen in a house dust mite exposure model. In addition, enhanced Th1 and Th1 immune responses are seen in experimental allergic encephalomyelitis. Blockade of ICOSL rescues T cell ICOS surface expression and at least partially rescues both TFH numbers and the abnormal antibody production previously reported in these mice. Overall, we propose a novel regulation of the ICOS:ICOSL axis, with ADAM10 playing a direct role in regulating ICOSL as well as indirectly regulating ICOS, thus controlling ICOS:ICOSL-dependent responses. Additionally, we report a specific role for the metalloprotease ADAM10 on B cells in regulating both ICOSL and ICOS in a mouse model of increased humoral immunity using mir146a-/- mice and a model of lymphoproliferative disease using the well characterized lpr model. B6lpr mice lacking ADAM10 on B cells (A10Blpr) have decreased nodal proliferation and T cell accumulation compared to control B6lpr mice. Additionally, A10Blpr mice have a drastic reduction in autoimmune anti-dsDNA antibody production. In line with this, we found a significant reduction in follicular helper T cells (TFH) and germinal center (GC) B cells in these mice. We also show that lymphoproliferation in this model is closely tied to elevated ICOS levels and decreased ICOSL levels. Overall, our data not only shows a role of B cell ADAM10 in controlling autoimmunity, but also increases our understanding of the regulation of ICOS and ICOSL in the context of autoimmunity. Additionally, we found that ADAM17 is important for marginal zone (MZ) B cell development as well as responses to T-independent type 2 (TI2) immunizations. Mice which lack ADAM17 on B cells (A17B) have decreased MZ B cell numbers but have increased levels of antigen specific antibodies in response to TI2 Immunizations. ADAM17 also regulates the level of several surface molecules on plasma cells and MZ B cells necessary for their function and survival. We also show a role for ADAM17 in ILC2 responsiveness to IL-33. In vivo, mice that lack ADAM17 specifically on ILC2s (ADAM17ILC2-/-) exhibit decreased ILC2 expansion in response to intranasal IL-33 as well as Nippostrongylus brasiliensis (Nb) infection. However, ADAM17ILC2-/- mice have normal ILC2 numbers in a naïve state, suggesting this defect in ILC2 function is limited to cell activation. In vitro, ADAM17 inhibited ILC2s have an increased level of apoptosis and less IL-13 production in response to IL-33 compared to vehicle treated ILC2s. The defect in cytokine production following ADAM17 inhibition is not observed in response to IL-25 stimulation, suggesting this defect is limited to IL-33 stimulation Mechanistically, ADAM17 inhibition in ILC2s specifically causes a defect in IL-33 mediated ERK activation, potentially explaining the defective survival and IL-13 production following ADAM17 inhibition in these cells. Additionally, ADAM17 regulates the level of surface IL1R2 which may affect IL-33 signaling in ILC2s.

Le modèle murin le plus utilisé pour le diabète de type 1 est la souris NOD. L’activation des lymphocytes T autoréactifs vis à vis des cellules béta nécessite la reconnaissance par le TCR de l’auto antigène présenté par le CMH ainsi que des signaux de co stimulation. Nous apportons la preuve que la voie de costimulation ICOS/ICOSL est indispensable au développement du diabète chez la souris NOD. En effet, les souris invalidées pour le gène Icos ou IcosL sont protégées du diabète. Nous avons démontré que cette protection est liée à un défaut d’activation des LT diabétogènes. De façon inattendue, nous avons observé chez ces souris ICOS-/- et ICOSL-/- une neuromyopathie. Cette pathologie se développe parallèlement au diabète chez la souris ICOSL+/+. Sur le plan histologique, le muscle strié périphérique et le nerf périphérique est envahi par un infiltrat lymphocytaire et par des cellules présentatrices d’antigène. Nous avons démontré par des expériences de transfert adoptif que la neuromyopathie est une maladie auto-immune données, nous avons étudié les souris NOD ICOSL-/- CIITA-/-. Ces souris sont dépourvues de lymphocytes T -CD4+ et ne développent pas de neuromyopathie ni de diabète. De même, nous avons étudié les souris NOD ICOSL-/- béta2m-/-. Ces souris sont dépourvues de lymphocytes T-CD8+ et développent une neuromyopathie. Cette déviation de l’auto-immunité est liée à l’interaction entre les LT et les lymphocytes B via le signal ICOS/ICOSL. Nous avons prouvé via des expériences de transfert et de chimères que l’absence de signal ICOS/ICOSL entre les lymphocytes T et les lymphocytes B oriente l’auto-immunité vers le système nerveux périphérique et le muscle strié. Enfin, l’analyse du spectre de spécificité des anticorps présent chez la souris ICOSL-/- par western blot puis par spectrométrie de masse a précisé les cibles antigéniques de la myopathie. L’invalidation de la voie ICOS/ICOSL conduit donc à une déviation de l’auto-immunité du pancréas vers le muscle et le système nerveux périphérique. Ces données prouvent que la voie ICOS/ICOSL est indispensable à l’initiation du diabète, mais aussi au contrôle de l’auto-immunité / Costimulation pathways are described as central in T cell activation and the control of autoimmune responses. We previously reported that NOD mice that are deficient for the icosl gene are protected from diabetes, but instead develop a spontaneous autoimmune neuromyopathy. The general phenotype of the neuromyopathy observed in ICOSL-/- NOD mice is globally similar to that observed in ICOS-/- and ICOS-/-ICOSL-/- double knockout NOD mice. The neuromyopathy is observed in 100% of female mice by the age of 35 weeks. The neuropathy remains limited to the peripheral nerve tissue. The disease is characterized by an infiltration of immune cells: CD4+ T cells, CD8+ T cells, dendritic cells and B lymphocytes, but does not extend to the central nervous system. A similar infiltrate is seen in muscles. Autoimmune neuromyopathy can be transfer to naive recipients by T lymphocytes. Transfer is achieved in NOD.scid recipient mice by CD4+ T-cells, although not by CD8+ T-cells, isolated from 35 week old ICOSL-/- NOD. The predominant role of CD4+T-cells is further demonstrated in this model by the observation that CIITA-/-ICOSL-/- NOD mice do not developed the neuromyopathy. By contrast, ȕ2m-/-ICOSL-/- NOD mice develop a neuromyopathy. We obtained evidence (in chimeric mice) that the interaction between antigen-presenting cells (APC) and T lymphocytes via ICOS/ICOSL is a prerequisite to the development of diabetes, while the loss of the interaction between T lymphocytes and APC play a key role in the development of nervous and muscular autoimmunity. Finally, the spectrum analysis of antibodies specificity in mouse ICOSL-/- with Western blot and mass spectrometry indicated the antigenic targets of myopathy. Altogether, our data indicate that the deviation of autoimmunity in NOD mice from the pancreas to muscles and the peripheral nervous system in the absence of ICOS/ICOSL signal is dependent on the loss of the physiological interaction between T cells and APC

Autoimmune Polyendocrine Syndrome Type 1(APS I) is a monogenic autosomal recessive autoimmune disorder which is the result of mutations in the autoimmune regulator (AIRE) gene. Symptoms of the disease include circulation of multiple organ specific autoantibodies, which leads to the breakdown of several tissues, including the adrenal cortex and the parathyroid glands. The patients also develop a number of non-endocrine disorders. This study has investigated the peripheral tolerance mechanisms controlled by the AIRE gene in Aire deficient mice, an animal model of the disease. The B cell Activating Factor (BAFF), which is a cytokine involved in B cell survival and growth, is elevated in Aire-/- mice, resulting in an increased release of autoantibodies and B cell proliferation. Therefore the BAFF level differences between TCR-/- and B6 mice was studied, and the results showed significantly higher levels of BAFF in TCR-/- mice. This is not in accordance with earlier studies. ICOS and ICOSL are involved in the activation of follicular T helper cells. The expression of ICOSL on different subpopulations of DC from mice was studied to evaluate the possible influence of AIRE expression on the T cells in the spleen. The results showed that ICOSL is significantly higher expressed in peripheral 33D1+ DCs in Aire-/- mice, showing that AIRE has a role in the over-activation of the follicular T helper cells, which can lead to autoantibody production and inflammation. These results show that AIRE is involved in peripheral tolerance.

Les myopathies inflammatoires (MI) représentent un groupe hétérogène de maladies caractérisépar une faiblesse musculaire chronique et symétrique associée à une augmentation du taux sérique decréatine phosphokinase (CPK). Les MI sont actuellement subdivisées en 5 entitées : les dermatomyosites,les myopathies nécrosantes auto-immunes, la myosite à inclusion, la polymyosite et les myosites dechevauchement. A ce jour, le diagnostic des MI repose sur l’association de signes cliniques, decaractéristiques anatomopathologiques sur la biopsie musculaire et la présence d’auto-anticorps (aAc). Eneffet, la découverte d’aAc spécifiques et/ou associés aux myosites (MSA/MAA) a considérablementamélioré le diagnostic et le pronostic de la maladie. Cependant, un nombre non négligeable de patientsatteints de MI sont séronégatifs pour les MSA/MAA connus. Par ailleurs, la biopsie musculaire nécessaireau diagnostic est parfois guidée par imagerie par résonance magnétique (IRM), bien qu’il n’ait pas étéprouvé que les données d’imagerie soient corrélées aux signes histologiques. Enfin, le traitement des MIrepose sur l’utilisation d’immunosuppresseurs systémiques, une approche non spécifique de laphysiopathologie de la maladie. Les modèles animaux de MI les plus utilisés sont induits et nonspontanés : ils reposent principalement sur l’immunisation d’animaux contre des protéines telles que lamyosine, la protéine C ou l’histidyl-tRNA synthétase.Les souris NOD (non obese diabetic) sont le modèle classique d’étude du diabète de type 1.Lorsque ces souris sont invalidées pour la voie de costimulation lymphocytaire ICOS/ICOSL, les souris nedéveloppent plus de diabète mais présentent alors une atteinte musculaire. Dans ce travail, nous avonsétudié le phénotype et caractérisé l’atteinte musculaire des souris NOD Icos-/- et NOD Icosl-/-. Nous avonsainsi établi le 1er modèle murin spontané de MI, dont la physiopathologie est médiée par leslymphocytes T CD4+ et la sécrétion d’IFN-γ. Par ailleurs, ces souris présentent un déficit en lymphocytes Trégulateurs. Nous avons également identifié 4 auto-antigènes (aAg) candidats cibles d’aAc chez ces souris.La recherche des aAc correspondants aux aAg orthologues dans le sérum des patients atteints de MI apermis d’identifier, pour l’un d’entre eux, une minorité d’individus séropositifs grâce au développementd’un nouveau test ALBIA (addressable laser bead immunoassay). Il pourrait donc s’agir d’un nouveaubiomarqueur. Dans la perspective de nouvelles évaluations thérapeutiques, nous avons établi desdonnées préliminaires montrant que l’interleukine 2 à faibles doses permet de retarder l’apparition de lamaladie. Enfin, nous avons mis à profit ce modèle et démontré la corrélation entre les données généréespar IRM et par analyse histologique de l’inflammation, confortant le rôle de cette technique d’imagerie àla fois pour le diagnostic et le suivi des MI. / Inflammatory Myopathies (IM) are a heterogeneous group of diseases characterized bychronic and symmetrical muscle weakness associated to increased creatine phosphokinase (CPK)levels, according to entity concerned. Currently, IM are divided into 5 main entities:dematomyositis, immune-mediated necrotizing myopathies, inclusion body myositis, polymyositisand overlap myositis. Nowadays, IM diagnosis is based on clinical signs associated to pathologicfeatures on muscle biopsy and presence of auto-antibodies (aAb). Indeed, the discovery of myositisspecific and/or associated auto-antibodies (MSA/MAA) had considerably improve disease diagnosisand prognosis. However, substantial proportion of IM patients do not display any knownMSA/MAA. Furthermore, diagnosis requires muscle biopsy. This biopsy is sometimes guided bymagnetic resonance imaging (MRI), even though correlation between MRI findings and pathologicalfeatures is not established. Lastly, therapeutics used in IM treatment are systemicimmunosuppressive agents, i.e. not specific to IM pathophysiology. Animal models of IM are mainlybased on active immunization against different proteins as myosin, C protein orhistidyl-tRNA synthetase, while spontaneous models are required to identify pathophysiologicalmechanisms that new therapeutics should target.NOD (non obese diabetic) mice are the main model of type 1 diabetes. When invalidatedfor ICOS/ICOSL costimulation pathway, these mice do not develop diabetes but present musculardisorders. In this work, we study Icos-/- and Icosl-/- NOD mice phenotype and characterize theremuscle lesion. Thus, we have established this model as the first paradigm of IM. Pathophysiologicalstudy in these mice demonstrated that disease is CD4+ T cell dependent and associated to IFN-γproduction. Furthermore, we shown a quantitative defect in regulatory T cells. We have alsoidentified 4 candidate autoantigens (aAg) in Icos-/- and Icosl-/- NOD mice. Searching forcorresponding aAb against ortholog proteins in patients with IM, we identified for one of them, alow percentage of seropositive individuals using a new ALBIA (addressable laser beadimmunoassay). It could be identified as a new biomarker. In order to evaluate new therapies, weestablished preliminary data showing that low dose interleukin 2 therapy allow to delay diseaseonset. Lastly, we took advantage of this new model to demonstrate the correlation betweenMRI findings and histological inflammation features, confirming the valuable role of MRI for thediagnosis and monitoring of IM.

Le modèle murin le plus utilisé pour le diabète de type 1 est la souris NOD. L'activation des lymphocytes T autoréactifs vis à vis des cellules béta nécessite la reconnaissance par le TCR de l'auto antigène présenté par le CMH ainsi que des signaux de co stimulation. Nous apportons la preuve que la voie de costimulation ICOS/ICOSL est indispensable au développement du diabète chez la souris NOD. En effet, les souris invalidées pour le gène Icos ou IcosL sont protégées du diabète. Nous avons démontré que cette protection est liée à un défaut d'activation des LT diabétogènes. De façon inattendue, nous avons observé chez ces souris ICOS-/- et ICOSL-/- une neuromyopathie. Cette pathologie se développe parallèlement au diabète chez la souris ICOSL+/+. Sur le plan histologique, le muscle strié périphérique et le nerf périphérique est envahi par un infiltrat lymphocytaire et par des cellules présentatrices d'antigène. Nous avons démontré par des expériences de transfert adoptif que la neuromyopathie est une maladie auto-immune données, nous avons étudié les souris NOD ICOSL-/- CIITA-/-. Ces souris sont dépourvues de lymphocytes T -CD4+ et ne développent pas de neuromyopathie ni de diabète. De même, nous avons étudié les souris NOD ICOSL-/- béta2m-/-. Ces souris sont dépourvues de lymphocytes T-CD8+ et développent une neuromyopathie. Cette déviation de l'auto-immunité est liée à l'interaction entre les LT et les lymphocytes B via le signal ICOS/ICOSL. Nous avons prouvé via des expériences de transfert et de chimères que l'absence de signal ICOS/ICOSL entre les lymphocytes T et les lymphocytes B oriente l'auto-immunité vers le système nerveux périphérique et le muscle strié. Enfin, l'analyse du spectre de spécificité des anticorps présent chez la souris ICOSL-/- par western blot puis par spectrométrie de masse a précisé les cibles antigéniques de la myopathie. L'invalidation de la voie ICOS/ICOSL conduit donc à une déviation de l'auto-immunité du pancréas vers le muscle et le système nerveux périphérique. Ces données prouvent que la voie ICOS/ICOSL est indispensable à l'initiation du diabète, mais aussi au contrôle de l'auto-immunité