Regulation of Germinal Center Reactions by B and T Cells

Abstract: Break of B cell tolerance to self-antigens results in the development of autoantibodies and, thus, leads to autoimmunity. How B cell tolerance is maintained during active germinal center (GC) reactions is yet to be fully understood. Recent advances revealed several subsets of T cells and B cells that can positively or negatively regulate GC B cell responses in vivo . IL-21-producing CXCR5 + CD4 + T cells comprise a distinct lineage of helper T cells—termed follicular helper T cells (T FH )—that can provide help for the development of GC reactions where somatic hypermutation and affinity maturation take place. Although the function of T FH cells is beneficial in generating high affinity antibodies against infectious agents, aberrant activation of T FH cell or B cell to self-antigens results in autoimmunity. At least three subsets of immune cells have been proposed as regulatory cells that can limit such antibody-mediated autoimmunity, including follicular regulatory T cells (T


Introduction
Negative selection during the development of B cells in the bone marrow and T cells in the thymus leads to the deletion of self-reactive B and T cells.Although some of the self-reactive B cells can escape from negative selection in the bone marrow, they are seldom activated due to the lack of proper help from T cells, since most of self-reactive T cells in the periphery are in an anergic state.These processes-termed central and peripheral tolerance-represent a primary mechanism by which the immune system prevents the development of autoimmunity.During active immune responses in the periphery, antigen-specific T:B cell interaction induces somatic hypermutation of the B cells in the complementarity determining regions.Multiple rounds of somatic hypermutations not only enable the generation of high affinity antibodies, but also diversify the repertoire of B cell receptors.As a result, some of the newly generated B cells possibly acquire B cell receptors that recognize self-antigens.Hence, the somatic hypermutation process is a double-edged sword that may lead to the generation of high affinity antibodies, as well as auto-reactive B cells in the periphery.Importantly, this T: B interaction mainly occurs in a specialized region of B cell area, termed germinal center (GC).Thus, understanding the regulation of GC responses is crucial for vaccine development, as well as for the treatment of antibody-mediated autoimmune diseases.
Recent advances have clearly demonstrated the crucial role of CXCR5 + follicular helper T cells (T FH ) in GC responses.T FH cells promote GC responses by providing developmental and survival signals, as well as factors important for B cells, which eventually become memory B cells and long-lived antibody secreting plasma cells [1].However, unnecessary activation of T FH and B cells against self-antigens may induce autoimmune diseases, such as systemic lupus erythematosus (SLE), rheumatoid arthritis (RA), Sjögren syndrome, and juvenile dermatomyositis [2].
The identity of cells that control T FH and GC B cell responses are not fully understood.Recent studies have simultaneously discovered at least three specialized immune cells that specifically suppress GC responses, namely follicular regulatory T cells (T FR ), Qa-1 restricted CD8 regulatory T cells, and regulatory B cells (B REG ) [3].Despite the differences in their lineages, all three types of regulatory cells express CXCR5, just like T FH cells.Hence, the balance between T FH cells and the regulatory cells likely determine the magnitude and duration of GC B cell responses.Understanding how T FH cells and these regulatory cells control GC B cell responses will pave the way for the development of novel therapeutic targets in vaccine design, as well as for the treatment of antibody-mediated autoimmune diseases.

Follicular Helper T Cells
The help of CD4 + T cells is required for GC formation, Ig class-switching, and antigen specific memory B cell and plasma cell production [4].The concept of T FH cells was first described in humans as CD4 + T cells in secondary lymphoid tissues (tonsil) that express CXCR5 and localize in B cell follicles, especially in germinal centers (GCs) [5][6][7].GCs are histologically specialized structures that develop within B cell follicles of secondary lymphoid tissues where somatic hypermutation, selection of high affinity B cells, class switch recombination, as well as plasma cell and memory B cell differentiation mainly occur [1].CXCR5 expression allows T FH cells to migrate into B cell follicles in response to a CXCL13 gradient [7,8], and thus it serves as a surface marker of this T H subset.In the past, CXCR5 + T H cells were thought to be a subpopulation of Th2 cells, since they were able to express IL-4 [9][10][11].However, recent studies have shown that CXCR5 + T FH cell generation was intact in STAT6 −/− , as well as IL-4 −/− mice [9,12].Moreover, T FH are normal in STAT4 −/− and Rorc −/− mice.Hence, the differentiation of T FH cells is independent of the Th1, Th2 and Th17 lineage programs.More recently, B cell lymphoma 6 (Bcl-6) has been reported as an essential transcriptional repressor for T FH cell differentiation [13][14][15].
Differentiation of GC B cells and T FH cells is likely initiated in the interfollicular (IF) zone, where initially activated CD4 + T cells are further primed by CXCR5 expressing dendritic cells [16,17].Interaction between antigen-specific T cells and B cells also occurs in the IF zone for the first two to three days after immunization, prior to their migration to the follicles to form GCs [16].In the GCs, B cell division is restricted to the dark zone (DZ), while B-T interaction occurs in the light zone (LZ).B cell assistance from T FH cells in the LZ facilitates B cell return to the DZ, which is accompanied by clonal expansion of B cells in the DZ [18].Unlike GC B cells, T FH cells have been shown to continually emigrate into follicles and neighboring GCs [19].Moreover, newly differentiated T FH cells can migrate into preexisting GCs and augment ongoing GC responses [19].These polyclonal colonization and invasive properties of T FH cells might be critical for prolonged GC B cell responses.T FH cells express CXCR5, inducible co-stimulator (ICOS), programmed cell death protein 1 (PD-1), Bcl-6, basic leucin zipper transcription factor ATF-like (BATF), signal transducer and activator of transcription 3 (STAT3), c-Maf and interferon regulatory factor 4 (IRF4), as well as cytokines IL-21, IL-4, and IL-10 [1,20,21].Other surface molecules, such as CD40 ligand (CD40L), BTLA, CD84, and cytoplasmic adaptor protein SLAM-associated protein (SAP), are also expressed in T FH cells (Figure 1) [5,6,13,14,[22][23][24][25][26][27].

Transcription Factors and Molecules Required for the Differentiation of T FH Cells
In the last five years, a number of studies have identified multiple transcription factors that are required for T FH differentiation.As mentioned above, Bcl-6 has been reported as a master transcriptional regulator for T FH lineage commitment [13][14][15].As a transcriptional repressor, expression of Bcl-6 can inhibit Th1, Th2, and Th17 differentiation.Interestingly, some of these transcription factors, such as STAT3, BATF, and IRF4 are also necessary for Th17 cell commitment.In addition, IL-21 is the most important effector cytokine for T FH cells, and it is also expressed in Th17 cells.Therefore, T FH and Th17 cells share many common features including developmental requirements and effector cytokines.

Bcl-6
Bcl-6 was first described in a subset of CD4 + T cells in GCs in human tonsils [28].CD4 + T cells from Bcl-6 deficient mice failed to differentiate into T FH cells, while constitutive expression of Bcl-6 induced T FH cell generation in vivo [13][14][15].Bcl-6 drives T FH cell differentiation by suppressing the differentiation of the other T H subsets.For instance, Bcl-6 suppresses GATA3 expression [29], and can directly bind to Tbx21 (encoding T-bet) and Rorc (encoding RORγt) promoter regions to suppress the transcription of these genes [15].Moreover, Bcl-6 suppresses Blimp-1 and a cluster of microRNAs which suppress T FH generation [15].Blimp-1 (encoded by Prdm1 gene) and its antagonist Bcl-6 are reciprocally expressed in T FH cells and other subsets of T helper cells [13,[30][31][32].Enforced Blimp-1 expression in CD4 + T cells suppresses Bcl-6 expression and generation of T FH cells [13].However, over-expression of Bcl-6 is likely not sufficient to induce IL-21 and CXCR5 expression on T cells [1].The mechanism of Bcl-6 induction during T FH differentiation remains unclear.

BATF
BATF belongs to the activator protein 1 (AP-1) superfamily and is originally known to be required for Th17 differentiation [39].Of note, BATF-deficient mice also show defects in T FH cells and GC responses [40,41].During T FH cell differentiation, BATF directly regulates Bcl-6 and c-Maf expression by binding to the promoter regions of these genes [41].Defective generation of T FH cells in BATF-deficient CD4 + T cells can be restored by simultaneous expression of Bcl-6 and c-Maf [41].

STAT3
STAT3 is the main transducer of IL-6R and IL-21R signals in T and B cells [42,43].These two signaling pathways are important for T FH differentiation and affect Bcl-6 expression in stimulated CD4 T cells [12,14].In T FH cells, STAT3 can bind to the Bcl-6 promoter and induce Bcl-6 expression [44][45][46].On the other hand, STAT3 also induces Blimp-1, an antagonist of Bcl-6 in CD4 + T cells [45].STAT3deficient CD4 + T cells are shown to be defective in T FH cell differentiation [12], while another study showed normal population of CXCR5 + CD4 + T cells in the absence of STAT3 [47,48].In human T cells, STAT3 does not seem to be important for T FH cell differentiation and function [49,50].Thus, it is likely that although STAT3 is generally required for optimal T FH differentiation, T FH cells can be generated in a STAT3-independent fashion.

IRF4
IRF4, like BATF and STAT3, is also known as a Th17 cell transcription factor [49,51,52]. IRF4 can bind to the Il21 promoter and play a crucial role during T FH cell development [45,53].Defect in the T FH population has been reported in IRF4-deficient mice, however, it is not clear whether this was T FH cell specific, or due to a general defect in CD4 + T cells [45].In cooperation with BATF, IRF4 binds to the AP1-IRF composite element and stimulates IL-21 expression during Th17 differentiation [54,55].However, whether the same complex mediates the expression of T FH transcription factors, such as Bcl-6 and c-Maf, remains to be determined.A number of studies have also demonstrated that deregulation of IRF4 is associated with auto-reactive B cell responses.For instance, Pernis and colleagues have identified the IRF4-binding protein (IBP) as a negative regulator of IRF4.Mice deficient in IBP develop spontaneous autoimmune phenotypes associated with enhanced responsiveness of T cells to low levels of stimulation [56,57].Mechanistically, IBP inhibits the binding of IRF4 to the transcriptional regulatory regions of IL-17 and IL-21, and thus, IBP-deficient T cells produce enhanced amounts of IL-17 and IL-21 [57].The increased production of IL-21 in IBP deficient mice results in the increased expression of the Aicda gene (encoding AID) in B cells [58].These results strongly suggest a crucial role for IRF4 in T FH cell differentiation and autoimmune B cell responses.

microRNAs
MicroRNAs of the miR-17~92 family have originally been reported as negative regulators of T FH differentiation [15].On the contrary, two recent studies have demonstrated that the miR-17~92 family is essential for T FH differentiation.For instance, Kang et al. described that mice with T cell-specific deletion of miR-17~92 have a significantly diminished T FH population, while mice with T cell-specific transgenic expression of the same microRNA family developed a fatal immunopathology with spontaneous T FH responses [59].Mechanistically, the miR-17~92 family has been shown to contribute to the migration of T FH cells into B cell follicles by suppressing the expression of phosphatase PHLPP2 [59].In addition, Baumjohann et al. have demonstrated that miRNA-17~92 suppresses the expression of Rora, thus inhibiting the induction of Th17-related genes during T FH cell differentiation [60].Collectively, these studies clearly demonstrate that the miR-17~92 cluster is necessary for T FH lineage specification in vivo.

Cytokines
In addition to its indispensable role in Th17 cell differentiation [61], IL-6 has also been shown to play a role in T FH cell differentiation through its capacity to induce IL-21 by T cells [12,47,62].Both IL-6 and IL-21 signaling occurs through the activation of STAT3 and STAT3-deficient CD4 + T cells show defects in T FH differentiation [12].An initial study showed that IL-6 induces the transcription of Bcl6 and CXCR5 [14], while other studies demonstrated normal development of T FH cells in the absence of IL-6 signaling [47,48,63].In addition, it has recently been described that the increase of T FH cells and GC responses during the late stage of chronic LCMV infection requires IL-6, and that late blockade of IL-6 signaling delays viral clearance [64].In the absence of IL-6 signaling, virus-specific CD4 + T cells express significantly decreased Bcl-6, indicating a crucial contribution of this cytokine to T FH lineage programming during viral infections [64].Another study showed that the absence of either IL-6 or IL-21 alone does not limit the T FH differentiation, while removing both signals substantially abrogates T FH cells.These observations suggest a redundant function for these two cytokines in this process [48], which might be crucial for late maintenance of T FH cells.
IL-2 has been reported to play a suppressive role in T FH differentiation [65][66][67] by inducing STAT5 mediated Blimp-1 [65,67].Moreover, in Th1 polarization conditions, a high concentration of IL-2 inhibits Bcl-6 expression by regulating STATs and Foxo binding to the Bcl-6 promoter.In addition, T-bet interacts with Bcl-6 to form a Tbet-Bcl6 complex, resulting in the inhibition of Bcl-6 dependent gene repression [66].Toxoplasma gondii and viral infections enhance humoral responses by increasing T FH cells in the absence of T-bet [68,69], indicating that the ratio of T-bet and Bcl-6 in T helper cells is a critical determinant of Th1 and T FH cell lineages.

CXCR5
CXCR5 represents a reliable marker of T FH cells and its expression helps them migrate into the B cell follicles in response to CXCL13 (Figure 1) [7,8].During differentiation, T FH cells down-regulate CCR7 to avoid their relocation to the T cell zone [1,25,72].A recent study demonstrated that expression of CXCR5 on DCs was required for optimal T FH and Th2 differentiation in response to H. polygyrus infection [17].Deficiency of CXCR5 in T cells up-regulates Blimp-1 and decreases T FH cell frequency [13,32,73,74].CXCR5-deficient T cells fail to migrate into B cell follicles and thus fail to induce GC B cells [25,72,75].Bcl6-deficient T cells do not express CXCR5 [14,15].Kitano et al. demonstrated that up-regulation of Bcl-6 and CXCR5 on all T cells were initiated two days after immunization, and peaked at day 3 [76].In contrast, Liu and colleagues reported that, while Bcl-6 expression was gradually increased from day 2 to day 7, the expression of CXCR5 was dramatically increased at day 2, and reached a plateau at day 3 and was maintained at a high level in activated T cells [77].Notably, over-expression of Bcl-6 alone minimally increased the expression of CXCR5 on CD4 + T cells [27,77].Future studies will be needed to identify one or more transcription factors that directly trigger the expression of CXCR5.

ICOS
ICOS is up-regulated on CD4 + T cells after CD28 co-stimulation [78,79].ICOS provides an important signal for the survival of effector CD4 + T cells, as well as for the differentiation of T FH cells.ICOSL is constitutively expressed on most of DCs and B cells [78,80,81].ICOS signaling recruits PI3K to its cytoplasmic tail, which in turn triggers the expression of IL-21, IL-4, c-Maf and CXCR5 [31,35,[82][83][84][85][86].ICOS-deficient mice have a reduced number of T FH cells and form immature GCs after immunization [35,87,88].ICOS has recently been shown to directly control follicular recruitment of activated T helper cells by follicular bystander B cells, rather than by antigen presenting DCs, or cognate B cells [89].ICOS engagement induces coordinated pseudopod formation and increases persistent T cell migration at the border between the T-B zone in vitro and in vivo [89].As a result, in the absence of ICOSL on follicular bystander B cells, activated T helper cells cannot develop into T FH cells [89].

CD40L-CD40
CD40 is involved in multiple stages of B cell activation and differentiation.CD40L, which is the only ligand of CD40, is highly expressed in activated CD4 + T cells [1].Deficiency of CD40L or CD40, or patients with mutations in CD40LG, has a decreased number of T FH cells and a failure of GC formation [84,90,91].CD40 is critical for B cell activation, proliferation and survival, and CD40L-CD40 engagement is critical for the maintenance of GC B cells [1,92].It has been reported that CD40L together with IL-21 or IL-4 is required for the maintenance of GC B cells.In the absence of CD40L, GC B cells differentiate into plasma cells [93][94][95][96].Furthermore, the CD40-CD40L interaction is important for migration of T cells into follicles since T FH cells from CD40L −/− mice failed to migrate into B cell follicles, which can be restored by anti-CD40 [97].Therefore, the CD40-CD40L interaction is bidirectional and CD40L signaling to the CD4 + T cells is essential for the migration, priming and maintenance of T FH cell [97][98][99].

SAP:SLAM Family Receptor
SAP plays multiple roles in T FH function.First, SAP is required for the formation of T-B conjugates, which is critical for T FH differentiation [100,101].SAP-deficient CD4 + T cells are unable to form stable conjugates with cognate B cells and fail to differentiate into T FH cells [10,26,[100][101][102][103].Importantly, patients with X-linked lympho-proliferative disease have mutations in SH2D1A (encoding SAP), and they exhibit impaired T FH cell function and poor humoral immunity [104,105].In addition, SAP is also required for the induction of IL-4 by T FH cells, which is independent of Th2, but requires SLAM-SAP-PKCθ signaling [10,106].
Additional molecules in the T-B cell interaction include the SLAM family of receptors: SLAM (CD150), CD84 (SLAMF5), Ly108 (SLAMF6;NTB-A in humans), Ly9 (CD229, SLAMF3) and 2B4 (CD244, Natural Killer cell receptor) [107].These SLAM family receptors can recruit SAP which is a cytoplasmic adaptor molecule and activate signal cascade through PKCθ, BCL-10, NF-κB and FYN [105].CD84 is a SAP-binding SLAM family receptor that is up-regulated on both mouse and human T FH cells and GC B cells [22,101].After primary integrin interaction between B and T cells, CD84 stabilizes the B:T cell conjugates and helps T FH function and GC formation in vivo [101].Ly108 is a SAP-binding SLAM family member and has two distinct isoforms [1].Ly108 has a role in B cell negative selection and T-B cells adhesion [101,108].Deletion of Ly108 in CD4 + T cells reversed the Sh2d1a −/− phenotype by eliminating the SAP requirement for GCs.Its inhibitory function is dependent on immunotyrosine switch motifs (ITSMs) and SHP-1 [109].Recruitment of high levels of SHP-1 at the T:B synapse enables Ly108 to limit T:B cell adhesion [109].Ly9 is constitutively expressed at a high level on CD4 + T cells and there is no defect in T FH differentiation, GCs, and antibody responses in Ly9-deficient mice [110].SLAM is known to be important for Th2-independent IL-4 expression in GC T FH cells, although IL-4 is likely not required for T FH differentiation [10,111].

IL-21
IL-21 has been demonstrated to be essential for T FH cell differentiation and function in vitro and in vivo [112,113].IL-21 production in CD4 + T cell is induced by IL-21 itself, or IL-6, IL-12 and IL-27 [12,62,114,115].IL-21 increases the expression Bcl6 and CXCR5 transcripts in CD4 + T cells in vitro [12,14].IL-21-deficient mice exhibit relatively normal frequency of T FH cells upon immunization, however, the number of T FH cells rapidly declines [112], indicating that IL-21 signaling is essential for the maintenance of differentiated T FH cells.Importantly, IL-21 from T FH cells drives the differentiation of plasma cells from activated B cells in GC [44,[116][117][118][119], and it also promotes the proliferation of GC B cells [112,113].IL-21 is also important for the affinity maturation of immunoglobulins, but it does not affect the formation of memory B cells [112,113].Furthermore, IL-21 signaling triggers immunoglobulin isotype switching to produce IgG3, IgA and IgG1 in human B cells, or IgG1 in murine B cells [120][121][122].

IL-4
Th2 cytokine IL-4 has been thought as a B cell survival and differentiation factor [123].A series of studies have demonstrated that T FH cells, rather than Th2 cells, provide help for B cell differentiation and maturation [1].By using IL-4 reporter mice, Reinhardt et al. showed that the majority of CD4 + T cells secreting IL-4 after Leishmania major infection are T FH cells and that IL-4 production is induced by SAP:SLAM interaction in a Th2-independent manner [10,11,106].IL-4 triggers the Ig isotype switch to IgG1 and IgE [1,124,125].IL-4 signaling in B cells induces Bcl-XL, an anti-apoptotic Bcl-2 family gene, and strongly enhances glucose uptake [125,126].Therefore, IL-4 production by T FH cells mediates the survival and proliferation of GC B cells.

T FH Cells and Autoimmunity
While T FH cells are necessary for humoral immune responses to infectious agents and cancerous cells, excessive T FH responses may lead to the induction of self-reactive B cells [102,127].Indeed, increased T FH responses are tightly associated with systemic autoimmunity, such as SLE, in both humans and mice [127][128][129].Spontaneous generation of GCs and expansion of T FH cells are key features of SLE [2,23,102,[130][131][132][133].

T FH Cells in Animal Models of Antibody-Mediated Autoimmunity
Sanroque mice have a single recessive mutation in the Roquin gene that encodes a RING-type ubiquitin ligase protein that interrupts a repressor of ICOS expression [23].The mutation of sanroque causes excessive numbers of T FH cells with high expression of ICOS and IL-21in T cells that lead to the development of SLE-like pathologies [23].Adoptive transfer of T FH cells from sanroque mice into wild-type mice triggers spontaneous GC formation and production of autoantibodies [134].SAP deficiency or Bcl-6 deletion in the sanroque mice results in the reduction of T FH cells and IL-21 production, and thus alleviates lupus-like symptoms [102,135,136].Of note, deletion of Roquin in hematopoietic cells results in deregulation of immune homeostasis, such as increased GC B cells and effector memory T cells in the secondary lymphoid organs, but does not lead to spontaneous autoimmunity [137].Thus, the exact role of Roquin in the generation of self-reactive T FH cells remains to be determined.BXSB.Yaa mice also exhibit spontaneous GC formation and T FH cells [44,136].These SLE-like autoimmune phenotypes are associated with increased IL-21 production due to the duplicated Tlr7 gene that mediates excessive signaling in response to self-RNA [135].Genetic deletion of the IL-21 receptor in BXSB.Yaa mice significantly decreases the production of autoantibodies, T FH cell numbers and disease severity [136].This result supports the concept that IL-21 is required for auto-reactive B cell differentiation [112,113], and that blockade of this cytokine might be a promising therapeutic approach [138].
The MRL/MpJ-Fas lpr/lpr /J (MRL lpr ) mouse strain is also a widely used as an experimental model of human lupus.Similar to Sanroque and BXSB.Yaa mice, MRL lpr mice exhibit increased IL-21 production.Of note, the production of autoantibodies in these mice is from B cells in the extrafollicular foci, and this process is mediated by extrafollicular helper T (T eFH ) cells [139,140].ICOS signaling is essential for the function of T eFH cells, demonstrated by the fact that ICOS deficiency in MRL lpr mice results in diminished production of autoantibodies [140][141][142].Deficiency in IL-21R in MRL lpr mice leads to a dramatic decrease of T FH , T eFH , GC B cells, plasma cells, and plasmablasts [134].
The NZB/W F1 model is another classical experimental autoimmune model that exhibits a spontaneous lupus-like disease [143].NZB/W F1 mice also show increased T FH and GC B cell responses that appear to be dependent on the ICOS/ICOSL pathway.Accordingly, treatment with anti-ICOSL ameliorates the disease severity by decreasing T FH and GC B cell responses [144].
BXD2 mice also exhibit a lupus-like phenotype.This strain has been established by inbreeding the intercross progeny of C57BL/6K and DBA/2J mice for more than 20 generations [145].BXD2 mice spontaneously produce pathogenic auto-antibodies and develop glomerulonephritis and erosive arthritis [146].CD4 + T cells in BXD2 mice express increased IL-17 levels and Il17-deficient BXD2 mice produce significantly reduced auto-antibodies [133].Therefore, unlike the other animal model of lupus, the lupus-like phenotype of BXD2 mice is known to be IL-17-dependent.Mechanistically, B cells in BXD2 mice exhibit increased expression of IL-17RA compared to C57BL/6 mice and thus quickly activate the canonical NF-κB signaling pathway upon IL-17 [147].Inhibition of NF-κB signaling diminishes IL-17 induced chemotactic arrest of B cells in response to CXCL12 [147].Interestingly, the number of IL-17RA expressing CXCR5 + ICOS + T FH cells is significantly increased in BXD2 mice, and blockade of IL-17 signaling reduces T FH :B cell interaction and the generation of autoantibody producing B cells in BXD2 mice [148].

T FH Cells in Human Autoimmune Diseases
While the role of T FH cells in mouse autoimmunity has been established, the role of T FH cells in humans remains largely unexplored.High levels of class-switched auto-antibodies and abnormal GC B cell populations in patients with autoimmune diseases strongly suggest the involvement of T FH cells in the pathogenesis of autoimmune diseases [139,149].Increased frequencies of circulating T FH -like cells (CXCR5 hi PD-1 hi ICOS hi CD4 + ) are found in patients with SLE, Sjögren's syndrome, RA, and juvenile dermatomyositis.In addition, the frequencies of T FH -like cells are positively correlated with autoantibody titers, as well as disease symptoms, in SLE patients [150][151][152][153].These circulating T FH -like cells share phenotypic and functional properties with GC T FH cells in follicles; but they express low levels of BCL6 and IL-21 [150].While Bcl-6 is gradually down-regulated as the GC response progresses in mouse T FH cells [76], BCL-6 is rapidly re-expressed after TCR stimulation in circulating human CXCR5 hi central memory CD4 + T cells [83].It is possible that the circulating T FH -like cells may act as T FH precursors with the potential for rapid CXCR5-mediated follicular access and B cell helper functions [83,151].Further studies will be required to demonstrate the role of T FH cells in the pathogenesis of antibody-mediated diseases in humans.

Regulatory Cells that Control GC B Cell Responses
While help from T FH cells promotes the generation of high-affinity IgGs and memory B cells, excessive activation of T FH and GC B cells can lead to autoimmunity as evident in a wide range of animal models of autoimmune disorders, as well as in humans.While the existence of suppressor T cells, including Foxp3 + T cells, has been well established over the last two decades, the existence and function of regulatory cell subsets specialized for GC B cell responses and T FH cells have only recently received attention.These include IL-10-producing regulatory B cells (B REG ), CXCR5 + Bcl6 + follicular regulatory cells (T FR ), and Qa-1-restricted CD8 + regulatory T cells (Figure 2).Although their role in autoimmunity and the mode of their suppressive activity are still under investigation, evidence from animal models strongly suggests that these cells are necessary for preventing excessive germinal center reactions in vivo.

Regulatory B Cells
Provision of ICOSL by B cells is required for the differentiation of T FH cells [12].Thus, B cells are generally considered as positive regulator of GC reactions.However, secreted antibodies are known to influence GC responses by limiting the acquisition of antigens by B cells [154].In addition, recent studies have identified a few subsets of B cells, termed regulatory B cells (B REG ), which exhibit immunoregulatory functions [155].

Identification and Development of Regulatory B Cells
Mice with a genetic deficiency of B cells are more susceptible to experimental autoimmune encephalomyelitis (EAE) [156].Subsequent studies revealed that IL-10 production by B cells plays an immunoregulatory role in animal models of inflammatory diseases, including EAE, inflammatory bowel disease and rheumatoid arthritis [157][158][159].This evidence indicates that B cells have regulatory properties.These IL-10 producing B REG cells seem to be heterogeneous, as no appropriate surface marker(s) or master transcription factor(s) has been described to define these cellular subsets [155].
Marginal-zone (MZ) B cells are shown to produce IL-10 in response to CpG stimulation and ameliorate disease severity in a murine model of lupus [160].Phenotypically, a subset of CD1d hi CD5 + B cells produces IL-10 and B-1a, MZ B and transitional 2-MZ precursor (T2-MZP) cells share this phenotype [161].IL-10 secretion is largely restricted in CD1d hi CD5 + B cells which consist only 1~2% of splenocytes in wild type mice.CD19 + CD23 + CD21 + CD1d hi T2-MZP cells also express IL-10 [162].Similarly, some of human B cells can produce IL-10 [163,164].In particular, CD19 + CD24 hi CD38 hi B cells contain the highest fraction of IL-10 producing B cells upon CD40 stimulation in human peripheral blood from healthy individuals [165].In addition, CD24 hi CD27 + B cells are also known to produce IL-10 [166].Interestingly, CD19 + CD24 hi CD38 hi B cells are associated with immature B cells, while CD24 hi CD27 + B cells are related to memory B cells [165,166].Thus, it is likely that multiple subsets of B cells are able to produce IL-10 in mice, as well as in humans.B REG cells are known to suppress inflammatory responses by producing IL-10, or by directly interacting with pathogenic T cells via a cell-to-cell contact dependent manner [167].Activation of B cells with certain TLR agonists triggers the production of IL-10 and inhibits dendritic cell-mediated activation of T cells in vitro [168].Interestingly, mice deficient in TLR2 and TLR4, or MyD88 in B cells, show a significantly delayed recovery from EAE, indicating that those TLR signals activate the regulatory function of B cells [168].Human B cells express TLR9, and stimulation with CpG plus anti-Ig synergistically induces IL-10 production by human B cells [169].CD40-CD40L interaction is required for B-cell mediated suppression.Agonistic anti-CD40 induces the differentiation of IL-10 producing B cells in splenocytes in an animal model of collagen-induced arthritis mice [159].In humans, blood B cells treated with CD40L are shown to induce Foxp3 + T REG cells via an unknown mechanism [170].In addition, CD40L stimulation increases the CD19 + CD24 hi CD38 hi B cell population that can inhibit the differentiation of Th1 cells via IL-10 [165].Notably, B cells from patients with SLE were insensitive to CD40L and produced limited amount of IL-10 [165].Together, these findings strongly suggest that the CD40-CD40L interaction is critical for the activation of B REG cells.
B cell activating factor (BAFF) is a TNF family protein and plays a key role in B cell maturation and survival [155].A study with BAFF transgenic mice suggests that BAFF can induce Foxp3 + T REG cells to suppress T cell responses in a B cell dependent manner [171].Moreover, in vitro cultures, with a low dosage of BAFF, can induce CD1d hi CD5 + B REG cells, and in vivo treatment with BAFF also increases the number of B REG cells in the marginal zone [172].

Regulatory Mechanism of B REG Cells
The mechanisms by which B REG cells employ their regulatory functions during the immune response have been explored.Two different B REG subsets, CD19 hi CD1d hi CD5 + and CD19 + CD23 + CD21 + CD1d hi T2-MZP cells, are rare in normal conditions, however, they repress both the T cell proliferation and Th1 cytokines (IFN-γ and TNF-α) production through IL-10 [161,162,172].B REG cells can convert effector T cells into regulatory Tr1 cells in vitro in an IL-10 dependent manner [173,174].
B REG cells are also known to control the balance between Foxp3 + and IL-17 producing T cells [175,176].For instance, B REG cells induced by Schistosoma mansoni infection suppressed allergic airway inflammation by increasing pulmonary infiltration of Foxp3 + T REG cells in an IL-10 dependent manner [175,177].Moreover, mice deficient in IL-10 in B cells develop severe arthritis and increased pro-inflammatory T cells (Th1/Th17) and decreased Foxp3 + T REG cells [176,178].In vitro induced B REG cells suppress the differentiation of Th17 from naïve T cells by down-regulating the phosphorylation of STAT3 [179].
B cells can express FasL and other death-inducing ligands to promote activation-induced cell death [167].CD5 + B cells highly express FasL and IL-10 and their regulatory functions come from their Fas-FasL mediated killing ability [167].The expression of FasL on CD5 + B cells is increased by stimulating them with the schistosome antigens IL-4 and IL-10 [180].These FasL + CD5 + B cells are IL-5R high and can be expanded by CD40L and IL-5 stimulation without losing FasL-dependent killing capacity [181].A recent study showed that the frequency of FasL + CD5 + B cells is inversely correlated with the severity of the disease in an animal model of collagen-induced arthritis [182], suggesting that FasL + B cells may have a role in the suppression of autoimmune diseases.Hence, although IL-10 serves as a key mechanism for the suppressive activity of B REG cells, other mechanisms of suppression, including cell-to-cell contact dependent suppression, are also involved in their regulatory capacity.

B REG Cells in Animal Models of Autoimmune Diseases
B REG cells mediate immunosuppression in many types of autoimmunity [155].Rheumatoid arthritis is associated with infiltration of activated T cells, B cells and macrophages that eventually trigger continuous destruction of cartilage and bone structure [183].Collagen-induced arthritis (CIA) is triggered by CD4 + T cells infiltrating into the synovial membrane and by B cells producing collagenspecific antibodies [184].Depletion of B cells by CD20 monoclonal antibody treatment prior to collagen immunization delays the onset of arthritis.On the contrary, transfer of in vitro activated B cells significantly reduces the incidence and severity in a DBA mouse model of arthritis [159].B cells from IL-10-deficient mice failed to protect recipient mice from arthritis [159].Moreover, adoptive transfer of B REG cells that were previously expanded ex vivo in the presence of BAFF suppresses the development of arthritis and relieves the disease severity in CIA mice [172,179].These findings suggest that B REG cells can ameliorate the severity of RA in experimental animal models.
Both T cells and B cells are involved in the pathogenesis of SLE [155].Interestingly, it has been shown that B cell depletion in young NZB/W F1 (four weeks) mice promotes disease onset, while B cell depletion in older mice (12-28 weeks old) delays the disease progression [185].IL-10-producing B REG cells are known to be heavily expanded in young NZB/W F1 mice [186], an observation that might explain the different role of B cell depletion on disease severity in NZB/W F1 mice.CD19 −/− NZB/W mice display delayed autoantibodies production; however, these mice show early nephritis development and poor survival rate due to the lack of B REG cells [187].Accordingly, transfer of splenic B REG cells from NZB/W mice into CD19 −/− NZB/W mice significantly increases survival rates.Similarly, transfer of anti-CD40 antibody induced B REG cells into MPL lpr mice ameliorates the disease severity and increases survival rate in an IL-10-dependent manner [174].Therefore, B REG cells can efficiently suppress different types of antibody-mediated experimental autoimmune diseases.

Follicular Regulatory T Cells
Foxp3 + regulatory T cells (T REG ) are a subset of CD4 + T cells and are necessary for immunological self-tolerance and homeostasis [188].Previous studies demonstrated that T REG cells are found in B cell follicles and GCs [189] and directly suppress B cell responses and auto-reactive B cells [190,191].Over the last five years, a series of studies have unveiled that distinct subsets of T REG cells selectively mediate the suppression of Th1, Th2, and Th17 responses in a CXCR3, IRF4, and STAT3-dependent mechanism, respectively [188,192].More recently, three independent groups simultaneously discovered the existence of T FR cells-a specialized subset of Foxp3 + T REG cells-that control GC reactions in vivo [193][194][195].

Identification and Development of T FR
Foxp3-deficient scurfy mice exhibit a profound population of spontaneous GC B cells as early as four weeks of ages [193].Similarly, the size of T FH population in scurfy mice is significantly increased compared to wild-type mice.These observations clearly demonstrate that Foxp3 + T REG cells are essential for the maintenance of B cell tolerance to self-antigens in the periphery.Importantly, approximately 10%~15% of the CXCR5 + CD4 + T cell population expresses Foxp3.CXCR5 + Foxp3 + T cells express Bcl-6 and Blimp-1, and they are absent in Bcl6-deficient, but not Blimp1-deficient mice [193,194].Such CXCR5 + Bcl6 + Foxp3 + T cells are termed 'follicular regulatory T cells' or 'T FR '.Importantly, T FR cells are present in all secondary lymphoid organs, but not in the thymus, and appear to express the transcription factor, Helios, that has been shown to be exclusively expressed by thymusderived T REG cells.Thus, it is likely that T FR can be differentiated from thymus-derived T REG cells in the periphery as a result of certain inflammatory signals.Although Bcl-6 is required in this process, the type(s) of inflammatory signals that are driving the differentiation of T FR cells remains to be determined.The surface phenotype of T FR cells resembles that of T FH .T FR cells express Bcl-6, CXCR5, PD-1, ICOS, and BTLA, but they lack CD40L, IL-4, and IL-21 [193,194].Another difference between T FR and T FH cells is that the former express Blimp-1 together with Bcl-6, while the latter only express Bcl-6.As a subset of T REG , T FR cells express GITR, CTLA4, CD25 and KLRG1.However, T FR cells do not express CXCR3, indicating that they are distinct from CXCR3 + T REG cells which are shown to be specialized for suppressing type I immune responses.Similar to that of T FH cells, differentiation of T FR cells depends on CD28, ICOS, and SAP [194].Moreover, T FR cells are absent in B cell-deficient mice, indicating that B cells provide crucial signals during T FR differentiation in vivo [194].Interestingly, Sage et al. recently unveiled a regulatory role for PD-1 during T FR differentiation.They showed that PD-1-deficient mice harbor a significantly increased T FR population, defined as CXCR5 + ICOS + Foxp3 + T cells [196].Moreover, they showed that PD-1-deficient T FR cells exhibit greater suppression of immunization-induced GC reactions in vivo.Hence, although T FR cells express PD-1, their interaction with PDL1 seems to inhibit the expansion and suppressive activity of T FR cells in vivo.It would be interesting to determine the involvement of the transcription factors, cytokines, and microRNA clusters that are known to mediate T FH differentiation in the differentiation of T FR cells.

Mechanism of T FR Suppressive Activity
T REG cells from Cxcr5 −/− , Bcl6 −/− , Sh2d1a −/− mice are significantly less efficient in suppressing T cell-dependent antibody production in vivo, compared to wild-type T REG cells [193,194].Therefore, T FR cells play an essential role in controlling GC reactions [194][195][196].Interestingly, a study by Linterman et al. showed that T FR cells control GC B cell responses by suppressing the differentiation of T FH cells [194], while our own study propose that T FR might directly suppress B cells, rather than through T FH cells [193].These differences might be due to the difference in the experimental systems.The former study utilized mixed bone marrow chimeras with 1:1 ratio of Sh2d1a −/− and Foxp3 DTR bone marrow to generate an in vivo system lacking T FR cells [194].In the latter study, T REG cells from Bcl-6 −/− or Cxcr5 −/− mice and naïve CD4 + T cells were co-transferred into Tcrb −/− mice to establish mice lacking T FR cells [193].Thus, it remains to be determined whether T FR cells directly suppress T FH , B cells, or both, during GC reactions in vivo.
The molecular mechanism by which T FR cells regulate GC reactions remains unclear.Nevertheless, current studies give some evidence that high levels of CTLA4, GITR, and IL-10 expression in T FR cells might be important for their regulatory function [193,194,197].Cretney et al. showed that a certain population of T REG cells residing in mucosal sites expresses Blimp-1, and that Blimp-1 induces the expression of IL-10 in T REG cells [198].Deletion of Blimp-1 caused impaired T REG activation and homeostasis.Increased transcription levels of Prdm1 and Il10 in T FR cells indicate that Blimp-1 expression mediates IL-10 expression in T FR cells [194].As described above, PD-1 is related to a suppressive function in T FR cells.Although deletion of PD-1 does not affect T FR cell migration into GCs, PD-1 deficient mice have higher numbers of T FR cells in the lymph nodes and these cells exhibit increased suppressive ability [196].Thus, PD-1 negatively regulates the suppressive function of T FR cells, as well as the development of the cells.Further studies will be needed to define the molecular mechanism of T FR -mediated suppression of GC reactions.

Qa-1 Restricted CD8 + T Cells
In addition to Foxp3 + T REG cells, CD8 + regulatory (CD8 + T REG ) cells have been known to suppress B cell responses by suppressing Th2 immunity [199].Subsequent studies demonstrated that Qa-1 restricted CD8 + T REG cells inhibit the immune responses mediated by Qa-1 expressing CD4 + T cells.Accordingly, Qa-1-deficient mice appeared to be more susceptible to experimental autoimmune diseases, including EAE, RA, and lupus [200,201].Strikingly, a recent study showed that Qa-1 is exclusively expressed on T FH, but not on the other subsets of CD4 + T cells, and that CD8 + T REG cells specifically inhibit T FH responses in vivo [202].

Identification of Qa-1 Restricted CD8 + T REG Cells
Early studies have provided fundamental evidence for the existence of CD8 + T REG cells that suppress T cell-dependent B cell responses in a Qa-1-dependent manner [199,203,204].In addition, a series of animal studies with experimental autoimmune encephalomyelitis also revealed the immune-regulatory function of Qa-1-restricted CD8 + T cells [205].Qa-1 is the murine homolog of non-classical the MHC class 1b molecule, human leukocyte antigen-E (HLA-E) [205].Engagement of the Qa-1/peptide complex by TCR activates and expands Ag-specific CD8 + T cells, while its binding to the CD94/NKG2A receptor decreases the activities of CD8 + T, NK, and NKT cells [205].
Importantly, the HLA-E restricted regulatory mechanism has also been reported in human autoimmune diseases.For instance, CD8 + T cells from patients with recent-onset T1D exhibit defects in the suppression of auto-reactive CD4 + T cells, a condition that can be restored by stimulation with DC primed with the HLA-E binding peptide [206].Moreover, the frequency CD8 + T cells that specifically recognize and lyse activated myelin-reactive CD4 + T cells in an HLA-E restricted manner appeared to be largely reduced in the peripheral blood of multiple sclerosis patients during disease exacerbation, compared with patients in remission and healthy individuals [207].These clinical observations strongly suggest a critical contribution of CD8 + T REG cells to the prevention of autoimmunity.

Mechanism of Qa-1 Restricted CD8 + T REG -Mediated Suppression on GC Reactions
Qa-1 can bind to two different receptors with opposing functions.Qa-1 can act as a MHC I molecule and bind to TCR and activate, as well as expand, Ag-specific CD8 + T cells.On the other hand, when Qa-1-Qdm (Qa-1 determinant modifier) binds to the CD95/NKG2A receptor, expressed by CD8 + T, NK, and NKT cells, it attenuates the activities of the cells [205].A peptide from Heat Shock Protein 60 (Hsp60) is dominantly bound to Qa-1 and Qa-1-peptide/TCR signaling activates antigenspecific CD8 + T cells [205].Replacement of an amino acid at Qa-1 position 227 (D to K) disrupts Qa-1 binding to the TCR/CD8 co-receptor, but has no effect on the inhibitory NKG2A receptor on CD8 and NK cells [202,208].Accordingly, mice with this Qa-1 single mutation (D227K) do not have Qa-1 restricted CD8 + T REG cells.These mice exhibit significantly increased numbers of T FH cells in the secondary lymphoid organs and develop SLE-like disease [202,205].
Qa-1 restricted CD8 + T REG cells express CXCR5 and ICOSL and migrate to B cell follicles, and attenuate the function of highly Qa-1 expressing T FH cells [202].Unlike conventional CD4 + T REG cells, these CD8 + T REG cells do not express Foxp3, and their suppressive function is mediated by perforin and IL-15 [202,205].Treatment with anti-IL-15 inhibits the suppressive activity of CD8 + T REG cells [202].Subsequent studies demonstrated the suppressive role of CD8 + T REG cells in a B6-Yaa mouse model of lupus and a mouse model of collagen induced arthritis [209,210].Notably, in vitro expanded CD8 + T REG cells successfully inhibit CIA by reducing auto-reactive T FH and Th17 cells via a perforin-dependent mechanism [210].
Killer cell immunoglobulin-like receptors (KIR) in humans, a functional counterpart of the Ly49 receptor in mice, is expressed in about 4%~5% of CD8 + T cells in humans [211,212].Similar to mouse Ly49 + CD8 + T cells, human KIR + CD8 + T cells recognize HLA-E, and express perforin [211].The role of this KIR + CD8 + T cell population remains to be determined.

Translational Potentials
Generation of high-affinity antibodies and strong memory B cells improve vaccine efficacy.On the other hand, suppression of auto-antibody production would ameliorate systemic autoimmune diseases.Thus, the immune cell subsets discussed in this review might provide a wide range of translational opportunities for the development of immunotherapy for human diseases.

Vaccine Design
Considering the crucial contributions of T FH cells to GC B cell responses and memory B cells, enhancement of T FH differentiation/function would improve the efficacy of vaccination [213].Recent studies showed that certain adjuvants enhanced the generation of T FH cells.For instance, vaccination with a nanoparticle (NP) delivery system promotes high-titer, high-avidity Ab responses to malaria antigens by enhancing the number of T FH cells, compared to a US FDA-approved adjuvant monophosphoryl lipid A (MPLA) [214].In addition, IFN-α has been shown to enhance the expansion of T FH cells in an animal model of adenoviral-vector based vaccination [215].
Interestingly, Zheng et al. have recently shown that immunization with a DNA vaccine containing the mGITRL gene significantly enhanced the T FH population and increased the levels of Ag-specific IgG to RagB of Porphyromonas gingivalis (P.gingivalis) [216].GITR/GITRL signaling can act like co-stimulatory molecules in TCR signal mediated T cell activation and proliferation [217].However, the role of the GITR signaling pathway in T FH, or GC B cell, responses is not clear.Modulating SAP and SLAM family receptor expression on T or B cells could be another target for optimizing vaccine efficacy since SAP:SLAM family receptor interaction is critical for long-term B:T FH interaction during GC reactions [218].Notably, over-expression of EAT2-a SLAM adapter protein-in DCs and macrophages enhances the induction of Ag-specific immune responses [219].
Similarly, blockade of inhibitory pathways during T FH and GC B cell responses might also offer attractive targets, since inhibitory co-stimulators are involved in chronic infections and cancer as immunosuppressive mechanisms [220].For instance, since PD-L1 is known to be highly expressed in various tumors and chronic infections [220], and since its ligand PD-1 is highly expressed on T FH cells, blockade of PD-1/PD-L1 would improve the expansion and function of T FH cells in tumor-bearing or infected hosts.In addition, it has been described that blocking B7-H1, but not B7-DC, increases the differentiation of T FH cells and enhances antigen specific Ig responses [221].Another study also demonstrated that blockade of PD-L1 and LAG3 signals in Plasmodium yoelii infected mice rapidly clears malaria by enhancing humoral immune responses with an increased number of T FH and GC B cells [222].
Ex vivo expansion of tumor-infiltrating lymphocytes has been established in order to obtain a large number of tumor-specific T cells.Administration of tumor-specific antibodies is one of the most successful immunotherapeutic approaches for the eradication of tumor cells in vivo.As described above, the molecular and cellular factors required for T FH cell generation are well documented.Thus, it seems possible to obtain a large number of tumor-specific T FH cells.It would be interesting to investigate whether transfusion of tumor-specific T FH cells would enhance anti-tumor immunity by inducing anti-tumor antibodies.

Autoimmune Diseases
As discussed above, in Section 2.3, exaggerated GC B cells and T FH responses represent key pathophysiologic features of antibody-mediated autoimmune diseases in humans.Despite differences in the cellular origin and suppressive mechanism, the three types of suppressor cells discussed in this review share a common outcome when it comes to their activities: inhibition of GC reaction.It is not clear if defects in these suppressor cell populations lead to autoimmunity in humans.Nevertheless, it is plausible to surmise that in vivo expansion, or administration of the suppressor cells, would be beneficial to alleviate the problem of production of autoantibodies.
Importantly, transfer of circulating blood T FR cells results in the suppression of Ag-specific IgG responses, and PD-1-deficient T FR cells show stronger suppression [196].Similarly, adoptive transfer of Qa-1 restricted CD8 + T REG cells and B REG cells induce diminished antibody production [155,202,209].As a next step, further studies will be needed to address if transfer of these GC suppressor cells can ameliorate the production of autoantibodies in animal models of lupus and other antibody-mediated disorders.The availability of only a limited number of suppressor cells is likely the main obstacle preventing their use in in vivo experimental studies.Given that T and B cells can be easily expanded, finding the right conditions for ex vivo expansion (or for differentiation) would enable us to overcome this obstacle.A series of future studies will be needed before GC suppressor cell-based immunotherapy can be developed as a therapy for personalized medicine.

Conclusions
The diverse types of immune cells involved in GC B cell responses reflect an orchestrated regulation of this process.In the past decade, there have been a number of milestone discoveries describing the regulation of GC reactions.Identification of the T FH cell lineage and the discoveries of T FR cells, CD8 + T REG cells, and B REG cells have broadened our understanding of the complex balance of B cell immunity and tolerance.We expect that future studies will define the suppressive mechanisms by which each suppressor cell subset inhibits GC B cell responses.Various types of expanded or engineered immune cells are currently under pre-clinical and clinical investigations.As discussed above, the suppressor cells of the GC reaction offer great potential for translational research and may pave the way for the development of novel therapeutics for autoimmune diseases, as well as infections and cancers [212].

Figure 1 .
Figure 1.Schematic view of germinal center reactions.(a) Naïve CD4 + T cells interact with antigen-presenting dendritic cells (DCs) in the T cell zones.(b) Activated T cells transiently express CXCR5 and migrate to the T-B border.(c) Interaction of CXCR5 + T cells with Ag-specific B cells and follicular DCs further promote T FH differentiation, as well as the formation of germinal center (GC).(d) In GC, T FH cells induce clone expansion, somatic hypermutation, and class switching of B cells.(e) The interaction of T FH and B cells leads to the generation of memory B cells and long-lived antibodyproducing plasma cells.Transcription factors, Bcl-6, c-Maf, BATF and IRF4, in T FH cells direct the expression of T FH signature genes, including CXCR5, ICOS, IL21, and PD1.Interactions of CD28:CD86, CD40L:CD40, PD-1:PD-1L, SLAM:SAP and ICOS: ICOSL between T FH and B cells are required for GC formation.(f) Follicular regulatory T cells (T FR ), Qa-1 restricted CD8 + regulatory T cells, and IL-10 producing regulatory B (B REG ) cells are known to suppress GC responses.

Figure 2 .
Figure 2. Diverse regulatory cells that control GC B cell responses.In order to maintain B cell tolerance to self-antigens and to control the size of GC reactions, the immune system establishes multiple subsets of regulatory cells, such as T FR cells, Qa-1 restricted CD8 + regulatory T cells, and regulatory B cells.(a) At least three subsets of B REG cells are known to exert immunosuppressive activity: CD1d hi CD5 + B cells, CD19 + CD24 hi CD38 hi B cells, and CD19 + CD23 + CD21 + CD1d hi T2-MZP B cells.They can be stimulated by TLR/CD40 signaling and secrete IL-10.(b) T FR cells express Bcl-6 and Blimp-1 in addition to Foxp3.Phenotypically, they express T FH -related molecules such as CXCR5, PD-1, ICOS, as well as the T REG -associated molecules CTLA4, GITR and IL-10.(c) T cell receptor of Qa-1 restricted CD8 + regulatory T cells recognize the MHC class Ib molecule, Qa-1, that is expressed exclusively on T FH cells.This TCR/Qa-1-peptides interaction triggers the suppressive activity of CD8 + regulatory T cells and limits GC reactions by inhibiting the function of T FH cells.