BCR specificity, affinity and signaling are the most important currently identified factors in B10 cell development. B10 cell regulation of inflammation and autoimmunity is Ag specific 23 30 36 . The importance of BCR diversity is demonstrated by the fact that B10+ B10_PRO cells are reduced by approximately 90% in transgenic mice with a fixed BCR 37 . Signaling through the BCR appears critical during early development in vivo. CD19-deficient mice (where BCR signaling is decreased) have a 70 to 80% decrease in B10+B10_PRO cells 30 . In contrast, B10 cells are expanded in human CD19 transgenic mice (where the overexpression of CD19 augments BCR signaling). The absence of CD22, which normally dampens CD19 and BCR signaling 38 , also results in increased B10 cell numbers. Ectopic B cell expression of CD40L (CD154) in transgenic mice, which induces increased CD40 signaling 39 , also increases B10 cell numbers. CD22^−/− mice that also ectopically express CD40L show dramatically enhanced numbers of CD1d^hiCD5⁺ B cells and B10 cells 40 . The induction of IL-10⁺ B cells with regulatory activity by T cell immunoglobulin domain and mucin domain protein 1 (TIM-1) ligation 41 further highlights the importance of BCR signaling in B10 cell development. BCR signaling and TIM-1 are closely related. BCR ligation induces TIM-1 expression on B cells 41 42 , and TIM-1 ligation appears to enhance BCR signaling since it increases antibody production both in vitro and in vivo 43 . The importance of BCR-related signals is further highlighted by the observation that the stromal interaction molecules 1 (STIM1) and 2 (STIM2) are required for B cell IL-10 production 44 . Remarkably, B cells lacking both stromal interaction molecule proteins failed to produce IL-10 after BCR stimulation in the presence of PMA and ionomycin for 5 hours 44 . All of the above indicate that BCR-related signals are particularly important in B10 cell development.

Despite the requirement for BCR expression and function during mouse B10 cell development, B cell stimulation with mitogenic anti-IgM antibody alone does not induce cytoplasmic IL-10 expression. The combination of anti-IgM stimulation with CD40 ligation and LPS or CpG significantly reduces IL-10 competence 37 . BCR-generated signals thus inhibit the abilities of LPS or CpG and CD40 ligation to induce cytoplasmic IL-10 production. Whether BCR stimulation inhibits the induction of IL-10 competence by inducing B cells to mature or differentiate down a divergent pathway or diverts intracellular signaling is unknown. Another possibility is that the signals generated by mitogenic anti-IgM BCR cross-linking are too intense and that low-affinity Ag--BCR interactions drive B10_PRO cell development in vivo.

A recent study revealed the importance of IL-21, major histocompatibility complex class II (MHC-II) and CD40 during cognate interactions with CD4⁺ T cells in B10 cell development 36 . Ex vivo stimulation of purified spleen CD19⁺ B cells with IL-21 induced 2.7-fold to 3.2-fold higher B10 cell frequencies, and 4.4-fold to 5.3-fold more IL-10 secretion compared with stimulation with media alone. Remarkably, IL-21 induced B10 cells to produce IL-10 without the need for stimulation with phorbol esters and ionomycin. Interestingly, IL-21 induced a threefold increase in IL-10⁺ B cells within the splenic CD1d^hiCD5⁺ B cell subset, but did not induce IL-10⁺ B cells within the CD1d^loCD5^-- B cell subset. Both B10 cells and non-B10 cells expressed IL-21R at similar levels, and ex vivo B10, B10pro and CD1d^hiCD5⁺ B cell numbers were similar among IL-21R-deficient (IL-21R^--/--), MHC-II-deficient (MHC-II^--/--) and CD40-deficient (CD40^--/--) mice. Nevertheless, IL-21R, MHC-II and CD40 appear to be required for B10 cell effector functions, at least in experimental autoimmune encephalomyelitis (EAE) 36 . Regulatory B10 cell function therefore requires IL-21R signaling, as well as CD40 and MHC-II interactions, potentially explaining Ag-specific B10 cell effector function 37 .

Although cognate interactions with CD4⁺ T cells are important for B10 cell effector functions 36 , T cells do not appear to be required for B10 cell development. B10 cells are present in T cell-deficient nude mice, and their frequencies and numbers are approximately fivefold higher when compared with wildtype mice. This observation is strengthened by the fact that MHC class I and MHC-II molecules and CD1d expression are not required for B10 cell development 37 . The presence or absence of T cells in vitro also does not affect the frequency of B10 cells. Although increased B10 cell frequencies in T cell-deficient mice suggest that T cells might actually inhibit B10 cell development, it is equally possible that the immunodeficient state of these mice allows subclinical inflammation that induces B10 cell generation. The role of T cells in B10 cell development in vivo is thereby complex and, although T cells are not required for B10 cell development, cognate interactions between CD4⁺ T cells and B10 cells are required for B10 cell effector function.

B10 cells can be driven to produce IL-10 by TLR4 (LPS) or TLR9 (CpG oligonucleotides) ligands. Mouse B10_PRO cells acquire the ability to function like B10 cells after in vitro maturation following stimulation with LPS, but not CpG, in the presence or absence of agonistic CD40 mAb 32 . TLR4 and TLR9 signaling through myeloid differentiation primary response gene 88 (MyD88) is necessary for the optimal maturation and IL-10 induction of B10pro and B10 cells following LPS stimulation and LPS or CpG stimulation, respectively 37 . Nevertheless, MyD88 expression is not an absolute requirement for B10 cell development in vivo, since B10 cells develop normally in MyD88^--/-- mice 37 . Specifically, numbers of B cells with the capacity to produce IL-10 are equivalent in wildtype and MyD88^--/-- mice when their maturation or IL-10 production are measured following CD40 ligation or PMA plus ionomycin stimulation, respectively, demonstrating that B10 _PRO and B10 cells are present at normal frequencies in MyD88^--/-- mice. Thereby, while TLR signaling is not required for B10 cell development, MyD88 expression is required for LPS to induce optimal B cell IL-10 expression and secretion in vitro.

The involvement of TLR signals in B10-cell IL-10 production was recently demonstrated 45 . IL-10 production by B cells, stimulated by contact with apoptotic cells, results from the engagement of TLR9 within the B cell after recognition of DNA-containing complexes on the surface of apoptotic cells by the BCR. An earlier study also highlights the effects of apoptotic cells on B cell IL-10 production, where apoptotic cells protected mice from developing collagen-induced arthritis (CIA) by the induction of IL-10-producing regulatory B cells 46 . Cell death products may therefore represent one of the physiologic triggers for B10 cell development by providing a combination of BCR and TLR signals. Additional non-TLR/non-BCR signals (such as alarmins) released from dying cells may be also involved but their identities remain to be determined.

Although certain transcription factors are involved at some point in B10 cell development, it is important to stress that there is no known transcription factor signature unique to B10 cells. Following a transient period of IL-10 transcription characterized by increased expression of the blimp1 and irf4 transcription factors along with decreased expression of pax5 and bcl6, a significant but small fraction of B10 cells can differentiate into antibody-secreting cells producing IgM and IgG polyreactive antibodies that are enriched for autoreactivity to single-stranded or double-stranded DNA and histones 47 . Whether B10 cells can produce and secrete IL-10 repeatedly remains to be determined.

B10_PRO cells and B10 cells have been recently identified in humans 48 and their responses to LPS, CpG and CD40 ligation appear to follow the general scheme of mouse B10 cell development (Figure 1). One notable difference in mouse versus human B10 cell development is the lack of response of mouse B10_PRO cells to CpG compared with their human counterparts. Human B10_PRO cells can be driven to develop ex vivo into B10 cells with LPS or CpG stimulation, or CD40 ligation. Interestingly, BCR ligation augmented human B cell IL-10 responses to CpG in one study 49 . This finding is in discordance with our findings in both humans 48 and mice 37 , where BCR-generated signals inhibit the abilities of LPS or CpG and CD40 ligation to induce cytoplasmic IL-10 production. Whether human B10 cells develop into antibody-secreting cells or enter the memory pool (memory B10 cells, B10_M) remains to be determined.

The most critical unsolved issue relates to the nature of antigenic stimuli driving B10 cell development. The identification of B10-cell BCR specificity is imperative since it will provide new insights into their early development. The autoreactive nature of mouse B10-cell BCRs 47 suggests that autoantigens may be driving early B10 cell development and that B10 cells may represent one of the ways enabling the immune system to peripherally tolerate autoantigens. B cells responding to autoantigens in an IL-10-dependent regulatory way can potentially limit inflammatory responses and limit autoimmune phenomena (see later section on B10 cell regulatory effects and Figure 2). Cell death products, by providing simultaneously both antigenic and nonantigenic stimuli, may represent one of the physiologic triggers for B10 cell development. The clearance of antigenic products of dying cells by noncomplement-fixing IgM polyreactive/autoreactive antibodies (such as those made by mouse B10 cells) in an IL-10-rich environment would be beneficial since it could potentially limit inflammatory responses to self-Ags. Additional unidentified antigenic and nonantigenic stimuli are probably involved in B10 cell development. The identification of such stimuli will provide additional insights ito B cell development that may prove invaluable for the future manipulation of B10 cells for treating autoimmune disease. Another important question is whether B10 cells enter the B cell memory pool during their development. This question is suggested by human studies demonstrating that B10_PRO cells and B10 cells share phenotypic features with memory B cells (see later section on Human B10 cell phenotype).

Although a variety of cell surface markers have been proposed 31 32 , there is no known surface phenotype unique to B10 cells and, currently, the only way to identify these cells is functionally by intracellular IL-10 staining 35 . Only a small portion of B cells (that is, ~1 to 3% of splenic B cells in wildtype C57BL/6 mice) produce IL-10 following PMA and ionomycin stimulation, implying that not all B cells are competent to produce IL-10. Intracellular cytokine staining combined with flow cytometric phenotyping shows that mouse spleen B10 cells are enriched within the small CD1d^hiCD5⁺ B cell subset, where they represent 15 to 20% of the cells in C57BL/6 mice. This phenotypically unique CD1d^hiCD5⁺ subset shares overlapping cell surface markers with a variety of phenotypically defined B cell subsets such as CD5⁺ B-1a B cells, CD1d^hiCD23⁻IgM^hiCD1d^hi marginal zone B cells, and CD1d^hiCD23⁺IgM^hiCD1d^hi T2 marginal zone precursor B cells, which all undoubtedly contain both B10_PRO cells and B10 cells 23 26 30 50 . Mouse B10 cells are predominantly IgD^lowIgM^hi, and <10% co-express IgG or IgA, but they can differentiate into antibody-secreting cells secreting polyreactive or Ag-specific IgM and IgG 47 . IL-10+ B cells were recently shown to be enriched in the TIM-1⁺ compartment and TIM-1⁺ B cells are enriched in the CD1d^hiCD5⁺ compartment 41 . However, IL-10⁺ B cells are also present in the TIM-1^-- compartment and TIM-1⁺ B cells are present in the non-CD1d^hiCD5⁺ compartment. Intracellular cytoplasmic IL-10 staining thereby remains the only current way to visualize the entire subset of IL-10-competent B cells. Nonetheless, the isolation of CD1d^hiCD5⁺ B cells or other phenotypically defined B cell subsets where B10 cells are enriched currently provides the best current means for isolating a viable B cell population that is significantly enriched for B10 cells and can be used for adoptive transfer experiments and functional studies in mice.

The IL-10-producing B cell subset characterized in humans normally represents <1% of peripheral blood B cells 48 . Peripheral blood B10 cells and B10_PRO cells are highly enriched in the CD24^hiCD27⁺ B cell subset, with approximately 60% also expressing CD38. Similar total numbers of IL-10⁺ B cells have been described in the CD24^hiCD38^hi and CD24^intCD38^int B cell subsets 51 . A separate study showed that B10 cells did not fall within any of the previously defined B cell subsets, but they were enriched in the CD27⁺ and the CD38^hi compartments 49 . Human B10 cells also highly express CD48 and CD148 48 . CD48 is a B cell activation marker 52 and CD148 is considered a marker for human memory B cells 53 . CD27 expression is another well-characterized marker for memory B cells, although some memory B cells may be CD27^-- 54 55 56 . The CD27⁺ B cell subset can also expand during the course of autoimmunity and has been proposed as a marker for disease activity 54 56 . The CD24^hiCD148⁺ phenotype of B10 cells and B10_PRO cells may thereby indicate their selection into the memory B cell pool during development, or they may represent a distinct B cell subset that shares common cell surface markers with memory B cells. Consistent with a memory phenotype, the proliferative capacity of human blood B10 cells in response to mitogen stimulation is higher than that for other B cells 48 , as is seen for mouse B10 cells 37 . Human transitional B cells are rare (2 to 3% of B cells) in adult human blood and are generally CD10⁺CD24^hiCD38^hi cells that are also CD27-negative 55 56 ; since CD10 expression is a well-accepted marker for most cells within the transitional B cell pool 57 , its absence on B10 cells suggests that these cells are not recent emigrants from the bone marrow. In summary, human B10 cells share phenotypic characteristics with other previously defined B cell subsets, and, currently, there is no known surface phenotype unique to B10 cells.

EAE is an established model of multiple sclerosis induced by immunization with myelin peptides (such as myelin oligodendrocyte glycoprotein) leading to demyelination mediated by auto-Ag-specific CD4⁺ T cells 76 77 . B cells were shown over a decade ago to have regulatory properties during the induction of EAE, with genetically B cell-deficient mice developing a severe nonremitting form of the disease 21 . However, these B cell regulatory effects were recently shown not to be IL-10 dependent 34 . Nonetheless, other studies highlight the importance of B cell-derived IL-10 in EAE. Specifically, EAE severity during the late phase of disease increases in B cell-deficient μMT mice that do not fully recover from their disease when compared with wildtype mice, and the adoptive transfer of wildtype B cells but not IL-10^--/-- B cells normalizes EAE severity in μMT mice 22 . Disease recovery is dependent on the presence of autoantigen-reactive B cells, and B cells isolated from mice with disease produced IL-10 in response to autoantigen stimulation. In the absence of Ag-specific B cell IL-10 production, the proinflammatory T_H1-mediated immune responses persist and mice do not recover from the disease.

The EAE model demonstrates the complexity of regulatory mechanisms mediated by different cell subsets during different stages of the disease. When B cells from wildtype mice are depleted by CD20 mAb treatment 7 days before EAE induction, there is an increased influx or expansion of encephalitogenic T cells within the central nervous system and exacerbation of disease symptoms 23 . This effect is related to B10 cell depletion since similar effects are observed with selective B10 depletion by means of CD22 mAb 60 . The adoptive transfer of Ag-specific (myelin oligodendrocyte glycoprotein-sensitized) B10 cells into wildtype mice also reduces EAE initiation dramatically. The protective effect is IL-10 dependent since the adoptive transfer of CD1d^hiCD5⁺ B cells purified from IL-10^−/− mice does not affect EAE severity. B10 cell effector functions in EAE require IL-21 along with cognate interactions with CD4⁺ T cells since the adoptive transfer of CD1d^hiCD5⁺ B cells into CD19^--/-- mice from IL-21R^--/--, MHC-II^--/-- or CD40^--/-- mice prior to the induction of EAE does not alter disease course 36 . Once disease is established, adoptive transfer of B10 cells does not suppress ongoing EAE. B10 cells thereby appear to normally regulate acute autoimmune responses in EAE. In contrast to the role of B10 cells in early disease, T_REG depletion enhances late-phase disease. Therefore, in EAE, depending on the stage of the disease, different regulatory mechanisms are involved in limiting inflammatory responses, with B10 cells regulating disease initiation and T_REGS being involved predominantly in the regulation of late-phase disease.

IL-10-producing B cells regulate intestinal inflammation in inflammatory bowel disease 26 . Early studies showed that B cells and their autoantibody products suppress colitis in T cell receptor alpha chain-deficient mice that spontaneously develop chronic colitis, while B cells are not required for disease initiation 78 . B cells with upregulated CD1d expression in the gut-associated lymphoid tissues of mice with intestinal inflammation were subsequently demonstrated to be regulatory 25 . This IL-10-producing B cell subset appears during chronic inflammation in T cell receptor alpha chain-deficient mice and suppresses the progression of intestinal inflammation by downregulating inflammatory cascades associated with IL-1 upregulation and signal transducer and activator of transcription 3 (stat3) activation rather than by altering polarized T_H cell responses. The adoptive transfer of these mesenteric lymph node B cells also suppresses inflammatory bowel disease through a mechanism that correlates with an increase in T_REG subsets 67 . Oral administration of dextran sulfate sodium solution to mice is widely used as a model of human ulcerative colitis. Dextran sulfate sodium-induced intestinal injury is more severe in CD19^--/-- mice (where B10 cells are absent) than in wildtype mice 79 , and these inflammatory responses are negatively regulated by CD1d^hiCD5⁺ B cells producing IL-10. B10 cells therefore emerge during chronic inflammation in mouse models of inflammatory bowel disease, where they suppress the progression of inflammatory responses and ameliorate disease manifestations.

CIA is a model for human rheumatoid arthritis that develops in susceptible mouse strains immunized with heterologous type II collagen emulsified in complete Freund's adjuvant 80 81 . CIA and rheumatoid arthritis share in common an association with a limited number of MHC-II haplotypes that determine disease susceptibility 82 83 . B cells are important for initiating inflammation and arthritis since mature B cell depletion significantly reduces disease severity prior to CIA induction but does not inhibit established disease 84 . Several studies on CIA demonstrate the negative regulatory effects and therapeutic potential of B10 cells.

Activation of arthritogenic splenocytes with Ag and agonistic anti-CD40 mAb induces a B cell population that produces high levels of IL-10 and low levels of IFNγ 85 . The adoptive transfer of these B cells into DBA/1--T cell receptor--β-Tg mice, immunized with bovine collagen (type II collagen) emulsified in complete Freund's adjuvant, inhibits T_H1 responses, prevents arthritis development, and is effective in ameliorating established disease. The adoptive transfer of CD21^hiCD23⁺IgM⁺ B cells from DBA/1 mice in the remission phase prevents CIA and reduces disease severity through IL-10 secretion 86 ; a significant but less dramatic therapeutic effect on CIA progression is seen when cells from naïve mice are adoptively transferred. In addition, the adoptive transfer of ex vivo expanded CD1d^hiCD5⁺ B cells in collagen-immunized mice delays arthritis onset and reduces disease severity, accompanied by a substantial reduction in the number of T_H17 cells 61 . Co-culture of CD1d^hiCD5⁺ B cells with naive CD4⁺ T cells suppresses T_H17 cell differentiation in vitro, and co-culture of CD1d^hiCD5⁺ B cells with T_H17 cells results in decreased proliferation responses in vitro. Furthermore, the adoptive transfer of T_H17 cells triggers CIA in IL-17^--/-- DBA mice; however, when T_H17 cells are co-transferred with CD1d^hiCD5⁺ B cells, the onset of CIA is significantly delayed. Finally, in a different study, administration of apoptotic thymocytes along with ovalbumin peptide and complete Freund's adjuvant to mice carrying an ovalbumin-specific rearranged T cell receptor transgene (DO11.10 mice) up to 1 month before the onset of CIA resulted in an increase in ovalbumin-specific IL-10 secretion and is protective for severe joint inflammation and bone destruction 46 . Activated spleen B cells responded directly to apoptotic cell treatment in vitro by increasing secretion of IL-10, and inhibition of IL-10 in vivo reversed the beneficial effects of apoptotic cell treatment 46 .

B cell-negative regulatory effects are important in NZB/W mice, a spontaneous lupus model, since mature B cell depletion initiated in 4-week-old NZB/W F1 mice hastens disease onset, which parallels depletion of B10 cells 87 . B10 cells are phenotypically similar in NZB/W F1 and C57BL/6 mice, but are expanded significantly in young NZB/W F1 mice 87 . In wildtype NZB/W mice, the CD1d^hiCD5⁺B220⁺ B cell subset, which is enriched in B10 cells, is increased 2.5-fold during the disease course, whereas CD19^--/-- NZB/W mice lack this CD1d^hiCD5⁺ regulatory B cell subset 88 . Finally, the potential therapeutic effect of B10 cells in lupus is highlighted by the prolonged survival of CD19^--/-- NZB/W recipients following the adoptive transfer of splenic CD1d^hiCD5⁺ B cells from wildtype NZB/W mice 88 . Studies in the NZB/W spontaneous lupus model therefore suggest that B10 cells have protective and potentially therapeutic effects.

In the MRL.Fas(lpr) mouse lupus model, B cell-derived IL-10 does not regulate spontaneous autoimmunity 89 . B cell-specific deletion of IL-10 in MRL.Fas(lpr) mice indicates that B cell-derived IL-10 is ineffective in suppressing the spontaneous activation of self-reactive B cells and T cells during lupus. The severity of organ disease and survival rates in mice harboring IL-10-deficient B cells were unaltered. MRL.Fas(lpr) IL-10 reporter mice illustrate that B cells comprise only a small fraction of the pool of IL-10-competent cells. In contrast to previously published studies from our laboratory and elsewhere, putative regulatory B cell phenotypic subsets, such as CD1d^hiCD5⁺ and CD21^hiCD23^hi B cells, were not enriched in IL-10 transcription. This observation suggests fundamental differences in the pathogenesis and immune dysregulation in the NZB/W lupus model compared with the MRL.Fas(lpr) model.

Studies on B10 cells and mouse models of diabetes are limited to the nonobese diabetic (NOD) mouse, a spontaneous model of type 1 diabetes in which autoimmune destruction of the insulin-producing pancreatic β cells is primarily T cell mediated 90 . Although B cells clearly have a pathogenic role in disease initiation 91 , B cells activated in vitro can maintain tolerance and transfer protection from type 1 diabetes in NOD mice 92 93 . The adoptive transfer of BCR-stimulated B cells into NOD mice starting at 5 to 6 weeks of age both delays the onset and reduces the incidence of type 1 diabetes, while treatment at 9 weeks of age delays disease onset. Protection from type 1 diabetes requires B cell IL-10 production since the adoptive transfer-activated NOD-IL-10^--/-- B cells do not confer protection from type 1 diabetes or the severe insulitis in NOD recipients. The therapeutic effect of adoptively transferred activated NOD B cells correlates with T_H2 polarization. The limited data above suggest that B10 cells may be protective in preventing establishment of type 1 diabetes in NOD mice.