Regulatory B10 Cells Differentiate into Antibody-Secreting Cells After Transient IL-10 Production In Vivo

B lymphocytes mediate humoral immunity through their production of secreted Ab but are also central regulators of CD4⁺ T cell activation by serving as APCs and providing costimulatory molecules and cytokines that regulate cellular immune responses during T cell expansion, memory formation, and cytokine production (1). However, B cells and specific B cell subsets can also negatively regulate immune responses (2). The absence or loss of these regulatory B cells exacerbates disease symptoms in diverse models of inflammation and autoimmunity, predominantly through the production of the regulatory cytokine IL-10 (3–11).

A specific subset of regulatory B cells was recently found to inhibit inflammation, autoimmunity, and innate and adaptive immune responses through the production of IL-10 (8, 9, 12, 13), a potent and pleiotropic cytokine (14). We call these B cells “regulatory B10 cells” because IL-10 is required for their negative regulatory function (2), and additional B cell subsets with unique regulatory properties also exist. For example, IL-12–producing B cells regulate intestinal inflammation (15). In mice, regulatory B10 cells are functionally identified by cytoplasmic IL-10 expression following in vitro stimulation with LPS, PMA, and ionomycin (L+PI), with monensin (L+PIM) included in the cultures to block IL-10 secretion (8, 9). Spleen B10 cells are found at low frequencies (1–5%), where they are predominantly found within the phenotypically unique CD1d^hiCD5⁺CD19^hi B cell subpopulation (8–10). Regulatory B10 cells share overlapping cell surface markers with multiple other phenotypically defined B cell subsets (B1a, marginal zone, and marginal zone precursor cells), potentially consistent with their localization within spleen follicles and marginal zones (16). B10 cells are presumed to be functionally mature because they are competent to express IL-10 after 5 h of ex vivo stimulation, and they proliferate rapidly following in vitro or in vivo activation (12, 17). Additional B cells within the CD1d^hiCD5⁺ B cell subpopulation acquire the ability to function like B10 cells during 48 h of in vitro stimulation with either agonistic CD40 mAb or LPS (17). These B10 progenitor (B10pro) cells are then able to express cytoplasmic IL-10 following L+PIM stimulation for 5 h. Regulatory B10 cell functions are Ag restricted in vivo (8, 9), with B10pro and B10 cells requiring diverse AgRs (BCR) for their development (17). Spleen B10 cell numbers increase significantly during inflammation and autoimmunity, with the adoptive transfer of Ag-primed CD1d^hiCD5⁺ B cells suppressing inflammation and disease in mouse models (8, 9, 11, 17, 18). Human blood B10 and B10pro cells that parallel their mouse counterparts are equally rare and represent a subset of the circulating CD24^hiCD27⁺ memory B cell subset (12). Thus, the capacity of human and mouse B10pro and B10 cells to express IL-10 is central to their regulatory function.

IL-10 reporter mice have been developed to examine regulatory T cell IL-10 expression and cell fates. In Tiger mice, an internal ribosomal entry site-GFP construct follows the genomic il10 coding sequence, resulting in cytoplasmic GFP expression during il10 transcription (19). Similarly, 10BiT mice express Thy1.1 under the control of il10 BAC-transgene regulatory elements, leading to cell surface Thy1.1 expression following IL-10 production (20). In the current studies, IL-10 reporter expression was used to track regulatory B10 cell induction and fates in Tiger and 10BiT mice, with the findings that regulatory B10 cells only transiently express IL-10 prior to their terminal differentiation into clonally diverse Ab-secreting plasmablasts and plasma cells that contribute significantly to the serum Ab pool. Thereby, regulatory B10 cells limit inflammation and immune responses by the production of IL-10 and contribute to humoral immunity.

C57BL/6 and Rag2^−/− mice were from National Cancer Institute-Frederick (Bethesda, MD). Tiger mice (19) were from The Jackson Laboratory (Bar Harbor, ME). A gene dose-dependent decrease in IL-10 production was not observed in homozygous Tiger mice, which occurs with T cells (19). Hemizygous 10BiT mice were as described (20). Mice were housed in a specific pathogen-free barrier facility with end-point analyses carried out between 8–14 wk of age. Mice were given (i.p.) sterile LPS in PBS (25 μg, Escherichia coli, clone 0111:B4; Sigma, St. Louis, MO), CFA, or IFA (200 μl 1:1 emulsified mixture with PBS; Sigma); Imject Alum (200 μl 1:1 emulsified mixture with PBS; Pierce, Rockford, IL); or alum with 2,4,6, trinitrophenyl-keyhole limpet hemocyanin (TNP-KLH, 50 μg/200 μl; Biosearch Technologies, Novato, CA). All studies and procedures were approved by the Duke University Animal Care and Use Committee.

B cells enriched (>95% CD19⁺) from single-cell tissue suspensions by MACS selection using CD19-microbeads (Miltenyi Biotec, Auburn, CA) were cultured in complete medium (RPMI 1640 medium containing 10% FBS, 1% HEPES, 1% l-glutamine, 1% Pen/Strep, and 0.1% 2-ME). Sterile LPS (10 μg/ml), goat F(ab′)₂ anti-mouse IgM Ab (5 μg/ml; Jackson ImmunoResearch, West Grove, PA), and CD40 mAb (2 μg/ml, clone HM40-3; BD Pharmingen, San Jose, CA) were added to cultures where indicated.

Single-cell leukocyte suspensions were stained with predetermined optimal Ab concentrations, as described (21), with cytoplasmic IL-10 expression assessed as described (22). Abs included anti-mouse IL-10 (JES5-2A5), CD138 (281-2), CD43 (S7), CD38 (90), and GL7 (Ly-77) mAbs from BD Pharmingen; CD16/CD32 (FcBlock), FITC-, PE-, PE.Cy5-, PE.Cy7-, biotin-, or allophycocyanin-conjugated anti-mouse B220 (clone RA3-6B2), CD19 (eBio1D3), CD1d (1B1), CD5 (53-7.3), Thy1.1 (HIS51), Thy1.1 (OX-7), CD21/35 (eBio8D9), and CD23 (B3B4) mAbs from eBioscience (San Diego, CA); anti-mouse IL-10 (JES5-16E3), CD19 (6D5), and CD16/32 (TruStain) from BioLegend (San Diego, CA); and goat anti-mouse IgM Ab (Southern Biotech, Birmingham, AL). In some instances, streptavidin conjugated to PE.Cy5 or PE.Cy7 (eBioscience) was used to reveal biotinylated Ab binding. Anti-mouse IgG1, IgG2a, IgG3, and IgA Abs were from Southern Biotech. Anti-mouse Blimp-1 mAb (3H2-E8) was from Novus Biologicals (Littleton, CO). Data were collected on a FACSCantoII flow cytometer (BD Biosciences, Franklin Lakes, NJ) and analyzed using FlowJo Software (TreeStar, Ashland, OR).

Adoptive transfers of syngeneic spleen B cell populations were as described (22). For some experiments, purified spleen CD19⁺ B cells were first cultured overnight with LPS in complete medium, washed twice, and suspended in sterile PBS prior to i.v. injection through lateral tail veins.

RNA extracted from enriched spleen B cells was used to generate cDNA, with relative transcript levels determined by reverse transcriptase quantitative real-time PCR of triplicate samples, as described (9). Thy1.1 transcripts were amplified using forward (5′-CGTTGGCGCACCAGGAGGAG-3′) and reverse (5′-TGGAGAGGGTGACGCGGGAG-3′) primers. Other primers were as described: gapdh and il10 (9), xbp1 (23), bcl6 (24), and blimp1, irf4, and pax5 (25). Cycle conditions were as follows: one denaturation step of 94°C for 2 min, followed by 40 cycles of 94°C for 30 s, 60°C for 30 s, and 72°C for 1 min. PCR products were controlled for purity by analyses of their melting curves. Expression threshold values (ΔCt) for each transcript were determined by normalizing to gapdh expression within each sample group.

Sera were collected weekly, with Ag-specific Abs quantified by ELISA using DNP-BSA. Serum IgM and IgG levels, autoantibody levels, and 2,4,6-trinitrophenyl (TNP)- or DNP-specific Abs were quantified by ELISA, as described (21, 26). Ab-secreting cell (ASC) frequencies from cell sorter-purified B10 and non-B10 cells were determined using ELISPOT assays, as described (27).

Purified spleen B cells from three mice were stimulated with LPS (10 μg/ml), PMA (50 ng/ml), and ionomycin (1 μg/ml) for 5 h. IL-10–secreting cells were identified using the Mouse IL-10 Secretion Assay Kit (Miltenyi Biotec). Individual IL-10⁺λ⁻CD19⁺ cells were sorted into single wells of 96-well PCR plates using a FACSAria II cell sorter (BD Biosciences). cDNA was synthesized with Ig H and L chain transcripts amplified using nested PCR primers, as described (28). PCR products were purified (QIAquick PCR Purification Kit; Qiagen, Valencia, CA) and cloned (StrataClone PCR Cloning Kit, Agilent Technologies, La Jolla, CA) before sequencing (Duke University DNA Analysis Facility). Productive Ig rearrangements were compared against germline Ig sequences according to the Ig Basic Local Alignment Search Tool database (National Center for Biotechnology Information, Bethesda, MD) and analyzed using the Immunogenetics V-query and Standardization tool (29) to determine V(D)J gene family usage. Mutation frequencies were determined using germline V, D, and J sequences from Ig Basic Local Alignment Search Tool. When L chain sequences obtained from adjacent wells were identical, only one sequence was reported. V_H-D-J_H and V_K-J_K transcript alignments and phylogenetic trees based on average percent identity were constructed using ClustalW2 (30).

Data are shown as means (± SEM). The two-tailed Student t test was used to identify significant differences between sample means.

Spleen GFP⁺ or cytoplasmic IL-10⁺ B cells were not observed in Tiger mice at frequencies significantly above background levels in monensin-treated B cells from wild-type mice (Fig. 1A, 1B). However, GFP⁺ and cytoplasmic IL-10⁺ B cell frequencies increased significantly after ex vivo stimulation using L+PIM for 5 h. GFP⁺ or IL-10⁺ B cells represented between 2 and 3% of spleen B cells in both Tiger and wild-type mice. Furthermore, 72 ± 3% of IL-10⁺ B cells from Tiger mice expressed readily measurable GFP in these assays. Likewise, the majority of GFP⁺ B cells expressed IL-10 (Fig. 1C). In comparison with spleen, significantly fewer IL-10– or GFP-competent B10 cells were found within peripheral or mesenteric lymph nodes after L+PIM stimulation (Fig. 1D). Thus, GFP mimicked cytoplasmic IL-10 expression by most B10 cells during 5-h induction assays.

Agonistic CD40 signals provided during 48-h in vitro cultures rendered B10pro cells competent to express IL-10 when subsequently stimulated with L+PIM. Under these conditions, similar frequencies of cytoplasmic IL-10⁺ (7.3 ± 0.2%) and GFP⁺ (6.3 ± 0.1%) B10+B10pro cells were enumerated (Fig. 1E). By contrast, LPS induced both B10pro cell maturation and B10 cell IL-10 secretion during 48-h assays (17). Under these conditions, the frequency of GFP⁺ B cells (9.3 ± 0.1%) was consistently higher than the frequency of cytoplasmic IL-10⁺ B cells (7.8 ± 0.5%), whereas BCR ligation did not induce B10pro maturation into GFP-competent B cells. Thus, GFP expression was more durable than IL-10 expression following prolonged (48 h) LPS stimulation due to IL-10 secretion and/or relative differences in protein turnover.

A small fraction of spleen CD19⁺ B cells (0.16 ± 0.02%) from 10BiT mice expressed cell surface Thy1.1⁺ ex vivo relative to background staining in wild-type mice (Fig. 2A, 2B). However, significantly increased Thy1.1⁺ (0.9 ± 0.1%, p < 0.01) and IL-10⁺ (1.8 ± 0.4%, p < 0.05) B10 cell frequencies were found after 5 h of L+PIM stimulation. Only 30 ± 2% of IL-10⁺ B cells from 10BiT mice expressed measurable Thy1.1 in these assays, whereas 47 ± 4% of the Thy1.1⁺ B cells expressed IL-10 (Fig. 2A–C). Mesenteric lymph nodes had the highest frequencies of Thy1.1⁺ B cells (2.1 ± 0.2%) when observed directly ex vivo (data not shown), as shown for T cells in mesenteric lymph nodes of 10BiT mice (20). Mesenteric lymph node Thy1.1⁺ B10 cell frequencies were also higher following 5 h of L+PIM stimulation, but the highest numbers of Thy1.1⁺ B cells were in the spleen (Fig. 2D). To determine whether the il10 and thy1.1 genes were transcribed with similar kinetics in 10BiT spleen B cells, their transcripts were measured after in vitro LPS stimulation. Both transcript levels increased congruently in CD1d^hiCD5⁺ B cells and peaked at 24 h relative to CD1d^loCD5⁻ cells (Fig. 2E). Thus, the temporal delay in cell surface Thy1.1 expression relative to cytoplasmic IL-10 was likely due to Thy1.1 processing and cell surface transport during the 5-h assays.

CD40-induced B10pro cell maturation did not induce nascent cell surface Thy1.1 expression or change the kinetics of Thy1.1 expression induced by PMA, ionomycin, and monensin (PIM) stimulation. A normal portion of B cells cultured with CD40 mAb for 48 h expressed cytoplasmic IL-10 after L+PIM stimulation for 5 h, whereas Thy 1.1 expression was only modestly induced (Fig. 2F). However, a higher fraction of 10BiT B cells expressed Thy1.1 than expressed IL-10 after 48-h cultures with LPS plus 5 h of PIM stimulation. Thus, cell surface Thy1.1 expression served as a more durable marker than did IL-10 induction, with a large portion of the B10 cells having terminated IL-10 expression during the 48-h LPS cultures.

To evaluate B10 cell expansion in vivo, wild-type mice were given IFA or CFA, alum, or low-dose LPS; spleen B10 cell numbers were enumerated 3 d later by IL-10 staining after 5 h of monensin or L+PIM treatment. Freund’s adjuvants did not drive B10 cell expansion, whereas B10 cell numbers increased 2–3-fold after alum and LPS treatments (Fig. 3A). When Tiger mice were given LPS, ex vivo IL-10⁺ or GFP⁺ B10 cell frequencies and numbers remained low but expanded 2–4-fold relative to their frequencies in littermates given only PBS (monensin treatment, Fig. 3B). Following 5 h of in vitro L+PIM stimulation, there were 2–3-fold increases in IL-10⁺ or GFP⁺ B10 cell frequencies and numbers relative to control mice, with most B10 cells expressing both IL-10 and GFP. Thus, GFP served as a reliable reporter for IL-10 expression in Tiger mice.

After 3 d of LPS treatment in vivo, Thy1.1⁺ and IL-10⁺ B cell frequencies and numbers in 10BiT mice increased by 4- and 2-fold, respectively (Fig. 3C). However, the higher frequencies and numbers of Thy1.1⁺ B cells relative to IL-10⁺ cells demonstrated that Thy1.1 expression served as a more durable B cell marker than did IL-10 expression, because half of the Thy1.1⁺ B cells had already lost the capacity to express IL-10 following in vitro L+PIM stimulation. Thus, ongoing and terminated IL-10 production in vivo was reported by B cell Thy1.1 expression in 10BiT mice.

After in vivo low-dose LPS treatment for 3 d, the phenotype of spleen IL-10⁺, GFP⁺, or Thy1.1⁺ B cells remained predominantly IgM^hiCD1d^hiCD5⁺CD19^hiCD23^lowCD38^hiB220^hi (Fig. 3D), consistent with the ex vivo phenotype of B10 cells from untreated wild-type mice (8, 10). However, variable frequencies of LPS-induced B10 cells also expressed the CD43- and GL7-activation markers (31), suggesting that LPS drives a subset of the reporter-positive B10 cells toward an ASC phenotype.

Spleen ASCs are predominantly found within the rare CD138^hiB220^int/lo B cell subset (27). However, CD138 staining is lost under the conditions used to visualize cytoplasmic IL-10⁺ cells. Therefore, Tiger and 10BiT mice were used to determine whether in vivo LPS treatment induced B10 cells to differentiate into ASCs. In Tiger mice, GFP⁺ B cells expanded in vivo after LPS treatment but predominantly remained CD138^low (Fig. 4A). Rare GFP⁺ B cells (<2%) were found within the CD138^hiB220^int/lo B cell subset in untreated Tiger mice, with LPS inducing significant numbers of GFP⁺ B cells (16%, p < 0.01) that peaked 1 d after LPS treatment and subsequently declined (Fig. 4B). By contrast, a significant portion of Thy1.1⁺ B cells (17–40%) in 10BiT mice expressed CD138 after 2–3 d of LPS treatment (Fig. 4A). Before receiving LPS, 14% of CD138^hiB220^lo B cells expressed Thy1.1, with almost half of the CD138^hiB220^lo B cells expressing Thy1.1 2 d after LPS treatment (Fig. 4C). Thus, Thy1.1⁺ B cells contributed significantly to the ASC pool following LPS treatment.

Because some pre-B cells, immature B cells, and plasma cells express CD43, GL-7, and CD138 (32), an association between B10 cells and ASCs was more rigorously tested. Thy1.1⁺ B10 cells purified from LPS-treated 10BiT mice spontaneously secreted IgM in ELISPOT assays at 5.5-fold higher frequencies than did Thy1.1⁻ B cells (Fig. 4D). IgG-secreting cells were not detectable within the Thy1.1⁺ or Thy1.1⁻ B cell subset under these conditions (data not shown). Furthermore, Thy1.1⁺ B cells from LPS-treated 10BiT mice expressed transcripts for the plasma cell-associated transcription factors blimp1 (also known as prdm1), xbp1, and irf4 at 2–6-fold higher levels than did Thy1.1⁻ B cells (Fig. 4E). Likewise, pax5 and bcl6 transcripts were markedly reduced in Thy1.1⁺ B cells relative to Thy1.1⁻ B cells, suggesting that reporter-positive B10 cells adopt an ASC or plasma cell fate.

B10 cell Blimp-1 expression was also measured during IL-10 induction. CD1d^hiCD5⁺ B cells (B10 cell-enriched) from wild-type mice expressed significantly higher il10 and blimp1 transcript levels relative to CD1d^loCD5⁻ B cells after 5 h of L+PI stimulation (Fig. 4F). Similarly, CD1d^hiCD5⁺ B cells cultured with CD40 mAb for 48 h expressed significant il10, blimp1, and irf4 transcripts relative to CD1d^loCD5⁻ B cells following 5 h of L+PI stimulation. Independently, blimp1 transcripts were significantly increased in purified IL-10⁺ B10 cells compared with IL-10⁻ B cells after 5 h of L+PI stimulation (Fig. 4G). Measurable B10 cell intracellular Blimp-1 protein expression was confirmed by immunofluorescence staining in comparison with non-B10 cells (Fig. 4H), using described methods (33). Intracellular Blimp-1 expression increased when purified B cells were cultured in the presence of LPS for 24 h, with ∼2-fold higher Blimp-1 levels in IL-10⁺ B cells than in IL-10⁻ B cells (Fig. 4I). Thus, B10 cells expressed Blimp-1 before initiating the ASC-differentiation program.

IL-10 induces human plasma cell differentiation in vitro (34–36). To determine whether autocrine IL-10 drives mouse B10 cell development or differentiation, the 10BiT transgene was bred into an IL-10^−/− background to create 10BiT.IL-10^−/− mice. Spleen Thy1.1⁺ B cell frequencies were identical in both 10BiT and 10BiT.IL-10^−/− mice after in vitro stimulation with agonistic CD40 mAb or LPS for 48 h (Fig. 5A). Identical frequencies of IgM ASCs were also found within the spleen Thy1.1⁺ subsets of 10BiT and 10BiT.IL-10^−/− mice following in vivo LPS treatment (Fig. 5B). ASC frequencies within the spleen CD1d^hiCD5⁺ subset were also equivalent in LPS-treated IL-10^−/− and wild-type mice, with the B10 cell-enriched CD1d^hiCD5⁺ B cells containing a higher frequency of ASCs compared with CD1d^loCD5⁻ B cells. Thus, autocrine IL-10 was not required for either B10 cell development or ASC differentiation.

Although spleen B10 cells are predominantly cell surface IgM^hi (Fig. 3D), B10 cells coexpressing IgG2c, IgG3, and IgA were overrepresented in the B10 cell subset relative to non-B10 cells (Fig. 5C). Therefore, the relative contribution of B10 cells to the ASC pool was assessed using GFP⁺ B10 cells purified from Tiger mice. Spleen B cells were stimulated for 5 h with L+PI to induce GFP expression, sorted into GFP⁺ and GFP⁻ fractions, and cultured overnight with LPS prior to ELISPOT analysis. Consistent with the B10 cell ASC potential demonstrated in 10BiT mice (Fig. 4D), GFP⁺ B10 cells were also a major source of IgM ASCs (Fig. 5D). Thus, a large portion of B cells in both Tiger and 10BiT mice produced IL-10 prior to ASC differentiation.

To determine whether B10 cells produce Ag-specific Ab, Tiger mice were immunized with the T cell-dependent Ag TNP-KLH in alum. Seven days later, spleen B cells were stimulated for 5 h with L+PI to induce GFP expression, with purified GFP⁺ and GFP⁻ cells assessed for anti-TNP IgM and IgG ASC potential. GFP⁺ B cells from both unimmunized and TNP-immunized Tiger mice produced TNP-reactive IgM, indicating that some reactivity was attributable to polyreactive or natural Abs (Fig. 5E). TNP-reactive IgG was only produced by GFP⁺ B cells from immunized mice. Thereby, B10 cells produced both polyreactive IgM and Ag-specific IgM and IgG.

To determine whether B10 cells contribute to serum Ig, equal numbers of spleen GFP⁺ B10 cells or GFP⁻ non-B10 cells were transferred from unimmunized Tiger mice into Rag2^−/− hosts. Serum IgM and IgG were first detected in mice given GFP⁺ cells after 1 and 4 d, respectively, and the levels increased thereafter (Fig. 5F). In mice receiving non-B10 cells, IgM and IgG were detected after 4 and 6 d, respectively. At day 10 posttransfer, serum IgM levels from Rag2^−/− mice that had received GFP⁺ B10 cells were significantly higher than those of untreated control Rag2^−/− mice or Rag2^−/− mice given non-B10 cells. Serum IgG levels in Rag2^−/− recipients given either B10 or non-B10 cells were below the levels found in wild-type mice (Fig. 5G). Rag2^−/− recipients given B10 cells produced IgM, but not IgG, reactive with TNP, further confirming that B10 cells produce polyreactive IgM. Serum IgM from these mice also reacted with nuclear Ags, including ssDNA, dsDNA, and histone proteins. IgM or IgG autoantibodies were not detected in sera from Rag2^−/− mice given non-B10 cells. Thus, B10 cells contributed to the serum IgM and IgG pools, including IgM Abs with autoreactive/polyreactive specificities.

PCR methods were used to obtain an unbiased representation of the IgH and IgL repertoires of single IL-10⁺λ⁻ CD19⁺ cells from wild-type mice. Both H and L chain transcripts revealed the use of diverse V_H and V_K family members (Fig. 6, Tables I, II). V_H1 (J558) was the most frequently observed V_H family, reflecting the predominance of this family within the Ig locus. Germline sequences without mutations encoded 84% of 50 representative V_H-D-J_H sequences and 91% of 69 representative V_K-J_K sequences. Thereby, B10 cells express diverse BCRs that were predominantly germline encoded (Fig. 7).

These results demonstrate that the B10 cell subset regulates inflammatory immune responses through the production of IL-10, as well as maintains a capacity for plasma cell differentiation. Following a transient period of IL-10 production, a significant fraction of B10 cells initiated the genetic and phenotypic program leading to ASC differentiation in vitro and in vivo (Figs. 4, 5). B10 cells produced Ag-specific Abs and represented a significant source of serum IgM and IgG (Figs. 5D–F), as well as contributed polyreactive and autoreactive Ab specificities (Fig. 5G), consistent with the broad diversity of their expressed BCRs (Fig. 6). Hence, B10 cells do not define a distinct B cell lineage committed exclusively to IL-10–dependent immunoregulation. Instead, Ag-specific in vivo signals select B10pro cells, which develop into IL-10–competent B10 cells that secrete IL-10 in response to Ag exposure and/or TLR signaling before plasma cell differentiation (Fig. 7). Thus, B10 cells regulate acute inflammation and immune responses by the transient production of IL-10 and may have the capacity to clear their inducing Ags by producing polyreactive and/or Ag-specific Ab.

The BCR repertoire of spleen B10 cells was remarkably diverse, involving a wide spectrum of V_H, D, and J_H elements, normal frequencies of noncoded nucleotide insertions, as well as considerable complementarity-determining region 3 diversity (Fig. 6, Tables I, II). Regulatory B10 cell BCRs were predominantly germline encoded, with no somatic mutations in most clones. Thereby, spleen B10 cell V_H use was similar to that observed for conventional spleen B cells (37) and did not exhibit the skewed pattern associated with peritoneal cavity B-1a cells (38, 39). Although different selective and/or developmental forces may ultimately shape the regulatory B10 cell BCR repertoire, the current findings demonstrated that IL-10–competent B cells are generated in response to diverse foreign and self-Ags, including a T cell-dependent Ag. Some B10 cells also produced “natural” IgM Ab that was characteristically polyreactive (Fig. 5E, 5G). Consistent with their IgM^hiIgD^lo phenotype (Fig. 3D) and ability to clonally expand rapidly in vitro (12, 17), it is likely that B10 cells contribute substantially to the short-lived plasma cell pool that develops rapidly following Ag encounter. Regulatory B10 cells also develop at normal frequencies in T cell-deficient mice (17), suggesting that many respond to T cell-independent Ags and are unlikely products of germinal center reactions. Germinal center-independent B cell isotype switching may apply to B10 cells, as described (40, 41), although it remains possible that some B10 cells are recruited into germinal centers. It is also unknown whether B10 cells re-enter the memory B cell pool after IL-10 production, because methods are not available to track B10 cells after they lose Thy1.1 expression. Regardless, B10 cell production of diverse Ab products following transient IL-10 production highlights their functional plasticity.

There were significant changes in B10 cell expression of the blimp1, xbp1, irf4, pax5, and bcl6 transcription factors following activation in vivo, which paralleled ASC differentiation (Fig. 4E). Upregulated B10 cell expression of blimp1 and irf4 (Fig. 4F–I) may be of considerable functional significance, because these transcription factors cooperatively induce regulatory T cell differentiation and il10 gene expression (42). The Blimp-1 transcriptional repressor is well known for its role in promoting plasma cell differentiation (43), with IRF4 required for blimp1 expression (44). Blimp-1 may also exert its normal function as a transcriptional repressor and stop IL-10 expression during B10 cell differentiation into ASCs. Identifying the overlapping upregulation of il10, blimp1, and irf4 by B10 cells highlights the potential importance of these transcription factors for regulatory B10 cell function, although other B cells also upregulate blimp1 and irf4 as they differentiate.

Based on their unique phenotypes and ability to proliferate rapidly following mitogenic stimulation, it is likely that mouse and human regulatory B10 cells represent subsets of Ag-experienced B cells (12, 17). Despite high IgM expression by most B10 cells (Fig. 3D), some B10 cells have undergone isotype switching (Fig. 5C). Furthermore, B10 cells do not develop in transgenic mice with fixed AgRs, and genetic alterations that regulate BCR signaling significantly influence B10 cell numbers (17, 45–47). Because only a small subset of B cells have the capacity to produce IL-10 in vivo or in vitro (Figs. 1, 2), and not all ASCs expressed IL-10 before differentiation (Fig. 4A–C), specific in vivo signals must be required to induce IL-10 competence. This may explain why potent BCR ligation alone does not induce B10pro cells to mature into B10 cells in vitro but may instead drive these cells toward different functional programs (Figs. 1E, 2F) (17). Because neither CD40 nor MyD88 expression is absolutely required for B10 cell development in vivo (17), it is likely that these signals and polyclonal mitogens, such as LPS, expand B10pro and B10 cells subsequent to Ag encounter. Consistent with this, murine CMV infection leads to the development of IL-10–expressing CD138^hi B cells by 7 d (48). Salmonella infection also results in the rapid development of IL-10–expressing CD138^hi B cells, which is maximal at day 1 postinfection (49). Thereby, pathways that modify intrinsic BCR signals drive IL-10 competence and B10 cell differentiation (16).

B10 cell Ab production in vitro and in vivo suggests that B10 cells contribute significantly to the serum IgM and IgG Ab pool after transient IL-10 secretion. The spleen marginal zone and B1a cell subsets also contribute significantly to Ab responses. In fact, spleen marginal zone B cells, by virtue of their preactivated state and topographical location, join B1 B cells to generate a wave of IgM-producing plasmablasts during early responses to blood-borne Ags (50, 51). B10 cells also proliferate rapidly following in vitro or in vivo activation (12, 17) and rapidly convert to plasmablasts (Figs. 4, 5). Because the regulatory B10 cell, B1a, and marginal zone B cell subsets share overlapping cell surface markers, it is not possible to ascertain whether individual members of any one of these functionally or phenotypically defined subsets is the primary source of natural, polyreactive, autoreactive, or Ag-specific Ab. Furthermore, spleen B10pro cells are predominantly found within the CD1d^hiCD5⁺ subset of B cells, so it is not possible to remove B cells that have the functional capacity to become IL-10 competent from either the CD5⁺ B1a or the CD1d^hi marginal zone subsets for functional studies. Thus, B1a, marginal zone, and B10 B cells share the capacity to produce Abs in vivo and contribute to early innate- and subsequent adaptive-immune responses.

B10 cell Ab secretion may also contribute to their immunosuppressive functions in vivo. Soluble Abs can quickly reduce Ag load and promote Ag clearance by opsonization or complement-mediated phagocytosis. In addition, bound Ab can directly interfere with Ag recognition by other cell types, effectively reducing the availability of activation signals via Ag neutralization. Autoantibodies can also be important negative regulators of intestinal inflammation and suppress colitis (52, 53). Thus, B10 cells may exhibit two waves of protection that are first IL-10 dependent and, subsequently, Ab dependent. For example, B10 cell IL-10 production inhibits the initial pathology associated with experimental autoimmune encephalomyelitis induction (9, 18), whereas other investigators defined a subsequent wave of B cell-mediated immunosuppression in this model that is both Ag specific and enhanced by CD40 signals (4, 54). Because B10 cells can produce autoantibodies (Fig. 5G), it is possible that their Ab products reduce inflammation and disease through a second wave of Ag clearance. Also, B10 cells primarily produced germline-encoded IgM Abs that are likely to be of low affinity and nonpathogenic, which may be optimally suited to neutralize self-Ags, preempt pathogenic IgG production, and contribute to the suppression of autoimmunity (Fig. 7). Consistent with this, treatment of MRL^lpr mice with unmutated IgM autoantibodies confers protection against lupus nephritis (55). Further characterization of the B10 cell repertoire will be important for understanding both B10 cell development and expansion, particularly during autoimmune disease. Defining the BCR ligands and other signals important for B10 cell expansion and subsequent Ab production may also lead to new therapies for treating both inflammatory and autoimmune conditions.

We thank Drs. Eric Weimer and Garnett Kelsoe for help with the experiments, interpretation of the results, and writing of the manuscript.

T.F.T. is a consultant for MedImmune/AstraZeneca, Inc. and shareholder and consultant for Angelica Therapeutics, Inc. All other authors have no financial conflicts of interest.