Characterization of a rare IL-10–competent B-cell subset in humans that parallels mouse regulatory B10 cells

B cells are generally considered to positively regulate immune responses by producing antigen-specific antibody and helping to induce optimal CD4⁺ T-cell activation.¹  However, B cells and specific B-cell subsets can also negatively regulate immune responses in mice.^2-6  The absence or loss of these regulatory B cells exacerbates disease symptoms in contact hypersensitivity, experimental autoimmune encephalomyelitis, chronic colitis, collagen-induced arthritis, and lupus-like models of autoimmunity.^7-15  In many of these cases, B cells regulate inflammation, asthma, and T cell–mediated autoimmunity through the production of interleukin-10 (IL-10).^8-10,12-16  Both human and mouse IL-10 exhibit numerous pleiotrophic activities in vitro and in vivo, including suppression of both Th1 and Th2 polarization and inhibition of antigen presentation and proinflammatory cytokine production by dendritic cells, monocytes, and macrophages.¹⁷ 

In mice, a subset of IL-10–competent regulatory B cells can be functionally identified by their ability to express cytoplasmic IL-10 after 5 hours of in vitro stimulation with lipopolysaccharide (LPS), phorbol myristate acetate (PMA), and ionomycin, with monensin included in the cultures to block IL-10 secretion.^12,13  These IL-10–competent B cells have been labeled as B10 cells to identify them as the predominant, if not exclusive, source of B-cell IL-10 production and to distinguish them from other regulatory B-cell subsets that may also exist.⁵  For example, inducible IL-12–producing B cells regulate intestinal inflammation.¹⁸  B10 cells are found within the spleens of naive wild-type mice at frequencies of 1% to 3%, where they predominantly represent a subset of the phenotypically unique CD1d^hiCD5⁺CD19^hi B-cell subpopulation that shares overlapping cell surface markers with multiple phenotypically defined B-cell subsets.^11-14,19,20  Additional B cells within the CD1d^hiCD5⁺ B-cell subpopulation acquire the ability to function like B10 cells during 48 hours of in vitro stimulation with LPS or agonistic CD40 monoclonal antibody (mAb).⁵  These B10 progenitor (B10pro) cells are then able to express cytoplasmic IL-10 after stimulation with PMA, ionomycin, and monensin for 5 hours.²¹  B10 cells also require diverse B-cell antigen receptors for their development,²¹  and their regulatory functions are Ag-restricted in vivo.^12,13  Spleen B10-cell numbers increase significantly in diabetes- and lupus-prone mice,^14,21  and the adoptive transfer of antigen-primed CD1d^hiCD5⁺ B cells reduces inflammation during contact hypersensitivity and autoimmune disease.^12,13,22 

The identification and characterization of an IL-10–producing B-cell subset in mice raise the issue of whether B cells with these functional properties exist in humans. Studies of B-cell IL-10 production in humans have yielded diverse results that are currently difficult to unify into a coherent model.^23-28  It is also unknown whether human B10 cells share overlapping physiologic triggers with mouse B10 cells that lead to IL-10 production and their expansion in vitro.^12,13,21  Therefore, the purpose of the current study was to enumerate and characterize the IL-10 competent B10 and B10pro cell subsets in humans.

Heparinized blood was obtained from healthy donors (age, 14-73 years) or from patients. Patients with rheumatoid arthritis met the American College of Rheumatology 1987 revised classification criteria.²⁹  Patients with systemic lupus erythematosus satisfied the 1982 classification criteria.³⁰  Patients with primary Sjögren syndrome fulfilled the American-European consensus group criteria.³¹  Patients with autoimmune vesiculobullous skin disease, including bullous pemphigoid, pemphigus foliaceus, pemphigus vulgaris, and dermatitis herpetiformis, had typical clinical and histologic findings, with diagnostic findings on direct immunofluorescence of perilesional skin or oral mucosa.^32,33  Patients with multiple sclerosis fulfilled the 2005 McDonald criteria for relapsing remitting or primary progressive multiple sclerosis³⁴  or secondary progressive multiple sclerosis as defined using the Lublin and Reingold criteria.³⁵  Tissues were obtained anonymously from persons without identifiable hematologic disorders, with the purified B cells immediately cryopreserved (> 90% cell viability). Cryopreserved cord blood samples were obtained from the Duke University Stem Cell Laboratory and the Carolinas Cord Blood Bank.

Blood mononuclear cells were isolated by centrifugation over a discontinuous Lymphoprep (Axis-Shield PoC As) gradient. Cell numbers were quantified by hemocytometer, with relative lymphocyte percentages among viable cells (based on scatter properties) determined by flow cytometry. In some experiments, B cells were enriched using RosetteSep (StemCell Technologies) following the manufacturer's protocols. CD19-mAb-coated microbeads (Miltenyi Biotec) were used to purify blood B cells by positive selection following the manufacturer's instructions. When necessary, the cells were enriched a second time using a fresh magnetic-activated cell sorting column to obtain more than 99% purities.

Written informed consent was obtained in all cases according to the Declaration of Helsinki. All protocols were approved by the respective ethics boards: Institutional Review Board of Duke University Medical Center, the Human Protection Committee of Dana-Farber Cancer Institute/Harvard Medical School, University of Rochester Institutional Review Board, and the Institutional Review Board of St Luke's-Roosevelt Hospital Center Institute for Health Sciences.

Antihuman mAbs included: IgD (IA6–2) from BD Biosciences; CD21 (BU33), CD22 (RFB4), CD23 (D.6) from Ancell; IgM (MHM-88), CD1d (51.1), CD5 (UCHT2), CD19 (HIB19), CD24 (ML5), CD25 (BC96), CD27 (O323), CD38 (HIT2), CD40 (HB14), CD48 (BJ40), and CD148 (A3), tumor necrosis factor-α (TNF-α; MAb11), functional grade CD3 (HIT3a), and phycoerythrin-conjugated or unconjugated anti–IL-10 (JES3–19F1, JES3–9D7) mAbs from BioLegend. Antihuman IgM Ab was from Jackson ImmunoResearch Laboratories.

Single-cell suspensions were stained on ice using predetermined optimal concentrations of each antibody for 20 to 60 minutes and fixed as described.³⁶  Cells with the light scatter properties of lymphocytes were analyzed by 2–6 color immunofluorescence staining and FACScan, FACSCalibur, or FACSCanto II flow cytometers (BD Biosciences). Dead cells were excluded from the analysis based on their forward- and side-light scatter properties and the use of LIVE/DEAD Fixable Violet Dead Cell Stain Kits (Invitrogen). All histograms are shown on a 4-decade logarithmic scale, with gates shown to indicate background isotype-matched control mAb staining set with less than 2% of the cells being positive. Blood CD24^hiCD27⁺ and CD24^lowCD27⁻ B cells were isolated using a FACSVantage SE flow cytometer (BD Biosciences) with 90% to 95% purities.

Intracellular IL-10 analysis by flow cytometry was as described.¹²  Briefly, cells were resuspended (2 × 10⁶ cells/mL) in medium (RPMI 1640 media containing 10% fetal calf serum, 200 μg/mL penicillin, 200 U/mL streptomycin, and 4mM L-glutamine [all from Invitrogen]) and stimulated with LPS (10 μg/mL, Escherichia coli serotype 0111: B4; Sigma-Aldrich), CpG (ODN 2006, 10 μg/mL; Invivogen), or other Toll-like receptor (TLR) agonists (TLR1, Pam3CSK4, 1 μg/mL; TLR2, heat-killed Listeria monocytogenes, 10⁸ cells/mL; TLR3, polyriboinosinic acid/polyribocytidylic acid, 10 μg/mL; TLR5, S typhimurium flagellin, 1 μg/mL; TLR6, Pam2CGDPKHPKSF, 1 μg/mL; TLR7, imiquimod, 1 μg/mL; TLR8, ssRNA40, 1 μg/mL; Invivogen), CD40L (1 μg/mL; R&D Systems), antihuman CD40 mAb (1 μg/mL; BioLegend), PMA (50 ng/mL; Sigma-Aldrich), ionomycin (1 μg/mL; Sigma-Aldrich), brefeldin A (BFA, 1 × solution/mL; BioLegend), monensin (2mM; eBioscience), and antiIgM antibody (10 μg/mL) as indicated in 48-well flat-bottom plates before staining and flow cytometry analysis. For analysis of cell proliferation, lymphocytes were stained with carboxyfluorescein succinimidyl ester (CFSE) Vybrant CFDA SE fluorescent dye (5 μM; Invitrogen) according to the manufacturer's instructions. For IL-10 detection, Fc receptors were blocked using FcγR-binding inhibitor (eBioscience). Stained cells were fixed and permeabilized using a Cytofix/Cytoperm kit (BD Biosciences) according to the manufacturer's instructions and stained with anti–IL-10 mAb.

Secreted IL-10 was quantified by enzyme-linked immunosorbent assay (ELISA). Purified B cells (4 × 10⁵) were cultured in 0.2 mL of complete medium in a 96-well flat-bottom tissue culture plates. Culture supernatant fluid IL-10 concentrations for triplicate samples were quantified using IL-10 OptEIA ELISA kits (BD Biosciences) following the manufacturer's protocols.

In some experiments, IL-10-secreting blood B cells were identified after 4 hours of in vitro stimulation using an IL-10 secretion detection kit (Miltenyi Biotec) with subsequent staining for CD19 expression before cell sorting into IL-10⁺CD19⁺ and IL-10⁻CD19⁺ populations. RNA extraction, cDNA generation, and real-time PCR analysis were as described.¹³  IL-10 primers were: sense 5′-CTTCGAGATC TCCGAGATGC CTTC-3′, antisense 5′-ATTCTTCACC TGCTCCACGG CCTT-3′. GAPDH primers were: sense 5′-GCCACCCAGA AGACTGTGGA TGGC-3′ and antisense 5′-CATGTAGGCC ATGAGGTCCA CCAC-3′.

Blood CD24^hiCD27⁺CD19⁺ and CD24^lowCD27⁻CD19⁺ B cells from healthy controls were purified by cell sorting and stimulated with CpG (10 μg/mL) plus CD40L (1 μg/mL) for 24 hours. Naive CD4⁺CD25⁻ T cells were purified by magnetic-activated cell sorting (Miltenyi Biotec) and cultured in replicate wells either alone (1 × 10⁶/mL) or with CD40/CpG-stimulated B cells (1 × 10⁶/mL) in tissue culture plates coated with CD3 mAb (0.5 μg/mL) for 72 hours with either BFA or phorbol ester and ionomycin plus BFA (PIB) added during the final 5 hours. The cells were stained for cell surface CD4 and cytoplasmic TNF-α expression before analysis by flow cytometry. Alternatively, CD40/CpG-stimulated sorted B-cell populations (1 × 10⁶/mL) were cultured for 20 hours with equal numbers of blood monocytes that had been purified by plastic adherence or magnetic-activated cell sorting, with anti–IL-10 mAb (10 μg/mL) added to some cultures. The cultured cells were then washed, stimulated with LPS (1 μg/mL) for 4 hours in the presence of BFA, and stained for cell surface CD14 and cytoplasmic TNF-α or interferon-γ expression before analysis by flow cytometry.

All data are shown as mean ± SEM. Significant differences between sample means were determined using the Student t test.

Mouse B10-cell frequencies have been described, with background cytoplasmic IL-10 staining of less than or equal to 0.2% commonly observed with B cells from IL-10^−/− mice.^12,13,21  A similar strategy was therefore optimized to enumerate human blood IL-10–competent B cells. B cells spontaneously expressing IL-10 were below the threshold for reliable quantification by immunofluorescence staining, even when the cells were cultured with BFA to stop Golgi transport (Figure 1A). However, a distinct subset of cytoplasmic IL-10⁺ B cells was observed at low 0.25% to 2% frequencies after ex vivo stimulation using PIB (Figure 1A-B). Stimulation with PIB for 5 hours induced 0.8% ± 0.1% of blood B cells on average to express IL-10 (n = 14, 1.9 ± 0.3 × 10⁻³ B10 cells/mL, Figure 1B-C). B-cell stimulation with TLR agonists did not substantially alter mean B10-cell numbers, although adding either CpG oligonucleotides (CpG, TLR9 agonist) or LPS to the PIB cultures increased IL-10⁺ B-cell frequencies in some persons (Figure 1A-C). Stimulation beyond 5 hours or using 10-fold higher PMA or ionomycin concentrations resulted in extensive B-cell death, which complicated B10-cell enumeration. Background IL-10 mAb staining was also reduced by the exclusion of cell doublets and dead cells from the analysis. BFA was also used to block IL-10 secretion rather than monensin because it optimized human B-cell cytoplasmic IL-10 expression (Figure 1A). B cells cultured in BFA served as negative controls because they gave results similar to isotype control mAb staining. Thus, blood B10 cells were rare but readily quantified in healthy humans.

In mice, B10pro cell maturation into IL-10-competent B10 cells is induced by 48 hours of stimulation with either LPS or agonistic CD40 mAb.²¹  Human blood B10pro cells capable of maturing into IL-10–competent cells after in vitro culture were also identified. The total frequency of B10 and B10pro cells (B10 + B10pro) is quantified in this assay, as the B cells that acquire IL-10 competence in vitro (eg, matured B10pro cells) cannot be differentiated from preexisting B10 cells that inherently express cytoplasmic IL-10 after 5 hours of PIB stimulation. Culturing human B cells in media alone for 48 hours resulted in approximately 0.2% of B cells expressing cytoplasmic IL-10 after PIB stimulation during the last 5 hours of culture (Figure 1B). Adding LPS, CpG, or recombinant CD40 ligand (CD40L, CD154) alone, together, or in combination with BFA to the 48-hour cultures did not increase IL-10-producing B10-cell frequencies. However, after PIB stimulation during the last 5 hours of culture, B10 + B10pro-cell frequencies increased to 0.6% ± 0.1%, 1.9% ± 0.4%, 0.8% ± 0.1%, 1.2% ± 0.2%, and 4.1% ± 1.0% after 48 hours of TLR1 agonist, LPS, TLR6 agonist, TLR7 agonist, or CpG stimulation, respectively (Figure 1E left panel). Compared with media or most TLR agonists alone, the addition of CD40L to the cultures significantly enhanced mean B10 + B10pro-cell frequencies (Figure 1E right panel). CD40L induced higher IL-10⁺ B-cell frequencies than agonistic CD40 mAb (Figure 1D). Thus, dual CD40 and TLR stimulation induced the highest frequencies of B10pro cells to become IL-10–competent B10 cells, with the highest frequencies (7.0% ± 1.4%) and numbers of IL-10⁺ cells (1.6 ± 0.3 × 10⁴ cells/mL, n = 14) induced after 48 hours of CD40L plus CpG stimulation. B10-cell percentages relative to all B cells decreased with age among 60 healthy persons, with a significant inverse correlation between B10 + B10pro-cell frequencies and age (P < .05, data not shown). Thus, B10pro cell maturation in vitro optimized IL-10–competent B-cell enumeration.

Newborn blood contained both B10 and B10pro cells. Mean B10-cell frequencies in cord blood after 5 hours of CpG + PIB stimulation were 0.45% ± 0.14% (Figure 2A). B10 + B10pro-cell frequencies were similar or higher in cord blood relative to adult blood after culture with CD40L and TLR agonists: TLR1 (2.6% ± 0.6%), LPS (7.6% ± 1.8%), TLR6 (4.2% ± 1.4%), or TLR9 (CpG, 9.6 ± 2.3%) agonists with PIB added during the final 5 hours of culture. B10 cells were also found within spleens (0.31% ± 0.06%, n = 4, CpG + PIB) and tonsils (0.31% ± 0.11%, n = 3, CpG + PIB) of adults without known disease (Figure 2B). Thus, newborn and adult blood and tissues contain quantifiable numbers of B10 and B10pro cells.

The time course of blood B-cell IL-10 induction was assessed in vitro by quantifying IL-10 transcripts. By 12, 24, and 48 hours, B-cell stimulation with CD40L plus CpG induced 6.8-, 24-, and 5.9-fold higher IL10 transcript levels, respectively, than was observed for unstimulated B cells (P < .05; Figure 3A). Blood B cells that were actively secreting IL-10 after in vitro stimulation expressed IL10 transcripts at 19-fold higher levels than were detected in IL-10⁻ B cells (Figure 3B). Thus, IL10 gene transcription parallels the induction of B10-cell cytoplasmic IL-10 expression.

The response of human B10 + B10pro cells to CD40L, CpG, and antigen receptor-generated signals was also examined. Compared with CD40L alone, CpG induced the highest levels of B10pro cell maturation into IL-10–competent B10 cells, which was further increased when both CD40L and CpG were added to the cultures (Figure 3C). By contrast, IgM ligation using mitogenic antibody did not induce cytoplasmic IL-10 expression but actually inhibited the B10pro cell maturation effects of CpG + CD40L stimulation. In vitro IgM signals also inhibit mouse B10pro cell maturation and cytoplasmic IL-10 induction.²¹  Among TLR agonists, LPS and CpG were also the most potent stimuli for inducing IL-10 secretion by human blood B cells (Figure 3D). The addition of CD40L to the cultures was not required but enhanced LPS- and CpG-induced IL-10 secretion. Thus, similar signals induce human and mouse B10pro and B10 cells to mature and express cytoplasmic IL-10 in vitro.

Whether human B10 cells represent a unique or known B-cell subset was determined by analyzing their cell surface phenotype. Cell surface IgM, IgD, CD1d, CD5, CD10, CD19, CD21, CD22, CD23, CD24, CD25, CD27, CD38, and CD40 densities did not change when B cells were stimulated with PIB, LPS + PIB, or CpG + PIB for 5 hours and/or permeabilized. Furthermore, the transport of newly synthesized proteins to the cell surface is inhibited by the addition of BFA to the cultures. Therefore, these markers were used to phenotype freshly isolated B10 cells. Half of blood B10 cells expressed high IgM levels and low IgD levels (Figure 4A). Both CD24 and CD27 expression was high on the majority of B10 cells, whereas IL-10⁻ B cells expressed either high or low density CD24 and CD27. CD19 and CD25 expression was also higher on B10 cells than IL-10⁻ B cells. Otherwise, the remaining cell surface markers were absent or expressed similarly by both B10 cells and IL-10⁻ B cells. The same results were obtained after PIB, LPS + PIB, or CpG + PIB stimulation (data not shown). Thereby, freshly isolated blood IL-10⁺ B10 cells were predominantly found within the CD24^hi or CD27⁺ B-cell subpopulations.

The phenotype of blood B10 + B10pro cells was also assessed after 48 hours of culture with CD40L + CpG and 5 hours of LPS + PIB stimulation. Prolonged B-cell stimulation induced significant changes in the cell surface phenotype of both IL-10⁺ and IL-10⁻ B cells, with most B cells induced to express CD25 and CD38 (Figure 4B). Nonetheless, B10 + B10pro cells on average expressed higher densities of CD1d, CD19, CD20, CD21, CD23, CD24, CD25, CD27, and CD38 compared with IL-10⁻ B cells, consistent with an activated phenotype. Spleen B10 cells were also predominantly CD27⁺, although the expression of most cell surface molecules was similar if not identical for B10 cells and IL-10⁻ B cells (Figure 4C). Spleen B10 + B10pro cells and IL-10⁻ B cells also had similar phenotypes after 48 hours of stimulation in vitro (Figure 4D). Thereby, blood B10 cells represented a subset of the CD24^hiCD27⁺ subpopulation, whereas IL-10 expression remained the best marker for categorizing B10 cells and B10 + B10pro cells.

In addition to being predominantly CD24^hi and CD27⁺, most B10 cells also expressed additional cell surface markers of activation (CD48^hi) and memory (CD148^hi) that were not affected by the 5-hour culture conditions used to induce B-cell cytoplasmic IL-10 expression (Figure 5A-B). Even when the spectrum of blood donors was compared, B10 cells were always predominantly CD24^hi and CD27⁺ (Figure 5C). Cell surface IgD, CD27, and CD38 expression profiles have also been used frequently to define human B-cell subsets.^37,38  However, when blood B10 cells were analyzed based on their CD38 versus IgD, or CD27 versus IgD expression profiles, IL-10⁺ B10 cells from representative blood donors did not fall into distinct subpopulations (Figure 5C).

Because the CD24^hiCD27⁺ B-cell subpopulation represented 24% ± 5% (n = 7) of blood B cells on average, it was assessed whether CD24 and CD27 could be used as overlapping markers to enrich for B10 cells. When purified CD24^hiCD27⁺ and CD24^lowCD27⁻ B cells were cultured individually with LPS + PIB for 5 hours to induce IL-10 expression, B10-cell frequencies were more than 10-fold higher within the CD24^hiCD27⁺ subpopulation compared with CD24^lowCD27⁻ B cells (Figure 5D). To determine whether ex vivo B10pro cells also localize within the CD24^hiCD27⁺ subpopulation, purified CD24^hiCD27⁺ and CD24^lowCD27⁻ B cells were cultured individually to induce B10pro cell maturation. Again, the frequency of IL-10⁺ B10 + B10pro cells was more than 10-fold higher within the CD24^hiCD27⁺ subpopulation compared with CD24^lowCD27⁻ cells (Figure 5E). The capacity of freshly isolated CD24^hiCD27⁺ and CD24^lowCD27⁻ B cells to secrete IL-10 was also assessed. Again, IL-10 was predominantly produced by the CD24^hiCD27⁺ B-cell subpopulation in response to CD40L, LPS, LPS + CD40L, CpG, or CpG + CD40L stimulation for 72 hours (Figure 5F). Thus, blood B10 and B10pro cells predominantly represented a small subset of cells within the CD24^hiCD27⁺ B-cell subpopulation.

Because B10 and B10pro cells express markers for activated and memory B cells, their ability to proliferate in response to mitogens was assessed. Purified blood B cells were labeled with CFSE before in vitro culture. There was little, if any, IL-10⁺ or IL-10⁻ B-cell proliferation during 48-hour cultures regardless of whether the cells were stimulated with LPS or CpG. However, IL-10⁺ B cells proliferated significantly in response to CpG at 96 hours, whereas IL-10⁻ B-cell proliferation was modest (Figure 5G). Thus, blood B10 and B10 + B10pro cells preferentially proliferated in response to mitogen stimulation, consistent with their CD24^hiCD27⁺ phenotype.

Resting or activated mouse B10 cells do not significantly influence mitogen-driven T-cell activation or proliferation in vitro.²²  However, activated B10 cells do inhibit TNF-α production by antigen-specific CD4⁺ T cells during antigen-driven in vitro assays. In vitro T-cell–B-cell coculture systems were therefore assessed to determine whether human B10 cells regulate mitogen-driven T-cell function. First, purified blood CD24^hiCD27⁺ B cells or CD24^lowCD27⁻ B cells were stimulated with CD40L and CpG for 24 hours, washed extensively, and added to cultures of purified CD4⁺ T cells that were stimulated with plate-bound CD3 mAb for 72 hours. In these assays, both CD24^hiCD27⁺ B cells and CD24^lowCD27⁻ B cells inhibited CD4⁺ T-cell expression of TNF-α equally (Figure 6A). Unstimulated B cells did not affect CD4⁺ T-cell expression of TNF-α in these assays, and the addition of anti–IL-10 mAb to cultures containing activated CD24^hiCD27⁺ or CD24^lowCD27⁻ B cells did not rescue TNF-α expression by CD4⁺ T cells (data not shown). Thus, activated human B cells can inhibit mitogen-induced TNF-α production by CD4⁺ T cells through IL-10-independent pathways that are not unique to CD24^hiCD27⁺ B cells.

Even though mouse B10-cell regulation of T-cell function may require antigen-specific in vitro assays that are not amenable to human studies, mouse monocyte cytokine production is directly regulated by B10 cells through IL-10-dependent pathways that are not antigen-specific (M.H. et al, manuscript in preparation). The ability of human B10 cells to regulate innate monocyte responses was therefore assessed. Remarkably, TNF-α production was significantly reduced when monocytes were cultured with activated blood CD24^hiCD27⁺ B cells, and this effect was blocked by anti–IL-10 mAb (Figure 6B). The addition of anti–IL-10 mAb to monocytes cultured alone did not affect TNF-α production (data not shown), whereas CD24^lowCD27⁻ B cells variably affected monocyte TNF-α production through IL-10–independent pathways. Thus, IL-10 production by CD24^hiCD27⁺ B cells regulates monocyte TNF-α production.

To determine whether blood B10-cell numbers are altered during inflammation and autoimmune disease, B10 and B10pro cells were examined in 91 patients with systemic lupus erythematosus, rheumatoid arthritis, primary Sjögren syndrome, autoimmune vesiculobullous skin disease, or multiple sclerosis (Figure 7A-B). Most patients were undergoing active treatment with immunomodulatory agents and/or low doses of prednisone (Table 1). B10-cell frequencies were significantly higher in one pemphigus vulgaris patient (PV07) and 1 dermatitis herpetiformis patient (DH02) not undergoing immunosuppressive therapy. Two lupus patients (SLE02, SLE06) and 1 rheumatoid arthritis patient (RA03) also had significantly higher B10-cell frequencies, but retrospective evaluation of their disease status, autoantibody profile, and treatment regimen did not indicate why these persons had higher B10-cell frequencies. No patients expressed significantly lower B10-cell frequencies than age-matched controls. Mean B10-cell frequencies were significantly higher for rheumatoid arthritis patients after culture with CpG but not for other patient groups (Figure 7B), even though B10-cell frequencies increased similarly in most cases after either LPS or CpG stimulation (Figure 7C left panel).

Mean B10 + B10pro-cell frequencies from patients with autoimmune disease were significantly higher than controls after either CD40L + LPS or CD40L + CpG stimulation (Figure 7A-B). Multiple patients had significantly higher B10 + B10pro-cell frequencies, including 2 patients not undergoing therapy (SLE04, PF03; Table 1). No patients expressed significantly lower B10 + B10pro-cell frequencies relative to age-matched controls. B10 + B10pro-cell frequencies increased after either LPS or CPG stimulation, but the scatter of the results was broad, suggesting inherently different patient sensitivities to LPS and CpG stimulation (Figure 7C right panel). Patients with high blood B10-cell frequencies did not necessarily have high B10 + B10pro-cell frequencies after either LPS or CpG stimulation (data not shown). Likewise, B10 or B10 + B10pro-cell frequencies did not correlate with CD27⁺ B-cell frequencies (data not shown). Relative B10 and B10 + B10pro-cell frequencies did correlate with the intensity of cytoplasmic IL-10 expression, but only one patient generated significantly higher (P < .05) cytoplasmic IL-10 expression levels on a per-cell basis relative to controls and other patients (Figure 7D). Thus, blood B10 and B10pro cell numbers were not decreased in patients with systemic or organ-specific autoimmune disease compared with healthy controls but were significantly increased in some patients.

These studies demonstrate the existence of human IL-10–competent B10 cells, which were readily identified by their ability to express cytoplasmic IL-10 after in vitro stimulation for 5 hours (Figure 1). Moreover, IL-10 production by human B10 cells regulated cytokine production by monocytes in vitro, demonstrating a functional link between regulatory B cells and the innate immune system (Figure 6B). Peripheral blood B10-cell frequencies were characteristically low in most persons, consistent with their low frequencies in mice. Human B10pro cells were also identified at low frequencies by their ability to express IL-10 after in vitro maturation during 48-hour cultures. Remarkably, the adaptive and innate activation pathways that induced human B10 and B10pro cell generation, maturation, cytoplasmic IL-10 expression, and IL-10 secretion were similar to those used to characterize mouse regulatory B10 cells. Specifically, human B10 cells responded to phorbol ester/ionomycin, LPS, and CpG stimulation, with B10pro cell maturation in response to CD40-, LPS-, and CpG-induced signals. Previous studies of IL-10 production by human B cells have predominantly studied bulk B-cell populations using stimulation and assay conditions that were not optimized for quantifying or characterizing individual B cells that were competent to express IL-10.^25,26,39-43  Nonetheless, the results herein demonstrate that rare B10 and B10pro cells that are competent to express IL-10 exist in human blood and can be quantified in vitro.

IL-10–competence remains the best phenotypic marker for defining human B10 cells. However, freshly isolated blood B10 and B10pro cells were also predominantly CD24^hiCD27⁺, with approximately 60% also expressing CD38 (Figure 5A,C). Others have found similar total numbers of IL-10⁺ B cells in the CD24^hiCD38^hi and CD24^intCD38^int B-cell fractions,²⁸  in agreement with the current findings (Figure 5C). B10 cells also expressed CD48 and CD148 at high levels (Figure 5A). CD48 is up-regulated on activated B cells,⁴⁴  and CD148 is considered a marker for human memory B cells.⁴⁵  CD27 expression is also a well-characterized marker for memory B cells, although some memory B cells may be CD27⁻.^38,46,47  The CD27⁺ B-cell subpopulation can also expand during the course of autoimmunity and may serve as a marker for disease activity.^38,47  However, B10-cell frequencies did not parallel the size of the blood CD27⁺ memory B-cell pool in normal donors or patients (data not shown). Thus, the CD24^hiCD27⁺ phenotype of B10 and B10pro cells may 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 blood B10 cells in response to mitogen stimulation was higher than that for other B cells (Figure 5F), as is seen for mouse B10 cells.²¹  Moreover, IL-10 was also predominantly secreted by ex vivo CD24^hiCD27⁺ B cells (Figure 5G). However, most CD24^lowCD27⁻ B cells were not IL-10 competent, even after 48 hours of LPS or CpG stimulation along with CD40 ligation. Human transitional B cells are also rare (2%-3% of B cells) in adult blood and are generally CD10⁺CD24^hiCD38^hi cells that are also IgD⁺CD27⁻.^48,49  Given that CD10 expression is a well-accepted marker for most cells within the transitional B-cell pool,⁵⁰  its absence on B10 cells suggests that these cells are not recent emigrants from the bone marrow. Thereby, in addition to inherent IL-10 competence, B10 cells demonstrate elevated proliferative responses that may reflect prior antigen stimulation.

Because IL-10 is critical for B cell–regulatory activity in mice, the current studies demonstrate that B10 cells were functionally competent to express IL-10 in healthy persons and 91 autoimmune disease patients (Figure 7A-B). Blood B10-cell frequencies in most patients with lupus, rheumatoid arthritis, Sjögren syndrome, autoimmune skin disease, and multiple sclerosis were not significantly different from those observed in healthy controls, although mean B10 + B10pro-cell frequencies were significantly increased. Multiple patients also had significantly higher B10 and/or B10 + B10pro-cell frequencies, including systemic lupus and pemphigus patients with either untreated or severe disease (Table 1). Consistent with this, IL-10 production by blood B cells is reported to be higher in patients with rheumatoid arthritis, lupus, and systemic sclerosis.^23,24,26  Moreover, elevated B10/B10pro-cell frequencies in humans parallel what has been found during inflammation^12,13  and autoimmunity in mice.^14,21  Although patient cohorts with recent-onset disease and clinically active disease across multiple organ systems will be needed to fully assess the relationship of blood B10-cell numbers with clinical, laboratory, and treatment status, none of the patients or patient groups had significantly lower blood B10-cell numbers than age-matched healthy controls. Because multiple sclerosis and lupus patient's B cells are reported to produce decreased amounts of IL-10,^27,28  the current studies demonstrate that a careful enumeration of IL-10–competent B10-cell frequencies within patients will be required to interpret experimental results, particularly when mixed populations of B cells are assayed functionally. Blair et al have also functionally characterized human blood CD24^hiCD38^hi B cells stimulated in vitro with CD40L expressing CHO cells and shown that they reduce CD4⁺ T-cell expression of interferon-γ and TNF-α after T-cell stimulation with CD3 mAb.²⁸  By contrast, we find that both activated CD24^hiCD27⁺ and CD24^lowCD27⁻ B cells reduce TNF-α expression after CD4⁺ T-cell stimulation with CD3 mAb through IL-10–independent pathways (Figure 6A). Nonetheless, we have found that mouse spleen CD1d^hiCD5⁺ B cells can inhibit CD4⁺ T-cell interferon-γ and TNF-α expression, which is completely dependent on IL-10 expression, but this requires B10-cell activation and is only observed with antigen-specific T-cell activation.²²  Furthermore, mouse spleen CD1d^hiCD5⁺ B cells regulate the antigen-presenting capacity of dendritic cells in vitro and can thereby also regulate CD4⁺ T-cell activation indirectly.²²  Moreover, the ability of human B10 cells to influence innate monocyte function (Figure 6B) expands their regulatory role during immunity and disease. Thus, it is probable that IL-10 produced by human B10 cells will have pleiotrophic regulatory effects on the immune system, as occurs in mice.

In conclusion, the current findings demonstrate the existence of a small but significant subset of CD24^hiCD27⁺ B cells that is preprogrammed in vivo to express IL-10 after ex vivo maturation/stimulation. Monitoring the numbers and ability of individual B10 cells to produce IL-10 will become even more informative once the in vivo physiologic triggers of B10 cell-regulatory activity are identified. The identification of antigen-specific B10 cells may also facilitate a further understanding of their relevance to immune responsiveness because antigen receptor specificity is important for mouse B10-cell development and in vivo functional activity in the regulation of inflammation and autoimmunity.^8,12,13 

The authors thank Robert Streilein, Edna Scarlett, and Yen-Yu Lin for help with patient procurement and analysis, Jacquie Bryant for statistical analysis, and Dr Max Cooper for helpful advice.

This work was supported by the National Institutes of Health (AI56363) and the Southeastern Regional Center of Excellence for Emerging Infections and Biodefense (grant U54 AI057157, T.F.T.; grants CA122645 and CA130805, S.H.B.; and grants CA132110 and HL091749, P.M.S.).

National Institutes of Health

Contribution: Y.I., T.M., M.H., D.J.D., K.Y., G.M.V., and T.F.T. made substantial contributions to conception and design of the study and acquisition of the data, and drafted the article; P.M.S., S.H.B., C.M.M., A.D.W., R.P.H., and E.W.S.C. provided critical patient materials, information, and input into the design of the study; and all authors reviewed the data and manuscript critically for intellectual content, analysis, and interpretation of the data, and have given final approval of the version to be published.

Conflict-of-interest disclosure: T.F.T. is a paid consultant for MedImmune Inc and a paid consultant and shareholder for Angelica Therapeutics Inc. The remaining authors declare no competing financial interests.

Correspondence: Thomas F. Tedder, Department of Immunology, Box 3010, Duke University Medical Center, Durham, NC 27710; e-mail: thomas.tedder@duke.edu.