Distinct Profiles of Human B Cell Effector Cytokines: A Role in Immune Regulation?¹

Strict control of B cell activation is essential to permit the rapid production of high affinity protective Abs while preventing autoreactivity. The early cognate interaction between independently activated Ag-specific T and B cells is pivotal in regulating the B cell response in both health and disease states (1). For T-dependent Ags, the consensus model of normal B cell activation involves two stages that follow rapidly, occurring within 12 h in the mouse (2). The first step is typically triggered in the B cell areas of secondary lymphoid structures by the binding of specific Ag to the B cell receptor (BCR).³This primes the B cell for migration and subsequent T cell encounter within the T cell zone. BCR signaling alone induces limited proliferation, but T cell help, principally through CD40/CD40-ligand (CD40L) interactions, triggers vigorous extrafollicular proliferation, leading to germinal center (GC) formation and, ultimately, differentiation into memory and plasma cells (1, 2). This paradigm ensures specificity: B cells not receiving specific BCR stimulation before meeting CD40L (an event not requiring cognate interaction) are rendered subject to apoptosis by up-regulation of fas (3, 4).

The outcome of T/B cell interactions is importantly influenced by the cytokine milieu (5), which is traditionally considered to be dominated by T lymphokines, the polarization of which is predetermined during the T/dendritic cell interaction. The ability of different T cell cytokines to influence B cell differentiation is well recognized (1, 5, 6). However, activated B cells themselves can produce significant quantities of cytokines that may contribute to the prevailing environment (7, 8). Indeed, there is growing interest in the potential of B cells to modulate immune responses, and it is attractive to postulate that they may accomplish this through regulating their profiles of effector cytokine secretion.

We hypothesized that B cells under different physiologically relevant conditions would produce distinct effector cytokine patterns. Literature review identified that an integrated profile of cytokines produced by human B cells under varying modes of stimulation has not been defined. Prior studies used a variety of methodologies to assess B cell responses following activation through the BCR or via CD40 engagement (2, 3, 4, 9, 10, 11), and reports on cytokine profiles produced by B cells under these conditions have been, at times, conflicting. This suggested to us that modifying the specific stimulation paradigm might significantly alter B cell cytokine output. We therefore studied ex vivo human B cells across a wide range of stimulations and established an optimized activation paradigm that matches current thinking of the anticipated effects of isolated or dual CD40- and BCR-mediated stimulation on outcomes of B cell proliferation, fas expression, and viability. We demonstrate that activated B cells can produce an array of cytokines, the bias of which is altered by the precise combination and strength of contributing signals. We propose a model in which B cells appropriately activated through both the BCR and CD40 produce high levels of IL-6, TNF-α, and lymphotoxin (LT), a combination that promotes GC development and amplifies T cell responses. In contrast, bystander B cells activated through CD40-mediated T cell stimulation without preceding Ag/BCR engagement produce negligible amounts of proinflammatory cytokines, but secrete significant levels of the immune regulatory cytokine IL-10, which would act to suppress the local inflammatory response.

Venous blood from healthy controls was drawn into EDTA. PBMC were separated using density centrifugation on Ficoll-Paque (Pharmacia Biotech, Uppsala, Sweden). CD19⁺ B cells were purified using magnetic cell sorting (MACS; Miltenyi Biotec, Auburn, CA). Preparations were typically >97% CD19⁺ by flow cytometry (CD3⁺ contamination <2%, negligible (<1%) monocyte (CD64⁺) contamination). B cells were washed twice to remove EDTA in the separation buffers, and resuspended in RPMI 1640 supplemented with 10% FCS, 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mmol/L l-glutamine (all from Sigma-Aldrich, St. Louis, MO). Cells were plated at 7.5 × 10⁵ cells/ml in a total volume of 200 μl in U-bottom 96-well plates.

For BCR cross-linking, we selected polyclonal goat anti-human IgG and IgM (Jackson ImmunoResearch Laboratories, West Grove, PA) capable of stimulating both surface IgM⁺ and IgG⁺ cells. We initially confirmed that this Ab induced the expected pattern of synergism with respect to B cell proliferation, when combined with IL-4 and CD40 engagement (1, 9) (data not shown). For CD40-mediated stimulation, we used CD40L on stably transfected cell mouse fibroblasts (L cells, kindly donated by Y. Liu, DNAX, Palo Alto, CA), because it has been shown that interactions with soluble CD40L may fail to elicit the full range of CD40-mediated events (10). L cells were lethally irradiated before coculture (1000 Gy). CD40L expression by flow cytometry was stable over the study period. For most activation protocols, B cells were cultured with the appropriate stimuli for 48 h at 37°C in 5% CO₂ in a humidified atmosphere. Aliquots of supernatant were removed and frozen at −70°C until batched analysis. Fresh medium was added, 1 μCi of tritiated thymidine being included in selected wells. Proliferation was assessed as the uptake of thymidine over 18 h in the pulsed wells, measured using a beta scintillation counter. Cells from parallel wells without added [³H]thymidine were collected at the same time point (66 h) for flow cytometric analysis.

For the staggered dual stimulation paradigms, cells were first incubated with anti-IgG and M for 24 h and then transferred with their medium to a preadhered monolayer of CD40L-transfected L cells from which all medium had been removed. Cells were then cultured for a further 48 h before collecting supernatants. Wells were subsequently treated, as described above (total culture time, 90 h).

Supernatants were analyzed as duplicate samples from replicate wells. IL-4, IL-6, IL-10, IL-12, TNF-α, and IFN-γ were assayed using OptEIA ELISA kits (BD PharMingen, San Diego, CA), following the manufacturer’s protocol. LT was assayed using ELISA plates coated overnight with 2 μg/ml anti-human LT (clone 259-238-8; BD PharMingen). After washing and blocking of the plate with 10% FCS, samples were incubated for 2 h at room temperature. After washing, 1 μg/ml biotinylated anti-human LT (clone 359-81-11; BD PharMingen), with avidin-conjugated HRP, was added for 1 h at room temperature. After further washing, color was developed using the tetramethylbenzidine reagents as for the OptEIA kits. The sensitivities and linear ranges of the cytokine ELISA were: IL-6, sensitivity 200 pg/ml, range 300-3000 pg/ml; IL-10, sensitivity 20 pg/ml, range 50–500 pg/ml; TNF-α, sensitivity 20 pg/ml, range 50–500 pg/ml; LT sensitivity, 20 pg/ml, range 50–500 pg/ml.

Total cellular RNA was precipitated from cells lysed in TRIzol (Life Technologies, Burlington, Ontario, Canada) following 24-h stimulation with BCR cross-linking alone or CD40-mediated stimulation alone, or after 24-h CD40-mediated stimulation preceded by 24-h BCR cross-linking. RPA was performed using Riboquant hCK2b and hCK3 primers following the manufacturer’s protocol without modification (BD PharMingen). Gels were exposed on a phosphor imager and scanned, and the resulting images were analyzed by ImageQuant software.

The following directly conjugated Abs were used: anti-CD3 FITC, anti-CD19 Cy, anti-CD95 PE, anti-CD154 FITC (all from BD PharMingen), and anti-CD64 PE (DAKO, Glostrup, Denmark). Staining was conducted in PBS/2% FCS. Cells were analyzed without delay on a BD FACScan flow cytometer using CellQuest software. Cell viability and apoptosis were measured using differential annexin V/propidium iodide binding with a commercial kit (BD PharMingen). This was performed at 66 or 90 h from the start of the culture period depending on the stimulation paradigm, as described above.

The current consensus is that BCR engagement alone induces minimal B cell proliferation, CD40 stimulation alone results in substantial proliferation, and, in combination, BCR and CD40 engagement exhibit synergy (1, 5, 12, 13). We sought to identify concentrations of stimulating reagents that reproduce this expected pattern.

Titration of BCR-mediated signaling revealed minimal B cell proliferation at concentrations of anti-BCR Ab <0.5 μg/ml. A dose-dependent increase in proliferation was noted at Ab concentrations >0.5 μg/ml (Fig. 1,a). The B cell proliferation observed at the higher Ab concentrations was independent of complementary signals, and was thus not considered representative of the anticipated in vivo state. As shown in Fig. 1 b, we observed significant induction of B cell proliferation across a broad range of CD40-mediated signaling (CD40L-transfected L:B cell ratios). Activation was further confirmed on flow cytometry as increases in forward and side scatter. At L:B cell ratios below 1:20, not all B cells were activated, and there were significant numbers of dead cells, confirmed by annexin V/propidium iodide binding (data not shown). We attributed this to the inability of all the B cells to physically interact with a limiting number of L cells.

The importance of carefully defining the strength of signal became apparent upon combining BCR and CD40 signals. Synergy was only seen at relatively low levels of each. As shown in Fig. 1,c, at CD40L-transfected L:B cell ratio of 1:40, BCR engagement across a broad range of concentrations resulted in synergistic proliferation. However, at a higher L:B cell ratio (1:10), titrating in the BCR cross-linking Ab produced little additional proliferation. We thus identified the optimal signals as BCR cross-linking Ab at 0.5 μg/ml and L cells at a ratio of 1:15 B cells. This combination consistently reproduced the expected proliferative responses from multiple individuals (Fig. 2,a). We also found the expected effects on fas up-regulation (4) (Fig. 2 b). Whereas fas expression on unstimulated B cells was negligible and not up-regulated by BCR cross-linking alone, fas was greatly up-regulated on B cells stimulated with isolated CD40 engagement. This CD40-mediated increase was dramatically attenuated by dual BCR- and CD40-mediated stimulation.

Using the paradigm defined above, we next tested our hypothesis that human B cells produce distinct patterns of effector cytokines under conditions designed to mimic Ag engagement (BCR cross-linking), bystander B cell activation (CD40 alone), or full dual stimulation. To approximate the in vivo delay described between Ag engagement and T cell help (2), we further studied staggered dual stimulation, in which 24-h BCR engagement precedes CD40-mediated signaling. Unstimulated B cells produced either no measurable cytokine or small amounts of IL-6 alone (mean = 65 pg/ml, n = 8). BCR cross-linking alone did not stimulate significant cytokine production (IL-6 mean = 89 pg/ml, p = NS, n = 8). IL-10, IL-12, TNF-α, LT, IFN-γ, and IL-4 all remained undetectable.

CD40 stimulation alone using CD40L-transfected L cells significantly stimulated the production of IL-10 (p = 0.0004, n = 10). As reported previously (14), IL-6 was also significantly induced compared with unstimulated cells (p = 0.0002, n = 10). Only minimal amounts of TNF-α and LT were induced. Simultaneous dual stimulation (Fig. 3, 40X) significantly suppressed the IL-10 production, while substantially inducing TNF-α, LT, and further IL-6. Interestingly, staggered dual activation amplified these distinct B cell cytokine profiles (Fig. 3, 40X), resulting in a further increase in the production of TNF-α, LT, and IL-6, while essentially eliminating the IL-10 production. IL-4, IFN-γ, and IL-12 were not detected under the above range of conditions. The reciprocal cytokine profiles observed with CD40 engagement alone compared with the staggered dual stimulation occurred, despite similar proliferative responses (nonsignificant difference in thymidine uptake, n = 8). Fas expression was equally reduced in the simultaneous and staggered dual stimulation protocols (fas mean fluorescence intensity = 195 in staggered dual stimulation; 194 in simultaneous; 599 with CD40 stimulation alone: n = 6, p = 0.0004).

This pattern of cytokine secretion was supported at the mRNA level by RPA, implying regulation, at least in part, at the transcriptional level (Fig. 4). With BCR cross-linking alone, mRNA was not detected for any of the cytokines measured. Staggered dual stimulation induced transcription of TNF-α, LT, and IL-6, but minimal IL-10. In contrast, CD40 stimulation alone induced higher IL-10 transcription, lower transcription of LT and IL-6, and no detectable transcription of TNF-α.

When used alone, even high concentrations of BCR cross-linking Ab (25 μg/ml) failed to induce significant production of LT or IL-6 (Fig. 5). Dual CD40- and BCR-mediated stimulation induced secretion of LT and IL-6 in a dose-dependent and additive fashion. The pattern of TNF-α secretion (data not shown) mimicked the LT pattern. In contrast to the above cytokines, the regulation of IL-10 secretion followed a more complex pattern: High concentrations of BCR cross-linking Ab alone induced significant IL-10 production. The CD40-mediated induction of IL-10 was inhibited by low doses of BCR cross-linking Ab, with a maximum reduction at 0.5 μg/ml. Further increasing the BCR-mediated stimulus resulted in increased IL-10 secretion, ultimately to levels higher than those induced by CD40 stimulation alone. This, however, occurred only at BCR cross-linking strengths previously shown to be supraphysiological in as far as the proliferation they induce is greater than that seen with CD40, and resistant to CD40-induced synergism.

The potential of B cells to produce a variety of cytokines is well established (7), but it is difficult to synthesize the existing results into an integrated model of the in vivo state. The use of nonphysiological mitogenic stimuli, cell lines rather than primary cultures, and a focus on a restricted profile of cytokines limits the interpretation of existing data. In this study, we established an in vitro activation model of directly ex vivo isolated human B cells reproducing the anticipated in vivo effects of BCR- and CD40-mediated signaling in terms of B cell proliferation, fas expression, and survival. We deemed this important, as preliminary experiments suggested that the range of stimuli achievable in vitro may not always reproduce in vivo phenomena. We believe we have effectively mimicked the earliest stages of B/T cell interaction.

CD40-mediated signaling through interaction with CD40L on activated T cells is known to be a critical trigger for proliferation and differentiation of B cells that have encountered their Ag (3). However, there is a risk that B cells, which all express CD40 constitutively (15), might interact with CD40L, up-regulated on T cells involved in other immune reactions. Such isolated CD40 signaling promotes B cell proliferation, while simultaneously targeting the cells for apoptosis through up-regulation of fas. Subsequent interaction with fas ligand, expressed on the CD40L-bearing T cells (4), limits undesired polyclonal B cell expansion, through the elimination of B cell clones that are not rescued by concurrent ligation of their BCR.

In contrast, BCR ligation combined with CD40-mediated T cell help promotes ongoing B cell development and plasma cell differentiation, usually via GC formation. Although these two signals are essential prerequisites for GC formation, they are not alone sufficient in vitro (16). In vivo, CD40/CD40L interactions are transient, and are followed in time by additional ligand-pair interactions and cytokine signals, which drive subsequent stages of T and B cell maturation. These later signals are not included in our current model, which reflects B cell responses occurring at the initial phase of B/T cell interaction. Thus, our study ascribes patterns of effector cytokine production to human B cells at their early stages of activation.

Isolated CD40-mediated stimulation is known to stimulate B cell IL-6 production (14), but, as we show, the amount is low, only a fraction of that produced with dual stimulation. B cell IL-6 acts as an autocrine growth factor, but it also affects T cells. It is proposed that IL-6 produced by APCs might be the primary in vivo stimulus inducing the Th2 phenotype, and may thus serve as a means for B cells to summon their own help via IL-4 (17).Importantly, IL-6 may also block the suppressor effect of CD4⁺CD25⁺ regulatory T cells (18). The relatively inefficient induction of IL-6 by CD40 alone is in keeping with this being an unproductive B/T cell interaction. IL-6 is markedly up-regulated upon dual BCR/CD40-mediated stimulation, especially with staggered dual stimulation. During an appropriate cognate interaction, this would serve to amplify both B and T cell responses, with a bias in the latter toward Th2-type B cell help.

TNF-α is a pluripotent cytokine produced by, and capable of stimulating, many cell types, including B, T, and dendritic cells. It is a necessary factor for B cell proliferative responses to both CD40 and BCR stimulation (19, 20). The target of TNF-α secreted by B cells may thus be autocrine or paracrine, but the net result would be stimulatory for the B/T cell interaction. LT is an absolute requirement in vivo for early B cell differentiation and function (21) and may be complementary with TNF-α in promoting GC formation (22). We therefore predicted that B cell production of these cytokines would occur only in the context of an appropriate B cell response.

Confirming this prediction, we show that CD40 stimulation alone is not an efficient stimulus for either TNF-α or LT, but combined with BCR stimulation both cytokines are potently induced, particularly in the staggered protocol. Because dual stimulation is a prerequisite for GC formation, the high production of both cytokines in this context represents a novel mechanism by which B cells can actively contribute to protective immune responses. We also noted that in our studies of ex vivo isolated peripheral blood B cells, we detected no TNF-α upon BCR stimulation alone (at either the protein or mRNA level). Others have reported BCR signaling to be sufficient to up-regulate TNF-α gene transcription from splenic B cells (23). We propose that this may reflect differences in the B cell populations studied, including the presence in the spleen of already activated B cells, contrasting with the generally quiescent state of circulating human B cells (24).

In sharp contrast to the other cytokines studied, we observed that B cell IL-10 secretion was maximally induced by isolated CD40-mediated signaling, and was abrogated in the dual stimulation paradigm. This would appear at odds with previous reports that IL-10 can be efficiently induced by combined BCR- and CD40-mediated stimulation (7) and act as a B cell growth and differentiation factor, promoting isotype switching and plasma cell formation (25, 26). However, the apparent discrepancy can be reconciled when considering both the context, as well as the stage, of B cell activation under study. We propose that, at the earliest stages of T/B cell interaction, as modeled in our current study, IL-10 production by appropriately stimulated B cells (dually stimulated via BCR and CD40 engagement) would be premature because it would terminate the expansion of the activated B cells, diverting them to plasma cell differentiation (27). IL-10 in this context might also be expected to impair the ability of T cells to provide optimal help to the recently activated B cells (25). In contrast, IL-10 production at the earliest stage of bystander B cell activation (CD40 engagement alone) would appropriately suppress the undesired immune response. IL-10 is well established as a regulatory cytokine that suppresses APC and T cell activation, and the importance of B cell-produced IL-10 in limiting inappropriate immune responses in vivo has recently been described in murine disease models (25, 28, 29).

The careful titration experiments conducted in our study underscore the importance of strength of signal in defining B cell effector responses. We first defined the strength of BCR-mediated stimulation that reproduced the expected physiologic responses of activated B cells, and further defined the strength of CD40-mediated signaling that would synergize with this BCR signal. Under these dual stimulation conditions, which we believe mimic the in vivo paradigm of T cell-dependent B cell activation, the addition of BCR signaling clearly abrogated the CD40-mediated induction of IL-10. When the signal strengths delivered through either the BCR or CD40 were increased, we found that BCR engagement could indeed amplify the CD40-mediated induction of IL-10, but only at doses of BCR cross-linking Ab, which were independently capable of inducing vigorous B cell proliferation and, as such, are not likely representative of typical T-dependent B cell responses.

In summary, the distinct cytokine profiles induced in the early setting of bystander activation vs the dual activation paradigm are well predicted in our model and point to the active roles that B cells are likely to play in regulating local immune responses. The inhibition of IL-10 production and the induction of TNF-α and LT, seen with simultaneous BCR/CD40 stimulation, are accentuated by staggered dual stimulation. This conforms to the consensus two-stage model of B cell activation. B cells fully activated by serial engagement of BCR and CD40 produce IL-6, TNF-α, and LT, a combination capable of amplifying both T and B cell responses. In contrast, B cells inappropriately activated by CD40 alone fail to produce proinflammatory cytokines and instead secrete IL-10. We propose that this CD40 induced IL-10 functions to suppress the early immune response and limit inappropriate bystander B cell activation. Emerging studies have demonstrated the importance of in vivo IL-10 in limiting inappropriate autoimmune responses in animal models (28, 29), and, to our knowledge, the present study is the first to implicate B cell IL-10 in the down-regulation of human immune responses. Our findings reconcile several apparent inconsistencies in prior studies reporting on B cell cytokines. They underscore the impact that the strength, mode, and sequence of stimulation have on the profile of human B cell effector cytokine production, and advance our fundamental understanding of how B cells may actively participate in the regulation of human immune responses.

We thank Dr. Trevor Owens for his insightful comments, and Linda Gilbert and Sherry Hebert for their help in preparation of the manuscript.