Of relevance to both protective and pathogenic responses to Ag is the recent finding that soluble molecules of the innate immune system, i.e., IL-4, B cell-activation factor of the TNF family (BAFF), and C3, exhibit significant synergy in promoting the clonal expansion of human B2 cells following low-level BCR ligation. Although IL-4, BAFF, and C3dg each contribute to early cell cycle entry and progression to S phase, only BAFF promotes later sustained viability of progeny needed for continued cycling. The present study sought to further clarify the mechanisms for BAFF’s multiple functions. By comparing BAFF and a proliferation-inducing ligand (APRIL) efficacy at different stages in the response (only BAFF binds BR3; both bind transmembrane activator and calcium modulator and cyclophilin ligand interactor (TACI) and B cell maturation Ag, the early role was attributed to BR3, while the later role was attributed to TACI/B cell maturation Ag. Importantly, BAFF- and APRIL-promoted viability of cycling lymphoblasts was associated with sustained expression of cyclooxygenase 2 (COX-2), the rate-limiting enzyme for PGE2 synthesis, within replicating cells. Supernatants of cultures with BAFF and APRIL contained elevated PGE2. Although COX-2 inhibitors diminished daughter cell viability, exogenous PGE2 (1–1000 nM) increased the viability and recovery of lymphoblasts. Increased yield of viable progeny was associated with elevated Mcl-1, suggesting that a BAFF/APRIL → TACI → COX-2 → PGE2 → Mcl-1 pathway reduces activation-related, mitochondrial apoptosis in replicating human B2 cell clones.
Due to their recirculating nature, B lymphocytes of the B2 subset will undoubtedly contact peripheral Ags (foreign or self) in sites remote from follicles of lymph nodes and spleen. The consequences of Ag engagement with the BCR may be B cell deletion, anergy, or full activation. One means by which full activation is assured is though contemporaneous receipt of coactivating signals from Ag-specific Th cells. However, under most circumstances, this is likely a rare peripheral event, because nonlymphoid tissues lack the mechanisms within secondary lymphoid organs for facilitating contacts between rare Ag-specific B cells and T cells. It is considerably more likely that costimuli for full activation will initially come from the nonspecific innate immune system in the form of 1) Ag-binding C3, which promotes coclustering of BCR and the CD21/CD19 complex (1, 2, 3), 2) B cell-activating factor (BAFF)3 (from macrophages, dendritic cells, and neutrophils) (4, 5, 6, 7), and 3) IL-4 (from mast cells, basophils, neutrophils and NK-like cells) (8, 9, 10, 11, 12, 13, 14). These effectors of innate immunity are abundant in all peripheral tissues where they serve as sentinels for the presence of infectious agents.
In sites of inflammation, B cells will undoubtedly confront more than one product of the innate immune system. This prompted a recent evaluation of the consequences of B cell engagement with a C3dg-bearing Ag (limiting dose of moderately multivalent anti-IgM:anti-CD21:dextran (dex)) in a milieu with both IL-4 and BAFF. Interestingly, notable synergy was noted between low-level BCR:CD21 coligation and engagement of BAFF and IL-4 receptors in promoting clonal expansion of CFSE-labeled human B2 cells (15). The resulting progeny were not plasmablasts or plasma cells, but rather, exhibited characteristics of APCs, i.e., high expression of CD86 (B7.2) and class II MHC (15). It is suspected that a similar expansion of Ag-specific B cell APC in vivo may contribute to the genesis of ectopic lymphoid tissue characterized by both activated B and T lymphocytes. Such tertiary lymphoid tissue is characteristic of organ-specific autoimmunity (16, 17, 18, 19, 20, 21, 22, 23, 24, 25) and often found in sites of localized, chronic infection with microbes, e.g., Helicobacter and Borrelia (26, 27, 28).
BAFF was found to have at least two roles in promoting the clonal expansion of human B2 cells exposed to limiting C3dg-bound Ag and IL-4 (15). First, BAFF synergized with the above ligands in optimizing the initial G1 → S-phase transition of resting B cells (15). Studies with murine B cells suggest that this may represent BAFF-mediated up-regulation of the early cell cycle proteins, cyclin D2 and cdk4 (29). Additionally, later signaling (≥day 3) via a BAFF receptor(s) was critical for sustained cycling of progeny (15). The two roles of BAFF are consistent with past in vitro and in vivo observations that BAFF not only augments day 2–3 DNA synthesis following BCR engagement (5, 6, 30), but also is required for optimal progression of germinal centers following immunization of mice with T cell-dependent (TD) Ags (32, 33, 34).
Which of the three known BAFFR, i.e., BR3, transmembrane activator and calcium modulator and cyclophilin ligand interactor (TACI), and B cell maturation Ag (BCMA) (5) are responsible for mediating the early and late BAFF signals in the above synergistic response is not clear. It is likely that the early effects are mediated by BR3 because resting B2 cells express high levels of BR3 but minimal expression of TACI or BCMA (15, 34, 35), and furthermore, other studies have shown that BR3 signaling augments early DNA synthesis following high-level BCR engagement by anti-IgM Ab or Staphylococcus aureus cells (34, 35). However, human B2 cells triggered by limiting BCR: CD21 ligand significantly up-regulate TACI within 32 h (15), raising the possibility that TACI and BR3 might jointly contribute signals for B cell activation and proliferation. As a means of discerning which of the three BAFF receptors are involved in the early- and late-stage signaling, the present study compared the temporally distinct effects of BAFF to that of a proliferation-inducing ligand (APRIL) (5). Like BAFF, APRIL is a TNF-family member produced by activated macrophages and dendritic cells and specific for both TACI and BCMA (5); however, unlike BAFF, APRIL does not bind BR3 (36). Thus, whether or not APRIL mimics the effects of BAFF at the early and late stages of the synergistic response will implicate either BR3 or TACI/BCMA.
The present study has additionally examined a possible mechanism for BAFF-promoted viability and cycling of B2 cell lymphoblasts. This involved assessing the role of cyclooxygenase 2 (COX-2), the rate-limiting enzyme for synthesis of PGE2 (37), recently shown to be critical for the sustained viability and growth of multiple lineages of transformed cells (38, 39, 40, 41, 42, 43, 44). Although the inducible COX isoform, COX-2, had not been thought to be expressed in B lymphocytes, Phipps et al. (41) reported that a unique subpopulation of biphenotypic B cells in the mouse peritoneal cavity express COX-2 and, more recently, that virtually all human peripheral blood B cells up-regulate COX-2 after high-level engagement of BCR and/or CD40 (46).
In this study, we show for the first time that viable replicating progeny of human B2 cells receiving synergistic stimuli from low-level BCR:CD21 coligation, IL-4, and BAFF exhibit significantly heightened levels of COX-2 protein. Late signals by BAFF, and particularly APRIL, are important for sustained COX-2 expression in daughter cells and maximal PGE2 production. Based on the above as well as inhibitory effects of several COX-2 inhibitors and stimulatory effects of low levels of exogenous PGE2, it appears that innate immune system-driven cycling of untransformed human B lymphocytes is, at least in part, dependent on sustained viability involving a TACI → COX-2 → PGE2 pathway.
Materials and Methods
The soluble mAb:dex conjugates used in this study were synthesized and characterized as described previously (3, 47). Each high m.w. dex molecule was covalently linked, via stable thioether bonds, to two distinct mAb, i.e., a mAb of low-, intermediate-, or high-affinity for human IgM (or IgG1 control mAb) and THB-5 anti-human CD21 mAb (or IgG2a control mAb). The three IgG1 anti-IgM mAbs used to construct the conjugates are all specific for proximal (or identical) epitopes on the Cμ2 domain; their Fab′ affinities range from Ka ∼2 × 106 M−1 (mAb P24), Ka = 2 × 107 M−1 (mAb Mu53), and Ka = 5 × 108 M−1 (mAb HB57 (DA4.4)) (3). The soluble conjugates have been determined to have 10–12 anti-IgM mAb (or control mAb) and 10–12 anti-CD21 mAb (or control mAb) per dex molecule (3). Unless indicated, cultures in these studies were routinely stimulated with a limiting concentration (0.01 μg/ml) of either intermediate-affinity anti-IgM:anti-CD21:dex or high-affinity anti-IgM:anti-CD21:dex.
Cytokines and other culture reagents
The rhBAFF was provided by Dr. S. Kalled (Biogen Idec) (4849), were used at a concentration of 500 ng/ml. PGE2 and PGI2 (Cayman Chemical) stocks were stored at −70°C in ethyl alcohol and diluted in culture medium just before use. The specific COX-2 inhibitors, celecoxib (Celebrex; Pharmacia/Pfizer) and CAY 10404, SC-58125, and NS-398 (Cayman Chemical); and the nonspecific COX inhibitor, indomethacin (Sigma-Aldrich), were stored at −70°C in DMSO and diluted just before use.
Purification of human B2 cell populations
Tonsils from 2- to 15-year-old donors, obtained by elective tonsillectomy, were used according to institutional review board guidelines (provided by Dr. S. McCormick and J. Jusino, New York Eye and Ear Infirmary, New York, NY). High-density tonsil follicular (B2) cells were purified as described (15, 50). The cells were ≥90% IgM+ with ≤2% CD3+ or CD16+ cells and when assessed, were ≥97% CD19+ and CD21+, and nearly all IgD+ and CD23+. Spleens were obtained from the National Disease Research Interchange and Cooperative Human Tissue Network, following removal for trauma, processed into single-cell suspensions, and cells were frozen at −150°C until B2 cell purification just before culture. For B2 cell purification, T cell-depleted suspensions were subjected to anti-CD43 and anti-CD27 magnetic bead negative selection (Miltenyi Biotech) to further deplete non-B cells and B1 cells (CD43+) and switched and subepithelial (marginal zone-like) memory B cells (CD27+). In more recent experiments, splenocytes are routinely selected for high-density (55 and 75% Percoll interface) before the CD27 and CD43 depletion step, with no change in conclusions. Additionally, where indicated, purified B cell populations were treated with leucine methyl ester to eliminate potentially contaminating macrophages, neutrophils, and NK-like cells (51).
B cells were cultured in an enriched medium, as described recently (15). In experiments involving two culture stages, cells were initially stimulated at 3 × 106 cells per 2 ml in round-bottom 15-cc culture tubes with the indicated reagents. After 32 h, cells were washed and recultured in 96-well plates at 2 × 105 cells per 200 μl of medium supplemented only with IL-4.
Assays for B cell DNA synthesis, viability, and division
B cells were pulsed with [3H]thymidine during the last 12 h of a 68-h culture period and harvested as described (3); mean cpm ± SEM of triplicate cultures is shown. To assess viability, cells were harvested, washed, fixed with 1% paraformaldehyde, and analyzed by flow cytometry. Apoptotic cells and viable cells were distinguished by differences in forward and side scatter after prior gating to remove debris (50). This scatter-based assay for distinguishing the viability of untransformed human B cells has good correspondence with results obtained by FITC-annexin staining (50) and intracellular staining with Ab to active caspase 3 (15, 52). To assess clonal replication, cultures of CFSE-labeled cells were harvested on day 6 or 7 (unless indicated otherwise), processed as above, and subjected to division gating by flow cytometry (FACScan; BD Biosciences). Increasing division number is represented by diminished CFSE fluorescence (53, 54). As discussed elsewhere (15), in all cultures, a subset of CFSE-labeled cells routinely die during the first 2 days of culture. These are evidenced by their failure to process short-lived CFSE-labeled proteins, as typically occurs early in culture (53, 54). As a result, they exhibit significantly greater levels of CFSE than the viable undivided population and are seen as a right shoulder of the undivided peak in cell populations gated as apoptotic on the basis of forward scatter/side scatter. In most experiments, division was assessed by determining the percentage of the total cell population (viable plus apoptotic), or alternatively, the total viable population, which exhibited a given level of CFSE fluorescence. In some experiments, the absolute recovery of cells within each division subset was calculated through adding a known standardizing quantity of beads to the recovered cell population analyzed by flow cytometry (54).
Cell surface staining for BAFF/APRIL receptors
Assessments of the membrane expression of BR3, TACI, and BCMA were performed as described recently (15) with mAbs specific for hBR3 (clone 9–1), hTACI (clones C4D7.4 and A1G11.4), and hBCMA (C4E2.2-1; donated by Dr. S. Kalled, Biogen Idec).
Intracellular staining for COX-2, COX-1, and Mcl-1
Cells from days 4 or 6–7 CFSE-labeled cultures were stained intracellularly using protocols described recently (15). Indirect staining assays involved the first stage addition of mouse mAb (0.5 μg Ab/∼105 cells/50 μl): anti-Mcl-1 (BD Pharmingen) or IgG1 control, MOPC-21. This was followed by second-stage incubation with RPE-conjugated goat F(ab′)2 anti-mouse IgG (H + L), preabsorbed to remove Abs reactive with human Ig (Southern Biotechnology Associates). Direct staining assays involved the addition of PE-anti-COX-2 (Cayman Chemical) or PE-IgG1 control (BD Pharmingen), and on occasion, PE-anti-COX-1 (Cayman) or PE-anti-COX-2 (mAb sc-19999; Santa Cruz Biotechnology) (0.5 μg PE-Ab per assay as above). To test for COX-2 Ab specificity, PE-labeled mAb (Cayman Chemical) was preincubated with assay buffer or COX-2 peptide (representing the immunogen) (each at 10 μg/ml) for 30–40 min before use in intracellular staining. Two-color flow cytometry was used to compare levels of PE (FL2) label on cells with differing levels of CFSE (FL1). Due to the low level of CFSE (CFDA-SE; Molecular Probes) used for staining (1 μM; needed to preclude loss in culture viability), compensation was readily achieved.
Immunoblotting experiments to assess COX-2 levels
Viable cells from day 6 cultures stimulated by limiting BCR:CD21L and varying cytokine combinations were isolated by Ficoll-Hypaque centrifugation for preparation of lysates as described earlier (15). In each experiment, equal amounts of lysate protein from each condition were loaded onto 10% SDS-PAGE gels and subsequently transferred to nitrocellulose. COX-2 was detected with anti-COX-2 mAb (clone 33; BD Pharmingen), followed by HRP-conjugated goat anti-mouse IgG and ECL. Blots were stripped and analyzed for actin as a loading control. Relative levels of each protein were determined by densitometric scanning, as described elsewhere (15).
Culture supernatants were obtained at the time of cell harvesting and frozen at −20°C until assay with a previously described inhibition RIA specific for PGE2 using 3H-labeled PGE2 and anti-PGE2 Ab (Sigma-Aldrich) (55).
BAFF is more effective than APRIL at delivering early signals that promote initial cell cycle progression of human B2 cells
A previous study by Craxton et al. (6) had shown BAFF to be more effective than APRIL in promoting early DNA synthesis in human B cell cultures triggered by a high dose of anti-IgM Ab. The data in Fig. 1⇓A show that this also is the case when human B2 cells are triggered by a suboptimal dose of anti-IgM:anti-CD21:dex (hereafter referred to as BCR:CD21L) and optimal IL-4. This was evidenced both by 1) the dose needed for maximal effects (∼20 ng/ml vs ≥500 ng/ml for BAFF and APRIL, respectively) and 2) the maximal response achieved. Further indications that BAFF, and not APRIL, promotes early cell cycle progression came from a two-phase culture experiment (Fig. 1⇓B). Although early (0–32 h) exposure to BAFF (50 ng/ml) augmented day 3 DNA synthesis, APRIL had little to no effect at this early interval (even at a 5-fold greater concentration than BAFF). Thus, even under these triggering conditions, BR3 is the major BAFFR involved in promoting the G1 → S-phase transition of resting human B2 cells (34, 35). A likely downstream consequence of early BR3 signals is the up-regulation of early G1 cell cycle proteins, cyclin D2 and cdk4 (29). The early BAFF effect does not represent enhanced viability, because as noted elsewhere (15, 52), the presence of BAFF has only a minor, if any, effect on the viability of resting or activated human B2 cells during the first 3 days of culture. Additionally, BAFF-dependent increases in BCR or CD21, which might facilitate signaling by limiting surrogate Ag, were not observed (Ref. 56 and data not shown).
APRIL is more effective than BAFF at delivering delayed signals that promote recovery of viable daughter cells
To compare APRIL and BAFF efficacy at providing delayed signals to replicating cells, the division state and viability of CFSE-labeled B2 cells was analyzed. Cultures received a pulse of medium, BAFF or APRIL, at 44, 78, and 108 h after stimulation with BCR:CD21L plus IL-4 (Fig. 1⇑C). The extent of clonal proliferation following 6–7 days of culture was discerned by monitoring the incremental decreases in CFSE fluorescence characteristic of each successive division (15, 53, 54). As seen in Fig. 1⇑C, although delayed addition of either BAFF or APRIL resulted in increased viable daughter cell yield, APRIL typically showed greater effects.
An increase in daughter cell yield might be the result of signals that promote cell viability and/or directly promote cell cycling. The data in Fig. 2⇓ indicate that APRIL significantly augments the viability of replicating lymphoblasts, similarly to BAFF (15). Using an approach described earlier (15), this was evidenced by gating cultured CFSE-labeled cells, consisting of both viable and apoptotic cells (57, 58), into division subsets (Fig. 2⇓A) and analyzing for percentage recovery (Fig. 2⇓B) and percentage viability (Fig. 2⇓C) within each division subset. Increases in daughter cell viability were readily noted even upon a delay of 96 h in the addition of APRIL and were typically most manifest in extensively replicated B cells (Fig. 2⇓C). Importantly, APRIL augments viability/recovery of daughter cells, even when optimal concentrations of response-augmenting BAFF are present. Although the effect has been most pronounced at very high doses of APRIL (625 ng/ml in Fig. 2⇓), it also is typically quite evident in experiments in which the dose of supplementary APRIL is restricted to 200–300 ng/ml. Sustained viability of replicating cells will undoubtedly have a positive impact on the yield of subsequent progeny. Whether late signaling by APRIL/BAFF also directly promotes the proliferation process is unclear.
Statistical analysis of the data from six experiments (by Student’s paired t test) indicated that the percentage recovery of total cells representing ≥3 divisions is significantly greater (p < 0.05) when both macrophage/dendritic cell cytokines are present than when only BAFF or only APRIL are added. Additionally, although viability tends to decline with increasing division, regardless of the cytokine combinations present, the viability of the cells representing two, three, and four divisions is significantly greater when BAFF and APRIL are jointly present.
Taken together, the previous (15) and present observations suggest that the innate immune system’s B cell tropic molecules, i.e., C3, IL-4, BAFF, and APRIL, each contribute in somewhat distinctive ways to maximize clonal expansion when recirculating B2 cells confront limiting Ag in an inflammatory setting. C3dg-Ag, IL-4, and BAFF have important roles in promoting a first round of replication (15). Additionally, as shown in this study, BAFF, and generally to a greater degree, APRIL, initiate signals that promote continued clonal expansion, at least in part, through sustained daughter cell viability.
BCMA-Fc and TACI-Fc abrogate APRIL-dependent clonal expansion
To confirm that APRIL’s effects were dependent on the preparation’s specificity for BAFFR, a neutralizing dose of BCMA-Fc or TACI-Fc was added to cultures stimulated 2 days earlier. Data in Fig. 3⇓, A and B, show that both the viability and replication-promoting effects of APRIL are significantly reduced by either BCMA-Fc or TACI-Fc treatment. Although addition of TACI-Fc to day 2 cultures with IL-4 and BAFF (no APRIL) was quite effective at neutralizing BAFF’s viability-promoting effects (Fig. 3⇓B) (15), BCMA-Fc was notably less effective. The latter likely reflects BAFF’s low intrinsic affinity for monomeric BCMA (59).
APRIL promotes further clonal replication in B2 cell cultures triggered by low-affinity as well as high-affinity BCR engagement
Past studies from this laboratory had revealed that the synergy between BCR:CD21 coligation, IL-4 and BAFF, resulted in significant clonal expansion of a fraction of IgM+ cells, even under conditions of low BCR:ligand affinity (15). The experiment in Fig. 4⇓A confirms these findings; indicates that this is apparent when the absolute number of recovered cells are calculated (by use of standardizing beads); and furthermore, shows that, with APRIL additionally present, the effect is accentuated by further cycles of replication. The relative effect of having only IL-4 plus APRIL (no BAFF) present was not evaluated in the experiment in Fig. 4⇓A. Nevertheless, results from the additional experiment in Fig. 4⇓B indicate that BCR:CD21L-triggered cultures supplemented with IL-4 and APRIL (limiting concentration of 200 μg/ml) generate absolute yields of highly replicated cells comparable to, or slightly exceeding, that in cultures supplemented with IL-4 and optimal amounts of BAFF.
Although BR3 levels decline with successive replications, low-level TACI shows transient increases associated with replication
The above functional studies suggested that, although BR3 is critical early in activation, TACI and/or BCMA functions prominently later during the rapid replication phase. To determine whether this transition is associated with kinetic changes in the relative expression of receptors, two-color flow cytometric analyses were performed with CFSE-labeled B cells and mAbs specific for BR3, TACI, and BCMA at various intervals after activation with BCR:CD21L plus IL-4. (Staining of BAFF-containing cultures was not informative because BAFF interferes with the binding of the specific mAbs; data not shown.) Fig. 5⇓ shows representative dot blots and summarized data on the kinetics of BR3 and TACI expression within viable cells, as a function of division status. BCMA expression is not shown because it was weakly and not reproducibly expressed on dividing cells (albeit the mAb bound strongly to a BCMA+ cell line). The results indicate that BR3 is highly expressed on the undivided population at day 2 (Fig. 5⇓A) and sustained within the cells that fail to divide during the remainder of the culture period (Fig. 5⇓B). As lymphoblasts undergo successive divisions, levels of BR3 are diminished. This contrasts with TACI in two respects. First, TACI expression declines in the nondividing population from the up-regulated levels expressed at day 2 (15) (Fig. 5⇓). Furthermore, albeit expressed at low levels, TACI appears to be transiently up-regulated in lymphoblasts undergoing division.
Human B2 cells replicating under the influences of BAFF and/or APRIL display increased levels of COX-2
In an effort to unravel the mechanism(s) for the late augmenting effect of APRIL and BAFF, cultures were assessed for expression of the rate-limiting enzyme in PGE2 synthesis, i.e., COX-2, recently linked to malignant growth (40, 42, 43, 44). To facilitate comparisons of COX-2 in B cells with differing division history and viability, CFSE-labeled cell cultures were fixed, permeabilized, and stained intracellularly for COX-2 using PE-conjugated mAb. Interestingly, although peripheral blood cells have been recently noted to up-regulate COX-2 protein within 24 h after activation with a relatively high dose (10 μg/ml) anti-IgM and/or CD40L (20- to 30-fold increase by both Western blotting and flow cytometry) (46), only low-level up-regulation of COX-2 was detected in tonsil B2 cells activated for 24 h by limiting BCR:CD21L (0.01 μg/ml mAb protein) plus IL-4 plus BAFF or BCR:CD21L alone (data not shown). Nevertheless, the COX-2 levels rose substantially by days 4–6 within viable cells undergoing replication (Fig. 6⇓). Specificity for COX-2 was indicated by the ability of a soluble peptide corresponding to the immunogenic epitope to inhibit staining (Fig. 6⇓A) and reproducibility with another anti-COX-2 mAb (sc-19999 PE; Santa Cruz Biotechnology) (data not shown). Although in some experiments, COX-1 levels slightly rose with replication, this was not consistently observed (Fig. 6⇓A).
Following multiple COX-2 analyses in similarly activated human B2 cell populations, several consistent observations were made. Thus, 1) COX-2 is barely evident within the apoptotic lymphoblasts (Fig. 6⇑C), although the latter stain well for caspase 3 (15); 2) COX-2 is maximally expressed within replicating lymphoblasts in cultures containing BAFF, and especially APRIL (Fig. 6⇑, B and D); 3) COX-2 appears to decline in viable lymphoblasts following an optimum noted with two to three division cycles; and finally, 4) within stimulated cultures containing BAFF and/or APRIL, COX-2 is generally expressed in the subpopulation of viable, dividing lymphoblasts at two to three times the levels found in viable, nondividing lymphoblasts. Lesser COX-2 expression in the nondividing population is not due to a failure to receive activation signals: nearly all cells exhibit heightened expression of CD23, CD86, and LFA-1 and diminished expression of CD180 after 3 days of activation with low-dose BCR:CD21L plus IL-4 plus BAFF (15, 56) (data not shown). Immunoblotting studies (Fig. 6⇑E) showed heightened levels of COX-2 in viable lymphoblasts from day 6 BCR:CD21L-triggered cultures containing additional BAFF/APRIL, compared with viable lymphoblasts from cultures supplemented only with IL-4. Taken together, these new findings suggest that innate immune system cytokines, which function in augmenting daughter cell viability and replication, are strongly linked to the sustained expression of COX-2 within the viable dividing population.
B2 cell cultures stimulated with low-dose BCR:CD21L produce heightened levels of PGE2 in a cytokine-modulated manner
To evaluate whether activated cultures of human B2 cells release the major physiologic product of COX-2 function, i.e., PGE2, supernatants were taken from cultures at the time of harvesting for CFSE division slicing, and PGE2 levels were assessed by inhibition RIA (55). The data in Fig. 7⇓A (Exp 1 and 2) show that stimulation with low-dose BCR:CD21L alone consistently elevated the levels of PGE2 above that of control cultures with medium alone. The levels were further heightened by the presence of BAFF or APRIL (Exp 1–3). In addition, in cultures containing IgG control:dex, heightened PGE2 was detected upon supplementation with BAFF or APRIL. The latter indicates that BCR:CD21 cross-linking is not obligatory for PGE2 production by BAFF/APRIL and suggests that low levels of TACI expressed on resting B cells (15) might mediate this.
Although BCR:CD21 stimulation, in a milieu with only BAFF or APRIL, generated supernatants with significant levels of PGE2, compared with cultures with BCR:CD21L alone (Fig. 7⇑A), the further presence of IL-4 significantly reduced these levels (Fig. 7⇑B; Exp 1–3). This suggests that, following certain stimuli, IL-4 might have a dampening effect on B2 cell PGE2 synthesis in B2 lymphocytes, as reported in myeloid cells (60).
An analysis of the kinetics of PGE2 production in B2 cells cultures with 1) IL-4 and BAFF (no BCR ligand), 2) BCR:CD21L plus IL-4 plus BAFF, or 3) BCR:CD21L plus IL-4 plus BAFF plus APRIL revealed that BCR:CD21L-triggered cultures with APRIL had the most uniformly up-regulated expression of PGE2, i.e., highest at day 3 and at the termination of the experiment, i.e., day 11 (Fig. 7⇑D). This is consistent with APRIL’s effects at augmenting COX-2 production in activated cultures (Fig. 6⇑). Suggesting that COX-2, and not COX-1, is responsible for the PGE2 present in B2 cell cultures are the findings that PGE2 levels are significantly reduced in the presence of the selective COX-2 inhibitors, celecoxib and SC58125 (Fig. 7⇑C).
Although the above experiments indicate that cultures receiving a limited BCR:CD21 stimulus produce heightened levels of PGE2 and that innate immune system cytokines modulate this response, some caution must be used interpreting the data obtained from supernatant analysis. First, PGE2 is quite labile (61, 62), and thus the amount of PGE2 detected at the time of the RIA may not be an accurate representation of the amount generated in culture. Additionally, cells may serve as a sink for synthesized PGE2. This is particularly relevant when one considers that differences exist in the density of activated B cells as responses evolve (Fig. 4⇑) and that certain stimuli can up-regulate B cell membrane receptors for PGE2 (63). Nevertheless, even with these caveats, it does appear that several stimuli associated with optimal B cell proliferation do induce B cell PGE2 production.
Exogenous PGE2 augments human B2 cell clonal expansion following activation by low-dose BCR:CD21L, IL-4, and BAFF
Given the correlation between COX-2 expression, PGE2 production, and BAFF/APRIL-facilitated viability of replicating lymphoblasts, it was of interest to examine whether the major byproduct of COX-2 activity, i.e., PGE2, could influence human B2 cell clonal expansion when supplied exogenously. Cell division and viability were examined after 6–7 days in low-dose BCR:CD21L plus IL-4 ± BAFF-stimulated CFSE-labeled cultures pulsed with various concentrations of PGE2 (or ETOH vehicle control) at days 0, 2, and 5. Multiple pulses were administered due to labile nature of PGE2. The three representative experiments in Fig. 8⇓A, including one which used spleen B2 cell populations treated with leucine methyl ester to remove potentially contaminating myeloid cells (Exp 3), show that PGE2 differentially affects the response, depending on the dose administered. Doses from the lowest concentration tested in these experiments, i.e., 20–1000 nM, augmented both the viability and overall yield of extensively replicated daughter cells. However, in striking contrast, doses >1000 nM were suboptimal, and concentrations ≥5000 nM were quite suppressive. Other experiments indicate that the lower threshold at which exogenous PGE2 potentiates the viability and recovery of daughter cells in cultures stimulated with low-dose BCR:CD21L plus IL4 plus BAFF was ∼1 nM (data not shown). Further studies examining the temporal requirement for response-augmenting PGE2 found that PGE2 could exhibit significant augmenting effects when added at day 2; however, if delayed until day 4, little effect was noted on the day 6–7 response (data not shown). Fig. 8⇓B shows pooled results for daughter cell yield and viability in cultures pulsed with 100-1000 nM PGE2 (or vehicle controls) at days 0 and/or 2 (n = 12 experiments). Statistical analysis of these data indicates that both the increased recovery of daughter cells, and particularly the increased viability of daughter cells, attained by the addition of PGE2 are highly significant (p < 0.01). These effects of PGE2 were not apparent with an alternative product of COX-2 enzymatic activity, i.e., PGI2 (prostacyclin), at a concentration of 1000 nM (Fig. 8⇓C). Taken together, these experiments indicate that the downstream product of COX-2 enzymatic activity can, at least in part, mimic the effects of delayed signals initiated by BAFF/APRIL.
Inhibitors of COX-2 reduce B2 cell clonal expansion and daughter cell viability in cultures stimulated with low-dose BCR:CD21L, IL-4, and BAFF
To confirm the importance of COX-2 function in the innate immune system-driven B2 cell response, we examined the effects of several selective COX-2 inhibitors on the yield and viability of CFSE-labeled daughter cells after 6–7 days of culture. The data in Fig. 9⇓ show that, when added at the beginning of the culture, each of four selective COX-2 inhibitors (celecoxib, SC58125, CAY10404, and NS-398 at a dose of 100–150 μM) reduced both the viability and yield of progeny with multiple divisions, compared with cultures receiving DMSO vehicle control. Indomethacin exhibited similar effects (data not shown). Importantly, in contrast with the effects on lymphoblasts with ≥2 divisions, the COX-2 inhibitors minimally affected the viability of undivided cells and cells that had divided once (Fig. 9⇓, C and D). An assessment of the absolute yield of viable daughter cells representing ≥3 divisions (Fig. 6⇑E) further shows that a pulse of celecoxib on day 2 of culture notably reduced the yield of viable progeny.
The COX-2 inhibitor dose that consistently suppressed the viability/yield of daughter cells (100 μM) is higher than the minimal pharmacologic doses reported for effective blockade of PG production (65). Nevertheless, it is compatible with the dose of celecoxib and/or SC58125 shown to optimally inhibit viability of B cell lymphomas (38) and to promote the apoptosis of activated T cells from systemic lupus erythematosus patients (64). Part of the discrepancy may be due to the labile nature of these inhibitors, even when stored in nondefrosted aliquots at −70°C in DMSO. Because functions of COX-2 that are independent of PGE2 synthesis have been reported (66, 67), it also is possible that the latter contribute to the observed suppression by pharmacologic agents that engage COX-2. To better resolve this, cultures receiving the COX-2 inhibitors were supplemented with exogenous PGE2. The data in Fig. 9⇑G show that exogenous PGE2 significantly reversed the suppression of daughter cell viability, suggesting that the effects of the COX-2 antagonists on daughter cell viability must, in large part, reflect an abrogation of PGE2 synthesis.
Thus, when taken together, several lines of evidence strongly suggest that a COX-2/PGE2 pathway is involved in the BAFF/APRIL-sustained viability of replicating clones. First, BAFF and APRIL could promote expression of both COX-2 and PGE2. Second, the augmenting effects of exogenous PGE2 appear to mirror late signaling effects of APRIL/BAFF. Third, COX-2 inhibitors impair the viability of highly replicated cell, in a manner reversed by exogenous PGE-2. Finally, two of two experiments have indicated that incorporation of neutralizing anti-PGE2 mAb (Cayman Chemical) (68) into cultures at a dose of 10 μg/ml causes specific, albeit partial, inhibition of APRIL-promoted responses (data not shown). The partial inhibition represents the limited neutralizing capacity of this dose of mAb, because the same mAb only partially inhibited the response attributable to exogenously added PGE2 (40 nM). The initial description of this neutralizing Ab indicated that a 10-fold molar excess over PGE2 was required for reversing the functional activity of PGE2 by 50% (68); with 40 nM PGE2, this is calculated to be 60 μg/ml mAb, substantially more than the dose available for use in this study.
Augmented clonal expansion of human B2 cells in the presence of BAFF, APRIL, or PGE2 is associated with heightened levels of Mcl-1 within the viable replicating B cells
We had demonstrated recently that the BAFF-dependent increase in daughter cell viability and replication within cultures triggered by low-dose BCR:CD21L plus IL-4 was linked to sustained expression of the antiapoptotic molecule, Mcl-1 (but not Bcl-2 or Bcl-x) within the replicating viable cells (15). Although IL-4 was critical for the large rise in Mcl-1 expression at day 3, BAFF was important for assuring that a high proportion of the divided cells sustained Mcl-1 expression (15). Given the present evidence that APRIL and PGE2 can both also contribute to augmented B2 cell clonal expansion, we assessed whether these late-acting stimuli also promoted Mcl-1 expression within the dividing B cell population.
Fig. 10⇓ shows the relative intracellular expression of Mcl-1 within cells with differing division history following 6 days of culture with BCR:CD21L plus IL-4 ± BAFF (day 0) ± APRIL or PGE2 (day 2). The pooled data from three experiments indicate, as described in Ref. 15 , that BAFF augments Mcl-1 levels within the divided cell population but has negligible effects on the undivided population. Furthermore and importantly, delayed signals from APRIL or PGE2 further increase Mcl-1 expression within highly replicated cells. These observations were evident both when the viable cell population alone was assessed (Fig. 10⇓B) and when the total cells (which include apoptotic cells with dramatically lowered Mcl-1 expression) (15) were analyzed (Fig. 10⇓A).
The present study provides several new mechanistic insights into how recirculating human B2 cells are driven to begin and to continue physiologically-relevant replication under the influences of macrophage/dendritic cell-derived BAFF and APRIL. First, in lymphocytes receiving early signals via limiting C3dg-bound Ag and IL-4, BAFF promotes the G1 → S-phase transition (15, 29) through a receptor not shared with APRIL. This implicates the BAFFR, BR3, and is consistent with 1) the substantially higher expression of BR3 (vs TACI and BCMA) on resting human B2 cells (15, 34, 35) and 2) past studies showing that antagonist mAbs to BR3 block day 3 DNA synthesis in cells stimulated with relatively high-dose anti-IgM Ab or SAC (35, 36). Second, both BAFF and APRIL promote the later sustained viability of lymphoblast progeny, implicating a downstream role for TACI and/or BCMA in the clonal expansion period. Third and importantly, later signaling by BAFF and APRIL leads to up-regulated expression of COX-2/PGE2 within nontransformed replicating B2 lymphoblasts, phenomena shown here to be critical for sustaining the viability/replication of daughter cells. Finally, human B2 cells proliferating under the influences of BAFF, APRIL, or PGE2 manifest heightened levels of Mcl-1, an antiapoptotic molecule of the Bcl-2 family known to be selectively up-regulated by COX-2 and PGE2 (38, 69). Taken together, the new findings suggest that a BAFF/APRIL → COX-2 → PGE2 → Mcl-1 pathway contributes to the continued viability and cycling of human B2 cells following encounters with limiting C3dg-coated Ag in an inflammatory environment rich in IL-4, an important early-acting product of the innate immune system (15).
Although both BAFF and APRIL individually augmented B2 cell replication, the most extensive response was noted when both macrophage/dendritic cell factors were present. In part, this undoubtedly reflects BAFF’s unique contributions at facilitating the earlier G1 → S-phase transition (15). Additionally, this reflects APRIL’s greater effectiveness at initiating later signals for sustaining viability and replication. The basis for APRIL’s heightened effects are not resolved in this study but likely represent this molecule’s recently appreciated ability to bind to cell-associated proteoglycans (70). A presentation of APRIL on cell membranes should increase its functional valency and, hence, triggering potential (70). It appears less likely that the optimized response noted with the combination of BAFF and APRIL is dependent on the formation of APRIL:BAFF trimers (71) because APRIL alone was generally more effective than BAFF alone at augmenting yield of viable extensively replicated cells.
The most important discoveries of this study were the related observations that BAFF and APRIL promote the expression of COX-2 in cycling human B2 cells and, furthermore, that COX-2-derived PGE2 contributes to BAFF/APRIL-augmented viability/replication. Although other separate lines of study have suggested that 1) COX-2 contributes to the rapid growth/development of malignant cells (38, 39, 40, 41, 42, 43, 44), and 2) BAFF and APRIL are associated with cell growth and viability, both in normal and transformed cells (52, 72, 73), the present study represents the first demonstration, of which we are aware, that the two are linked, i.e., that BAFF/APRIL function, at least in part, involves the up-regulation of COX-2/PGE2.
Although a direct relationship between BAFF/APRIL-induced second messengers and enhanced COX-2 transcription has not yet been shown, a biochemical basis for such a linkage is suggested upon comparing the intracellular signaling molecules activated upon TACI or BCMA ligation with molecules known augment COX-2. Thus, the TRAFs that bind the cytoplasmic tails of TACI and BCMA are potent activators of NF-κB and MAPK (74, 75), and TACI, through its unique associations with CAML, activates NFAT (76). All the above integrate quite well with the fact that COX-2 promotor activity is highly regulated by NF-κB, NFAT, and MAPK (77, 78, 79).
A number of past studies have investigated the effects of exogenous PGE2 on B cell proliferation and Ab synthesis. Although PGE2 suppressed B cell responses under certain culture conditions (80, 81), under other conditions, this eicosanoid augmented late [3H]thymidine synthesis (80, 82), Ab synthesis, and/or isotype switching (46, 83, 84, 85, 86). Importantly, plasma cell differentiation and H chain-class switching are quite dependent on prior B cell division (87). Given the new information provided in this report, we strongly suspect that the past evidence for PGE2 augmentation may, at least in part, have been explained by PGE2-promoted viability of replicating B cells.
Interestingly, although APRIL−/− mice exhibit no major immunological defects, several consequences of APRIL under- or overexpression are consistent with an in vivo role for APRIL at sustaining B cell proliferation responses to T cell-independent (TI) Ags. Mice deficient in APRIL exhibit significantly lower baseline levels of serum IgA, impaired IgA responses to both TI-2 and TD Ags, and slightly impaired IgM responses to TI-2 Ags (88). In contrast, mice expressing an APRIL transgene have higher baseline levels of IgM, higher IgM and IgG responses to TI-2 Ags, and augmented IgM responses to virus Ag (89). Given that 1) IgM and IgA Abs are the most TI isotypes under physiologic conditions (90, 91), and that 2) both isotype switching and plasma cell differentiation are phenomena requiring significant cell division (87), APRIL’s major function may be to serve as an adjunct to BAFF in sustaining B cell clonal proliferation when T cell CD40 signals are limiting.
Although the present study has not clearly distinguished whether TACI and/or BCMA is responsible for the late viability-promoting effects of BAFF/APRIL, there are several reasons to suspect that TACI is involved. First, TACI is up-regulated from the low levels in resting B2 cells within 2 days of activation by BCR:CD21L (15) and, albeit negatively modulated by IL-4 throughout the response (P.K.A. Mongini, unpublished results; and Ref. 15), undergoes transient up-regulation as B cells replicate (Fig. 5⇑). BCMA is not reproducibly observed on these replicating cells upon surface staining. Second, TACI deficiency, but not BCMA deficiency, quite significantly impairs in vivo immune responses to TI-2 Ag (92). Third, APRIL-induced in vitro isotype switching to IgG1, IgA, and IgE in cultures of IgM-positive B cells is intact if the B cells are from BCMA−/− mice, but ablated if from TACI−/− mice (93). Nevertheless, a role for BCMA in the BAFF/APRIL-sustained viability of progeny from activated human memory B cells was suggested recently (52). Given several differences between the present and the later study, it is possible that relative use of TACI vs BCMA by cycling cells may represent 1) the naive vs memory phenotype of the responding precursor; 2) the triggering stimuli; and 3) importantly, the differentiation stage of the progeny, i.e., lymphoblasts (15) or alternatively, plasmablasts (52).
Although mice deficient in APRIL or TACI exhibit impairments in certain B cell responses, quite interestingly, the TD Ag-induced response of these deficient mice is characterized by an increase in the frequency and size of germinal centers and subsequent IgG Ab formation (88, 92, 94). The reasons for the Janus-like, opposing functions of TACI remain unclear, but the fact that responses to TD Ags are selectively inhibited, together with the observation that certain anti-TACI Abs significantly inhibit day 3 DNA synthesis triggered by anti-CD40 plus IL-4 (95), suggest that APRIL/TACI signals are particularly antagonistic for cells receiving signals via CD40.
An intriguing possibility is that the inhibitory effects of TACI signaling in CD40-driven responses reflects an overabundance of B cell-synthesized PGE2. This is suggested from the present findings that PGE2 affects the viability/recovery of replicating B2 cells in a very dose-dependent manner, with high doses significantly inhibiting daughter cell viability and replication (Fig. 8⇑). The additional strong suggestion from this study that TACI up-regulates COX-2/PGE2 production, taken together with other findings that PGE2 is more abundant in human B cell cultures optimally triggered by CD40, compared with BCR (46), suggests that combined signaling via TACI and CD40 could generate supraoptimal levels of PGE2. Studies are being initiated to evaluate this.
Some insight into how the BAFF/APRIL → COX-2 → PGE2 pathway promotes human B2 cell clonal expansion came from noting that the Bcl-2 family anti-apoptotic molecule, Mcl-1 (but not Bcl-2 or Bcl-x) (15) was up-regulated in daughter cells replicating under the influences of either BAFF (15), APRIL, or exogenous PGE2 (present study). These findings are consistent with prior reports that BAFF promotes Mcl-1 expression in plasma cells and transformed myelomas (96, 97) and, furthermore, that Mcl-1 is uniquely regulated by COX-2/PGE2 in transformed cell lines (38, 69). Mcl-1 binds to the proapoptotic Bcl-2 family members, Bim and Bak, with high affinity (98, 99, 100) and thus likely plays an important role in precluding mitochondrial apoptosis within the cycling B2 population.
Of interest, Mcl-1 up-regulation is achieved through distinct receptors at varying stages in the innate immune system-driven B2 cell response. Before division, IL-4R signaling is primarily responsible for Mcl-1 elevation in cells triggered by limiting BCR:CD21L (15). Although cycling as lymphoblasts or plasmablasts, a TACI and/or BCMA pathway appears to assume the major role. The likelihood that TACI and BCMA both up-regulate Mcl-1 is consistent with their common binding to the signaling molecules, TNFR-associated factor 2 (TRAF2), TRAF5, and TRAF6 (5, 74, 101). The latter TRAF molecules are also bound, either directly or indirectly, by CD40, the molecule primarily responsible for T cell-driven B cell clonal expansion (102), which also up-regulates COX-2 (46, 60) and Mcl-1 (103, 104). Thus, the available data suggest that a TRAF → COX-2 → PGE2 → Mcl-1 pathway is shared by TNF family members that promote both innate immune system-driven and T cell-driven proliferation. Consistent with a critical role for this pathway in sustaining B cell clonal expansion is the heightened Mcl-1 expression within tonsil germinal centers (105).
It is of interest that both COX-2 (Fig. 6⇑D) and Mcl-1 (15) were highly down-regulated within apoptotic daughter cells within 1 day of the beginning of activation-related apoptosis on day 5 (15). The latter observations are consistent with a COX-2 half-life of ∼3 h and an Mcl-1 half-life of ∼1 to a few hours (106, 107, 108, 109). Both the rapid up-regulation and short half-life of Mcl-1 have been proposed to make it an effective checkpoint molecule for modulating cellular progression along a given pathway (107, 108). Undoubtedly, the relatively short half-life of its upstream regulator, COX-2 (t½ ∼3 h) (109), contributes to the fine-tuning of Mcl-1-based cell viability.
Importantly, several recent clinical investigations have demonstrated that heightened BAFF/APRIL expression is quite characteristic of sites of ectopic B cell growth, e.g., the inflamed joints, salivary glands, and brain of patients with rheumatoid arthritis, Sjogren’s syndrome, and multiple sclerosis, respectively (110, 111, 112, 113, 114). Given the present observations, one might predict that replicating B cells in these sites (16, 17, 18, 19, 20) should express high levels of COX-2 and respond to autocrine and paracrine PGE2 within these localized environments. Indeed, it is of interest that the concentrations of PGE2 reported within rheumatoid arthritis synovial fluids, i.e., 0.1 to 1 nM (115, 116), and within the saliva of individuals with Sjogren’s syndrome (∼10 nM) (117), are at the lower threshold and well within, respectively, the dose range for augmenting viability/replication of human B2 cells. It is, therefore, possible that combination therapies that interfere with the function/formation of TACI (BCMA) and COX-2 might be useful in controlling existing autoimmune disease and in reducing the propensity for malignant B cell transformation, as in Sjogren’s syndrome (18, 19).
We thank Jyoti Patel for her expert performance of assays for PGE2 and Dr. Michael Brunda for useful comments during the preparation of this manuscript.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by National Institutes of Health Grant R01 AI052189 (to P.K.A.M.), and in part by the Intramural Program of the National Institutes of Health, National Institute of Allergy and Infectious Diseases.
↵2 Address correspondence and reprint requests to Dr. Patricia Mongini, Department of Rheumatology, Hospital for Joint Diseases, New York University Medical Center, 301 East 17th Street, New York, NY 10003. E-mail address:
↵3 Abbreviations used in this paper: BAFF, B cell-activating factor; APRIL, a proliferation-inducing ligand; TACI, transmembrane activator and calcium modulator and cyclophilin ligand interactor; COX-2, cyclooxygenase 2; dex, detran; TI, T cell independent; TD, T cell dependent.
- Received September 30, 2005.
- Accepted March 15, 2006.
- Copyright © 2006 by The American Association of Immunologists