Nonsteroidal anti-inflammatory drugs (NSAIDs) are widely used for the treatment of inflammatory diseases and target cyclooxygenases 1 and 2 (Cox-1, Cox-2) that are responsible for PG production. Newer Cox-2-selective drugs have been heavily prescribed to quench inflammation. Little is known about whether or not these drugs influence human B lymphocytes and their ability to produce Ab. We report herein that activated human B cells not only highly express Cox-2 and produce PGs, but that the NSAID indomethacin and Cox-2-selective drugs profoundly inhibit the ability of human B cells to produce IgG and IgM in vitro. Human blood B cells highly express Cox-2 mRNA and protein and produce PGs after activation with CD40L, pansorbin, or CD40L plus BCR engagement. Cox-2 is also highly expressed by human tonsil B cells, as shown by immunohistochemistry. Cox-inhibiting drugs modestly affect purified B cell proliferation but profoundly reduce Ab production. The ability of whole blood to produce IgM and IgG following stimulation is also strongly inhibited. In support that Cox-2 plays a seminal role in B lymphocyte Ab production, Cox-2 knockout mice have 64% less IgM and 35% less IgG than normal littermate controls. These findings support that NSAIDs and the new Cox-2-selective drugs have an unsuspected target, the B cell, and attenuate Ab production in humans. Use of NSAIDs may therefore influence autoantibody production in autoimmune diseases and may dampen humoral immunity in response to antigenic challenge/vaccination.
Prostaglandins are critical mediators of inflammation that affect both humoral and cell-mediated immune responses. For example, PGE2 enhances Ab production and promotes type 2 immune responses (1, 2). PGE2 has been shown to directly promote Ig class switching in B cells acting through the EP2 and EP4 PGE2 receptors (3). Although it has been well established that PGs derived from accessory cells, such as macrophages and fibroblasts, signal in a paracrine manner to B cells, what has not been known was whether human B cells could synthesize PGs and thus signal in an autocrine manner. The recent finding that activated T cells express cyclooxygenase-2 (Cox-2)3 (4), an inducible enzyme that catalyzes a series of reactions to generate PGs, led us to hypothesize that human B cells express Cox-2 and therefore synthesize PGs upon activation. Indeed, this hypothesis is supported by our previous findings that proinflammatory signals increase Cox-2 expression and PG production in a mouse biphenotypic B/macrophage cell type (5).
PG endoperoxide synthase, also referred to as Cox, catalyze a Cox (bis-oxygenase) reaction in which arachidonic acid is converted to PGG2, and a subsequent peroxidase reaction in which PGG2 undergoes a two-electron reduction to PGH2, the common precursor to all prostanoids, including PGE2, PGD2, PGF2α, PGI2, and thromboxane. Two forms of Cox are known: a constitutive type (Cox-1), present in most cells and tissues, supporting prostanoid synthesis required for tissue homeostasis, and an inducible isoenzyme (Cox-2) expressed in response to cytokines, growth factors, and stressors such as hypoxia and free radicals (6). Cox-2 is the predominant isoform contributing to high levels of PGE2 found in chronic inflammatory conditions. Interestingly, elevated Cox-2 levels have been reported in autoimmune diseases, such as systemic lupus erythematosus, where chronic inflammation persists at multiple sites in the body. This explains the clinical utility of highly selective Cox-2 inhibitors such as celecoxib (Celebrex) and rofecoxib (Vioxx) to reduce the pain associated with inflammation (7).
In the study reported herein, we demonstrate that human B cells strongly express Cox-2 mRNA and protein and produce PGs upon activation. In addition, nonsteroidal anti-inflammatory drugs (NSAIDs) (Cox-1/Cox-2 inhibitors) and the highly selective Cox-2 drugs markedly reduce Ab production. These studies suggest a novel autoregulatory pathway for B cell activation. In addition, such studies suggest that patients taking NSAIDs or selective Cox-2 inhibitors may have suboptimal Ab responses to vaccination.
Materials and Methods
Culture conditions and reagents
l-glutamine, and 50 μg/ml gentamicin. B cells were stimulated with nothing or with recombinant human CD40L (8), pansorbin (Staphylococcus aureus Cowen I strain; Sigma-Aldrich), or rabbit anti-human F(ab′)2 anti-IgM Ab (Jackson ImmunoResearch Laboratories). Arachidonic acid, (Nu-Chek Prep) dissolved in 100% ethanol, was diluted to working concentrations in culture medium.5 M 2-ME, 10 mM HEPES, 2 mM
Small molecule Cox inhibitors
Indomethacin, a Cox-1/Cox-2 inhibitor, was purchased from Sigma-Aldrich. SC-58125, a highly selective Cox-2 inhibitor, was purchased from Cayman Chemical. Experiments were also performed using NS-398, a highly selective Cox-2 inhibitor purchased from Cayman Chemical (data not shown). Results were very similar to that obtained with SC-58125. Stock concentrations (10 mM) of indomethacin dissolved in 95% ethanol were made fresh each day and diluted to working concentrations in culture medium. SC-58125 and NS-398 were dissolved in DMSO (10 mM stock stored at −20°C) and diluted to working concentrations in culture medium. Pharmacological doses of these drugs that have been previously shown to block PG production range from 5 to 40 μM (9, 10). Dose response experiments (1, 5, 10, 20, 40, and 80 μΜ) were conducted to determine the optimal dose for PG inhibition using the supplier’s protocol for the PG screening kit (Cayman Chemical).
B lymphocyte preparation
Normal human B lymphocytes were isolated from peripheral blood obtained from healthy donors at the University of Rochester. Ethical permission for blood collection was obtained from the Research Subjects Review Board at the University of Rochester. The isolation of highly purified B lymphocytes has been previously described (11). In brief, buffy coats were separated, and samples were diluted with an equal volume of 1× PBS. PBMCs were separated by Ficoll-Paque (Amersham Biosciences AB) gradient centrifugation. Leukocyte layers were pooled to single tubes and washed with PBS. CD19 Dynabeads (Dynal Biotech) were added to the PBMCs and placed on an orbital shaker at 4°C for 30 min. Then CD19 Dynabead cell rosettes were captured on a magnet (Dynal Biotech). The bead cell rosettes were resuspended in RPMI 1640 medium and transferred to a 2-ml tube. Of the volume used for CD19 Dynabeads, 25% was the volume of Detachabeads added to the rosettes. Detachabead CD19 (Dynal Biotech) is a polyclonal anti-F(ab′)2 to release cells from Dynabeads CD19. The Detachabeads with rosettes were placed on an orbital shaker for 1 h at 4°C. The CD19-positive cells were washed and counted. Viability was determined using the trypan blue exclusion method. B lymphocytes isolated in this manner are >98% surface CD19 positive and <2% CD3 or CD14 positive (as determined by flow cytometry).
To determine the percentage of B lymphocytes expressing Cox-1 and Cox-2, whole blood or freshly isolated B cells were incubated in the presence or absence of CD40L plus 10 μg/ml anti-IgM Ab for 24 h. RBC were lysed using 1× FACS lysing solution (BD Biosciences) before staining with Abs. Whole blood or purified B lymphocytes were stained with mouse anti-human CD19 APC (BD Biosciences) in cold PBS buffer with sodium azide (0.02%) and BSA (0.3%) for 20 min at 20°C. Intracellular staining was performed with a fixation and permeabilization kit (Caltag Laboratories) using the supplier’s protocol. Intracellular Cox-1 and Cox-2 proteins were detected by PE-labeled mouse anti-human Cox-1 Ab (Cayman Chemical) and FITC-labeled mouse anti-human Cox-2 Ab (Cayman Chemical). The percentage of CD19-positive B cells expressing Cox-1 and Cox-2 was determined by flow cytometric analysis using Cell Quest software on a FACSCalibur flow cytometer (BD Biosciences).
CD40L plus anti-IgM (5 × 104)-activated B cells were cytospun onto glass microscope slides and stained for surface CD19 and intracellular Cox-2 expression as described above for immunofluorescence. Images were visualized on an Olympus BX51 system microscope, and photographs were taken using SPOT camera with SPOT RT software (SPOT Diagnostic Instruments).
Paraffin-embedded tonsil tissue sections (4 μm) were obtained from Department of Pathology and Laboratory Medicine at the University of Rochester. Slides were processed for immunohistochemistry. In the primary Ab reaction step, slides were incubated with the monoclonal mouse anti-human CD20 Ab (Zymed Laboratories) or the polyclonal mouse anti-human Cox-2 Ab (Cayman Chemical). Multiple tonsil tissue sections were stained. For positive controls, sections of prostate tissue expressing the Cox-2 protein were included in each staining procedure. Images were visualized on an Olympus BX51 system microscope, and photographs were taken using SPOT camera with SPOT RT software.
Real time RT-PCR
Total RNA was extracted using Qiagen RNAeasy mini kit according to the supplier’s protocol. RNA was measured on a Bio-Rad SmartSpec 3000. RNA (1 μg) was reverse transcribed using Moloney murine leukemia virus RT primed by an oligo(dT) primer. PCRs for human Cox-1, Cox-2, and GAPDH were performed using Platinum Taq
Total cellular protein was harvested from rigorously purified B lymphocytes immediately following isolation or after culture for 24 h. B cells were lysed in protein isolation buffer (1% IGEPAL, 150 mM sodium chloride, 50 mM Tris, 10% protease inhibitor mixture) and quantified by bicinchoninic acid protein assay kit (Pierce). Thirty micrograms of protein was fractionated by 7.5% SDS-PAGE and electrophoretically transferred onto nitrocellulose membranes. After blocking in 10% nonfat powdered milk in 1× PBS/0.1% Tween 20, membranes were incubated with either a primary mouse anti-human Cox-2 (Cayman Chemical), mouse anti-human Cox-1 (Cayman Chemical), or a mouse anti-actin (Oncogene Research Products) Ab for 2 h at 20°C. After washing in PBST buffer, membranes were incubated with secondary goat anti-mouse-HRP conjugated Ab (Jackson ImmunoResearch Laboratories) for 1 h at 20°C. Bands were visualized with ECL (Amersham Biosciences) and autoradiography film according to the manufacturer’s protocol.
Measurement of PG production
Triplicate samples of B lymphocyte cell culture supernatants were analyzed by highly specific competitive enzyme immunoassays (EIA) (Cayman Chemical) to determine the concentration of PGE2 and PGF2α after 36 h of incubation.
B lymphocytes were cultured in flat-bottom 96-well plates. Cells were incubated for 48 h to 5 days in triplicate at 4 × 105 cells/ml. Arachidonic acid (5 μM) was added to cells that were stimulated with CD40L plus anti-IgM Ab. Cells were treated with nonselective Cox-1/Cox-2 and highly selective Cox-2 inhibitors and compared with untreated (vehicle) cells as a control. Plates were harvested after 2, 3, 4, and 5 days, and incorporation of [3H]thymidine was determined for a 24-h period by scintillation spectroscopy.
Mitochondrial membrane permeability
One million B lymphocytes were incubated in duplicate with CD40L plus anti-IgM Ab and treated with vehicle (DMSO or ethanol) or drug (up to 20 μM indomethacin or 20 μM SC-58125). After 48 h, 40 nM 3,3′-dihexyloxacarbocyanine iodide (DiOC6) (3) (Molecular Probes) was diluted in culture medium and added to the cells for 15 min. Cells were harvested, washed in 1× PBS, and immediately analyzed on a FACSCalibur flow cytometer (BD Biosciences). Cells with intact mitochondrial membrane potential incorporate DiOC6 into the mitochondria.
Ig production assays
Highly purified B lymphocytes (5 × 10512). Supernatants were harvested and tested for IgM and IgG levels as described above for highly purified B cells.
Ptgs2-knockout (Cox-2 deficient) mouse studies
Cox-2-deficient (B6.129P2-Ptgs2tm1Smi) mice (8- to 12-wk-old) and their wild-type barrier colony controls were purchased from Taconic Farms. The Animal Care and Use Committee of the University of Rochester approved all mouse protocols. Mice were anesthetized with sodium pentobarbital (60 mg/kg). Eye bleeds were performed to harvest peripheral blood in heparin-coated hematocrit tubes. Whole blood was centrifuged, and plasma was isolated for analysis of Ig levels by mouse-specific ELISAs (Bethyl Laboratories). Mouse B lymphocytes were isolated as previously described (13). In brief, normal mouse splenic B lymphocytes were isolated from wild-type and Cox-2-deficient mice. Erythrocytes were lysed with ammonium chloride buffer, and the suspension was depleted of adherent cells by two successive rounds of adherence on Falcon tissue culture dishes (BD Biosciences) at 37°C for 2 h. Nonadherent cells were collected by rinsing the plates with warm medium. T cells were depleted with anti-T cell mixture consisting of Abs 30-H12 (anti-Thy 1.2), 3.155 (anti-CD8), and GK 1.5 (anti-CD4) and low toxicity baby rabbit complement (Cedarlane Laboratories) followed by incubation at 37°C. The cells were washed and counted, and viability was determined using the trypan blue exclusion method. B lymphocytes isolated in this manner are >97% surface B220 positive.
Activation of human B lymphocytes up-regulates Cox-2 protein expression
The expression of Cox-1 and Cox-2 on healthy donor human B lymphocytes from whole blood (defined by their expression of CD19) was determined by a sensitive and rapid flow cytometric assay (14). We first used this technique to evaluate the expression of Cox-1 and Cox-2 in unstimulated whole human blood B cells. One hundred percent of the CD19+ gated cells from freshly isolated whole blood expressed Cox-1, although <2% expressed Cox-2 (Fig. 1⇓A). In contrast, after 24 h ex vivo stimulation with CD40L plus anti-IgM Ab at 37°C, 96% of the CD19+ gated cells expressed Cox-2. The expression of Cox-1 was no different between the unstimulated and stimulated B lymphocytes.
Encouraged by the findings of constitutive Cox-1 and inducible Cox-2 expression on human B lymphocytes, we next studied Cox-2 expression on purified B cells. Freshly isolated B lymphocytes, as well as B lymphocytes stimulated in vitro for 24 h (with CD40L plus anti-IgM Ab), were stained for surface CD19 expression and for intracellular Cox-1 and Cox-2 expression (Fig. 1⇑B). Density plot analysis indicates that there was no change in the percentage of Cox-1-positive cells before and after stimulation. In contrast, only 3% of freshly isolated B cells expressed Cox-2, whereas after 24 h of in vitro stimulation, >95% of activated B cells were Cox-2 positive (Fig. 1⇑B). In Fig. 1⇑C, cytospin preparations of highly purified activated B cells were stained for surface CD19 (red) and intracellular Cox-2 expression (green). Dual-color fluorescence microscopy was next used to confirm that activated B cells were in fact positive for Cox-2. CD19+ B cells expressing Cox-2 are shown as red and green cells.
Since in vitro activated B cells expressed Cox-2, we next investigated whether there was a population of activated B cells that express Cox-2 in vivo. To assess this, histologically normal tonsil tissue (obtained from tonsillectomy samples) was stained with Abs against Cox-2 and CD20 (L-26 clone) in formalin-fixed, paraffin-embedded, sequential tissue sections. As Fig. 2⇓ demonstrates, Cox-2 was coexpressed with CD20 on B lymphocytes.
Constitutive Cox-1 and inducible Cox-2 mRNA and protein levels from activated human B lymphocytes
To determine whether the increase in Cox-2 protein expression after B cell activation is due to an increase in Cox-2 mRNA, total RNA from highly purified human peripheral blood B cells was next analyzed by real time RT-PCR analysis. Quantitative analysis of Cox-1 and Cox-2 mRNA expression in human B lymphocytes was performed by normalization of cycle threshold values for the target genes to GAPDH mRNA levels. Fig. 3⇓A (gel electrophoresis) first shows that steady-state Cox-1 mRNA levels were constitutively expressed and remained unchanged during B cell activation. These findings are consistent with the cellular maintenance functions of Cox-1. In contrast, Cox-2 mRNA was not detectable in unactivated B cells (lane 1). However, activation of B cells with CD40L and/or anti-IgM Ab (as single or combined treatments) resulted in the substantial induction of steady-state Cox-2 mRNA levels (lanes 2–4). A significant 4-fold induction of Cox-2 mRNA was seen as early as 4 h following treatment with stimuli (data not shown). After 8 h of activation, a 9-fold increase in Cox-2 mRNA compared with that of nonactivated B cells was seen as shown in Fig. 3⇓B.
We next analyzed the expression pattern of Cox-1 and Cox-2 proteins from freshly isolated blood B cells by Western blot analysis. As shown in Fig. 3⇑C, unactivated B cells expressed Cox-1 but not Cox-2. These results are consistent with the flow cytometry results for Cox-1/Cox-2 shown in Fig. 1⇑B. Consistent with our quantitative PCR data, Cox-2 protein expression was highly up-regulated in B lymphocytes activated by pansorbin, CD40L, anti-IgM Ab, or by a combination of signals from CD40L plus anti-IgM Ab (Fig. 3⇑C). Both Cox-1 and Cox-2 had the same mass (70 and 72 kDa, respectively), as the Cox-1 and Cox-2 enzyme from IL-1β-stimulated L828 human fibroblasts. β-Actin levels were used in densitometry analysis as a protein-loading control. Densitometry of Western blots shows a 25- to 30-fold relative increase in Cox-2 protein levels in activated B cells compared with that of untreated controls after 24 h (Fig. 3⇑D). As expected, Cox-1 protein expression was unchanged following activation with B cell stimuli. Taken together with the data presented above, our findings demonstrate that B cell stimulation through either CD40 or the BCR transduces a signal(s) that up-regulates both Cox-2 mRNA and protein expression.
B lymphocytes synthesize PGE2 and PGF2α, via Cox-2
Given that activated B cells express Cox-2 and that Cox-2 catalyzes a series of reactions resulting in PG synthesis, we next determined whether activated B cells have the capacity to produce and secrete the PGs, PGE2 and PGF2α. Freshly isolated B cells were incubated with and without the nonspecific B cell activator pansorbin or with CD40L ± anti-IgM Ab for 36 h. The supernatants were then assayed for PGE2 and PGF2α using a specific EIA analysis. In Fig. 4⇓A, PGE2 levels in the medium were dramatically increased up to 10-fold following B cell treatment with pansorbin, CD40L, and/or anti-IgM Ab (single or dual stimulation). Similarly, PGF2α levels were increased upon B cell activation, although not to the same extent as PGE2. To determine whether or not PGE2 and PGF2α production was synthesized mainly via Cox-2, B lymphocytes were treated with either indomethacin, a nonselective Cox-1/Cox-2 inhibitor, or with SC-58125, a highly selective Cox-2 inhibitor (15). As shown in Fig. 4⇓B, indomethacin blocks the production of nearly all the PGE2 and PGF2α. Treatment with the Cox-2-selective inhibitor, SC-58125, resulted in ∼85% decrease in PGE2 and PGF2α produced, indicating that the majority of PGE2 and PGF2α were produced by the inducible Cox-2 (Fig. 4⇓B). The use of another selective Cox-2 inhibitor, NS-398, supports these findings as this drug also reduced PG synthesis by up to 85% of the PGs produced (data not shown). These data demonstrate that Cox-1/Cox-2 inhibitors significantly blunt PGE2 and PGF2α production seen after B cell activation.
Cox-2 inhibition dramatically reduced IgM and IgG production by activated B lymphocytes
When B cells are activated, they differentiate into Ig-secreting plasma cells (16). Ig production by human B cells was assayed by measuring the amount of Ig present in culture supernatants of B cells stimulated by selected activators. As expected, when B lymphocytes were incubated with CD40L plus anti-IgM, 3300 ± 205 ng/ml and 2033 ± 328 ng/ml IgM and IgG, respectively, were released into the culture supernatants after 7 days. PGs were previously reported to enhance Ab production by promoting class switching in B cells (17). However, it has not been determined whether or not B cell-derived PGs promote Ab production.
Since we showed that human B cells expressed Cox-2 upon activation and produced PGs, we hypothesized that these PGs enhanced Ab production in an autocrine fashion. To test this, we evaluated Ab production after activation of purified B cells in the presence of the Cox-1/Cox-2 inhibitor indomethacin (added to culture every other day for 7 days) or a highly selective Cox-2 inhibitor SC-58125 (added once at start of culture). The data in Fig. 5⇓A demonstrate that in vitro activated B cells failed to optimally produce Ab when incubated with indomethacin or SC-58125, compared with vehicle-treated cells. A 50% reduction in IgM and a 95% reduction in IgG synthesis were measured following inhibition of Cox-1/Cox-2 in CD40L plus anti-IgM Ab-activated B cells, with similar results seen with CD40L plus pansorbin-stimulated cells. The selective Cox-2 inhibitor NS-398 reduced Ab production to the same extent as SC-58125 (data not shown). These compelling findings demonstrate that human B cell Cox-2 is an important autocrine signal for promoting optimal Ig production in vitro.
Given that Cox-1/Cox-2 inhibitors greatly reduced Ab production by purified B cells, we next investigated whether or not these drugs reduced Ab production by whole blood ex vivo. Whole human blood was washed two times to remove plasma and was then diluted in culture medium (see Materials and Methods). Ig production was measured after whole blood was activated and treated with or without Cox-1/Cox-2 inhibitors for 7 days. Fig. 5⇑B shows that activation of whole blood with CD40L plus anti-IgM Ab (or pansorbin) increased IgM and IgG levels compared with that of unactivated blood. As anticipated from the data with purified B cells, a significant reduction in IgM (up to 80%) and in IgG (up to 90%) levels was seen following treatment with Cox-1/Cox-2 inhibitors. Herein, our findings reveal that B cell activation-induced Cox-2 activity and the subsequent increase in PG production are important immunoregulatory signals for optimal Ab production by purified B lymphocytes in vitro and by whole blood cells ex vivo.
Cox inhibitors modestly reduce B lymphocyte proliferation
Pharmacological concentrations of indomethacin and the selective Cox-2 inhibitors that inhibited PG synthesis (see Fig. 4⇑B) and reduced Ab production (see Fig. 5⇑) were used to test the ability of Cox inhibitors to affect B cell proliferation using a [3H]thymidine incorporation assay. Cox inhibitors did not influence B cell proliferation in response to CD40L or anti-IgM Ab stimulation (data not shown). However, both the nonselective and selective Cox-2 inhibitors, indomethacin and SC-58125, modestly reduced proliferation of CD40L plus anti-IgM Ab stimulated B cells. Treatment with Cox inhibitors showed a significant 35% decrease in proliferation after 48 h and a 25% reduction after 4 and 5 days in culture as compared with vehicle treated cells (Fig. 6⇓A). The reduction in proliferation following Cox-1/Cox-2 inhibition on days 2, 4, and 5 were also seen by MTT assay (data not shown).
We next wanted to determine whether or not the reduction in B cell proliferation (as shown by [3H]thymidine incorporation and the MTT assay) was due to an increase in early apoptotic cells or cell death. Therefore, the DiOC6 incorporation assay was used to measure the percentage of live and dead B cells following Cox-1/Cox-2 inhibition by flow cytometry. Early apoptotic and dead cells have a permeable mitochondrial membrane and do not incorporate DiOC6. As shown in Fig. 6⇑B, there was no significant change in the percentage of CD40L plus anti-IgM Ab-stimulated B cells that incorporate DiOC6 following treatment with indomethacin or SC-58125 for 48 h. There was no change in the number of dying/dead cells after 3, 4, and 5 days of incubation with drug (data not shown). Thus, indomethacin and SC-58125 reduced day 4 proliferation of CD40L plus anti-IgM Ab-stimulated B lymphocytes, but did not induce apoptosis or B cell death. These data reveal that the Cox inhibitory drugs did not elicit mitochondrial membrane permeability, a hallmark of apoptosis, and suggest that the modest reduction in proliferation may account for some, but not all of the dramatically reduced IgM and IgG levels (Fig. 5⇑).
Reduced IgM and IgG levels in peripheral blood of Cox-2-deficient mice
We next determined whether or not B cell Cox-2 activity was important for Ab production in vivo by comparing peripheral blood plasma Ig levels in Cox-2-deficient mice with that of normal littermate controls. Peripheral blood was harvested in heparinized hematocrit tubes (see Materials and Methods). Cox-2-deficient mice (n = 5) had significantly lower levels of IgM (p < 0.01) and IgG (p < 0.05) in their plasma compared with that of control mice (n = 5) exposed to the same diet and environment. As shown in Fig. 7⇓B, we determined that the reduction in plasma Ab levels in Cox-2-deficient mice was likely due to a defect in the B lymphocyte. In support of this, in vitro LPS-stimulated B cells from Cox-2-deficient mouse spleen produced ∼75% less IgM and ∼60% less IgG compared with wild-type control. It has been previously reported that elevated Cox-1 levels may compensate for the loss of Cox-2 in Cox-2-deficient mice compared with wild-type controls (18). Our findings suggest that elevated Cox-1 activity may account for the slightly elevated IgM levels produced by the unstimulated Cox-2 knockout B cells after 7 days in culture. However, the in vitro LPS-stimulated Cox-2 knockout B cells show a marked reduction in IgM levels compared with wild-type controls, which supports our hypothesis that the increase in Cox-2 activity by activated B lymphocytes is required for optimal Ab production. These in vivo and in vitro findings using a Cox-2-deficient model provide strong support for the human B cell data, which demonstrate that Cox-2 is required for optimal Ab production.
The CD40-CD40L interaction and BCR cross-linking signal a series of complex events essential for B cell activation, proliferation, and differentiation to Ab-producing cells (16, 19). Our findings that activated B cells express Cox-2, secrete PGE2 and PGF2α, and produce Ig in a PG-dependent fashion supports a novel mechanism through which CD40-CD40L interaction and BCR cross-linking activate normal human B cells. The results presented here also demonstrate that the Cox-2 isoform is preeminent over Cox-1 for the stimulated production of PG products by activated human B lymphocytes. Understanding the mechanisms involved in regulating Cox activity in B lymphocytes has clinical significance because pharmacological agents that inhibit Cox enzyme activity are commonly used for acute and chronic inflammatory conditions (15). We report for the first time that there are direct effects of Cox-1/Cox-2 inhibition, as well as selective Cox-2 inhibition, on B lymphocytes. Activated B cells have not been considered primary targets for NSAIDs and Cox-2-selective inhibitors because normal B cells were not previously known to express Cox-2 or to produce PGs.
Following exposure to CD40L plus anti-IgM Ab (or pansorbin) for 7 days, B lymphocytes become activated, proliferate, and differentiate, thus increasing their production of IgM and IgG (20). Our findings of drastically reduced IgM and IgG levels following incubation with indomethacin or the selective Cox-2 inhibitor SC-58125 supports the concept that the B cell Cox-2-derived products are required for optimal Ab production. A few studies have begun to address whether or not Cox inhibitory drugs influence humoral immune responses in animal models. For example, a study using DBA/1J mice demonstrated that in vivo administration of indomethacin, a nonselective Cox-1/Cox-2 inhibitor, following OVA immunization, reduced OVA-specific IgG Ab production and splenocyte proliferation (21). Comparable results were obtained in a Mycobacterium butyricum-induced adjuvant arthritis rat model, which showed a significant reduction in plasma anti-Mycobacterium IgG levels following in vivo administration of two different selective Cox-2 inhibitors (22). These findings suggested an important role for Cox-2-derived products in promoting a humoral immune response. Finally, an in vitro study using pokeweed mitogen-stimulated human PBMCs demonstrated a significant reduction in IgM and IgG production following treatment with the Cox-1/Cox-2 inhibitor indomethacin (23). These studies support the concept that Cox-1/Cox-2-derived products are important signals for Ab production in humans. However, the report of decreased Ig levels following indomethacin treatment of stimulated PBMCs did not investigate whether there were direct effects of these drugs on B cells. Rather, these reports suggested an indirect effect on B cells as a result of reduced macrophage or T cell-derived factors (23). We conclude that the decrease in Ig production is not solely due to decreases in macrophage and T lymphocyte-associated Cox activity but rather due to reduced PG synthesis by B cells. We have observed up to a 15% reduction in syndecan-1 (CD138) expression in B cells activated with CD40L plus anti-IgM Ab and treated with Cox-2 inhibitors (data not shown). We interpret these data as supporting the hypothesis that Cox-1/Cox-2 inhibitors reduce B cell Ab production, in part, by attenuating B cell differentiation to plasma cells.
PGE2 modulates many aspects of inflammation and immune responses including inhibition of IL-2 synthesis by T cells, inhibition of IL-12 production in macrophages and dendritic cells, and increased vascular permeability and vasodilation (24, 25, 26). Our laboratory previously reported that PGE2 enhances CD40L and IL-4-induced mouse B lymphocyte production of IgG1 and IgE (27). However, it is not clear at present whether these same PG-induced class-switch mechanisms occur in human B cells. Our findings suggest that human B cell-derived PG products signal by an autocrine mechanism to enhance B cell proliferation and Ab production. Further investigation of the mechanisms by which B cell Cox-2-derived products modulate human B cell effector functions is necessary. Based on our results of dramatically reduced IgG levels following Cox-2 inhibition, we speculate that most cells failed to undergo class switching required for IgG synthesis. Our data suggest that the widespread use of NSAIDs and Cox-2-selective drugs may alter B cell function and the immune response in ways that are not yet fully appreciated.
Emerging data show that the growth-promoting properties of Cox-2 in normal physiological responses is exploited in malignancy and autoimmune disorders. Wun et al. (28, 29) recently reported elevated Cox-2 expression in human B cell lymphomas. Cox-2 was also implicated as playing a role in extranodal marginal cell lymphomas of mucosa-associated lymphoid tissue (30). These low-grade B cell malignancies are thought to arise in a background of chronic B cell activation, although development has been linked to several chronic bacterial and/or microorganism infections (31). We speculate that the increased Cox-2 levels seen in B cell lymphomas were a result of inducible Cox-2 expression by normal activated B cells before their conversion to a malignant state. However, the mechanisms by which excess eicosanoid production by elevated Cox-2 promotes malignancy are unknown. Interestingly, anti-apoptotic functions of Cox-2 were recently reported in human lupus T cells as Cox-2 inhibitors were reported to reduce cell viability. These drugs were also capable of suppressing IgG autoantibody production by mixed lymphocyte cultures in mouse lupus models, whereas Cox-2-deficient mice were incapable of developing autoimmune arthritis (32, 33). Our compelling finding of reduced Ab production by Cox-2-deficient mice and following selective Cox-2 inhibition of human B cells suggests that these agents may be suppressing autoantibody production in lupus models via direct effects on B cells. Thus, it will be important to further evaluate these drugs as potential therapeutic agents to control abnormal B lymphocyte proliferation seen in non-Hodgkin lymphoma as well as the abnormal Ab production seen in autoimmune diseases.
The findings reported herein also have important implications for the use of Cox-1/Cox-2 inhibitory drugs following vaccinations, where the goal is to promote a humoral immune response. Although these drugs are commonly used to alleviate the pain associated with injection of the vaccine, our findings suggest that there may be an adverse effect on Ab production and/or the immune response following secondary exposure. Because this report is the first to show Cox-2 activity in human B lymphocytes and to demonstrate the important role of these enzymes in promoting Ab production, additional studies are warranted to study the effects of NSAIDs and selective Cox-2 inhibitors on human B lymphocyte effector functions in vivo.
The authors have no financial conflict of interest.
We thank Pat Bourne for assistance with the human tonsil tissue staining.
↵1 This work was supported by U.S. Public Health Service Grant DE011390, The Leukemia Society of America, Environmental Health Sciences Center Grant P30ES01247, Toxicology Training Grant T3ES07026, and the James P. Wilmot Discovery Fund.
↵2 Address correspondence and reprint requests to Dr. Richard P. Phipps, Box 850, Department of Environmental Medicine, University of Rochester School of Medicine and Dentistry, 601 Elmwood Avenue, Rochester, NY 14642. E-mail address:
↵3 Abbreviations used in this paper: Cox, cyclooxygenase; EIA, enzyme immunoassay; NSAID, nonsteroidal anti-inflammatory drug.
- Received September 15, 2004.
- Accepted December 27, 2004.
- Copyright © 2005 by The American Association of Immunologists