Drs. Sibilia and Wachsmann contributed equally to this work.
BAFF synthesis by rheumatoid synoviocytes is positively controlled by α5β1 integrin stimulation and is negatively regulated by tumor necrosis factor α and toll-like receptor ligands
Version of Record online: 28 SEP 2007
Copyright © 2007 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 56, Issue 10, pages 3202–3214, October 2007
How to Cite
Alsaleh, G., Messer, L., Semaan, N., Boulanger, N., Gottenberg, J.-E., Sibilia, J. and Wachsmann, D. (2007), BAFF synthesis by rheumatoid synoviocytes is positively controlled by α5β1 integrin stimulation and is negatively regulated by tumor necrosis factor α and toll-like receptor ligands. Arthritis & Rheumatism, 56: 3202–3214. doi: 10.1002/art.22915
- Issue online: 28 SEP 2007
- Version of Record online: 28 SEP 2007
- Manuscript Accepted: 22 JUN 2007
- Manuscript Received: 7 DEC 2006
- Bristol-Myers Squibb, Roche, and Pfizer
It was recently demonstrated that synoviocytes (FLS) from rheumatoid arthritis (RA) patients express BAFF transcripts that are up-regulated by tumor necrosis factor α (TNFα) and interferon-γ (IFNγ). Thus, BAFF increases in RA target cells might be related to activation of the receptors of innate immunity. The purpose of this study was to determine whether ligands of Toll-like receptor 2 (TLR-2), TLR-4, TLR-9, and α5β1 integrin are able to induce BAFF synthesis by RA FLS.
Quantitative reverse transcription–polymerase chain reaction analyses and enzyme-linked immunosorbent assays were performed to evaluate BAFF messenger RNA induction and BAFF release from FLS after stimulation by ligands for TLR-2, TLR-4, TLR-9, α5β1 integrin (bacterial lipopeptide [BLP] palmitoyl-3-cysteine-serine-lysine-4, lipopolysaccharide [LPS], CpG, and protein I/II, respectively), TNFα, and IFNγ.
In contrast to IFNγ, neither TNFα, LPS, BLP, nor CpG induced the de novo synthesis and release of BAFF by FLS. Priming of cells with IFNγ did not have a synergistic effect on BAFF synthesis by FLS stimulated with bacterial products known as pathogen-associated molecular patterns. Moreover, we found that IFNγ-induced BAFF synthesis is inhibited by simultaneous stimulation with either TLR ligands or TNFα. We also showed that interplay between TLRs, TNF receptors, and IFNγ signaling induces the expression of suppressor of cytokine signaling 1 (SOCS-1) and SOCS-3 and reduces IFNγ-dependent STAT-1 phosphorylation, which might explain this inhibition. In contrast, we demonstrated that stimulation of α5β1 integrin can induce BAFF synthesis and release per se and that stimulation of this pathway has no inhibitory effect on IFNγ-induced BAFF synthesis.
Our findings indicate that BAFF secretion by resident cells in target organs of autoimmunity is tightly regulated by innate immunity, with positive and negative controls, depending on the receptors and the pathways triggered.
BAFF is known to play a central role in the maturation and survival of B cells as well as in antibody synthesis (1). BAFF is produced as a membrane form or is secreted by cells of hematopoietic origin, essentially, monocytes, dendritic cells, macrophages, and stimulated neutrophils (2, 3). It exerts its activity by interacting with 3 receptors: the transmembrane activator and calcium-modulator and cyclophilin ligand activator, or TACI, the B cell maturation antigen, and the BAFF receptor. These receptors are present almost exclusively on B lymphocytes, although some populations of T cells can also express them (4).
The finding that BAFF-transgenic mice develop autoimmune manifestations with similarities to systemic lupus erythematosus and Sjögren's syndrome in humans suggested a critical role of BAFF in autoimmune diseases (5–7). Elevated levels of BAFF have been detected in serum and synovial fluid samples from patients with rheumatoid arthritis (RA), and it was recently demonstrated that overproduction of BAFF by dendritic cells and macrophages may play a crucial role in the pathogenesis of experimental arthritis (8, 9).
Interestingly, BAFF has also been found to be expressed by cells of mesenchymal origin. This cytokine, which is present at low levels in normal human brain tissue as compared with lymphoid tissue, is constitutively produced by astrocytes (10). Its production is increased in sclerotic cerebral lesions and in cerebral primary lymphomas. This overexpression is dependent on the presence of different proinflammatory factors, such as tumor necrosis factor α (TNFα) and interferon-γ (IFNγ). Ohata et al (11) recently demonstrated that fibroblast-like synoviocytes (FLS) from RA patients express BAFF transcripts that are equally strongly up-regulated by TNFα and IFNγ. Since previous studies have shown that FLS participate in the formation of synovial pseudofollicles by secreting stromal cell–derived factor 1 and CXCL13, which attract B cells to the synovial cavity, BAFF expression by FLS points out the importance of these resident cells as a source of long-term B cell survival factor (12–14).
Most studies have focused on the role of cytokines or classic proinflammatory mediators on BAFF expression. Among the cytokines tested, only IFNγ and, to a lesser extent, TNFα induced BAFF release from stimulated astrocytes. This induction was strongly increased with a combination of the 2 cytokines (10). Concerning FLS, higher BAFF levels were obtained after treatment with either IFNγ or TNFα. An important notion is that trace amounts of IFNγ could prime FLS and enhanced their sensibility to TNFα, which is mainly present in the RA synovium. Other cytokines, such as interleukin-1β (IL-1β), IL-15, IL-17, and IL-18, failed to induce BAFF synthesis de novo (11).
However, it cannot be excluded that environmental factors or other factors may also account directly or indirectly for BAFF secretion. Bacterial products known as pathogen-associated molecular patterns (PAMPs), such as lipopolysaccharide (LPS) or peptidoglycan, are known to activate FLS by interacting with cellular pattern-recognition receptors (PRRs) present on these cells (15, 16). A large number of PRRs, such as Toll-like receptor 2 (TLR-2), TLR-4, and TLR-9, as well as numerous integrins are expressed by FLS, and their expression is increased in response to inflammatory stimuli (16, 17). We previously showed that LPS, bacterial lipopeptide (BLP), and protein I/II, which are ligands of TLR-4, TLR-2, and α5β1 integrin, respectively, can trigger the release of IL-6 and IL-8 from activated FLS (18, 19).
In the present study, using FLS isolated from RA patients, we investigated the capacity of these PAMPs as well as unmethylated CpG motifs, which are ligands of TLR-9, IFNγ, and TNFα, to induce the synthesis and release of BAFF. Our data indicate that intracellular signaling pathways activated by TLR ligands or TNFα negatively regulate IFNγ-induced BAFF expression and release, whereas integrin signaling pathways stimulate BAFF secretion by resident cells of the target organs of autoimmunity.
MATERIALS AND METHODS
Cell culture media (RPMI 1640 and medium 199), fetal calf serum, penicillin, streptomycin, and amphotericin B were obtained from Invitrogen (Cergy-Pontoise, France). Human recombinant IFNγ was purchased from BD PharMingen (Le Pont de Claix, France). Synthetic BLP, palmitoyl-3-cysteine-serine-lysine-4 (Pam3CSK4), was obtained from EMC Microcollections (Tübingen, Germany). Streptococcal protein I/II was prepared as previously described (20). Potential endotoxins in the protein I/II preparation were removed using polymyxin B–agarose (Pierce, Rockford, IL) according to the manufacturer's recommendations. Protein I/II had an endotoxin content <0.01 ng/125 pM, as tested by the Limulus chromogenic assay. Purified phosphorothioate-modified CpG, including 5′-GGG-GGA-CGA-TCG-TCG-GGG-GG-3′ ODN-2216/type A CpG DNA and 5′-TCG-TCG-TTT-TGT-CGT-TTT-GTC-GTT-3′ ODN-2006/type B CpG DNA, were from Operon Technologies (Cologne, Germany). LPS from Salmonella abortus equi and type XI collagenase were obtained from Sigma (St. Quentin Fallavier, France). Nucleospin RNA II extraction kit was from Macherey-Nagel (Souffelweyersheim, France). The LightCycler FastStart DNA Master SYBR Green I kit was from Roche Applied Science (Penzberg, Germany). The enzyme immunoassay kits for human BAFF and IL-6 detection and human recombinant TNFα were from R&D Systems (Lille, France). The RayBio Cell-Based STAT-1 (Tyr701) enzyme-linked immunosorbent assay (ELISA) kit was from Tebu-Bio (Le Perray en Yvelines, France). Throughout this study, buffers were prepared with apyrogenic water obtained from Braun Medical (Boulogne, France).
Human FLS were isolated from RA synovial tissues obtained from 3 different patients at the time of knee joint arthroscopic synovectomy, as described previously (21). The diagnosis of RA was conformed according to the revised criteria of the American College of Rheumatology (formerly, the American Rheumatism Association) (22). FLS cultures were performed as previously described (23). FLS between the third and the ninth passages were used in the experiments. During that time, cultures consisted of a homogeneous population of fibroblastic cells that were negative for CD16, as determined by fluorescence-activated cell sorter analysis. Cell number and cell viability were determined by MTT assay as described elsewhere (24).
Stimulation of cells for total RNA extraction.
FLS (106 cells) were stimulated with 2 ml of complete medium containing the various activators: IFNγ (0.01, 0.1, and 1 ng/ml), TNFα (10 ng/ml), LPS (1 μg/ml), BLP (1 μg/ml), CpG A (10 μg/ml), CpG B (10 μg/ml), and protein I/II (125 pM). After a 24-hour or a 72-hour incubation period, cells were trypsinized and centrifuged (130g for 10 minutes, 4°C). Total RNA was extracted from the cell pellets using a Nucleospin RNA II extraction kit according to the manufacturer's instructions.
Stimulation of cells for BAFF assay.
FLS (2 × 105 cells) were stimulated with 1 ml of complete medium containing the various activators for 24, 48, and 72 hours. BAFF release was measured in culture supernatants by a heterologous 2-site sandwich ELISA according to the manufacturer's instructions.
Real-time quantitative reverse transcription–polymerase chain reaction (RT-PCR).
Total RNA isolated from FLS was reverse transcribed using a first-strand complementary DNA synthesis kit (Invitrogen) according to the manufacturer's instructions. Real-time quantitative RT-PCR was performed in 96-well plates in a total volume of 20 μl using a LightCycler FastStart DNA Master SYBR Green I kit and the following gene-specific primers: for BAFF (0.25 μM), 5′-TGA-AAC-ACC-AAC-TAT-ACA-AAA-AG-3′ and 5′-TCA-ATT-CAT-CCC-CAA-AGA-CAT-3′; and for GAPDH (0.4 μM), 5′-GGT-GAA-GGT-CGG-AGT-CAA-CGG-A-3′ and 5′-GAG-GGA-TCT-CGC-TCG-CTC-CTG-GAA-GA-3′. After an initial incubation at 96°C for 10 minutes, samples were subjected to 40 rounds of amplification for 10 seconds at 95°C, 15 seconds at 60°C, and 20 seconds at 72°C for BAFF and an initial incubation at 95°C for 10 minutes, with 35 rounds of amplification for 15 seconds at 95°C, 10 seconds at 57°C, and 10 seconds at 72°C for GAPDH, using a LightCycler instrument (Roche Applied Science). Amplification products were detected as an increased fluorescent SYBR Green signal during the amplification cycles. Results were obtained using SDS Software (PerkinElmer Applied Biosystems, St. Quentin, France) and evaluated using Excel software (Microsoft, Redmond, WA). Melting-curve analysis was performed to assess the specificity of the PCR products.
Total RNA extracted from 106 human FLS was incubated for various times with IFNγ (0.1 ng/ml) or with either LPS (1 μg/ml) or TNFα (10 ng/ml) plus IFNγ (0.1 ng/ml) and reverse transcribed. The following protocol was used: after an initial denaturation at 94°C for 5 minutes, samples were subjected to temperatures of 94°C for 1 minute, 56°C for 45 seconds, and 72°C for 1 minute for β-actin, and an initial denaturation at 94°C for 1 minute, followed by temperatures of 60°C for 1 minute, and 72°C for 2 minutes for suppressor of cytokine signaling 1 [SOCS-1] and SOCS-3, and were then subjected to an extension step for 5 minutes at 72°C. PCR products were separated in 2% agarose gels and visualized with ethidium bromide. The forward and reverse primers used were as follows: for SOCS-1, 5′-CCC-TGG-TTG-TAG-CAG-CTT-3′ and 5′-CAA-CCC-CTG-GTT-TGT-GCA-A-3′; for SOCS-3, 5′-TAA-GTA-TTG-GCC-AGT-CAG-GCG-3′ and 5′-TTC-CAT-CGC-TAC-ATT-CCT-3′, and for β-actin, 5′-CCA-ACC-GCG-AGA-AGA-TGA-CC-3′ and 5′-GAT-CTT-CAT-GAG-GTA-GTC-AGT-3′.
Detection of phosphorylated STAT-1 (Tyr701).
FLS (2 × 104 cells) were seeded into 96-well plates, incubated overnight at 37°C, and then incubated in 200 μl of complete medium containing the various activators for 15 minutes, 30 minutes, 1 hour, and 16 hours. Phosphorylated STAT-1 was detected using a RayBio Cell-Based STAT-1 (Tyr701) ELISA kit according to the manufacturer's instructions.
Values are reported as the mean ± SEM. The significance of the results was examined by analysis of variance, with pairwise comparison by Scheffe's method, using StatView software (SAS Institute, Cary, NC). P values less than 0.05 were considered significant.
Effect of IFNγ and TNFα on BAFF release by activated FLS isolated from RA patients.
In this study, we first verified the capacity of IFNγ (0.01, 0.1, and 1 ng/ml) to stimulate BAFF synthesis by RA FLS. Real-time quantitative RT-PCR was performed with RNA isolated from control and activated FLS. BAFF transcripts did not show any obvious change in response to 0.01 ng/ml of IFNγ, but stimulation with IFNγ at concentrations of 0.1 and 1 ng/ml resulted in an increasing amount of BAFF transcripts, which were detectable within 24 hours and continued to increase for up to 72 hours (6-fold mean increase at 24 hours and 60-fold at 72 hours for 0.1 ng/ml of IFNγ) (Figure 1A).
To determine whether increased BAFF messenger RNA (mRNA) expression led to BAFF release from FLS, we used a sandwich ELISA to measure BAFF levels in cell culture supernatants. IFNγ (0.1 and 1 ng/ml) induced a strong release of BAFF (Figure 1B), reaching mean ± SEM values of 300 ± 9.5 pg/ml and 400 ± 11.2 pg/ml, respectively, at 72 hours. These values represented 6-fold and 8-fold increases as compared with controls.
We then tested the effect of TNFα on BAFF synthesis. FLS were stimulated with 10 ng/ml of TNFα and real-time quantitative RT-PCR and ELISA were performed. As shown in Figures 1A and B, and in contrast with previous results, BAFF mRNA was not induced by TNFα in activated FLS even after 72 hours of stimulation. These findings were confirmed by ELISA, which showed that FLS failed to release significant concentrations of BAFF in response to TNFα at each time point examined. This was not the result of a failure of cell activation, since FLS produced IL-6 after 24, 48, and 72 hours of activation with TNFα (Figure 1C).
The inflammatory environment of the RA synovial cavity is complex: most cytokines are present in the synovial cavity and can interact with each other. Since previous studies suggested that IFNγ and TNFα could act synergistically to stimulate the expression of BAFF by activated FLS, we next tested whether prior exposure of FLS to IFNγ might prime the FLS response to TNFα. As shown in Figure 2, pretreatment of cells with 0.01 ng/ml of IFNγ did not induce BAFF mRNA production (Figure 2A) and BAFF release (Figure 2B). Moreover, pretreatment with 0.01 ng/ml of IFNγ 24 hours before the addition of TNFα did not prime the FLS to mount an enhanced response to TNFα, even when longer activation times were used. Conditioned FLS that were stimulated with 0.1 ng/ml of IFNγ and then treated with TNFα (10 ng/ml) had increased levels of BAFF mRNA and secreted more BAFF. This effect was observed 24 hours, 48 hours, and 72 hours after activation with TNFα, which corresponds to 48 hours, 72 hours, and 96 hours of IFNγ activation. Taken together, these results demonstrate that IFNγ and TNFα act synergistically in this regard (P < 0.001).
Inhibition of IFNγ-induced BAFF release from RA FLS activated by TNFα.
We then examined whether a combination of TNFα (10 ng/ml) and IFNγ (0.01 and 0.1 ng/ml) could influence the levels of BAFF mRNA expression by FLS, since it is likely that FLS are exposed to both cytokines in the RA synovial cavity. No effect was observed with 0.01 ng/ml of IFNγ (Figure 3). Surprisingly, concomitant stimulation of FLS with IFNγ (0.1 ng/ml) and TNFα (10 ng/ml) resulted in a strong inhibition of BAFF mRNA expression (Figure 3A). We also tested BAFF release by activated FLS and found impaired BAFF secretion by FLS activated with IFNγ (0.1 ng/ml) (Figure 3B). This inhibition was particularly significant after 48 hours and 72 hours (P < 0.001). These results indicate that TNFα has an antagonistic effect on IFNγ-induced BAFF synthesis by FLS stimulated simultaneously with both cytokines.
Effect of LPS, BLP, and CpG on BAFF release by activated FLS isolated from RA patients.
We next determined whether LPS (1 μg/ml), BLP (1 μg/ml), and CpG A and CpG B (10 μg/ml) might also play a role in BAFF release by FLS. Quantitative real-time RT-PCR was performed under the same conditions as for TNFα and IFNγ activations. As shown in Figure 4A, there was no induction of BAFF mRNA expression in RA FLS in response to either LPS, BLP, CpG A, or CpG B after 24 and 72 hours. Similar results were obtained by ELISA (Figure 4B): none of the 4 bacterial stimuli were able to elicit BAFF release from activated FLS at 24, 48, and 72 hours, although LPS and BLP induced the release of IL-6 from these cells (Figure 4C).
We also investigated whether prior exposure of FLS to IFNγ could prime their response to the different PAMPs. FLS were stimulated with 0.01 and 0.1 ng/ml of IFNγ 24 hours before the addition of PAMPs. We observed that pretreatment of cells with 0.01 and 0.1 ng/ml of IFNγ did not have any effect on BAFF mRNA production or BAFF release in response to PAMPs at all time points examined (data not shown).
Inhibition of IFNγ-induced BAFF release from activated RA FLS by LPS, BLP, and CpG.
We studied the effect of concomitant stimulation with IFNγ (0.1 ng/ml) and either LPS, BLP, CpG A, or CpG B. As shown in Figure 5A, stimulation of FLS with IFNγ (0.1 ng/ml) and each of the different PAMPs strongly reduced the transcript levels of BAFF as compared with FLS activated with the same amount of IFNγ alone. The amount of BAFF secreted paralleled the level of transcription (Figure 5B). As seen previously, this effect was especially notable after 48 hours and 72 hours of stimulation. These results indicate that ligands of TLRs 2, 4, and 9 inhibit BAFF synthesis induced by IFNγ and then exert an antagonistic effect on this synthesis (P < 0.001).
Effect of protein I/II on BAFF synthesis and release by activated FLS isolated from RA patients.
BAFF mRNA and protein expression in RA FLS in response to protein I/II, a ligand of integrin α5β1, was then examined. Cells were activated with protein I/II for 24 and 72 hours, and PCR and ELISA were performed under the same conditions as for the LPS, BLP, and CpG analyses. As shown in Figures 5C and D, treatment with protein I/II, in contrast to the other PAMPs tested, induced BAFF mRNA expression and release from RA FLS. This release began at 24 hours and reached a mean ± SD of 125 ± 10 pg/ml at 72 hours (P < 0.001). Moreover, we showed that simultaneous treatment of FLS with protein I/II and IFNγ (0.1 ng/ml), did not inhibit BAFF mRNA production and BAFF release in response to IFNγ at any of the time points studied (Figures 5C and D).
Effect of TLR ligands or TNFα on IFNγ-induced STAT-1 phosphorylation.
Next, we analyzed the mechanisms responsible for the negative regulatory effects of stimulation with TLR and TNF receptors on IFNγ-mediated signaling. Tyrosine phosphorylation of STAT-1 is necessary for dimerization and subsequent transcriptional activation by IFNγ. We therefore evaluated the tyrosine phosphorylation status of STAT-1 after simultaneous activation of either TLR-2, TLR-4, and TLR-9 ligands plus IFNγ or TNFα plus IFNγ. STAT-1 phosphorylation was already detectable at 15 minutes after IFNγ stimulation (Figure 6A, right). STAT-1 phosphorylation remained elevated at 16 hours and declined after 24 hours. Costimulation with either LPS, BLP, or CpG plus IFNγ resulted in a strong inhibition of STAT-1 phosphorylation on Tyr701, which began after 1 hour and remained elevated at 16 hours after stimulation (Figure 6A). Protein I/II had no effect on STAT-1 phosphorylation induced by IFNγ (data not shown). In contrast, STAT-1 phosphorylation was first up-regulated at 1 hour after costimulation with TNFα (data not shown) and then was reduced at 16 hours (Figure 6A).
We speculated that SOCS-1 and SOCS-3 could be responsible for the negative regulation of STAT-1 phosphorylation. We studied SOCS-1 and SOCS-3 mRNA expression in RA FLS stimulated for 1 hour and 16 hours with either LPS or TNFα plus IFNγ. Costimulation with LPS strongly enhanced the expression of mRNA for SOCS-1 after 16 hours and of mRNA for SOCS-3 after 1 hour and 16 hours, as compared with IFNγ stimulation alone (Figure 6B). TNFα enhanced the expression of mRNA for SOCS-1 and SOCS-3 at 16 hours only (Figure 6C).
The role of B cells in autoimmunity has undergone a renaissance after the demonstration of the efficacy of B cell depletion in RA (25). Since BAFF plays a pivotal role in B cell activation in autoimmune diseases and might be secreted by resident cells of target organs and given the current knowledge on the role of innate immunity in autoimmune diseases, we investigated the role of inflammatory cytokines and ligands of innate immune receptors in BAFF secretion by fibroblast-like synoviocytes.
We examined whether activation of TLR-2, TLR-4, and TLR-9, which recognize various components of gram-positive and gram-negative bacteria, can induce BAFF synthesis. Our results clearly showed that activation with LPS, BLP, CpG A, and CpG B failed to induce BAFF transcription and release, even after extended stimulation times. Moreover, the fact that there was no BAFF expression after 72 hours completely rules out an indirect effect of these PAMPs on the synthesis of molecules that could act by an autocrine effect.
PAMPs also interact with PRRs that do not belong to the TLR family. FLS express various integrins, and in this study, we also explored the ultimate implication of α5β1 integrin, a well-known receptor of several bacterial PAMPs, in BAFF synthesis (26–28). Protein I/II, an adhesin derived from viridans streptococci, is an established α5β1 integrin ligand that induces the release of classic proinflammatory cytokines, such as IL-6 and IL-8, by FLS (18). An important result in this study was the finding that protein I/II induced BAFF transcription and secretion by activated FLS. This result suggests that in our model, a bacterial component can stimulate BAFF synthesis through the integrin signaling pathway, whereas interactions between TLRs and LPS, BLP, and CpG do not participate in BAFF synthesis.
Previous studies showed that IL-6 and IL-8 release by protein I/II–activated FLS requires crosstalk between myeloid differentiation factor 88 and focal adhesion kinase, which contributes to the TLR and integrin signaling pathways, respectively (19). It is, however, conceivable that protein I/II signaling involved in BAFF synthesis might use distinct downstream signaling molecules belonging to the α5β1 integrin pathway. To our knowledge, this is the first demonstration of BAFF induction by an integrin. Recently, BAFF-induced stimulation of NF-κB was shown to induce the up-regulation of integrin and promote the sequestration of BAFF-mediated autoreactive B cells in the splenic marginal zone (29). A similar sequestration of B cells in the rheumatoid synovium might be related to the integrin-mediated BAFF secretion and reciprocal BAFF-mediated integrin up-regulation (positive regulatory loop). This crossregulation of BAFF and integrins might also contribute to the increased proliferation of FLS in RA, which might also be related to the autocrine effect of BAFF on FLS. Indeed, FLS not only secrete BAFF, but they also have BAFF receptors (Nagatani K, et al: unpublished observations).
IFNγ plays a central role in BAFF synthesis. This cytokine is produced in RA by CD4+ T cells or by subsets of CD8+,CD40LT cells or CD4+ T cells that express killer cell immunoglobulin-like receptor 2DS2 and natural killer cell receptor 2D. Immunohistochemical studies have shown the reinforced expression of STAT proteins in rheumatoid synovial tissues, which suggests the importance of the IFNγ/JAK/STAT pathway (30, 31). We therefore assessed the role of IFNγ in BAFF secretion by FLS. We first confirmed the findings of recent studies by Ohata et al (11), who showed that in FLS, IFNγ caused BAFF synthesis and release and that this effect was particularly significant between 48 hours and 72 hours.
Under most conditions, the production of TNFα correlates with the production of IFNγ. Thus, we also investigated the role of TNFα in BAFF secretion. TNFα did not stimulate either BAFF mRNA synthesis or BAFF secretion in FLS in our experiments. This finding is not consistent with those reported by Ohata (11), who showed that TNFα induced a significant increase in BAFF mRNA and protein in rheumatoid FLS. There is some controversy in the literature concerning the role of TNFα in mediating BAFF synthesis by cells of mesenchymal origin. For example, Ittah et al (32) demonstrated that BAFF is not expressed by TNFα-stimulated salivary gland epithelial cells from patients with primary Sjögren's syndrome. In another study, BAFF expression and production by astrocytes were not up-regulated by TNFα (10). The fact that in our model, TNFα is unlikely to function as an inducer of BAFF synthesis by FLS, could not be due to impaired activity of TNFα insofar as FLS release high concentrations of IL-6 in response to TNFα. The reason for this discrepancy is therefore presently unknown.
Since TNFα, IFNγ, and bacterial components are probably present subsequently or concomitantly in the synovial cavity at specific stages of RA, we investigated whether prior or concomitant exposure to IFNγ, TNFα, or PAMPs could act in synergy in the induction of BAFF by activated FLS, as has been reported for IFNγ and TNFα. Ohata et al (11) showed that priming of FLS with low or high concentrations of IFNγ enhanced the sensitivity of the TNF receptors on FLS, allowing TNFα to induce a greater BAFF synthesis. We also found that pretreatment of cells with IFNγ favored the effect of TNFα, since we observed an increase in BAFF synthesis after stimulation with TNFα. We also demonstrated that prior exposure to IFNγ did not definitely influence the response of FLS to LPS, BLP, and CpG, insofar as no increase in BAFF was noted as compared with stimulation with IFNγ alone.
A striking observation was that concomitant stimulation with TNFα and IFNγ belatedly (48 hours and 72 hours) inhibited IFNγ-induced BAFF synthesis. The ability of TNFα to exert a negative regulation was previously reported by Huong et al (33), who studied IFNα and IFNβ signaling, and recent results reported by Zhakarova and Ziegler (34) demonstrated that TNFα exerts a negative regulatory loop on IL-12 and IL-23 production by macrophages and dendritic cells during the later stage of inflammation.
Interestingly, we also demonstrated that stimulation of TLRs with PAMPs inhibited IFNγ-induced BAFF synthesis. Similar results were obtained with exposure to LPS, BLP, or CpG, which exert a negative regulation of BAFF synthesis. This effect is specific to the TLR and TNF pathways, since protein I/II, which stimulates the integrin signaling pathway, did not inhibit BAFF synthesis induced by IFNγ.
We showed in the present study that ligands of TLR-2, TLR-4, and TLR-9 inhibit IFNγ-induced STAT-1 tyrosine phosphorylation, which was only observed 1 hour after costimulation. Similar results were obtained with TNFα, but the inhibitory effect was observed later. This could explain the weak increase in BAFF synthesis detected after 24 hours of costimulation with TNFα and IFNγ. Protein I/II failed to modify STAT-1 phosphorylation. This is consistent with the observation that costimulation with protein I/II had no effect on IFNγ-induced BAFF synthesis. It also supports a role of STAT-1 dephosphorylation in this inhibitory mechanism. Recently, Dalpke et al (35) demonstrated that costimulation with IFNγ with either LPS, CpG, or lymphotoxin A led to the inhibition of STAT-1 phosphorylation and that the inhibition was dependent on the induction of SOCS-1 and SOCS-3. SOCS proteins are classic feedback inhibitors of the JAK/STAT pathway. Both SOCS-1 and SOCS-3 can inhibit JAK phosphorylation of STAT-1. Whereas SOCS-1 functions by binding directly to JAK proteins, SOCS-3 binds to phosphorylated tyrosine on the receptor. SOCS-1 and SOCS-3 are also induced by JAK/STAT-independent stimuli, such as LPS, CpG, TNFα, and insulin (36, 37). Indeed, we demonstrated that costimulation with either LPS or TNFα enhanced SOCS-1 and SOCS-3 mRNA expression. Thus, we believe that the inhibitory effects of LPS and TNFα on BAFF secretion are related to STAT-1 dephosphorylation by SOCS-1 and SOCS-3.
The crosstalk between TLR and IFNγ signaling pathways, however, might be dependent on the cell type. Recent data reported by Zhao et al (38) showed that TLR-2, TLR-3, and TLR-4 ligands and IFNγ synergistically activate macrophages, resulting in maximal production of proinflammatory cytokines.
IL-6, which is produced after stimulation of FLS with LPS, BLP, and TNFα, might also participate indirectly in the inhibition of IFNγ signaling, since the inhibitory effect is detected after 48 hours and 72 hours (39). The inhibitory effect of CpG on BAFF synthesis is not dependent on IL-6, since CpG failed to induce IL-6 release by activated FLS, as has also been observed in leukocytes (40), whereas they can directly induce the expression of SOCS-1 and SOCS-3 (41).
Overall, our findings showed that BAFF secretion by resident cells of target organs of autoimmunity is tightly regulated by a complex network involving innate immunity and cytokines, with positive and negative controls that are dependent on the receptors and pathways triggered. An understanding of these complex interactions is important with regard to therapeutic applications.
Dr. Wachsmann had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Alsaleh, Messer, Semaan, Gottenberg, Sibilia, Wachsmann.
Acquisition of data. Alsaleh, Messer, Semaan, Sibilia, Wachsmann.
Analysis and interpretation of data. Alsaleh, Messer, Semaan, Gottenberg, Sibilia, Wachsmann.
Manuscript preparation. Alsaleh, Boulanger, Gottenberg, Sibilia, Wachsmann.
Statistical analysis. Alsaleh.
- 23NF-κB and the MAP kinases/AP-1 pathways are both involved in interleukin-6 and interleukin-8 expression in fibroblast-like synoviocytes stimulated by protein I/II, a modulin from oral streptococci. Cell Microbiol 2001; 3: 703–12., , , , , .