B cell-activating factor
primary Sjögren's syndrome
salivary gland epithelial cells
systemic lupus erythematosus
B cell activating factor (BAFF) plays a key role in promoting B lymphocyte activation. We investigated whether danger signals induce BAFF secretion by cultured salivary gland epithelial cells (SGEC), which are the target of primary Sjögren's syndrome, a prototypic systemic autoimmune disease. SGEC cultures were established from minor salivary glands obtained from ten patients with pSS or sicca symptoms. BAFF mRNA and protein were measured after stimulation of the different Toll-like receptors (TLR) by agonists or viruses. The expression of TLR2, –3, and –7 was detected in SGEC. Poly (I:C) (a synthetic TLR3 agonist) and reovirus-1 (a dsRNA virus) induced high expression of BAFF mRNA (multiplied by a factor of 246 ± 39 (SEM) and 347 ± 66, respectively) and of BAFF protein secretion (58.49 ± 4.34 pg/mL and 69.73 ± 5.67). Inhibition of both the endosomal (by chloroquine) and IFN (by anti-IFNAR antibody) pathways partly inhibited BAFF expression. Treatment with both dsRNA virus and poly (I:C) induced high levels of BAFF mRNA and protein expression by SGEC, through pathways dependent on and independent of TLR and dependent on and independent of IFN. BAFF induction by target organs of autoimmune diseases after viral infection may be a link between innate immunity and autoimmunity.
Increased expression of B cell activating factor (BAFF), a cytokine also called B lymphocyte stimulator (BLyS) 1, may explain pathogenic B cell activation in several systemic autoimmune diseases and lymphomas. BAFF plays a crucial role in B cell maturation, plasma cell survival, antibody response promotion and immunoglobulin-class switch recombination 2. Its involvement in the pathogenesis of autoimmune diseases is demonstrated by BAFF transgenic mice, which develop autoimmune diseases mimicking systemic lupus erythematosus (SLE) and primary Sjögren's syndrome (pSS) and have a rate of B cell lymphoma twice that of control mice 3. In humans, patients with SLE and pSS have elevated serum levels of BAFF 4, 5, and some studies report correlations between serum levels of BAFF and of autoantibodies 5.
SLE and pSS are prototypic systemic autoimmune diseases; the latter is characterized by xerostomia and keratoconjunctivitis sicca due to lymphocytic infiltration of salivary and lachrymal glands, with local secretion of autoantibodies 6, 7. Systemic complications involving joints, skin, lungs, kidneys and the nervous system occur in one third of patients. The 20-fold increased risk of non-Hodgkin B cell lymphoma 8 in patients with pSS is one illustration of the relation between autoimmunity and B cell lymphoma. pSS shares numerous pathogenic features with SLE, including an elevated serum level of BAFF and, of special interest, an interferon (IFN) signature 9, 10. Moreover, plasmacytoid dendritic cells (pDC), which may be called professional IFN-α-secreting cells, are present in target organs of the autoimmune process, i.e., the skin in SLE and salivary glands in pSS 9, 11.
Toll-like receptor (TLR) agonists are the principal exogenous stimuli that drive type-I IFN secretion. TLR recognize pathogen-associated molecular patterns (PAMP) including PGN (TLR2 ligand), viral double-stranded DNA (dsDNA) containing unmethylated CpG motifs (TLR9 ligand), viral dsRNA (TLR3 ligand), and viral single-stranded RNA (ssRNA; TLR7 and –8 ligands) 12. TLR-induced signaling spreads to several adaptors and downstream effectors and usually culminates in activation of NF-κB, mitogen-activated protein kinase (MAPK) pathways, and IFN-responsive factors (IRF), as well as production of inflammatory and immune cytokines 12. Under physiological conditions, BAFF is secreted by myeloid cells, especially monocytes, macrophages, and polynuclear cells 13. In patients with autoimmune diseases, elevated BAFF levels are found in the target organs. In pSS, for example, they have been shown in the salivary glands and specifically in the salivary epithelial cells, as demonstrated by immunohistochemistry and long-term culture of salivary gland epithelial cells (SGEC) 9, 14, 15. This SGEC secretion of BAFF increases substantially after stimulation with type-I and type-II IFN 15.
An association with viral infection has long been suspected for pSS as for many other autoimmune diseases 16. In this study, we assessed the effect of viral infection and of chemical TLR agonists on BAFF secretion by SEGC and demonstrate for the first time that dsRNA viral infection and TLR3 agonists directly induce these cells to express and secrete high levels of BAFF. We also investigated the mechanisms of BAFF induction and conclude that numerous pathways are involved: both dependent on or independent of TLR and dependent on or independent of IFN.
Results and discussion
TLR expression by ductal SGEC
Cells obtained from minor salivary gland biopsies were cultured for 4 weeks. The finding that they were 100% positive for cytokeratin (CK) 7 and CK19 and negative for CK20 confirmed that they were salivary ductal epithelial cells (data not shown). Given that TLR2 recognizes both bacterial peptidoglycan and vaccinia virus 17, that TLR3 and TLR7 are PAMP receptors recognizing viral RNA, and that TLR9 recognizes viral DNA, we sought to determine whether SGEC express these receptors. RT-PCR showed that the cultured SGEC expressed TLR2, TLR3, and TLR7 but not TLR9 (Fig. 1A).
Other studies assessing TLR in salivary glands 18 or SGEC 19 report expression of TLR2, –3, and –4, with no difference for TLR3 between pSS patients and controls. No study, however, has evaluated the presence of TLR7 and –9 in SGEC. The target cells of autoimmune diseases may play an active role, as synoviocytes do in rheumatoid arthritis 20 and astrocytes in multiple sclerosis 21. The presence of TLR3 and –7 in SGEC is thus of major interest because it may enable us to identify potential stimuli that target them.
Effect of TLR ligands on BAFF expression in ductal SGEC
To assess whether TLR ligands induce BAFF expression, SGEC were treated with PGN (for TLR2 ligand), poly (I:C) (TLR3), R837 (TLR7) and CpG (TLR9) for 24 or 48 h. Significant expression of BAFF mRNA and protein was induced by treatment with poly (I:C) but not PGN, R837 or CpG (Fig. 1B and C). Next we investigated the effect of reovirus-1, a dsRNA virus, and HSV1, a dsDNA virus, on SGEC expression and secretion of BAFF. Both viruses easily infected SGEC. Reovirus-1 induced BAFF mRNA expression and protein secretion, whereas HSV1 induced BAFF mRNA expression but not protein secretion (Fig. 1B and C).
DNA viruses may target cells without any TLR9, by stimulating TLR2, –3, –7 or –8, either during replication or via the virus envelope or core 17. BAFF mRNA expression and protein secretion after treatment with poly (I:C) and reovirus-1 prompted us to study the mechanisms of BAFF induction downstream, after TLR3 stimulation.
Effect of TLR3 inhibition on BAFF expression in ductal SGEC
Our first hypothesis was that poly (I:C) and reovirus-1 induced expression of BAFF mRNA and protein by stimulating TLR3 and synthesizing type-I IFN. TLR3 is expressed in the endosomal compartment and needs an acid environment for its maturation. To test whether BAFF induction by poly (I:C) or reovirus-1 depended on TLR3, we pretreated SGEC from ten patients (each sample processed in duplicate) with chloroquine used to inhibit endosomal acidification. Chloroquine significantly reduced the BAFF expression induced by poly (I:C) and reovirus-1, both mRNA (p=0.007 and p=0.005, respectively) (Fig. 2A) and protein secretion (p=0.02 and p=0.005) (Fig. 2B). However, this inhibition was only partial (ranging from 43% to 54% for mRNA expression and from 24% to 30% for protein secretion), and was greater with reovirus-1 than with poly (I:C). These results suggest that BAFF induction does not depend exclusively on TLR3.
The failure of chloroquine to inhibit BAFF expression completely was not due to its ineffectiveness in inhibiting acidification. Indeed, chloroquine very efficiently inhibited the induction of SEGC secretion of three inflammatory cytokines (IL-6 and IL-8, which depend essentially on the NF-κB pathway, and RANTES, mostly regulated by IRF3). IL-6, IL-8, and RANTES were significantly up-regulated after reovirus-1 infection, and chloroquine reduced their levels back to baseline (Fig. 3).
Effect of inhibiting type-I IFN on BAFF expression in ductal SGEC
As expected, reovirus-1 infection and poly (I:C) stimulation both induced IFN-β expression in SGEC (Fig. 2C). To test whether BAFF induction by poly (I:C) or reovirus-1 treatments depended on the presence of type-I IFN, we treated SGEC with an anti-IFN-α receptor 1 (IFNAR1) antibody before poly (I:C) stimulation or reovirus-1 infection. The efficacy of this antibody was confirmed by the substantial reduction in SGEC secretion of BAFF after type-I IFN stimulation (Fig. 2D). Anti-IFNAR1 pretreatment reduced both BAFF mRNA expression and protein secretion after either poly (I:C) stimulation (p=0.07 and p=0.01, respectively) or reovirus-1 infection (p=0.005 and p=0.009) (Fig. 2A and B). Again, however, this inhibition was only partial and was greater with reovirus-1 (47% for mRNA expression and 25% for protein secretion) than poly (I:C) (27% for mRNA expression and 12% for protein secretion). These results suggest that BAFF induction does not depend exclusively on type-I IFN.
Hypothesis about pathways involved in epithelial cell secretion of BAFF after innate immunity stimulation
Because chloroquine and anti-IFNAR1 antibody only partially inhibited BAFF induction after reovirus-1 infection, we used them together in one experiment. The combination of the two agents did not increase the level of inhibition (data not shown). This suggests that different pathways are probably involved. These alternative pathways could involve retinoic acid-inducible gene I (RIG-I) and melanoma differentiation-associated gene 5 (MDA5), two intracellular RNA sensors mediating TLR-independent induction of type-I IFN or NF-κB by actively binding to RNA viruses and poly (I:C) 22.
New insight into relations between viral infection and autoimmunity
Our results reinforce the hypothesis that viral infection may be one of the first events of pSS pathogenesis, leading to BAFF expression by epithelial cells, which in turn induces activation of B cells and secretion of antibodies in genetically predisposed subjects. An association with viral infection has long been suspected for pSS, as for many other autoimmune diseases 16. The frequency of viral carriage in the mouth has led to extensive investigation of this hypothesis for pSS. However, to date, there is no consensus about the increased frequency of any particular virus in this disease. Our data suggest that various viral infections may lead to the same consequence: an increase in BAFF levels and thus the activation of autoimmune response in genetically susceptible subjects.
Patients with pSS, those with sicca symptoms and some signs of autoimmunity but no definite diagnosis of pSS, and controls with sicca symptoms and no sign of autoimmunity showed no difference in SGEC expression and secretion of BAFF levels after viral infection (data not shown). However, because of the complexity of the experiment, we could test only ten patients. Moreover, SGEC were infected after 4 weeks of in vitro culture, thereby avoiding the potential effect of the in vivo infiltrate and cytokine environment. Thus, SGEC expression of BAFF after viral infection is probably common in patients and healthy subjects. Nevertheless, the possibility of that BAFF expression is higher in patients with pSS requires further evaluation.
Interestingly, the possibility of BAFF expression and secretion by epithelial cells other than salivary cells after stimulation by type-I IFN and synthetic dsRNA has recently been demonstrated in a bronchial epithelial cell line and in human airway epithelial cells 23. Likewise, commercially available oral epithelial cells are reported to express BAFF after poly (I:C) stimulation, and this expression leads to class switching of tonsil B cells 24. These authors also show that other epithelium, such as epidermal and intestinal epithelium, expresses BAFF after poly (I:C) stimulation.
In conclusion, this study demonstrates for the first time that human SGEC secrete BAFF after viral infection or synthetic RNA stimulation and do so through pathways dependent on and independent of TLR and dependent on and independent of IFN. It suggests that the increase of BAFF in the target organs of autoimmune diseases may be the consequence of innate immune stimulation of resident cells that results in early activation of B lymphocytes.
These results also open up physiopathological perspectives for studying other organ-specific autoimmune diseases. Resident cells of target organs, such as synovial cells in rheumatoid arthritis 20 or astrocytes in multiple sclerosis 21, secrete BAFF after IFN stimulation. In these diseases, the direct activation of resident cells by danger signals may be a very early stage of pathogenesis. Lastly, this study provides an additional rationale for BAFF therapeutic blockade in autoimmune diseases.
Materials and methods
Ten consecutive patients referred for sicca symptoms underwent biopsies of the minor salivary gland. Three patients had pSS according to the criteria of the European-American Consensus Group (EACG) 25; four had features of autoimmune response (antinuclear antibodies or lymphocytic infiltrates in salivary glands), but disease did not meet either the EACG criteria or criteria for other autoimmune diseases; and three had only subjective symptoms of dryness without any autoimmune abnormalities. None of the patients had corticosteroid or immunosuppressive treatments. The study received approval from the local ethics committee, and informed consent was obtained from all study subjects.
Anti-IFNAR1 and virus production
The 64G12 mAb, a mouse IgG1 that inhibits both binding and biological activity of all the human type-I IFN tested, was prepared by immunizing mice with a recombinant protein corresponding to the extracellular domain of the IFNAR1 chain, as described previously 26.
Stocks of the human strain reovirus-1 (Lang) and herpes simplex virus-1 (HSV1), were prepared from supernatants of infected Vero cells cultured in RPMI 1640–2% FBS at a multiplicity of infection of 0.1. At 72 h post-infection, cells were frozen and thawed once.
SGEC cultures and treatment
Biopsies of minor salivary glands provided SGEC for primary cultures, as described previously 27. After 4 weeks of culture, cells at 70–80% confluence were dissociated with 0.125% trypsin-EDTA and transferred to 12-well plates coated with collagen type-I, at 380 × 103 cells per well. The next day, cells were treated with peptidoglycan (PGN) (2.5–5 µg/mL; for TLR2), poly (I:C) (30 µg/mL; for TLR3), R837 (1–75 µg/mL; for TLR7) (Sigma, Saint Quentin Fallavier, France) and CpG 2084 (5 µg/mL; for TLR9) (Operon Biotechnologies GmbH, Cologne, Germany). Cells were also infected with reovirus-1, 4 × 103 PFU/mL, or HSV1, 5 × 103 PFU/mL. To prevent endosomal acidification or type-I IFN pathway induction, respectively, chloroquine (10 µM for 2 h at 37°C) or anti-IFNAR1 (50 µg/mL for 30 min at 37°C) was used before in vitro treatment. All experiments were performed in duplicate.
RNA extraction and expression analysis
Total RNA was isolated from SEGC with use of the RNeasy Mini kit (Qiagen, Courtaboeuf, France) and digested by DNase I (Qiagen), according to the manufacturer's instructions. cDNA synthesis involved use of Enhanced Avian HS RT-PCR (Sigma-Aldrich, Saint Quentin Fallavier, France). PCR amplification of TLR2, –3, –7 and 9 was done as previously described 28.
Real-time quantitative RT-PCR
BAFF, IFN-β and β-actin cDNA levels were determined with Light Cycler-based kinetic quantitative RT-PCR (Roche Diagnostics), as previously described 15. Amplification primers for IFN-β were as follows: 5′-TGCTCTCCTGTTGTGCTTCTCC-3′ and 5′-CATCTCATAGATGGTCAATGCGG-3′. Primers for BAFF were specific to full-length BAFF, excluding any amplification of ΔBAFF.
Detection and quantification of cytokine secretion
The levels of BAFF, IL-6, IL-8 and RANTES in supernatants of cultures of unstimulated or stimulated SGEC were determined with ELISA kits from R&D Systems (Lille, France).
Results are shown as mean ± SEM. The Wilcoxon signed-rank test was used to compare two related samples or for repeated measurements of a single sample. Statistical comparisons were performed with StatView, version 5.0 (SAS institute Inc). A p value <0.05 was considered significant.
We thank Joel Plumas who provided cDNA of pDC for positive controls of TLR7 and –9 and for quantitative PCR of IFN-β. Funding for the study came from the Société Française de Rhumatologie, Réseau de Recherche clinique INSERM and Agence Nationale de la Recherche.
The authors declare no financial or commercial conflicts of interest.