Local administration of glucocorticoids decrease synovial citrullination in rheumatoid arthritis

Introduction: Protein citrullination is present in the rheumatoid synovium, presumably contributing to the perpetuation of chronic inflammation, in the presence of specific autoimmunity. As a result, the present study examines the possibility that effective anti rheumatic treatment will decrease the level of synovial citrullination. Methods: Synovial biopsies were obtained from 11 rheumatoid arthritis (RA) patients before and after 8 weeks of treatment with 20 mg methotrexate weekly, 15 RA patients before and two weeks after an intra-articular glucocorticoid injection and eight healthy volunteers. Synovial inflammation was assessed by double-blind semi quantitative analysis of lining thickness, cell infiltration and vascularity using a 4 points scale. Expression of citrullinated proteins (CP) with the monoclonal antibody F95 and peptidylarginine deiminase (PAD) 2 and 4 was assessed immunohistochemically by double blind semi-quantitative analysis. In vitro synovial fluid (SF), peripheral blood (PB) mononuclear cells (MC) and synovial explants obtained from RA patients were incubated with dexamethasone and analyzed by immunhistochemistry for expression of CP as well as PAD2 and PAD4 enzymes. Results: Presence of synovial CP was almost exclusive in RA compared to healthy synovium and correlated with the degree of local inflammation. Treatment with glucocorticoids but not methotrexate alters expression of synovial CP and PAD enzymes, in parallel with a decrease of synovial inflammation. Ex vivo and in vitro studies suggest also a direct effect of glucocorticoids on citrullination as demonstrated by the decrease in the level of citrullination and PAD expression following incubation of SFMC and synovial explants with dexamethasone. Conclusion: Synovial citrullination and PAD expression is dependent on local inflammation and targeted by glucocorticoids. Introduction Rheumatoid arthritis (RA) is a chronic inflammatory disease characterized by the presence of highly specific anti-citrullinated protein antibodies (ACPA) [1]. These antibodies recognize several different proteins that are citrullinated. Citrullination is the conversion of peptidylarginine to peptidylcitrulline through a calcium dependent process catalyzed by the peptidylarginine deiminase (PAD) enzymes. Five PAD isotypes have been described in humans (PAD1, PAD2, PAD3, PAD4 and PAD6), which are expressed in a variety of tissues, but only PAD2 and PAD4 have been found to be expressed in inflamed synovial tissue of RA and other inflammatory arthritides [2]. Despite the high specificity of ACPA in RA in comparison to other arthritides and other inflammatory diseases [3], the presence of CP is not restricted to RA synovial tissue [4] [5], but rather associated with inflammation in general [6]. Synovial citrullination appears therefore not to be essential for the pre disease phase of induction of specific anti citrulline immunity. On the other hand protein citrullination enhances the HLA binding capacity of synovial derived proteins and promotes NF-kappaB and TNF production in the presence ACPA [7]. This suggests that local synovial citrullination might be essential in a later phase of the disease process, contributing to occurrence and perpetuation of chronic synovitis in the presence of specific anti citrulline antibodies. It is therefore conceivable that downregulation of synovial citrullination by any means will contribute to resolution of local inflammation. We hypothesize that effective anti rheumatic treatment with either anti inflammatory, intra-articular glucocorticoids (GC) or a disease modifying anti rheumatic drug such as methotrexate (MTX) will decrease synovial citrullination in vivo. As a result, the present study aims to investigate any direct effect of these drugs on protein citrullination. Materials and methods Patients Twenty-six patients meeting the 1987 American College of Rheumatology criteria for RA [8] were recruited for this study. In a first group, 11 patients (6 women and 5 men, median age 56 years, range 33-78 years) with new diagnosed RA (symptom duration less than 1 year) previously DMARD naïve were started on MTX 10 mg weekly and reached a stable dose of 20 mg after 2 weeks increasing the dose with 5 mg every week. Synovial biopsy samples were obtained by arthroscopy from all patients before and after a median of eight weeks of treatment. Clinical evaluation of the therapeutic response according to EULAR response criteria was performed on a median of 3 months after methotrexate initiation. In a second group, 15 RA patients (11 women and 4 men, median age 63 years, range 34-83 years) with active knee arthritis independent of disease duration received an intra-articular injection of 40 mg of triamcinolone hexacetonide and synovial biopsy samples were obtained by arthroscopy before and after two weeks after intra-articular treatment. In this second group, associated DMARD treatments were stable for at least 2 weeks before initiation of treatment and throughout the entire study period. Clinical evaluation of the therapeutic response was performed, by macroscopic scoring of the level of inflammation during arthroscopy. Non steroidal anti inflammatory drugs and per oral prednisolone to a maximum dose of 10 mg daily were permitted in both groups. Additionally, synovial biopsy samples were obtained by arthroscopy in 8 healthy volunteers. All procedures where approved by the Northern Stockholm Ethical Review Board and informed consent was obtained from all the participants in the study. Synovial biopsies handling Synovial biopsy samples were snap-frozen during arthroscopy in dry-ice cooled isopentane. Serial cryostat sections (7 μm) were fixed for 20 min with 2% (v/v) formaldehyde and stored at -70oC. Cell culturing Paired peripheral blood mononuclear cells (PBMC) and synovial fluid mononuclear cells (SFMC) from RA patients (n=6) were cultured as duplicates in RPMI 1640 supplemented with 20% heat inactivated fetal bovine serum (FBS), 2mM L-glutamine, 50 IU/ml penicillin and 50μg/ml streptomycin (Gibco, CA, USA) and incubated at 37 o C with 5% CO2 in a humidified atmosphere. Cells were grown at a density of 1x10 6 per ml in polypropylene tubes and incubated with dexamethasone for 24 hours at a final concentration of 1 and 100 μM DXM. After treatment cells were washed, resuspended in PBS and seeded on glass slides (Thermo Scientific diagnostic microscopic glass slides, Braunschweig, Germany). Adherent cells were fixed for 20 min at 4 o C with 2% (volume/volume) formaldehyde (Merck, Darmstadt, Germany). Synovial explants Synovial tissue pieces obtained from an orthopedic hip replacement surgery in an RA patient were selected based on maximal macroscopic score of inflammation, dissected and seeded on 24 wells plates in RPMI supplemented with 10% heat inactivated fetal calf serum (FCS), 2mM L-glutamine, 50 IU/ml penicillin and 50μg/ml streptomycin (Gibco, CA, USA) and incubated at 37 o C with 5% CO2 in a humidified atmosphere for 24 hours with 100μM of DXM. Tissue pieces were then transferred to 6 well plates and cultured for additional 5 days with fresh prepared serum supplemented with the same concentration of the drugs. At the end of the incubation medium was removed and tissue pieces were cryosectioned, fixed for 20 minutes at 4 o C with 2% (volume/volume) formaldehyde (Merck, Darmstadt, Germany) and stored at -80C. Immunohistochemistry and histological evaluation Presence of citrullinated proteins (CP) was detected using a mouse IgM monoclonal antibody (F95) which was raised against a decacitrullinated peptide linked to the carrier protein, keyhole limpet hemocyanin [9], [10], [11]. For detection of PAD enzyme expression we used two PAD4 (SN823 and SG1467) [12], [13] and one PAD2 (SN665) [12] rabbit polyclonal antibodies and one additional rabbit polyclonal antibody (ROI002) (CosmoBio, Tokyo, Japan) to identify PAD2 expression [14]. Appropriate negative controls were used for each antibody. Two independent observers (DM and AIC) who were unaware of the sample’s identity scored the presence of CP and PAD enzymes in synovial tissue using a four-point scale (0 = no staining, 1 = low amounts of staining, 2 = moderate amounts of staining, 3 = high amounts of staining). In parallel, histologic scoring of the degree of inflammation was performed, using a four-point scale, by double blind semi-quantitative analysis (DM and SR) of the lining thickness, infiltration level and vascularity in serial hematoxylin eosin staining. In every instance, the final scores represent the mean of the 2 observations. The presence of CP and PAD enzymes in SFMC and PBMC was assessed by manual counting of positive cells and expressed as percentage of positive cells out of total number of the cells. Statistical analysis Statistical analysis was performed using the Wilcoxon test for comparison of paired samples, the Mann-Whitney test for comparison of independent samples and Spearmen rank correlation test. All comparisons were planned and therefore no Bonferroni correction was applied. Differences between proportions were analyzed by Fisher’s exact test. In vitro experiments results were analyzed using one-way Anova analysis of variance as appropriate. P values less than 0.05 were considered as statistically significant. Results Intracellular but not extracellular presence of citrullinated peptides is RA specific. It has been previously suggested that the intracellular pattern of F95 staining is a specific trait of RA as compared to healthy synovium [11]. To further investigate this we analyzed the immunohistochemical expression of citrullinated peptides (as detected by F95 antibody) and the PAD2 and PAD4 enzymes in baseline RA samples obtained before initiation of treatment, as compared to healthy synovial biopsies. Intracellular F95 staining was readily detected in RA samples (24/28, 85.7%) and absent in healthy synovium (0/8, p<0.05). Extracellular F95 staining was present in RA (24/28, 85.7%) and to a lesser extent in healthy synovial tissue (1/8, 12.5%, p<0.05). PAD2 and PAD4 expression was detected in all baseline RA samples (100%) and a vast majority of the healthy synovial biopsies (7/8, 87.5% positive for ROI002 antibody, 6/8, 75% positive for SN665 antibody, 8/8, 100% positive for SG1467 antibody and 7/8, 87.5% positive for SN8233 antibody) and with a general higher expression of all investigated molecules in RA as compared to healthy synovium (figure 1, p<0.05 for all investigated antibodies). Synovial citrullination and PAD expression correlate with the degree of inflammation In order to further analyze the relationship between inflammation and citrullination we analyzed the correlation between presence of total and intracellular CP, as well as PAD enzymes and the degree of local inflammation in baseline samples obtained from all RA patients and healthy individuals (n=34). As expected most of the investigated markers correlated with the degree of lining thickness, cell infiltration and vascularity (table 1). Intra-articular GC, but not MTX decreases synovial citrullination and PAD4 expression. The observed correlation between inflammation and citrullination prompted us to investigate the effect of anti rheumatic treatment on synovial citrullination and PAD expression. Local administration of GC significantly decreased both total and intra cellular expression of CP from a median score of 2 (range 0-3) to a median score of 1 (range 0-2), p<0.05. PAD4 expression also decreased after treatment from a median score of 2 (range 1-3) to a median score of 1 (range 0-2.5), p<0.05, when evaluated by the SG1467 antibody and to a lesser extent when evaluating by the SN823 antibody (p=0.06), while no changes were observed for PAD2 expression following intra-articular administration of GC (figure 2). The GC effect on citrullination and PAD expression was paralleled by a significant decrease (p<0.05) in the lining thickness (from a median score 1, range 0-2 to a median score of 0.5, range 0-1) and cell infiltration (from a median score 3, range 2-3 to a median score of 2, range 1-3), but not vascularity. All patients were good clinical responders as evaluated by macroscopic investigation of the joint inflammation at the time of arthroscopy and retrospective scoring of photo records. In contrast methotrexate treatment had no effect on either synovial inflammation or local expression of CP and PAD (figure 3), despite a good clinical response in 9/11 treated patients with a significant decrease in the DAS28 score from a mean±SEM of 5.5±0.3 to a mean±SEM of 3.4±0.4 after 3 months of treatment. Interestingly the same was true when only EULAR responders (n=9) were included in the analysis. In vitro GCs have a direct effect on cellular expression of CP, potentially through a PAD dependent mechanism. Apart from their inflammation damping effect, GC may also directly target the process of citrullination. To investigate this, we tested the effect of DXM on CP and PAD enzymes expression in vitro in SFMC and PBMC paired samples of RA individuals. Confirming our in vivo results, DXM decreased the expression of CP, PAD4, but also PAD2, in SFMC (figure 4) but not PBMC (data not shown). The effect of DXM is dose-dependent and present at doses as low as 1 μM (figure 5). No such effects were observed for methotrexate when tested in the same systems (additional file 1). GC effect can be reproduced ex vivo in synovial explants biopsies. SFMC are in vitro surrogate replacements of the synovial biopsy complex milieu, though lacking important characteristics such as the local complex intracellular interplay. To circumvent this caveat, we established explants from a synovial biopsy obtained from an RA patient at open surgery for hip replacement and tested the effect of GC to further confirm our results. DXM treatment of the explants decreased the expression of CP and PAD4 but not PAD2, while parallel histological investigation did not reveal major changes in the histological composition of the samples. Once again no such effects were observed for methotrexate (additional file 2). Discussion Protein modification through post translational citrullination in the rheumatoid joint is thought to play an important role in perpetuation of local chronic inflammation in the presence of specific anti citrulline immunity. This is the first report showing that synovial citrullination is actively modulated by anti rheumatic treatment. Previous studies characterized synovial expression of citrullination showing a general increase in the level of citrullination in the presence of active inflammation [2, 4-6, 11, 15]. These early reports used one particular method for detection of citrullination, consisting in the chemical modification of the citrullinated tissue proteins to allow their recognition by an antibody raised against similarly modified citrullinated proteins [16]. However this technique is no longer available and in our hands the only other antibody performing well in immunohistochemistry is F95 [10]. As with the anti modified citrulline antibody, F95 is supposed to recognize a large array of citrullinated molecules independent of the amino acid context. We have previously demonstrated that this antibody is able to specifically recognize at least one RA relevant citrullinated antigen, namely citrullinated fibrinogen [17]. However a somewhat more restricted pattern of citrullination as compared to the anti modified citrulline antibodies has been observed both in synovial biopsies [11] and other tissues, such as lungs of smoking healthy individuals [18]. Interestingly using this new antibody, De Rycke et al were able to describe a RA specific intracellular F95 staining [11]. We confirm this finding by the virtually absence of F95 positive cells in synovial biopsies of healthy individuals. Only few reports on PAD synovial expression are currently available [2, 11, 19]. The most thorough investigation of synovial PAD expression was performed by Foulquier et al [2] showing correlation between expression of both PAD2 and PAD4 expression with the degree of synovial inflammation. We confirm these findings and identify the same correlation for the CP expression. The novel finding of the current study is the reduction in citrullination and PAD expression induced by intra-articular GC, which was not observed after MTX treatment. One potential explanation for this difference is the different time points to obtain the follow up biopsy, 2 weeks after treatment initiation with GC as compared to 8 weeks for MTX. However, the two different time points have been chosen in accordance with the clinical expectation of maximal effect of the administered drug, which would theoretically increase the chance to observe a change, not only for GC, but also for MTX. We were able to also demonstrate a direct effect of GC independent of inflammation in our synovial explants. This is further supported by the results of the in vitro experiments, where we observed a dose dependent effect of GC despite obvious difficulties in standardization of a quantitative analysis using immunohistochemistry on cells. Interestingly while GC decreased citrullination and PAD4 expression in SFMC, no such effect was observed in PBMC. This could be due on one hand to the lower baseline levels and, as a consequence, to the lower sensitivity to detect changes of expression of the investigated molecules in PBMC as compared to SFMC and, on the other hand, to different regulatory mechanisms and cellular activation state of PBMC as compared to SFMC, as previously suggested [20]. Despite major advances in understanding the central role of citrullination and anti citrulline immunity in RA pathogenesis, we face a striking lack of knowledge regarding regulatory factors responsible for induction, perpetuation and/or amelioration of the process of citrullination, both locally in the joint and even more general in other tissues where citrullination occurs either under physiological (skin) [21] or pathological conditions (lungs of smokers) [18, 22]. It is generally accepted that citrullination accompanies inflammation and it has been suggested that induction of citrullination by inflammatory stimuli such as TNF occurs following PAD4 activation and induction of signaling pathways dependent on NF-kB [23]. Interestingly, the anti inflammatory effects of GC are at least partially dependent on NF-kB as demonstrated by their lack of effect in animal models of acute inflammation in NF-kB knock-out mice [24]. In contrast, MTX appears to have limited effects on NF-kB activation as demonstrated by high levels of activation in PBMC of MTX treated RA patients that can be reversed by anti TNF agents [25]. These findings suggest that GC might affect citrullination through PAD4 downregulation through a NF-kB dependent mechanism. Conclusion The inflamed RA synovium is characterized by high expression of intracellular CP and PAD4 enzyme that is reversed through GC, but not MTX treatment. Further investigation of the exact mechanism of action and comparison with other anti rheumatic drugs is warranted. Abbreviations: CP, citrullinated proteins; PAD, peptidylarginine deiminase; SF; synovial fluid; PB, peripheral blood; MC, mononuclear cells; RA, rheumatoid arthritis, ACPA, anticitrullinated protein antibodies, DMARD, disease modifying anti rheumatic drugs; GC glucocorticoids, MTX, methotrexate; FCS, fetal calf serum; DXM, dexamethasone. Competing interests The authors have no competing interests to declare. Authors contribution DM, AIC and SR participated in study design, collection and interpretation of the data, and manuscript writing. ME and EaK participated in collection of data. APN and GP participated in data collection, interpretation of the data and manuscript writing. DM and SR have contributed equally to this work. All authors have read approved the final manuscript.