Progressive destruction of articular structures and chronic inflammation of synovial joints are major pathophysiological outcomes of rheumatoid arthritis (RA) 1. As the disease progresses, destruction of joint cartilage and, eventually, loss of joint function cause excessive morbidity and disability. Current approaches to drug therapy for RA focus predominantly on the alleviation of inflammation, pain, and disease progression. Among the medicinal strategies, nonbiological disease-modifying antirheumatic drugs (DMARDs) (for example, azathioprine, gold sodium thiomalate, chloroquine phosphate, and methotrexate [MTX]), steroidal anti-inflammatory drugs (SAIDs) (for example, prednisolone and methylprednisolone), and nonsteroidal anti-inflammatory drugs (NSAIDs) (for example, piroxicam and diclofenac) have already been successfully employed. The new group of biologics specifically targets inflammatory cytokines (for example, tumour necrosis factor [TNF] inhibitor etanercept) or receptors 23.

Despite recent progress in controlling inflammation, little cartilage repair has yet to be observed. Probably, suppression of inflammation is not sufficient to restore joint structure and function, and significant cartilage repair may be achieved only by activation of local chondrocyte regeneration 4. This underlines the need to identify distinct genes of RA-related chondrocyte dysfunction and to elucidate potential molecular mechanisms, markers, and pharmacological targets in human chondrocytes that might be involved in cartilage regeneration and suppression of inflammation. Gene expression profiling may be of help here to offer a better molecular understanding of chondrocyte dysfunction and regeneration and to disclose new therapeutic strategies 5.

Key mediators of joint destruction are RA synovial fibroblasts (RASFs), which directly destroy cartilage by secreting matrix-degrading enzymes 67. Numerous studies on the gene expression and protein secretion of RASFs have elucidated potent diagnostic and therapeutic targets in RASFs that mediate direct joint destruction and inflammation 8910111213. Recent studies have offered insight into the mechanisms of drug action; the molecular effects on RASFs following treatment with frequently used antirheumatic drugs were determined by genome-wide expression profiling 14.

Beyond direct cartilage destruction, RASFs maintain inflammation in synovial joints and induce chondrocyte dysfunction by releasing proinflammatory cytokines, in particular TNF-alpha and interleukin (IL)-1-beta, and catabolic mediators 615. Inflammatory and catabolic stimuli from RASFs cause indirect cartilage destruction; a disturbed tissue homeostasis and a shift to catabolic mechanisms lead to suppressed matrix synthesis and induce the production of degradative mediators by chondrocytes, such as matrix metalloproteinases (MMPs), prostaglandins, and nitric oxide 1617. Recently, we determined the RASF-induced expression profile in human chondrocytes that disclosed genes that are related to cartilage destruction and that involve marker genes of inflammation/nuclear factor-kappa-B (NF-κB) signalling, cytokines, chemokines and receptors, matrix degradation, and suppressed matrix synthesis 18. Although much is known about RASFs as key mediators of cartilage destruction in RA, researchers have scarcely analysed the molecular mechanisms of cartilage regeneration induced by antirheumatic treatment. Thus, the aim of this study was to establish an interactive in vitro model that comprehensively illustrates the diversity of antirheumatic drug effects on human chondrocytes and that offers the opportunity for parallel and future drug testing. To reveal marker and target genes for stimulation of cartilage/chondrocyte regeneration and suppression of inflammation was an additional goal of this study.

In the present study, human chondrocytes were cultured in alginate beads and were stimulated with supernatant of RASFs, healthy donor synovial fibroblasts (NDSFs), and drug-treated RASFs, respectively. Genome-wide microarray analysis was performed to determine RA-related gene expression and antirheumatic drug response signatures in human chondrocytes. Real-time reverse transcription-polymerase chain reaction (RT-PCR) and enzyme-linked immunosorbent assay (ELISA) were performed for validation of microarray data.

To our knowledge, this is the first genomic study that analysed the molecular mechanisms of RA-related chondrocyte dysfunction and regeneration in response to treatment with commonly used antirheumatic drugs. The study was based on a previously published in vitro model 18 that induced gene expression of inflammatory and destructive mediators and repressed regeneration and matrix formation when chondrocyte bead cultures were exposed to the secreted factors of a well-characterised RASF cell line. RASF-stimulated human chondrocytes showed a disturbed homeostasis on the molecular level, and the RA-induced gene expression signatures were considered to be relevant for chondrocyte dysfunction in RA.

In the present study, we show that the different classes of drugs exhibited distinct effects on the RA-induced signature in human chondrocytes. SAIDs were most effective to revert the molecular changes from the 'diseased' to the 'healthy' pattern. SAIDs were followed by the DMARDs azathioprine, gold sodium thiomalate, and MTX, whereas the NSAIDs and the DMARD chloroquine phosphate had the least effects. Thus, the molecular data reflect the clinical experience of therapeutic efficiency. Furthermore, best responses were associated with a strong suppression of proinflammatory cytokines (IL-6 and IL-23A), chemokines (CCL-20 and CXCL-8), and COX-2. Patterns typical for putative regenerative processes with downregulation by RASF supernatant and reversion or even upregulation upon treatment were seen mostly for matrix genes but also for some factors, which may be inductors of regeneration like CTGF 30 or CYR-61 31. The model that we applied for this study was selected as a compromise of the advantages for standardisation, availability, and comparability with parallel and future testing of different drugs and the disadvantage of an in vitro cartilage model and a transformed RASF cell line.

The RASF cell line certainly does not reflect all aspects of the in vivo situation and may differ in some aspects from primary fibroblast cultures of RA patients. Nevertheless, previous studies have demonstrated that the RASF cell line is a prototype of activated SFs expressing genes and secreting inflammatory cytokines (for example, IL-1α, IL-1β, IL-6, IL-11, IL-16, IL-18, CXCL-1–3/Gro-α-γ, CXCL-8/IL-8, CCL-2/monocyte chemoattractant protein-1, basic fibroblast growth factor, and leukaemia inhibiting factor) and matrix-degrading enzymes (for example, MMP-1, cathepsin-B, and cathepsin-L) associated with the pathomechanism of RA 14182122. The current setting of the model is focused on the effects of RASFs on chondrocyte gene expression and excludes the impact of lymphocytes and macrophages. This has the advantage that effects can be precisely attributed to the secretome of RA fibroblasts. A composed model with several different cell types would impede functional interpretation of the signature, especially with respect to the fact that lymphocyte presence and activity can vary widely from patient to patient and such differences cannot be addressed without knowing the individual components. Thus, stepwise development of such signatures with secretomes of other cell types will account for these additional effects and will help to interpret, at the end, the effects of drugs on more complex settings like cocultures of different cell types or even inflamed tissues. In addition, during recent years, activated RASFs have been determined to be the key players of cartilage destruction in RA by perpetuating the proinflammatory environment in synovial joints and by destroying cartilage matrix and chondrocyte homeostasis 6715. Therefore, the model reflects major mechanisms related to RA-induced cartilage destruction. The alginate bead culture of chondrocytes was chosen because (a) human chondrocytes could be cultured batchwise in a phenotype-stabilising surrounding, (b) human chondrocytes could be stimulated batchwise with conditioned supernatant of untreated and drug-treated SFs, and (c) total RNA could be isolated easily from human chondrocytes after isolation from the alginate.

The different soluble factors secreted by RASFs were considered to mediate the RA-induced gene expression pattern in human chondrocytes. Treatment of the RASF cell line with antirheumatic drugs was shown to repress many of these proinflammatory factors, with SAIDs being most effective 14. Therefore, we hypothesise that antirheumatic drugs exert their effects predominantly on RASFs and their secretome but may also act on human chondrocytes directly like SAIDs.

This study disclosed SAIDs to be most effective in reverting RA-induced gene expression in human chondrocytes even beyond the 'healthy' level, in particular the expression of genes associated with inflammation/NF-κB (BCL2-A1 and COX-2) and cytokine/chemokine activity (CXCL-8/IL-8 and IL-6). This is consistent with the pathway analysis that revealed SAIDs to target particularly pathways involved in cytokine/chemokine activity and ECM remodelling. SAID treatment of RASFs has been shown to repress the synthesis of the proinflammatory IL-1β and CXCL-8/IL-8 in SFs 14 and thus prevent human chondrocytes from stimulation by these factors. This is in line with the clinical application of SAIDs for intra-articular injection to suppress inflammation and disease activity in the short term. Consistent with our results, general effects of SAIDs include inhibition of the synthesis of inflammatory cytokines, reduction of COX-2 expression, and immune suppression 323334. Furthermore, glucocorticoid treatment of RA patients has been shown to reduce progression of cartilage destruction 3536. Thus, SAID therapy is appropriate to achieve rapid suppression of cartilage destruction and to control symptoms. For long-term treatment, however, adverse effects provoke serious problems 3233.

Hierarchical clustering and principal components analyses revealed that the DMARDs azathioprine, gold sodium thiomalate, and MTX effectively revert the RA-related chondrocyte gene expression toward the level of 'healthy' expression. Again, genes involved in inflammation/NF-κB signalling, cytokine/chemokine activity, immune response, proliferation/differentiation, and matrix remodelling were predominantly involved. In accordance with these results, DMARDs are described to interfere with the disease process, thereby impeding both the inflammatory and the destructive processes in RA 2. Among the DMARDs, MTX is regarded as the most effective cornerstone of RA therapy 3738. Interestingly, only marginal differences between these three DMARDs were found although their molecular modes of action differ. Azathioprine is a prodrug that is converted via 6-mercaptopurine by a series of transferases, kinases, and reductases to exert cytotoxicity by DNA incorporation and by inhibiting purine de novo synthesis 39. Apparently, RASFs were capable of metabolising azathioprine into active compounds because cytotoxic effects of azathioprine on RASFs were determined by MTS cell proliferation assay (data not shown). MTX is a folate analogue and inhibits methylation processes, and gold sodium thiomalate has a complex pharmacology, which recently was reported to interfere with COX-2 transcription 4041. In vivo, these actions may differ with respect to the cell type (suppression of lymphocyte proliferation, induction of apoptosis in activated T cells and monocytes, and inhibition of macrophage activation) or the kinetics of drug action.

Treatment with chloroquine phosphate, in contrast, resulted in only minor effects. Chloroquine phosphate is an antimalarial drug used in the treatment of RA. The exact mechanism of drug action remains unknown. This study showed that chloroquine phosphate repressed the expression of selected RA-related genes associated with inflammation/NF-κB signalling and immune response. However, the expression of the majority of RA-related genes was not reproducibly influenced following treatment with chloroquine phosphate. In agreement with the results of this study, chloroquine phosphate has been described to fail in inhibiting radiographic joint destruction and to have a relatively slow onset of action when compared with other DMARDs 3442.

NSAIDs were described to relieve joint swelling but to have only minor effects on disease progression and cartilage breakdown. As symptomatic agents, NSAIDs were supposed to inhibit prostaglandin synthesis by COXs and to help to control symptoms but to fail to retard or even to heal RA-related joint destruction 2343. This is in line with the results from microarray analysis that determined only minor effects for diclofenac and moderate effects for piroxicam on chondrocyte RA-related gene expression. Piroxicam effects contrasted with those of diclofenac by substantially repressing COX-2 expression in chondrocytes. Interestingly, diclofenac treatment of RASFs has been shown to have minor effects on the expression profile of disease-related genes in SFs 14. Probably, diclofenac exerts the RA-related effects on nonfibroblastic cell types that infiltrate the synovium rather than on SFs 44.

Overall, molecular differences of the various drugs are reflected by the clinical experience that steroids and the three DMARDs MTX, azathioprine, and gold sodium thiomalate, but not chloroquine phosphate or NSAIDs, may be effective in inhibiting radiographic joint destruction 23344243. Furthermore, SAIDs are best at suppressing the RA-induced changes in chondrocytes but are associated with many side effects. Thus, new therapeutic strategies may focus especially on molecular effects induced by steroids and the effective DMARDs but not induced or insufficiently induced by chloroquine phosphate and NSAIDs. This qualitative difference of drugs is best reflected by the expression pattern of COX-2 and several cytokines/chemokines (IL-6, CXCL-8/IL-8, IL-23A, and CCL-20). Inversely, CTGF, CYR-61, and TXNIP are suppressed by RASFs and reversely induced by the effective drugs.

Inhibition of the transcription of COX-2 can be considered as more effective than blocking of the enzyme activity. Transcription of COX-2 was highest suppressed compared with all other genes by SAIDs, which were also the most effective drugs. Interestingly, prostaglandin E₂ synergistically with IL-23 was reported to favour human Th17 expansion 45 and prostaglandin E₂ may also induce IL-23 in bone marrow-derived dendritic cells 46. Here, we find both increased COX-2 and IL-23 expression in chondrocytes after stimulation by RASFs, suggesting that chondrocytes may contribute to the development of Th17 cells. This T-helper cell type is discussed in several aspects of proinflammatory activities and may trigger a positive feedback loop via IL-6 47, another candidate found to be induced in chondrocytes by RASFs and also involved in the differentiation of Th17 cells. This suggests that cartilage, when stimulated by RASF supernatant, may contribute to the proinflammatory network of secondary immune reactions. With COX-2 (prostaglandins), IL-23A, and IL-6 as inductors of this process, targeting these molecules could be favourable.

Concerning induction of regenerative processes, CTGF and CYR-61 may be potential candidates. CYR-61 is reported as a regulator of chondrogenesis 31 and belongs to the same CCN family of molecules as CTGF 48. However, CTGF is reported to be involved in many fibrotic processes, including lung 49 and kidney 50, indicating that only local application could be advisable.

As long as a causative therapy of RA is not available, the aim will be to find the appropriate combination of targets that is most effective in remitting or even healing of RA-related cartilage destruction. Since sensitive diagnostic tools to assess cartilage destruction/repair are rare, in vitro models for the evaluation of drug efficiency and the identification of potent targets are necessary. Candidate genes that evolved in this study for new therapeutic approaches include suppression of specific immune responses (COX-2, IL-23A, and IL-6) and activation of cartilage regeneration (CTGF and CYR-61). Further studies are needed to investigate the influence of immune cells on chondrocytes to complete the molecular effects induced by inflammatory processes in arthritis.

This in vitro study provides comprehensive insight into the molecular mechanisms involved in RA-induced chondrocyte dysfunction and in drug-related chondrocyte regeneration. Our findings indicate that RA-relevant stimuli result in the molecular activation of inflammatory and catabolic processes in human chondrocytes that is again reverted by antirheumatic treatment. Molecular differences of the various drugs are reflected by the clinical experience. Furthermore, this study provides evidence of numerous pharmacological marker genes to define and induce chondrocyte integrity and regeneration, including established genes (COX-2, CXCR-4, IL-1RN, IL-6/8, MMP-10/12, and TLR-2) and new genes (ADORA2A, BCL2-A1, CTGF, CXCR-7, CYR-61, HSD11B-1, IL-23A, MARCKS, MXRA-5, NDUFA4L2, NR4A3, SMS, STS, TNFAIP-2, and TXNIP). Candidate genes that evolved in this study for new therapeutic approaches include suppression of specific immune responses (COX-2, IL-23A, and IL-6) and activation of cartilage regeneration (CTGF and CYR-61). Thus, from the molecular point of view, the present study helps to elucidate the role of chondrocytes during cartilage destruction in RA and during antirheumatic therapy and may contribute to the development of novel therapeutic chondro-protective agents and strategies.

ADORA2A: adenosine A2A receptor; BCL2-A1: BCL2-related protein-A1; CCL-20: chemokine (C-C motif) ligand-20; COX: cyclooxygenase; CTGF: connective tissue growth factor; CXCR-4: chemokine (C-X-C motif) receptor-4; CYR-61: cysteine-rich angiogenic inducer-61; DMARD: disease-modifying antirheumatic drug; ECM: extracellular matrix; ELISA: enzyme-linked immunosorbent assay; GAPDH: glyceraldehyde 3-phosphate dehydrogenase; GCOS: GeneChip Operating Software; GEO: Gene Expression Omnibus; HSD11B-1: hydroxysteroid (11-beta) dehydrogenase-1; IC₂₀: 20% inhibitory concentration; IL: interleukin; IL-1RN: interleukin-1 receptor antagonist; KEGG: Kyoto Encyclopaedia of Genes and Genomes; MARCKS: myristoylated alanine-rich protein kinase C substrate; MIP-3α: macrophage inflammatory protein-3-alpha; MMP: matrix metalloproteinase; MTS: 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium; MTX: methotrexate; MXRA-5: matrix-remodelling associated-5; NDSF: normal (healthy) donor synovial fibroblast; NDSFsn: supernatant of untreated normal (healthy) donor synovial fibroblast; NDUFA4L2: NADH dehydrogenase (ubiquinone) 1 alpha subcomplex, 4-like 2; NF-κB: nuclear factor-kappa-B; NR4A3: nuclear receptor subfamily 4, group A, member 2; NSAID: nonsteroidal anti-inflammatory drug; PCR: polymerase chain reaction; PLA2G2A: phospholipase A2 group IIA; PTX3: pentraxin-related gene; RA: rheumatoid arthritis; RASF: rheumatoid arthritis synovial fibroblast; RASFsn: supernatant of untreated rheumatoid arthritis synovial fibroblast; RIPK2: receptor-interacting serine-threonine kinase 2; RMA: Robust Multichip Analysis; RSAD2: radical S-adenosyl methionine domain containing 2; RT-PCR: reverse transcription-polymerase chain reaction; SAID: steroidal anti-inflammatory drug; SDF-1: stromal cell-derived factor-1; SF: synovial fibroblast; STAT: signal transducer and activator of transcription; STS: steroid sulfatase; TGF-β: transforming growth factor-beta; TLR: Toll-like receptor; TNF: tumour necrosis factor; TNFAIP-2: tumour necrosis factor-alpha-induced protein-2; TXNIP: thioredoxin interacting protein; VCAN: chondroitin sulfate proteoglycan 2; WISP2: WNT1 inducible signalling protein 2.

CK is an employee of TransTissueTechnologies GmbH (Berlin, Germany). CK, TH, and MS hold a patent related to the content of this manuscript: 'An in vitro cell interaction culture system for testing and developing medicaments' (DE 10 10 54 20 B4). MS is a shareholder of CellServe GmbH (Berlin, Germany) and works as a consultant for BioTissue Technologies AG (Freiburg, Germany). The other authors declare that they have no competing interests.

KA helped to perform the gene expression data processing, participated in the design and coordination of the study, helped to draft the manuscript, and helped to conduct the cell culture experiments and to perform the microarray and ELISA experiments. TH helped to perform the gene expression data processing, participated in the design and coordination of the study, and helped to draft the manuscript. KA and TH contributed equally to this article. CL helped to perform the gene expression data processing, participated in the design and coordination of the study, and shared responsibility for collection, assembly, and analysis of data and data interpretation. JR helped to perform the gene expression data processing and participated in the design and coordination of the study. LM shared responsibility for collection, assembly, and analysis of data and data interpretation. AW helped to conduct the cell culture experiments and to perform the microarray and ELISA experiments. MS and CK helped to conceive the study and participated in its design and coordination. All authors read and approved the final manuscript.