Synovial expression of Th17-related and cancer-associated genes is regulated by the arthritis severity locus Cia10.

Rheumatoid arthritis (RA) is a complex autoimmune disease characterized by chronic joint inflammation that commonly leads to joint destructive changes, disability and reduced quality of living. RA affects 1% of the population and is regulated by both genetic and environmental factors 1 . Most of the genetic studies have focused in the identification of susceptibility genes. The HLA-DRB1 shared-epitope association with RA susceptibility has been known for over two decades, and in recent years several non-MHC susceptibility alleles have been identified, including PTPN22 2 , CTLA4 3 , TNFAIP3 and TRAF1 4 , 5 . Yet, little is known about the genetic regulation of disease severity and joint damage, which are among the best predictors of disease outcome and disability. We have identified the non-MHC arthritis severity and joint damage quantitative trait locus (QTL) Cia10 on rat chromosome 2, studying an intercross between MHC-identical DA (severe arthritis) and ACI (arthritis-resistant) rats 6 . DA.ACI(Cia10) congenic rats, which are identical to DA except for the presence of ACI alleles at the Cia10 interval, were generated using a genotype-guided strategy and found to develop a significantly milder form of pristane-induced arthritis (PIA), and to preserve normal joint architecture compared to DA rats. DA.ACI(Cia10) congenics also had significantly reduced synovial tissue levels of mRNA of pro-inflammatory cytokines central to RA pathogenesis 7 . During the refining of the congenic interval towards positional cloning of the Cia10 gene we aimed to identify molecular pathways and potential candidate genes involved in protecting DA.ACI(Cia10) congenics, using genome-wide microarray gene expression profiling of synovial tissues. In the present study we report the presence of ACI alleles at the Cia10 interval correlates with reduced expression of genes implicated in Th17 responses, including IL-17A. Additionally, congenics had reduced expression of genes involved in the regulation of cancer growth and invasion, which may suggest new targets for therapies aimed at reducing synovial pannus growth and cartilage and bone invasion and destruction.

DA (arthritis-susceptible) and ACI (arthritis resistant) rats were originally purchased from Harlan Sprague Dawley (Indianapolis, IN). DA.ACI(Cia10) congenic rats were generated as previously described 7 . Briefly, a 52.6 Mbp region from the ACI strain containing the Cia10 interval on chromosome 2 ( Figure 1A ) was introgressed into DA through genotype-guided breeding. This strategy selected for ACI alleles at the Cia10 interval while at the same time excluding donor genome contamination at other loci known to regulate arthritis severity 6 , 55 . After five backcrosses offspring heterozygous at the Cia10 interval were intercrossed to generate Cia10 homozygotes for ACI alleles (DA.ACI(Cia10)) to be used in experiments. The experimental protocol was reviewed and approved by the FIMR Institutional Animal Care and Use Committee. Male DA (n=12) and DA.ACI(Cia10) (n=6) congenic rats 8-12 week-old were anesthetized and injected intradermally at the base of the tail with 150 μl of pristane (MP Bio, Solon, OH) divided into two injection sites (day 0) 7 , 56 . Arthritis severity was assessed using a previously described 80-point scoring system that uses the sum of scores to generate an arthritis severity index (ASI) 7 . Ankle synovial tissues were collected for analysis 21 days post-induction of PIA. Synovial tissue was homogenized in lysis buffer using a rotor-stator homogenizer and total RNA was extracted using RNeasy (Qiagen, Valencia, CA) and including a DNase treatment step, according to manufacturer’s instructions. RNA was quantified and assessed for purity using the NanoDrop spectrophotometer (Rockland, DE), and RNA integrity was verified using the BioAnalyzer 2100 (Agilent, Palo Alto, CA). The microarray protocol was previously described 40 . Briefly, 200 ng of total RNA was amplified and biotinylated using the TotalPrep labeling kit (Ambion, Austin, TX). RNA samples were hybridized to the RatRef-12 Expression BeadChip (Illumina, San Diego, CA), which contains 22,524 probes covering 21,922 rat genes selected primarily from the NCBI RefSeq database (Release 16). Hybridization was carried out in Illumina IntelliHyb chambers, followed by washing and staining with Cy3-streptavidin. The array was scanned on a high-resolution Illumina BeadArray reader using a two-channel 0.8 μm resolution confocal laser scanner. Protocols were previously optimized for use with the Illumina Whole-Genome Expression platform. Illumina Bead studio software (Version 2.0) was used to extract data, which was normalized using the cubic spline algorithm. In order to reduce the false positive rate, only genes expressed by all 18 synovial tissues were used for analyses. Mean intensity values for genes detected in all 18 samples were Log 2 -transformed and expression data from DA and DA.ACI(Cia10) compared using the t-test. Fold-differences were calculated by elevating 2 to the difference between the means of log 2 -transformed DA and log 2 -transformed DA.ACI(Cia10) values [2 (mean Log 2 DA − mean Log 2 Cia10) ], and vice-versa for DA.ACI(Cia10) fold-differences. Genes with a P-value ≤0.01 (t-test) and a fold-difference ≥1.5 between DA and DA.ACI(Cia10) were considered significantly different. Ingenuity IPA 5.5.1 program (Ingenuity, Redwood City, CA, USA) and publicly available databases such as Pubmed and Online Mendelian Inheritance in Man (OMIM) were used for pathway discovery and analysis. Enrichment for biological functions and disease groups was determined with the IPA software and calculated using the Fisher’s exact test with a cutoff p-value of ≤0.05, using the Benjamini-Hochberg correction. We also looked for genes a) expressed in synovial tissues of one strain and not by the other, or b) predominantly in one strain and not in the other, since those could represent even more significant differences relevant to disease pathogenesis. None of the genes met criteria “ a ” (expressed only by one strain and not by the other), but several met criteria “ b ”. cDNA was synthesized from the same RNA samples hybridized to the RatRef-12 Expression BeadChip. 2μl of total RNA was used for cDNA synthesis using the Superscript III kit (Invitrogen Carlsbad, CA). cDNA was diluted 1:10 for qPCR. Rat specific primers and probes were designed using the Universal ProbeLibrary (Roche, Indianapolis, IN) ( table 1 ). Gapdh was used as an internal control and all samples were run in duplicate. The average threshold cycle (Ct) values were used to analyze relative gene expression of Cd53, Fcgr1a, Gp49b, Timp2 Reck, Nov and Gstm1. Probes were used at a final concentration of 100nM, 5’ ends were labeled with FAM and 3’ ends with TAMRA. Primers were used at 200 nM with Eurogentec qPCR mastermix (Eurogentec, San Diego, CA). Reactions were run on an ABI 7700 qPCR thermocycler at 50°C for 2 minutes, 95°C for 10 minutes, and 45 cycles of 95°C for 0.15 minutes and 60°C for 1 minute. Relative expression of genes was normalized to Gapdh in each sample (ΔCt), and ΔCt values were used for t-test analysis. qPCR fold-differences were calculated using the 2 -ΔΔCt method 57 . The miRBase.org database was used to identify the microRNAs located within the Cia10 interval, and TargetScan.org was used to predict their targets. The eight microRNAs contained within the Cia10 and with differentially expressed predicted targets were sequenced as described below. Splenic genomic DNA (gDNA) was extracted using DNeasy Blood and Tissue Kit (Qiagen). Primers were designed using the Whitehead Institute Primer3 website ( http://frodo.wi.mit.edu/primer3/input.htm ) to amplify 500-1000bp regions surrounding eight known microRNA sequences located within the Cia10 interval ( table 1 ). PCR products were sequenced and analysed with the DNASTAR sequencing analysis software (Madison, WI). Synovial tissues were collected from a different group of DA and from DA.ACI(Cia10) joints 21 days after the induction of PIA for fibroblast-like synoviocyte (FLS) isolation. FLS were isolated as previously described 40 . Cells from passage four or greater (>95% FLS purity) were used for the experiments. 4.6×10 4 cells per cell line per strain were co-transfected with a NFκB luciferase reporter plasmid (NFκB binding sequences were originally cloned into a Promega pGL2 construct at Dr. L. Klampfer’s laboratory, Albert Einstein College of Medicine, Bronx, NY) and a Renilla internal control plasmid (Promega, Madison, WI) using Lipofectamine 2000 (Invitrogen). Control cells were transfected with a luciferase plasmid without a promoter. Following transfections cells were cultured overnight on FBS-free medium, followed by stimulation with IL-1β 10ng/ml for 48h. The NFκB reporter activity was measured with the TD 20/20 Luminometer (Turner Designs, Sunny vale, CA). Arbitrary units of the Renilla reporter were used for internal normalization. The t-test was used to compare the expression of genes expressed by all samples. Chi-squared and Fisher exact tests were performed using SigmaStat 3.0 (SPSS, Chicago, IL).

DA.ACI(Cia10) congenic rats developed significantly milder arthritis compared with DA, with a median ASI of 12.3 compared with 30.4, respectively (P=0.002 Mann-Whitney test) ( Figure 1B ). These observations were in agreement with our previous studies 7 . 7,593 (34.6%) of the 21,922 genes present in the array were expressed by all 18 synovium samples and were used for analysis. 310 of the 7,593 genes (4.08%) expressed by all synovial tissues met the criteria for differential expression. 120 of these 310 genes (38.7%) were expressed in significantly higher levels in DA.ACI(Cia10) compared with DA rats. 190 of the 310 genes (61.3%) were expressed in significantly increased levels in DA synovium, and were reciprocally reduced in DA.ACI(Cia10). The genes with the most significantly increased expression in the DA.ACI(Cia10) congenics and reduced expression in DA included Mup4 (4.88-fold, P=0.0067), Aldh1a1 (3.7-fold P=0.009), Rpesp (3.56-fold, P=0.0082), Igfbp6, (3.56-fold, P=0.0068), Gstm1 (2.58-fold, P=0.0029), Timp2 (2.15-fold, P=0.0024) and Reck (2.1-fold, P=0.005) ( table 2 ). Among the 190 genes with significantly reduced expression levels in the synovial tissues of DA.ACI(Cia10) congenics and increased expression in DA was Gp49b (5.39-fold, P=0.0023), Cd53 (3.28-fold, P=0.004), Fcgr1a (2.94-fold, P=0.007), Emr1 (2.78-fold, P=0.00002) and Rbbp7 (2.68-fold, P=0.0051) ( table 2 ). Seven of the differentially expressed genes were confirmed with qPCR ( figure 2 ). Analyses of the 310 differentially expressed genes revealed an over-representation of ten disease categories and/or molecular and cellular functions as detected by the IPA pathway analyses ( table 3 ). These categories included cancer (89 genes), cell death (68 genes), reproductive system disease (this category predominantly included endometriosis and genes implicated in cancer of the reproductive system or breast) (57 genes), inflammatory disease (53 genes), hematological disease (37 genes) and others ( table 3 ). The IPA network/pathway detection analyses identified seven additional functional groups over-represented among the differentially expressed genes ( supplemental table 1 ). These included “Embryonic Development, Tissue Development, Cell Death” (27 genes), “Inflammatory Response, Connective Tissue Disorders, Tissue Morphology” (20 genes) and “Cell Cycle, Connective Tissue Development and Function, Organ Morphology” (19 genes), all with significant scores ( supplemental table 1 ). We used three approaches to identify differentially expressed cytokines, chemokines and other known inflammatory mediators or receptors: a) genes expressed by all 18 synovial samples and meeting the criteria for significance described in the Methods section (P≤0.01 plus ≥1.5-fold-difference; b) genes expressed by only one of the strains and not the other; c) genes predominantly expressed in one strain and not in the other. Using these combined parameters we identify 21 cytokines, chemokines, receptors or related genes of relevance ( table 4 ). IL-6 was preferentially expressed in 92% of DA (n=11) and only in 33% (n=2) congenic’s synovial tissues ( table 4 ). IL-11 was expressed in significantly increased levels in DA, compared with congenics (1.66-fold, P=0.007). These two pleiotropic cytokines are also expressed in increased levels in RA synovium, and are known to activate NFκB and osteoclasts, and to increase cell tolerance to oxidative stress, all phenomena that take place in the arthritic synovial tissues 8 - 11 . Cxcl10, Ccl12, and Ccl21b among others, were preferentially expressed, or expressed in significantly increased levels in DA, compared with congenics. Chemokine receptors Ccr6, Cx3cr1 and Cxcr3, as well as cytokine receptors, IL-1rl1 (IL-33r) and IL-3ra were also up-regulated in the synovial tissues from DA rats. Interestingly, Ccl21b, Ccr6 and Stat4 have been associated with RA susceptibility 12 . Additional important mediators of the inflammatory response up-regulated in DA included Csk, a key signaling gene that regulates T and B cell proliferation, activation, and migration that interacts with the RA susceptibility gene Ptpn22 2 . Four of the above genes, including IL-6, Stat4, Ccr6 and Cxcl10, have been implicated in the differentiation, activity or chemotaxis of Th17 cells, a cell type central to autoimmune disease pathogenesis. We next looked for the three genes more specifically related to Th17 cells: IL-17A, IL-17F and Rorc (RORgt). The Illumina RatRef microarray does not contain probes for IL-17A and IL-1F. Therefore we used qPCR to study the expression of these two genes. IL-17A was preferentially expressed in DA synovial tissues (93%), compared with DA.ACI(Cia10) congenics (43%), and this difference was statistically significant (p=0.025; Fisher Exact test, table 4 ). IL-17F was not expressed in either strain’s synovial tissues. Rorc (RORgt) expression was not significantly different between the DA and DA.ACI(Cia10) both in the microarray and qPCR (data not shown). Therefore, our results suggest that the Cia10 gene in involved in the regulation of differentiation or synovial tissue homing of Th17 cells. Tgfa, Tgfb2 and the receptor Tgfbr3 were up-regulated or predominantly expressed in DA.ACI(Cia10) congenics, compared with DA ( table 4 ). Tgfb and its receptors are known to suppress immune responses and to induce the differentiation of Treg cells 13 . Several genes up-regulated in DA.ACI(Cia10) congenics are known to interact with, or to be regulated by Tgfb, further suggesting a role for this gene in mediating protection in the congenics ( figure 3 ). The Vdr gene was predominantly expressed in congenics (80%) compared with DA (33%) ( table 4 ). IPA pathway and gene interaction analyses revealed that several of the most significantly differentially expressed genes were induced by or interacted with the Vdr, including Tgfb2 ( figure 3 ). The Vdr is known to inhibit the expression of cytokines like IL-6, which is in agreement with our observations of an inverse correlation between Vdr-Tgfb2 and IL-6. Ccl11 (Eotaxin) expression was increased in the protected DA.ACI(Cia10) congenics (1.93-fold, P=0.003). Interestingly, Ccl11 levels have been shown to correlate with reduced radiographic damage in RA via yet unknown mechanisms 14 . Ccl11 is a chemoattractant to eosinophils and Th2 cells, and modulates monocyte activity. 89 genes of the 310 (28.7%) differentially expressed genes are implicated in cancer and cancer-related phenotypes. These included oncogenes, cell-cycle regulators, tumor-suppressor genes, proteases and others. The 27 genes with the strongest literature support for a role in cancer are listed on table 5 . Among the cancer-favoring genes up-regulated in DA there were eight oncogenes or tumorigenesis genes, including Rbm3, Ikbke, Rcl, Kras, four cell proliferation enhancing genes (Mcts1, Aplnr, Ppap2a, and Ier3) and three invasion-associated genes (Ctsd, Cd53 and Loxl1). Two additional mediators of cancer invasion and metastasis, Cxcl10 and Cxcr3, were preferentially expressed by DA synovial tissues [DA=12 (100%), DA.ACI(Cia10)=3 (50%)] ( table 4 ). IL-11, which as described above was expressed in increased levels in DA ( table 4 ), has also been implicated in the pathogenesis of carcinomas 15 . Histone deacetylases (Hdac1, Hdac2 and Rbbp7) had increased expression in DA, and are also commonly up-regulated in cancers and considered to regulate cell differentiation and proliferation 16 . Nine of the genes up-regulated in DA.ACI(Cia10) are known to suppress a cancer phenotype. Specifically, four of these genes (Pawr, Per1, Igfbp6 and Thbd) negatively regulate cell proliferation and growth, and three genes (Rgs4, Nov and Tgfbr3) negatively regulate cell invasion ( table 4 ). Matrix metalloproteinases (MMPs) are known to increase invasion of cancer and synovial cells and are key mediators of cartilage and joint destruction in RA. Timp2 (2.15-fold, P=0.0024) and Reck (2.14-fold, P=0.005), two MMP inhibitors, were also significantly up-regulated in DA.ACI(Cia10). Four of the cancer-related genes (Cd53, Nov, Reck and Timp2) were confirmed with qPCR ( figure 2 ). Therefore, our results suggest a general increased expression of genes known to favor cell proliferation and invasion in DA, while the congenics has an increased expression of anti-proliferation and anti-invasion/anti-destruction (MMP inhibitors) genes. These observations provide new insight into the molecular processes regulating the reduced synovial hyperplasia and reduced cartilage and joint damage that we had previously reported on DA.ACI(Cia10) congenics compared with DA. 68 cell death regulatory genes ( table 3 ), including 43 apoptosis genes (13.9% of all differentially expressed) had a significantly different expression between DA and DA.ACI(Cia10) ( supplemental table 2 ). However, a similar number of pro-apoptosis [DA=11 and DA.ACI(Cia10)=8] and anti-apoptosis [DA=12 and DA.ACI(Cia10)=10] genes was up-regulated in DA and in DA.ACI(Cia10). Therefore, while several apoptosis genes were differentially expressed, there was not a clear suggestion of an increased or reduced apoptotic gene expression pattern in either strain. Genes with anti-oxidant properties (Cat, Mt3, Nfe2l1, and Sqstm1), including Gstm1, which has been associated with disease severity and articular damage in RA 17 , were amongst the most significantly up-regulated genes in DA.ACI(Cia10) ( figure 3 and table 4 ). Ten of the differentially expressed genes are known to interact with the NFκB pathway ( supplemental table 3 ). These included three NFκB activators (Ikbke, Ccl21b and Ikbkap) and one inhibitor (Ucp2) up-regulated in DA. As described above, IL-6 and IL-11, which are known to activate NFκB, were also expressed in increased levels, or preferentially expressed in DA. Among the genes up-regulated in DA.ACI(Cia10) congenics, there were those that reduce activation (Cat) and transcription (Pawr and Thbd) of NFκB. Additionally, several genes up-regulated in DA.ACI(Cia10) congenics have been reported to interfere with the NFκB pathway ( Supplemental figure 1 ). Therefore, we looked for 244 NFκB/C-rel target genes previously reported by others 18 and determined that 78 were expressed by all of our samples (32%). Only one of these genes (Acs14) was differentially expressed (2.6-fold up-regulated in DA.ACI(Cia10), P=0.003), suggesting that NFκB was not differentially activated between the two strains. An in vitro NFκB luciferase reporter assay performed in cultured FLS obtained from DA and DA.ACI(Cia10) rat synovial tissues and stimulated with IL-1β did not detect any significant difference in NFκB activity between the two strains, in agreement with the lack on an NFκB expression signature ( supplemental figure 1 ). 204 of the 7,593 genes expressed by all samples were contained within the Cia10 QTL region on rat chromosome 2. Sixteen of these 204 (7.8%) genes were differentially expressed between DA and DA.ACI(Cia10) congenics, which was more than the 76 (1%) that would have been expected by chance (P=0.018, Chi-square test). Seven of these sixteen genes were expressed in increased levels in DA.ACI(Cia10) and included Pde5a, Palmd, Tspan2, Nes, Adamtsl4 and two antioxidant genes Gstm1 (2.58–fold, P=0.0029) and Gstm2 (2.70–fold, P=0.0028; table 6 ). Nine of the above 16 differentially expressed genes were up-regulated in DA, and conversely down-regulated in DA.ACI(Cia10) congenics, and included Cd53, Fcgr1a, Shc1, Tnfaip8l2, gene similar to Gapdh, Sass6, Hist2h2ac, Ncu-g1 and gene similar to Nrf2 ( table 6 ). Cd53 (3.28-fold, P=0.004) and Fcgr1a (2.94-fold, P=0.007) were amongst the most significantly differentially expressed genes. The differential expression of Cd53, Fcgr1a and Gstm1 was confirmed using qPCR ( figure 2 ). Thirteen of the differentially expressed genes located within the DA.ACI(Cia10) interval are located in the same cytogenetic band on chromosome 2 (2q34) suggesting the possibility that there could be a polymorphism within the region affecting mRNA transcription or stability of multiple genes. None of the genes contained within the Cia10 interval were preferentially expressed in one strain versus the other. It was hypothesized that part of the differential gene expression pattern could be explained by polymorphisms in a miRNA located within the Cia10 interval. The polymorphism could lead to an alteration of the miRNA expression or function and modulation of target genes. Eight miRNAs are predicted to reside within the Cia10 interval on rat chromosome 2 (miR9-1, miR15b, miR16, miR92b, miR137, miR186, miR190b and miR760). While seven of these microRNAs had differentially expressed targets, none of them had more targets genes than would have been expected by chance (mir9-1: n=11, mir15b: n=16, mir16: n=16, mir92b: n=9, mir137: n=9, mir186: n=5, mir760: n=6). For further confirmation, we sequenced these seven miRNAs and their surrounding DNA (500-1000bp), which contains regulatory regions. However, no polymorphisms were detected between DA and ACI.

Increased disease severity and articular damage are associated with increased risk for disability, deformities and reduced life expectancy in patients with RA 19 , 20 . Yet, the genes that regulate disease severity and articular damage remain largely unknown. We have been interested in the identification of these genes and consider that they have great potential to generate new and perhaps better targets for therapies aimed at preserving joint integrity and function. In the present study we used synovial tissues from DA rats, which develop severe arthritis (PIA) with pronounced synovial hyperplasia and cartilage and bone destruction, and synovial tissues from arthritis-protected and non-erosive DA.ACI(Cia10) congenics. These two strains are genetically identical except for the presence of ACI alleles at the Cia10 interval, underscoring the magnitude of the effect of this single locus on clinical disease, histologic joint damage 7 and gene expression. DA.ACI(Cia10) and DA had significant differences in the expression of inflammatory mediators. That difference was not broad, but instead limited to a very specific set of genes, most of which have been involved in the regulation of cartilage and bone erosions, and articular damage. IL-6 and IL-11, which belong to the IL-6 family of genes and signal through gp130, were among the cytokines with significantly reduced expression in DA.ACI(Cia10) congenics. Both IL-6 and IL-11 are expressed in increased levels in RA synovium, and while IL-6 has been clearly implicated in arthritis pathogenesis and joint damage 8 , 9 , the literature on IL-11 is contradictory 10 , 11 . The chemokines Ccl12, Ccl21b and Cxcl10, and chemokine receptors Ccr6, Cx3cr1 and Cxcr3 were also expressed in reduced levels in congenics. Ccl12 is chemotactic for Ccr2-expressing monocytes and neutrophils, which have a key role in the development of synovial hyperplasia and joint erosions and damage 21 . Ccl21b is chemotactic for Ccr7-expressing T and B cells, stimulates FLS, and has been implicated in the organization of germinal center-like structures in synovial tissues of RA 22 , 23 . Cx3cr1 is expressed by myeloid cells, and its blockade ameliorates rodent arthritis and reduces articular damage 24 . Cxcl10 and its receptor Cxcr3 was predominantly expressed in DA, and these genes mediate T cell and mast cell chemotaxis, and increase FLS invasion in an autocrine and paracrine manner (Laragione et al. unpublished observations). Furthermore, blockade of Cxcr3 ameliorates adjuvant-induced arthritis in rats, also reducing joint erosion and damage 25 . Synovial tissues from DA.ACI(Cia10) congenics had increased expression of Tgfb2 and Ccl11. Tgfbr3 was also preferentially expressed in congenics’ synovial tissues. Tgfb is considered a suppressor of T cell responses and has a central role in the differentiation of Treg cells. Furthermore, treatment with Tgfb ameliorates autoimmune arthritis in rodents while anti-Tgfb antibodies exacerbate disease 26 . Ccl11, also known as Eotaxin-1, is an eosinophil chemoattractant expressed by T-cells and fibroblasts. Increased serum levels of Ccl11 have been associated with reduced radiographic damage progression in early RA 14 . Lastly, the Vdr was predominantly expressed in the synovial tissues of congenics. The Vdr has known IL-6-suppressive 27 , Th17 differentiation-inhibitory properties 28 , and Tgfb-inducing activities ( figure 3 ), which matches the scenario identified in this study. Vdr agonists also reduced arthritis severity 29 . Together, our findings suggest that Cia10 directly or indirectly regulates the expression of genes implicated in arthritis severity and joint damage, with increased levels or preferential expression of pro-inflammatory and joint damage-favoring genes in DA, while the opposite takes place in congenics. While IL-23, IL-17-F and RORgt (Rorc) were not differentially expressed in synovial tissues collected on day 21, four of the genes up-regulated or preferentially expressed in DA, and conversely down-regulated in DA.ACI(Cia10) congenics, suggest an increased presence of IL-17 producing or inducing cells: a) IL-17A was expressed in nearly all DA samples (93%), but only in 43% of congenics; b) Ccr6 is a cell surface marker of Th17 T cells 30 ; c) Stat4 mediates IL-23r signaling, which is required for the generation of Th17 cells; and d) Cxcr3 is expressed by mast cells, which have been considered the main source of IL-17 in the synovial tissue 31 . Furthermore, DA synovial tissues preferentially expressed IL-6, which is a key cytokine for differentiation of Th17 cells, but lacked Vdr expression, which is an inhibitor of Th17 differentiation. These observations raise the possibility that Cia10 could be a new gene controlling arthritis severity indirectly via the regulation of the differentiation and/or homing of Th17 cells to the joint. Four of the genes with the most significantly reduced expression in DA.ACI(Cia10) congenics are either associated with genetic susceptibility to RA (Ccr6, Ccl21b, Stat4) 12 , 32 , or as in the case of Csk, directly interact and regulate the activity of a RA susceptibility gene (PTPN22) 2 . Furthermore, Gstm1 null alleles have been associated with RA severity and radiographic erosive damage 17 , and DA synovial tissues did in fact have significantly reduced expression compared with DA.ACI(Cia10). However, sequencing of DA Gstm1 gene revealed no deletions, insertions or significant sequence variants that might explain the reduced expression (data not shown). Taken together, our results suggest that the Cia10 gene could interact with RA susceptibility genes to regulate their expression and perhaps function, raising the possibility of potential epistatic interactions. NFκB is a central regulator of synovial hyperplasia and articular damage 33 , 34 , and several RA susceptibility genes are involved in this pathway 35 . A subset of the differentially expressed genes in the present study, including IL-6, IL-11 and Ccl21b, are known activators of NFκB, and NFκB-interacting genes were differentially expressed, raising the possibility that Cia10 might be involved in the regulation of NFκB activity. However, we were not able to detect a differentially expressed NFκB transcriptional signature 18 . For further confirmation a NFκB luciferase reporter assay was studied in FLS from DA and DA.ACI(Cia10) congenics stimulated with IL-1β, and did not detect any significant difference in transcriptional activity. Therefore, our results suggest that Cia10 does not have a major effect on the regulation of the NFκB pathway. There are several similarities between the pathogenesis of RA synovial hyperplasia, its invasion and destruction of cartilage and bone, and the behavior of cancers 36 - 38 . Previous studies from our group and others have identified increased expression of cancer-related genes in FLS from arthritic rats and from patients with RA 39 - 41 . In the present study one of the predominant differentially expressed groups was related to cancer and cancer phenotypes. Genes involved in oncogenesis (Kras, Ikbke, St14), cell proliferation (Cdc25a, Ier3, Ppap2a), cancer invasion (Cd53, Ctsd, Cxcl10, Cxcr3, Loxl1) 42 - 46 and histone deacetylation (Hdac1, Hdac2 and Rbbp7) 16 were expressed in increased levels in DA and in reduced levels in congenics. On the contrary of the cancer-favoring genes up-regulated in DA, genes up-regulated in DA.ACI(Cia10) were protective against a cancer phenotypes. For instance, Nov, Rgs4 and Tgfbr3, which are inhibitors of cancer cell migration and invasion 47 - 49 , were expressed in increased levels in DA.ACI(Cia10). Two MMP inhibitors (Timp2 and Reck) were up-regulated in DA.ACI(Cia10) and down-regulated in DA. Reck negatively regulates cancer cell invasion, metastasis and angiogenesis in cancer, and its expression is reduced in cancer cells 50 , 51 . Like in cancers and in DA synovium, Reck is expressed in reduced levels in RA synovial tissues, compared with OA controls 52 . Timp2 is a negative regulator of MMPs, angiogenesis, cancer cell growth, invasion and metastasis 53 , 54 . Taken together, our results show that the presence of DA alleles at Cia10 increases the expression of genes favoring cancer development, proliferation and invasion, while the opposite is seen in the presence of ACI alleles. These observations suggest that these Cia10-regulated cancer genes could be involved in the regulation of synovial hyperplasia and pannus invasion and destruction of cartilage and bone. Survival and apoptosis abnormalities have been described in cancer, in arthritic synovial tissues and in autoimmunity. We considered that the synovial hyperplasia in DA, and the lack of in DA.ACI(Cia10) could be attributed to differences in expression of apoptosis genes. While many apoptosis and cell survival genes were differentially expressed, there was not a clear bias towards having increased numbers of pro-apoptosis versus anti-apoptosis genes in either strain. Thirteen of the sixteen differentially expressed genes located within the DA.ACI(Cia10) interval map to the same cytogenetic band on chromosome 2 (2q34) raising the possibility that a polymorphism within the region could affect a regulatory element acting in cis to influence the transcription of these closely located genes. While cis -acting transcription factor binding site polymorphisms have been a rare explanation for autoimmunity or other complex traits, it is certainly a possibility worth considering. We also considered that a polymorphism in a microRNA located within the Cia10 interval might explain part of the differences in synovial gene expression. The Cia10 interval contains eight predicted microRNAs but sequencing these microRNAs revealed no polymorphisms. Additionally, analyses of the expression of predicted microRNA targets revealed no significant differences between the strains, making the microRNA hypothesis unlikely to explain the Cia10 effect. In conclusion, we have determined that Cia10 regulates the expression of a unique set of inflammatory mediators, including markers of Th17 cells, IL-6, IL-17A and IL-11, oxidative stress regulators and a cancer-associated gene expression signature. These observations suggest a mechanism of action for the Cia10 gene, and several new potential prognostic biomarkers and targets for therapies aimed at reducing disease severity and joint damage in RA.

FLS from DA (n=4) and DA.ACI(Cia10) (n=4) were transfected with a NFκB luciferase reporter construct and stimulated with IL-1β (10ng/ml) or control vehicle for 48h. Values represent arbitrary units of luciferase luminescence.