TET3 expression in synovial membranes and FLS of patients with RA
To understand the role of TET family proteins in RA progression, we first analyzed the expression profiles of the TET1/2/3 proteins, 5mC, and 5hmC in the synovial membranes of RA patients, and compared these with those of osteoarthritis (OA) patients. Immunohistochemical analysis confirmed the expression of TET2, TET3, and 5hmC in both the RA and OA synovial membranes. Among the TET proteins, TET3 exhibited the highest expression in the RA patients (Fig. 1A). In addition, quantitative analysis revealed that both TET3 and 5hmC expression were higher in the RA patients when compared with the OA patients, while the expression levels of TET2 were similar between the two patient groups (Fig. 1B).
Next, we attempted to identify the cell types expressing TET proteins within the synovial tissues. The cell type markers for FLS (CD55) and monocytes/macrophages (CD68) were found to co-localize with TET2 and TET3, while TET1 expression levels were low in both of these cell types (Fig. 1C). TET3 was highly expressed in the superficial layer of the synovial membrane and clearly co-stained with CD55, but not with CD68 (Fig. 1C). Conversely, co-staining with TET2 and CD68 were co-stained, but not CD55 and TET2 (Fig. 1C). As TET3 expression in the synovial membrane was visible in the FLS, we went on to evaluate TET3 expression in a primary culture of FLS from RA patients. Immunohistochemical analysis of the cultured FLS showed higher expression levels of TET3 than TET1 and TET2 (data not shown). RA FLS presented with increased TET3 expression when compared to the FLS from OA patients (Fig. 1F). Collectively, these findings suggest that RA activation in FLS is associated with the expression of TET3 and not the other TET proteins.
Pro-Inflammatory cytokines induce TET3 expression in FLS samples
As TET3 is highly expressed in FLS samples from RA patients, we reasoned that the associated pro-inflammatory cytokines are likely to function as TET3 inducers. The cultured RA FLS were treated with pro-inflammatory cytokines, and then TET1/2/3 expression was assessed at the mRNA level. Among the TET members, only TET3 exhibited significant induction in responses to TNFα, IL-1, and IL-17 (Fig. 2A). In contrast, no increase in TET1 or TET2 expression was recorded for any of the nine cytokines used in this assay.
TNFα induces TET3 expression and the hydroxylation of methylated DNA
We then tested whether pro-inflammatory cytokines potentiate TET3 function in the putative DNA demethylation process. Given the clinical impact of increased TNFα during RA progression, we selected TNFα as the most likely to induce a response (Fig. 2A). First, we investigated the effect of TNFα on TET3 regulation at the protein level. Western blot confirmed that TNFα increased TET3 expression at the protein level, which was further supported by the quantitative image analysis using ImageJ software (Fig. 2B). In addition, the obtained results suggest that TNFα induced this response at the transcript level as there was no change in TET3 mRNA turnover (Supplementary Figure S1). These results were consistent with the significant accumulation of nuclear TET3 protein in cultured RA FLS at 48 h post-TNFα stimulation (Fig. 2C). This increase was also accompanied by an increase in 5hmC (Fig. 2D), suggesting that increases in pro-inflammatory cytokines during RA progression stimulate TET3 expression, leading to the hydroxylation of methylated DNA in RA FLS.
Genes regulated by both TET3 and TNFα are relevant to RA
Persistent exposure to pro-inflammatory cytokines has been reported to transform FLS into cells that produce a variety of arthritogenic molecules [14]. We then went on to use a gene microarray analysis (21,448 genes) to profile the global gene regulation in cultured RA FLS (n = 3) treated with TNFα in the presence or absence of anti-TET3 small interfering RNA (siRNA) [i.e., TET3- knockdown (KD)] in an effort to determine whether TET3 mediates this TNFα induced transformation. The KD efficiency of TET3 by siRNA was confirmed using qPCR and Western blotting (Supplementary Figure S2A and B). The gene expression array analysis was conducted using four groups [control siRNA (siCTL)/TNFα(–), siCTL/TNFα(+), TET3-KD (siTET3)/TNFα(–), or siTET3/TNFα(+)], and the genes with significant expression differences (ANOVA F-test, P < 0.05) were used in the hierarchical analysis (Supplementary Figure S3). When we compared the siCTL/TNFα(+) and the siCTL/TNFα(–) groups, the number of genes regulated by TNFα stimulation at the cutoff point [|log fold change (FC)| >0.58, false discovery rate (FDR)-corrected F-test, P < 0.3)] is shown in Fig. 3A, with 280 upregulated genes and 185 downregulated genes, respectively. When we compared the siTET3/ TNFα(+) and the siCTL/TNFα(+) group, the number of genes affected by TET3-KD under TNFα treatment was 180, with 93 genes upregulated and 87 downregulated, respectively. There were an estimated 95 genes that are likely to be regulated by both TNFα and TET3, with 52 of these being upregulated and 43 downregulated, respectively (Fig. 3A, Supplementary Table S1 and S2). The 52 upregulated genes encode several factors associated with RA progression, including those associated with neutrophil migration [such as C-X-C motif chemokine ligand 8 (CXCL8) and CXCL5], cell migration [such as Myocardin (MYOCD), Calponin 1 (CNN1), and Integrin Subunit Beta 3 (ITGB3)], amplifying inflammation [such as Leukemia Inhibitory Factor (LIF) and Interleukin 1 Beta (IL1B)], proto-oncogenes [KIT Proto-Oncogene, Receptor Tyrosine Kinase (KIT), RELB Proto-Oncogene, nuclear factor-kappa B (NF-κB) Subunit (RELB)], Interferon (IFN)-inducible genes [Interferon Induced Protein 44 Like (IFI44L), 2′-5′-Oligoadenylate Synthetase 1 (OAS1), and Radical S-Adenosyl Methionine Domain Containing 2 (RSAD2)]. Those genes with variations known to increase RA risk are also listed in this figure [TNF alpha-induced protein 3 (TNFAIP3) and fatty acid desaturase 2 (FADS2)].
Next, we went on to complete functional ontology and KEGG pathway analyses on these genes (Fig. 3B, Supplementary Table S3 and S4), and confirmed the expected enrichment of the “TNF signaling” pathway [16.1-fold (P = 2.79E−06, FDR = 0.003)]. Significant enrichments were detected in the known signaling pathways associated with RA progression, including “NOD-like receptor signaling,” “NF-kappa B signaling,” “Cytokine-cytokine receptor interaction” and “Chemokine signaling” pathways (Fig. 3B). As the CXC-chemokines are well-recognized facilitators of RA progression, we went on to perform a cluster analysis and showed that several C-C motif chemokines are likely regulated by both TET3 and TNFα (Fig. 3C).
TET3 facilitates the mobility of activated FLS
To verify gene regulation of the candidate chemokines by these factors, real-time qPCR was performed using cultured RA FLS to assess the mRNA expression levels of CXCL8, CCL2, receptor activator of nuclear factor kappa-B ligand (RANKL), osteoprotegerin (OPG), matrix metalloproteinase 1 (MMP1), and MMP13 (Fig. 3D). Although the expression of RANKL, MMP1, and MMP13 (up-regulated) as well as OPG (down-regulated) were shown to be TNFα-dependent, they appeared to be TET3 independent (Fig. 3D). As TNFα is well known to induce the production of inflammatory mediators [5], the production of the other mediators during RA FLS culture were also evaluated for TET3 dependence (Fig. 3E). Although IL-1, IL-17A, and TNFα were below the detection level in this study, presumably due to the detection limits of the culture media, induction of CXCL8 and CCL2 by TNFα were both confirmed to be linked to TET3 expression (Fig. 3E). Increases in IL-6 and VEGF in response to TNFα were observed, but were shown to be TET3 independent (Fig. 3E). One of the most distinct features of RA progression is cell migration and invasion of activated FLS under persistent stimulation by pro-inflammatory cytokines [5]. Given this, we then investigated whether TET3 is indeed involved in increasing the cellular mobility of FLS following TNFα stimulation using cultured RA FLS. The scratch assay was used to assess cell migration and invasion [15], and TET3-KD abrogated the effect of TNFα stimulation on FLS cell mobility (Fig. 3F). Thus, TET3 seems to mediate the action of a subset of TNFα target genes responsible for pannus formation in progressed RA joints.
Haploinsufficiency of TET3 (TET3
+/−) attenuates RA progression in an RA mouse model induced by K/BxN serum
Given the findings in both the clinical samples and cultured FLS, we hypothesize that TET3 expression in FLS facilitates RA progression in the joints. To further address this point, we attempted to illustrate TET3 function in the intact joints of an RA mouse model. The development of a typical RA-like phenotype was achieved following murine treatment with K/BxN serum [16] (Fig.4), and a TET3 gene-depleted CL57BL/6 line was used as no overt abnormality with normal reproductive ability has been observed in mice with TET3 haploinsufficiency [17, 18]. TET2/3 and methylated DNAs were clearly stained in the synovial tissues of the joints in wild-type (WT) mice. K/BxN serum transfer was found to upregulate the expression levels of TET2, TET3, and 5hmC (Fig. 4A), consistent with the findings from the human RA clinical samples (Fig. 1). In TET3+/− mice, K/BxN serum transfer was unable to induce significant expression of TET3 and did not affect TET2 expression levels (Fig. 4A). Acute arthritis was induced following K/BxN serum transfer in WT and TET3+/− mice (Fig. 4B, C). However, the progression of arthritis was clearly aborted in TET3+/−-K/BxN mice (Fig. 4B). Histological analysis suggested that TET3 haploinsufficiency attenuates the hallmarks of arthritis progression, including reducing synovial inflammation and FLS proliferation following bone destruction (Fig. 4C). Marked bone erosion was obvious in the arthritic WT-K/BxN mice, but not in the TET3+/−-K/BxN mice when evaluated using micro-computed tomography (Fig. 4D). Furthermore, K/BxN serum transfer potently induced the spread of tartrate-resistant acid phosphatase (TRAP)-positive mature osteoclasts in the border area between the inflamed synovial membrane and bone, but this effect was much less obvious in the TET3+/−-K/BxN mice (Fig. 4E). In contrast, TET3 silencing by transfection of FLS with siRNAs significantly suppressed the cell proliferation of FLS than those transfected with control siRNA (Supplementary Fig S4). These findings support the in vivo significance of TET3 function in facilitating the progression of arthritis and pannus formation.