Metabolomic analysis of RA SF
Samples of RA (n = 10) and OA (n = 10) SF were analyzed using an LC-MS approach. A total of 481 variables were identified after searching the Human Metabolome Database (HMDB). Orthogonal Partial Least Squares-Discriminant Analysis (OPLS-DA), the most frequently used multivariate statistical method, was applied for metabolomic analysis. The R2Y of the OPLS-DA model was 1, and the Q2 was 0.96, indicating that the model was stable and reliable. The OA and RA OPLS-DA score plots were separated, indicating that the model could discriminate metabolites between the OA and RA groups (Fig. 1A). Volcano plots were constructed by analyzing the fold change (FC) and p values from the t-test results, and the variable importance in projection (VIP) scores from OPLS-DA. Metabolites meeting the criteria of VIP > 1, FC > 2 or < 0.5 and p < 0.05 were defined as differentially expressed metabolites DEMs. The contents of the following metabolites were significantly elevated in RA SF samples compared with OA samples: 2-aminobenzoic acid, 2-hydroxy-3-methylbutyric acid, 9,12,13-TriHOME, creatinine, cyclic GMP (guanosinc monophosphate), epinephrine sulfate, guanidinosuccinic acid, guanosine, isobutyryl-L-carnitine, methionine sulfoxide, N-acetyl-L-tyrosine, norepinephrine sulfate, phenol, phenylpropanolamine, S-adenosylhomocysteine, tauroursodeoxycholic acid and UDP. In addition, the content of pyrophosphate was significantly decreased in RA samples (Fig. 1B). Using the metabolite set, we identified 8 variable network modules (green (n = 51), black (n = 31), brown (n = 56), blue (n = 66), turquoise (n = 144), red, yellow (n = 54), and gray (n = 27) via WGCNA. Each leaf in the tree represents one metabolite (Fig. 1C). Correlations between these metabolite network modules and clinical prognostic data (sex, age, rheumatoid factor (RF) level, anticyclic citrullinated peptide (anti-CCP) level, and Kellgren-Lawrence (K&L) scale score were analyzed. The turquoise module was found to be markedly positively correlated with RF (r = 0.71, p = 0.0004) and anti-CCP (r = 0.67, p = 0.001); the red module was found to be markedly positively correlated with RF (r = 0.88, p < 0.0001) and anti-CCP (r = 0.84, p < 0.0001) levels (Fig. 1D). Cytoscape 3.6.1 was used to visualize the metabolite network in the red module. A total of 49 variables were identified in the red module: UDP, urocanic acid, thiosulfate, sulfide, ribitol, pyridoxal, pipecolic acid, phytanic acid, p-hydroxyphenylacetic acid, PC(18:3(6Z,9Z,12Z)/24:1(15Z)), paraxanthine, orotidine, ornithine, nicotinic acid, nandrolone, monoisobutyl phthalic acid, methylamine, lysoPC(14:1(9Z)), L-serine, L-proline, L-octanoylcarnitine, L-arginine, isovalerylglycine, imidazoleacetic acid, guanidine, erythritol, erucic acid, D-ribose, dimethylsulfide, dimethyl-L-arginine, dimethylglycine, D-glucuronic acid, cholesterol sulfate, chenodeoxycholic acid glycine conjugate, CE(18:2(9Z,12Z)), behenic acid, ascorbic acid, 9-OxoODE, 9,12,13-TriHOME, 7-dehydrocholesterol, 5-methoxydimethyltryptamine, 4-hydroxyphenylpyruvic acid, 4-aminophenol, 4,7,10,13,16-docosapentaenoic acid, 1-phenylethylamine, 1-methylhistidine, 14,15-DiHETE, and 1,3-diaminopropane. The results are shown in a Venn diagram (Fig. 1E). Two metabolites (UDP and 9,12,13-TriHOME) were found in both the red module identified by WGCNA and among the DEMs identified by volcano plot analysis (Fig. 1F), indicating that they might play an important role in promoting the pathological progression of RA. By analyzing the VIP scores, odds ratios (ORs) in logistic regression and FC values, we focused on the pathogenic role of UDP in the subsequent study (Fig. 1G). The above data are also provided in Supplementary Table 2.
UDP level in RA peripheral blood and SF
The UDP concentration was measured in the plasma of RA patients (n = 36) and healthy volunteers (n = 36) using the Transcreener UDP Assay. Compared with those in the blood of healthy volunteers (27.1 ± 2.1 ng/mL), the UDP levels in the blood of patients with RA were significantly increased (46.8 ± 7.8 ng/mL, p = 0.017) (Fig. 2A). Pearson correlation analysis was performed, and no correlation was detected (R = -0.18) between the blood UDP level and serum RF level (Fig. 2B) or between the UDP level and serum anti-CCP level (R =-0.04) (Fig. 2C). Receiver operating characteristic (ROC) analysis was used to examine possible associations between blood UDP levels and RF and anti-CCP levels in RA patients. In the ROC analysis, an area under the ROC curve (AUC) of > 0.9 was considered excellent; 0.8–0.9, very good; 0.7–0.8, good; 0.6–0.7, average; and < 0.6, poor. The AUC value of UDP was 0.97, which was thus considered to indicate an excellent diagnostic test for RA (Fig. 2D). In addition, the UDP concentration was measured in RA SF (n = 36) and OA SF (n = 36). Compared with that in OA SF (0.5 ± 0.5 ng/mL), the UDP level in RA SF was significantly increased (23.1 ± 32.9 ng/mL, p < 0.0001) (Fig. 2E). Pearson correlation analysis detected a moderate positive correlation (R = 0.75) between the SF UDP level and serum RF level (Fig. 2F) and between the SF UDP level and serum anti-CCP level (R = 0.76) (Fig. 2G).
The effect of UDP on RA FLSs
Cultured RA FLSs (n = 5) and OA FLSs (n = 5) were treated with different concentrations of UDP (0 µM, 10 µM, 50 µM and 100 µM) for different times (0 h, 6 h, 12 h and 24 h). The CCK-8 assay showed increased proliferation of RA FLSs in the presence of 10 µM, 50 µM and 100 µM UDP (p = 0.002, 0.0003, and 0.0002, respectively) compared with that in the PBS-treated control group (Fig. 3A). However, the CCK-8 assay did not show a significant change in the proliferation of OA FLSs in the presence of 10 µM, 50 µM and 100 µM UDP (p = 0.96, 0.90, and 0.81, respectively) compared with that in the PBS-treated control group (Fig. 3B). The Annexin V/PI apoptosis analysis showed that the apoptosis rate of RA FLSs was decreased in the presence of UDP (100 µM) compared with that in the PBS-treated control group (p = 0.0316) (Fig. 3C). However, the Annexin V/PI apoptosis analysis detected little change in OA FLS apoptosis in the presence of UDP (p = 0.3317) compared with that in the PBS-treated control culture (Fig. 3D).
We used wound healing and Transwell assays to evaluate the effect of UDP on the migration of RA FLSs. The wound healing assays showed that RA FLS migration was increased in the presence of UDP (100 µM) (p = 0.0022) compared with that of control FLSs treated with PBS (Fig. 3E). The Transwell assays also showed that RA FLS migration was increased in the presence of UDP (100 µM) (p = 0.004) (Fig. 3F) compared with that of PBS-treated control FLSs. However, the wound healing assay showed only a slight change in OA FLS migration in the presence of UDP (100 µM) (p = 0.59) compared with that of control FLSs treated with PBS (Fig. 3G). The Transwell assay also showed only a slight change in OA FLS migration in the presence of UDP (100 µM) (p = 0.32) compared with that of FLSs treated with PBS (Fig. 3H).
We used flow cytometry to examine proinflammatory cytokine production in the FLS cultures. The assay showed significantly elevated IL-6 levels in the culture medium of RA FLSs in the presence of 100 µM UDP (2203 ± 712.2, p = 0.044), but the concentrations of IL-2, IL-4, IL-10, TNF-α and IFN-γ were not significantly changed (14.04 ± 6.09, p = 0.99; 20.17 ± 14.19, p = 0.59; 29.8 ± 25.13, p = 0.61; 99.47 ± 17.07, p = 0.75; and 18.43 ± 17.61 pg/mL, p = 0., respectively), compared with those in the PBS-treated control culture (861 ± 294.5, 13.95 ± 7.81, 15.97 ± 9.13, 25.53 ± 21.84, 104.2 ± 32.92, and 18.44 ± 11.84 pg/mL, respectively) (Fig. 4A). Compared with those in the control culture (14.3 ± 7.77, 41.9 ± 35.6, 1039 ± 230.2, 19.9 ± 15.6, 86.27 ± 46.17, and 14.87 ± 6.20 pg/mL), the respective concentrations of IL-2, IL-4, IL-6, IL-10, TNF-α and IFN-γ in the culture medium of OA FLSs were not significantly changed in the presence of 100 µM UDP (14.44 ± 12.44, p = 0.98; 43.53 ± 34.55, p = 0.74; 1239 ± 559.4, p = 0.40; 17.73 ± 13.19, p = 0.27; 87.47 ± 37.36, p = 0.85; and 15.33 ± 5.65 pg/mL, p = 0.83) (Fig. 4B).
An IL-6 ELISA was performed to verify the above results. This assay also showed significantly increased IL-6 levels in the culture medium of RA FLSs in the presence of 10 µM, 50 µM and 100 µM UDP (p = 0.022, 0.0205, and 0.0003, respectively) compared with that in samples from cells treated with PBS (Fig. 4C). However, ELISA showed only slight changes in IL-6 levels in the culture medium of OA FLSs in the presence of 10 µM, 50 µM and 100 µM UDP (p = 0.559, 0.088, and 0.010, respectively) (Fig. 4D). The above results indicated that UDP activated RA FLSs and induced IL-6 secretion but did not similarly affect OA FLSs.
Regulation of P2Y6 expression on the effects of UDP
The abundance of P2Y6 mRNA in RA and OA synovial tissue was analyzed in datasets from the GEO database (dataset type: expression profiling by array). The expression data (expression values) of 10 patients with RA and 6 OA controls from dataset GDS5402/208373_s_at, 5 patients with RA and 5 OA controls from dataset GDS2126/38222_at, 10 patients with RA and 10 OA controls from dataset GDS5401/208373_s_at, and 13 patients with RA and 10 OA controls from dataset GDS5403/208373_s_at were analyzed in SPSS software v.21.0. Significantly increased transcription of P2Y6 was found in RA synovial tissues (259.6 ± 24.53, n = 38) compared with OA samples (171.3 ± 18.85, n = 31) (Student’s unpaired t-test, p = 0.0076) (Fig. 5A).
In the present study, the expression level of P2Y6 in synovial tissues was also examined by real-time PCR, Western blotting, immunofluorescence and immunohistochemistry. P2Y6 mRNA expression was significantly increased in RA synovial tissues (n = 9) compared with OA synovial tissues (n = 9) (p = 0.0447) (Fig. 5B). In addition, western blot analysis showed that P2Y6 protein expression was significantly increased in RA synovial tissues (n = 5) compared with OA synovial tissues (n = 5) (p = 0.0005) (Fig. 5C). Immunofluorescence analysis of synovial tissues showed that P2Y6 protein expression was significantly increased in RA synovial tissues (n = 5) compared with OA synovial tissues (n = 5) (p = 0.0002) (Fig. 5D). Immunohistochemical analysis showed that P2Y6 protein expression was significantly increased in RA synovial tissues (n = 5) compared with OA synovial tissues (n = 5) (p = 0.0005) (Fig. 5E).
To verify the effect of UDP on RA FLSs, RA FLSs (n = 5) were cultured with both UDP (100 µM) and MRS2578 (10 µM), the chemical inhibitor of P2Y6. Cell proliferation was investigated using RTCA, apoptosis was detected with an annexin V/PI apoptosis assay, and IL-6 secretion was measured using a flow cytometric bead assay. Compared with that of RA FLSs cultured with PBS, RA FLS proliferation was increased in the presence of UDP (p = 0.0007) and was decreased in the presence of MRS2578 alone (p = 0.0017) and of UDP and MRS2578 together (p = 0.0024). The above results demonstrated that RA FLS proliferation was decreased in the presence of MRS2578 alone (p < 0.0001) and of UDP and MRS2578 together (p < 0.0001) (Fig. 6A). Compared with that in the PBS-treated culture, RA FLS apoptosis was decreased in the presence of UDP alone (p = 0.0440), increased in the presence of MRS2578 alone (p = 0.0285), and unchanged in the presence of UDP and MRS2578 together (p = 0.0917). The above results demonstrated that RA FLS apoptosis was increased in the presence of MRS2578 alone (p = 0.0339) and of UDP and MRS2578 together (p = 0.0476) (Fig. 6B). In addition, compared with that in the PBS-treated culture, IL-6 secretion in RA FLSs was increased in the presence of UDP (p = 0.0051) and was decreased in the presence of MRS2578 alone (p = 0.0021) and of UDP and MRS2578 together (p = 0.0023) (Fig. 6C). The wound healing assay showed that compared with that in the PBS-treated control culture, RA FLS migration was increased in the presence of UDP (p = 0.0016) and was decreased in the presence of MRS2578 alone (p = 0.0094) and of UDP and MRS2578 together (p = 0.0007). The above results demonstrated that RA FLS migration was decreased in the presence of MRS2578 alone (p < 0.0001) and of UDP and MRS2578 together (p = 0.0031) (Fig. 6D). Moreover, the Transwell assay showed that compared with that in the PBS-treated control culture, RA FLS migration was increased in the presence of UDP (p = 0.0003) and was decreased in the presence of MRS2578 (p = 0.0088) and of UDP and MRS2578 together (p = 0.0328). The above results demonstrated that RA FLS migration was decreased in the presence of MRS2578 alone (p = 0.0014) and of UDP and MRS2578 together (p = 0.0063) (Fig. 6E).
The effect of UDP on CIA
Rats induced with collagen II were simultaneously injected with UDP, MRS2578 or both UDP and MRS2578. The disease score as assessed by toe swelling was increased, and radiological signs (soft tissue swelling, new bone formation and marginal osseointegration) and histochemical staining were significantly enhanced in CIA rats compared with normal control rats (p < 0.0001), indicating successful establishment of CIA in the rats. The disease score was significantly increased in CIA rats treated with UDP (p = 0.0052) but decreased in CIA rats treated with MRS2578 (p = 0.0193) or both UDP and MRS2578 (p = 0.0331) compared with CIA rats. In addition, compared with that in CIA rats treated with UDP, the disease score was decreased in CIA rats treated with MRS2578 (p = 0.0002) or with both UDP and MRS2578 (p = 0.0002). The disease score did not differ significantly between CIA rats treated with MRS2578 alone and CIA rats treated with both UDP and MRS2578 (p = 0.7332) (Fig. 7A, B). Compared with that in normal control rats, paw inflammation in CIA rats was significantly exacerbated, indicating successful establishment of CIA in the rats. Paw inflammation was significantly exacerbated in CIA rats treated with UDP (p = 0.0058) but alleviated in CIA rats treated with MRS2578 (p = 0.024) or with both UDP and MRS2578 (p = 0.004) compared with CIA rats. Compared with that in CIA rats treated with UDP, paw inflammation was alleviated in CIA rats treated with MRS2578 (p < 0.0001) or with both UDP and MRS2578 (p < 0.0001). Paw inflammation did not differ significantly between CIA rats treated with MRS2578 alone and CIA rats treated with both UDP and MRS2578 (p = 0.51) (Fig. 7C).
SF was also collected from rats on day 20 after the first UDP injection, and the cytokine concentrations were measured by flow cytometry. Compared with the corresponding concentrations in the normal control group, the concentrations of IL-6, GM-CSF and TNF-α in the CIA control group were significantly increased (p = 0.0299, 0.0250, and 0.0376, respectively), the concentrations of IL-10 were significantly decreased (p = 0.0317), and the concentrations of IL-2, IL-4, IL-5, IL-13, and IFN-γ were not significantly changed. Compared with that in CIA rats, the IL-6 level was significantly increased in CIA rats treated with UDP (p = 0.0053) but decreased in CIA rats treated with MRS2578 (p = 0.0040) or with both UDP and MRS2578 (p = 0.0114). Compared with that in CIA rats treated with UDP, the IL-6 level was decreased in CIA rats treated with MRS2578 (p = 0.0002) or with both UDP and MRS2578 (p = 0.0004). However, the IL-6 level did not differ significantly between CIA rats treated with MRS2578 alone and CIA rats treated with both UDP and MRS2578 (p = 0.2901). Moreover, the concentrations of IL-2, IL-4, IL-5, IL-10, IL-13, GM-CSF, IFN-γ and TNF-α did not differ significantly (Fig. 7D). Cytokine expression in the SF of rats was also evaluated by IL-6 ELISA. The IL-6 level was significantly increased in CIA rats treated with UDP (p = 0.0486) but decreased in CIA rats treated with MRS2578 (p = 0.0467) or with both UDP and MRS2578 (p = 0.293) compared with CIA rats. Compared with that in CIA rats treated with UDP, the IL-6 level was decreased in CIA rats treated with MRS2578 (p = 0.0005) or with both UDP and MRS2578 (p = 0.0002). However, the IL-6 level did not differ significantly between CIA rats treated with MRS2578 alone and CIA rats treated with both UDP and MRS2578 (p = 0.908) (Fig. 7E).
We examined UDP levels in the peripheral blood and SF of CIA rats using the fluorescence polarization method. The UDP level in peripheral blood was significantly higher in CIA rats than in normal control rats (p = 0.0163). Compared with that in CIA rats, the peripheral blood UDP level was significantly increased in CIA rats treated with UDP (p = 0.0482) or both UDP and MRS2578 (p = 0.0362) but was not significantly different in CIA rats treated with MRS2578 (p = 0.3711). Compared with that in CIA rats treated with UDP, the peripheral blood UDP level was decreased in CIA rats treated with MRS2578 (p = 0.0312) but was not significantly different in CIA rats treated with both UDP and MRS2578 (p = 0.9421). The UDP level in peripheral blood was significantly higher in CIA rats treated with both UDP and MRS2578 than in CIA rats treated with MRS2578 (p = 0.0231) (Fig. 8A). The UDP level in SF was significantly higher in CIA rats than in normal control rats (p = 0.0090). Compared with that in CIA rats, the UDP level in SF was significantly increased in CIA rats treated with UDP (p = 0.0066) or both UDP and MRS2578 (p = 0.0074) but was not significantly different in CIA rats treated with MRS2578 (p = 0.4286). Compared with that in CIA rats treated with UDP, the UDP level in SF was decreased in CIA rats treated with MRS2578 (p = 0.0022) but was not significantly different in CIA rats treated with both UDP and MRS2578 (p = 0.5821). Finally, the UDP level in SF was significantly higher in CIA rats treated with both UDP and MRS2578 than in CIA rats treated with MRS2578 alone (p = 0.0019) (Fig. 8B).
We also examined P2Y6 expression in synovial tissues of the model rats by real-time PCR and Western blot analysis. P2Y6 mRNA expression in synovial samples was significantly higher in CIA rats than in normal control rats (p = 0.0017). P2Y6 mRNA expression in synovial tissues did not differ significantly between CIA rats and CIA rats treated with UDP alone (p = 0.196), MRS2578 alone (p = 0.211) or UDP and MRS2578 together (p = 0.721). P2Y6 mRNA expression in synovial tissues did not differ significantly between CIA rats treated with UDP and CIA rats treated with MRS2578 alone (p = 0.0581) or with UDP and MRS2578 together (p = 0.1091). P2Y6 mRNA expression in synovial tissues did not differ significantly between CIA rats treated with MRS2578 alone and CIA rats treated with UDP and MRS2578 together (p = 0.348) (Fig. 8C). However, P2Y6 protein expression in synovial tissues was significantly higher in CIA rats than in normal control rats (p = 0.0049). However, P2Y6 protein expression in synovial tissues did not differ significantly between CIA rats and CIA rats treated with UDP alone (p = 0.476), MRS2578 alone (p = 0.308) or UDP and MRS2578 together (p = 0.545). Moreover, P2Y6 protein expression in synovial tissues did not differ significantly between CIA rats treated with UDP and CIA rats treated with MRS2578 alone (p = 0.128) or with UDP and MRS2578 together (p = 0.267). P2Y6 protein expression in synovial tissues also did not differ significantly between CIA rats treated with MRS2578 alone and CIA rats treated with UDP and MRS2578 together (p = 0.765) (Fig. 8D).