ORM2 expression is upregulated in RA patients by proinflammatory stimuli
A global analysis to determine which acute phase reactants are involved in rheumatoid inflammation has not yet been performed. To address this knowledge gap, we conducted a comparative analysis of DEPs in RA sera (106 genes), DEPs in RA urine (133 genes), and 46 genes defined as acute phase reactants by Gene Ontology (GO) (Fig. 1a). We identified 9 acute phase reactants upregulated in the sera and/or urine samples of RA patients. Among these genes, ORM2 exhibited the highest fold change. Thus, it was selected as a potential regulator involved in RA pathology (Fig. 1 and Supplementary Fig. 1). Indeed, the ORM2 expression levels were much greater in the sera and synovial fluids of the RA patients than in those of the OA patients (mean ± SEM: 3.5 ± 0.47 μg/mL vs. 0.26 ± 0.04 μg/mL in the sera and 8.7 ± 1.8 μg/mL vs. 0.12 ± 0.04 μg/mL in the synovial fluids) (Fig. 1b). In addition, immunohistochemical staining revealed that ORM2 was more highly expressed in RA synovia than in OA synovia, particularly in the lining layer and sublining leukocytes (Fig. 1c).
The lining of the synovium in RA patients is composed of macrophage-like synoviocytes and FLSs27. Immunofluorescence staining revealed that ORM2-expressing cells colocalized well with CD55+ and CD68+ cells, which are markers of FLSs and macrophages, respectively, indicating that these cells were the major cells expressing ORM2 (Fig. 1d and Supplementary Fig. 2a, b). The use of only the secondary Abs without either the anti-ORM2 Ab or the anti-CD68 Ab failed to stain the ORM2- or CD68-expressing cells, respectively (Supplementary Fig. 2c). Moreover, preabsorption of the anti-ORM2 Ab with recombinant ORM2 markedly inhibited ORM2 staining in CD68+ cells (Supplementary Fig. 2d), indicating the specificity of the anti-ORM2 Ab. To further confirm the presence of ORM2 in the RA synovium, we isolated FLSs and macrophages from synovial membranes and synovial fluids, respectively, and measured ORM2 expression levels. As a result, freshly isolated CD14+ cells from RA patients had significantly higher levels of ORM2 mRNA than FLSs from RA patients (RA-FLSs) and FLSs from OA patients (OA-FLSs) (Fig. 1e). The mean ORM2 mRNA expression levels in RA CD14+ cells were 12.5-fold and 27.8-fold higher than those in RA-FLSs and OA-FLSs, respectively (Fig. 1e). The expression of ORM2 mRNA was 2.2-fold higher in RA-FLSs than in OA-FLSs, but this difference was not statistically significant (p = 0.11) (Fig. 1e).
The RA synovium is heavily exposed to a variety of proinflammatory stimuli, including Toll-like receptor ligands and inflammatory cytokines. As shown in Fig. 1f, stimulation of synovial fluid mononuclear cells (SFMCs) from RA patients (RA-SFMCs) with LPS, IL-1β, or TNF-α resulted in a substantial increase in ORM2 mRNA expression levels—by 19.4-, 3.1-, and 5.2-fold, respectively—compared to treatment with medium alone. This increase in ORM2 mRNA expression levels was not observed in macrophages treated with IL-6 (Supplementary Fig. 3a). ORM2 transcript levels in RA-FLSs were also significantly increased by LPS, IL-1β, TNF-α, and TGF-β stimuli—up to 4.8-, 2.1-, 4.2-, and 2.9-fold, respectively—but not by other cytokines, such as IL-6, M-CSF, or IL-10 (Fig. 1f and Supplementary Fig. 3b). ORM2 protein production was also markedly increased in the synovial fluid CD14+ cells of RA patients and in RA-FLSs after stimulation with proinflammatory stimuli, including LPS, IL-1β, and TNF-α, as well as the profibrotic cytokine TGF-β (Fig. 1g, h).
In summary, ORM2 expression was upregulated in the extrahepatic sites, including synovial fluids and synovial membranes, of RA patients, and it was upregulated by proinflammatory stimuli. The major cell types producing ORM2 were synovial macrophages and fibroblasts.
ORM2 directly increases the production of IL-6, CXCL8, and CCL2 by RA-FLSs and macrophages
Next, we investigated whether ORM2, like several other acute phase reactants5,6,7,8,9, could functionally regulate inflammatory responses and contribute to RA pathogenesis. To this end, we treated RA-FLSs and macrophages with recombinant ORM2 and tested whether ORM2 could induce the production of proinflammatory cytokines. The results showed that recombinant ORM2 dramatically increased the production of IL-6, CXCL8, and CCL2 by RA-FLSs in a dose- and time-dependent manner (Fig. 2a). The effect of recombinant ORM2 persisted even when polymyxin B, a compound that blocks LPS-induced TLR4 activation, was present. In contrast, the LPS-induced increase in IL6 and CXCL8 expression levels was completely inhibited by polymyxin B (Supplementary Fig. 4a). After stimulation with 1 μg/mL ORM2 for 72 h, the levels of IL-6, CXCL8, and CCL2 produced by RA-FLSs increased 3.6-, 10.3-, and 3.4-fold, respectively, relative to those observed after treatment with medium alone (Fig. 2a). These increases did not seem to be due to cell proliferation since the number and viability of FLSs were not altered according to the results of trypan blue exclusion and the MTT assay, respectively, at 72 h after stimulation with 0.1 to 1 μg/mL ORM2 (Supplementary Fig. 4b). After treatment with exogenous ORM2, the IL6, CXCL8, and CCL2 mRNA expression levels also markedly increased by 6.4-, 10.7-, and 2.4-fold, respectively, compared to those in the medium-treated control (Fig. 2b), indicating that these increases were transcriptionally regulated. Similarly, ORM2-stimulated RA-SFMCs dose-dependently increased the production of IL-6 and TNF-α (Fig. 2c). In parallel, recombinant ORM2 time-dependently increased IL-6 and TNF-α secretion from the macrophages of healthy controls, which were differentiated from peripheral monocytes (Fig. 2d). The secretion of CXCL8 and CCL2 from healthy macrophages was also robustly promoted by exogenous ORM2 (Supplementary Fig. 4c). Furthermore, in RA synovial fluid, a modest correlation was found between the ORM2 concentration and the CXCL8 and CCL2 levels (Fig. 2e, f). Taken together, these results suggest that upregulated ORM2 in RA joints can directly stimulate RA-FLSs and macrophages, known as effector cells in RA, to induce the production of proinflammatory cytokines and chemokines, thereby further amplifying inflammatory responses.
NF-κB and p38 are major signals for ORM2-induced proinflammatory responses
We further investigated which signaling pathways are involved in the regulation of cytokine and chemokine production by ORM2. A number of studies have demonstrated that NF-κB and p38 MAP kinase are major signaling molecules responsible for IL-6, CXCL8, and CCL2 production in RA-FLSs28. To determine how ORM2 induces IL-6 and CXCL8 production, we treated RA-FLSs with recombinant human ORM2 in the presence of chemical inhibitors specific for NF-κB and p38 MAP kinase. As expected, NF-κB inhibitors such as pyrollidine dithiocarbamate (PDTC) and BAY117082 substantially suppressed ORM2-induced increases in IL6 and CXCL8 mRNA expression levels (Fig. 3a); the suppressive effect of the BAY inhibitor on IL6 upregulation was less pronounced than that on CXCL8 upregulation, suggesting that the NF-κB pathway partially contributes to the production of CXCL8 induced by ORM2. Additionally, the p38 MAP kinase inhibitor SB203580 strongly inhibited the upregulation of the IL6 and CXCL8 mRNAs in RA-FLSs stimulated with ORM2 (Fig. 3a). Similarly, in macrophages from healthy controls, the increase in IL6 and CCL2 expression levels induced by ORM2 was almost completely suppressed by SB203580, PDTC and BAY117082 (Supplementary Fig. 5). However, in contrast to NF-κB inhibitors, SB203580 failed to mitigate the increase in CXCL8 expression levels in macrophages. Taken together, these results suggest that ORM2 can increase the production of proinflammatory cytokines via the NF-κB and/or p38 MAP kinase pathway in RA-FLSs and macrophages.
In support of this, recombinant human ORM2 (1 μg/mL) time-dependently increased NF-κB translocation from the cytoplasm to the nucleus in RA-FLSs and upregulated the phosphorylation of NF-κB p65 (Fig. 3b, c), while it downregulated IκB expression up to 2 h after stimulation (Fig. 3c). Moreover, NF-κB p65 siRNA, but not control siRNA, markedly repressed the upregulation of IL6 and CXCL8 mRNA expression induced by ORM2 (Fig. 3d, e), indicating that NF-κB was a major signaling factor involved in this process. Moreover, as early as 10 min following stimulation with recombinant ORM2, phospho-p38 (p-p38) MAP kinase expression was sharply upregulated in RA-FLSs, as determined by Western blot analysis. This increase lasted for 1 h (Fig. 3f). Like NF-κB p65 siRNA, p38 siRNA also strongly inhibited the ORM2-induced increase in IL6 and CXCL8 mRNA levels (Fig. 3g, h), demonstrating that p38 MAP kinase is another major signaling molecule that mediates the promotive effect of ORM2 on IL-6 and CXCL8 expression levels in RA-FLSs. Notably, knockdown of either the NF-κB p65 or p38 MAP kinase transcript only partially reduced the ORM2-stimulated mRNA expression of the IL6 and CXCL8 (Fig. 3e, h). However, simultaneous knockdown of NF-κB p65 and p38 MAP kinase almost completely abolished the ORM2-induced increase in the mRNA expression levels of IL6 and CXCL8 (Fig. 3i, j), suggesting that both the NF-κB p65 and p38 MAP kinase signaling pathways are required for ORM2-induced IL-6 and CXCL8 production by RA-FLSs.
Glycophorin C is a receptor for ORM2 on synovial macrophages and FLSs
Given the ORM2-induced increase in proinflammatory factors in monocytes/macrophages and RA-FLSs (Figs. 2, 3), we next questioned whether a cell surface receptor(s) for ORM2 was present in these cells. If so, what receptor could induce cytokine and chemokine production upon ORM2 ligation? To the best of our knowledge, the specific cellular receptor of ORM2 has not been identified. To address this issue, we first utilized previously published protein-to-protein interactome databases (Supplementary Table 2). With these databases, 11 proteins were found to potentially interact with ORM2. Among these 11 proteins, GYPC has been reported to exhibit receptor activity in erythrocytes29. Therefore, we next explored whether RA-FLSs and macrophages expressed GYPC. As shown in Supplementary Fig. 6a, GYPC was expressed on peripheral monocytes freshly isolated from healthy donors, as determined by flow cytometry. Notably, the expression of this gene was significantly upregulated by stimulation with IL-1β, TNF-α, or LPS; in fact, a large portion of cultured monocytes (more than 80%) expressed GYPC on their surface (Supplementary Fig. 6a, b). In cultured macrophages differentiated from peripheral monocytes, TNF-α and LPS also significantly upregulated GYPC expression on the surface (Fig. 4a). However, IL-6 stimulation failed to upregulate GYPC expression in macrophages (Fig. 4a).
GYPC was also expressed in RA-FLSs, but its level was much lower than that in cultured monocytes (Fig. 4b). Like in peripheral monocytes, IL-1β and TNF-α, but not LPS or IL-6, upregulated GYPC expression in RA-FLSs (Fig. 4b and Supplementary Fig. 6c). RA-FLSs exhibit heterogeneity and consist of more than three different subtypes, including CD90+ and CD55+ FLSs30. In particular, CD55+ FLSs, located in the synovial lining, are involved in bone and cartilage damage and have a limited impact on inflammation, while CD90+ fibroblasts in the sublining layer induce more severe inflammation with minimal effects on bone and cartilage30. Immunofluorescence staining of RA synovial tissues revealed that GYPC-expressing cells colocalized well with CD68+ and CD90+ cells, indicating that synovial macrophages and sublining CD90+ FLSs are the major cell types that express GYPC (Fig. 4c and Supplementary Fig. 7). In contrast, only a modest percentage of the CD55+ cells in the lining layer expressed GYPC (Supplementary Fig. 7), suggesting that these cells are not the major FLS subtype that responds to GYPC stimulation. Furthermore, the colocalization of GYPC with macrophage or FLS markers varied (Supplementary Fig. 7), which might be due to the difference in GYPC expression levels depending on proinflammatory stimuli (Fig. 4a, b).
To test whether GYPC could actually mediate ORM2-induced IL-6 and CXCL8 expression, we further carried out knockdown experiments using GYPC siRNA. As a result, GYPC siRNA, but not the control siRNA, almost completely blocked the ORM2-induced increase in IL6 and CXCL8 mRNA expression in RA-FLSs (Fig. 4d, e). Secretion of IL-6 and CXCL8 from RA-FLSs in the presence of recombinant human ORM2 was also markedly hampered by GYPC siRNA (Fig. 4f). In contrast, TNF-α-induced IL6, and CXCL8 expression was rarely affected by GYPC siRNA, which excluded the possibility of nonspecific cellular toxicity caused by GYPC siRNA (Fig. 4g). Collectively, these results strongly suggest that GYPC can function as an ORM2 receptor in RA-FLSs.
To test this hypothesis, we first conducted a proximity ligation assay (PLA) using anti-ORM2 and anti-GYPC antibodies. PLA is a powerful experimental tool that facilitates the detection of protein interactions in situ with high specificity and sensitivity31. Robust red fluorescence was observed only when these two antibodies were simultaneously used to treat RA-FLSs in the presence of recombinant ORM2. Indeed, almost all of the observed RA-FLSs exhibited strong fluorescent signals (see Materials and Methods for details) (Fig. 5a). Such robust red fluorescence was not observed after a single treatment with either the anti-ORM2 or the anti-GYPC Ab. Moreover, in the absence of recombinant ORM2, although fluorescent signals were noted, their intensity was relatively modest (Supplementary Fig. 8, bottom panel). These findings demonstrated that there may be direct molecular interactions between exogenous ORM2 and GYPC on RA-FLSs. To confirm this, solid-phase ELISA was performed using recombinant GYPC (rGYPC) and recombinant ORM2 (rORM2), in which rGYPC was used to precoat the ELISA plate and rORM2 or bovine serum albumin (BSA) was subsequently added to the plate to react with GYPC. As a result, we observed a dose-dependent increase in optical density with increasing concentrations of rORM2, which was not observed with BSA (Fig. 5b). The optical density was further enhanced by increasing the amount of rGYPC coating on the plate (Fig. 5b), indicating that there was a direct molecular interaction between rGYPC and rORM2. We finally investigated whether the increase in cytokine and chemokine production induced by ORM2 could be blocked by the soluble form of rGYPC. As shown in Fig. 5c, the ORM2-induced increase in IL-6 and CXCL8 secretion by RA-FLSs was substantially mitigated by rGYPC pretreatment. This reduction was dose dependent. Taken together, these results, along with the flow cytometric analysis of GYPC and the PLA data, suggest that the soluble form of rGYPC interferes with the interaction between exogenous ORM2 and the membrane form of GYPC, which leads to the inhibition of IL-6 and CXCL8 upregulation induced by ORM2.
In summary, GYPC, a membrane glycoprotein required for erythrocyte shaping and stability29, is expressed in synovial macrophages and RA-FLSs. It can interact with ORM2 as a cell surface receptor to mediate the production of proinflammatory cytokines and chemokines.
ORM2 induces proinflammatory cytokine production by mouse macrophages and FLSs
Based on human data, we examined whether the regulation of cytokine/chemokine expression by ORM2 could be reproduced in a mouse model. As shown in Fig. 6a, the mean ORM2 expression was much higher (4.2-fold) in mice with collagen-induced arthritis (CIA), a representative animal model of RA32, than in control (vehicle-treated) mice, as determined by immunohistochemistry and qPCR analysis of synovial tissues (Fig. 6a). Immunofluorescence staining of the affected joints revealed that, similar to what was observed in RA synovium, ORM2 was strongly colocalized with CD55+ and F4/80+ cells in the affected joints of mice with CIA (Fig. 6b and Supplementary Fig. 9), confirming that synovial fibroblasts and macrophages are the major cells that produce ORM2. To validate the regulatory effect of ORM2 on the proinflammatory response in the mouse system, mouse bone marrow-derived macrophages (BMDMs) and synovial fibroblasts isolated from the affected joints of mice with CIA were treated with recombinant mouse ORM2. We found that recombinant mouse ORM2 dramatically increased Tnf and Il6 mRNA expression levels in BMDMs in a time-dependent manner (Fig. 6c). After treatment with 1 μg/mL mouse ORM2 for 6 h, the fold change relative to that in the medium alone was 12.4 for Tnf and 147.9 for Il6 (Fig. 6c). Similarly, compared with FLSs treated with medium alone, ORM2-stimulated mouse FLSs exhibited marked increases in Il6 and Ccl2 expression levels up to 14.5- and 16.0-fold, respectively (Fig. 6c). However, ORM2 stimulation failed to upregulate Il8 expression in mouse FLSs (data not shown), unlike in RA-FLSs. Moreover, ORM2 stimulation substantially increased the secretion of IL-6 and TNF-α in mouse BMDMs and of IL-6 and CCL2 in RA-FLSs in a time-dependent manner (Fig. 6d).
Taken together, our results demonstrate that ORM2 is also highly expressed in the inflamed joints of mice, particularly in synovial fibroblasts and macrophages, and directly promotes the secretion of proinflammatory cytokines and chemokines by these cells.
ORM2 promotes IL-1β-induced arthritis in vivo and reflects the inflammatory activity of RA
Since ORM2 has strong proinflammatory activity in vitro, we explored whether ORM2 could aggravate the severity of chronic arthritis in vivo. To this end, we generated a severe form of ORM2-accelerated arthritis by intra-articular administration of ORM2 (4 μg) into the knee joints of mice with suboptimal IL-1β-induced arthritis (Supplementary Fig. 10a), a model of chronic arthritis in which macrophages play a central role33. After inducing IL-1β-induced arthritis in mice, the Orm2 mRNA expression level increased up to 65.3-fold and 3.2-fold in the liver and affected joints, respectively, as determined by qPCR (Fig. 7a). Moreover, compared with vehicle injection, intra-articular injection of ORM2 markedly exacerbated the suboptimal severity of IL-1β-induced arthritis, as assessed by inflammatory cell infiltration and synovial hyperplasia (Fig. 7b). Interestingly, the infiltrated cells were strongly stained with the anti-F4/80 Ab but only modestly stained with the anti-NIMP-R14 Ab (Fig. 7c and Supplementary Fig. 10b). These findings suggest that the major cell types recruited by ORM2 injection are monocytes/macrophages, in accordance with the in vitro findings that ORM2 increases the CCL2 production necessary for monocyte recruitment34, in contrast with the finding that ORM2 has no effect on CXCL8 production for neutrophil recruitment35.
Finally, we investigated whether the serum ORM2 concentration could represent inflammatory activity and disease severity in RA patients (n = 90) (Supplementary Table 1). The results showed that the serum ORM2 concentration was positively correlated with simultaneously measured parameters for RA activity, including the CRP level (rho = 0.47) and disease activity score 28 (DAS28, rho = 0.28), while it was negatively correlated with the serum ALB concentration (rho = −0.29) and Hb concentration (rho = −0.32) (Fig. 7d). Moreover, serum ORM2 concentrations at baseline were significantly higher in the subgroup of RA patients (n = 14) with radiographic progression than in those without such progression (n = 76), which was determined using the serial X-ray images from patients collected over 2 years (median serum ORM2 levels [IQR]: 10,310 [4,710-12,870] ng/mL for the progression group and 1,090 [340-1,545] ng/mL for the nonprogression group; P < 0.001) (Fig. 7e). Taken together, these results suggest that the serum ORM2 concentration could indicate disease activity and progression in RA patients, suggesting that ORM2 has potential use as a diagnostic marker for RA.