TNF-α activates WNT5A, which recruits osteoclast precursors by MCP-1 production, but not directly activates osteoclastogenesis, in psoriatic arthritis

Psoriatic arthritis (PsA) results from joint destruction by osteoclasts. Promising ecacy of TNF-α blockage indicates its important role in osteoclastogenesis of PsA. WNT ligands actively regulate osteoclastogenesis. We investigated how WNT ligands activate osteoclasts amid the TNF-α milieu in PsA. Methods: We rst proled the expression of WNT ligands in CD14+ monocyte-derived osteoclasts (MDOC) from 3 PsA patients and 3 healthy controls (HC) and then validated the candidate WNT ligands in 32 PsA patients and 16 HC. Through RNA interference against WNT ligands in MDOC, we determined the mechanisms by which TNF-α exerts its effects on osteclastogenesis or chemotaxis. Results: The results showed numbers of CD68+WNT5A+ osteoclasts are increased in PsA joints. WNT5A was selectively upregulated by TNF-α in MDOC from PsA patients. However, direct osteoclastogenesis effect (RANK expression) by TNF-αwas not inhibited by WNT5A siRNA. Instead, CXCL1, CXCL16, and MCP-1 was selectively increased in supernatants of MDOC from PsA patients. RNA interference against WNT5A abolished the increased MCP-1 from MDOC and THP-1-cell-derived osteoclasts. The increased migration of osteoclast precursors (OCP) induced by supernatant from PsA MDOC was abolished by MCP-1 neutralizing antibody. WNT5A and MCP-1 expressions were decreased in MDOC from PsA patients treated by biologics against TNF-a but not IL-17. TNF-α recruits OCP WNT5A-mediated directly activates results indicate the of WNT5A in The previous showed that Wnt5a increased the expression of Rank and active osteoclastogenesis in a murine arthritis model. We then investigated whether RANK mediated osteoclastogenesis in PsA. As anticipated, the expression level of RANK was higher in MDOC following and RANKL+ TNF-α than following with We also explored we RNA expression level after M-CSF, RANKL, and TNF-α These results suggest that WNT5A does not directly contribute to active osteoclastogenesis and RANK expression in MDOC from PsA patients. This study revealed increased WNT5A expression in MDOC from PsA patients. TNF-α increases MCP-1 production from MDOC through WNT5A upregulation in patients with PsA. Anti-TNF-α agents decreased OCP recruitment through MCP-1 inhibition.


Introduction
Psoriatic arthritis (PsA) is a chronic in ammatory joint disease. Unrecognized joint damage can lead to permanent joint deformity and functional impairment [1][2][3]. The joint destruction is associated with multifocal bony erosion and resorption by active osteoclasts. Tissue osteoclasts are derived from precursors of the monocyte/macrophage lineage [4][5][6]. The differentiation of the osteoclasts from their monocyte precursors is tightly regulated by a cascade of integrated signaling steps, including the macrophage colony stimulating factor (M-CSF) and the receptor activator of NF-κB ligand (RANKL) [6]. M-Carlsbad, CA, USA) using samples of RNA from the CD14+ monocytes of HC and patients with PsA. The PCR program consisted of an initial denaturation at 95°C for 20 seconds followed by 45 cycles of qRT-PCR at 95°C for 3 seconds (denaturation) and 60°C for 30 seconds (annealing, extension, and reading uorescence). The primer sequences for the different WNT ligands are listed in Supplementary Table 1. Differentiation of osteoclasts (MDOC) from human circulatory CD14+ monocytes Puri ed human CD14+ monocytes were seeded at 2.5 × 10 5 cells/well in 96-well plates containing αminimum essential medium (α-MEM) with fetal bovine serum (FBS) (10%, v/v; Invitrogen, Waltham, MA) and M-CSF (20 ng/mL; PeproTech, Rocky Hill, NJ) for 3 days. RANKL (25 ng/mL) and TNF-α (50 ng/mL) (both from PeproTech, Rocky Hill, NJ, USA) were added every 3 days for 9 days to induce osteoclast differentiation. The osteoclasts were identi ed by staining with tartrate-resistant acid phosphatase (TRAP) on Day 13 using the Acid Phosphate Leukocyte Kit (Sigma, St. Louis, MO). TRAP-stained cells containing three or more nuclei were de ned as osteoclasts. [21] The numbers of osteoclasts were counted and averaged from four high-power elds (HPFs) (100×) per well.

Results
The demographics of patients with PsA and HC  Table 2).
Increased transcription and translation of WNT5A in MDOC and tissue osteoclasts from affected joints in

PsA patients
We conducted a small pilot study to pro le the transcriptional levels of all the WNT ligands using qRT-PCR in CD14+ monocytes and MDOC from PsA patients (n=3) and HC (n=3). Among the WNT ligands, the expression level of WNT5A was selectively increased 15-fold in MDOC from PsA patients compared to the level in MDOC from the HC (p<0.05) (Figure 1a). To validate the selective upregulation of WNT5A in MDOC from PsA patients, we measured the RNA expression level of WNT5A in MDOC from more PsA patients (n=32) and HC (n=16). The results showed that the transcriptional expression of WNT5A was higher in the MDOC from PsA patients than in those from the HC (p=0.0001) ( Figure 1b). As we had observed an increase in the expression of WNT5A mRNA in the MDOC from PsA patients, we next examined whether the level of the WNT5A protein was similarly increased using Western blotting. The results showed that the level of the WNT5A protein was higher in MDOC from PsA patients (n=3) than in those from HC (n=3) (p<0.05) (Figure 1c). Furthermore, we investigated whether WNT5A expression was increased in osteoclasts in the destructive joints of PsA patients. We collected joint tissues from PsA patients (n=5) and osteoarthritic patients (n=5) who had received joint replacement and stained them with WNT5A and CD68 using immunohistochemical staining. The results showed increased numbers of WNT5A-and CD68-expressing osteoclasts in the joints of the PsA patients compared to those of the osteoarthritis patients (p=0.0019) (Figure 1d). We con rmed that the expression of WNT5A was selectively increased in the osteoclasts of PsA patients.
TNF-α activates WNT5A pathway, which is independent of osteoclastogenesis in PsA WNT5A was highly expressed in the MDOC of the PsA patients. We investigated which cytokines contributed to this increased expression level. RNA samples from monocytes and MDOC were analyzed using qRT-PCR. The results show that the expression of WNT5A was signi cantly increased in MDOC by M-CSF, RANKL and TNF-α treatment compared to that with medium only and/or M-CSF+ RANKL treatment (p<0.05) (Figure 2a). These results indicate that TNF-α signi cantly increased the expression of WNT5A in MDOC from PsA patients. The previous study showed that Wnt5a increased the expression of Rank and active osteoclastogenesis in a murine arthritis model. [15] We then investigated whether RANK mediated osteoclastogenesis in PsA. As anticipated, the expression level of RANK was higher in MDOC following RANKL and RANKL+ TNF-α treatment than following treatment with medium only (p<0.05) ( Figure 2a). We also explored whether WNT5A regulated RANK or osteoclastogenesis. After WNT5A interference, we measured the expression levels of WNT5A and RANK using qRT-PCR and the number of osteoclasts using TRAP staining. The results showed that WNT5A mRNA was signi cantly downregulated in the WNT5A siRNA but not in the control siRNA group (p < 0.05) (Figure 2b). The number of TRAP+ osteoclasts was induced after combined M-CSF, RANKL, and TNF-α treatment, it was not changed by WNT5A RNA interference (Figure 2c and d). The increased expression level of RANK after combined M-CSF, RANKL, and TNF-α treatment was not inhibited by WNT5A siRNA (Figure 2e). These results suggest that WNT5A does not directly contribute to active osteoclastogenesis and RANK expression in MDOC from PsA patients.
Selective induction of MCP-1, but not CXCL1 or CXCL16, by WNT5A in MDOC from PsA patients We wanted to determine whether chemokine or cytokine production was increased in the MDOC of the PsA patients compared to those of the HC. The supernatants from the MDOC of the PsA (n=3) patients and HC (n=3) were analyzed using a multiplex chemokine assay. Among 36 chemokines, the results showed higher expression levels of CXCL1, CXCL16, and MCP-1 in the supernatants of the MDOC from the PsA patients than those from the HC (p<0.05) (Figure 3a and b). We further investigated whether WNT5A regulated the production of CXCL1, CXCL16, and MCP-1. The levels of the three cytokines in the supernatants of monocytes and MDOC with/without WNT5A RNA interference were measured using ELISA. The results show that the production of CXCL1, CXCL16 and MCP-1, was increased in MDOC from PsA patients and that the enhanced MCP-1 production (but not that of CXCL1 or CXCL16) in the MDOC from the PsA patients was signi cantly decreased by WNT5A blockade (Figure 3c, d and e). WNT5A regulated the production of MCP-1 in MDOC from PsA patients.
TNF-α increased MCP-1 production through WNT5A in THP-1-cell-derived osteoclasts We found increased expression of WNT5A in MDOC in patients with PsA. We next investigated whether the upregulation of WNT5A by TNF-α could be recapitulated in vitro and sought to decipher its mechanism. We differentiated THP-1 cells into osteoclasts to investigate whether WNT5A increased after TNF-α treatment in osteoclasts. RNA samples from THP-1-derived osteoclasts were analyzed using qRT-PCR. The results showed that combined M-CSF, RANKL, and TNF-α treatment induced WNT5A expression, which was signi cantly reduced in cells transfected with WNT5A siRNA (Figure 4a). In parallel, we investigated whether WNT5A interference abrogated the expression of RANK. The expression level of RANK increased after M-CSF, RANKL and TNF-α treatment, it is not decreased by WNT5A RNA interference. (Figure 4b). We then explored whether WNT5A interference abrogated the production of CXCL1, CXCL16, and MCP-1. The concentrations of CXCL1 for THP-1 cells with medium only, THP-1 cells with PMA treatment, THP-1-cell-derived osteoclasts, THP-1-cell-derived osteoclasts with control siRNA, and THP-1-cell-derived osteoclasts with WNT5A siRNA (100 µg/mL) were low and did not signi cantly differ among the groups (Figure 4c). Although the production of CXCL16 was induced after combined M-CSF, RANKL, and TNF-α treatment, it was not changed by WNT5A RNA interference (Figure 4d). Notably, the production of MCP-1 was induced after the combined M-CSF, RANKL, and TNF-α treatment but was decreased by more than 50% by WNT5A RNA interference (Figure 4e).

MCP-1 in supernatants of MDOC helps to recruit OCP in PsA
We observed increased MCP-1 in the supernatant of MDOC from PsA patients; we then explored whether this recruited more OCP. The supernatants of MDOC from PsA patients (n=5) were treated with MCP-1 antibody or a mock antibody. We tested the direct effects of supernatant from MDOC with/without MCP-1 antibody on monocyte migration using Transwell assays in which CD14+ monocytes from HC were added to the upper chamber and assessed for their response to chemotactic stimuli from the culture supernatant of MDOC from PsA patients, which was added to the lower chamber. The ratio of CD14+ monocyte migration was increased when the supernatants were added, and the amount of migration was similar to that induced by the addition of MCP-1 at 1500 pg/mL (Figure 5a). MCP-1 antibody treatment at 20 or 40 µg/mL signi cantly reduced the enhancement of the migration of MDOC by the supernatant. Furthermore, we wanted to determine whether MCP-1 in the supernatant of MDOC could recruit more CCR2+RANK-expressing OCP. Coupled with a chemotactic assay and ow cytometry to identify OCP, the data show that the percentages of CCR2+RANK-expressing OCP in peripheral CD14+ monocytes, the upper chamber and the lower chamber were 1.5±0.5%, 2±1.2%, and 1.5±1.1%, respectively (Figure 5b). The results indicate that the increased MCP-1 in the supernatants of the MDOC from PsA patients recruited high numbers of CCR2+RANK-expressing OCP.
Both WNT5A expression and MCP-1 production in MDOC of PsA patients were decreased by TNF-α blockade We observed that WNT5A-mediated MCP-1 production in MDOC recruits OCP in PsA. We next investigated whether TNF-α or IL-17 modulated WNT5A expression in MDOC in PsA. For this, MDOC were cultivated from PsA patients as described above. On Days 3 and 9, cells were treated with different concentrations of TNF-α or IL-17 inhibitors. WNT5A was upregulated as expected when CD14+ cells were treated with combined M-CSF, RANKL, and TNF-α. The upregulation of WNT5A was abolished by TNF-α blockers including etanercept and adalimumab, but not the IL-17 blocker secukinumab (Figure 6a). We were also interested in whether the production of MCP-1 from MDOC was mediated by TNF-α or IL-17. The concentrations of CXCL1, CXCL16, and MCP-1 in the supernatants of MDOC treated with or without TNFα blockers or IL-17 blockers were measured using ELISA. The results show that productions of CXCL1 and CXCL16 increased after M-CSF, RANKL, and TNF-α treatment, they were not changed by anti-TNF-α treatment (etanercept, 400 or 800 µg/mL, or adalimumab, 400 or 800 µg/mL) or anti-IL-17A treatment (secukinumab, 400 and 800 µg/mL) (Figure 6b and c). However, the production of MCP-1 was enhanced in cells that received combined M-CSF, RANKL, and TNF-α treatments. The enhancement was reduced by more than 50% when cells were treated with TNF-α blockers including etanercept and adalimumab, but not with the IL-17A blocker secukinumab (Figure 6d). Overall, we provide compelling evidence that TNF-α mediates WNT5A-dependent MCP-1 production, which drives the migration of OCP to the PsA joint.

Discussion
This is the rst study to investigate the role of WNT signaling in MDOC from patients with PsA. Our results reveal higher WNT5A expression levels in MDOC from PsA patients than in those from HC. In addition, WNT5A expression was increased in osteoclasts in the damaged PsA joints compared to that in those from osteoarthritis. MCP-1-mediated OCP migration could be abolished by WNT5A RNA interference and blocking TNF-α (but not by blocking IL-17).
The WNT signaling pathways are involved in physiological bone metabolism [14]. Our results show higher expression levels of WNT5A mRNA and protein in MDOC from PsA patients than in those from HC. The increased expression of WNT5A in the osteoclasts of destructive joints con rms its pathogenic role in active osteoclastogenesis in PsA patients. Previous studies showed that TNF-α induced the expression of WNT5A in human monocytes [21] and dental pulp cells [22]. WNT5A expression was increased in the synovial uids of patients with spondyloarthropathy compared to those with osteoarthritis [23]. Our results consistently show increased expression of WNT5A after TNF-α treatment in the MDOC of PsA patients. Wnt5a promotes osteoclast differentiation and function via RANK expression, thereby enhancing RANKL-induced osteoclastogenesis in mouse models [15]. The knockout of late-stage osteoclast-speci c Ror2, a protein downstream of Wnt5a, increases bone mass [24,25]. However, our results show that WNT5A interference did not decrease the expression of RANK and inhibit osteoclastogenesis. Instead, WNT5A regulates MCP-1 production, which recruits OCP. In fact, WNT5A has been reported to upregulate MCP-1 expression in macrophages [26]. The WNT5A treatment of human dental pulp cells increased the production of cytokines and chemokines, including IL-8, CXCL1, MCP-1, and CCL5 [22]. This result may re ect the chronic in ammatory burden in PsA.
Increased MCP-1 in the serum was found to be a potential biomarker for distinguishing PsA from osteoarthritis [28]. Our results show increased concentrations of MCP-1, CXCL1, and CXCL16 in the supernatants of MDOC from PsA patients. Furthermore, the results con rm that WNT5A interference could selectively inhibit the production of MCP-1 from MDOC in PsA patients. In addition, the results con rm that the production of MCP-1 by THP-1-cell-derived osteoclasts was regulated by WNT5A.
Researchers have shown that increased CXCL16 contributed to the retention of CXCR6+ Tc17 cells in PsA synovial uid [29]. CXCL16 could be produced as a transmembrane-bound chemokine by monocytes, macrophages, and dendritic cells; CXCL16 may contribute to their recruitment and persistence in the in amed PsA joint [29]. Cytological analysis revealed CXCL1 mRNA to be located mainly in monocytic cells in the synovial uid of PsA patients [27]. Hardaway et al. reported that CXCL1 could stimulate osteoclast differentiation in vitro [30].
MCP-1 was induced during TNF-α-mediated osteoclast differentiation. [31] Early work showed that the recruitment of monocytes to the bone surface is mediated by MCP-1 [32,33]. The link between mechanical strain and the onset of arthritis appears to depend on the local recruitment of Ly6 high in ammatory monocytes elicited by the mechanostress-induced MCP-1/CCR2 axis [34]. MCP-1 was reported to serve as a chemotactic signal for OCP via the CCR2 receptor [35]. Our results indicate that MCP-1 in the supernatants of MDOC from PsA patients recruits CCR2+RANK-expressing OCP. The number of OCP was decreased in PsA patients by successful anti-TNF-α treatment [8]. The good clinical responses to TNF-α inhibition in patients with PsA are well known, although the mechanisms could be multifactorial [36]. One previous study showed that the high numbers of OCP in the peripheral blood of PsA patients were decreased signi cantly by anti-TNF-α agents [37]. However, the impact of OCP regulation is not known. Anti-TNF-α treatment has been reported to decrease MCP-1 production from cultured mononuclear cells from the synovial uid of patients with PsA [38]. Our results show that anti-TNF-α agents (etanercept and adalimumab) decreased both WNT5A expression and MCP-1 production. Furthermore, the decreased recruitment of OCP through WNT5A represents one plausible mechanism independent of blocking osteoclastogenesis for the clinical e cacy of TNF-α inhibitors in treating PsA.

Conclusions
This study revealed increased WNT5A expression in MDOC from PsA patients. TNF-α increases MCP-1 production from MDOC through WNT5A upregulation in patients with PsA. Anti-TNF-α agents decreased OCP recruitment through MCP-1 inhibition.

Declarations
Ethics approval and consent to participate This study was approved by the Institutional Review Board of Chang Gung Memorial Hospital (IRB-201802336A3). All the individual data were anonymous before analysis, and informed consent was waived.

Consent for publication
Not applicable.

Availability of supporting data and materials
The authors con rm that the data supporting the ndings of the present study are available within the manuscript and the supplemental data.

Competing interests
The authors have declared that no competing interests exist.

Funding
The authors received grants from Ministry of Science and Technology of Taiwan (MOST 108-2314-B-182A-105 -MY3) and the Chang-Gung Memorial Foundation (CMRPG8J0411). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Authors' contributions SHL designed the study, conducted the experiments, analyzed the results and wrote the manuscript. CHL and CCH contributed to study design, analyzed the results and critically revised the paper. SCL contributed to data analysis. JCH, CYH, and WYC contributed to data acquisition.      PsA patients. On Day 13, the supernatants from MDOC were collected for further Transwell chemotaxis assays. (a) Supernatants were mixed with MCP-1 antibody at 20 or 40 ng/mL or corresponding isotype antibody at 40 ng/mL for the lower chamber. In other sets, culture medium supplemented with recombinant MCP-1 protein at 100 and 1500 pg/mL was placed in the lower chamber. The CD14+ monocytes from HC were placed above the lter (upper chamber) to allow chemotaxis for 1 hour. We measured the migration of CD14+ monocytes to the lower chamber using a Cell Counting Kit-8. (b) The numbers of CD14+ monocytes in the upper chamber and the lower chamber were determined. OCP, de ned as CCR2+RANK-expressing CD14+ monocytes, were identi ed by multicolor ow cytometry. The secondary antibodies were conjugated with different uorescent markers (RANK: FITC and CCR2: PE). The isotypes of individual antibodies were used as negative controls. The medium only or MCP-1 recombinant protein (1500 pg/mL) was added into the lower chamber. The percentages of CCR2+RANKexpressing CD14+ monocytes in the upper and lower chambers were measured. ** p indicates < 0.01.

Figure 6
Anti-TNF-α agents inhibit the expression of WNT5A and production of MCP-1 from supernatants of MDOC in PsA patients. MDOC were obtained from 5 PsA patients as described previously. MDOC were treated with anti-TNF-α (adalimumab at 400 and 800 μg/mL or etanercept at 400 and 800 μg/mL) and anti-IL-17a (secukinumab at 400 and 800 μg/mL) agents on Days 3 and 9. On Day 13, the supernatants and RNA samples from MDOC were collected. (a) The expression level of WNT5A in MDOC was measured using qRT-PCR. The levels of CXCL1 (b), CXCL16 (c), and MCP-1 (d) in the supernatants were measured via ELISA. * indicates p<0.05, ** indicates p < 0.01, *** indicates p < 0.001 and **** indicates p < 0.0001.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.