CD34+THY1+ synovial fibroblast subset in arthritic joints has high osteoblastic and chondrogenic potentials in vitro

Objective Synovial fibroblasts (SFs) in rheumatoid arthritis (RA) and osteoarthritis (OA) play biphasic roles in joint destruction and regeneration of bone/cartilage as mesenchymal stem cells (MSCs). Although MSCs contribute to joint homeostasis, such function is impaired in arthritic joints. We have identified functionally distinct three SF subsets characterized by the expression of CD34 and THY1 as follows: CD34+THY1+, CD34−THY1−, and CD34−THY1+. The objective of this study was to clarify the differentiation potentials as MSCs in each SF subset since both molecules would be associated with the MSC function. Methods SF subsets were isolated from synovial tissues of 70 patients (RA: 18, OA: 52). Expressions of surface markers associated with MSCs (THY1, CD34, CD73, CD271, CD54, CD44, and CD29) were evaluated in fleshly isolated SF subsets by flow cytometry. The differentiation potentials of osteogenesis, chondrogenesis, and adipogenesis were evaluated with histological staining and a quantitative polymerase chain reaction of differentiation marker genes. Small interfering RNA was examined to deplete THY1 in SFs. Results The expression levels of THY1+, CD73+, and CD271+ were highest and those of CD54+ and CD29+ were lowest in CD34+THY1+ among three subsets. Comparing three subsets, the calcified area, alkaline phosphatase (ALP)-stained area, and cartilage matrix subset were the largest in the CD34+THY1+ subset. Consistently, the expressions of differentiation markers of the osteoblasts (RUNX2, ALPL, and OCN) or chondrocytes (ACAN) were the highest in the CD34+THY1+ subset, indicating that the CD34+THY1+ subset possessed the highest osteogenic and chondrogenic potential among three subsets, while the differentiation potentials to adipocytes were comparable among the subsets regarding lipid droplet formations and the expression of LPL and PPARγ. The knockdown of THY1 in bulk SFs resulted in impaired osteoblast differentiation indicating some functional aspects in this stem-cell marker. Conclusion The CD34+THY1+ SF subset has high osteogenic and chondrogenic potentials. The preferential enhancement of MSC functions in the CD34+THY1+ subset may provide a new treatment strategy for regenerating damaged bone/cartilage in arthritic joints. Supplementary Information The online version contains supplementary material available at 10.1186/s13075-022-02736-7.

life expectancy [1]. The homeostatic responses of the joint environments, including the joint repairing, were impaired because of chronic inflammation and mechanical stress [2]. According to these clinical situations, the strategy to enhance tissue regeneration would be a possible strategy to construct complementary treatment.
Synovial fibroblasts (SFs) are contributing to the pathogenesis of RA and OA by secreting inflammatory cytokines, tissue degrading factors, and making pannus formation [2,3]. SFs also possess several characteristics of mesenchymal stem cells (MSCs), including self-renewal capacity and multi-lineage differentiation potentials to mesenchymal tissues [4,5]. MSCs isolated from synovial tissue can be more efficiently differentiated into chondrocytes than those from other tissues (e.g., bone marrow) [6,7]. In addition, MSCs are proliferated in response to mechanical stress or inflammatory cytokines [8,9], suggesting that they can be induced under disease conditions in RA and OA. Since MSC functions as osteogenesis and chondrogenesis were induced in case of bone damages or bone fractures [10][11][12], the therapeutic application of synovial MSCs would have a potent for the repairment of bone erosion in RA.
We previously reported that SFs in arthritic joints are composed of three functionally distinct subsets, CD34 − THY1 − , CD34 − THY1 + , and CD34 + THY1 + population, based on the expression of CD34 and THY1 [13]. The CD34 − THY1 + subset was found to be expanded in patients with RA, whereas the proportion of the CD34 + THY1 + subset was comparable between RA and OA. Both subsets possessed pathological functions, including proinflammatory cytokine secretions, high proliferation capacity, and enhanced invasiveness when compared with the CD34 − THY1 − subset. THY1 and CD34 are the MSC surface markers related to wound repair as well as lineage potentials in osteogenesis and chondrogenesis [14,15]. Since we and others have already elucidated that the functions of THY1 − /THY1 + SF are distinct [16][17][18], we hypothesized that three SF subsets have a different function as MSC and that baseline expression of THY1 in SF subsets would determine the differentiation potentials. Indeed, the THY1 + CD73 + SF subset presented higher chondrogenic potentials than the THY1 − CD73 + subset [19].
In the present study, we evaluated the MSC function of each SF subset regarding differentiation potential to osteoblast, chondrocyte, and adipocyte. Here we demonstrate that the CD34 + THY1 + subset has MSC potential than the others regardless of the background diseases, suggesting future therapeutic applications utilizing MSC function in this subset.

Patient recruitment and isolation of synovial cells
We obtained synovial tissues from surgeries of joint replacement for patients with OA and RA. We consecutively collected all the available OA and RA samples for this study. Written informed consent for this study and the ethics approval from the medical research ethics committee of Tokyo Medical and Dental University (approval number: M2000-979) were obtained. The synovial tissues were collected from 70 patients (RA: 18, OA: 52). All the patients with RA have been received treatment with disease-modified anti-rheumatic drugs (DMARDs) including biologics. Ten patients were administered steroids (average dose, 2.9 mg/day). Other characteristics are described in Table 1. In this study, we statistically analyzed combined data from RA and OA samples because of their limited sample numbers. Our previous study demonstrated that the altered proportions of SF subsets but not subsets themselves differed between RA and OA [13].
Tissue samples were collected consecutively from joint replacement surgeries to eliminate any bias as previously reported [13]. Briefly, joint tissues were obtained immediately after the surgeries, followed by removal of bone and adipose tissues with scissors. Synovial tissues were minced into small pieces and then subjected to enzymatic digestion. For cell culture, we digested tissues with 2 mg/mL collagenase type 4 (Worthington, NJ, USA), 0.8 mg/mL Dispase II, and 0.1 mg/mL DNase I (Roche, Basel, Switzerland) in Dulbecco's modified Eagle Roche's medium (DMEM) at 37 °C. After 15 min, we collected the supernatant and replaced it with a fresh enzyme mix. These procedures were repeated every 15 min for a total of 1 h. After lysing red blood cells with ACK-lysing buffer, obtained cells were treated with antibodies as described below then sorted by FACS Aria II and III (BD Biosciences, CA, USA) with 100-μm nozzle.

Antibodies
The

Flow cytometry analysis
The gating strategy of SF subsets was as shown (Supplementary Figure 1). While the mean ratios of CD34 − THY1 − , CD34 − THY1 + , and CD34 + THY1 + subsets in RA were 35.5%, 24.1%, and 24.8%, respectively, the mean ratios of these subsets in OA were 56.6%, 10.2%, and 17.2%. These results were compatible as previously reported [13]. The mean purity of each subset after sorting was 95.9%, 94.3%, and 98.0%, respectively. We evaluated the expression of MSC surface markers (THY1, CD34, CD73, CD271, CD54, CD29, and CD44) in the SF subsets. Considering the expression of MSC markers in the freshly isolated synovial cells, we evaluated the mean fluorescence intensity (MFI) as expression levels of these markers in the individual subsets by flow cytometry in the 12 consecutive samples (RA: 3, OA: 9) (Supplementary Figure 2).
For chondrogenic differentiation, 1.25-2.5 × 10 5 cells were placed in a 15-mL polypropylene tube (AGC Techno Glass Co., Ltd, Shizuoka, Japan) and centrifuged at 1500 × g for 5 min. The cells were cultured in a chondrogenic induction medium containing 1000 ng/ mL of BMP-2 (PeproTech, NJ, USA) and 10 ng/mL of transforming growth factor-β3 (PeproTech), incubated at 37 °C in 5% CO 2 for 3 weeks. All media were changed twice per week. Histological staining was performed with safranin O staining for chondrogenesis.
For adipogenesis, 7.0 × 10 3 cells/cm 2 are plated and cultured in StemPro ™ Adipogenesis Differentiation Kit (Gibco, Thermo Fisher Scientific, MA, USA) for 3 weeks. All media were changed twice per week. Oil-red staining was used to evaluate adipogenesis.
For chondrogenesis and adipocyte differentiation, we referred to the previous reports with some modifications [20,21]. Briefly, we performed some pellet culture at a density of 1.25 × 10 5 cells due to an imbalance in the number of each subset.

Quantitative real-time polymerase chain reaction (qPCR)
cDNA was synthesized with QuantiTect Reverse Transcription kit (Qiagen, Hilden, German). Quantitative polymerase chain reaction (qPCR) was performed with Brilliant III Ultra-Fast SYBR Green qPCR master mix (Agilent Technologies, CA, USA) on a LightCycler96 ® (Roche). The following primers are used as shown in Table 2.

Knockdown of gene expression by small interfering (si) RNA
Bulk SFs were seeded at 1.2 × 10 4 into 12-well cell culture plates and subsequently transiently transfected with 20 pM of THY1 or control small interfering (si) RNA (Thermo Fisher Scientific) using Lipofectamine RNAiMax (Thermo Fisher Scientific) according to the manufacturer's protocol. Cells were incubated with siRNA for 3 days and subjected to osteogenic differentiation as described above.

Cell survival/proliferation assay
Cell survival/proliferation was evaluated with a watersoluble tetrazolium salt (WST-8) colorimetric assay using Cell Counting Kit-8 (Dojindo) according to the manufacturer's protocol. Briefly, SFs transfected with siRNAs were incubated with 1% WST-8 for 6 h and absorbance of supernatants at 450 nm was measured by a plate reader.

CD34 + THY1 + subset expressed MSC surface markers
Considering the expression of surface markers in the freshly isolated synovial cells, both CD34 + THY1 + and CD34 − THY1 + subsets expressed THY1 with higher levels than the CD34 − THY1 − subset, as we have shown previously [13]. Interestingly, the expression level of THY1 in CD34 + THY1 + was the highest among the SF subsets regardless of the underlying diseases (Fig. 1A).
The expressions of CD73 and CD271 were the highest in the CD34 + THY1 + subset among three subsets (Fig. 1B), whereas the expression levels of CD54 and CD29, a surface marker with low osteogenic potentials derived from other cell types [22,23], were lower in the CD34 + THY1 + subset than the CD34 − THY1 − subset (Fig. 1B). The expression level of CD44 was not significantly different among the three subsets (Fig. 1B). We also analyzed data from RA and OA separately (Supplementary Figure 6). Although statistical significance was observed in only RA + OA and OA data, RA and OA showed the same tendency.
Regarding the transcriptions of MSC surface markers in microarray data, these were not significantly different among the three subsets (Supplementary Figure 3), indicating that these molecules are not regulated differently at the transcriptional levels. We also evaluated the expression level of surface THY1, CD34, and CD73 by flow cytometry after 4-week expansion for the subsequent differentiation experiments (Supplementary Figure 4). The expression of CD34 in the CD34 + THY1 − subset was lost after expansion, while that of THY1 and CD73 was comparable among the three subsets.

CD34 + THY1 + subset presented the highest osteogenic potentials
The average ratio of calcified area stained with alizarin red as red spots was 4 and 1.5 times higher in CD34 + THY1 + subsets than CD34 − THY1 − and CD34 − THY1 + subsets ( Fig. 2A, B). We also stained three SF subsets with ALP staining to verify whether observed calcification is due to the differentiation of CD34 + THY1 + SFs into osteoblasts. The activity of ALP, which was expressed as blue-stained area, was observed at twice higher levels in the CD34 + THY1 + subset ( Fig. 2A, B) than the CD34 − THY1 − subset.  To confirm the osteoblast differentiation by quantifying differentiation-associated genes, the expressions of mRNA levels were measured. The expression levels of RUNX2, the master regulator for osteoblast differentiation [24], were 2.8 times higher in the CD34 + THY1 + subset than the CD34 − THY1 − subset (Fig. 2C). As for expression levels of ALPL, which is a marker in the early stage of osteoblast differentiation, it was expressed 1.8 times higher in the CD34 + THY1 + subset than the other two subsets after the differentiation for 3 weeks Fig. 1 The expression of THY1, CD34, and MSC surface markers in the individual subset. A THY1 and CD34 expression. The mean fluorescence intensity (MFI) of THY1 and CD34 was evaluated with flow cytometry. Data was shown in total samples (OA+RA). B-D MSC surface markers. The MFI of MSC surface markers (CD73, CD271, CD54, CD29, and CD44) was evaluated with flow cytometry. OA and RA samples are plotted as white circles and white squares, respectively. Data were analyzed by Holm-Sidak's multiple comparison test for comparing each subset from the same samples (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001) (Fig. 2C). The expression levels of OCN, a differentiation marker in the mature osteoblast, were 1.7 times higher in the CD34 + THY1 + subset than the CD34 − THY1 − subset after the differentiation for 4 weeks (Fig. 2C). These findings indicate that THY1 + subsets, especially the CD34 + THY1 + subset, are the subset with the most superior osteogenic potentials in vitro. We also analyzed data from RA and OA separately (Supplementary Figure 7). Although statistical significance was observed in only RA + OA and OA data, RA and OA showed the same tendency.
Confirming the quantitative verification of chondrogenesis, expression levels of ACAN, which codes aggrecan, were evaluated. In both CD34 − THY1 + and CD34 + THY1 + subsets, ACAN expression levels were 2.8 and 8 times higher than in the CD34 − THY1 − subset, respectively (Fig. 3B). These findings indicated that THY1 + subsets, especially the CD34 + THY1 + subset, have the highest chondrogenic potential in vitro. We also analyzed data from RA and OA separately (Supplementary Figure 8). Although statistical significance was Osteogenic potentials of SF subsets. A Representative photographs of whole dishes stained with alizarin red and ALP. Individual subsets were incubated in the calcification differentiation medium for 3 weeks. B Quantification of the staining. Cells per donor were cultured in 3 dishes and the results of 10 donors for alizarin red and 11 donors for ALP were plotted. C Gene expression levels by qPCR. The expression levels of the genes were plotted as a ratio to the expression level of 18S of undifferentiated SF as a control. Cells per donor were cultured in 3 dishes and the results of 10 donors for RUNX2 and ALPL and 9 donors for OCN. OA and RA samples are plotted as white circles and white squares, respectively. The average data of each subset are plotted as black squares. Data were analyzed by Dunn's multiple comparison test for comparing each subset from the same patients (*p < 0.05, **p < 0.01) observed in only RA + OA and OA data, RA and OA showed the same tendency.

Adipocyte differentiation potential was not significantly different among the subsets
After culture in the adipocyte differentiation medium for 3 weeks, all SF subsets comparably presented lipid droplets (Fig. 4A). The expressions of adipocyte-associated genes, including LPL and PPARγ, which codes lipoprotein lipase and peroxisome proliferator-activated receptor γ, were evaluated. The expression levels of LPL and PPARγ transcripts were not significantly different (Fig. 4B). We also analyzed data from RA and OA separately (Supplementary Figure 9). Although statistical significance was not observed, RA and OA showed the same tendency. These findings indicated that osteogenic and chondrogenic potentials were relatively high in the CD34 + THY1 + subset, whereas adipogenic potential was comparable among the three subsets.

THY1 knockdown suppressed osteoblast differentiation
Since THY1 is one of the MSC surface markers and associated with lineage potentials in osteogenesis and chondrogenesis, we hypothesized that baseline expression of THY1 in SF subsets determined differentiation potentials. Therefore, we investigated the endogenous function of THY1 in the osteoblast differentiation by the depletion using RNA interference in bulk SFs.
THY1 knockdown had minor effects on the survival and proliferation of SFs as assessed by WST-8 assay (Supplementary Figure 5A). We also examined the effects of THY1 knockdown on the expression of other MSC markers (Supplementary Figure 5B). The expression of CD29 was significantly suppressed by THY1 knockdown. CD73 was also suppressed in most samples. We could not detect any changes in CD271 and CD54 expressions.
Compared with those treated with siRNA control, THY1-deficient SF demonstrated impaired calcification as well as attenuated ALP activity after 4-week culture in an osteogenic differentiation medium (Fig. 5A,   B). Expressions of THY1, RUNX2, and ALPL were also significantly suppressed (Fig. 5C). These findings supported the hypothesis that the expression of endogenous THY1 is inevitable for the MSC functions in the CD34 + THY1 + subset.

Discussion
In the present study, we identified significantly higher differentiation potency in the CD34 + THY1 + subset by evaluating the osteogenic and chondrogenic differentiation potentials. The pattern of the MSC-associated surface markers also supported the highest potential as MSCs in the CD34 + THY1 + subset.
Our findings were consistent with the previous reports that THY1 expression was associated with osteogenesis in bone marrow-derived MSCs [25]. Additionally, THY1 is involved in angiogenesis through the differentiation of endothelial cells [26]. In arthritic joints, perivascular SFs are exposed to Notch3 signaling from vascular endothelial cells, which is essential in the induction and maintenance of the expression of THY1 [27]. THY1 + subsets, which predominantly localize at the perivascular lesion in the synovium, may be induced due to angiogenesis followed by synovitis and joint damage. In contrast to the differentiation potentials, THY1 interacts with integrin and induces cell apoptosis via activation of the caspase 3/7 pathway [28,29]. Since THY1 expression was enhanced in fibroblasts when joint tissue was injured [30,31], these findings suggest that THY1 exerts to orchestrate the maintenance of joint homeostasis. In the arthritic joints, the CD34 + THY1 + subset might be induced compensatory by the mechanical stress/injury in the joints.
Some MSC markers including THY1 are functionally involved in the lineage commitment. Among them, CD73, an ecto-5′-nucleotidase, which produces extracellular adenosine, affects the osteoblast differentiation of MSC [32]. It is interesting to note that CD73 is highly expressed in THY1 + subsets (Fig. 1A) and that THY1 knockdown suppressed CD73 expression. Thus, CD73 in addition to THY1 may be also responsible for the high osteogenic differentiation potential of THY1 + subsets. There is one study demonstrating opposing effects of THY1 on the osteogenic differentiation. They employed MSCs harvested from dental pulp, adipose tissue, and amniotic fluid, and the lentiviral shRNA transduction to archive THY1 knockdown [33]. These differences may explain the opposite results by them and others [34] including our study.
In addition to THY1, we identified CD34 as a complementary marker to enrich MSCs from freshly isolated synovial cells. Since CD34, a marker for hematopoietic stem cells [35], was used as a negative marker for MSC isolation so far, CD34 is positive in freshly isolated bone marrow-derived MSC (BMSC) and CD34 + BMSCs produced greater fibroblast colony-forming units than CD34 − BMSCs [36]. CD34 is also expressed not only on hematopoietic stem cells but also 10% of circulating fibrocytes, which are recruited to the injured site and associated with inflammation and wound repair [37][38][39]. Although the potentials to differentiate osteoblasts and chondrocytes circulating fibrocytes have been analyzed in fibrotic lung tissue [40], the differentiation potentials in SFs, especially CD34 +/− populations, have not been compared. Therefore, this is the first report evaluating the MSC function in the CD34 + THY1 + double-positive SF subset.
In active arthritis, the osteogenic and chondrogenic differentiation potentials in THY1 + subsets may be overwhelmed or impaired due to inflammation. Although TNF and IL-6 blocking therapy have similar efficacy for RA patients [41], erosion repairment was observed most frequently under treatment with IL-6 blocking therapy [42,43]. Since MSC functions were enhanced in the presence of inflammatory cytokines, including IL-1β or TNF α [8,44], bone repair would result from the fine-tuning of inflammatory mediators in the arthritic joint. Suppression of several inflammatory cytokines ameliorates arthritis effectively, whereas these might not be beneficial for inducing MSC functions.
As the CD34 + THY1 + subset has a highly potent MSC function, these cell populations might be appropriate to be applied for joint repairing therapy. However, the CD34 + THY1 + subset would simultaneously contribute to the RA pathogenesis by the secretion of inflammatory cytokines and by enhanced proliferation potential [13]. We need to find the way to preferentially utilize the MSC function in THY1 + subsets or modify proinflammatory cytokine production in CD34 + THY1 + for further therapeutic application. To enhance MSC potentials, we would consider several approaches to induce the expression of THY1 + in SFs. Jagged 1 and Delta-like 4, which are one of the Notch3 ligands, may be useful for THY1 induction [27,45]. Stimulation of Notch3 signaling may be applicable to acquire osteogenic and chondrogenic differentiation potentials in SFs. A low oxygen environment is beneficial for MSC expansion and chondrogenesis by activating hypoxiainducible factor α and upregulation of THY1 [46].
Direct introduction of MSCs in the inflammatory joints has been proved to have immunosuppressive effects by induction of inducible regulatory T cells, which leads to inhibition of T cell proliferation and cytokine production [47]. In addition, their systemic administration improves arthritis by inhibiting osteoclast differentiation in the arthritic animal model [48]. Clinical trials by administrating human MSCs into joint spaces have been examined [4,49,50]. Alternatively, MSC-derived exosomes, but not MSCs themselves, would be candidates for a novel treatment strategy for RA and OA since they can promote chondrogenesis [51].
Our study comprises several limitations. First, more than half of the synovial tissues were derived from OA, and the comparison of differentiation potential between RA and OA was not sufficient. Second, osteogenic and chondrogenic potentials in the CD34 + THY1 + subset were not evaluated under the in vivo environment. Third, we had not analyzed the relevance of previously reported signaling pathways, including the Wnt pathway [52]. Fourth, we used bulk SFs for the THY1 knockdown experiment because of the limited cell number of each subset. Fifth, the functions of MSC surface markers other than THY1 were not examined.

Conclusions
The multipotency of differentiation in SFs has drawn attention of researchers as such MSC-like characteristics would expect us to develop novel therapeutic strategies for joint repair in RA and/or OA. The CD34 + THY1 + subset presented high osteogenic and chondrogenic differentiation potentials. This subset also most intensively expressed MSC-positive surface markers at the protein level, CD73 and CD271 as well as THY1. The preferential enhancement of MSC functions in the CD34 + THY1 + subset may provide a new treatment strategy in the regeneration of damaged bone/cartilage in advanced RA and/or OA.