2.1 Chemicals

Human recombinant tumor necrosis factor (TNF)-α and platelet-derived growth factor (PDGF) were purchased from R&D Systems (Minneapolis, USA). An IDO inhibitor [1-methyl-l-tryptophan (1-MT)] and CD73 inhibitor (α,β-methylene ADP, APCP) were purchased from Sigma-Aldrich (St Louis, USA). A CD39 inhibitor (sodium polyoxotungstate, POM1) was supplied by Tocris Bioscience (Bristol, UK). In some experiments, GMSCs were pretreated with 1-MT (500 μmol/L), APCP (100 μmol/L), and POM1 (100 μmol/L).

2.2 GMSC, human dermal fibroblast, and BMSC culture

Human gingiva tissue samples were obtained through routine dental procedures at the Department of Dentistry at Penn State University Hershey College of Medicine and the First Affiliated Hospital, Zhejiang University School of Medicine. Human GMSCs were prepared from these samples in accordance with the protocol in our previous study.²¹ All experimental protocols were approved by the Medical Ethics Committees of the Institutional Review Boards at both institutes. Informed consent was obtained from donors before the study. GMSCs were used in experiments from the third to sixth passage.

A human dermal fibroblast cell line (ATCC, PCS-201-012) and BMSCs (ATCC, PCS-500-012) were obtained from ATCC (https://www.atcc.org/). They were used as negative and positive controls of GMSCs.

2.3 Flow cytometry

The following fluorescent dye-conjugated mouse antibodies were used for flow cytometric analysis: mouse anti-human CD90, anti-CD29, anti-CD44, anti-CD73, anti-CD105, anti-CD39, anti-CD4, anti-CD34, anti-CD45, anti-CD14, and anti-CD11b (Biolegend, San Diego, USA). Cell subsets were stained with monoclonal antibodies and the isotype control and analyzed on a FACSCalibur flow cytometer using CellQuest Software (Becton Dickinson). Plots were generated using FlowJo Software (Treestar Inc, Ashland, USA).

2.4 RA FLSs

Synovial tissue specimens were collected from six female patients with active RA, who fulfilled the ACR 1987 revised criteria for the classification or the 2010 ACR/European League Against Rheumatism criteria of RA^{28, 29} and underwent synovectomy or joint replacement surgery. All patients provided written informed consent before the procedure. RA fibroblast-like synoviocytes were isolated from the synovial tissue of each RA patient separately. The synovial tissue was finely minced and digested with 1 mg/mL type I collagenase (Sigma-Aldrich) in serum-free Dulbecco's modified Eagle's medium (DMEM) at 37°C with 5% CO₂ for 2 h. The cell suspension was filtered through a sterile cell strainer (BD Biosciences, USA). Synoviocytes were collected and rinsed by centrifugation. The pellet was resuspended in DMEM containing 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 μg/mL streptomycin, and transferred to a 100 mm dish. At 90% confluence, RA FLSs were trypsinized and passaged to another dish. Passage 4–6 RA FLSs were used in experiments.

2.5 Semiquantitative real-time PCR

RA FLSs were stored in TRIzol reagent (Invitrogen, China) at −80°C. Total RNA was extracted from the RA FLSs. cDNA was prepared from 1 μg RNA and amplified by semiquantitative real-time PCR using SYBR Green (Takara, Japan). GAPDH was used for normalization. Data were analyzed by the 2^−ΔΔCT method. Each sample was analyzed in triplicate, and the analysis was repeated three times. The primers for target genes were as follows.

IL-8, forward: 5ʹ-GACCACACTGCGCCAACAC-3ʹ

reverse: 5ʹ-CTTCTCCACAACCCTCTGCAC-3ʹ,

IL-6, forward: 5ʹ-ACTCACCTCTTCAGAACGAATTG-3ʹ

reverse: 5ʹ-CCATCTTTGGAAGGTTCAGGTTG-3ʹ,

MMP-1, forward: 5ʹ-GCTAACAAATACTGGAGGTATGATG-3ʹ

reverse: 5ʹ-ATTTTGGGATAACCTGGATCCATAG-3ʹ,

MMP-3, forward: 5ʹ-AGCAAGGACCTCGTTTTCATT-3ʹ

reverse: 5ʹ-GTCAATCCCTGGAAAGTCTTCA-3ʹ,

MMP-13, forward: 5ʹ-TCCTGATGTGGGTGAATACAATG-3ʹ

reverse: 5ʹ-GCCATCGTGAAGTCTGGTAAAAT-3ʹ

MCP-1, forward: 5ʹ-CAGCCAGATGCAATCAATGCC-3ʹ

reverse: 5ʹ-TGGAATCCTGAACCCACTTCT-3ʹ,

GAPDH, forward: 5ʹ-GGAGCGAGATCCCTCCAAAAT-3ʹ

reverse: 5ʹ-GGCTGTTGTCATACTTCTCATGG-3ʹ,

2.6 Transwell invasion assay

RA FLSs were starved for 24 h in serum-free DMEM. A transwell assay was performed in 6.5 mm transwell chambers with 8 μm pores. Briefly, the bottom surface of each membrane was precoated with matrigel. Then, RA FLSs (1 × 10⁴) were added to the upper chamber, and complete DMEM with 10% FBS and GMSCs or human dermal fibroblast cell was added to the lower chamber. After incubation for 24 h at 37°C, the upper surface of each membrane was removed and cleaned with a cotton swab. Cells that had migrated to the bottom side were carefully stained with crystal violet and counted under a light microscope.

2.7 5-Ethynyl-2′-deoxyuridine (EdU) assay

RA FLSs (1 × 10⁴) were starved in a serum-free medium for 24 h in the upper chamber of a transwell plate. GMSCs were seeded in the lower chamber. RA FLSs were treated with PDGF (20 ng/mL) for 72 h and then treated with 50 μmol/L EdU for 2 h. RA FLSs were fixed with paraformaldehyde and then stained with Apollo fluid and Hoechst 33342. EdU incorporation was assessed under a LEICA DMIRB fluorescence microscope in accordance with the manufacturer's instructions (RiboBio Co., Guangzhou, China). Image-Pro Plus software was used for analysis.

2.8 Mice

Seven to eight-week-old male DBA/1J mice were purchased from The Jackson Laboratory (Bar Harbor, USA). Mice were handled in accordance with National Institutes of Health guidelines for the use of experimental animals with approval from the Penn State University Hershey Medical Center for the use and care of animals. Our study followed standard biosecurity and institutional safety procedures (SUZ15-01-2).

2.9 Arthritis induction

Bovine type II collagen (CII; Condrex Inc) was emulsified in an equal volume of complete Freund's adjuvant (Sigma) containing 4 mg/mL heat-denatured mycobacterium (BD Biosciences). DBA1/J mice were intradermally injected at the base of the tail with 100 μg CII per mouse. Mice were euthanized by CO₂ asphyxiation and cervical dislocation at the indicated times. To determine whether GMSCs can treat CIA, mice received an intravenous injection of GMSCs (2 × 10⁶) in the tail vein 14 days after immunization. As negative and positive controls, human dermal fibroblast (2 × 10⁶) and BMSCs (2 × 10⁶) were intravenously injected into mice at the same time.

2.10 Clinical assessment and histopathological evaluation of experimental arthritis

Clinical signs of arthritis were evaluated every 2–3 days to determine arthritis incidence after immunization. Each paw was assessed and scored individually for arthritis severity using a scoring system (scale 0–4).^{3, 21} The total scores were calculated for all paws per mouse with a maximum score of 16.

Hind limbs were harvested from mice after sacrifice on Day 56. Tissue sections were prepared and stained with hematoxylin and eosin following fixation, decalcification, and embedding in paraffin. All slides were evaluated in a blinded fashion. Evidence of synovitis, pannus formation, and bone/cartilage erosion was assessed using a graded scale.²¹

2.11 Statistics

Data are expressed as means ± standard error of the mean (SEM). Data were analyzed using the unpaired t-test (Mann–Whitney) between two groups or one-way analysis of variance (ANOVA) for comparison among multiple groups followed by Tukey's test as appropriate. Differences were considered statistically significant at P < 0.05.

3.1 CD39 expression is higher in GMSCs

We analyzed classic surface markers of GMSCs and found that CD90, CD29, CD44, CD73, and CD105 were all highly expressed in GMSCs and human dermal fibroblasts. CD4, CD34, CD45, and CD11b were all weakly expressed in GMSCs and human dermal fibroblasts. Most importantly, CD39 was highly expressed in GMSCs, but not in human dermal fibroblasts (Figure 1).

3.2 GMSCs decrease mRNA levels of MMPs and pro-inflammatory cytokines of RA FLSs

High levels of pro-inflammatory cytokines and MMPs are crucial characteristics of RA FLSs. Therefore, we examined whether GMSCs regulated the expression of MMPs and pro-inflammatory cytokines in TNF-α-stimulated RA FLSs. IL-8, IL-6, MMP-1, MMP-3, MMP-13, and MCP-1 mRNAs were all decreased in the GMSC-treated group compared with the human dermal fibroblast group (Figure 2).

3.3 GMSCs suppress the proliferation of RA FLSs via CD39/CD73 signaling

To assess the influence of GMSCs on RA FLS proliferation, RA FLSs were stimulated by PDGF and incubated with various numbers of GMSCs (0, 1, 10, and 20 × 10³). Proliferation was then assessed by an EdU incorporation assay. GMSCs suppressed the proliferation of RA FLSs in a dose-dependent manner (Figure 3A,B). Furthermore, when the GMSCs were pretreated with CD39i or CD73i, such suppression was abrogated. GMSCs pretreated with IDOi still had the capacity to suppress RA FLS proliferation (Figure 3C).

3.4 GMSCs suppress RA FLS invasion

Next, we investigated the effect of GMSCs on RA FLS invasion. RA FLSs were incubated with GMSCs, and then a transwell assay was conducted to measure the effects on invasion. GMSCs markedly suppressed the FBS-induced invasion of RA FLSs through matrigel in a dose-dependent manner (Figure 4), indicating that GMSCs suppressed RA FLS invasion.

3.5 GMSC therapy delays arthritis onset and reduces arthritis scores in the CIA model

GMSCs and BMSCs exert anti-inflammatory effects in autoimmune diseases. Therefore, we investigated the effect of GMSCs in the CIA model. GMSCs and BMSCs significantly delayed arthritis onset and markedly reduced arthritis scores in the CIA model, whereas human dermal fibroblast did not have such effects. GMSCs were comparable to BMSCs in reducing arthritis scores (Figure 5A). To determine whether GMSCs prevented bone destruction in CIA mice, we stained joints with HE. A significant decrease in histological changes of synovitis, pannus formation, and destruction of cartilage and bone was observed in the ankle joints after incubation with GMSCs. However, there was no difference in histological changes between GMSC and BMSC groups (Figure 5B).