miRNA-148a-containing GMSC-derived EVs modulate Treg/Th17 balance via IKKB/NF-κB pathway and treat a rheumatoid arthritis model

Abstract

Mesenchymal stem cells (MSCs) have demonstrated potent immunomodulatory properties that have shown promise in the treatment of autoimmune diseases, including rheumatoid arthritis (RA). However, the inherent heterogeneity of MSCs triggered conflicting therapeutic outcomes, raising safety concerns and limiting their clinical application. This study aimed to investigate the potential of extracellular vesicles derived from human gingival mesenchymal stem cells (GMSC-EVs) as a therapeutic strategy for RA. Through in vivo experiments using an experimental RA model, our results demonstrate that GMSC-EVs selectively homed to inflamed joints and recovered Treg and Th17 cell balance, resulting in the reduction of arthritis progression. Our investigations also uncovered miR-148a-3p as a critical contributor to the Treg/Th17 balance modulation via IKKB/NF-κB signaling orchestrated by GMSC-EVs, which was subsequently validated in a model of human xenograft versus host disease (xGvHD). Furthermore, we successfully developed a humanized animal model by utilizing synovial fibroblasts obtained from patients with RA (RASFs). We found that GMSC-EVs impeded the invasiveness of RASFs and minimized cartilage destruction, indicating their potential therapeutic efficacy in the context of patients with RA. Overall, the unique characteristics - including reduced immunogenicity, simplified administration, and inherent ability to target inflamed tissues - position GMSC-EVs as a viable alternative for RA and other autoimmune diseases.

Keywords: Autoimmune diseases; Autoimmunity; Stem cell transplantation; Stem cells.

Figure 1. Human GMSC–derived EVs inhibit T cell responses in vitro.

(A) Electron micrograph analysis of the morphology of EVs. Scale bar: 200 nm. (B) Nanoparticle trafficking analyzed the diameters and concentration of EVs. (C) The EVs’ protein markers were detected by Western blot. (D) PKH67-labeled (green) GMSC-EVs were cocultured with CD3⁺ T cells under stimulation of soluble anti-CD3 and soluble anti-CD28 Abs after 1 days. Cells were harvested and stained with CM-DiI (red) and DAPI (blue), and then images were acquired by fluorescence confocal. (E) In vitro suppressive assay of T cell proliferation. (F) Th17-polarizing analysis. (G and H) Treg-polarizing analysis. (I) In vitro suppressive assay of cytokine production. Statistical significance was assessed by 1-way ANOVA with Dunnett multiple-comparison test in E–I. Data are shown as the means ± SD from 1 of 3 independent experiments. *P < 0.05; **P < 0.01.

Figure 2. Human GMSC–derived EVs protect against collagen-induced arthritis (CIA) model.

(A) Schematic diagram summarized the CIA modeling and G-EV administration. (B) The representative images of gross appearance of swollen hind paws at the endpoint of the experiment. (C–E) The paw thickness (C), incidence of arthritis (D), and arthritis severity scores (E) of CIA mice were monitored from day 15 to day 60 after immunization. (F and G) Ankle joint sections isolated from CIA mice at day 60 after immunization were stained with H&E and toluidine blue staining. Histopathologic scores were evaluated for features of synovitis, pannus, erosion, and cartilage matrix. The red arrows indicated the cartilage destruction of joints. Osteoclast distribution was quantified by tartrate acid resistant phosphatase (TRAP) staining. Scale bars = 500 μm. (H) Toe joint sections isolated from CIA mice at day 60 after immunization were imaged with μ-CT, and the structural damage was evaluated as bone volumes of the metatarsophalangeal joint indicated. (I) dLN cells isolated from CIA mice at day 60 after immunization for intracellular staining of IL-17A and Foxp3 by flow cytometry analysis. (J and K) Splenic cells isolated from CIA mice at day 60 after immunization were collected for the detection of the protein level of RORγt by Western blot analysis. (L–N) dLNs isolated from CIA mice at day 60 after immunization for intracellular staining of TNF-α and IL-10 in CD4⁺ cells by flow cytometry analysis. (O and P) Serum samples obtained from blood of CIA mice at day 60 after immunization were used for the detection of cytokines (O) and autoantibodies (P) by ELISA. Statistical significance was assessed by 1-way ANOVA with Dunnett multiple-comparison test in C–P. Data are mean ± SD, n = 5–8 mice. *P < 0.05; **P < 0.01; ****P < 0.0001.

Figure 3. In vivo tracking of human GMSC–derived EVs in CIA mice.

(A) Schematic illustration depicting the delivery of EVs to the joint via the tail vein for the treatment of CIA. (B) Twenty-four–hour following the administration of DiR-labeled (red) EVs in CIA mice 2, digital photo, and IVIS images were used to present the fluorescence signal. (C) Quantification of fluorescence percentage of joint in total for B. (D) In vivo imaging of mCherry-carried (red) EVs in CIA mice 24 hours after injection, and quantification of fluorescence percentage of joint in total. (E) In vivo imaging of DiR-labeled GMSC-EVs in CIA mice at 24 hours, 14 days, and 28 days after injection, and quantification of fluorescence percentage of joint in total. Left mouse received PBS as the control. Statistical significance was assessed with 2-tailed Student t test in C and D. Representative images from 3 separate experiments. ***P < 0.001; ****P < 0.0001.

Figure 4. Bioinformatics analysis of the miRNA expression profile of human GMSC–derived EVs.

(A) Flowchart illustrates the experimental procedures for removal of proteins or RNAs in GMSC-EVs. (B) Silver staining of polyacrylamide gel showed the protein profile GMSC-EVs upon different treatment procedures described in Methods. (C) The image of agarose gel showed the RNA profile GMSC-EVs upon different treatment procedures described in Methods. (D and E) In vitro suppressive assay of cytokine production. (F) The heatmap shows the miRNA expression profile of GMSC-EVs. (G) Volcano plot shows differentially expressed miRNAs. P < 0.05 and fold change ≥ 2 was considered statistically significant. (H) The pathway enrichment of the differentially expressed miRNAs was performed in online database DIANA-miRPath v.3. The x axis represents –log₁₀(P value); the y axis represents KEGG term; P < 0.05 was considered statistically significant. (I) The predicted miRNAs to regulate IKKB from different database TargetScan, miRWalk, and miRDB. (J) The miR-148a-3p level in GMSC-EVs were measured by qPCR. (K) Heatmap of the differentially expressed genes in RA-related publicly available data set GSE56649 (13 cases of RA and 9 healthy controls). Statistical significance was assessed by 1-way ANOVA with Dunnett multiple-comparison test in E and by 2-tailed Student’s t test in J. Data are shown as the means ± SD from 1 of 3 independent experiments. *P < 0.05; ***P < 0.001; ****P < 0.0001.

Figure 5. Blockage of miR-148a-3p in human GMSC–derived EVs disturbs the immunoregulatory properties.

(A) In vitro suppressive assay of T cell proliferation. (B and C) In vitro Th17-polarizing and Treg-polarizing assays. (D and E) Representative images of osteoclast generation under different conditions. TRAP⁺ osteoclast numbers of per area under different conditions were quantified. (F and G) In vitro suppressive assay of cytokine production. (H) qPCR for inflammation or tolerance phenotype of CD3⁺ T cells. (I–K) CIA mice received a single type of NC-GMSC-EVs or si-GMSC-EVs at days 0, 15, and 30 after immunization, and individual analysis was acquired at the endpoint of the experiment (day 60 after immunization). Scale bars = 500 μm. (I) Knee joint sections were stained with H&E and toluidine blue staining, and histopathologic scores were evaluated for features of synovitis, pannus, erosion, and cartilage matrix. (J) Toe joint sections were imaged with μ-CT, and bone volumes of the metatarsophalangeal joints were calculated. (K) Intracellular staining of IL-17A and Foxp3 in dLNs was detected by flow cytometry analysis. Statistical significance was assessed by 1-way ANOVA with Dunnett multiple-comparison test in A–G and I–K and by 2-tailed Student’s t test in H. (A–H) Data are shown as the means ± SD from 1 of 3 independent experiments. In I–K data are mean ± SD, n = 5-8 mice. **P < 0.01; ***P < 0.001; ****P < 0.0001.

Figure 6. miR-148a-3p–containing human GMSC–derived EVs modulate IKKB/NF-κB signaling pathway.

(A) Sequence alignment of miR-148a-3p and its putative target sites in the 3′ UTR of IKKB mRNA. Mutation was generated in the complementary sites for the seed region of miR-148a-3p, as indicated. (B) HEK293T cells were transiently cotransfected with IKKB WT or mutant 3′ UTR luciferase reporter plasmid and miR-148a-3p mimic for 48 hours, and luciferase activity was analyzed. (C and D) HEK293T cells were transiently transfected with negative control or miR-148a-3p mimic. Cells were collected at 48 hours, and the expression of IKKB or p-IKKB was detected by qPCR or Western blot respectively. (E and F) CD3⁺ T cells isolated from C57BL/6 mice were cocultured with NC-GMSC-EVs or si-GMSC-EVs under the activated condition. Cells were collected at 72 hours, and the expression of IKKB, p-IKKB, and p-NF-κB was detected by qPCR and Western blot. Statistical significance was assessed by 1-way ANOVA with Dunnett multiple-comparison test in C–F and by 2-tailed Student’s t test in B. Data are shown as the means ± SD from 1 of 3 independent experiments. *P < 0.05; **P < 0.01.

Figure 7. Effect of GMSC-derived EVs on xGvHD model in vivo.

(A) Schematic experimental setup for xGvHD. (B) Following the administration of DiR-labeled (red) EV injections to the xGvHD mice, digital photographs and IVIS images were used to present the major organs. (C) Quantification of fluorescence percentage of organs for B. (D–I) xGvHD mice received NC-GMSC-EVs or si-GMSC-EVs at days 0, 15, and 30. The weight (D), survival (E), and human CD3⁺ T cells in peripheral blood (F) of xGvHD mice were monitored from day 15 to day 60. (G) dLNs isolated from xGvHD mice at the 50th day were used to determine the human CD3⁺ percentage by flow cytometry analysis. (H) Liver, lung, and intestine of NOD/SCID mice collected at the 50th day were stained with H&E, and histopathologic severity scores were determined by lymphocyte invasion. Scale bars = 500 μm. (I) Sera were collected from blood of NOD/SCID mice at the 50th day, and the levels of TNF-α, IL-2, IFN-γ, IL-17A, IL-4, and IL-10 were detected by ELISA. B and C show representative in vivo tracking images from 3 separate experiments. Statistical significance was assessed by 1-way ANOVA with Dunnett multiple-comparison test in D and F–I and by log-rank test in E. In D–I data are mean ± SD, n = 10 mice. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Figure 8. GMSC-derived EVs protect against inflamed synovial fibroblast–mediated humanized animal model.

(A) Schematic experimental set-up for RASF-mediated humanized animal model. In the first operation, SCID mice were implanted with a cartilage-sponge complex under the left flank skin (primary implant). After 2 weeks, individual 5 × 10⁵ CM-DiI-labeled RASFs, 2 × 10⁶ GMSCs, and/or 100 μg GMSC-EVs were injected into the cartilage-sponge complex, and the implant was inserted into a s.c. space in the right flank skin (contralateral implant). (B and C) At day 60, the primarily and contralateral cartilages were collected, and the mean fluorescence intensity (MFI) of CM-DiI–labeled RASFs in primarily cartilages were quantified using ImageJ software to evaluate the invasiveness of contralateral RASFs after treatment with GMSCs or GMSC-EVs. Scale bars = 500 μm. (D and E) The contralateral and primary cartilages were collected and subjected to H&E staining to assess the invasiveness scores of inflammatory cells and the destruction of cartilages. The red arrows indicated the lesions of cartilage destruction caused by RASFs. Scale bars = 200 μm. Statistical significance was assessed by 1-way ANOVA with Dunnett multiple-comparison test in B–E. Data are mean ± SD, n = 5–6 mice. ****P < 0.0001.