Introduction

Rheumatoid arthritis (RA) is systemic autoimmune-mediated arthritis that is characterized by synovitis, which is capable of the permanent destruction of articular cartilage and bone1. Although the exact etiology is unknown, RA is considered a T cell-mediated disease in which various helper (CD4+) T cell subsets are implicated in the pathogenesis2,3. Among the various pathologic events occurring in inflamed joints, the irreversible destruction of joints is the most important clinical event because it is related to functional impairment and lower quality of life in RA patients. Increased osteoclastogenesis is the main cause of joint destruction, and RA patients have an increased risk of osteoporosis4. Therefore, RA treatment should target not only pathologic immune cell proliferation but also osteoclastogenesis suppression5. The current RA treatment guidelines recommend using conventional synthetic disease-modifying anti-rheumatic drugs (csDMARDs), biologic DMARDs (bDMARDs), and targeted synthetic DMARDs (tsDMARDs) to control pain and prevent structural damage in patients6,7. Although various DMARDs are used in RA treatment, unmet needs still exist because some RA patients do not achieve clinical responses, or considerable adverse events occur with the current treatment modalities8.

Mesenchymal stromal cells (MSCs) are multipotent stromal cells capable of differentiating into a variety of cell types, including osteoblasts, chondrocytes, and adipocytes under controlled culture conditions9,10. MSCs can be obtained from various tissues, and nasal turbinate is one of the most suitable tissues that contains MSCs11. MSCs are characterized by their plasticity and tropism and the expression of surface markers, such as CD73, CD90, and CD105, with the absence of typical hematopoietic markers, such as CD45 and CD11b9. MSCs are hypoimmunogenic because of their low expression of human leukocyte antigen (HLA) and co-stimulatory molecules. They also exert potent immunoregulatory actions by secreting cytokines and regulating immune cells, such as T cells and monocytes9. These properties suggest MSCs as a potential therapy for various inflammatory and autoimmune disorders. Indoleamine 2,3-dioxygenase (IDO) is the representative immunomodulatory enzyme regulating T cell proliferation, and MSCs produce IDO by interferon (IFN)-γ stimulation12. Several preclinical, and clinical studies showed the potential therapeutic role of MSCs in RA13 and clinical trials are ongoing. However, not all MSCs are effective for RA treatment because MSCs may have different immunomodulatory roles. Recent studies of MSC therapy have focused on overcoming donor variations in MSCs to enhance therapeutic effects14,15. Determining MSCs suitable for RA treatment and identifying potent MSC gene signatures for RA treatment may increase the efficacy and potential of MSC therapy for RA.

Most clinical trials using MSCs for RA have obtained them from adipose tissue16,17, bone marrow18,19, or umbilical cords20,21,22. Previously used MSCs had intrinsic limitations on acquisition (bone marrow [BM], adipose tissue), and the MSC yield was relatively small. In a previous study, we showed that inferior turbinate tissues removed during turbinate surgery, which is most commonly performed with ease under local anesthesia in the field of otolaryngology, could be used as a source of MSCs called human nasal turbinate-derived stem cells (hNTSCs)23. Surgical procedures on the inferior turbinate have a lower risk of adverse events, shorter recovery period, and less loss of blood than obtaining BM or adipose tissue biopsies, and the nasal turbinate possesses a higher density of MSCs than BM or adipose tissue24.

In the present study, we examined the therapeutic effect of hNTSCs in an RA mouse model. We also measured the engraftment and migration capacity of the hNTSCs and their role in suppressing T cell proliferation and osteoclastogenesis. Finally, we identified specific gene signatures of hNTSCs, which showed therapeutic effects in the RA mouse model.

Materials and methods

hNTSC culture

The study procedure utilizing hNTSCs was conducted in compliance with the Institutional Review Board of Seoul St. Mary’s Hospital (approval no. KIRB-00399_18-005), the Catholic University of Korea, informed consent regulations, and the Declaration of Helsinki. Before surgery, the participants provided written informed consent to participate in the study. hNTSCs were isolated from the discarded human patient nasal inferior turbinate tissue of 20 healthy controls (HCs) (Fig. 1A) who underwent partial turbinectomy, as previously described23. The tissue was washed with a saline solution and phosphate-buffered saline (PBS, Thermo Fisher Scientific, Waltham, MA, USA) and cut into small pieces, which were then plated in a culture dish and covered with a sterilized glass cover slide. The tissue was incubated at 37 °C in a humidified atmosphere containing 5% (v/v) CO2 in α-minimum essential medium (α-MEM, Thermo Fisher Scientific) supplemented with 1% (v/v) penicillin/streptomycin (antibiotics, Invitrogen) and 10% (v/v) fetal bovine serum (FBS, Thermo Fisher Scientific). The culture medium was changed every 2 days during 3 weeks. The glass cover slide was removed, and cells isolated from the tissue were harvested using 0.25% trypsin in 1 mM EDTA solution. Human bone marrow-derived stem cells (hBMSCs) were provided by The Catholic Institute of Cell Therapy (Catholic MASTER Cells, The Catholic University of Korea). The hBMSCs were isolated from healthy donors after approval by the Institutional Review Board of Seoul St. Mary’s Hospital (approval no. KC10CSSE0651). All stem cells were expanded in 6 passages for use in the experiments. Cells (9 × 103) were seeded in 24-well culture plates to analyze cell growth and measured for 4 days using an EZ-Cytox assay kit (DAEILLAB Co., Seoul, Korea, http://www.daeillab.co.kr). Absorbance at a wavelength of 450 nm was measured using a microplate reader (MolecularDevices Corporation, Sunnyvale, CA, USA). The other methodological processes are described in the Supplementary Methods section.

Fig. 1
figure 1

Highly proliferative, GRO-α and OPG-producing hNTSCs are multipotent. (A) Images of human nasal inferior turbinate tissue obtained from turbinectomy. (B) Microscopic images of hNTSCs in culture. Scale bar: 200 μm. (C) Images of hNTSCs stained with alizarin red S on day 21 of incubation in osteogenic differentiation medium (left) and oil red O on day 14 of incubation in adipogenic differentiation medium (right). Scale bars: 50 μm. (D) Growth rate of cultured hNTSCs and hBMSCs for 4 days after plating. Each bar represents relative cell growth (mean ± SD). ***P < 0.001. (E) The cytokine expression of hNTSCs and hBMSCs in culture. Cytokine secretion profiles were obtained by incubating the array membrane with hNTSC and hBMSC culture medium. (F) Quantification of GRO-α, osteoprotegerin (OPG), and angiogenin optical-scale levels. Bars represent relative spot density. The intensity of each cytokine spot in the array membrane was measured based on gray-scale intensity (mean ± SD). *P < 0.05. (G) Flow cytometric analysis of cell surface markers using anti-CD34, anti-CD44, anti-CD73, anti-CD90, and anti-CD105 antibodies in hNTSCs and hBMSCs. (H) Confocal microscopy images of cultured hNTSCs and hBMSCs stained for CD44, CD73, CD90, and CD105. Nuclei were labeled with DAPI. Scale bar: 50 μm. (I) Western blots of SDS−PAGE gels of proteins extracted from hNTSCs and hBMSCs using primary anti-CD44 and anti-105 antibodies. β-actin was used as a loading control. All images and data are representative of two or three independent experiments.

Full size image

Immunocytofluorescence staining

The expression of CD44 (1:200, FITC-conjugated, BD Biosciences, San Jose, CA, USA, 555478), CD73 (1:200, PE-conjugated, BD Biosciences, 550257), CD90 (1:200, PE-conjugated, BD Biosciences, 555596), and CD105 (1:200, PE-conjugated, BD Biosciences, 560839) in cultured hNTSCs and hBMSCs was determined by immunofluorescence staining. The cells were fixed with 4% (w/v) PFA and blocked with 1% (w/v) normal goat serum (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) and then incubated with CD44, CD73, CD90, and CD105 antibodies. The nuclei were labeled with DAPI (1:1000, Sigma-Aldrich), and fluorescence was observed using a Zeiss LSM510 confocal microscope (Carl Zeiss).

Flow cytometry for hNTSCs and hBMSCs

Cell suspensions prepared from hNTSCs and hBMSCs were prepared and then incubated with CD34 (1:100, PE-conjugated, BD Biosciences, 555822), CD44 (1:100, FITC-conjugated, BD Biosciences, 555478), CD73 (1:100, PE-conjugated, BD Biosciences, 550257), CD90 (1:100, PE-conjugated, BD Biosciences, 555596), and CD105 (1:100, PE-conjugated, BD Biosciences, 560839) antibodies for 60 min at 4 °C. The cells were resuspended in fluorescence-activated cell sorting (FACS) buffer and acquired through FACS Canto II (BD Biosciences) with DIVA software.

Western blots

The cells were lysed in RIPA buffer (Thermo Fisher Scientific) containing protease inhibitors (GenDEPOT Inc., Barker, TX, USA), centrifuged at 20,000g for 20 min at 4 °C, and the supernatant was collected. For western blots of CD44 and CD105 from the cells, protein samples were loaded onto NuPAGE 4–12% (w/v) and 12% (w/v) Bis- Tris Protein Gels (Thermo Fisher Scientific) and transferred to a polyvinylidene difluoride (PVDF) membrane (Roche, Mannheim, Germany). The membrane was incubated with primary anti−CD44 (1:500, Abcam, Cambridge, U.K., ab157107), anti-CD105 (1:500, BD Biosciences, 555690), and anti-β-actin (1:1000, Santa Cruz Biotechnology, SC47778) antibodies. The membranes were incubated with horseradish peroxidase-conjugated secondary antibodies and developed using enhanced chemiluminescence detection reagents (Thermo Fisher Scientific).

Animals

All animal research procedures were performed in accordance with the Laboratory Animals Welfare Act, the Guide for the Care and Use of Laboratory Animals, and the Guidelines and Policies for Rodent Experiments provided by the Institutional Animal Care and Use Committee (IACUC) in the School of Medicine, The Catholic University of Korea. (approval number: CUMS-2019-0201-05). The IACUC and Department of Laboratory Animals (DOLAs) at The Catholic University of Korea, Songeui Campus, were accredited by the Korea Excellence Animal Laboratory Facility from the Korea Food and Drug Administration in 2017 and acquired AAALAC International full accreditation in 2018. All in vivo animal experiments were performed in accordance with the ARRIVE guidelines. Male DBA/1J mice were purchased from Orient Biotechnology. The animals were maintained under specific pathogen-free conditions.

Induction of collagen-induced arthritis (CIA) and animal treatment

Type II collagen (CII) was dissolved in 0.1 N acetic acid (4 mg/mL) with gentle rotation overnight at 4 °C. Eight-week-old male DBA/1J mice, 8 weeks old, were immunized intradermally at the base of the tail with 100 µg chicken type II collagen (Chondrex) in complete Freund’s adjuvant (CFA, Chondrex). Two weeks later, the mice received intradermal tail boosts with 100 µg of CII emulsified in incomplete Freund’s adjuvant (IFA, Chondrexc)25. After assessing serum anti-type II collagen antibodies (1:1000 dilution) at 3 weeks after primary immunization, the mice were divided into groups and injected intravenously with hNTSCs (1 × 106 cells/100µL/mouse) or vehicle control (PBS). For the positive control group, mice were administered methotrexate hydrate (MTX, Sigma, 3 mg/kg) 3 times a week intraperitoneally. Arthritis severity was monitored in all mice for 8–9 weeks. Then, mice were euthanized by CO2 gas inhalation method for histological study.

Assessment of arthritis

Arthritis scores were determined by 3 independent observers. The mice were examined thrice weekly for the development and severity of joint inflammation. Arthritis was scored on a scale from 0 to 4 per limb (total 0 to 16) where 0 indicated no evidence of erythema and swelling, 1 was erythema and mild swelling confined to the mid-foot (tarsals) or ankle joint, 2 was erythema and mild swelling extending from the ankle to the mid-foot, 3 was erythema and moderate swelling extending from the ankle to the metatarsal joints, and 4 was erythema and severe swelling encompassing the ankle, foot, and digits. The mean arthritis index score was used to compare the data between the control and experimental groups. The arthritis incidence was calculated by assigning 25–100%, respectively, when the arthritis scores of 1–4 legs were more than 1 point each.

Analysis of anti-type II collagen antibodies

Blood was drawn from the orbital sinuses of the mice, and sera were stored at − 20 °C until use. Chicken CII (4 µg/mL in Tris-buffered saline) was coated in 96-well-microplates. After blocking, the serum samples were diluted 1:5000 (IgG1 detection at the end of CIA), 1:1000 (IgG2a detection for grouping at 3rd week of CIA), or 1:20,000 (IgG2a detection at the end of CIA) in Tris-buffered saline (pH 8.0) containing 1% bovine serum albumin and 0.5% Tween-20, and incubated in the microtiter plates for 1 h, after which the plates were washed 5 times. The cells were reacted with horseradish peroxidase (HRP)-conjugated anti-mouse IgG1 or IgG2a antibodies (Bethyl Laboratories), and the color was developed using TMB substrates (Thermo Fisher Scientific). Absorbance values were determined with ELISA microplate reader software (Gen 5, BioTek) at 450 nm.

Joint histology

Joint tissues of CIA mice were fixed in formalin (Sigma), decalcified in Calci-Clear Rapid Bone Decalcifier (National Diagnostics), and embedded in paraffin. Tissue sections with 7-µm thickness were prepared and stained with hematoxylin and eosin (H&E).

Micro-computed tomography (CT)

A femur (subchondral bone) and tarsal joint (talus bone) were scanned by SKYSCAN1272 ex-vivo micro-CT (Bruker microCT). The measurement conditions were as follows: X-ray source: 60 kV, 166 µA, AI 0.25 mm filter, rotation step 0.6°; and resolution: 10 μm pixels. Measurement and analysis of the scanned images were conducted by the 2nd Analysis Lab.

Engraftment of hNTSCs

Cultured hNTSCs from 2 donors and human normal dermal fibroblasts (hNDFs) (passage 5) were stained with DiR dye (DilC18(7), 1,1’-dioctadecyl-3,3,3’,3’-tetramethylindotricarbocyanine iodide, Invitrogen) according to the manufacturer’s instructions. hNTSCs or hNDFs carrying DiR (1 × 106 cells/10µL in saline) or DiR dye-only (0.5 mg/10µL in saline) were injected intraarticularly into the ankle joint of CIA mice at 8 weeks after primary immunization. We obtained serial images of the ankle joint by monitoring with the IVIS Lumina XRMS (PerkinElmer) the day after injection and weekly until no DiR signal was observed. Epi-fluorescence was measured under a specific range of 1.00 ~ 6.00 × 109 [p/s/cm2/sr]/[ µW/cm2].

Osteoclastogenesis

We obtained bone marrow cells from mouse femurs and tibias to induce osteoclastogenesis. Cells (1.6 × 106 cells/mL/well) were primarily seeded onto 24-well plates in 10% FBS (GW Vitek) and 1% penicillin-streptomycin (Gibco) containing α-MEM (Invitrogen). The suspended mouse bone marrow cells were then stimulated with 100 ng/mL of recombinant macrophage colony-stimulating factor (M-CSF, Peprotech). After 3 days, the adherent cells were harvested and further cultured with 50 ng/mL of M-CSF and 100 ng/mL of soluble receptor activator of NF-κB ligand (RANKL, Peprotech) in the absence or the presence of 1.25 × 104 hNTSCs/well pretreated with 10 µg/mL of mitomycin C (Sigma) for 4 h. We prepared 72 h culture supernatants from hNTSCs or hBMSCs seeded at a concentration of 1 × 105 cells/20 mL/75 cm2 culture flask in advance. hNTSC culture supernatant at 1 ~ 5% was added to mouse osteoclastogenesis cultures. The medium was replaced with fresh medium containing M-CSF, RANKL, and/or 1 ~ 5% hNTSC culture supernatant every 3 ~ 4 days for 2 ~ 3 weeks. Osteoprotegerin (OPG) concentrations were measured in culture supernatant collected for the last 3 days. The cells were fixed and stained for tartrate-resistant acid phosphatase and alkaline phosphatase 2 (TRAP & ALP) using a staining kit (Takara) according to the manufacturer’s instructions, omitting counterstaining with ALP and methyl green. The TRAP-positive multinucleated cells were counted 3 times by an examiner blinded to the previous osteoclast counts.

Statistics

Statistical analyses were performed using GraphPad Prism (Version 8, GraphPad, CA, USA). Statistical significance was determined by two-way analysis of variance (ANOVA) with Sidak’s multiple comparisons test for the relative growth rate (Fig. 1D), unpaired t-test for cytokine arrays (Fig. 1F); micro-CT analysis (Fig. 3D), migration assay (Fig. 4C), and two-way ANOVA with Turkey’s multiple comparisons test for arthritis scores in the collagen-induced arthritis (CIA) experiment and the incidence of CIA (Fig. 3A). The Mann-Whitney test was used for anti-CII IgG1/IgG2a ELISA (Fig. 3B) and the paired t-test was used for chemokine receptor qPCR analysis (Fig. 4B) and IDO qPCR analysis (Fig. 5B). One-way ANOVA with Turkey’s multiple comparison test was used to analyze T cell suppression (Fig. 5A,C), and TRAP cell counts (supplementary Fig. 2C ). P-values of less than 0.05 were considered significant.

Results

Biological characteristics of hNTSCs

Prior to evaluating the therapeutic effects of hNTSCs in RA, we investigated the characteristics of hNTSCs after 5 in vitro culture passages. Microscopic images showed that hNTSCs cultured in growth medium exhibited a fibroblast-like morphology, and some of them showed narrow, spindle-shaped morphology (Fig. 1B). The cells were induced to differentiate into osteogenic and adipogenic lineages by incubating cells in osteogenic and adipogenic differentiation medium to assess the multipotency of the hNTSCs. The expression of specific lineage phenotypes was analyzed by immunostaining. hNTSCs cultured in osteogenic differentiation medium showed high alizarin red S staining on day 21 of induction. Adipogenic differentiation was demonstrated by oil red O staining of hNTSCs cultured in adipogenic differentiation medium for 14 days (Fig. 1C). The growth of hNTSCs was examined by an EZ-Cytox assay throughout the 4-day culture in vitro and compared with that of hBMSCs, which are the most common stem cell type used in stem cell therapy. The hNTSCs showed a faster cell growth rate than hBMSCs during the 4-day culture period (Fig. 1D). We analyzed the basic cytokines secreted by hNTSCs or hBMSCs using human cytokine array kits consisting of 80 different cytokine and chemokine antibodies. The array data featured predominant hybridization signals for a number of cytokines, which are known as mesenchymal stem cells-related cytokines, such as interleukin (IL)-6, IL-8, monocyte chemoattractant protein-1 (MCP-1), tissue inhibitor of metalloproteinase-2 (TIMP-2), osteoprotegerin (OPG), and angiogenin (Fig. 1E). All spots in the array membrane were quantified by averaging 3 different membranes in each stem cell. The intensity values revealed a significantly higher level of cytokines and chemokines, such as growth-related oncogene-α (GRO-α, also known as C-X-C motif chemokine ligand 1, CXCL1), OPG, and angiogenin, in hNTSCs compared with hBMSCs (Fig. 1F). hNTSCs showed higher proliferative capacity than hBMSCs, production of GRO-α, OPG, and angiogenin, and multipotency to differentiate into osteogenic and adipogenic lineages. Cell surface markers of hNTSCs and hBMSCs were examined by flow cytometric analysis to measure the expression of several mesenchymal stem cell markers. Both hNTSCs and hBMSCs were negative for hematopoietic stem cell marker CD34 and positive for MSC markers, CD44, CD73, CD90, and CD105 (Fig. 1G). Moreover, immunofluorescence staining for the cell surface markers showed higher levels of CD44, CD73, CD90, and CD105 in hNTSCs and hBMSCs (Fig. 1H). Western blots of SDS-PAGE gels demonstrated similar expression levels of MSC markers in hNTSCs and hBMSCs (Fig. 1I, Supplementary Fig. 1).

Cluster analysis of hNTSCs by scRNA-seq

We obtained non-normalized raw data set was obtained on 33,538 genes, 23,646 hNTSCs from 4 donors. After filtration of data, we obtained a feature-barcode matrix with 21,302 genes, 23,456 hNTSCs. cDNA libraries were prepared using the chromium controller according to the 10× Single Cell 3′ v3 protocol (10× Genomics) (Fig. 2A). SC gene expression data were analyzed using Cell Ranger v3.1.0 (10× Genomics) and Seurat v3 standard integration workflow (CCA & anchor). To correct for the effects of stem cell cycle heterogeneity, scaling and regressing out cell cycle effects was performed using Seurat Cell-Cycle Scoring and Regression workflow (Fig. 2B). Based on Seurat batch effect correction by the CCA method and cell cycle regression out by alternative workflow, cell aggregates displayed on a uniform manifold approximation and projection (UMAP) revealed 11 molecularly distinct clusters (Fig. 2C). The cell cluster (C) frequency was as follows: C0, 21.5%; C1, 18.2%; C2, 17.2%; C3, 13.4%; C4, 10.1%; C5, 7.5%; C6, 5.0%; C7, 3.2%; C8, 2.1%; C9, 1.1%; and C10, 0.7%.

Fig. 2
figure 2

scRNA-seq reveals 11 clusters of hNTSCs. (A) Overview of the experimental procedure. Human NTSCs were subjected to single-cell suspension, library construction, and sequencing. (B) Workflow for data analysis using CellRanger pipeline and Seurat analysis. (C) Raw data (n = 23,456 cells) were clustered and UMAP projection of 11 clusters of 4 donor-derived hNTSCs was performed. (D) Representative mesenchymal stem cell marker expression of 4 donor-derived hNTSC aggregates. x axis: cell clusters C0 to 10, y axis: surface markers (E) Functional annotation analysis for significant probes was performed using g: Profiler.

Full size image

We analyzed all clusters of hNTSCs showing stem cell-like phenotypes by expressing CD13, CD29, CD44, CD73, CD81, CD90, CD105, and CD166, while not expressing CD18, CD31, CD34, and CD45 (Fig. 2D). In functional annotation analysis, hNTSCs were enriched in cell cycle/cell division (C0, C1, C8), cell/tissue development (C2, C5, C9), extracellular matrix organization (C5), cell population proliferation (C3), response to chemicals/organic substrates (C2, C3, C9), RNA/mRNA regulation (C8), DNA repair/replication/DNA metabolic process (C4), protein localization (C6, C7), sequestering of iron ion (C6), and cellular response to stress (C10) (Fig. 2E). In the scRNA sequencing assay, hNTSCs showed a stem cell-like phenotype with high proliferative potential.

hNTSCs suppress arthritis in the CIA model

CIA mice were treated with vehicle (negative control), methotrexate (MTX, positive control), or hNTSCs, and DBA/1J mice were not immunized with type II collagen. The arthritis score was significantly suppressed in both MTX and hNTSC-treated groups compared to the vehicle-treated group, and the incidence of arthritis was only lowered in the hNTSC-treated group (Fig. 3A). Serum levels of anti-CII IgG1 and IgG2a were significantly lower in hNTSC-treated mice than in vehicle-treated mice (Fig. 3B). Representative H&E staining of ankle joints is presented in Fig. 3C, and joint destruction was less severe in hNTSC- or MTX-treated mice than in vehicle-treated mice. hNTSC-treated mice showed tendency for higher bone volume with less bone surface (less erosion) than the vehicle group in micro-CT of talus bones, but these were non-significant (Fig. 3D). These were similarly observed in micro-CT findings of subchondral femur bones (Fig. 3E).

Fig. 3
figure 3

hNTSCs attenuate the development of arthritis in mice. (A) Arthritic scores and incidence in vehicle (PBS, n = 12), hNTSCs (1 × 106 cells/0.1 mL/mouse, n = 12) and methotrexate hydrate (MTX, 3 mg/kg, n = 12)-treated CIA mice. Mice were euthanized 8 weeks after primary immunization. (B) Serum levels of anti-CII-specific IgG1 and IgG2a were determined by ELISA. (C) Joint histology was assessed by staining with hematoxylin & eosin (original magnification, x40). (D) Micro-CT images of talus bone from vehicle- or hNTSCs-treated mice. The scatter dot plot shows the talus volume and surface. (E) Micro-CT images of subchondral femur bone from vehicle- or hNTSC-treated mice. Data are shown as the mean ± SD. ns not significant; *P < 0.05; **P < 0.01; ***P < 0.001.

Full size image

Engraftment and migration ability of hNTSCs

We selected two hNTSCs, hNTSCs#10 and hNTSCs#433, which showed excellent therapeutic effects in CIA mice to evaluate engraftment and migration capacity. hNDFs and DiR dye were used to compare engraftment duration. DiR dye-stained hNTSCs, hNDFs, and DiR dye-only were injected into the ankle joints of CIA mice 8 weeks after the first immunization against type II collagen. The DiR dye-stained hNTSCs showed longer maintenance of engraftment (hNTSCs#10 for 8 weeks and hNTSCs#433 for 12 weeks) than DiR dye-stained hNDFs, but it was maintained for only 3 weeks (Fig. 4A). Next, we stimulated hNTSCs with IL-1β and TNF-α to measure chemokine receptor expression under inflammatory conditions. The expression levels of CXCR1, CXCR4, and CXCR7 were increased by IL-1β stimulation compared to those in the unstimulated condition. The up-regulation of chemokine receptor-related genes (CXCR1, CXCR4, CCR2B, CCR8, and XCR1) was also detected in the TNF-α stimulated condition (Fig. 4B). Then, we evaluated the migration capacity of hNTSCs by adding ligands for chemokine receptors, SDF-1 (ligand for CXCR4), IL-8 (ligand for CXCR1 and 2), and IP10 (ligand for CXCR3). The migration capacity significantly increased under SDF-1-added conditions (Fig. 4C).

Fig. 4
figure 4

Engraftment of hNTSCs in inflamed joint tissues. (A) DiR dye-labeled hNTSCs (#10, #433), hNDFs (normal human dermal fibroblasts) (1 × 106 cells/site), or DiR dye-only was injected into the ankle joint of mice at 8 weeks after primary immunization. DiR signal was measured weekly by optical in vivo imaging until no more signal was detected. (B) mRNA levels of various chemokine receptors in hNTSCs. hNTSCs were stimulated by medium alone or 10 ng/mL of IL-1β or TNF-α for 72 h. The mRNA levels of chemokine receptors were determined by real-time PCR. The numbers represent the relative ratio of mRNA levels in hNTSCs stimulated by IL-1β or TNF-α. (C) hNTSCs migrated in vitro according to the chemokine gradient. A Boyden chamber assay was performed to assess the migration capacity of hNTSCs stimulated with SDF-1, IL-8, or IP-10, as described in the Materials and Methods section. The chemokine concentration was 150 ng/mL. Cells were incubated for 6 h, fixed, and stained with DiffQuik. Migrated NTSCs were analyzed by pixels using ImageJ software. Scale bar: 200 μm. The data are expressed as the mean ± SD. **P < 0.01.

Full size image

Suppressive effect of hNTSCs on CD4+ CD25− T cell proliferation and osteoclastogenesis

CD4+ CD25− T cells were harvested from healthy donors and then cultured under T cell proliferative conditions (anti-CD3 and anti-CD28-stimulating conditions) with hNTSCs or hBMSCs. CD4+CD25− T cell proliferation was suppressed by both hNTSCs and hBMSCs at 1:2 dilution (Fig. 5A). IDO gene expression was significantly increased in hNTSCs by IFN-γ (20 ng/mL) stimulation compared to non-stimulated cells (Fig. 5B). We compared CD4+ CD25− T cell proliferation under hNTSC-added conditions with or without an IDO inhibitor (1-MT, 100 µM) to determine whether hNTSCs suppressed CD4+ CD25− T cell proliferation via IDO. The proliferation of CD4+ CD25− T cells was suppressed by hNTSCs, which was attenuated by adding 1-MT (Fig. 5C), suggesting that IDO produced by hNTSCs was responsible for T cell suppression. We used mice tibia-derived monocytes to evaluate the role of hNTSCs and hNTSC-derived enzymes on osteoclastogenesis. The TRAP+ multinucleated osteoclast count decreased when hNTSCs and hNTSC culture media were added (Supplementary Fig. 2A,B). The suppression of osteoclastogenesis by the addition of hNTSCs or hNTSC culture media was prevented by adding anti-OPG antibody (Supplementary Fig. 2C).

Fig. 5
figure 5

hNTSCs inhibit human CD4 + T cell proliferationin vitro. (A) Human CD4 + CD25- responder T cells (5 × 104 cells) from two healthy controls were labeled with CellTrace Violet (CTV) and cultured with/without BM-MSC or hNTSCs (5:1, 2:1 ratio) in the presence of human CD3/CD28 T activator Dynabeads for 4 days. CTV dye was analyzed by flow cytometry. The representative results are shown in the upper panel. The percentage of proliferating CD4 + CD25- T cells is shown in the lower panel. Data are expressed as the mean ± SD. *P < 0.05. (B) mRNA levels of IDO were greatly increased by treatment with IFN-γ. hNTSCs (1 × 105 cells/well) were cultured with medium alone or IFN-γ (20 ng/mL) for 24 h. The mRNA levels of IDO were measured by real-time PCR. The data are expressed as the mean ± SD. ***P < 0.001. (C) CTV dye-labeled human CD4 + CD25- responder T cells (5 × 104 cells) were cultured with or without hNTSCs (2.5 × 104 cells) and/or human CD3/CD28 T activator Dynabeads and/or 100 µM of 1-MT for 6 days. CTV dye was analyzed by flow cytometry. Representative results are shown in the upper panel. The percentage of CD4 + CD25- responder T cells that divided more than 4 times is shown in the right panel. The data are expressed as the mean ± SD. ***P < 0.001.

Full size image

Gene expression signature of hNTSCs with anti-arthritic effects

We subdivided hNTSCs into two groups: (1) group A, which did not have anti-arthritic effects in the CIA model (N = 11), and (2) group B, which showed anti-arthritic effects in the CIA model (N = 7). The baseline characteristics of donors of group A and B did not show significant difference. The heatmap analysis and microarray DEG analysis revealed that group B hNTSCs showed up-regulated genes (SSTR1, KRTAP1-5, HAS2, CXCL1, and CPED1) and down-regulated genes (GSTT2B, MGP, OR10G2, DCLK1, ADAMTS10, SULF2, CFD, C4B, RBP1, and CCN5) compared to the gene signature of group A hNTSCs (Fig. 6A,B). Then, we performed RT-qPCR to validate the DEG between group A and B hNTSCs. Similar to previous results (Fig. 6A,B), several genes, including HAS2, KRTAP1-5, and CXCL1, were up-regulated in group B hNTSCs, whereas GSTT2B and C4B were relatively down-regulated in group B hNTSCs (Fig. 6C).

Fig. 6
figure 6

Identification and validation of specific gene signatures that could discriminate hNTSCs with superior efficacy. (A) Heatmap analysis revealed 15 differential expressed genes in hNTSCs by immunomodulatory capacity. Group A: 11 donor-derived hNTSCs with no efficacy against CIA; Group B: 7 donor-derived hNTSCs with significant suppression of CIA. (B) Differential gene expression levels between Group A_hNTSCs and Group B_hNTSCs. Featured genes are marked. (C) Validation of featured gene expression by the Taqman qPCR method. Sixteen donor-derived hNTSCs were tested in duplicate (mean ± SD). *P < 0.05; **P < 0.01.

Full size image

Discussion

In the present study, we used hNTSCs in an RA mouse model and demonstrated the anti-arthritic effects of hNTSCs. hNTSCs showed higher cell proliferation potential than hBMSCs and a longer engraftment period than hNDFs. hNTSCs also reduced effector T cell proliferation by IDO production and attenuated osteoclastogenesis by inducing OPG. The most remarkable strength of the present study is that a beneficial RA gene signature of hNTSCs was identified by microarray DEG and validated by RT-qPCR. hNTSCs showed high expressions of functional annotation for cell proliferation and tissue repair processes. The hNTSCs that potentiated RA treatment showed specific gene signatures: (1) the increased genes were KRTAP1-5, HAS2, and CXCL1, and (2) the decreased genes were GSTT2B and C4B. These results could support and guide the successful development of novel RA therapy using MSCs.

hNTSCs attenuated arthritis severity in the RA mouse model (Fig. 3A). Previous studies also showed the therapeutic effects of MSCs in preclinical experiments in RA animal models26,27,28,29,30,31,32,33,34. These previous studies used MSCs from various tissues, including umbilical cords27,29,32,33,34, BM31, adipose tissue26,30, and embryonic tissue28. The present study was the first preclinical study to demonstrate the therapeutic effects of nasal turbinate-derived MSCs in an RA animal model. The proliferative capacity and secretory function of CXCL1 and OPG were higher in hNTSCs than in hBMSCs (Fig. 1D–F). hNTSCs suppressed osteoclastogenesis by inducing OPG (Fig. 6). Increased osteoclastogenesis results in more pronounced joint destruction and an increased risk of osteoporosis in RA, and these harmful consequences of RA (joint destruction and osteoporosis) reduce the quality of life and restrict the daily activities and working ability of patients1,4. Therefore, suppressing osteoclastogenesis is an important treatment goal in RA. Garimella et al. previously showed the suppressive role of MSCs on osteoclastogenesis, which was mediated by reductions in pro-inflammatory cytokine production26. In the present study, we revealed that hNTSCs could reduce osteoclastogenesis and found another mechanism, which was mediated by OPG induction by hNTSCs.

Stem cells can be divided into two types: (1) embryonic stem cells and (2) adult stem cells. Because embryonic stem cells have ethical issues, MSCs from various sources have been identified and tried in clinical trials. MSCs were first used for regenerative medicine and then tried in various autoimmune and inflammatory diseases after the discovery of the immunomodulatory effects of MSCs9. Approximately 66% of RA patients fail to achieve remission or low disease activity by the most widely used treatments (bDMARD and TNF inhibitors)35. Thus, unmet needs for RA treatment still exist1,36. Several phase I and II studies have been conducted based on the immunomodulatory functions of MSCs13. Three studies showed that combination therapy with MSCs and csDMARDs significantly reduced disease activity in RA patients. However, appropriate control groups were not included19,20,21,22. These studies also evaluated additional therapeutic effects of MSCs in the context of keeping underlying csDMARD therapy19,20,22, but did not prove the effects of MSC therapy alone in RA. Another study demonstrated that American College of Rheumatology (ACR) 20 response achievement tended to be higher in RA patients treated with adipose tissue-derived MSCs, but clinical efficacy was non-significant compared to the control group17. Therefore, the clinical efficacy of MSCs in RA treatment is still unproven, and efforts to improve the immunomodulatory and anti-inflammatory effects of MSCs have been tried13. Discriminating effective MSCs and their gene signatures, then engineering MSCs toward specific gene signatures mimicking effective MSCs may increase their therapeutic potential in RA37. In the present study, we identified hNTSCs with better therapeutic potential in an RA mouse model and revealed specific gene signatures of these hNTSCs. Microarrays were used to reveal up-regulated genes (KRTAP1-5, HAS2, and CXCL1) and down-regulated genes (GSTT2B and C4B) and these were validated by RT qPCR. CXCL1 induced cyclooxygenase-2 in RA synovial fibroblasts via NF-κB signal38. C4B deficiency was more frequently observed in seropositive RA patients than in non-RA controls39. HAS2 over-expression in stromal cell reduced RANKL expression40. Association between these gene signatures and RA pathogenesis is unclear, yet. Although we could not unravel the mechanism of how these hNTSC gene signatures showed better therapeutic effects, the present findings may lead to engineering and modification of MSCs to become more efficient for RA treatment. Small sample size of RT-qPCR is limitation of present study, and further studies are needed to validate the gene signature of efficient hNTSCs for RA treatment and to investigate the association between specific hNTSC gene signatures and their therapeutic potential in RA.

The immunomodulatory and anti-inflammatory functions of MSCs are well-known, and MSCs showed regulatory roles in effector T cell and regulatory T cell (Treg) proliferation by cell-to-cell contact or paracrine activity9. The CD4+CD25+Foxp3+ Treg population was increased in RA patients by umbilical cord MSC therapy20. The down-regulation of type 17 helper T cells, activated CD4+ T cells, and the up-regulation of Treg cells were observed in several RA animal model experiments26,28,30,32,34. IDO is a representative immunomodulatory molecule produced by MSCs and is inducible by IFN-γ stimulation9,13. Previous studies demonstrated that the induction of IDO by MSCs reduced arthritis severity in RA animal models27,28. We also demonstrated the suppression of effector helper T cells (CD4+CD25− T cells) by hNTSCs, which was mediated by IDO induced by hNTSCs.

Conclusions

In conclusion, we revealed that hNTSCs ameliorated arthritis in CIA mice, suppressed effector helper T cell proliferation induced by IDO, and reduced osteoclastogenesis via OPG production. Specific gene signatures (increased KRTAP1-5, HAS2, and CXCL1, decreased GSTT2B and C4B) were present in hNTSCs with anti-arthritic effects. Our findings suggest hNTSCs for RA therapy, and more improved MSC therapy may be developed by modifying and engineering MSCs by considering the gene signatures of hNTSCs with anti-arthritic effects.