MicroRNA-663 induces immune dysregulation by inhibiting TGF-β1 production in bone marrow-derived mesenchymal stem cells in patients with systemic lupus erythematosus

Abstract

Mesenchymal stem cells (MSCs) are critical for immune regulation. Although several microRNAs (miRNAs) have been shown to participate in autoimmune pathogenesis by affecting lymphocyte development and function, the roles of miRNAs in MSC dysfunction in autoimmune diseases remain unclear. Here, we show that patients with systemic lupus erythematosus (SLE) display a unique miRNA signature in bone marrow-derived MSCs (BMSCs) compared with normal controls, among which miR-663 is closely associated with SLE disease activity. MiR-663 inhibits the proliferation and migration of BMSCs and impairs BMSC-mediated downregulation of follicular T helper (T_fh) cells and upregulation of regulatory T (T_reg) cells by targeting transforming growth factor β1 (TGF-β1). MiR-663 overexpression weakens the therapeutic effect of BMSCs, while miR-663 inhibition improves the remission of lupus disease in MRL/lpr mice. Thus, miR-663 is a key mediator of SLE BMSC regulation and may serve as a new therapeutic target for the treatment of lupus.

Introduction

Mesenchymal stem cells (MSCs), originally isolated from the bone marrow stroma, exhibit potent immunoregulatory function by inhibiting the proliferation and function of major immune cells, including T and B lymphocytes, natural killer cells and dendritic cells.¹ MSCs have been shown to have great potential for clinical applications in the treatment of immune disorders. Systemic lupus erythematosus (SLE) is a prototypic autoimmune disease characterized by systemic autoantibody production and inflammatory cell infiltration in target organs.² Previously, we have shown that bone marrow-derived MSCs (BMSCs) from SLE patients exhibit abnormalities in immune modulation,^3,4,5,6,7 and allogeneic normal MSCs show therapeutic effects in both lupus mice and severe refractory SLE patients.^8,9,10,11,12 However, despite ample evidence for the therapeutic potential of MSCs in SLE, the mechanisms by which MSCs exert their immunomodulatory effects are incompletely understood.

MicroRNAs (miRNAs) are a novel class of endogenous, non-coding small RNAs of ~19–25 nucleotides in length and have been recognized as important negative modulators of genes in eukaryotic organisms. SLE patients have unique miRNA signatures in peripheral blood cells, body fluid and target tissues, such as kidney when compared with normal controls or patients with other diseases.^13,14,15 Several miRNAs, including miR-155, miR-146a and miR-126, have been shown to affect the functions of T and B cells, thereby modulating autoimmune pathogenesis.^16,17 Recently, several studies have identified critical roles for miRNAs in the proliferation, migration and differentiation of MSCs, suggesting that they might have an impact on the acquisition of reparative MSCs phenotypes.¹⁸ However, the roles of miRNAs in MSCs during autoimmune pathogenesis remain to be elucidated.

Follicular T helper (T_fh) cells aid effector B cells and augment autoimmunity.¹⁹ In contrast, regulatory T (T_reg) cells appear to inhibit B-cell responses by suppressing the activities of T_fh cells, and by doing so, prevent lupus development.^20,21 The imbalance between T cell subsets is important in autoimmune disease including SLE pathogenesis,²² but the underlying mechanisms remain unclear. In this study, using comparative miRNAs profilings in BMSCs, we identified miR-663 as a candidate miRNA for SLE pathogenesis. MiR-663 downregulated the secretion of transforming growth factor β1 (TGF-β1) by BMSCs, and consequently increased the percentage of T_fh cells but decreased the frequency of T_reg cells in vitro. Importantly, the therapeutic effect of BMSC transplantation in MRL/lpr mice was substantially improved and impaired after the inhibition and overexpression of miR-663 in BMSCs, respectively. Thus, miR-663 in MSCs may serve as a new target for autoimmune diseases associated with a T_fh/T_reg cell imbalance.

Materials and methods

Patients and controls

A total of 13 SLE patients (mean age of 34.9±7.6 years) were included in the BMSC study. BMSCs from 4 SLE patients were used for miRNA array analysis, and RT-PCR was performed with BMSCs from another 9 SLE patients. All patients fulfilled 4 or more criteria according to the revised 1997 American College of Rheumatology criteria for SLE⁴⁶ and had SLEDAI (Systemic Lupus Erythematosus Disease Activity Index) scores greater than 6 at the time of bone marrow extraction.⁴⁶ For the BMSC study, 10 normal controls (mean age of 38±6 years) were recruited as normal controls, and 3 patients with primary Sjögren’s syndrome (SS, mean age of 42±7 years) were recruited as disease controls with efforts to match age and gender. In total, for PBMCs and serum RT-PCR analysis, 39 normal controls (mean age of 34±5 years), 22 SS patients (mean age of 39±8 years), 28 SLE patients (mean age of 37±7 years), and 9 other types of connective tissue diseases (mean age of 41±9 years) were included in the analysis. The demographic and clinical characteristics of these patient samples are presented in Supplementary Table 3. All participants provided written consent to participate in the study, which was approved by the Ethics Committee of the Affiliated Drum Tower Hospital of Nanjing University Medical School.

BMSC culture

Bone marrow mononuclear cells were isolated from bone marrow obtained from the iliac crest from all the patients and normal controls by using 1.073 g/ml Ficoll separation medium (TBD, China), and resuspended at a density of 2 × 10⁷ cells per 25 cm² dish in low glucose Dulbecco’s Modified Eagle’s Medium (L-DMEM) (Gibco, USA) supplemented with 10% heat inactivated fetal bovine serum (FBS) (Invitrogen, USA) and 1% antibiotic-antimycotic solution for adherent screening culture at 37 °C in a humidified 5% CO₂ incubator (MCO-15AC, SANYO, Japan). Medium containing non-adherent cells was replaced after 48 h and then every 3 days. Cells grown to 90% confluency were recovered with 0.25% trypsin-ethylenediaminetetraacetic acid (EDTA) (Gibco, USA) and replated at a density of 1 × 10⁶ per 25 cm². At passage 4, cells were identified by flow cytometry (FCM) based on positive staining (>95%) for CD29, CD44, CD90, CD73 and CD166 and negative staining (<2%) for CD45, CD34, CD19, HLA-DR and CD133 (Supplementary Fig. 1).

MRL/lpr mice

Female MRL/lpr mice were purchased from the Model Animal Research Center of Nanjing University. All mice were fed standard chow diet at all ages and maintained in a temperature-controlled room with a 12-h light/dark cycle according to the approved protocol by the Affiliated Drum Tower Hospital of Nanjing University Medical School Committee for the use and care of experimental animals. To observe the in vivo effects of miR-663, BMSCs were transfected with miR-663-C, miR-663-M, and miR-663-I eukaryotic expression vectors and intravenously injected (1 × 10⁶) into 16-week-old MRL/lpr mice, with the FLS injection group (1 × 10⁶) as the background control as described previously.^27,28

Eight weeks later, serum samples and urine from each genotype of female mice were collected. The mice were then killed by cervical dislocation, and kidney and lymph node were collected and spleen was weighed. The value (mg) was divided by the body weight (g) and then multiplied by 10 to determine the splenic index. Lumbar vertebrae and limbs were excised to obtain mononuclear cells from bone marrow, and then mouse BMSCs (mBMSCs) were cultured in vitro to prepare for the next proliferation and migration study by CFSE staining and the trans-well assay, respectively, as described below. Total IgG, IgG anti-ds-DNA and ANA were measured using a commercial ELISA kit (R&D Systems, Inc., MN) according to the manufacturer’s protocol. Cytokine levels were detected using enzyme-linked immunosorbent assay kits. Proteinuria in fresh urine was examined using the Coomassie blue staining assay (Bayer, Elkhart, IN).

Immunohistochemistry

Half of the kidney was fixed in 10% formaldehyde and embedded in paraffin. Three-micrometer-thick sections of kidney tissue were cut and observed for various morphological lesions after hematoxylin-eosin (H&E) staining. Glomerular pathology was evaluated by assessing 20 glomerular cross-sections (gcs) per kidney and scored for each glomerulus on a semi-quantitative scale (0–3).⁴⁷ The other half of the kidney was embedded in Tissue-Tec OCT medium, frozen in liquid nitrogen, and stored at −70 °C until sectioning. Five-micrometer-thick frozen sections were fixed with 4% paraformaldehyde and blocked with 2% BSA. Subsequently, slides were stained with goat anti-mouse IgG (1:150 dilutions; Abcam, Cambridge, UK) or goat anti-mouse complement C3 (1:150 dilutions; Abcam, Cambridge, UK). The sections were then stained with FITC-labeled anti-goat antibodies and digitally photographed using a fluorescence microscope fitted with a digital camera (Cannon Power shot G10, Cannon, Inc.). The fluorescence intensity within the peripheral glomerular capillary walls and mesangial region were scored on a scale from 0 to 3 (0=none; 1=weak; 2=moderate; 3=strong).⁴⁷ At least 10 glomeruli/section were analyzed.

Lymph node samples were fixed for 2 h in 4% paraformaldehyde on ice, incubated in six changes of sucrose buffer overnight and embedded in Tissue-Tek OCT compound (Sakura Finetek). Sections were blocked with a Streptavidin and Biotin Blocking Kit (Vector Laboratories) and then stained with an antibody against mouse Foxp3 (Abcam, Cambridge, UK), followed by donkey anti-mouse antibody labeled with fluorescein isothiocyanate (FITC) (Abcam, Cambridge, UK). Then, the sections were stained with a biotinylated antibody against mouse Bcl6 (Santa Cruz) followed by streptavidin-HRP conjugates and TSA tetramethylrhodamine. Next, the sections were stained with 4 ′,6-diamidino-2-phenylindole (DAPI, Sigma, USA) for 3–5 min. Finally, the stained sections were digitally photographed using a fluorescence microscope fitted with a digital camera (Cannon Power shot G10, Cannon, Inc.) using a × 20 objective.

MicroRNA array analysis

Total RNA containing small RNA was extracted from BMSCs using a mirVana miRNA Isolation Kit (Ambion, Austin, TX, USA). The purity and concentration of the RNA were determined from OD260/280 readings using a spectrophotometer (NanoDrop ND-1000). The RNA integrity and concentration were determined by capillary electrophoresis using the RNA 6000 Nano Lab-on-a-Chip kit and the Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA). Only RNA extracts with RNA integrity number values ?6 underwent further analysis. MiRNA profiling was performed using a GeneChip miRNA Array (Affymetrix, Santa Clara, CA, USA). The array comprised 7815 probe sets, of which 6703 provided miRNA coverage of human, mouse, rat, canine and others (71 organisms) from the Sanger miRNA database (V.11) and an additional 922 encompassed human snoRNAs and scaRNAs (from the Ensembl database and snoRNABase). Control targets were also included in the array containing 95 background probe sets, 22 oligonucleotide spike-in control probe sets and 10 identical probes for human 5.8s rRNA, and hybridization control probe sets. Microarray experiments were conducted according to the manufacturer's instructions. Briefly, 1 μg total RNA was labeled with the Biotin FlashTag Biotin Labeling Kit (Affymetrix). The labeling reaction was hybridized on the miRNA Array in an Affymetrix Hybridization Oven 640 (Affymetrix) at 48 °C for 16 h. The arrays were stained with a Fluidics Station 450 using fluidics script FS450_0003 (Affymetrix) and then scanned on a GeneChip microarray scanner (Affymetrix).

For the microarray data analysis, after removing the adaptor sequence, miRNA probe outliers were defined according to the manufacturer's instructions (Affymetrix) and further analyzed for data summarization, normalization and quality control using the miRNA QC Tool software (www.affymetrix.com). To determine the significant differentially expressed genes, Significance Analysis of Microarrays (SAM, version 3.02) was performed. To select differentially expressed genes, we used threshold values ?2, a ?−two fold change and an FDR significance level of <5%. The data were Log2-transformed and median-centered by gene using the adjust data function in the CLUSTER 3.0 software and then analyzed by hierarchical clustering with average linkage. Finally, tree visualization was performed using Java Treeview (Stanford University School of Medicine, Stanford, CA, USA).

Quantitative real-time polymerase chain reaction (RT-PCR) analysis

Total RNAs in passage 4 BMSCs or PBMCs were extracted using TRIzol Reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. MicroRNAs from each serum sample were extracted using the mirVana miRNA Isolation Kit (Ambion, Austin, TX, USA) according to the manufacturer's protocol. RNA integrity was determined using formaldehyde denaturalization agarose gel electrophoresis. RNA concentrations were measured using a smartspecTM plus spectrophotometer (BIO-RAD, Hercules, CA, USA). cDNA was generated using SuperScript III First Strand Synthesis SuperMix (Takara, Dalian, China), a TaqMan MicroRNA Reverse Transcript Kit (Applied Biosystems), or the Multiplex RT pool set (Systems Biosciences). TaqMan probes for individual miRNAs were purchased from Applied Biosystems. Internal housekeeping controls were β-actin, U6 or miR-16. Quantification of mRNA and mature miRNA was performed on an ABI 7500 FAST real-time PCR detection system (Applied Biosystems, USA) as previously described. Specific primer oligonucleotides (TaKaRa, Dalian, China) were used, and the relative expression of target genes was calculated with the 2^−△△Ct method.

FCM analysis

The following antibodies were applied in this study: fluorescein isothiocyanate (FITC)-conjugated anti-human CD4, HLA-DR (BD Biosciences), CD34 and CD44 (BD eBioscience), phycoerythrin (PE)-conjugated anti-human CD4 (BD Biosciences), CD45, CD29, CD166, CD138 and Foxp3 (eBioscience), allophycocyanin (APC)-conjugated anti-human CD25 (BD Biosciences), CD19 (eBioscience), B220 (eBioscience), phycoerythrin-Cy7-conjugated anti-human IFN-γ (eBioscience) and their isotype-matched control antibodies (mouse IgG1, mouse IgG2a).

Western blot analysis and ELISA assay

We used antibodies recognizing human Smad2/3, p38/MAPK, Akt and their phosphorylation forms, p-Smad2/3, p-p38/MAPK, p-Akt and GAPDH (Cell Signaling Technology Inc., 1:1000) to examine the concentrations of proteins in MSC lysates. We detected the amounts of active and total TGF-β1 in the conditioned medium and/or human serum with ELISA kits (eBioscience or BioLegend) according to the manufacturers’ instructions.

Plasmid construction, transfection and reporter assay

Plasmid vectors (Supplementary Fig. 2a, b) containing miR-663 negative control (miR-663-C), pri-miR-663 (miR-663-M) and inhibitor-miR-663 (miR-663-I) were purchased from Invitrogen (Life Technologies, USA) using the following sequences (Supplementary Table 6). Transient transfection of these plasmids into BMSCs was performed by electroporation using the 4D-nucleofector system (Lonza, Germany) according to the manufacturer’s instructions.

DNA fragments of TGF-β1 with or without 3′-UTR (144 bp) extracted from the cDNA library were cloned into the PGL3 luciferase vector (Promega, USA) via the XbaI 1934 site (Supplementary Fig. 3e). Primers for TGF-β1 without the 3′-UTR were as follows: forward primer, 5′--3′; reverse primer, 5′--3′. Primers for full-length TGF-β1 were: forward primer, 5′--3′; reverse primer, 5′--3′. Transfection of HEK293T cells with miRNAs was performed using Lipofectamine 2000 (Invitrogen, USA). HEK293T cells were co-transfected with the luciferase reporter constructs (200 ng) and appropriate miRNA plasmids. After 36 h, the cells were washed and lysed with passive lysis buffer (Promega, USA), and f-luc and renilla luciferase (r-luc) activities were determined using the dual-luciferase reporter assay system (Promega, USA). Relative reporter activity was obtained by normalization to the r-luc activity.

Cell proliferation and apoptosis assay

Briefly, for the proliferation assay, 10⁶ cells/ml (BMSCs or purified T cells) were incubated with 3 μM carboxyfluorescein diacetate succinimidyl ester (CFSE, Invitrogen, Camarillo, CA, USA). For the apoptosis assay, miR-663-transfected BMSCs (1 × 10⁶/well) were stained with the Annexin V/7AAD apoptosis detection kit Annexin V (BD Biosciences). After culturing for 4–5 days, cells were collected for examination by FCM, and then they were analyzed using Flowjo v10.0.7.

Trans-well migration assay

BMSCs at a density of 5 × 10⁵ cells/ml in 0.2 ml L-DMEM (Gibco, USA) without FBS (Invitrogen, USA) were added to the upper chamber of a 6.5-mm-diameter trans-well insert (8 μM pore size, Millipore). The lower chamber in 24-well plates contained 0.5 ml of L-DMEM (Gibco, USA) with 10% FBS. After incubation at 37 °C and 5% CO₂ for 12 h, the upper surface of the membrane was gently scraped to remove non-migrating cells. Cells on the lower surface of the membrane that had migrated into the lower compartment of the chamber were then fixed in 4% paraformaldehyde for 10 min and stained with Giemsa solution or DAPI (Sigma, USA) for 3–5 min. The number of migrating cells was quantified using Image Pro-Plus 6.0 in five random morphological fields per well. The values were averaged and then multiplied by the ratio per microscopic field area to the bottom area per 24 wells.

BMSC-PBMC co-culture

PBMCs were isolated from heparinized venous blood by Ficoll-Paque gradient centrifugation (Takara, Dalian, China) and co-cultured with or without pre-plated miR-663-transfected BMSCs at a ratio of 10:1 in 96-well flat-bottomed plates in the presence of soluble anti-human CD3 (1 μg/ml) and anti-human CD28 (1 μg/ml) antibodies in a final volume of 200 μl RM1640 medium (Gibco, USA). Recombinant human TGF-β1 (5 ng/ml; R&D Systems, USA) or anti-human TGF-β1 antibody (10 μg/ml; R&D Systems, USA) was added to the culture system to determine the role of TGF-β1 in the miR-663-mediated effect on BMSCs. After 72 h, non-adherent cells were collected for FCM analysis, and supernatants were collected for cytokine measurement. The adherent BMSCs were washed and lysed with TRIzol (Takara, Dalian, China) for RT-PCR analysis. To rule out possible cell–cell contact effects, a trans-well system (0.4 μM pore size, Millipore) was applied.

Differentiation assay

PBMCs were isolated from peripheral blood using Ficoll density-gradient centrifugation. CD4⁺CD25^- T cell subsets or naïve CD4⁺ T cells were isolated and purified using a human CD4⁺CD25⁺ regulatory T cell isolation kit or naïve CD4⁺ T cell isolation kit (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer’s instructions. Then, CD4⁺CD25^- T cells (1 × 10⁶/well) were cultured with soluble anti-CD3 (1 μg/ml) and anti-CD28 (1 μg/ml) antibodies, with the addition of recombinant human TGF-β1 (10 ng/ml; R&D Systems, USA) and IL-2 (100 U/ml; Peprotech, USA) to induce T_reg cell conversion. After culturing for 5– 6 days, the cells were collected for the measurement of CD4⁺CD25⁺ percentages by FCM.

In addition, isolated naive CD4⁺ T cells (1 × 10⁶/well) were stimulated with soluble anti-CD3 (1 μg/ml) and anti-CD28 (1 μg/ml) antibodies, with the addition of recombinant human IL-2 (100 U/ml; Peprotech, USA), IL-6 (20 ng/ml; Peprotech, USA), anti-IL-4 (10 μg/ml; R&D Systems, USA), anti-IFN-γ (10 μg/ml; R&D Systems, USA) and anti-TGF-β (10 μg/ml; R&D Systems, USA). After 5–6 days, the cells were collected for the measurement of CXCR5⁺⁺PD-1⁺⁺CD4⁺ T cell or CXCR5⁺⁺PD-1⁺⁺ Foxp3⁺CD4⁺ T cell percentages by FCM.

In vivo homing assay

To determine the in vivo homing capacity, BMSCs were labeled with PKH26 (Sigma, USA) according to the manufacturer’s protocol, with a labeling efficacy >98%. Labeled cells were infused into 24-week-old MRL/lpr mice, and the mice were killed 24, 48 and 72 h after infusion, with organs including heart, liver, spleen, lung, kidney and lymph nodes collected and wrapped in the dark at −80 °C. To quantify BMSC engraftment in organ tissue, samples were stained with Giemsa solution or 4′,6-diamidino-2-phenylindole (DAPI, Sigma, USA) for 3–5 min, fixed in 4% paraformaldehyde (Electron Microscopy Sciences, PA), and embedded in 30% sucrose/PBS and in Tissue-Tek OCT Compound (Sakura Finetek, CA). Fifteen sections per organ were analyzed using a fluorescence inverted microscope (Axio observer A1, ZEISS, Germany), and images were acquired with an objective magnification of × 10 (× 40 total magnification) using an Olympus DP30BW camera (Olympus, Japan).

Statistical analysis

The data are shown as the mean±SEM. Differences between two groups were determined using an unpaired Student’s t-test if the variance was normally distributed. Comparisons among three or more groups were conducted using one-way ANOVA. Data were calculated using GraphPad Prism 5 software, and a value of P<0.05 was considered statistically significant.

Results

The miR-663 level is increased in SLE BMSCs

To determine whether there were differentially expressed miRNAs in SLE BMSCs, we performed comparative miRNA screenings between SLE patients (n=4) and normal controls (NOR, n=4) using human miRNAs arrays containing 194 miRNAs (Supplementary Table 1). Compared with normal controls, 8 miRNAs were highly expressed and 4 miRNAs were decreased in SLE BMSCs (Figure 1a, Supplementary Table 2). Next, the data from miRNAs arrays were confirmed by real-time PCR (RT-PCR) in another set of BMSCs from 9 SLE patients and 6 age- and gender- matched normal controls. Of the verified miRNAs, 4 (miR-663, miR-638, miR-214 and miR-574-3p) remained increased and 2 (let-7 f and miR-374a) remained reduced in SLE BMSCs (Figure 1b). It has previously been reported that several miRNAs, including miR-155, miR-126, miR-125a, miR-146a, miR-150, miR-181a and miR-21, in peripheral blood or kidneys are linked to SLE patients.¹⁶ However, we found that none of these SLE-associated miRNAs was abnormally expressed in SLE BMSCs, except that the miR-125a level was decreased in the patient group (Figure 1c). Thus, SLE patients may present a distinct BMSC miRNA signature compared with normal controls.

Figure 1.

MiR-663 expression is increased in SLE BMSCs. (a) Screening of miRNAs in BMSCs from SLE patients and normal controls (NOR). (b) RT-PCR validation of miRNA expressions in BMSCs from 9 SLE patients and 6 NOR. (c) Expression of SLE-associated miRNAs in BMSCs. (d and g) Association of miR-663 expression in BMSCs (d) and serum (g) with the SLE disease activity index (SLEDAI) score. (e,f) MiR-663 expression in serum and peripheral blood mononuclear cells (PBMCs) from SLE patients, Sjögren's syndrome (SS) patients and NOR. For others, patients with rheumatoid arthritis (RA) and dermatomyositis (DM) were included. (h) MiR-663 expression in BMSCs from patients with SS and umbilical cord-derived MSCs (UCMSCs). All data represent the mean±SEM. *P<0.05, **P<0.01, ***P<0.001.

Among the 6 verified miRNAs (Figure 1b) in SLE BMSCs, miR-663 had previously been implicated to participate in immune regulation.²³ Our data showed that the miR-663 level was positively correlated with SLE clinical spectrums, including the SLE disease activity index (SLEDAI, Figure 1d) and the erythrocyte sedimentation rate (ESR) and levels of C-reactive protein (CRP) and complement 3 (C3, data not shown), suggesting that miR-663 is potentially involved in the pathogenesis of SLE. Consistent with the observations in BMSCs, miR-663 expression in SLE serum, but not in peripheral blood mononuclear cells (PBMCs), was also increased and associated with the SLEDAI score (Figures 1e–g, Supplementary Table 3). Unlike SLE, patients with Sjögren's syndrome (SS) did not show high levels of miR-663 in their BMSCs (Figure 1h). Consequently, miR-663 was selected for further investigation.

MiR-663 is involved in BMSC proliferation, apoptosis and migration

To investigate the biological functions of miR-663 in BMSCs from normal controls, eukaryotic expression vectors (Supplementary Fig. 2a, b; Supplementary Table 5) of pri-miR-663 (miR-663-M) and inhibitor-miR-663 (miR-663-I) were generated. Compared with the negative control vector (miR-663-C)-transfected BMSCs, miR-663 production was increased in miR-663-M-transfected BMSCs and decreased in the miR-663-I group (Supplementary Fig. 2c, d). MiR-663-M ectopic expression significantly suppressed proliferation (Figures 2a and b), promoted apoptosis (Figures 2c and d) and decreased the migration capacity (Figures 2e and f) of BMSCs in vitro. Conversely, inhibition of miR-663 by miR-663-I transfection resulted in opposite effects (Figure 2). As increased apoptosis, reduced proliferation and migration constituted the characteristic abnormalities of SLE BMSCs,^3,4,6 the increased expression of miR-663 in BMSCs may contribute to the pathogenesis of SLE.

Figure 2.

MiR-663 affects BMSC proliferation, apoptosis and migration in vitro. (a and b) Proliferation status of normal BMSCs (miR-663-N), normal BMSCs transfected with control vector (miR-663-C), pri-miR-663 (miR-663-M) or inhibitor-miR-663 (miR-663-I) by the CFSE assay. (c and d) Apoptotic status of BMSCs among the 4 groups using the Annexin V and 7-AAD staining assay. Dot plots showing the frequency of apoptosis (Annexin V⁺7-AAD^- early apoptotic cells and Annexin V⁺7-AAD⁺ late apoptotic cells) of BMSCs among different groups. (e and f) Migrated BMSC numbers per field in the 4 groups. The yellow arrow refers to the BMSC nucleus stained with Giemsa solution. All data represent the mean±SEM. n=6, *P<0.05, **P<0.01.

MiR-663 inhibits the immunoregulatory effect of BMSCs on T_fh/T_reg cells

To explore whether miR-663 could affect the immunoregulatory function of BMSCs, BMSCs were co-cultured with peripheral blood mononuclear cells (PBMCs) of NOR at a ratio of 1:10 for 3 days. Compared with miR-663-C-transfected BMSCs, there was no alteration of the T helper (Th) 1, Th2 and Th17 subsets after co-culturing with miR-663-M or miR-663-I-transfected BMSCs (Figure 3a, Supplementary Fig. 3a-b). However, the frequency of T_reg cells (CD25⁺Foxp3⁺/CD4⁺ T cells, Figure 3b) was upregulated, while the frequency of T_fh cells (CXCR5⁺⁺PD-1⁺⁺/CD4⁺ T cells, Figure 3c, Supplementary Fig. 3c) and plasma cells (CD19^-CD138⁺ cells, Figure 3d) were downregulated in miR-663-I-transfected BMSC co-cultures. In contrast, a decreased frequency of T_reg cells but increased frequency of T_fh cells, as well as plasma cells were observed in miR-663-M-transfected BMSC co-cultures (Figures 3b–d, Supplementary Fig. 3c). Consequently, the ratio of T_fh/T_reg was significantly increased in the miR-663-M groups but decreased in the miR-663-I groups (Figure 3i), suggesting that miR-663 controls the BMSC-mediated T_fh/T_reg cell balance.

Figure 3.

MiR-663 downregulates the immunoregulatory effects of BMSCs in vitro. (a–d) Percentages of Th17 cells (IL-17 A⁺/CD4⁺ T cells, (a), T_reg cells (CD25⁺Foxp3⁺/CD4⁺ T cells, (b), T_fh cells (CXCR5⁺⁺PD-1⁺⁺/CD4⁺ T cells, (c), and plasma cells (CD19^-CD138⁺ cells, (d) after co-culturing of pre-stimulated naïve T cells with miR-663-related vector-transfected BMSCs. (e) The percentage of proliferated Foxp3⁺ cells in sorted CD4⁺CD25⁺ T cells after co-culture with different BMSCs. (f) The absolute number of Foxp3⁺CD4⁺CD25⁺ T cells increased in the presence of miR-663-I BMSCs. (g and h) The effects of miR-663-related BMSCs on T_reg cell differentiation and the conversion of T_reg cells to T_fh cells. (i) The ratio of T_fh (CXCR5⁺⁺PD-1⁺⁺/CD4⁺ T cells) to T_reg (CD25⁺Foxp3⁺/CD4⁺ T cells) after co-culturing of pre-stimulated naïve T cells with miR-663-related BMSCs. All data represent the mean±SEM. n=6, *P<0.05, **P<0.01, ***P<0.001.

To understand how miR-663 in BMSCs interferes with T_fh/T_reg cells, we co-cultured miR-663-transfected BMSCs with CD4⁺CD25⁺ and CD4⁺CD25^- T cells isolated from normal human PBMCs. The proliferation rate and absolute number of Foxp3⁺CD4⁺CD25⁺ T cells were elevated in the miR-663-I group, but were reduced in the miR-663-M group compared with miR-663-C co-cultures (Figures 3e and f; Supplementary Fig. 3d), suggesting that miR-663 suppresses BMSC-mediated natural (thymic) T_reg cells (nT_reg) growth. In addition, our data showed that miR-663-I BMSCs increased the differentiation of T_reg cells (iT_reg) from CD4⁺CD25^- T cells, but miR-663-M BMSCs decreased T_reg cell differentiation (Figure 3g). Since it has been recently reported that T_reg cells may be converted into T_fh cells,²⁴ we next investigated whether miR-663 could also control the phenotypic conversion of T_reg cells to T_fh cells. As expected, miR-663-M-transfected BMSCs significantly increased the conversion of CD4⁺CD25⁺ T cells to T_fh cells compared with miR-663-C-transfected BMSCs, while there was a trend toward a decrease in T_fh cell conversion in the miR-663-I group (Figure 3h). Thus, miR-663 in BMSCs may downregulate the frequency of T_reg cells not only by inhibiting nT_reg cell proliferation and iT_reg cell differentiation but also by converting T_reg cells toward a T_fh cell phenotype.

MiR-663 acts through targeted regulation of the TGF-β1 3’UTR

It has been widely recognized that miRNAs exert their functions through the regulation of target genes. To identify candidate miR-663 target genes, two computational methods, Targetscan (http://www.targetscan.org) and PITA (http://genie.weizmann.ac.il/pubs/mir07/mir07_prediction.html), were applied, and CapitalBio Molecule Annotation System V3.0 was used to perform pathway analysis on the putative miR-663 target genes (http://bioinfo.capitalbio.com/mas3/). According to the prediction, several molecules, including TGF-β1 were proposed as potential targets of miR-663 (Supplementary Fig. 4a).

Interestingly, among the predicted targets, only TGF-β1 mRNA (Supplementary Fig. 4b) and protein levels (Figure 4a) were significantly lower in SLE BMSCs than that in normal BMSCs, and TGF-β1 derived from BMSCs was involved in upregulating T_reg cells (Figure 4b) and promoting BMSC migration (Figure 4c). In addition, both active and total protein levels of TGF-β1 secreted by BMSCs were further elevated in the miR-663-I group (Figure 4d). MiR-663-I transfection markedly increased TGF-β1 mRNA expression in BMSCs but had no effect on TNF-α, IFN-α, IL-1β, JUND, GRID2D and EPHB3 (Supplementary Fig. 4c). To further confirm the role of TGF-β1 in miR-663-mediated BMSC dysfunction, exogenous human recombinant TGF-β1 and anti-TGF-β1 antibody were applied. Consistently, miR-663-M-modified dysregulation of T_reg cell differentiation and dysfunction of BMSC migration capacity were restored by recombinant TGF-β1, while the increased T_reg cell differentiation (Figure 4e) and BMSC migration capacity (Figure 4f) by miR-663-I transfected BMSCs were abolished by anti-TGF-β1 antibody.

Figure 4.

TGF-β1 is a direct target of miR-663. (a) SLE BMSCs produce fewer active and total TGF-β1 proteins than normal BMSCs after culturing for 24 hours. (b and c) The role of TGF-β1 in the immunoregulatory effect of BMSCs on T_reg cells (b) and the BMSC migration capacity (c). (d) Active and total TGF-β1 protein levels in cultured supernatants of various miR-663 related BMSCs. (e and f) The role of miR-663 in the immunoregulatory effect of BMSCs on T_reg cells (e) and BMSC migration capacity (f) acting through the regulation of TGF-β1. (g and h) The effect of miR-663 and TGF-β1 on the Akt, p38/MAPK and Smad2/3 pathways in BMSCs and UCMSCs. (i) The role of Akt inhibitor (GSK690693), p38/MAPK inhibitor (SB203580) and Smad inhibitor (SB431532) in the migration capacity of miR-663-related BMSCs. (j) Five potential miR-663 binding sites predicted in TGF-β1 3′-UTR. (k) The TGF-β1 3′-UTR is involved in miR-663-regulated gene expression. All data represent the mean±SEM. n=6, *P<0.05, **P<0.01, ***P<0.001.

Several signaling pathways have been reported to participate in TGF-β1-induced BMSC migration.^25,26 Our data showed that miR-663-M significantly suppressed the phosphorylation of Akt and p38/MAPK in BMSCs, which could be abrogated by exogenous human recombinant TGF-β1, while miR-663-I led to an opposite effect (Figures 4g and h). When treated with Akt inhibitor (GSK690693), p38/MAPK inhibitor (SB203580) and Smad inhibitor (SB431532), only GSK690693 and SB203580 significantly decreased the migration capacity of miR-663-I BMSCs (Figure 4i), suggesting that Akt and p38/MAPK signaling pathways play a critical role in miR-663-mediated BMSC dysfunction.

Bioinformatics analyses revealed five potential miR-663 binding sites in the TGF-β1 3′-untranslated regions (3′-UTR, Figure 4j). To experimentally validate whether the binding between miR-663 and TGF-β1 3′-UTR is necessary for the transcription of TGF-β1, luciferase reporters containing the wild type or mutated 3′-UTR of human TGF-β1 (Supplementary Fig. 4d) were co-transfected with miR-663-C or miR-663-M into HEK293T cell lines. In cells expressing a luciferase construct containing the wild-type TGF-β1 3′-UTR, miR-663-M transfection markedly reduced luciferase activity and miR-663-I transfection significantly increased luciferase activity (Figure 4k). However, in cells with the mutated TGF-β1 3′-UTR, co-transfection of either miR-663-M or miR-663-I did not alter the reporter activity, supporting that binding to the TGF-β1 3′-UTR is crucial for the regulation of TGF-β1 expression by miR-663.

MiR-663 impairs the therapeutic effects of BMSCs in MRL/lpr mice

Systemic infusion of BMSCs has been shown to be beneficial for lupus mice.^10,11 Since miR-663 participated in the immunoregulatory effect of BMSCs, we next investigated whether inhibition of miR-663 could augment the therapeutic effects of BMSCs in MRL/lpr mice. Treatment with miR-663-I-transfected BMSCs significantly reduced the size of spleens and lymph nodes (Figures 5a and b) and decreased serum levels of total immunoglobulin (IgG) and IgG anti-double-stranded DNA antibodies (anti-ds-DNA) (Figures 5c and d) in MRL/lpr mice compared with those treated with miR-663-C-transfected BMSCs. Renal impairments were also ameliorated in the miR-663-I group, as shown by lower proteinuria (Figure 5e), reduced glomerular enlargement and hypercellularity (Figure 5f) and less IgG and complement 3 (C3) deposition in the peripheral capillary loops (Figures 5g and h). In contrast, the therapeutic effect of BMSCs in MRL/lpr mice was almost abolished when BMSCs were transfected with miR-663-M, with serological alterations and organ involvement similar to the mice treated with fibroblast-like synoviocytes (FLS), which were used as negative controls based on previous studies.^27,28 Taken together, these in vivo findings demonstrate that miR-663 overexpression impairs the therapeutic effects of MSCs in MRL/lpr mice.

Figure 5.

MiR-663 overexpression impairs the treatment effects of BMSCs in MRL/lpr mice. (a–e) Lymph node size (a), spleen index (b), total IgG (c), anti-ds-DNA levels (d), and proteinuria (e) levels among different groups. (f–h) Representative images of renal pathology and immunohistochemical analysis of MRL/lpr mice after treatment with miR-663-related BMSCs. All data represent the mean±SEM. n=8 for each group at the beginning of the experiment. For detection at 24 weeks, n(FLS)=6, n(miR-663-M)=7, n(for other groups)=8. *P<0.05, **P<0.01, ***P<0.001.

Inhibition of miR-663 restores the T_fh/T_reg cell balance in vivo

Next, we studied the mechanisms of the in vivo effect of miR-663 on BMSCs. To examine the homing capacity of BMSCs in MRL/lpr mice, miR-663-M or miR-663-I-transfected BMSCs were labeled with PKH26 and then infused intravenously. After 24 h, massively trapped BMSCs within the lung and liver rather than inflammatory organs, including kidney, lymph nodes and spleen were present in the miR-663-M group. In contrast, a significantly enhanced BMSC homing capacity to kidney, lymph node and spleen was observed in the miR-663-I group (Figure 6a; Supplementary Fig. 5-10). A similar effect was detected at 48 and 72 h (data not shown).

Figure 6.

MiR-663 contributes to the T_fh/T_reg cell imbalance in vivo. (a) MiR-663-related human BMSC homing capacity to different organs in MRL/lpr mice at 24 h after infusion. (b and c) Migration and proliferation capacity of mouse BMSCs (mBMSCs) from MRL/lpr mice at 8 weeks after treatment of different human BMSCs. (d–i) Percentages of T_reg (d,f), T_fh (e,g) and expression levels of Bcl6 mRNA (h) as well as the ratio of T_fh/T_reg (i) in mononuclear cells from the spleen of MRL/lpr mice after treatment with different BMSCs. (j and k) The immunofluorescence intensity ratio of Bcl6⁺/Foxp3⁺ cells in the germinal center (GC) of lymph nodes from MRL/lpr mice after treatment with different BMSCs. Foxp3⁺ cells (green) were present in the Bcl6⁺ germinal center area (red) in MRL/lpr mice. All data represent the mean±SEM. n=3 for each group in (a), n=6 for each group in (b,c); for other images, n(FLS)=6, n(miR-663-M)=7, n(for other groups)=8. *P<0.05, **P<0.01, ***P<0.001.

To determine the role of miR-663 in BMSC migration, BMSCs from MRL/lpr mice (mBMSCs) were collected 8 weeks after BMSC or FLS treatment (Supplementary Fig. 11a). We found the migration capacity of mBMSCs increased after treatment with miR-663-I-transfected BMSCs by the trans-well assay but decreased in the miR-663-M group compared with mice treated with miR-663-C-BMSCs (Figure 6b,Supplementary Fig. 11b). Thus, downregulation of miR-663 could not only enhance the homing capacity of infused BMSCs but also facilitate the migration of host mBMSCs. In addition to the migration capacity, the proliferation rate of mBMSCs was also increased after infusion of miR-663-I-transfected BMSCs (Figure 6c).

The interaction between activated T_fh cells and long-lived plasma cells would promote autoantibody production, which is involved in lupus onset.¹⁹ Consistent with the in vitro data, not only the mRNA level of B-cell lymphoma/leukemia 6 (Bcl6), a well-recognized transcription factor defining the Tfh lineage but also the frequencies of T_fh cells and plasma cells were elevated in the miR-663-M group compared with the miR-663-C group (Figures 6d–h; Supplementary Fig. 11g, j). Similarly, the ratio of T_fh/T_reg increased in the miR-663-M group and decreased in the miR-663-I group (Figure 6i). To determine the location of T cell subsets in germinal centers (GC), T_fh cells (Bcl6⁺) and T_reg cells (Foxp3⁺) were stained using immunofluorescence histochemistry in lymph nodes from MRL/lpr mice. The intensity ratio of Bcl6⁺/Foxp3⁺ cells was significantly lower in the miR-663-I group compared with the miR-663-C group (Figures 6j and k), supporting the role of miR-663 in the regulation of T_fh/T_reg cells. Similar to the in vivo findings, there were no differences in Th1, Th2 and Th17 cells in PBMCs and protein levels of IFN-γ, IL-4 and IL-17A in the serum among these groups (Supplementary Fig. 11c–e, 12c,e,f). However, the total TGF-β1, IL-10 and IL-21 level remained upregulated 8 weeks after the infusion of miR-663-I-transfected BMSCs (Supplementary Fig. 12a,b,d). Thus, miR-663 serves as a vital factor in the MSC-regulated T_fh/T_fr cell balance.

Discussion

The bone marrow microenvironment, especially BMSCs, is critical for the acquisition and maintenance of immunological balance.²⁹ MiRNAs, which target messenger RNA for cleavage or translational repression, have recently been shown to play an important role in the regulation of MSC functions.^18,30 Although, miRNA profiling in PBMCs has been performed in individual autoimmune diseases,^31,32 the role of MSC-derived miRNAs in autoimmune pathogenesis is largely unknown. Here, we identified 8 highly expressed miRNAs in BMSCs from patients with SLE, a prototypical autoimmune disease, among which 4 miRNAs (miR-663, miR-638, miR-214 and miR-574-3p) remained upregulated after verification. These data differ from those in other tissues, in which miR-155, miR-l26 and miR-146a have received extensive attention,^13,16,33 suggesting a distinct miRNA signature in SLE BMSCs.

The characterization of miRNAs present in BMSCs may be relevant not only as a signature of the cell type but also for understanding their biological activities. We further characterized miR-663 because this miRNA is potentially associated with immune functions²³ and relevant to SLE disease activity (Figure 1d). MiR-663 has been demonstrated to play a critical role in anti-malignancies^34,35,36 because it induces the differentiation but suppresses the proliferation and invasion of tumor cells. Moreover, miR-663 is important for the key events induced in endothelial cells by stress agents and oxidized lipids,³⁷ and thus it may have a potential role in the development of atherosclerosis. However, the effect of miR-663 in MSCs remains largely unknown. In this study, we identified that miR-663 is not only associated with the apoptosis, proliferation and migration capacity of BMSCs, but it is also involved in the immunoregulatory function of BMSCs in the T_fh/T_reg cell balance both in vitro and in vivo.

Many miR-663 target genes have been predicted by bioinformatics analysis, among which several have been confirmed in specific cell types. MiR-663 may decrease endogenous activator protein-1 (AP-1) activity, at least in part by directly targeting the JUNB and JUND transcripts in THP-1 cells.²³ Moreover, myosin light chain 9 (MYL9) in human vascular smooth muscle cells,³⁸ TGF-β1 in A549 cells,³⁹ and heparin sulfate proteoglycan 2 (HSPG2) in human breast cancer cells⁴⁰ have been identified as downstream targets of miR-663. Here, we evaluated the effects of miR-663 on the expression of predicted targets (TGF-β1, JUND, MYL9, GRID2D, EPHB3) and several well-known pro-inflammatory cytokines, including TNF-α, IFN-α and IL-1β, in BMSCs. Our data showed that in addition to MYL9, only TGF-β1 expression was regulated by miR-663 via binding to the 3’UTR.

TGF-β1, the most abundant isoform of the TGF-β family involved in immune regulation, can be produced by bone marrow cells. TGF-β1 is an important multifunctional cytokine that is involved in the control of several biological processes, including cell proliferation, differentiation, migration, apoptosis and, most importantly, immune modulation.⁴¹ While the role of TGF-β1 in T_reg cell generation is widely accepted,⁴² the underlying molecular mechanism remains unclear. In this study, we showed that miR-663 participated in BMSC-regulated T_reg differentiation through the modulation of TGF-β1 production, which might be related to the p38/MAPK and Akt signaling pathways.

T_fh cells have recently been highlighted for their crucial role in humoral immunity as well as their abnormal control leading to the induction of autoimmune diseases.¹⁹ Multiple cytokines, signaling molecules and transcription factors have been reported to be involved in T_fh cell differentiation.¹⁹ A recent study has illustrated that TGF-β can induce miR-10a expression in T_reg cells, which helps attenuate the phenotypic conversion of inducible T_reg cells into T_fh cells,²⁴ implicating a possible link between TGF-β1 and T_fh cells. MSCs may also interfere with the viability and differentiation of follicular lymphoma-infiltrating T_fh cells, yet the underlying mechanism remains unclear. The present data support that the imbalance between T_reg and T_fh cells in SLE patients may be linked to overexpression of miR-663 in their BMSCs. Inhibition of miR-663 in BMSCs could restore the production of T_reg and T_fh cells through the secretion of TGF-β1. To further support this notion, we showed that in MRL/lpr mice, the therapeutic effect of BMSCs transplantation was augmented after treatment with miR-663-I-transfected BMSCs, together with elevated TGF-β1 expression and a reversed frequency of T_fh and T_reg cells.

T_fr cells, a specialized subset of T_reg cells that share features of T_fh and T_reg cells, have been found to localize in follicular regions of GC and dampen GC responses by limiting the numbers of both T_fh and B cells.^43,44 The balance between T_fh and T_fr cells in the GC environment likely represents a key factor in the generation of both high-affinity protective antibodies and pathogenic autoantibodies.^20,45 Our data showed that similar to T_reg cells, the percentage of T_fr cells (Foxp3⁺CXCR5⁺⁺PD-1⁺⁺/CD4⁺) was downregulated in the miR-663-M group (Supplementary Fig. 13a). Moreover, the numbers of T_fr cells (Foxp3⁺Bcl6⁺) in lymph nodes and the expression levels of CTLA-4 (cytotoxic T-lymphocyte-associated protein 4), a direct mediator of T_fr function,⁴⁴ were also decreased in spleen and GC in MRL/lpr mice after treatment with miR-663-M-transfected BMSCs (Supplementary Fig. 13b-d), suggesting that T_fr cells may represent a main T_reg cell subset regulated by miR-663.

In addition to the direct role on T_reg cell dysregulation and the T_fh/T_reg cell balance, transplanted MSCs may also act through the regulation of host MSCs to exert their functions, as recent studies have indicated that the exchange of miRNAs through microvesicles between neighboring cells is an integral part of MSC communication with tissue-injured cells.¹⁸ Consequently, we observed the alterations of the migration capacity and proliferation rate of mouse BMSCs (mBMSCs) 8 weeks after the infusion of miR-663-M or miR-663-I-transfected normal BMSCs. Thus, infused BMSCs may act directly or through the modulation of host BMSCs to exert their immunoregulatory functions. The miR-663-TGF-β1-T_fh/T_reg could help not only to elucidate the cause of abnormalities in SLE BMSCs but also to explain the therapeutic function of normal allogeneic MSCs in refractory SLE patients.

Author contributions

LS designed, coordinated and supervised the study. LG carried out most of the experiments, performed the data acquisition and analysis, and wrote the manuscript. XF contributed to the data interpretation and manuscript drafting. KZ, XG, WC, SS and NS participated in the study design. SW, GY, XT, WC and DW participated in human sample collection and breeding of MRL/lpr mice. All authors read and approved the final manuscript.

Conflict of interest

The authors declare no conflict of interest.

Electronic supplementary material

Supplementary Information for this article can be found on the Cellular & Molecular Immunology website 10.1038/cmi.2018.1