Overexpression of soluble RAGE in mesenchymal stem cells enhances their immunoregulatory potential for cellular therapy in autoimmune arthritis

Mesenchymal stem cells (MSCs) are attractive agents for cellular therapy in rheumatoid arthritis (RA). The receptor for advanced glycation end products (RAGE) serves as a pattern recognition receptor for endogenous inflammatory ligands. Soluble RAGE (sRAGE) is a truncated form of RAGE that functions as a decoy and acts as an anti-inflammatory molecule. The aim of this study was to determine whether sRAGE has therapeutic effects and the mechanisms active in sRAGE-overexpressing MSCs (sRAGE-MSCs) in an experimental model of RA. sRAGE-MSCs were generated by DNA transfection of human adipose tissue-derived MSCs (Ad-hMSCs). MSCs showed increased expression of VEGF, IL-1β, IL-6, and HMGB-1 under inflammatory conditions. However, sRAGE-MSCs showed significantly lower production of these proinflammatory molecules. Expression of immunomodulatory molecules such as IL-10, TGF-β, and indoleamine 2, 3-dioxygenase was higher in sRAGE-MSCs than in mock-MSCs. sRAGE-MSCs showed enhanced migration potential. Transplantation of sRAGE-MSCs into arthritic IL-1Ra-knockout mice markedly suppressed inflammatory arthritis, decreased Th17 cells, and reciprocally increased regulatory T cells. The differentiation of IFN-γ+CD4+ and IL-17+CD4+ cells was inhibited by incubation with sRAGE-MSCs compared with mock-MSCs. These findings suggest that sRAGE overexpression in Ad-hMSCs optimizes their immunoregulatory properties, which may be useful as a novel cellular therapy for RA.

Currently, RA treatment remains a significant unmet medical need 6 despite myriad therapeutic advances in biologics that have remarkable efficacy and acceptable safety profiles. Although biologics are more effective than synthetic disease-modifying antirheumatic drugs (DMARDs) in treating RA, a subset of patients achieves only partial remission [7][8][9][10] . Mesenchymal stem cells (MSCs) are cells of stromal origin that are present in various tissues including bone marrow, peripheral blood, umbilical cord blood, and adipose tissues. MSCs can exert profound immunoregulatory effects by modulating the proliferation and cytokine production of T and B cells, maturation of dendritic cells, and activity of NK cells [11][12][13][14] . Much recent research has focused on MSCs, and the results have encouraged the clinical application of MSCs in immunotherapy for autoimmune disorders including Crohn's disease, type 1 diabetes, lupus erythematosus, and Sjögren syndrome [15][16][17][18] . MSCs can reduce the activity of Th17 cells and promote the differentiation of Treg cells 19 .
Although MSCs show beneficial effects in autoimmune disorders because of their anti-inflammatory activity, several preclinical studies have raised significant concerns about their therapeutic application in human RA. A recent study reported that intravenously infused MSCs induce inflammatory responses in vivo 20 . Moreover, MSCs fail to exhibit immunomodulatory properties when infused into the inflammatory micromilieu of autoimmune arthritis 21 . Therefore, the therapeutic potential of MSCs in experimental arthritis remains unclear 22,23 . Although MSC-based therapies have been effective to a certain degree, increasing evidence demonstrates that MSCs do not always achieve immunoregulatory function and that MSCs may promote immune responses only under certain conditions 24 . Therefore, optimization of MSC function may increase the opportunities for their clinical application in the treatment of autoimmune diseases including RA.
The receptor for advanced glycation end products (RAGE), a multiligand cell surface protein, is a member of the immunoglobulin (Ig) superfamily. RAGE participates in inflammatory and immune responses, inducing leukocyte recruitment. Soluble RAGE (sRAGE), a preventer of the RAGE signaling pathway, exerts anti-inflammatory activity by quenching RAGE ligands. sRAGE has been shown to reduce inflammation and to ameliorate several inflammatory diseases 25,26 . The blood levels of sRAGE are significantly lower in RA patients than in healthy controls, and synovial fluid from RA patients treated with DMARDs contained higher sRAGE concentrations compared with that from patients not treated with DMARDs, which shows compromised inflammatory control in active RA 27 .
Although autophagy induction in MSCs can optimize their clinical application by inhibiting the prevention of apoptosis [28][29][30] and MSC treatment has potential therapeutic activity in RA 31 , MSCs fail to inhibit the immune modulation involved in the inflammatory condition of autoimmune arthritis 21 . MSCs in proinflammatory niches have immunoregulatory properties that can be induced by inflammatory cytokines such as tumor necrosis factor-α and establish an immunosuppressive microenvironment 32 . Thus, suppression of the MSC-induced inflammatory response is important for the use of MSCs in RA therapy because these cells are involved in the inflammatory response. It has been suggested that RAGE induces inflammatory responses by promoting HMGB-1 33 and is expressed by MSCs 34 . Additionally, MSCs can be differentiated into cells that display a proinflammatory phenotype 35 . However, sRAGE can induce improvement of several inflammatory disorders 25,26 , and treatment with MSCs improves acute lung injury by decreasing the level of RAGE induced by lipopolysaccharide (LPS) 36 .
IL-1 receptor antagonist (IL-1Ra) prevents binding of IL-1, acts as an inhibitor of IL-1, and suppresses the progression of experimental arthritis 37 . The IL-1Ra-deficient strain derived from BALB/c mice spontaneously develops an excessive immune inflammatory response 38 and increased frequency of Th17 cells 39 . IL-17 expression in T cells is increased by IL-1Ra deficiency 40 , and the unregulated IL-1 signaling mediated by IL-1Ra deficiency leads to the development of arthritis, excessive bone annihilation, and joint inflammation 41 .
To optimize the therapeutic potential of MSCs in RA, we hypothesized that sRAGE overexpression in adipose tissue-derived human MSCs (Ad-hMSCs) may augment their immunoregulatory properties in a murine model of RA. sRAGE was overexpressed by DNA transfection into Ad-hMSCs (sRAGE-MSCs). The anti-inflammatory, migratory, and differentiation potential of sRAGE-MSCswas quantified and compared with that of mock-MSCs. The in vivo immunoregulatory potential of sRAGE-MSCs was studied in IL-1Ra-knockout (IL-1Ra-KO) mice, an experimental model of RA. Finally, we evaluated the mechanisms underlying the augmented anti-arthritic effects of sRAGE-MSCs with respect to regulation of the Th17/Treg cell balance.

Results
Ad-hMSCs produce inflammatory mediators including high-mobility group box-1 (HMGB-1) when stimulated with LPS. Ad-hMSCs were first stimulated with LPS, and the mRNA expression of inflammatory mediators was measured and comparison with that of non-stimulated cells. The transcript levels of VEGF, IL-1β, IL-6, and HMGB-1 in Ad-hMSCs increased significantly after LPS stimulation (Fig. 1A). Significantly higher concentrations of vascular endothelial growth factor (VEGF), IL-1β , IL-6, and HMGB-1 were found in the in culture supernatants from Ad-hMSCs treated with LPS compared with those from non-stimulated Ad-hMSCs (Fig. 1B). Western blot analysis showed that LPS stimulation increased the production of HMGB-1 and RAGE by Ad-hMSCs compared with the control unstimulated cells (Fig. 1C).

Overexpression of sRAGE in Ad-hMSCs inhibits the expression of proinflammatory cytokines
and increases the expression of immunomodulatory mediators. We wanted to ascertain whether sRAGE overexpression could optimize the immunoregulatory properties of Ad-hMSCs. sRAGE was overexpressed in Ad-hMSCs by transfection of an sRAGE expression vector ( Fig. 2A), which significantly decreased the expression and production of HMGB-1 (Fig. 2B). sRAGE overexpression also reduced the gene expression of proinflammatory cytokines such as VEGF, IL-1β, and IL-6 in Ad-hMSCs (Fig. 2C, upper panel). The mRNA levels of immunomodulatory mediators including IL-10, TGF-β, IDO, and HGF were markedly higher in sRAGE-MSCs compared with mock-treated MSCs (Fig. 2C, lower panel). Confocal microscopy also confirmed the induction of IL-10 and indoleamine 2,3-dioxygenase (IDO) in sRAGE-MSCs (Fig. 2D).
Scientific RepoRts | 6:35933 | DOI: 10.1038/srep35933 sRAGE overexpression in Ad-hMSCs enhanced their migratory capacity. Autophagy increases cell survival by inducing lysosomal degradation of damaged cytoplasmic organelles or cytosolic aggregates 42 . Several recent studies have revealed that autophagy induction in MSCs can optimize their clinical application by inhibiting the prevention of apoptosis [28][29][30] . Transmission electron microscopy revealed that sRAGE overexpression in Ad-hMSCs caused an increase in the appearance of double-membrane autophagic vacuoles (autophagosomes) in the cells compared with mock-MSCs (Fig. 3A). sRAGE-MSCs expressed significantly greater mRNA levels of chemokine (C-C motif) receptor 1 (CCR1), CCR3, CCR4, CCR7, chemokine (C-X-C motif) receptors(CXCR1), and CXCR4, which are involved in the migratory capacities of homing and tracking 43 compared with the mock-treated Ad-hMSCs (Fig. 3B). The in vitro migratory capacity of sRAGE-MSCs toward the chemokine stromal-derived factor-1 (SDF-1) was also significantly higher than that of mock-MSCs (Fig. 3C).

Overexpression of sRAGE does not affect the differentiation potential of Ad-hMSCs.
To determine whether the overexpression of sRAGE in MSCs can affect their potential to differentiate into multiple cell lineages, MSCs were cultured in different conditions: adipogenic medium (Fig. 4, upper panel), osteogenic medium (Fig. 4, middle panel), or chondrogenic medium (Fig. 4, lower panel) for 2 or 3 weeks. MSCs, mock-MSCs, and sRAGE-MSCs showed similar patterns of differentiation into adipocytes, osteoblasts, and chondrocytes, which suggests that sRAGE-MSCs maintain their multipotency.

sRAGE-MSCs exert greater therapeutic activity in vivo.
We then investigated whether sRAGE-MSCs have increased immunomodulatory effects in vivo. The arthritis scores and arthritis incidence were compared between three groups: IL-1Ra-KO mice (control), mock-MSC-treated IL-1Ra-KO mice, and sRAGE-MSC-treated IL-1Ra-KO mice. sRAGE-MSCs exerted the greatest therapeutic effect in vivo compared with the control and mock-MSC-treated mice (Fig. 5A). Serum levels of total IgG, IgG1, and IgG3 were lower in mice treated with sRAGE-MSCs than in the other Histological examination showed that the paws and ankles from arthritis mice treated with sRAGE-MSCs exhibited the lowest degree of inflammation (Fig. 5C). Interestingly, compared with the controls, the expression of IL-17, IL-6, HMGB-1, and RAGE in inflamed joints was significantly higher in mock-MSC-treated IL-1Ra-KO mice and lower in joints of sRAGE-MSC-treated mice (Fig. 5D). Furthermore, the RAGE level in serum of controls was similar to that in mock-MSC-treated IL-1Ra-KO mice, whereas the concentrations of HMGB-1 in serum were significantly higher in mock-MSC-treated IL-1Ra-KO mice than in controls. (Fig. 5E). However, sRAGE-MSC treatment significantly reduced the expression of IL-17, IL-6, HMGB-1, and RAGE in inflamed joints and the concentrations of HMGB-1 and RAGE in serum.

sRAGE overexpression in Ad-hMSCs reciprocally regulates the Th17 and Treg cell balance.
T cells are implicated in the pathogenesis of RA, and IL-17 secreted by Th17 cells is a treatment target in RA. To identify whether Th17 and Treg cell populations were altered in IL-1Ra-KO mice treated with sRAGE-MSCs, IL-17-expressing (mainly Th17) and CD25 + Foxp3 + (mainly Treg) cells among CD4 + T cells in spleens were analyzed by confocal microscopy at 7 weeks after CII immunization. Spleens from mice treated with sRAGE-MSCs had more Treg cells and reciprocally fewer Th17 cells compared with spleens from control mice and mock-MSC-treated mice (Fig. 6A). STAT3 and STAT5 are critical transcription factors for Th17 and Treg cell differentiation, respectively. As shown in Fig. 6A, spleens from sRAGE-MSC-treated mice has significantly fewer pSTAT3Y705 (phosphorylated at tyrosine 705)-expressing and pSTAT3S727 (phosphorylated at serine 727)-expressing cells compared with the other two groups. By contrast, spleens from sRAGE-MSC-treated mice had significantly more STAT5-expressing CD4 + T cells (Fig. 6A). Flow cytometry showed that spleens of sRAGE-MSC-treated mice had significantly fewer interferon-γ (IFN-γ )-expressing CD4 + T (Th1) cells and IL-17-expressing CD4 + T (Th17) cells but more CD4 + CD25 + Foxp3 + (Treg) cells (Fig. 6B) compared with the other groups.
To examine the immunoregulatory properties of sRAGE-MSCs related to the control of Th17/Treg cells in vitro, activated murine CD4 + cells were stimulated with anti-CD3 and anti-CD28 monoclonal antibodies (mAbs) (Th0 condition) and then cocultured with medium only (nil), mock-MSCs, or sRAGE-MSCs, and followed by flow cytometry to analyze the expression of Foxp3 and IL-17. sRAGE overexpression in Ad-hMSCs inhibited the Th17 population and reciprocally induced Foxp3+ Treg differentiation in vitro (Fig. 6C) compared with the nil condition and mock-MSCs. The concentrations of IFN-γ and IL-17 in the culture supernatants were lowest in murine CD4 + T cells cocultured with sRAGE-MSCs compared with the other two groups of T cells (nil and mock-MSCs) (Fig. 6D). Next, we confirmed the immunoregulatory effects of sRAGE-MSCs in human CD4 + T cells that had been stimulated under the Th0 condition (anti-CD3 and anti-CD28 mAbs). sRAGE-MSCs inhibited the differentiation of Th1 and Th17 cells from naïve T cells and reciprocally increased the Foxp3 + Treg cells (Fig. 6E).

Discussion
Blockade of RAGE signaling can reduce inflammation, and sRAGE is a prominent target in the development of new treatments in inflammatory diseases. However, little is known about the biological roles of sRAGE in RA. Although MSCs are recognized as immunosuppressive agents in inflammatory disorders, they are also known to participate in the inflammatory process and in induction of immune responses 44 . Here, we investigated the therapeutic efficacy and underlying mechanisms of sRAGE-overexpressing MSCs in a murine model of inflammatory arthritis. The main observation of this study was that sRAGE-MSCs attenuated rheumatoid inflammation through reciprocal regulation of Th17 and Treg cells. To our knowledge, this is the first study to report data suggesting that sRAGE overexpression can optimize the immunotherapeutic potential of Ad-hMSCs; this may be a novel remedial strategy for the treatment of RA.
Although MSCs are considered to have potential for RA treatment by inhibiting the CII-reactive T cell responses 45 , naïve MSCs fail to elicit immunosuppressive and therapeutic functions in vivo 46 . It is well documented that the inflammatory micromilieu of RA can inhibit the therapeutic function of MSCs 21 . In our experiments, sRAGE-MSCs attenuated the clinical severity and histological changes of autoimmune arthritis, whereas naïve MSCs failed to show this effect. We also demonstrated that sRAGE overexpression in Ad-hMSCs maintained their migration and differentiation capacity. These data indicate that overexpression of sRAGE in Ad-hMSCs can refine their immunoregulatory potential while preserving their unique properties.
RAGE incites the inflammatory response by interacting with HMGB-1. Signal molecules whose release is initiated by RAGE bind to HMGB-1 and induce the expression of several proinflammatory cytokines 47 . HMGB-1 is a nonhistone nuclear protein that is ubiquitously present in eukaryotic cells 48,49 . It is released extracellularly by passive diffusion from necrotic cells and activates macrophages to trigger signals that drive the pathogenesis of autoimmune and inflammatory disorders by forming immunostimulatory complexes with cytokines and endogenous and exogenous factors [50][51][52] . The pathological role of HMGB-1 has been demonstrated convincingly. HMGB-1 levels in blood, synovium, and synovial fluids are increased in RA patients compared with OA patients and healthy people [53][54][55] . Animal studies have shown similar results 56 . HMGB-1 antagonists attenuate autoimmune arthritis in animal models, which supports the notion that HMGB-1 may be a treatment target in RA 57 . In our study, the serum and synovial concentrations of HMGB-1 were reduced in sRAGE-MSC-treated arthritic mice, whereas naïve MSC-treated mice showed significantly elevated HMGB-1 levels compared with vehicle-treated arthritic mice. These results may explain the lack of therapeutic efficacy of MSCs in RA.
Previous studies have shown that sRAGE inhibits the expression of inflammatory mediators in vitro and in vivo 58,59 . Moreover, administration of sRAGE exerts a therapeutic effect by downregulating chronic and severe inflammation in various animal models of inflammatory diseases such as atherosclerosis and inflammatory bowel disease 60,61 . A recent paper reported that sRAGE has therapeutic activity by attenuating the immune inflammatory response caused by HMGB-1 62 . The function of sRAGE is related to the inhibition of RAGE signaling in the inflammatory response 26 . In our investigation, sRAGE-overexpressing MSCs decreased the levels of inflammatory mediators and improved the symptoms of experimental arthritis. These results suggest that the therapeutic function of sRAGE-MSCs in arthritic mice can inhibit RAGE signaling in inflammatory conditions. Several proinflammatory cytokines including IL-6, IL-1β , and VEGF are involved in RA pathogenesis and therapy 63,64 . However, IDO, IL-10, hepatocyte growth factor (HGF), and TGF-β exert immunosuppressive activity and cause an anti-inflammatory response [65][66][67] . Autophagy increases cell survival by inducing lysosomal degradation of damaged cytoplasmic organelles or cytosolic aggregates 42 . Several recent studies have reported that induction of autophagy in MSCs can optimize their clinical application by inhibiting the prevention of apoptosis [28][29][30] . In our study, overexpression of sRAGE in Ad-hMSCs decreased the expression of HMGB-1, IL-6, IL-1β , and VEGF, and increased the number of autophagosomes and IL-10, HGF, and TGF-β expression. These results suggest that sRAGE-MSCs can inhibit inflammation while promoting immunoregulatory responses. Thus, optimizing the immunoregulating capacity of sRAGE-MSCs may improve their therapeutic function in autoimmune arthritis.
Th17 and Treg cells are important T-cell subsets that can regulate RA pathogenesis. Th17 cells play a key role in inflammation of autoimmune arthritis 68 . The balance between Th17 and Treg cells is distorted in the peripheral blood of RA patients 69 , and Th17/Treg cell imbalance is a significant treatment target in RA. Several lines of evidence indicate that restoring Th17/Treg cell balance improves the severity of RA and reduces joint inflammation [70][71][72] . Here, we found that sRAGE-MSCs ameliorated the clinical severity of inflammatory arthritis in mice by regulating the Th17/Treg cell balance and the corresponding transcription factors. These results suggest that the  Collectively, these results provide the first evidence of MSC optimization through sRAGE overexpression to suppress inflammatory disease. sRAGE-MSCs ameliorated the clinical severity of autoimmune arthritis by regulating the Th17/Treg cell balance and by inhibiting extracellular HMGB-1 activity compared with the activities of naïve MSCs. We conclude that this strategy to optimize the therapeutic potential of MSCs may help to control the inflammatory milieu of the rheumatoid synovium. The refined immunoregulatory function of sRAGE-MSCs identified in our research may have potential in the clinical applications of MSCs for cell therapy in RA.

Materials and Methods
Mice. IL-1Ra-KO mice with a BALB/c background were kindly provided by Prof. Yoichiro Iwakura (University

Induction of arthritis and treatment with sRAGE-MSCs.
To augment the arthritis severity, 0.1 mL of an emulsion containing 100 μ g of bovine CII and complete Freund's adjuvant (Arthrogen-CIA; Chondrex, Seattle, WA, USA) was injected intradermally into the base of the tail. After CII immunization, mice were injected intravenously with 1 × 10 6 Ad-hMSCs or sRAGE-MSCs once per week for 3 weeks. Clinical signs of arthritis in IL-1Ra-KO mice were monitored visually and scored twice per week as reported previously 73 . The final arthritis score was calculated as the sum of scores from all four limbs, which were assessed by three independent individuals blinded to the experimental groups.
Histological analysis. Histological analysis was conducted to determine the extent of joint damage. Mice joint tissues were fixed in 4% paraformaldehyde, decalcified in EDTA solution, embedded in paraffin, and sectioned. The sections were dewaxed using xylene and dehydrated through an alcohol gradient. Endogenous peroxidase activity was quenched with methanol-3% H 2 O 2 .Sections were stained routinely with hematoxylin-eosin. Culture of human adipose tissue-derived MSCs. Adipose tissue was digested with RTase (4 mL/g fat; K-STEM CELL Co., Ltd., Seoul, Korea) for 60 min at 37 °C. The digested tissues were filtered through a 100-μ m nylon sieve to remove cellular debris and then centrifuged at 470 × g for 5 min. The pellet obtained was resuspended in RCME cell attachment medium (K-STEM CELL) and cultured overnight at 37 °C in a humidified atmosphere with 5% CO 2 . After 24 h, the cultures were washed with phosphate-buffered saline (PBS) to remove nonadherent cells. The medium was changed to RKCM cell growth medium (K-STEM CELL) containing 5% fetal bovine serum (FBS; Invitrogen, Carlsbad, CA, USA). The cells were cultured for 4 days until 90% confluent (passage zero). The cells were expanded for two or three passages and used for experiments.

Construction of sRAGE vector and transfection.
A fragment of the human sRAGE gene was synthesized by GenScript Corporation, with codon optimization for expression in mammalian cells. The sequence was then subcloned into the HindIII and XhoI sites of pcDNA3.1+ (Invitrogen,Life Technologies, Grand Island, NY, USA). The vector constructs were transfected into MSCs using the X-tremeGENE HP Transfection Reagent (Roche, Mannheim, Germany), according to manufacturer's recommendations.

Flow cytometry.
Totalsplenocyteswere obtained from aseptically collected mousespleen described previously 74 . The cells were harvested in Roswell Park Memorial Institute (RPMI)-1640 medium supplemented with 5% fetal bovine serum. Mononuclear cells (1 × 10 6 ) were cultured without treatment as a negative control or with anti-CD3 and anti-CD28 (BD Pharmingen) for 3 days. The cells were stained with various combinations of fluorescence-tagged antibodies against CD4, CD25, IFN-γ (BD Biosciences), Foxp3, and IL-17 (eBioscience). Before intracellular staining, cells were restimulated for 4 h with phorbol myristate acetate (25 ng/mL) and ionomycin (250 ng/mL) in the presence of GolgiSTOP (BD Biosciences). Intracellular staining was performed using a kit (eBioscience) following the manufacturer's protocol. Stained cells were analyzed on a FACSCalibur apparatus (BD Biosciences).
Immunohistochemistry. Immunohistochemistry was performed using a VECTASTAIN ABC kit (Vector Laboratories, Burlingame, CA, USA). Tissue sections were incubated overnight at 4 °C with the primary antibodies against RAGE, HMGB-1, and IL-17, probed with a biotinylated secondary antibody, and then stained with a streptavidin-peroxidase complex for 1 h. The final color product was developed using DAB chromogen (Dako, Carpinteria, CA, USA). Statistical analysis. The data were analyzed using IBM SPSS Statistics 20 for Windows (IBM Corp., Armonk, NY, USA). One-way analysis of variance followed by Bonferroni's post hoc test was used to compare differences between three groups. The Mann-Whitney U test was used to compare numerical data between two groups. P < 0.05 was accepted as significant. Data are presented as mean ± standard deviation (s.d.).