Mesenchymal Stem/Stromal Cells Protect against Autoimmunity via CCL2-Dependent Recruitment of Myeloid-Derived Suppressor Cells

Mesenchymal stem/stromal cells (MSCs) have multiple regulatory effects on innate and adaptive immunity (1–3). Therefore, exogenous administration of MSCs is a promising tool for the treatment of immunological and inflammatory disorders. Indeed, a large number of clinical trials are under way to harness the cells for treating patients with autoimmune diseases such as type 1 diabetes, multiple sclerosis, Crohn’s disease, or systemic lupus erythematosus (http://www.clinicaltraial.gov).

Previous studies suggest that MSCs or the cell-derived factors exert their immunomodulatory functions through direct suppression of the immune system (1–3). However, exogenously administered cells are rapidly cleared by the host and do not engraft in the long-term (3, 4). Thus, direct effects of MSCs or the cell-derived factors on the immune system are short-lived and do not completely account for the immune modulation and tissue regeneration observed in vivo with MSC treatment. In this context, there are indications that MSCs activate endogenous immunoregulatory networks, for example, through the induction of regulatory T (Treg) cells, and thereby trigger immune tolerance (5–7).

Myeloid-derived suppressor cells (MDSCs) are another type of immune regulatory cell that expands during cancer, infection, or inflammation (8, 9). MDSCs comprise heterogeneous populations of immature myeloid cells (IMCs) and are defined functionally by immune suppressive properties and phenotypically by the coexpression of the myeloid lineage markers CD11b and Gr1 (Ly6G or Ly6C), as well as the lack of MHC class II (MHC II) molecule (8). Although MDSCs have a deleterious role in cancer by inhibiting antitumor immunity (10–13), it is possible that the cells have beneficial effects in autoimmunity by limiting T cell responses (8, 14). In support of this possibility, recent studies have demonstrated that MDSCs accumulate in tissues with autoimmune diseases, including multiple sclerosis (15–17), inflammatory bowel disease (18, 19), uveoretinitis (20), collagen-induced arthritis (21), and type 1 diabetes (22). However, the role of MDSCs in autoimmune diseases is not fully elucidated (14).

In the present study, we hypothesized that 1) MSCs stimulate the endogenous immunoregulatory system through induction of MDSCs, and 2) MDSCs induced by MSCs protect against autoimmunity. To address these hypotheses, we used a mouse model of experimental autoimmune uveitis (EAU), a well-established model for human autoimmune intraocular inflammation (23). We found that i.v. infused MSCs migrated to draining lymph nodes (dLNs) in mice with EAU and recruited MHC II^loLy6G⁻Ly6C^hiCD11b⁺ cells to dLNs in a CCL2-dependent manner. The Ly6C^hiCD11b⁺ cells potently suppressed Th1 and Th17 cell responses and ameliorated EAU. Thus, our findings demonstrate a novel mechanism of MSCs in the induction of the intrinsic immunoregulatory system and provide a potential role of MDSCs in the modulation of autoimmunity.

Six-week-old female B6 mice (C57BL/6J, H-2^b) were obtained from Orient Bio (Seongnam, Korea) and housed in a specific pathogen-free environment in the animal facilities of Biomedical Research Institute, Seoul National University Hospital. All animals were used at 6–8 wk of age. Animals were treated in strict accordance with the Association for Research in Vision and Ophthalmology statement for the use of animals in ophthalmic and vision research (http://www.arvo.org/About_ARVO/Policies/Statement_for_the_Use_of_Animals_in_Ophthalmic_and_Visual_Research). The experimental protocols were approved by the Institutional Animal Care and Use Committee of Seoul National University Biomedical Research Institute (approval no. 13-0104-C1A1).

Human bone marrow–derived MSCs (hMSCs) were obtained from the Center for the Preparation and Distribution of Adult Stem Cells (http://medicine.tamhsc.edu/irm/msc-distribution.html) that supplies standardized preparations of MSCs enriched for early progenitor cells under the auspices of an National Institutes of Health/National Center for Research Resources grant (P40 RR 17447-06). The acquisition and use of hMSCs were approved by the Institutional Review Board of Texas A&M Health Science Center and Seoul National University Hospital Biomedical Research Institute. The passage two hMSCs from one donor were used in all experiments. The cells consistently differentiated into three lineages in culture, were negative for hematopoietic markers (CD34, CD36, CD117, and CD45), and were positive for mesenchymal markers CD29 (95%), CD44 (>93%), CD49c (99%), CD49f (>70%), CD59 (>99%), CD90 (>99%), CD105 (>99%), and CD166 (>99%). The cells were cultured in complete culture medium (CCM) containing 17% FBS (Life Technologies, Grand Island, NY), 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin (Invitrogen/Life Technologies, Carlsbad, CA) until 70% confluence was reached and then harvested with 0.25% trypsin/1 mM EDTA (Invitrogen/Life Technologies) at 37°C for 2 min. After washing, the cells were used for in vitro experiments or resuspended in balanced salt solution (BSS, Lonza/BioWhittaker, Walkersville, MD) at a concentration of 10,000 cells/μl for injection in vivo.

The mouse MSCs were prepared from the marrow of long bones of hindlimbs of 8-wk-old female B6 mice using the previously described method (24) with modification. Briefly, the bone marrow extracts were obtained by flushing the cut ends of the bones, and cell suspension was filtered through a 70-μm strainer to remove bone spicules. After centrifugation, the cells were resuspended in CCM and distributed into culture plates at a density of 1 × 10⁶ cells/cm². The plates were cultured undisturbed at 37°C with 5% CO₂ in a humidified chamber (normoxia condition) for 3 d. Then, the nonadherent cells were removed by gentle swirling, and adherent cells were replaced with CCM. After an additional 4 d of culture, the plates were washed with serum-free medium and harvested with 0.25% trypsin/1 mM EDTA at 37°C for 2 min before injection.

EAU was induced in C57BL/6J with H-2^b haplotype mice using the standardized method as previously described (25). Briefly, mice were immunized with s.c. injection into a footpad of 250 μg human interphotoreceptor retinal binding protein (IRBP) peptide 1–20, GPTHLFQPSLVLDMAKVLLD (20 mg/ml; Peptron, Daejeon, Korea) that was emulsified in CFA (Sigma-Aldrich, St. Louis, MO) containing Mycobacterium tuberculosis (2.5 mg/ml; BD Difco, Franklin Lakes, NJ). Simultaneously, the mice received an i.p. injection of 0.7 μg pertussis toxin (300 μl; Sigma-Aldrich).

Tissues were fixed in 10% formaldehyde and embedded in paraffin. Serial 4-μm-thick sections were cut, deparaffinized, and stained with either H&E or Abs according to the manufacturers’ protocols. The Abs used were mouse anti-human nuclei (Cy3 conjugate, 1:50) (MAB1281C3, Millipore, Billerica, MA), anti-human mitochondria (1:50) (MAB1273, Millipore), and Alexa Fluor 488 goat anti-mouse IgG (1:500) (A11001, Molecular Probes/Life Technologies, Eugene, OR). A DAPI solution (Vectashield mounting medium; Vector Laboratories, Burlingame, CA) was used for counterstaining. Retinal pathology scores were assessed in the H&E-stained slides on a scale of 0–4 using the criteria previously defined by Caspi (23).

Tissues were lysed in RNA isolation reagent (RNA-Bee; Tel-Test, Friendswood, TX) and sonicated with an ultrasound sonicator (ultrasonic processor; Cole-Parmer Instruments, Vernon Hills, IL). Total RNA was extracted using an RNeasy mini kit (Qiagen, Valencia, CA), and double-stranded cDNA was synthesized by reverse transcription (high-capacity RNA-to-cDNA kit; Applied Biosystems/Life Technologies, Carlsbad, CA). Real-time amplification was performed using TaqMan Universal PCR Master Mix (Applied Biosystems/Life Technologies). Human- or mouse-specific PCR probe sets were purchased from Applied Biosystems (TaqMan gene expression assay kits).

A standard curve was generated by adding serial dilutions of hMSCs to mouse tissue as previously described (4). Briefly, 10–100,000 hMSCs were added to one mouse popliteal LN, respectively. Following RNA extraction, cDNA was generated using 1 μg total RNA, and real-time amplification was performed using a human-specific GAPDH primer and probe set (TaqMan gene expression assays, ID no. GAPDH HS99999905_05). The values were normalized to total eukaryotic 18S rRNA (TaqMan gene expression assays, ID no. Hs03003631_g1). The standard curve was made based on human-specific GAPDH expression from a known number of hMSCs added to one mouse LN.

A single-cell suspension was collected by mincing the tissue between the frosted ends of two glass slides in RPMI 1640 medium (Welgene, Daegu, Korea) containing 10% FBS (Life Technologies) and stained with the following fluorescence-conjugated anti-mouse Abs: CD4, CD25, Foxp3, IFN-γ, IL-17, IL-10, CD11b, Ly6C, Ly6G, MHC II (H-2^b) (all from eBioscience, San Diego, CA). For intracellular staining, the cells were stimulated for 5 h with 50 ng/ml PMA and 1 μg/ml ionomycin in the presence of GolgiPlug (BD Pharmingen, San Diego, CA) and stained. The cells were then assayed for fluorescence using a FACSCanto flow cytometer (BD Biosciences, San Diego, CA). Data were analyzed using the FlowJo program (Tree Star, Ashland, OR).

Tissues were cut into small pieces and lysed in Pro-Prep protein extraction solution (Intron Biotechnology, Seongnam, Korea). The samples were sonicated on ice with an ultrasound sonicator (Cole Parmer Instruments), and the cell-free supernatant was collected after centrifugation at 12,000 rpm for 15 min. The supernatants were assayed for the concentration of IL-17 (mouse DuoSet kit; R&D Systems, Minneapolis, MN), IFN-γ (mouse DuoSet kit; R&D Systems), and CCL2 (human Quantikine set; R&D Systems) according to the manufacturer’s protocols.

A single-cell suspension prepared from LN and blood of EAU-immunized mice after RBC lysis was stained with anti-CD11b and anti-Ly6C Abs (eBioscience) and sorted using a flow cytometer (BD FACSAria III cell sorter; BD Biosciences) (purity of >95%). CD4⁺ T cells were purified from LNs of B6 mice using CD4 MACS beads (Miltenyi Biotec, Bergisch Gladbach, Germany) in accordance with the manufacturer’s instructions (purity of >95%). The cells were cultured in RPMI 1640 media (Welgene) containing 10% FBS (Life Technologies), 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin (Invitrogen/Life Technologies).

The cells were collected on microscopic slides using a cytospin machine and stained with Giemsa solution (GS-500; Sigma-Aldrich).

For proliferation assays, CD4⁺ T cells purified from LNs were labeled with 5 μM CFSE (1 μl/ml; Invitrogen/Life Technologies) in RPMI 1640 medium (Welgene) for 10 min at 37°C in the dark. After washing, the CFSE-labeled CD4⁺ T cells were cultured in RPMI 1640 medium containing 10% FBS (Life Technologies) in plates coated with 2.5 μg/ml anti-CD3 and 2.5 μg/ml anti-CD28 mAbs (eBioscience) with or without Ly6C^hiCD11b⁺ or Ly6C^loCD11b⁺ cells in a Transwell for 5 d. The CFSE fluorescence was measured in CD4⁺ T cells with flow cytometry (FACSCanto flow cytometer; BD Biosciences).

For apoptosis assays, CD4⁺ T cells cultured with or without Ly6C^hiCD11b⁺ or Ly6C^loCD11b⁺ cells in a Transwell for 5 d were stained with a combination of propidium iodide (PI; Molecular Probes/Life Technologies) and annexin V (ANX; FITC annexin V apoptosis detection kit; BD Pharmingen) and analyzed by a FACSCanto flow cytometer.

For differentiation assays, CD4⁺ T cells were cultured in anti-CD3/anti-CD28–coated plates with or without Ly6C^hiCD11b⁺ or Ly6C^loCD11b⁺ cells in a Transwell. To stimulate differentiation into Th1 cells, 20 ng/ml recombinant murine IL-12 and 2.5 μg/ml anti–IL-4 Ab (both from R&D Systems) were added to the culture. To stimulate differentiation into Th17 cells, 50 ng/ml recombinant mouse IL-6, 5 ng/ml recombinant mouse TGF-β1, and Abs against IFN-γ (2.5 μg /ml) and IL-4 (2.5 μg /ml) (all from R&D Systems) were added. After 3 d of culture, CD4⁺ T cells were restimulated, and the surface marker or intracellular cytokine staining was performed as described above. The cells were analyzed for expression of CD4, CD25, Foxp3, IL-17A, and IFN-γ by a FACSCanto flow cytometer (BD Biosciences).

For adoptive transfer experiments, Ly6C^hiCD11b⁺ cells isolated as mentioned above were injected into EAU mice through the tail vein (5 × 10⁵ cells/mouse).

For a depletion study, 100 μg anti-mouse Gr-1 mAb (clone RB6-8C5; StemCell Technologies, Carlsbad, CA) or rat IgG2a isotype control (553926; BD Pharmingen) was i.p. injected to mice three times at days −1 (24 h prior to immunization), 0 (the day of immunization), and 1 (the day after immunization) (26–30).

The hMSCs were transfected with small interfering RNA (siRNA) for CCL2 (sc-43913; Santa Cruz Biotechnology, Santa Cruz, CA) or scrambled siRNA (sc-37007; Santa Cruz Biotechnology) with a commercial kit (Lipofectamine RNAiMAX, Invitrogen/Life Technologies) per the manufacturers’ instructions. The knockdown efficiency of CCL2 in hMSCs was 86–90%.

In vitro migration of Ly6C^hiCD11b⁺ cells isolated from mice as above was evaluated in 24-well plates with 3 μm pore Transwell supports (CytoSelect 24-well cell migration assay; Cell Biolabs, San Diego, CA). One million mouse MDSCs (>95% Ly6C^hiCD11b⁺ cells) were plated in 300 μl serum-free CCM in the upper compartment, and 500 μl human CCL2 (100 ng/ml, 279-MC-010/CF; R&D Systems) was added to the lower compartment. Plates were incubated at 37°C with 5% CO₂ for 18 h. The cells that migrated to the underside of the filter and lower compartment were quantified using CyQuant GR fluorescent dye (Cell Biolabs) according to the manufacturer’s protocol.

The data are presented as mean ± SD or SEM. Comparisons of two values between the groups were made using the two-tailed Student t test, and comparisons of more than two means were made using a one-way ANOVA (Prism; GraphPad Software, La Jolla, CA). Differences were considered significant at p < 0.05.

To determine whether MSCs protect the retina from EAU, we immunized mice with s.c. injection of the retina-specific Ag IRBP into footpads and injected hMSCs (1 × 10⁶ cells/mouse) or the vehicle (BSS) into tail vein immediately after immunization. At days 1, 7, 14, and 21 postimmunization, the mice were humanely killed and the tissues were extracted for analysis (Fig. 1A).

Histologic examination showed that the disease appeared in the eyes of EAU-immunized mice at day 14 and reached a peak by day 21, as demonstrated by inflammatory cell infiltration, photoreceptor layer disorganization, retinal folds, and edema (Fig. 1B, 1C). In contrast, there were little structural damage and inflammatory infiltrates in the eyes of hMSC-treated mice (Fig. 1B, 1C). Disease scores assigned by retinal pathology (23) were markedly lower in the hMSC-treated mice compared with the BSS-treated mice (Fig. 1C).

We further measured the levels of proinflammatory cytokines in the eyes. Real-time RT-PCR showed that transcript levels of IFN-γ, IL-17A, IL-2, IL-6, and IL-1β were the highest at day 14, and the level of TNF-α transcript was increased at day 21 (Fig. 1D). The levels of the inflammatory cytokines in the eyes were significantly lower in the hMSC-treated mice than in the BSS-treated mice (Fig. 1D).

Similar results were obtained with mouse bone marrow-derived MSCs (mMSCs) or hMSCs from a second donor. An i.v. injection of mMSCs or second donor-derived hMSCs was also effective in preventing retinal damage and reducing the levels of inflammatory cytokines in the eyes of mice with EAU (Supplemental Fig. 1A, 1B).

Taken together, the results demonstrate that a single i.v. injection of MSCs at the time of immunization blocks the development of EAU.

Th1 and Th17 cells are central to the pathogenesis of autoimmunity involving autoimmune uveitis (31, 32). We therefore investigated whether hMSCs ameliorate EAU through the suppression of Th1 and Th17 cells.

By monitoring IFN-γ⁺CD4⁺ cells and IL-17⁺CD4⁺ cells in popliteal LNs (dLNs), blood, and retina with flow cytometry, we observed three temporally distinct phases: an initial expansion of Th1 and Th17 cells in dLNs during the postimmunization days 1–7, a subsequent increase of Th1 and Th17 cells in the blood at day 7, and a later infiltration of the cells in the retina with a peak at day 14 (Fig. 2A, 2B). An i.v. infusion of hMSCs markedly reduced the number of IFN-γ⁺CD4⁺ and IL-17⁺CD4⁺ cells in dLNs at days 1 and 7 (Fig. 2A, 2B). Similarly, the number of IFN-γ⁺CD4⁺ and IL-17⁺CD4⁺ cells in the blood and retina was significantly decreased by i.v. hMSCs (Fig. 2B).

The differentiation of Th1 and Th17 cells from naive CD4⁺ T cells is directly regulated by cytokines that are produced by activated dendritic cells (DCs) or monocytes/macrophages (33–37). It has been reported that MSCs can repress the activation of DCs or monocytes/macrophages (2). Therefore, we further examined whether hMSCs diminish the levels of cytokines derived from DCs or monocytes/macrophages and thereby inhibit Th1 and Th17 differentiation. Consistent with the flow cytometric data shown in Fig. 2B, the levels of IFN-γ and IL-17A, which are Th1 and Th17 cell–derived cytokines, were significantly lower in dLNs of the hMSC-treated mice at days 1 and 7, compared with the BSS-treated controls (Fig. 2C). However, there were no differences in the levels of IL-1β, IL-6, IL-12A, IL-12B, and IL-23 in dLNs, which are cytokines derived from DCs or monocytes/macrophages and direct Th1 and Th17 differentiation (33–37), either at day 1 or 7 between the hMSC- and BSS-treated mice (Fig. 2C).

Previous studies have reported that MSCs induce Treg cells and inhibit T cell proliferation and differentiation (5–7). In our experimental setting, the number of CD4⁺CD25⁺Foxp3⁺ cells was increased by EAU immunization (Supplemental Fig. 2). However, the number of CD4⁺CD25⁺Foxp3⁺ cells in dLNs, blood, or spleen was not different with or without i.v. hMSCs (Supplemental Fig. 2). Thus, the results indicate that Treg cells do not mediate the effects of hMSCs in suppressing EAU.

Recent studies have identified the novel type of IMCs, known as MDSCs, that restrain T cell activation. MDSCs consist of two main subsets: Ly6G⁺Ly6C^loCD11b⁺ cells (granulocytic MDSCs) and Ly6G⁻Ly6C^hiCD11b⁺ cells (monocytic MDSCs) (38–40). It has been suggested that monocytic MDSCs are the main immunosuppressive myeloid cells (15–17, 26, 38, 41). Thus, we next explored the possibility that i.v. infusion of MSCs might affect the population of monocytic MDSCs.

We conducted a time course analysis of MHC II^loLy6G⁻Ly6C^hiCD11b⁺ cells in dLNs, blood, spleen, lung, and retina after EAU immunization with or without i.v. hMSCs. Of note was the finding that there was a dramatic increase of MHC II^loLy6G⁻Ly6C^hiCD11b⁺ cells in dLNs of the hMSC-treated mice at the postimmunization day 1 (Fig. 3A, 3B). Similar findings were observed with mMSCs (Supplemental Fig. 1C). The MHC II^loLy6G⁻Ly6C^hiCD11b⁺ cells were increased in the blood of the hMSC-treated mice, but not in the spleen, lung, or retina (Fig. 3B). Also, the MHC II^loLy6G⁻Ly6C^hiCD11b⁺ cells that were increased by hMSCs expressed a high level of an immunoregulatory cytokine, IL-10 (Fig. 3A, 3C).

We further examined the gene expression profile of cytokines that are known to mediate the immunosuppressive functions of monocytic MDSCs: inducible NO synthase (iNOS) and arginase (Arg1) (9, 38–40, 42, 43). We found ∼80-fold and 400-fold increases of iNOS and Arg1 transcripts, respectively, in dLNs of the hMSC-treated mice, compared with the BSS-treated controls (Fig. 3C). Also, there was a significant upregulation in dLNs of mouse-specific transcripts for PG-endoperoxide synthase 2, M-CSF, and GM-CSF by i.v. hMSCs, all of which mediate MDSC functions and induce MDSC expansion (Fig. 3C) (8, 38, 44, 45).

To directly test the effects of hMSC-induced MDSCs on T cell proliferation and apoptosis in vitro, we isolated Ly6C^hiCD11b⁺ and Ly6C^loCD11b⁺ cells from dLNs and peripheral blood of the hMSC-treated mice at day 1 after immunization. The CFSE-labeled CD4⁺ cells purified from popliteal LNs of naive mice were stimulated with anti-CD3/anti-CD28 mAbs and cocultured with either Ly6C^hiCD11b⁺ or Ly6C^loCD11b⁺ cells in a Transwell system for 5 d. Flow cytometric analysis of CFSE dilution revealed that CD4⁺ cell proliferation was significantly inhibited by coculture with Ly6C^hiCD11b⁺ cells (Fig. 4A). Additionally, PI and ANX staining showed that the number of PI⁺ANX⁺ cells was significantly increased in CD4⁺ cells upon coculture with Ly6C^hiCD11b⁺ cells compared with coculture with Ly6C^loCD11b⁺ cells, indicating that CD4⁺ cell apoptosis was elicited by Ly6C^hiCD11b⁺ cells (Fig. 4B).

Next, we tested whether hMSC-induced MDSCs affect T cell differentiation toward Th1 and Th17 cells. The anti-CD3/anti-CD28–activated CD4⁺ cells were primed with Th1 or Th17 polarizing conditions and cocultured with either Ly6C^hiCD11b⁺ or Ly6C^loCD11b⁺ cells in a Transwell for 3 d. The number of IFN-γ⁺CD4⁺ and IL-17⁺CD4⁺ cells was significantly lower in CD4⁺ cells cocultured with Ly6C^hiCD11b⁺ cells than in those cocultured with Ly6C^loCD11b⁺ cells (Fig. 4C, 4D). However, coculture with Ly6C^hiCD11b⁺ did not increase the number of CD4⁺CD25⁺Foxp3⁺ cells among CD4⁺ cell population (Supplemental Fig. 3), which is consistent with the in vivo finding that conventional Treg cells (CD4⁺CD25⁺Foxp3⁺ cells) were not affected by hMSCs in EAU mice (Supplemental Fig. 2).

Therefore, the data collectively demonstrate that i.v. administration of hMSCs recruits monocytic MDSCs capable of inhibiting T cell proliferation and Th1/Th17 differentiation and inducing T cell apoptosis to dLNs of mice with EAU.

To confirm the effects of hMSC-induced MDSCs on EAU in vivo, we adoptively transferred Ly6C^hiCD11b⁺ cells that were isolated from dLNs and blood of hMSC-treated EAU mice through tail vein injection (5 × 10⁵ cells/mouse) right after immunization. Wright–Giemsa staining showed that the isolated Ly6C^hiCD11b⁺ cells were largely mononuclear cells (Fig. 5A). EAU was markedly suppressed in mice treated with Ly6C^hiCD11b⁺ cells as demonstrated by decreased inflammatory infiltration in the retina (Fig. 5B) and reduced the levels of the proinflammatory cytokines in the eye (Fig. 5C). In accordance with the reduced disease severity, the number of IFN-γ⁺CD4⁺ and IL-17⁺CD4⁺ cells in dLNs was significantly lowered by adoptive transfer of Ly6C^hiCD11b⁺ cells (Fig. 5D, 5E).

To further evaluate whether MDSCs are required for the action of hMSCs in suppressing EAU, we depleted mice of Ly6C^hiCD11b⁺ cells by i.p. injections of neutralizing Abs against Gr-1 (RB6-8C5) three times at days −1 (24 h prior to immunization), 0 (the day of immunization), and 1 (the day after immunization) (26–30). Immediately after EAU immunization, hMSCs were injected into the mice (Fig. 5F). The effects of hMSCs in attenuating pathological features of EAU and reducing the levels of the proinflammatory cytokines in the eye were less marked in the anti-Gr-1–treated mice, compared with the isotype IgG-treated controls (Fig. 5F, 5G).

Taken together, these results suggest that i.v. hMSCs recruit monocytic MDSCs into dLNs of mice with EAU and thereby prevent the development of EAU.

To explore the mechanisms by which hMSCs recruit MDSCs to dLNs of EAU mice, we evaluated the engraftment of hMSCs in dLNs. Quantitative RT-PCR assays for human-specific GAPDH revealed that ∼1.8% of hMSCs were present in dLNs of the hMSC-treated EAU mice at day 1 after i.v. infusion (Supplemental Fig. 4A). Additionally, immunohistochemical staining for human-specific mitochondria and human nuclear Ag confirmed the presence of hMSCs in dLNs of the EAU-immunized mice (Fig. 6A).

Previous studies have demonstrated that the chemokine CCL2 and its major receptor CCR2 signaling is critical for the recruitment of Ly6C^hiCD11b⁺ monocytes to tumors or injured tissues (46–49). Also, it has been shown that MSCs suppress inflammation or promote tumors by producing CCL2 (49–51). Based on these observations, we next investigated whether hMSCs exert their EAU-suppressing effects by attracting MDSCs to dLNs through the production of CCL2. To prove this hypothesis, we first investigated whether the recruitment of MHC II^loLy6G⁻Ly6C^hiCD11b⁺ cells is dependent on CCL2 produced by hMSCs. First, we confirmed that human CCL2 is capable of attracting mouse Ly6C^hiCD11b⁺ cells in vitro (Supplemental Fig. 4B). Next, we knocked down the expression of CCL2 in hMSCs by transient transfection with siRNA and injected the cells with CCL2 knockdown into EAU-immunized mice (Fig. 6B). The level of human CCL2 was elevated in dLNs and plasma at the postimmunization day 1 in the mice that received hMSCs with scrambled siRNA, and significantly decreased in those that received hMSCs with CCL2 siRNA (Fig. 6B). Of note, assays of dLNs showed that i.v. infusion of hMSCs with CCL2 siRNA did not increase the number of MHC II^loLy6G⁻Ly6C^hiCD11b⁺ cells in dLNs at day 1, whereas hMSCs with scrambled siRNA markedly increased the number of MHC II^loLy6G⁻Ly6C^hiCD11b⁺ cells (Fig. 6C). We next evaluated whether hMSCs with CCL2 knockdown are still capable of attenuating EAU. As expected, hMSCs with CCL2 siRNA did not diminish IFN-γ⁺CD4⁺ or IL-17⁺CD4⁺ cells in dLNs (Fig. 6D), and they were not effective in suppressing intraocular inflammation and preventing retinal damage in mice with EAU (Fig. 6E, 6F).

Overall, the data demonstrate that hMSCs mobilize MDSCs to dLNs in a CCL2-dependent manner and thereby prevent the development of EAU.

Our data demonstrate a novel mechanism by which MSCs modulate autoimmunity: activation of the endogenous immunoregulatory system of myeloid cells.

MDSCs are a heterogeneous group of IMCs that are characterized by their myeloid origin, immature state, and most importantly by potent biological activities to suppress T cells (8). Under physiological conditions, IMCs generated in the bone marrow quickly differentiate into mature myeloid cells, including granulocytes, macrophages, or DCs. Conversely, in pathological states such as cancer, infection, trauma, or autoimmune diseases, MDSCs dramatically expand, accumulate in peripheral lymphoid organs, and migrate to sites of the disease where they contribute to immune suppression. Because MDSCs were originally described in tumor-bearing mice nearly two decades ago (52, 53), the cells have been predominantly explored in the setting of tumors as one of the major factors responsible for tumor-associated immune defects and as an attractive target for developing cancer immunotherapy through the inhibition of MDSC expansion or function. However, as we demonstrated in this study, MDSCs, which serve a deleterious role in cancer and chronic infection, can have favorable effects on autoimmunity and transplantation. In support for the potential benefits of MDSCs in immune-related diseases, several studies demonstrated that MDSCs suppress immune responses in models for multiple sclerosis (15–17) and autoimmune arthritis (21), and induce immune tolerance of kidney or cardiovascular allografts (54, 55).

Our observation that MSCs suppress immune responses by recruiting MDSCs is consistent with the findings reported by Ren et al. (49), who found that tumor-infiltrating MSCs or TNF-α–pretreated MSCs recruit CD11b⁺Ly6C⁺ monocytes to tumors and thereby facilitate tumor growth in lymphoma-bearing mice. Although they did not directly test the effects of CD11b⁺Ly6C⁺ monocytes on immune cells, Ren et al. convincingly showed that in an inflammatory environment, MSCs repress the adaptive immune system and recruit monocytes through the production of CCL2. In our study, we further demonstrated that monocytes recruited by MSCs are immature as shown by the lack of MHC II Ag, and they suppress T cell responses in vitro and in vivo, all of which are characteristic of MDSCs. Of note was that MDSCs recruited by MSCs ameliorated EAU in our study, whereas they promoted tumors in the study by Ren et al. These results suggest that the bipartite interaction between MSCs and monocytes can have positive or negative effects depending on the disease. Specifically, in our Transwell coculture assays, monocytic MDSCs recruited by MSCs potently inhibited CD4⁺ T cell proliferation and differentiation and induced apoptosis, indicating that MDSCs act through the release of soluble factors. Also, in dLNs of EAU mice treated by MSCs, an increase in monocytic MDSCs was accompanied by a huge upregulation of iNOS and Arg1. In agreement with our findings, previous studies have reported that monocytic MDSCs suppress Ag-independent T cell responses without direct cell–cell contact and through the expression of iNOS, arginase, and other immunoregulatory cytokines (8, 38–40, 42, 43). This function of monocytic MDSCs contrasts with that of granulocytic MDSCs, which use reactive oxygen species and suppress T cells through Ag-specific interaction via close cell–cell contact (8, 38, 56, 57). However, given a marked functional heterogeneity and plasticity of MDSCs (16, 38), the mechanisms implicated in T cell suppression by MSC-induced MDSCs remain to be clarified.

The CCL2/CCR2 signaling axis has been implicated in the trafficking of tumor-associated macrophages and MDSCs into tumors, hence supporting tumor growth via immunosuppressive mechanisms (58). Moreover, tumor cell–derived CCL2 has been shown to stimulate metastasis and angiogenesis (48, 59), whereas CCL2- or CCR2-blocking agents improve therapeutic responses to immunotherapeutic regimens (47, 60). We found in the present study that CCL2 produced by MSCs mobilizes immunosuppressive MDSCs into dLNs of EAU-immunized mice and modulates Th1/Th17 autoimmune responses, subsequently preventing EAU. Similarly, previous studies showed that MSCs ameliorate experimental autoimmune encephalitis through a metalloproteinase-mediated paracrine proteolysis of CCL2 leading to PD-L1 and inhibition of Th17 cells (51). These findings suggest that CCL2-dependent immune suppression by MSCs can aggravate tumors, whereas it can attenuate autoimmune disorders. Therefore, importantly, note that a therapeutic approach using MSCs can be detrimental or beneficial depending on the context of a disease. Our data echo the concern that MSC therapies should be used in clinical trials with caution, based on more complete understanding of their mechanisms of action (61).

One of the unsolved questions is the mechanism on how MSCs home to injured tissues. We here found that ∼1.8% of i.v. hMSCs were present in dLNs of EAU-immunized mice at day 1, concomitant with the afferent phase of immune responses where inflammation occurs predominantly in dLNs. Previous experiments indicate that most MSCs are entrapped in the lung after i.v. infusion, and ∼10% of hMSCs are detected in the lung at day 1 (4). However, our results demonstrate that MSCs did not increase MDSCs in the lung of EAU-immunized mice as opposed to a remarkable increase of MDSCs in dLNs. Therefore, it is likely that MSCs in an inflammatory microenvironment, that is, dLNs in EAU-immunized mice, are activated to secrete CCL2 and mobilize immunosuppressive myeloid cells into the site of inflammation. Further investigations are needed to elucidate the mechanisms responsible for the homing of i.v. administered MSCs to sites of diseases and licensing of MSCs to exhibit their immunomodulatory features.

A previous study reported that syngeneic or allogeneic rat MSCs dramatically reduced the severity of EAU induced by IRBP in rats when the cells were administered before the onset or at the peak of disease, but not after disease stabilization (62). Although in the present study we evaluated the effects of MSCs injected at the time of EAU immunization, additional investigation on the effects and mechanisms of MSCs injected at later time points would be beneficial for clinical translation.

In conclusion, our study provides a novel insight into how MSCs, which are resident stromal cells in the bone marrow or other tissues, modulate excessive activation of lymphoid immune cells and maintain tissue homeostasis. The cross-talk between MSCs and endogenous IMCs appears to be one of the immunoregulatory mechanisms of MSCs. A better understanding of the complexity and significance of this bipartite interaction in different disease contexts would be critical for successful and safe translation of MSCs into the clinic.

The authors have no financial conflicts of interest.