All the culture products were provided by ThermoFisher (France) unless mentioned. In this study, the induced pluripotent stem (huiPS) cells were provided by Dr. I. Petit (INSERM U976, Paris) obtained from the reprogramming of human adult fibroblasts (29) or were produced in the laboratory (30). These cells were grown into homogeneous colonies on feeder mouse embryonic fibroblasts (MEFs) treated with mitomycin C (Sigma, France). The culture medium of huiPS cells consisted in 85% DMEM/F12, 15% knockout serum replacement, l-glutamine 100 mM, β-mercaptoethanol 0.1 mM, and bFGF 10 ng/ml (Invitrogen or Peprotech, France). The huiPS cells were passaged one to two times per week by splitting colonies in dissociation buffer (DMEM containing Collagenase type IV 2 mg/ml) without detaching the feeder MEF.

Human iPS-derived mesenchymal stromal cells (huiPS-MSC) were obtained by spontaneous differentiation of huIPS cells. For this, huiPS cells were maintained in huiPS medium without bFGF until the huiPS colonies overgrew. Without passaging them, the differentiating cells were maintained for the next 4–6 days in an “MSC” culture medium containing 30% DMEM, 30% F12, 10% serum FcII (Hyclone, ThermoFisher, France), NEAA 10 mM, Na pyruvate 1 mM, penicillin (1 U/ml)/streptomycin (1 μg/ml), glutamine 1 mM, β-mercaptoethanol 100 μM, ascorbic acid 50 μg/ml (Sigma-Aldrich, France), and huEGF 10 ng/ml. The differentiating cells were then dissociated in PBS 0.05% trypsin–EDTA, and put back in culture in the “MSC” medium. Only few cells collected (<10%) were able to survive and grow (medium was changed 1 or 2 days later to removed dead cells and floating cells). The “MSC” medium was then changed every 3–4 days. Ten to fifteen days later, the cells were passaged (passage 1) and analyzed for the expression of MSC markers (usually 80–90% of cells with MSC phenotype are recovered). These huIPS-MSCs were then maintained in culture and passaged one to two times per week in “MSC” medium. Although the huiPS-MSCs obtained could be maintained up to passage 10, they were used before they reached passage 5.

Human PBMCs were obtained from the EFS (Etablissement Français du Sang) from healthy platelet donors. After separation on a Ficoll gradient, the cells were immediately used or frozen and stored in liquid nitrogen.

Human-induced pluripotent stem-MSCs’ differentiation into adipocytes and chondrocytes was performed using specific differentiation media (StemProAdipocyte medium and StemProChondrogenesis medium, ThermoFisher) according to the manufacturer’s instructions. Osteoblast differentiation was performed in αMEM medium containing 5% Hyclone serum supplemented with ascorbic acid 50 μg/ml, β-glycerophosphate 110 μM, and dexamethasone 0.1 μM. After 14–21 days of culture and fixation, the cells were treated with either Alcian Blue 1% for coloration of chondrocyte matrix, Alizarin Red 2% for osteoblast-derived matrix or Red Oil for presence of lipid droplets as a marker of adipocytes.

The cells were labeled for expression of surface markers with fluorescent antibodies in PBS plus 1% FCS and 2 mM EDTA for 30 min at 4°C and were analyzed on a FACS Canto II with the Diva (Becton-Dickinson, France) or the FlowJo (FlowJo, LLC) softwares. The anti-CD90 FITC (clone 5E10), anti-CD105 PeCy7 (clone SN6), and anti-CD3 eFluor450 (clone UCHT1) antibodies were provided by eBioscience SA (France). The anti-CD73 APC (clone AD2), anti-CD4 PeCy7 (clone SK3), and anti-CD8 PerCP-Cy5 (clone SK1) antibodies were provided by Becton-Dickinson.

For intracytoplasmic staining of cytokine expression, human T cells were stimulated for 4 h with Ionomycin (10 μg/ml) and PMA (1 μg/ml) in the presence of Brefeldin A (10 μg/ml) [as already described (31)]. T cells were labeled with the anti-CD4 PeCy7 and anti-CD8 PerCP-Cy5 antibodies, fixed and permeabilized with the Fix/Perm buffer (eBiosciences). They were incubated with antibodies specific (from eBioscience) for the following cytokines: IFNγ (clone 45.B3), TNFα (clone MAb11), and IL-10 (clone Jes3-9D7). A nuclear labeling of fixed and permeabilized human CD4⁺ and CD8⁺ T cells was performed to detect the expression of the FoxP3 transcription factor (APC, clone 259D/C7, Becton-Dickinson). In some experiment, the following antibodies were used CD3 BV510, CD4 PerCP-Cy5, CD127 PeCy7, Neuropilin (Nrp1) BB 515, Helios AF 488, and FoxP3 APC (all from Becton-Dickinson).

Responder PBMC labeled with 0.4 μM CFSE (ThermoFisher) was stimulated by allogeneic manner in MLR with irradiated (40 Gy) PBMCs from two different donors labeled with 0.4 μM APC-Cell Tracer (eBioscience) and cocultured with or without irradiated (60 Gy) huiPS-MSCs (ratio 1 huiPS-MSC for 10 immune cells) in a u-bottom 96 well plates for 5 days. The cells then were collected and labeled with the anti-CD4 PeCy7 and anti-CD8 PerCP antibodies and then analyzed by flow cytometry. We excluded stimulator cells stained with the APC-Cell Tracer. The proliferation of CD4⁺ and CD8⁺ responder T cells was measured by the dilution of the CFSE marker. In some experiments, the PBMCs were put in the insert of a Transwell culture system (Nunc, 0.45 μm pores), the huiPS-MSC being put in the lower part. A rat anti-human blocking IL-10 antibody (clone JES3-19F1, Becton-Dickinson) and an isotype rat control were used (10 μg/ml) to inhibit the potential role of IL-10 in the immunosuppression.

NOD/SCID/IL2Rγ KO (NSG) mice were purchased from Charles River Laboratory. Animals were maintained in accordance with the general guidelines of the institute. They were injected ip with 10 × 10⁶ human PBMC and were treated with or without 1 × 10⁶ huiPS-MSCs by ip injection once a week for 3 weeks. In some experiments, mice received only the huiPS-MSCs. Mice were sacrificed between weeks 5 and 7 after PBMC injection. Peritoneal fluid, blood, and spleen were collected for flow cytometry analysis. Approval for the use of mice in this study was obtained from our local Institutional Ethic Committee for Laboratory Animals (CIEPAL-Azur, NCE/2013-102).

We determined by ELISA the concentrations of human IL-1a, IL-6, IL-8, IL-2, and IFNγ (Development kit, Peprotech) in MLR culture media according to the manufacturer’s instructions.

Total RNA was extracted and reverse transcribed (SuperScript II Reverse Transcriptase, Invitrogen), and real-time RT-PCR was performed on an StepOnePlus Fast real-time PCR system (SybR Green, Applied Biosystems) on triplicates as described. Results were normalized to the different housekeeping genes (ACTIN, GAPDH, and UBIQUITIN) on the same plate. Differences in gene expression were calculated using the 2^−ΔΔCt method.

Results are presented as mean ± SD. The level of statistical significance was determined by the unpaired two-sample Student’s t-test. p Values <0.05 were considered statistically significant.

Because previous differentiation protocols of MSCs from pluripotent cells involve many steps ranging from embryonic body formation to cell sorting or both (20, 22, 32–36), we set up a simple two-step protocol that would generate rapidly mesodermal-derived cells (Figure 1A). Cultured human iPS cells were kept in absence of bFGF for 7 days, followed by a change for an ectodermal/mesodermal medium (37) for the next 2 weeks. Eighty to ninety percent (at passage 1) and 100% (at passage 2) of the recovered cells (Figure 1B) express surface antigens (Ags) known to be expressed by tissue-derived MSCs. They were positive for CD44, CD73, CD90, CD105, and HLA-ABC Ags but negative for the endothelial CD31 marker, the hematopoietic and immune related CD34, CD45 markers, HLA-DR antigens, and CD80 and CD86 co-stimulatory molecules (Figure 1B; Figure S1 in Supplementary Material). Three different human iPS cell lines, prepared from different donors and from different tissues [skin fibroblast (29, 38) or myoblasts (30)] behaved identically, confirming that this protocol is applicable for multiple human iPS cells.

In addition, the cells recovered and further kept in culture were capable of differentiation into the classical mesenchymal-derived cells (osteoblasts, chondrocytes, and adipocytes) (Figure 1C) when cultured with appropriate differentiation media, suggesting that the huiPS-MSC population contained multipotent cells and correspond to bona fide MSCs. Finally, the huiPS-MSCs we have generated secreted high and sustained amount of IL-6 and IL-8 cytokine/chemokine but low amount IL-1α (as tested by ELISA) (Figure 1D), a cytokine profile shared by tissue-derived MSCs and associated with their role in tissue repair (39, 40). Altogether, these results highlight that the huiPS-MSCs generated with our protocol shared strong similarities with in vitro maintained MSCs derived from adult tissues (2, 3).

Besides their multipotent characteristics, tissue MSCs display in vitro immunosuppressive activities. To test the immunosuppressive properties of the huiPS-MSCs, we analyzed their action on the proliferation of human T lymphocytes stimulated in an allogeneic manner (Figure 2A). The stimulation of PBMCs in MLR with allogeneic antigen-presenting cells (PBMC from a secondary donor—in here named alloAPC) resulted in CD4⁺ and CD8⁺ T-cell proliferation, which was significantly reduced in coculture with huiPS-MSCs. We further observed a significant reduction in the % of CD25-expressing CD4⁺ and CD8⁺ T cells, indicating a diminished proportion of activated T cells in coculture with huiPS-MSCs (Figure 2B). Another activation marker (CD69) was on the contrary expressed on a higher % of CD4⁺ and CD8⁺ T cells (Figure 2B). Known to be an early activation marker, its maintained expression might indicate that the immunosuppression required to be effective the early activation of the T cells. It is also expressed on memory T cells, which would suggest that the remaining T cells in the coculture might acquire such a “memory” phenotype. We also analyzed by RT-qPCR the relative expression of mRNAs coding for cytokines as well as some surface receptors involved in such immune reactivities (Figure S2 in Supplementary Material). As expected, gene expression signature of activated T cell was clearly reversed upon exposure with huiPS-MSCs. Indeed, compared with the RNA expression by huiPS-MSCs alone or in coculture with unactivated T cells, the relative expressions of activated T-cell cytokines (IL-2, IFNγ, TNFα, and TNFβ) were reduced in the MLRs in the presence of huiPS-MSCs (Figure S2A in Supplementary Material). Measured by ELISA (Figure 2C), the IL-2 and IFNγ production was indeed reduced in the cocultures with huiPS-MSCs confirming at the protein level the lower activation of T lymphocytes observed in such conditions. Furthermore, RNA expression of other genes coding for cytokines were affected (Figure S2A in Supplementary Material): we observed an increased expression of inflammatory IL-1α and IL-1β, IL-6 (cytokines that were shown to sustain the immunosuppressive activity of MSCs), as well as TGF-β and LIF known for their immunosuppressive functions. The mRNA coding for the cytokine IL-10 was also overexpressed. However, its immunosuppressive function appeared to be not involved in the T-cell immunosuppression. Indeed, using anti-IL-10-blocking antibodies during the initial MLR in the presence of huiPS-MSCs did not affect the inhibition of the T-cell proliferation (Figure S3A in Supplementary Material). Tested in transwell assay, we observed that the level of inhibition of the CD4⁺ T-cell proliferation was partially reduced (Figure S3B in Supplementary Material). This indicated that both cellular contacts between activated T cells and huiPS-MSCs, and soluble factors were involved in the immunosuppressive function of the huiPS-MSCs. On the contrary, the proliferation of the CD8⁺ T-cell population appeared to be less inhibited in the transwell assay, suggesting that their immunosuppression required a direct contact between T cells and huiPS-MSCs. The relative RNA expression of surface molecules involved in the T-cell activation and their polarization into effector cells such as OX40 and its ligand (OX40L), as well as CD47 were diminished, while on the contrary, those of LAG3 and CTLA4, potentially involved in immunosuppressive functions were increased in the MLR in the presence of huiPS-MSCs (Figure S2B in Supplementary Material). Altogether, these observations might indicate a change in the polarization of the T-cell populations activated in the presence of huiPS-MSCs.

Using intracytoplasmic cytokine detection by flow cytometry in activated T lymphocytes, we observed a dramatic reduction in the proportion of CD4⁺ T lymphocytes producing IFNγ and TNFα corresponding to Th1 cells in MLR with huiPS-MSCs (Figure 3A). Furthermore, we detected the presence of CD4⁺ Treg cells among the T populations recovered from the cocultures. As shown in Figure 3B, the MLR without huiPS-MSCs generated only a small amount of FoxP3⁺ CD4⁺ T cells, while in the presence of huiPS-MSCs this percentage strongly increased up to 16%. FoxP3 being susceptible to reflect human T-cell activation, we confirmed the presence of CD4⁺ Treg cells since we observed an increased detection of a population of CD4⁺ T cells expressing a high level of the CD25 and a low level of CD127 markers (Figure 3C). This CD4⁺ CD25^hi CD127^lo T-cell population further contained a higher proportion of FoxP3 cells (Figure 3C). To better define the phenotypic characteristic of the Treg cell population induced in the presence of huiPS-MSCs, we showed that the neuropilin1 (Nrp1) surface marker was not particularly expressed by such Treg cells and that the Ikaros family member Helios was clearly highly expressed in the whole CD4⁺ T cells population (reflecting a global activation) (Figure S4 in Supplementary Material). Finally, this CD4⁺ T-cell population recovered from the cultures in the presence of the huiPS-MSCs was then assayed in a secondary MLR. As shown in Figure 3D, this CD4⁺ T-cell population containing Treg cells was very efficient to inhibit CD4⁺ T-cell proliferation. Indeed, we observed 50 and 70% inhibition of cell proliferation when the ratios of Treg cells over CD4⁺ T cells were 1 for 3 and 1 for 2, respectively. We thus demonstrated that the FoxP3⁺ CD4⁺ T cells generated during the coculture with the huiPS-MSCs are immunosuppressive CD4⁺ Treg cells.

Altogether, these results highlight the immunosuppressive activity in vitro of the huiPS-MSCs on T-cell stimulation that induces a switch in T-cell cytokine polarization and the generation of Treg cells.

We further tested the immunosuppressive activity of huiPS-MSCs in a model of human T-cell expansion in immune-deficient NSG mice (41). First, we determined whether ip-injected huiPS-MSCs could be detected in different compartments in NSG mice. HLA-ABC⁺ huiPS-MSCs were labeled with CFSE and injected ip in NSG mice. Despite a lower level of expression of CD73 on recovered cells, we showed that CFSE⁺ HLA-ABC⁺ huiPS-MSCs could be detected up to 7 days after injection not only within the peritoneal cavity but also among the circulating cells and splenocytes (Figures 4A,B). This indicated that the injected huiPS-MSCs remained viable for at least 7 days within these mice and were able to circulate at least up to the spleen.

In a second step, we tested their impact in vivo on the expansion of human T cells. NSG mice were injected with human PBMC, were then treated or not with three infusions of huiPS-MSCs (through ip injection at 1-week intervals), and were sacrificed between weeks 5 and 7. After sacrifice, cells collected from the peritoneal cavity, those circulating in the blood and those present in the spleen were analyzed by FACS analysis (not shown and Figure 5) on the basis of expression of the human CD45 Ag. Within the peritoneal cavity, human CD45⁺ cells represented about 65% of total cells recovered from control or huiPS-MSCs-treated mice; more than 95% of them were CD3⁺ T cells (not shown). This indicated that among the injected PBMC, T cells were the main human cell population able to expand and colonize the mice. Accordingly, we observed in the blood and spleen that more than 90% of human CD45⁺ cells were CD3⁺ T cells with expected proportion of CD4⁺ and CD8⁺ cells, indicating a similar rate of expansion. But when mice were treated with huiPS-MSCs, the percentage of circulating human T cells was significantly reduced about 1.8-fold (Figure 5A) leading to a 1.6-fold reduced accumulation of total T cells within the spleen (Figure 5B). Interestingly, the proportion of both CD4⁺ and CD8⁺ T cells was changed, the CD8⁺ T cells being significantly more affected by the huiPS-MSCs treatments (Figure 5B).

Intracytoplasmic labeling of cytokines produced by the human T cells recovered from the spleen was performed. We showed that untreated mice displayed high percentages of human inflammatory IFNγ⁺ TNFα⁺ Th1 cells, while little or none produced the anti-inflammatory IL-10 cytokine (Figure 6A). By contrast, in mice treated with the huiPS-MSCs, the proportion of Th1 cells was substantially reduced, while the one of T cells producing IL-10 was increased more than sixfolds. In parallel, FoxP3⁺ CD3⁺ Treg cells were absent in non-treated animals whereas they were systematically detected in huiPS-MSCs injected mice (Figure 6B).

Altogether, these data demonstrated that the huiPS-MSCs were able to limit the human T-cell expansion in vivo, along with a reduced Th1 inflammatory cytokine profile, the presence of IL-10-producing T cells and the generation of FoxP3⁺ Treg cells.