Mesenchymal stromal cells induce regulatory T cells via epigenetic conversion of human conventional CD4 T cells in vitro

TRANSLATIONAL AND CLINICAL RESEARCH Mesenchymal stromal cells induce regulatory T cells via epigenetic conversion of human conventional CD4 T cells in vitro Rita I. Azevedo 1 | Ekaterina Minskaia 1 | Ana Fernandes-Platzgummer 2 | Ana I. S. Vieira 1 | Cláudia L. da Silva 2 | Joaquim M. S. Cabral 2 |Jo ~ ao F. Lacerda 1 1 Instituto de Medicina Molecular, Faculdade de Medicina da Universidade de Lisboa, Lisbon, Portugal 2 Department of Bioengineering and iBBInstitute for Bioengineering and Biosciences, Instituto Superior Técnico, Universidade de Lisboa, Lisbon, Portugal Correspondence Rita Isabel Azevedo, PhD, Instituto de Medicina Molecular, Avenida Professor Egas Moniz, 1649-028 Lisbon, Portugal. Email: [email protected]oa.pt Funding information Fundaç~ ao para a Ciência e a Tecnologia (FCT)/ Ministério da Ciência, Tecnologia e Ensino Superior (MCTES); Fundaç~ ao para a Ciência e Tecnologia, Portugal Abstract Regulatory T cells (Treg) play a critical role in immune tolerance. The scarcity of Treg therapy clinical trials in humans has been largely due to the difficulty in obtaining sufficient Treg numbers. We performed a preclinical investigation on the potential of mesenchymal stromal cells (MSCs) to expand Treg in vitro to support future clinical trials. Human peripheral blood mononuclear cells from healthy donors were cocultured with allogeneic bone marrow-derived MSCs expanded under xenogeneicfree conditions. Our data show an increase in the counts and frequency of CD4 + CD25 high Foxp3 + CD127 low Treg cells (4and 6-fold, respectively) after a 14-day coculture. However, natural Treg do not proliferate in coculture with MSCs. When purified conventional CD4 T cells (Tcon) are cocultured with MSCs, only cells that acquire a Treg-like phenotype proliferate. These MSC-induced Treg-like cells also resemble Treg functionally, since they suppress autologous Tcon proliferation. Importantly, the DNA methylation profile of MSC-induced Treg-like cells more closely resembles that of natural Treg than of Tcon, indicating that this population is stable. The expression of PD-1 is higher in Treg-like cells than in Tcon, whereas the frequency of PDL-1 increases in MSCs after coculture. TGF-βlevels are also significantly increased MSC cocultures. Overall, our data suggest that Treg enrichment by MSCs results from Tcon conversion into Treg-like cells, rather than to expansion of natural Treg, possibly through mechanisms involving TGF-βand/or PD-1/PDL-1 expression. This MSC-induced Treg population closely resembles natural Treg in terms of phenotype, suppressive ability, and methylation profile. KEYWORDS celland tissue-based therapy, DNA methylation, mesenchymal stromal cells, regulatory T lymphocytes Received: 3 January 2020 Accepted: 21 February 2020 DOI: 10.1002/stem.3185 This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. ©2020 The Authors. STEM CELLS published by Wiley Periodicals LLC on behalf of AlphaMed Press Stem Cells. 2020;38:1007–1019. wileyonlinelibrary.com/journal/stem 1007 1|INTRODUCTION Regulatory T cells (Treg) are a key population in immune tolerance and their potential use in the treatment of chronic inflammatory diseases has been increasingly investigated. In allogeneic hematopoietic stem cell transplantation (allo-HSCT), lower Treg counts are associated with the development of chronic graft vs host disease (cGVHD). 1,2 Adoptive immunotherapy with donor Treg has been reported to successfully prevent GVHD after allo HSCT. 3-5 Our group is currently coordinating a Horizon 2020-funded consortium, TREGeneration, in which parallel phase I/II clinical trials are assessing the safety and efficacy of donor Treg infusion in patients with steroidrefractory cGVHD (www.tregeneration.eu). Nevertheless, adoptive Treg therapy in humans still presents many challenges, particularly the difficulty in obtaining sufficient numbers of clinical grade Treg. Due to the paucity of Treg in the peripheral blood, the expansion of these cells in vitro has been investigated as a strategy to increase Treg yield. 3,6 Like Treg, mesenchymal stromal cells (MSCs) also constitute an important immunoregulatory population. MSCs are multipotent progenitor cells able to differentiate into diverse mesodermal lineages that can be isolated from a wide range of adult and fetal tissues. 7 MSCs play a suppressive role by inhibiting both innate and adaptive immune responses. 8 Previous studies suggest that MSC-mediated immunomodulation may be partly driven by Treg induction and/or expansion. 9-13 However, these reports have not assessed Treg yield in terms of absolute counts, 10,13 nor characterized the resulting Treg-like cells in detail. 9,11,12 Thus far, the potential of MSCs to recruit Treg has been poorly understood. In the present study, we sought to determine whether MSCs are able to induce and/or expand Treg in vitro, as well as the mechanisms of Treg enrichment by MSCs. 2|MATERIALS AND METHODS 2.1 |Cell isolation Peripheral blood mononuclear cells (PBMC) were isolated by FicollHypaque density gradient (Amersham Pharmacia Biotech) from peripheral blood of healthy volunteers, after informed consent with the approval of the Ethics Committee of Lisbon Academic Medical Centre. CD4 T cells were negatively selected using the EasySep Human CD4 + T-cell enrichment kit (StemCell Technologies), stained for CD3, CD4, CD25, and CD127, and subsequently sorted into Treg and Tcon cells using a FACSAria flow cytometer (BD Biosciences). 2.2 |Human bone marrow MSC cultures Bone marrow (BM) MSCs used in this study are part of the cell bank of the Stem Cell Engineering Research Group, iBB-Institute for Bioengineering and Biosciences at Instituto Superior Técnico. BM aspirates were obtained from healthy donors after written informed consent with the approval of the Ethics Committee of Instituto Português de Oncologia Francisco Gentil, Lisboa, Portugal. BM MSCs were isolated, as previously described, 14 and cryopreserved. BM MSCs expressed their characteristic immunophenotype, being positive for CD73, CD90, and CD105, and negative for CD31, CD34, CD45, and CD80. BM MSCs were thawed and seeded at 3000 cells/cm 2 using StemPro MSC SFM XenoFree medium on tissue culture flasks precoated with CELLstartCTS (Life Technologies), as described elsewhere. 7 For these studies, BM MSCs within passages 4 to 6 from five donors (both male and female; aged 26-60 years) were used. 2.3 |Coculture of MSCs and lymphocyte populations Cells were cultured in RPMI 1640 (Invitrogen) supplemented with 10% heat-inactivated human AB serum (Sigma-Aldrich), 100 U/mL penicillin, 100 mg/mL streptomycin, and 2 mM L-glutamine (Invitrogen), in the presence or absence of recombinant human IL-2 (10 U/mL; Peprotech). MSCs and PBMC (1.6 ×10 6 per well) or purified T-cell populations (3 ×10 6 total lymphocytes per well) were cultured in sixwell plates at 37C and 5% CO 2 in a humidified atmosphere. Supernatants were collected and fresh culture medium was added every 3 to 4 days. 2.4 |Phenotypic analysis Cells were stained with the following antihuman monoclonal antibodies: PD-L1 (130021) Alexa Fluor 488 (R&D Systems); PD1 Significance statement Regulatory T cells (Treg) have been increasingly recognized for their application in the treatment of chronic inflammatory diseases. However, Treg are very scarce and may need to be expanded in vitro to obtain clinically relevant numbers. The induction of Treg from conventional CD4 T cells (Tcon) is particularly appealing due to the abundance of this population. However, the stability and functional ability of induced Treg need to be thoroughly ensured prior to clinical application. Mesenchymal stromal cells (MSCs) have been suggested to recruit Treg in vitro, although the underlying mechanisms are so far poorly understood. This study demonstrates for the first time that MSCs are able to induce functional Treg through epigenetic modification of human Tcon in vitro. These data suggest that MSC-induced Treg have a potential application as a cellular therapy in clinical settings where immune tolerance is compromised. 1008 AZEVEDO ET AL. (EH12.1) BB515, CD15s (CSLEX1) PE-CF594 (BD Horizon); GITR (621) PE (Biolegend); CD25 (M-A251) PE-CY7 (BD Pharmingen); CD3 (OKT3) PerCPCY5.5; CD4 (RPAT4) APC; CD127 (eBioRDR5) APCeFlour780 (eBioscience). Intracellular staining for Bcl-2 (Bcl-2/100) FITC, Ki67 (20Raj1) FITC and FoxP3 (PCH101) e450 (eBioscience) was performed using fixation and permeabilization reagents from eBioscience. Samples were acquired on an LSR Fortessa flow cytometer (BD Biosciences) after fixation with 1% formaldehyde (Sigma-Aldrich). Data were analyzed using FlowJo9. 2.5 |Suppression assay Fresh autologous Treg, Tcon, and CD4 neg PBMC were purified by FACSorting. Tcon were labeled with 0.5 μM CellTrace CFSE cell proliferation kit (ThermoFisher). CFSE Tcon responder cells (1.5-2.5 ×10 4 per well) were plated with different suppressor populations at 1:0, 8:1, 4:1, 2:1, and 1:1 responder to suppressor ratios in 96-well round bottom plates. Cells were activated with the Treg suppression inspector kit (Miltenyi Biotec) at a 1:1 bead-to-cell ratio for 4 days before acquisition. 2.6 |Cytokine quantification in culture supernatants Supernatants were collected at different time points of MSC coculture. Cytokine levels were quantified by ELISA or multiplex analysis: human indoleamine 2,3-dioxygenase (IDO) ELISA kit (Sunred Biological Technology); DetectX Prostaglandin E2 EIA kit (Arbor Assays); human TGF-βsingle-plex HDR assay chemiluminescent; IL-10, TNF-α, and IFN-γQ-Plex multiplex ELISA arrays (Quansys). Image analysis was performed using the Q-View Imager. Samples were tested undiluted in duplicate. Levels of luminescent or pixel intensity units were measured by the Q-View Software. Assays were performed and analyzed by tebu-bio. 2.7 |Isolation of genomic DNA and bisulphite conversion Genomic DNA (gDNA) from Treg and Tcon was isolated with the Quick-gDNA MiniPrep kit (Zymo Research). Bisulphite (BS) conversion of gDNA from all cell populations was performed at the same time with EZ DNA methylation lightning kit (Zymo Research). TSDR and CAMTA1 gene regions were amplified with nonmethylation-, BS DNA-, strand-specific forward (Fw) and reverse (Rev) primers and cloned into pGEM-T Easy vector (Promega) via NcoI and NsiI restriction sites, as previously described. 15 Plasmid DNAs from 22 to 24 clones were isolated using plasmid NZYMiniprep kit (NZYtech) and 20 to 22 positive clones were sequenced using reverse SP6 primer: 50-GTG ACACTATAGAATACTC-30(Stabvida). Sequences (AB1 files containing chromatograms) were aligned to each gene region's reference sequence using SeqMan software (DNA Star Lasergene 8). All nonmethylated cytosines (C) were identified by the presence of a thymidine (T) in BS-converted sequences, whereas methylated C ( m C) were identified by a C. Efficiency of BS conversion was confirmed by conversion of non-CpG C to T. The percentage of methylation in each CpG position was determined by defining the proportion of m C in the total of 20. As both gene regions were amplified from the same BS DNA template, the methylation differences reflect the average methylation status of the cell population. 2.8 |Statistical analysis Statistical analysis was performed using GraphPad Prism 7 (GraphPad Software). Data are presented as mean with SE of mean. Statistical analysis was performed using oneor two-way analysis of variance (ANOVA), or paired ttest, as appropriate. Pvalues <.05 were considered significant. 3|RESULTS 3.1 |MSC coculture increases counts and frequency of Treg-like cells We investigated the ability of MSCs to support Treg survival and/or expansion in MSC+PBMC cocultures. We cocultured PBMC from healthy donors with allogeneic BM-derived MSCs expanded under xenogeneic-free conditions, in the presence or absence of IL-2. As a control, we cultured PBMC alone, with or without IL-2. We first compared two MSC:PBMC ratios (1:5 and 1:3) and found that MSC:PBMC 1:3 was the optimal ratio in terms of Treg yield for most of the donors tested (data not shown). At each time point, we assessed the frequency and absolute counts of different lymphocyte populations by flow cytometry, particularly of CD4 + CD25 high Foxp3 + CD127 low Treg and CD4 + CD25 − Foxp3 − conventional CD4 T cells (Tcon) (Figure 1). We found that MSCs at a 1:3 ratio induce an approximately 6-fold increase in the frequency of Treg within T cells compared to PBMC cultured alone (Figure 1A,B; P< .0001). Importantly, Treg counts also dramatically increased in the presence of MSCs, both compared to PBMC cultured alone (P= .0003) and to day 0 (P= .0035) (Figure 1C). In contrast, the frequency and counts of Tcon (Figures 1D,E), as well as the frequency of CD8 T cells (Figure 1F), were not increased in the presence of MSCs. In fact, total cell numbers decreased in MSC+PBMC cocultures (Figure 1G). Hence, MSC coculture appears to specifically increase Treg levels. 3.2 |MSC coculture does not induce Treg proliferation We then assessed if the increase in Treg counts observed in MSC +PBMC cocultures was due to Treg expansion. We evaluated cell MSCs INDUCE Treg VIA EPIGENETIC CONVERSION OF Tcon 1009 proliferation by Ki67 expression (Figure 2A-C). In the absence of IL-2, we found only negligible levels of proliferation within Treg (Figure 2A, B) and Tcon (Figure 2C), even in the presence of MSCs, which can be explained by the fact that MSCs are a suppressive population. This repression was overcome by the addition of IL-2 (Figure 2A-C). The lack of Treg proliferation suggests that MSCs increase the Treg pool by converting Tcon into induced Treg (iTreg), rather than by expanding natural Treg (nTreg). We also investigated if the MSC-induced increase in Treg numbers was associated with preferential cell survival by assessing the levels of the antiapoptotic protein Bcl-2 (Figure 2D,E). MSC coculture increased the levels of Bcl-2 within Treg (Figure 2D; P< .0001), but not within Tcon (Figure 2E), further suggesting a preferential impact of MSCs in Treg survival. 3.3 |MSCs induce Tcon to acquire a Treg-like phenotype Given that Treg are increased but do not proliferate in the presence of MSCs, we hypothesized that Treg enrichment might be driven by Tcon acquiring a Treg-like phenotype. To determine the mechanism by which MSCs increase the numbers of Treg-like cells, purified Tcon (CD3 + CD4 + CD25 − ) were labeled with CFSE. These CFSE Tcon were then remixed with the remaining autologous nonlabeled PBMC, allowing us to track the phenotype and proliferation of Tcon cells during MSC+PBMC coculture. On day 0, all CFSE + cells in the coculture were CD3 + CD4 + CD25 − Foxp3 − (Figure 3A). On day 10, a CD3 + CD4 + CD25 + Foxp3 + population was clearly observable within the CFSE + gate (Figure 3A). Gating on these induced Treg-like cells, CFSE FIGURE 1 MSC+PBMC cocultures increase the absolute counts and frequency of Treg cells. A, Representative dot-plots of CD25 and Foxp3 expression within CD4 T cells at days 0, 7, and 14 of PBMC alone (top panel) and MSC:PBMC 1:3 cocultures (bottom panel). Frequency of regulatory (Treg, CD4 + CD25 bright IL7R − Foxp3 + ) (B), conventional CD4 (Tcon, CD4 + CD25 − Foxp3 − ) (D), and CD8 T cells (CD4 − ) (F) was assessed by flow cytometry within CD3 + T cells before and after culture (n = 4 for each culture condition). Viable cell counts were assessed with a hemocytometer before and after culture. Absolute counts of Treg (C), Tcon (E), and total cells (G) are shown (n = 4 for each culture condition). Graphs show mean values plus or minus SEM. Statistical analysis was performed using two-way ANOVA (**P< .01, ***P< .001, ****P< .0001 between PBMC alone and MSC: PBMC 1:3) 1010 AZEVEDO ET AL. dilution revealed they were proliferating (Figure 3A). These results show that MSCs induce a population of Tcon to acquire a Treg-like phenotype, while proliferating. To further clarify this issue, we cocultured MSCs with either purified CFSE-labeled Treg (Figure 3B) or Tcon (Figure 3C). CD4 + T cells were labeled with CFSE, and then purified into Treg (CD3 + CD4 + CD25 bright CD127 low ) and Tcon (CD3 + CD4 + CD25 − ) by FACSorting. Each population was separately remixed with autologous unlabeled CD4 neg PBMC and cultured with or without MSCs, so that the only CD4 + T cells present in culture on day 0 were either CFSE Treg (Figure 3B) or CFSE Tcon (Figure 3C). On day 10 of CFSE Treg coculture with MSCs, the CFSE + population maintained the CD25 + Foxp3 + phenotype, showing brighter levels of CD25 and Foxp3 expression than on day 0 (Figure 3B; average fold increase in CD25 and Foxp3 MFI within CD4 T cells relative to day 0: 1.4 ± 0.2 and 1.3 ± 0.2, respectively). However, CFSE labeling remained unchanged, revealing that Treg do not proliferate in MSC coculture (Figure 3B). The frequency of CD25 + CD127 − Foxp3 + cells within CD4 T cells remained very similar to that observed on day 0 throughout the culture period, in the presence or absence of MSCs (Figure 3B). These data show that nTreg are maintained but not expanded by MSC coculture. As for CFSE Tcon, we again observed the induction of a proliferating Treg-like population and the consequent increase in the frequency of CD25 + CD127 − Foxp3 + cells during coculture with MSCs (Figure 3C). In contrast, when we gated on the CFSE + cells that maintained a CD25 − Foxp3 − Tcon phenotype, we found that these cells do not proliferate (Figure 3C). Therefore, Treg enrichment by MSCs appears to be due to induction of Treg-like cells from Tcon, rather than to Treg expansion. 3.4 |Treg-like cells induced by MSCs are suppressive To determine if MSC-induced Treg-like cells also resemble Treg in terms of functional ability, we performed suppression assays. We cocultured PBMC, purified Tcon or purified Treg with MSCs, and FACSorted the resulting Treg-like populations on day 10 (Figure 4A). FIGURE 2 MSC+PBMC coculture does not induce Treg proliferation in the absence of IL-2. A, Representative dot-plots of Ki67 expression within Treg cells (CD3 + CD4 + Foxp3 + ) at day 10 of MSC:PBMC 1:3, PBMC alone and PBMC with IL-2 cultures. Frequency of Ki67 + cells (B,C) and Bcl-2 median fluorescence intensity (MFI) (D,E) were assessed by flow cytometry within Tcon (CD3 + CD4 + Foxp3 − ) and Treg (CD3 + CD4 + Foxp3 + ) cells before and after culture (n = 4 for each culture condition). Graphs show mean values plus or minus SEM. Statistical analysis was performed using two-way ANOVA (****P< .0001 between PBMC alone and MSC:PBMC 1:3) MSCs INDUCE Treg VIA EPIGENETIC CONVERSION OF Tcon 1011 FIGURE 3 MSC coculture induces Tcon to acquire a Treg-like phenotype, while proliferating. A, Tcon (CD3 + CD4 + CD25 − ) cells were purified by FACSorting, labeled with CFSE and remixed with the remaining autologous nonlabeled PBMC. These PBMC containing CFSE Tcon were cocultured with MSCs. On day 10, CD25 + Foxp3 + cellswerefoundwithintheCFSE + gate. Further gating within these CFSE + Treg-like cells showed they were proliferating. Representative dot-plots are shown. B,C, CD4 + T cells were purified by magnetic cell separation and labeled with CFSE. B, CFSE Treg (CD3 + CD4 + CD25 bright CD127 low ) cells were purified by FACSorting, remixed with autologous nonlabeled CD4 neg PBMC, and cultured with or without MSCs. On day 10, gating on CD25 + Foxp3 + cells showed they were not proliferating. C, CFSE Tcon (CD3 + CD4 + CD25 − ) cells were purified by FACSorting, remixed with autologous nonlabeled CD4 neg PBMC, and cultured with or without MSCs. On day 10, gating on CD25 + Foxp3 + cells showed these Treg-like cells were proliferating, but not the cells that maintained a CD25 − Foxp3 − Tcon phenotype. Representative dot-plots are shown. B,C, Graphs show the meanfrequencyofTreg(CD25 + CD127 − Foxp3 + ) cells within CD4 T cells before and after culture, plus or minus SEM (n = 4 for each culture condition) 1012 AZEVEDO ET AL. The Treg population obtained after PBMC+MSC coculture, Treg (PBMC+MSC), encompasses both preexisting nTreg and MSC-induced Treg-like cells. The CD25 high CD127 low population purified after Tcon +MSC coculture consists exclusively of MSC-induced Treg-like cells: iTreg (Tcon+MSC). The Treg cells isolated after Treg+MSC coculture are nTreg: nTreg (Treg+MSC). We also tested the suppressive ability of the untouched cocultures, straight out of the well. These unsorted cocultures comprise all the cell types present in culture, including MSCs. The suppressive ability of the unsorted cocultures (Figure 4B) and of the Purified Treg-like cells (Figure 4C) was assessed in suppression assays using autologous fresh Tcon labeled with CFSE as responder cells. As a positive control, we used autologous fresh Treg. We observed a significant suppression of Tcon proliferation by all unsorted cocultures (Figure 4D, % divided cells at 1:1 Tcon:Treg ratio FIGURE 4 MSC-induced Treg-like cells are suppressive. Total PBMC, purified Tcon (CD3 + CD4 + CD25 − ), plus CD4 neg PBMC or purified Treg (CD3 + CD4 + CD25 bright CD127 low ) plus CD4 neg PBMC were cocultured with MSCs. A, Representative dotplots showing the unsorted cocultures (green gates) and the FACSorted CD25 bright CD127 low cells (red gates) from these cocultures on day 10. The suppressive function of either the unsorted cocultures (B) or of FACSorted CD25 bright CD127 low cells (C) was tested using autologous CFSE-labeled fresh Tcon as responder cells. D, Frequency of divided fresh Tcon responder cells in the presence of different suppressor populations (unsorted PBMC+MSC n = 6; Treg purified from PBMC+MSC cocultures n = 7; unsorted Tcon+MSC n = 5; induced Treg cells purified from Tcon+MSC cocultures n = 4; unsorted Treg+MSC n = 6; natural Treg purified from Treg+MSC cocultures n = 5; fresh autologous Treg n = 7) at a 1:1 ratio. Graphs show the mean frequency of divided cells plus or minus SEM. Statistical analysis was performed using oneway ANOVA followed by Dunnett's multiple comparisons test (***P< .001, ****P< .0001) MSCs INDUCE Treg VIA EPIGENETIC CONVERSION OF Tcon 1013 compared to Tcon alone, one-way ANOVA followed by Dunnett's multiple comparisons analysis, P< .0001). Purified Treg-like populations significantly suppressed autologous Tcon proliferation in a similar degree to that observed with fresh Treg (P< .0001): Treg (PBMC+MSC) P< .0001; iTreg (Tcon+MSC) P= .0002; nTreg (Treg +MSC) P< .0001 (Figure 4D). Thus, MSC-induced Treg-like cells are suppressive. 3.5 |MSC-induced Treg are epigenetically similar to nTreg We next investigated whether the DNA methylation profile of MSCinduced Treg resembles that of nTreg or of the Tcon they arose from. For that purpose, we purified CD25 + Foxp3 + Treg-like cells induced from Tcon+MSC coculture (iTreg), and compared them to fresh nTreg FIGURE 5 MSC-induced Treg-like cells resemble natural Treg epigenetically. Fresh Tcon (CD3 + CD4 + CD25 − ) and nTreg (CD3 + CD4 + CD25 bright CD127 low ) were purified by FACSorting. Purified Tcon plus CD4 neg PBMC were cocultured with MSCs. On day 10, Treglike cells (iTreg, CD25 bright CD127 low ) were FACSorted from Tcon+MSC cocultures. A, The methylation status of 13 CpG sites within the CAMTA1 intronic region 3 was analyzed on the coding DNA strand on fresh nTreg, fresh Tcon, and iTreg samples from three different donors. B,The average difference in the frequency of methylation for each CpG was calculated between fresh Tcon and fresh nTreg, between fresh Tcon and iTreg, and between fresh nTreg and iTreg: Δ(Tcon −nTreg), Δ(Tcon −iTreg), and Δ(nTreg −iTreg), respectively. Graphs show the frequency of methylation (C) and the average difference in the frequency of methylation (D) within CpGs 2 and 11 for each donor. Statistical analysis was performed using paired ttest (*P< .05, **P< .01) 1014 AZEVEDO ET AL. and fresh Tcon from the same donor. The Treg-specific demethylated region (TSDR) of the FOXP3 locus has been reported to be demethylated in nTreg, and mostly methylated in iTreg and Tcon cells. 16-19 We first analyzed the methylation status of 15 CpG sites within both the coding and template strands of the 700 bp region covering the TSDR in iTreg purified after Tcon+MSC coculture, as well as in fresh nTreg and fresh Tcon from the same donor. In all donors analyzed (two male, one female), we found no differences in methylation among fresh Tcon, fresh nTreg, and iTreg. Both coding and template strands had the average methylation levels for CpG 1 to 15 of 100% in all cell populations (data not shown). Since no differences were detected between our cell populations, we sought an alternative to TSDR methylation analysis. CAMTA1 (calmodulin-binding transcription activator 1) has been shown by us and others to distinguish Tcon from Treg subsets in both male and female donors. 15,16 In particular, CpG 2 and 11 are strikingly more demethylated in Treg. 15 We assessed the methylation status of 13 CpG sites within the 470 bp CAMTA1 intronic region 3 in fresh nTreg, fresh Tcon, and iTreg (Figure 5). MSC-induced Treg-like cells (iTreg) displayed a methylation pattern similar to that of fresh nTreg (Figure 5A). The average difference in the frequency of methylation for each CpG revealed higher levels of methylation in fresh Tcon compared to fresh nTreg and iTreg, whereas fresh nTreg and iTreg displayed no major differences (Figure 5B). Focusing on CpG 2 and 11, these sites were consistently more demethylated in fresh nTreg and in iTreg than in fresh Tcon (Figure 5C). The average difference in the frequency of methylation of CpG 2 and 11 between fresh nTreg and iTreg was significantly lower than the difference between fresh Tcon and nTreg (P= .035), and that between fresh Tcon and iTreg (P= .0216; Figure 5D). Hence, the epigenetic profile of MSC-induced Treg-like cells more closely resembles that of nTreg than of Tcon. 3.6 |PD-1/PDL-1 levels increase after MSC coculture To better understand the mechanism of Treg induction by MSCs, we analyzed the expression of functional markers associated with Treg induction/expansion. CD15s expression identifies activated and highly suppressive Foxp3 high effector Treg cells. 20 The basal frequency of CD15s + cells was significantly higher in fresh Treg than in fresh Tcon (Figure 6A; P< .0001). After MSC coculture, natural and induced Treg displayed a similar frequency of CD15s + cells, which was higher than that observed in the cells that maintained a Tcon phenotype (Figure 6A). GITR (glucocorticoid-induced tumor necrosis factorrelated receptor) is constitutively expressed at high levels on Treg and FIGURE 6 Treg-like cells express functional markers associated with Treg induction/ expansion. Purified Treg or Tcon plus CD4 neg PBMC were cocultured with MSCs. The frequency of CD15s (A, n = 6), GITR (B, n = 4), and PD-1 (C, n = 6) expressing cells was determined by flow cytometry within Treg and Tcon before (fresh Treg, fresh Tcon) and after 10 days of MSC coculture. Analysis was performed by gating on Treg (CD3 + CD4 + CD25 bright CD127 low ) cells from Treg plus MSC cocultures (nTreg [Treg+MSC]) or from Tcon plus MSC cocultures (iTreg [Tcon +MSC]), as well as on Tcon from Tcon plus MSC cocultures (Tcon [Tcon+MSC]). The frequency of PDL-1 expressing cells (D, n = 3) was assessed on fresh MSCs (day 0) and after a 10-day culture alone (MSC alone) or with PBMC (PBMC+MSC). Graphs show mean values plus or minus SEM. Statistical analysis was performed using one-way ANOVA followed by Tukey's multiple comparisons test (*P< .05, **P< .01, ****P< .0001) MSCs INDUCE Treg VIA EPIGENETIC CONVERSION OF Tcon 1015