Instability of Foxp3 Expression Limits the Ability of Induced Regulatory T Cells to Mitigate Graft versus Host Disease Free

T-regulatory cells (Treg) are a critical component of the adaptive immune system and play a pivotal role in controlling inflammatory responses as well as preventing the development of autoimmunity (1). There are 2 subsets of Tregs, both of which are highly dependent on Foxp3 expression for their suppressive function, and have unique and overlapping characteristics. Natural Tregs (nTreg) differentiate in the thymus (2), are most commonly characterized by the constitutive expression of activation markers such as CD25, CTLA-4, CD134, CD103, and glucocorticoid-induced TNF receptor (GITR) and require high affinity interactions with self peptides (3–6). In contrast, induced Tregs (iTreg) arise from CD4⁺Foxp3⁻ conventional T cells that upregulate Foxp3 in the periphery upon activation in the context of TGF-β (7, 8). Moreover, iTregs can be generated in vitro by coculture with TGF-β after stimulation through the T-cell receptor and provision of interleukin (IL) 2 (9–11).

Graft versus host disease (GVHD) is the major complication of allogeneic bone marrow transplantation (BMT). GVHD is characterized by the expansion and differentiation of donor alloreactive T cells, the release of proinflammatory cytokines, and the recruitment of other effector cell populations, leading to the damage of host target organs (12–14). A number of studies have shown that the reconstitution of Tregs, which plays a pivotal role in regulating donor T-cell–mediated alloresponses, is severely impaired in both acute and chronic GVHD (15–18). This premise is further supported by studies which have shown that the adoptive transfer of nTregs is able to significantly attenuate GVHD severity, indicating that exogenous supplementation of these cells is an effective strategy for the re-establishment of transplantation tolerance (19–23). The clinical application of this approach, however, is constrained by the low frequency of these cells in the peripheral blood that makes these cells difficult to isolate (24). iTregs, on the other hand, can be generated in large numbers in vitro and are able to suppress alloreactive T-cell responses (25), raising the possibility that these cells might also be efficacious in vivo. However, the role of these cells in GVHD biology has yet to be fully examined, and many questions remain about their stability and ability to protect against disease.

iTregs can also be generated in vivo in response to inflammatory events (26, 27), although the specific contribution of these cells relative to nTregs is not known. Moreover, the extent to which the in vivo conversion of these cells from naive T-cell precursors is affected by the proinflammatory environment that accompanies most pathologic conditions has also not been defined. We have recently shown that IL-6 is one of the proinflammatory cytokines that is induced during GVHD and inhibits the generation of Tregs (28), but the effect of inflammatory cytokines in the local microenvironment on the conversion of iTregs from conventional T cells remains uncertain. The goal of the current study therefore was to define the role of iTregs during GVHD. Specifically, we sought to determine the efficacy of iTregs in the prevention of GVHD, whether there was differential protection afforded by in vitro versus in vivo iTregs, and to what extent, protection or lack thereof was attributable to the stability of Foxp3 expression and the underlying proinflammatory cytokine milieu.

C57BL/6 (B6; H-2^b, Thy1.2⁺), B6.PL (H-2^b, Thy 1.1⁺), and BALB/c (H-2^d, Thy1.2⁺) mice were bred in the Animal Resource Center (ARC) at the Medical College of Wisconsin (MCW) or purchased from Jackson Laboratories. Foxp3^EGFP mice (backcrossed to the B6 background for 6 generations) in which the Foxp3 gene is coupled to the enhanced green fluorescent protein (EGFP) were obtained from Dr. Calvin Williams (MCW) and have been previously described (29). All animals were housed in the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC)-accredited ARC of the MCW. Experiments were all carried out under protocols approved by the MCW Institutional Animal Care and Use Committee. Mice received regular mouse chow and acidified tap water ad libitum.

Anti–IL-6R antibody (MR-16-1) is a rat IgG antibody that has been previously described (28). Animals received a loading dose of 2 mg intravenously on the day of transplantation and then were treated with 0.5 mg weekly by intraperitoneal injection. Antibody was resuspended in PBS prior to injection. Rat IgG (Jackson ImmunoResearch Laboratories) was used as a control and administered at the same dose and schedule as MR-16-1.

Bone marrow (BM) was flushed from donor femurs and tibias with Dulbecco's modified Eagle's media (DMEM; Gibco-BRL) and passed through sterile mesh filters to obtain single-cell suspensions. BM was T-cell depleted (TCD) in vitro with anti-Thy1.2 monoclonal antibody plus low toxicity rabbit complement (C-six Diagnostics). The hybridoma for 30-H12 (anti-Thy1.2, rat IgG2b) antibody was purchased from the American Type Culture Collection. Host mice were conditioned with total body irradiation (TBI) administered as a single exposure at a dose rate of 82 cGy using a Shepherd Mark I Cesium Irradiator (J.L. Shepherd and Associates). Irradiated recipients received a single intravenous injection in the lateral tail vein of BM with or without added spleen cells.

Spleen and peripheral lymph node cells were collected from Foxp3^EGFP mice and sorted on a FACSAria (Becton-Dickinson). Sort purity for these studies consistently averaged 98% to 99%. Spleen, liver, lung, colon, and lymph node cells from transplant recipients were labeled with monoclonal antibodies (mAb) conjugated to fluorescein isothiocyanate (FITC), phycoerythrin (PE), PE-Cy5, or allophycocyanin (APC) that were obtained from BD Biosciences. Cells were analyzed on a FACSCalibur or LSRII flow cytometer with CellQuest or FACSDiva software (Becton-Dickinson). Data were analyzed using FlowJo software (TreeStar).

To generate iTregs in vitro, sorted CD4⁺EGFP⁻ T cells from Foxp3^EGFP mice were cultured in complete RPMI with immobilized anti-CD3 antibody (2.5 μg/mL), soluble anti-CD28 antibody (1 μg/mL), IL-2 (100 U/mL), and TGF-β (10 ng/mL) for 3 days. This typically resulted in the conversion of 60% to 70% of cells. GFP-expressing cells were then resorted for use in BMT studies. In some experiments, all-trans retinoic acid (ATRA; Sigma-Aldrich) was added to cultures at a concentration of 10 nmol/L for the entire 3 days. In other experiments, cells were labeled with Annexin-PE and 7-aminoactinomycin D (7-AAD) to assess the percentage of apoptotic cells in expanded cultures.

To isolate lamina propria lymphocytes, pooled colons were incubated in HBSS buffer (Gibco-BRL) supplemented with 2% FBS, EDTA (0.05 mmol/L), and 15 μg/mL dithiothreitol (Invitrogen) at 37°C for 30 minutes and subsequently digested in a solution of collagenase D (Roche Diagnostics; 1 mg/mL) in DMEM with 2% FBS for 75 minutes at 37°C. The resulting cell suspension was then layered on a 44%/67% Percoll gradient (Sigma). Liver and lung lymphocytes were isolated by collagenase D digestion followed by layering on a Percoll gradient.

Thy1.2⁺ cells (1 × 10⁵) were purified from B6 spleens using the magnetic-activated cell separation (MACS) system (Miltenyi Biotech) and cocultured with 5 × 10⁴ BALB/c dendritic cell–enriched stimulator cells in U-bottomed microwell plates (Becton-Dickinson) at 37°C. Stimulator cells were obtained by digestion of spleens with collagenase D (1 mg/mL) followed by positive selection of CD11c⁺ “dendritic cells” using the MACS system. Nonirradiated stimulator cells were then seeded into microwell plates. Flow-sorted iTreg or nTreg cells were added at varying ratios to wells containing dendritic and T cells. One microcurie of ³H-thymidine was added to triplicate wells for the final 12 to 18 hours prior to harvest. Thymidine incorporation was assessed using a Wallac 1450 Microbeta liquid scintillation counter (Perkin Elmer). Control wells consisted of responders, stimulators, and Tregs alone.

Representative samples of liver, colon, and lung were obtained from transplant recipients and fixed in 10% neutral-buffered formalin. Samples were then embedded in paraffin, cut into 5-μm thick sections, and stained with hematoxylin and eosin. A semiquantitative scoring system was employed to account for histologic changes in the colon, liver, and lung as previously described (30). All slides were coded and read in a blinded fashion.

Lymphocytes isolated from spleen, liver, lung and colon were stimulated with 50 ng/mL phorbol-12-myristate-13-acetate (PMA; Sigma) and 750 ng/mL ionomycin (Calbiochem) for 1 hour and then incubated with GolgiStop (BD Pharmingen) for an additional 4 hours. Cells were surface stained with Pacific Blue anti-CD4 and APC anti-Thy1.2, and then intracellularly stained with PE-labeled antibody to IL-17 and Alexa Fluor 700–labeled antibody to IFN-γ. All antibodies were obtained from Becton-Dickinson.

Group comparisons of Treg and T-cell populations, pathology scores, and thymidine incorporation were conducted using the Mann–Whitney U test. Survival curves were constructed using the Kaplan–Meier product limit estimator and compared using the log-rank rest. A value of P ≤ 0.05 was deemed to be significant in all experiments.

CD4⁺EGFP⁻ T cells were sorted from Foxp3^EGFP mice and cultured with immobilized anti-CD3 antibody, soluble anti-CD28 antibody, IL-2, and TGF-β. After 72 hours, approximately 60% of the cells expressed EGFP (Fig. 1A). Phenotypic analysis showed that iTregs expressed higher levels of CD4, CD25, and CD44 on their surface as compared with nTregs. Conversely, iTregs expressed lower levels of CD62L, which is typically downregulated upon cell activation, and CD103 than nTregs (Fig. 1B). Intracellular cytokine staining revealed that iTregs had negligible secretion of IFN-γ, IL-17A, or IL-10 (Fig. 1C). To confirm functionality, iTregs were sorted and compared with freshly isolated nTregs from Foxp3^EGFP animals for their ability to suppress an alloresponse. Sorted iTregs were stained with Annexin and 7-AAD to determine whether the cell culture process resulted in any apoptotic cells. Only 3% of cells were observed to be Annexin⁺7-AAD⁻ indicating no significant apoptosis in the cultured cells (data not shown). nTregs were obtained by sorting CD4⁺Foxp3⁺ T cells from Foxp3^EGFP mice. Fluorescence-activated cell sorting (FACS)-sorted iTregs and nTregs were then added to cultures at varying ratios of Tregs to effector T cells (T_eff). We observed that iTregs and nTregs suppressed alloreactive T-cell proliferation to the same extent and in a dose-dependent manner, indicating that these cells have comparable suppressive capability in vitro (Fig. 1D).

To determine in vivo functionality of in vitro differentiated Tregs, we examined whether iTregs were able to suppress GVHD. Lethally irradiated BALB/c mice were transplanted with TCD B6 BM cells alone, TCD BM plus whole splenocytes, or TCD BM and spleen cells along with in vitro differentiated Tregs at a 1:1 ratio. There was no difference in survival between GVHD control animals and mice that were supplemented with iTregs (Fig. 2A), indicating that iTregs did not attenuate GVHD severity. We also observed no difference in serial weight curves between these 2 cohorts of animals (Fig. 2B). To more fully examine the ability of iTregs to suppress GVHD, we repeated these studies with a reduced T-cell dose so that mice in all groups would survive and be able to be examined for GVHD-associated pathology. Histologic analysis of GVHD target organs 24 days posttransplantation revealed that iTregs affected no reduction in pathologic damage in the colon, liver, or lung of transplant recipients (Fig. 2C), further confirming that these cells had no significant protective role and that there was no differential effect on individual tissues. To assess whether lack of protection was attributable to an inadequate dose, we repeated these studies at an iTreg:effector T-cell ratio of 5:1. Again, there was no significant difference in long-term survival in animals that received an escalated dose of iTregs (Fig. 2D) indicating that the inability of these cells to abrogate GVHD did not appear to be due to an inadequate cell number. These results were in contrast to what was observed when mice were transplanted with nTregs. Specifically, transplantation of nTregs at a 1:1 ratio with conventional T cells significantly prolonged survival compared with animals that received no Tregs (Fig. 2E). Collectively, these data show that, in contrast to nTregs, iTregs are unable to prevent GVHD-associated mortality and pathologic damage.

Studies were then conducted to understand why in vitro differentiated Tregs were unable to protect mice from lethal GVHD. We reasoned that possible explanations for the lack of protection from GVHD by iTregs might be that these cells had limited survival or that downregulation of molecules such as CD62L might alter trafficking to GVHD target organs. To address this question, lethally irradiated BALB/c mice were transplanted with TCD B6 BM and spleen cells along with an equivalent number of iTregs. Cohorts of mice were sacrificed on days 7, 10, and 14 post-BMT and GVHD target organs were examined for the presence of EGFP⁺ cells. iTregs were detectable in the spleen, colon, liver, and lung of all recipients on days 7 and 10 posttransplantation (Fig. 3A). Both the percentage (Fig. 3B) and absolute number (Fig. 3C) of these cells was similar at both time points. Notably, iTregs were also present in the peripheral lymph nodes on day 10 indicating that, despite low CD62L expression, these cells could traffic into this tissue site early posttransplantation (Supplementary Fig. S1). By day 14, however, iTregs were essentially undetectable in all GVHD target organs. This was in contrast to what was observed when nTregs were transplanted into recipient mice. In these studies, examination of GVHD target organs showed persistence of nTregs in all tissue sites for up to 60 days posttransplantation with about 1% to 3% of all CD4⁺ T cells expressing EGFP (Fig. 3D and E). Thus, Foxp3-expressing iTregs had limited in vivo persistence in GVHD recipients that were not protected from pathologic damage.

Although the inability to detect iTregs early posttransplantation could have been due to reduced in vivo survival, an alternative explanation was that these cells persisted in recipient animals but had lost expression of Foxp3. In fact, plasticity in the Treg lineage has been recently described whereby Foxp3-expressing cells have been shown to be capable of reverting to a proinflammatory phenotype (31–33). To determine whether the latter explanation was valid, lethally irradiated BALB/c mice were transplanted with B6.PL (Thy 1.1) BM alone, B6.PL BM plus B6.PL whole splenocytes, or B6.PL BM and spleen cells plus B6 Foxp3^EGFP (Thy1.2⁺) iTregs. Using iTregs from congenic Thy1.2 mice allowed us to specifically identify the presence of these cells in otherwise fully donor-engrafted recipients and determine whether these cells had survived but lost EGFP expression. Cohorts of mice were sacrificed at days 3, 5, 7, 10, 14, and 21 and examined for the presence of Thy1.2⁺ cells that retained EGFP expression. We observed that at early time points (days 3, 5, and 7), there was a population of residual Thy1.2⁺ that were of host origin (H-2^d+; data not shown). By day 10, however, mice were completely donor T cell engrafted (Fig. 4A) and we therefore selected this time point for our analysis to eliminate any potentially confounding host populations. A population of CD4⁺Thy1.2⁺ cells was easily discernible in all of the tissue examined (Fig. 4A). The majority of these cells were EGFP⁻ (Fig. 4B), although a small population of Thy1.2⁺EGFP⁺ cells was present in all organs (Fig. 4C). To determine whether these revertant cells had acquired a proinflammatory phenotype, cells were stained for secretion of IFN-γ and IL-17. Whereas these cells produced a negligible amount of IL-17A, 20% to 35% of the revertant iTregs from all tissue sites secreted IFN-γ (Fig. 4D and E). This was noteworthy given that iTregs did not secrete either cytokine immediately after in vitro conversion (Fig. 1C) and was evidence that these cells had acquired a proinflammatory phenotype in vivo. Further longitudinal analysis revealed that iTregs were undetectable by day 21 posttransplantation (data not shown). To determine whether in vitro differentiated iTregs had the potential to induce GVHD, we transplanted animals with these cells alone in the absence of conventional donor T cells. iTregs were detectable in animals 10 days posttransplantation in all tissues sites (Supplementary Fig. S2A). However, mice transplanted with these cells had no mortality and survival was comparable to that observed in animals transplanted with BM cells alone (Supplementary Fig. S2B), indicating that these cells did not appear to contribute significantly to the induction of GVHD.

A possible confounding factor in our interpretation that phenotypic reversion had occurred in vivo was that small numbers of contaminating Thy1.2⁺ non–GFP-expressing cells may have been present in the iTreg sort and then expanded in recipient animals. Because the average sort purity was 98% to 99%, we assumed a worst case scenario whereby 3% of the transferred iTreg population was contaminated by non-iTreg cells. This amounted to an absolute number of 1.8 × 10⁴ cells. To address this question, lethally irradiated BALB/c mice were transplanted with B6.PL BM and whole splenocytes plus 1.8 × 10⁴ Thy1.2⁺ CD4⁺EGFP⁻ cells from Foxp3^EGFP mice. On day 10 posttransplantation, we observed only a negligible population of these cells in the spleen or GVHD target organs (i.e., ≤0.5%; Fig. 4F), indicating that the revertant cells were not the result of contaminating non-iTreg cells in the sort population.

On the basis of prior studies that showed that in vitro exposure to ATRA can enhance the induction of Foxp3 in conventional T cells (34,35), we examined whether ATRA could stabilize Foxp3 expression in iTregs after their adoptive transfer into recipient animals. To examine this question, CD4⁺Foxp3^EGFP− T cells were first cultured in TGF-β and IL-2 alone or in the presence of ATRA. Exposure to ATRA resulted in a modest 10% to 15% increase in Foxp3-expressing cells compared with cells cultured in the absence of ATRA (Fig. 5A). To determine whether RA-iTregs were more effective at suppressing an alloresponse than iTregs, RA-iTregs and iTregs were added to mixed lymphocyte culture (MLC) at varying ratios of Tregs to effector T cells. Both RA-iTregs and iTregs were observed to be equally suppressive as determined by ³H-thymidine incorporation (Fig. 5B). Because RA exposure had no additive effect on the suppressive function of iTregs in vitro, we examined whether these cells had any suppressive role in vivo. Lethally irradiated BALB/c mice were transplanted with B6.PL BM and whole splenocytes (0.6 × 10⁶ αβ T cells) plus 0.6 × 10⁶ Thy1.2⁺Foxp3^EGFP+ RA-iTregs or iTregs. Ten days after transplantation, the majority of RA-iTregs had lost GFP expression (Fig. 5B). Notably, although the number of RA-iTreg–derived Thy1.2⁺EGFP⁻ cells was significantly greater than the number of Thy1.2⁺EGFP⁺ cells, RA-iTregs had a comparable number of revertant cells as the iTreg control group (Fig. 5D). Moreover, revertant cells from the RA-iTreg population produced IFN-γ to the same extent as revertant cells from the control iTreg population (Figs. 5E vs. 4E). Thus, in vitro exposure to RA did not prevent phenotypic reversion in vivo.

A defining characteristic of GVHD is the overproduction of inflammatory cytokines that are able to modulate the intensity of the disease (36–38). We previously have shown that IL-6 is one inflammatory cytokine that has a critical role in modulating the balance between effector and Treg cells during GVHD (28). In fact, blockade of IL-6 signaling is able to significantly reduce GVHD-associated pathologic damage by increasing nTreg numbers and decreasing the absolute number of proinflammatory T cells. On the basis of these prior data, we considered that overproduction of IL-6 might be a factor that contributes to the phenotypic reversion of iTregs in vivo. To test this hypothesis, animals were transplanted with B6.PL BM and spleen cells along with B6 iTregs. Cohorts of mice were then administered either an anti–IL-6R or an isotype control antibody as described in Materials and Methods. On day 10 after transplantation, mice were sacrificed and the percentage of revertant iTregs was determined. Compared with mice treated with isotype antibody, there was a striking increase in the number and percentage of Thy1.2⁺ cells in animals treated with anti–IL-6R antibody (Fig. 6A and B). This resulted in a significant increase in the absolute number of both EGFP⁺ and EGFP⁻ cells compared with isotype control–treated mice. Notably, however, the majority of these Thy1.2⁺ cells also lacked expression of EGFP (Fig. 6A) and a similar percentage secreted IFN-γ (Fig. 6C), indicating that blockade of IL-6 signaling did not prevent phenotypic reversion in vivo.

To understand why treatment with anti–IL-6R antibody resulted in the preferential expansion of Thy1.2⁺ T cells, we examined IL-6R expression on naive CD4⁺ T cells and iTregs pre- and posttransplantation. We observed that the IL-6R was expressed on naive CD4⁺ T cells from donor animals but was not present on in vitro differentiated iTregs that had been cultured in IL-2 and TGF-β (Fig. 6D). Similarly, cells obtained from the spleen of animals 10 days posttransplantation revealed a lack of IL-6R expression on revertant cells which resulted in a significant increase in the number of these cells in all tissues (Fig. 6E). Thus, the expansion of revertant iTregs relative to donor-derived Thy1.1⁺CD4⁺ T cells appeared to be due to a lack of receptor expression that made these cells resistant to antibody blockade.

In vivo, conventional T cells are able to differentiate into either proinflammatory or regulatory cells depending upon the cytokine milieu to which they are exposed. During GVHD, there is an inflammatory environment that directs differentiation along T helper (Th) 1 and Th17 cytokine phenotypes and inhibits the in vivo induction of Tregs from CD4⁺Foxp3⁻ T cells (28). Because treatment with anti–IL-6R antibody was unable to prevent phenotypic reversion of in vitro differentiated iTregs, we examined whether in vivo iTregs were also resistant to IL-6 blockade. To address this question, lethally irradiated BALB/c mice were transplanted with B6 Rag-1 BM cells and sorted CD4⁺Foxp3^EGFP− T cells, and then treated with either anti–IL-6R or control antibody. Anti–IL-6R antibody–treated animals had a significant overall reduction in pathologic damage when compared with control-treated mice (Fig. 7A). Specifically, there was reduced pathology in the lung and colon in treated animals, with a trend for reduced damage observed in the liver. Commensurate with the attenuation of GVHD-related pathologic damage, there was a significant increase in the relative and absolute number of iTregs in GVHD tissue sites when examined 4 weeks posttransplantation (Fig. 7B and D). Collectively, these data showed that, in contrast to what was observed with in vitro differentiated iTregs, blockade of IL-6 signaling was able to augment the in vivo induction of iTregs and this was associated with a significant reduction in the severity of GVHD.

The use of Tregs to facilitate transplantation tolerance is an emerging area in clinical allogeneic stem cell transplantation. On the basis of preclinical murine studies that showed that the adoptive transfer of Tregs could mitigate GVHD, several groups of investigators have begun clinical trials to determine whether this approach will be efficacious in humans (39, 40). Obtaining sufficient numbers of these cells for transfer into individual patients as well as insuring that the acquired cell population is not contaminated by potentially alloreactive non-Treg cells have been ongoing challenges with this strategy (24). An alternative approach to circumvent these potential obstacles is predicated on the in vitro expansion of Tregs that can be converted from conventional CD4⁺ T cells in the presence of IL-2 and TGF-β. Under these conditions, large numbers of in vitro derived iTregs can be obtained for the subsequent administration into recipients. This strategy, however, is contingent upon the underlying assumption that these cells will function in a similar manner to nTregs with respect to suppression of GVH reactivity. In the current study, we examined this question to determine the functional capability of these cells to ameliorate GVHD in vivo. These studies showed that in vitro differentiated iTregs were not effective for the prevention of GVHD despite their ability to suppress alloreactive T cells responses in vitro. The failure to mitigate GVHD was attributable to limited in vivo survival, instability of Foxp3 expression, and the subsequent reversion of these cells to a Th1 cytokine phenotype.

Our results, along with those of other investigators, suggest that the ability of in vitro differentiated iTregs to suppress disease may be a function of the magnitude of the inflammatory milieu. For example, Haribhai and colleagues (26) showed that these cells could protect mice from colitis in a T-cell transfer model. Similarly, Zheng and colleagues (41) showed that in vitro differentiated iTregs could prolong survival in an autoimmune model characterized by a lupus-like syndrome. Both models, however, are characterized by less inflammation than is observed in GVHD where the administration of a lethal TBI conditioning regimen exacerbates the inflammatory milieu induced by a T-cell–mediated alloresponse. In that regard, our results are consistent with those of Koenecke and colleagues (42) who recently showed that alloantigen-specific iTregs generated in vitro after coculture with activated dendritic cells had a negligible effect on protection from lethal GVHD. The current study, however, extends these findings by showing that the loss of Foxp3 expression is associated with the acquisition of an inflammatory phenotype characterized by IFN-γ secretion with no evidence of IL-17 production. Thus, in the presence of an inflammatory environment induced by GVHD, there is a reversion of a suppressive T-cell population to one with a Th1 cytokine phenotype.

One of the primary observations of these studies was the limited in vivo survival of iTregs which were essentially undetectable after 2 weeks, even after accounting for revertant cells. This was observed in all GVHD target organs examined indicating no preferential localization of these cells at sites of pathologic damage. This finding is consistent with an earlier report by Selvaraj and Geiger (25) who showed that only a small percentage of iTregs were able to survive in vivo after the adoptive transfer into immunocompetent animals. In that study, iTregs disappeared from nearly all organs within the first 2 weeks with the exception of the bone marrow and lymph nodes. This was distinctly different from what we observed when nTregs were transferred into recipient animals where Foxp3⁺ cells were observed for up to 8 weeks post-BMT. In the latter case, the stability of Foxp3 expression correlated with protection from lethal GVHD.

Prior studies have shown that RA is able to induce differentiation of Tregs from conventional T cells through both direct and indirect effects. Hill and colleagues reported that RA promoted iTreg generation by relieving the inhibition that was mediated indirectly by CD4⁺CD44^hi T cells (43). Conversely, others have noted that RA can directly enhance TGF-β–mediated Foxp3 induction, whereas inhibiting the generation of Th17 cells (34,35). On the basis of these studies, we examined whether RA could enhance the stability of Foxp3-expressing iTregs in vivo and prevent phenotypic reversion in lethally irradiated transplant recipients. Although treatment with RA increased iTreg conversion, the in vitro exposure to this agent did not prevent phenotypic reversion in vivo after adoptive transfer. These results differ from what was observed in a murine model where animals immunized with ovalbumin had less reversion if they received RA-iTregs (44). The degree of inflammation and the extent of reversion in control animals, however, were notably less in that model and suggest that other cytokines and/or environmental factors are responsible for the phenotypic reversion which occurs after adoptive transfer of iTregs during GVHD.

We reasoned that IL-6 might be one cytokine that contributed to instability of Foxp3 expression given prior studies which have shown that IL-6 deleteriously affects the generation of iTregs induced by TGF-β by blocking Foxp3 expression (45). This has been shown to occur by upregulation of the TGF-β inhibitor SMAD7 (8). IL-6 has also been shown to play a pivotal role in driving the differentiation of naive T cells to become Th17 cells and in inhibiting the generation of CD4⁺Foxp3⁺ from CD4⁺Foxp3⁻ T cells (46). Moreover, Horwitz and colleagues (47) have reported that iTregs are more resistant to the inhibitory effects of IL-6 compared with nTregs. Thus, we considered that transfer of these cells might be more effective if the effects of IL-6 which is significantly increased during GVHD (28) could be blunted. Blockade of IL-6 signaling, however, had no effect on stabilization of Foxp3 expression or preventing phenotypic reversion. Further analysis revealed that in vitro differentiated iTregs did not express the IL-6R after either in vitro conversion or in vivo isolation of this same population after transfer. Downregulation of the IL-6R on iTregs has been shown to inhibit signaling though IL-6 (41), providing an explanation for why these cells were not affected by antibody administration. Downregulation of the receptor is also a potential explanation for the absence of IL-17 secretion in revertant iTregs as the inability to respond to IL-6 would prevent Th17 differentiation (41).

One of the main findings was the observation that in vitro and in vivo generated iTregs appear to be differentially regulated by IL-6. Specifically, blockade of IL-6 signaling was able to augment the in vivo induction of Tregs from the conventional CD4⁺ T-cell pool. Under these conditions, mice treated with antibody blockade had an increased number of Foxp3⁺ cells in all tissue sites due to enhanced in vivo conversion which correlated with augmented GVHD protection. Because naive T cells express the IL-6R, we surmise that, in contrast to in vitro differentiated iTregs, antibody administration was able to direct the differentiation of these cells toward a regulatory pathway due to their susceptibility to IL-6 signaling blockade. These results have direct clinical relevance given the availability of tocilizumab which is a humanized version of this antibody and has been Food and Drug Administration (FDA) approved for the treatment of patients with rheumatoid arthritis. Thus, administration of this antibody may be a means to augment Treg development in vivo without the need for more time consuming and costly cellular therapy approaches.

In summary, these studies show that in vitro differentiated iTregs are ineffective at attenuating the severity of GVHD. This lack of protection is characterized by instability of Foxp3 expression which results in reversion to an inflammatory Th1 cytokine phenotype. Reversion is not due to increased production of IL-6 nor is in vitro exposure to retinoic acid able to stabilize iTregs, indicating that other mechanisms are responsible for this unstable phenotype. These results have implications for the potential use of in vitro differentiated iTregs for the abrogation of GVHD and suggest that additional studies that are focused on augmenting the stability of Foxp3 expression in these cells is warranted before their use is contemplated in the clinical setting. Finally, blockade of IL-6 signaling is an effective approach to augment the generation of iTregs in vivo which may be a clinically feasible strategy to enhance Treg-mediated suppression and reduce GVHD severity.

No potential conflicts of interest were disclosed.

A. Beres designed and conducted research, analyzed data, and wrote the manuscript. R. Komorowski conducted all pathologic analysis. M. Mihara provided vital reagents. W.R. Drobyski designed and supervised research, analyzed data, and wrote the manuscript.

This research was supported by grants from the National Institutes of Health (HL064603, HL081650, and DK083358 to W.R. Drobyski) and by an award from the Midwest Athletes Against Childhood Cancer Fund (Milwaukee, WI; W.R. Drobyski).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.