|Home | About | Journals | Submit | Contact Us | Français|
Acute graft versus host disease (GvHD) is a major cause of mortality in allogeneic bone marrow transplantation (BMT), for which administration of FoxP3+ Treg cells has been proposed as a therapy. However, the phenotypic stability of Treg cells is controversial and cytokines that signal through the transcription factor STAT3 can inhibit FoxP3 expression. We assessed whether the elimination of STAT3 in T cells could limit the severity of GvHD, and if so, what mechanisms were involved. We found STAT3 limited the numbers of FoxP3+ Treg cells following allogeneic BMT by two pathways: instability of nTregs and inhibition of iTreg cell polarization from naïve CD4+ T cells. Deletion of STAT3 within only the nTreg cell population was not sufficient to protect against lethal GvHD. In contrast, transfer of STAT3-deficient naïve CD4+ T cells increased FoxP3+ Treg cells post-BMT and prevented lethality, suggesting that the consequence of STAT3-signaling may be greater for inducible rather than natural Treg cells during GvHD.
Allogeneic bone marrow transplantation (BMT) can be lifesaving therapy for treatment of relapsed acute leukaemia and lymphoma, as well as inherited disorders of the bone marrow and immune system (Atkinson, 1992). However, its efficacy is limited by the morbidity and mortality associated with graft versus host disease (GvHD). Acute GvHD is associated with activation of donor T cells and their secretion of proinflammatory cytokines, resulting in inflammation in the gut, liver and skin. The incidence of severe acute GvHD occurs in 50% of patients who receive an HLA-matched, but unrelated donor allograft and less than 20% of patients who develop severe GvHD survive past five years post transplant (Kernan et al., 1993). Consequently, development of effective treatment of GvHD remains an important goal in BMT.
Mouse models of GvHD after allogeneic bone marrow transplantation (BMT) have shown that both CD4+ T cells (T helper, Th cells) and CD8+ T cells (T cytotoxic, Tc cells) mediate acute GvHD (Vallera et al., 1994). The traditional model of Th cell differentiation into either Th1 or Th2 cells has now been modified by the identification of two additional lineages generated in the presence of transforming growth factor beta (TGF-β) (Weaver and Hatton, 2009). The first, STAT5-dependent T regulatory cells (Treg) cells, express the transcription factor FoxP3 and inhibit inflammatory responses (Sakaguchi, 2011). Treg cells can be found naturally (natural Treg: nTreg) in healthy mice where the majority are derived during thymic development. Alternatively, FoxP3+ cells can be induced (induced Treg: iTreg) from naïve Th cells by stimulating them in vitro in the presence of TGF-β and IL-2 (Chen et al., 2003). The second lineage consists of STAT3-dependent Th17 cells, which are characterized by secretion of IL-17 and are associated with autoimmune disease (Bettelli et al., 2007). The roles of these helper cell lineages in contributing to the pathology of GvHD remain controversial (Yi et al., 2009). Nonetheless, specific Th cell lineages may contribute to the organ specificity of GvHD, with Th1 cells mediating gut and liver GvHD (Nikolic et al., 2000), Th2 cells mediating lung GvHD (Nikolic et al., 2000) and Th17 cells responsible for skin GvHD (Carlson et al., 2009).
In contrast, Treg cells suppress a wide variety of T cell inflammatory diseases including experimental GvHD mediated by mouse (Cohen et al., 2002; Edinger et al., 2003) and human T cells (Amarnath et al., 2010; Mutis et al., 2006). Adoptive transfer of Treg cells has been proposed as a therapy to prevent GvHD (June and Blazar, 2006). However, the success of an adoptive Treg cell therapy response is reliant upon the fidelity of the transplanted cells, which has recently been called into question (Tang et al., 2012). This notion of nTreg instability remains a contentious issue with different groups reaching opposite conclusions (Rubtsov et al., 2010; Zhou et al., 2009).
Whereas STAT5 is a key positive regulator of FoxP3 expression, STAT3 is as an important inhibitor (Yao et al., 2007). STAT3 binds to a silencer element within the Foxp3 locus and is associated with a reduction in Smad3 binding (Xu et al., 2010). The importance of cytokines that signal through STAT3 has been demonstrated in murine models of GvHD. Lu et al found that allo-activated T cells induced in mouse models of acute GvHD are characterized by phosphorylation of STAT3 (Lu et al., 2008). IL-6, IL-21 and IL-23, cytokines that activate STAT3, are all necessary for the development of acute GvHD (Bucher et al., 2009; Chen et al., 2009; Das et al., 2009). Conversely, inhibition of IL-6 was associated with the appearance of induced Tregs; however, the potential effects of IL-6 on transferred nTreg cells were not investigated (Chen et al., 2009).
Given the potential therapeutic importance of Treg-based therapies, we set out to investigate the role of STAT3 in donor T cells during acute GvHD. We used T cells from Foxp3-GFP reporter mice, which allowed us track the in vivo fate of the nTreg and iTreg cell populations in the inflammatory milieu of acute GvHD. We show that nTreg cells lose expression of FoxP3 within this inflammatory environment and that this loss of FoxP3 is in part STAT3-dependent. Finally, the absence of STAT3 permitted the conversion of transferred naive CD4+ T cells to become iTreg cells, which correlated with strikingly improved survival during GvHD.
To explore the role of STAT3 in allogeneic BMT, mice with conditional deletion of Stat3 in T cells on a B6 background were used as a source of donor T cells. We analyzed the effect of transplanting allogeneic (BALB/c) animals with T-depleted bone marrow (BM) with or without T cells from STAT3-deficient or control animals. The majority of animals that received T-depleted BM supplemented with control T cells rapidly lost weight and died within 14 days post-BMT (Figure 1A and 1B). In contrast, all recipients of T-depleted BM alone or T-depleted-BM supplemented with STAT3-deficient T cells survived (Figure 1A and 1B) (P<0.001). Relative to control T cell recipients, STAT3-deficient T cell recipients had reduced overall GvHD by histologic analysis; STAT3-deficient T cell recipients were particularly protected against GvHD of the skin and colon (Figure 1C). Thus, T cell STAT3 signaling was necessary for the development of acute GvHD. We next sought to determine potential mechanisms underlying the differences in survival.
STAT3 is a critical factor for the generation of Th17 cells (Hirahara et al., 2010). A role of IL-17 has been explored in mouse models of acute GvHD, but its importance remains controversial (Kappel et al., 2009; Yi et al., 2008). We therefore investigated the expression of this cytokine in our model of acute GvHD. We observed a higher proportion of IFN-γ producing splenic CD4+ T cells in the setting of allogeneic BMT compared with syngeneic BMT (Figure 2A and 2B); the increase in IFN-γ secretion was less marked in STAT3-deficient T cells compared with T cells from control animals (P=0.006). However, there were few IL-17 secreting splenic CD4+ T cells in either syngeneic or allogeneic BMT; of note, the rare events that were detected consisted of IL-17+IFN-γ+ T cells (Figure 2A, lower left panel).
CD4+ cells from the gut lamina propria (LP) contain abundant numbers of IL-17+ cells in healthy mice (Ivanov et al., 2006). The colon was a site where there was a significant difference in the degree of acute GvHD pathology between mice that received STAT3-deficient T cells compared with control T cells (Figure 1C). We therefore investigated the influence of STAT3 in colonic LP T cells harvested during the second week post-BMT. Twelve days after syngeneic BMT using control T cells, we observed that approximately 10% of LP CD4+ lymphocytes secreted IL-17 (Figure 2C, top left panel); as predicted, recipients of STAT3-deficient syngeneic T cells had an approximate 90% reduction in LP CD4+ T cells capable of IL-17 secretion (P=0.0002) (Figure 2C, top right panel). Surprisingly, in allogeneic host animals receiving control T cells, there was a profound reduction in the number IL-17+ LP lymphocytes compared with that seen in mice receiving syngeneic control T cells (P=0.00012) (Figure 2D). In summary, recipients of control or STAT3-deficient allogeneic T cells each had minimal numbers of IL-17 secreting, LP CD4+ T cells (Figure 2C, lower panels). The few IL-17+ allogeneic T cells seen in the LP were IL-17+ IFN-γ+ T cells. Thus, the inflammatory colitis associated with acute GvHD, far from being dependent on Th17 cells, was associated with their reduction. These data further indicate that the ability of STAT3 to modulate acute GvHD was not explained simply by effects on Th17 cells, thereby implying an alternative mechanism.
We next turned to the fate of Treg cells after BMT as determined by FoxP3 expression within the transplanted T cells. An alternative possibility is that STAT3 was responsible for inhibiting FoxP3 expression (Xu et al., 2010). We hypothesized that STAT3-deficient Treg cells would be maintained in vivo, thereby reducing acute GvHD. Initially, studies were performed to rule-out the possibility that STAT3-deficient mice might contain nTregs that were quantitatively or qualitatively different from control nTregs. To facilitate these studies, FoxP3 expression was determined by GFP fluorescence as both STAT3-deficient and control animals contained a Foxp3-GFP reporter transgene. First, we found that the basal frequency of FoxP3+ cells was similar in control and STAT3-deficient mice (Figure S1A and S1B). Second, CpG methylation in the Treg specific demethylation region (TSDR) (Floess et al., 2007) was similar in sorted control and STAT3-deficient nTreg cells (Figure S1C). Third, we found that control and STAT3-deficient FoxP3+ T cells had a similar capacity to prevent naïve T cell proliferation after TCR stimulation (Figure S1D and S1E). In addition, control and STAT3-deficient FoxP3+ T cells had an equivalent capacity to inhibit naïve T cell proliferation (Figure S1F), IFN-γ secretion (Figure S1G), and IL-2 (Figure S1H) secretion in response to fully MHC disparate antigen presenting cells.
As STAT3-deficient nTreg cells were quantitatively and qualitatively similar to control nTreg cells in vitro, there existed sufficient validity to compare whether such T cell populations were differentially maintained in vivo post-BMT. Administration of control T cells into syngeneic hosts resulted in a frequency of FoxP3+ T cells in peripheral lymph nodes at day 14 post-BMT, which was modestly reduced in recipients of STAT3-deficient T cells (Figure 3A, left panels). In contrast, administration of control T cells into allogeneic hosts resulted in a low frequency of FoxP3+ T cells that was greatly diminished relative to recipients of STAT3-deficient T cells (P=0.0002) (Figure 3A, right panels). Similar results were obtained in an analysis of colonic lamina propria lymphocytes (Figure 3B): that is, a significant population of control lymphocytes expressed FoxP3 after syngeneic but not allogeneic BMT (P=0.011), whereas STAT3-deficient T cells yielded substantial populations of FoxP3+ cells after both syngeneic and allogeneic BMT.
We reasoned that the differential numbers of FoxP3+ cells observed after allogeneic transplant using control vs. STAT3-deficient T cells may have been due to two possibilities: (1) transferred control Treg cells lost FoxP3 expression; or (2) transferred naïve T cells were unable to be induced into a Treg phenotype after allogeneic BMT. In both cases, this would be due to STAT3 signaling. With respect to the first possibility, investigators have reached opposite conclusions in their respective models of immune-mediated, inflammatory disease (Oldenhove et al., 2009; Rubtsov et al., 2010; Zhou et al., 2009). We approached this issue with a variety of BMT conditions (syngeneic vs. allogeneic; T cell-depleted vs. T cell-replete).
In the setting of T cell-depleted allogeneic BMT, mice received a pure (99% FoxP3-GFP+) population of control or STAT3-deficient nTregs and the fate of the transferred FoxP3+ cells could be tracked using congenic markers, along with FoxP3-GFP expression. Two weeks after the nTreg cells were transferred, the mice remained healthy and over 90% of the post-BMT T cells remained FoxP3+ , irrespective of whether the transferred nTreg cells expressed STAT3 (Figure 4A). In a second experiment, irradiated syngeneic and allogeneic hosts received BM, nTreg-depleted T cells and FoxP3+ cells from control animals. In syngeneic host animals, the cells retained their FoxP3 expression (Figure 4B, left panel). In contrast, in the presence of effector T cells within allogeneic hosts, the transferred FoxP3+ cells lost FoxP3 expression (P=0.021) (Figure 4B, right panel) and acquired either IL-17 or IFN-γ expression. In a third experiment, we explored whether STAT3-deficient Treg cells were resistant to this conversion in allogeneic host animals. Irradiated allogeneic hosts were transplanted with BM, nTreg-depleted T cells and FoxP3+ cells from control or STAT3-deficient donors. Consistent with the prior experiments, most of the control nTreg cells lost FoxP3 expression (Figure 4C, left panel) whereas most of the STAT3-deficient nTreg cells retained FoxP3 expression (P=0.0002) (Figure 4C, right panel).
We next sought to determine if we could recapitulate the in vivo loss of FoxP3 in nTregs using an in vitro model that contained an inflammatory cytokines elaborated during acute GvHD. In preliminary experiments, various combinations of pro-inflammatory cytokines were tested (data not shown). As demonstrated in Figure 5A, FoxP3 expression was inhibited by IL-6 compared to cells cultured in cytokine free media. Of note, IL-27 also had the capacity to down-regulate FoxP3. IL-6 predominantly exerts its effect via STAT3 and we found that the ability of IL-6 to down-regulate FoxP3 was blocked in the absence of STAT3. In contrast with IL-6, IL-27 signals principally through STAT1 (Stumhofer et al., 2006). Accordingly, we found that the inhibitory effect of IL-27 on FoxP3 expression was in part independent of STAT3. We next measured cytokine expression in the converted nTreg cells (Figure 5B). We found that nTreg cells, which lost FoxP3 expression in the presence of IL-6, produced IL-17. These data are consistent with previous reports (Huber et al., 2008; Xu et al., 2007). The anti-inflammatory cytokine IL-10 is an activator of STAT3. However, addition of IL-10 alone or in conjunction with IL-6 did not alter the proportion of FoxP3+ cells recovered after six days in either control or STAT3 deficient cells (data not shown). This is consistent with previous reports demonstrating divergent functions of IL-6 and IL-10 despite the ability of both cytokines to activate STAT3 (Murray, 2006).
Several groups have previously shown that STAT5 is essential for expression of FoxP3. Its actions appear to be direct insofar as STAT5 binds multiple sites within the Foxp3 locus (Yao et al., 2007; Zorn et al., 2006). The third STAT5 binding site (Target III) has also been identified as a STAT3 binding site in both mice and humans (Xu et al., 2010; Zorn et al., 2006). We considered if activation of STAT3 would inhibit STAT5 binding to the Foxp3 gene. The data in Figure 5C demonstrate that significantly more STAT5 bound to this region of Foxp3 locus when STAT3 was not present (P=0.002).
The differential expression of IL-17 seen in nTreg cells that lost FoxP3 expression in the presence of IL-6 versus IL-27 led us to characterize the cytokine secretion profile of the transferred nTreg cell population. Irradiated allogeneic BALB/c mice received BM transplants containing a mixture of effector and natural T cells that could be differentiated by congenic markers. Ten days post transplant, IFN-γ and IL-17 expression was determined in the CD45.1+ (effector T cell) and CD45.2+ (nTreg) cell populations. As shown in Figure 2, a substantial population of effector T cells expressed IFN-γ and few cells stained positive for IL-17 (Figure 6A, left panels). In contrast, the control nTreg cells that had lost FoxP3 expression had a different pattern of cytokine expression: Relative to effector cells, these converted nTreg cells became IL-17 producers more frequently (P=0.024) and had less of a propensity to express IFN-γ (P=0.04) (Figure 6A, right panels, Figure 6B). It is important to note that the proportion of cells derived from the transferred CD45.2+ nTreg population made up less than 4% of the total CD4+ population, irrespective of whether the transferred cells were from control or STAT3-deficient animals (Figure S2). The total number of transferred FoxP3+ nTreg, FoxP3− IFN-γ+ and FoxP3− IL-17+ cells is indicated in Figure 6C. Although consistent with the data in Figure 4C and and5B,5B, only the differences in IL-17 expression were significant (P=0.046).
Recently, work by Miyao and colleagues have demonstrated that not all FoxP3−GFP+ nTreg cells express high amounts of CD25 and that CD25−FoxP3+ cells are less able to retain FoxP3 expression compared with CD25hi FoxP3+ cells (Miyao et al., 2012). In view of this we investigated the proportion of FoxP3+ nTreg cells that expressed CD25 in control and STAT3 deficient T cells, and found no difference in CD25 expression (Figure S3). Next we repeated the experiments described in Figure 6 using CD25hi FoxP3+ nTreg cells, which were transplanted into irradiated allogeneic host animals together with congenic naïve CD4+ T cells. Consistent with our earlier work, a substantial proportion of control CD25hiFoxP3+ nTreg lost FoxP3 expression after 10 days (Figure S3).
Next, we evaluated whether STAT3-deficient nTreg cells would be better able to control GvHD induced by co-administered T cells. We transplanted mice with nTreg cells and effector T cells, as previously described (Edinger et al., 2003); in this setting, we then compared the ability of wild type and STAT3-deficient nTreg cells to inhibit GvHD. The data in Figure 6D demonstrate that co-administration of STAT3-deficient nTregs, despite the maintenance of FoxP3+ expression in the natural Treg cell pool post-BMT, did not improve survival compared with recipients of wild type nTreg cells. From this, we concluded that neither the retention of FoxP3 expression, nor the expression of inflammatory cytokines by the converted control nTreg cells, was sufficient to explain the survival benefit seen in allogeneic mice transplanted with STAT3 deficient T cells.
We investigated the second possibility, namely, that STAT3-deficient naïve Th cells might be predisposed to become iTreg cells. The ability of naïve cells to acquire FoxP3 has been noted in a number of circumstances both in vitro (Chen et al., 2003) and in disease models in vivo (You et al., 2007). To address this, the fate of purified, naïve, FoxP3− Th cells isolated from control and STAT3-deficient donors was tracked after syngeneic and allogeneic transplantation. After syngeneic BMT, recipients of control and STAT3-deficient naïve, FoxP3− T cells had similar reconstitution with FoxP3+ T cells (Figure 7A, left panels). In marked contrast, after allogeneic BMT, FoxP3 induction was greatly diminished in recipients of control T cells, but relatively preserved in recipients of STAT3-deficient T cells (P=0.007) (Figure 7A, right panels). Thus, the inflammatory environment within allogeneic host animals appeared to inhibit the induction of FoxP3 expression in transferred naïve T cells and STAT3 was necessary for this inhibition.
We next explored whether a mixture of wild type and STAT3-deficient T cells would alter the ability of STAT3-deficient naive T cells to acquire FoxP3 in allogeneic hosts. To test this, equal proportions of naïve wild type CD45.1+ Th cells and CD45.2+ Th cells from control or STAT3-deficient animals were transferred into irradiated allogeneic hosts. Consistent with previous experiments, we found that there was no significant difference in the total number of recovered CD4+ T cells between all three groups; however, the ratio of CD45.1+/CD45.2+ cells was higher if the CD45.2+ population was STAT3-deficient (P=0.022) (Figure 7B). Next, we determined the percentage of naive T cells that had acquired FoxP3 in the CD45.2+ population. We found that the likelihood of acquiring FoxP3 expression was significantly higher (P=0.025) if the cells lacked STAT3 compared with control cells, irrespective of the presence of wild type T cells (Figure 7C). Similarly, the presence of STAT3 deficient cells did not alter the inability of wild type T cells to acquire FoxP3 (Figure S4A).
Finally, we evaluated whether the STAT3-dependent inhibition of iTreg differentiation post-transplant was associated with the development of lethal GvHD. Irradiated mice received allogeneic BM together with naïve CD4+ T cells. Allogeneic hosts that received wild-type, naïve CD4+ T cells rapidly lost weight and died in the first two weeks post-BMT (Figure 7D). In contrast, recipients of STAT3-deficient naïve CD4+ T cells were protected against GvHD lethality (P=0.037). After 10 days post-BMT, the total numbers of CD4+ T cells were determined: there were no significant differences regardless of whether the mice had received control or STAT3-deficient T cells (Figure S4B). Therefore, in the absence of transferred nTreg cells, the development of iTreg cells within the mice receiving STAT3 deficient T cells appeared to be sufficient to limit lethal GvHD.
Acute GvHD remains a principal cause of post-BMT morbidity and mortality. Supplementation of transplanted BM with nTreg cells has been proposed to be a strategy to reduce acute GvHD, in addition to other immune-mediated diseases (June and Blazar, 2006). Nonetheless, the potential success of these novel approaches is predicated on phenotypic stability of Treg cells and their resistance to the action of inflammatory cytokines elaborated during GvHD. In this study, we have demonstrated murine recipients of an allogeneic bone marrow transplant with T cells deficient in STAT3 had a significantly prolonged survival compared with animals that received an allogeneic transplant with control T cells.
The discovery of Th17 cells has provided insights into mechanisms of immune-mediated disease and the importance of STAT3 in the generation of Th17 cells has been substantiated in both mouse and humans (Hirahara et al., 2010). We therefore first considered that the beneficial anti-GvHD effect of transplanting STAT3-deficient T cells was likely related to their inability to produce IL-17. However, several findings argue against this as the major mechanism operational in our model. First, we found that cytokine expression after allogeneic BMT was similar in recipients of control or STAT3-deficient T cells. Second, we found a striking difference in the number of Th17 cells within the colonic lamina propria of syngeneic animals and allogeneic animals transplanted with wild type T cells; of note, this comparison was not explored in previous work (Bucher et al., 2009; Chen et al., 2009; Das et al., 2009). Compared with syngeneic hosts, there were remarkably few IL-17 secreting CD4+ T cells in allogeneic hosts despite severe acute GvHD. This indicated that IL-17 was not a major mediator of acute GvHD in our model, and is consistent with the finding that deleting Il17 in T cells has only minor effects in acute GvHD (Kappel et al., 2009). In summary, these findings suggest that the inflammatory colitis associated with acute GvHD is not dependent on Th17 cells, and further suggest that the beneficial effect of deleting STAT3 in donor T cells can not be attributed to inhibition of Th17 cells.
Previous studies have indicated that administration of nTreg cells can reduce the severity of murine acute GvHD and other autoimmune diseases (Cohen et al., 2002; Edinger et al., 2003). Given this literature, it is interesting to note that one of our most dramatic findings was the loss of FoxP3+ in transplanted nTregs with their subsequent conversion into cytokine producing effector cells. The phenomenon of Th cell lineage plasticity has been the subject of much interest and research activity (Wei et al., 2009). A number of groups have explored the stability of FoxP3 expression in nTreg cells and recent publications have made the subject more rather than less controversial. Several groups have shown that in the presence of inflammation, FoxP3 expression is either inhibited or lost (Oldenhove et al., 2009; Zhou et al., 2009). Zhou et al, using Foxp3-GFP reporter animals expressing a diabetogenic T cell receptor, demonstrated that one-third of adoptively transferred Treg cells lost FoxP3 expression and became effector cells (Zhou et al., 2009). In contrast, FoxP3 has been reported as able to induce its own expression, and nTreg cells have been shown to remain FoxP3+ following adoptive transfer into healthy host animals (Gavin et al., 2007). Using a similar reporter strategy, Rubtsov et al noted that nTreg cells remained FoxP3+ in a number of inflammatory environments, including Listeria monocytogenes infection, the NOD.BDC2.5 model of type 1 diabetes, and the K/BxN model of inflammatory arthritis (Rubtsov et al., 2010).
Recently, work by Miyao and colleagues have argued that a small proportion of FoxP3+ Th cells that are CD25− may be responsible for this controversy (Miyao et al., 2012). However, we believe the present work supports the contention that nTreg cells may truely be de-stabilized. We believe that our findings represent actual, rapid conversion of FoxP3+ cells into FoxP3− cells; that is, using flow cytometry sorting, we transferred pure populations of nTreg cells, and were able to reliably differentiate different populations of transplanted cells using congenic markers. In the absence of effector cells and the associated GvHD, more than 90% of the transferred nTreg maintained FoxP3 expression. When effector T cells were added, FoxP3 expression was maintained in syngeneic hosts but was essentially extinguished in allogeneic host animals. Furthermore, we obtained similar results using CD25hi nTreg cells; and thus, we do not believe the results are explained by the outgrowth of Foxp3+ CD25− cells. Thus, our findings suggest that the ability of STAT3 to sense an inflammatory environment is a key factor in the phenotypic destabilization of nTreg cells. This biology may have important clinical translational significance: in the setting of profound immune cell activation, the approach of merely providing Treg cells may prove problematic as long as STAT3-activating cytokines are present.
There remains the question as to precisely how STAT3 serves to inhibit FoxP3 expression in T cells. The present work substantiates previous findings that nTreg cells may be destabilized in vitro in a STAT3-dependent manner (Xu et al., 2007; Xu et al., 2010; Yang et al., 2008). Zorn and colleagues first identified a region in the Foxp3 locus that binds both STAT3 and STAT5 (Zorn et al., 2006) at a site that we subsequently referred to as target site III (Yao et al., 2007). Subsequently, Xu et al argued that this site resides within a second enhancer region and mediated the inhibitory actions of STAT3 on FoxP3 gene transcription (Xu et al., 2010). In the present work, we show that presence of STAT3 inhibits STAT5 binding to this element. In contrast with the action of STAT3, STAT5 is known both promote and maintain iTreg FoxP3 expression (Chen et al., 2011). In the setting of T. gondii infection Oldenhove and colleagues have demonstrated that Treg numbers decline, an effect that was reversed by the addition of IL-2, a potent inducer of STAT5 (Oldenhove et al., 2009). Forced expression of STAT5 in donor T cells has been shown protect mice from acute GvHD by enhancing the induction of Treg cells (Vogtenhuber et al., 2010). Taken together, our data support the notion that the inhibitory actions of STAT3 may act in part, by interfering with the ability of STAT5 to bind the Foxp3 locus and promote gene expression, thereby providing an important mechanism by which STAT3 can destabilize Treg cells.
We were struck by the finding that STAT3-deficient nTregs, in spite of their in vivo persistence, were no more effective than wild-type nTregs in ameliorating GvHD. Chaudry et al found that STAT3-deficient nTregs were functionally impaired in vivo, although their primary defect related to impaired ability to constrain IL-17-mediated pathology (Chaudhry et al., 2009). Consistent with this prior publication, we noted that STAT3-deficient nTregs suppressed T cell proliferation in vitro to a similar extent as control nTreg cells. Subsequently, it has been shown that STAT3-deficient nTregs have reduced in vivo efficacy due to a deficiency in IL-10 secretion (Chaudhry et al., 2011; Huber et al., 2011). As previous studies found that Treg cells inhibit GvHD in part through their secretion of IL-10 (Hoffmann et al., 2002), it is possible that a deficiency in IL-10 secretion in STAT3-deficient nTregs may account in part for their inability to inhibit GvHD in spite of preservation of FoxP3 expression.
Despite the fact that transferred STAT3-deficient nTreg cells were unable to improve survival during acute GvHD, administration of STAT3-deficient naïve T cells was sufficient to significantly improve survival even in the absence of any transferred nTreg cells. In our model of GvHD, wild type donor cells nearly universally became effector T cells. However, a substantial portion of transplanted STAT3-deficient T cells induced FoxP3 expression, thus acquiring an iTreg phenotype. This observation is consistent with our previous results using a T cell transfer model of colitis (Durant et al., 2010). Taken together, these results argue that STAT3-signaling cytokines limit the numbers of FoxP3+ T cells by two pathways: (1) enhanced instability of nTreg cells; and (2) failure to generate new iTregs from naïve CD4+ T cells. In the context of a profound immunologic disease such as GvHD, the appearance of iTreg cells from naïve precursors appears to be of paramount importance for improving survival, whereas the simple provision of even a STAT3-deficient nTreg may have limited therapeutic efficacy.
The present data help to further explain the beneficial effects of blocking IL-6 during GvHD (Chen et al., 2009). Transplantation with STAT3-deficient naïve T cells resulted in an accumulation of iTreg cells, similar to that seen during IL-6 blockade. In our hands, it was the promotion of iTreg accumulation rather than protection of the nTreg phenotype that represented the primary pathway whereby STAT3-deficient T cells limited the pathology associated with acute GvHD. These data provide further mechanistic rationale for efforts to prevent GvHD by direct inhibition of STAT3 expression or DNA binding, or indirect inhibition of STAT3 via combination neutralization of IL-6, IL-21 and IL-23.
CD45.1+ C57BL/6 (B6) and BALB/c mice were purchased from the Jackson Lab (Bar Harbor, ME). Foxp3-GFP B6 reporter mice were obtained from Dr. M Oukka, Harvard Medical School, Cambridge, MA (Bettelli et al., 2006) and Stat3fl/fl animals were obtained from Dr. Levy (University of New York) (Lee et al., 2002). Stat3fl/fl animals were initially crossed with CD4-Cre and subsequently with FoxP3 to generate Foxp3-GFP;CD4-Cre;Stat3fl/fl (STAT3KO, STAT3-deficient) and Foxp3-GFP;Stat3fl/fl littermate controls (control) (Durant et al., 2010).
BM was flushed from CD45.1 B6 donor femurs and tibias and T cell depleted using negative selection by MACS beads (Miltenyi Biotec, Bergisch Gladbach, Germany). Host allogeneic (BALB/c) or syngeneic (B6) mice were conditioned with total body irradiation of 950 cGy in two divided doses three hours apart before being rescued with 5×106 BM cells together with 2.5×106 T cells from either STAT3-deficient or control donors. For experiments where mice were transplanted with nTreg cells, CD4+ CD45.2+ FoxP3-GFP+ cells were purified by flow cytometry using a FACSaria (BD Biosciences, San Diego, CA). 5×105 Treg cells were transplanted in combination with 5×106 T depleted BM cells alone or in combination with 2×106 (figure 5A – C) or 1×106 (Figure 5D) CD25− CD45.1+ T cells isolated by MACS bead isolation. Flow cytometry was used to determine the proportion of FoxP3-GFP+ transplanted T cells that remained FoxP3+ two weeks post-BMT. The transferred B6 FoxP3 cells were differentiated from BALB/c host cells by expression of H-2Kb; both populations could be differentiated from transferred B6 BM and effector T cells by use of congenic markers (CD45.1 for transferred T depleted B6 BM and CD25− T cells; CD45.2 for transferred nTreg cells and for host identification).
Mice were analyzed 10 – 14 days after BMT. Splenocytes and peripheral lymph node cells were either analyzed directly or after sorting using Thy1.2+ antibody and MACS beads. Colonic lamina propria cells were isolated as described previously (Durant et al., 2010). Briefly, large intestines were removed, cleared from mesentery, fat and Peyer’s patches, cut into pieces and washed in HBSS w/o Ca2+ /Mg2+. After incubation in HBSS with EDTA, epithelial layer cells were removed and the remaining tissue was digested with liberase and DNAse I (both Roche, IN) at 37°C. Lamina propria lymphocytes were recovered from the supernatant and purified over a 40%, 80% percol gradient.
Unless stated otherwise, GFP fluorescence was used to determine FoxP3 expression peripheral lymphocytes both pre and post transplantation. Intracellular staining on stimulated cells was used to determine cytokine expression in cells recovered during the second week post transplant. Cells were first stimulated with phorbol ester and ionomycin (Sigma, St Louis, MO) alone for two hours; then, for the next two hours, golgiplug® was added (BD Biosciences, San Diego, CA). Flow cytometry staining antibodies for CD4, CD25, CD62L, CD45.1, CD45.2, H2Kb, IFN-γ and IL-17 were purchased from BD Biosciences (San Diego, CA). Data were acquired using a CyAn flow cytometry machine.
Natural Treg cells, isolated by flow cytometry sorting for CD4+ FoxP3-GFP+ expression, were stimulated in cell culture plates coated with anti-CD3 and anti-CD28 (10ug/ml each) for six days in cell culture media alone (RPMI, supplemented with 10% FCS, glutamine, 2-mecaptoethanol, penicillin and streptomycin) or in the presence of IL-6 (100ng/ml) or IL-27 (20ng/ml), the cytokines were added on days 1 and 4 of cell culture. Induced Treg cells were cultured from CD4+ naïve (CD62L+CD44− FoxP3-GFP−) in cell culture plates coated with anti-CD3 and anti-CD28 in cell culture media with TGF-β (10ng/ml), IL-2 (100iu/ml), anti-IL-4 (10ug/ml) and anti-IFN-γ (10ug/ml) for three days. Cells were subsequently sorted for GFP expression. Cytokines were purchased from RnD systems (MN, USA), anti cytokine antibodies were purchased from BioXCell (NH, USA).
ChIP was performed as described previously (Yao et al., 2007). Flow cytometry sorted CD4+ FoxP3-GFP+ induced Treg cells were stimulated with IL-2 (100iu/ml) and IL-6 (20ng/ml) for 2 hours followed by cross-linking for 15 min with 1% formaldehyde. The cells were harvested and lysed by sonication. After pre-clearing with protein A agarose beads (Upstate), cell lysates were immunoprecipitated with anti-H3K4m3 (ab8580, Abcam), anti-STAT5A/STAT5B (PA-ST5A and PA-ST5B, R&D Systems) overnight at 4°C. After washing and elution, crosslinks were reversed at 65°C for 4 h. The eluted DNA was purified, samples analyzed by quantitative-PCR with customer designed primers and probes directed against target site III (5’-ACAACAGGGCCCAGATGTAGA-3’, 5’-GGAGGTTGTTTCTGGGACATAGA-3’, and 5’-6FAM-CCCGATAGGAAAACA-3’) and site IV (5’-CACCAAAGGCTGGAAGCCT-3’, 5’-CAGACGAGCCTCCACAGAGTT-3’, and 5’-6FAMCCGTGCCTTGTCAGG-3’) as defined previously using a 7500 real time PCR system (both Applied Biosystems). The Ct value for each sample was normalized to corresponding input value.
Representative samples of skin, liver, intestine and lung were obtained from transplant recipients and fixed in 10% formyl saline. Samples were embedded in paraffin, sectioned and stained with hematoxylin and eosin. All slides were coded and read by an external pathologist in a blinded fashion. A four-point scale of GvHD severity was used to score the samples (specific histology criteria detailed in the supplementary experimental procedure section).
Unless stated otherwise, P values from data presented as histograms were determined using a two-tailed unpaired t test. P values from survival curve data were determined using the log rank test. Unless stated otherwise, histogram columns represent the mean values for each experiment and error bars indicate the standard deviation. Single asterisk (*) denotes P<0.05, double asterisk (**) denotes P<0.01, and triple asterisk (***) denotes P<0.001.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.