STAT3 Signaling in CD4⁺ T Cells Is Critical for the Pathogenesis of Chronic Sclerodermatous Graft-Versus-Host Disease in a Murine Model

Radiation- or chemotherapy-based conditioning prior to allografting results in damage of host tissues and induction of local and systemic inflammatory milieu that naive donor T cells first encounter after allogenic bone marrow transplantation (alloBMT). Thus, it is not surprising that naive T cells are highly virulent in inducing graft-versus-host disease (GVHD) and that modulation of the posttransplant inflammatory environment can influence acute GVHD development (1, 2). It has been suggested that this same environment may be essential for the development of chronic GVHD as well (3–5). The unifying feature of animal models of chronic GVHD is that tissue injury is mediated primarily by noncytotoxic CD4⁺ T cells (6–8). However, studies of the various models have provided conflicting data on the roles of mature donor graft-derived and hematopoietic stem cell (HSC)-derived CD4⁺ T cells that emerge from the stressed host thymus in the pathogenesis of chronic GVHD (5, 9–11). Additionally, the roles of distinct Th subsets (Th1, Th2, and Th17) remain controversial.

The effector differentiation of naive CD4⁺ T cells represents the outcome of cytokine-mediated activation of specific STAT proteins and the induction of lineage-specific transcription factors (12). In vitro, naive T cell differentiation is readily achieved by the addition of cytokines and cytokine-neutralizing Abs. However, in vivo and especially posttransplant, the situation is far more complex because T cells must integrate diverse signals that redirect their differentiation toward a distinct effector or regulatory lineage. Given the known role of STAT proteins in integrating cytokine signaling and regulating effector T cell differentiation and homeostasis, we examined the effect of STAT3 ablation on chronic GVHD development. We focused specifically on STAT3 because this transcription factor has emerged as a key regulator of the proinflammatory immune response during induction of several experimental autoimmune diseases (13, 14). Additionally, the signaling of several inflammatory cytokines, including IL-6, -10, -21, and -23, converges on the STAT3 pathway. Experiments were conducted using a well-characterized B10.D2→BALB/c MHC-matched, minor histocompatibility Ag-mismatched model that recapitulates the pathophysiology of chronic GVHD, especially the sclerodermatous form of the disease (15, 16). We studied the role of STAT3 signaling using CD4⁺ T cells in which STAT3 was selectively ablated by crossing CD4-Cre mice with STAT3^flox/flox mice on a B10.D2 background (13, 14, 17). By altering only STAT3 signaling in donor CD4⁺ T cells while leaving all other components of allograft intact (HSC and donor CD8⁺ T cells), we were able to specifically examine the importance of this pathway in CD4⁺ T cell differentiation and function. Our results demonstrate that STAT3 signaling in donor CD4⁺ T cells influences the development of chronic sclerodermatous GVHD by enabling sustained alloreactivity and negatively affecting the reconstitution of regulatory T cells (Tregs).

BALB/c-CD45.1 (H-2^d, CD45.1⁺, CD90.2⁺), B10.D2-Thy1.2 (H-2^d, CD45.2⁺, CD90.2⁺), B10.D2-Thy1.1 (wild-type [WT]; H-2^d, CD45.2⁺, CD90.1⁺), and B10.D2-Thy1.1-CD4-Cre × STAT3^flox/flox [STAT3^KO; CD45.2⁺, CD90.1⁺ (13)] were propagated in the animal facility at the Johns Hopkins University Cancer Research Building I. Thymectomized BALB/c (H-2^d, CD45.2⁺, CD90.2⁺) mice were obtained from the National Cancer Institute (Frederick, MD). All animals were 8- to 12-wk-old at the time of experimentation. All protocols were approved by the Animal Care and Use Committee of The Johns Hopkins University.

A single lethal irradiation dose of 775 cGy was administered using a [¹³⁷Cs] irradiator. Animals were reconstituted with 10⁷ T cell depleted (TCD) B10.D2-Thy1.2 bone marrow (BM) cells alone or they received TCD BM supplemented with a graft-versus-host (GVH) inoculum consisting of 1.8 × 10⁶ WT CD90.1⁺CD4⁺ and 0.9 × 10⁶ WT CD90.1⁺CD8⁺ T cells (WT inoculum) or STAT3^KO 1.8 × 10⁶ CD90.1⁺CD4⁺ and 0.9 × 10⁶ WT CD90.1⁺CD8⁺ T cells (STAT3^KO inoculum). The T cell dose reflects T cells found in 1.2 × 10⁷ B10.D2 donor splenocytes, a dose that reproducibly induces GVHD. T cell depletion was performed as previously described (18). Purified populations of donor T cells were obtained using T cell isolation kits (Dynabeads, Invitrogen, Carlsbad, CA). When necessary, cell sorting was done using a FACSAria cell sorter (BD Biosciences, San Jose, CA). In designated experiments, T cells were labeled with CFSE prior to injection into irradiated recipients, as previously described (18). The purity and viability of T cell isolates in all experiments exceeded 95%. Cells were injected in the lateral tail vein.

For in vivo CD25 depletion, chimeras were injected i.p. with 0.5 mg anti-CD25 (PC61) or isotype control Ab (IgG₁; both BioXcell, West Lebanon, NH). Injections were given every third day, for a total of six injections, starting on day 1 after transplantation (19).

At designated time points, animals were sacrificed, and organs of interest (spleens, mesenteric lymph nodes [MLNs], livers, skin, and thymuses) were collected. Single-cell suspensions of spleen and MLNs were prepared by passing the organs through a cell strainer. Minced livers were digested for 30 min in complete medium supplemented with Liberase and DNase (both purchased from Roche Applied Science, Indianapolis, IN), and leukocytes were isolated by density gradient centrifugation on Accuprep medium (Accurate Chemical, Oslo, Norway). Minced depilated back skin was incubated for 90 min in complete medium supplemented with Liberase, DNase, and hyaluronidase (Worthington, Lakewood, NJ). For thymic analysis, animals were exsanguinated, thymuses were harvested, and single-cell suspensions were prepared by passing the organs through a cell strainer. Abs against mouse CD4, CD8a, CD11b, CD11c, CD25, CD44, CD45, CD45.1, CD45.2, CD62L, CD69, CD90.1, CD90.2, CD103, Gr-1, Annexin V, Foxp3, IL-6 and -17, and IFN-γ and the corresponding isotype controls were obtained from Biolegend (San Diego, CA), BD Biosciences, eBioscience (San Diego, CA), or Invitrogen. For detection of cytokine production, cells were briefly restimulated with PMA and ionomycin, in the presence of monensin (GolgiStop, BD Biosciences), prior to staining for flow cytometric analysis. Annexin V staining was performed according to the manufacturer’s instructions (BD Biosciences) Analysis was performed using an LSR II or FACSCalibur flow cytometer (BD Biosciences). Data were analyzed using FACSDiva (BD Biosciences) and FlowJo (Tree Star, Ashland, OR) software.

To quantify clinical GVHD, we used a scoring system that sums changes in weight loss, posture, activity, fur texture, and skin integrity (20, 21). Assessment of cutaneous GVHD was done using a separate scoring system (22). GVHD was also evaluated by histopathological analysis of ear skin, liver, colon, and thymus. All samples were prepared for routine histological analysis, stained with H&E, and slides were read by experienced pathologists (skin by A.C.G.; liver, colon, and thymus by A.P-M), in a blinded fashion, using established scoring systems (22, 23). Microphotographs were acquired using an RT Spot camera (Diagnostic Instruments, Sterling Heights, MI) mounted on an Olympus BX51 microscope (Olympus, Melville, NY) with Spot advanced software, at a magnification of ×200.

Total RNA from snap-frozen tissues and sorted T cells was extracted using an RNeasy mini kit (Qiagen, Valencia, CA) and reverse transcribed using a qScript cDNA Synthesis Kit (Quanta BioSciences, Gaithersburg, MD). cDNA was subjected to PCR amplification using primers for 18s rRNA, IFN-γ, TNF-α, CCL2, CCL5, CCL20, CCR5, IL-6 and -17, T-bet, GATA3, and Foxp3 (24–28) and SYBR Premix Ex Taq (Takara Bio, Madison, WI) in a MyiQ light cycler (Bio-Rad, Hercules, CA).

Values are presented as the mean ± SEM or mean ± SD, as appropriate. Statistical differences were calculated using one-way ANOVA, the Student t test, or the log-rank test, as appropriate. A value of p < 0.05 was considered statistically significant.

Loss of STAT3 in graft-derived donor CD4⁺ T cells prevents chronic GVHD development in a B10.D2→BALB/c model.

To assess the role of STAT3 signaling in donor T cells (GVH inocula) during chronic sclerodermatous GVHD development, we used CD4⁺ T cells from donors in which conditional STAT3 deletion (Cre-Lox system) occurs solely in T cells without altering their number or phenotype (13, 14, 17). Cohorts of chimeras were constructed after conditioning of BALB/c-CD45.1 recipients and transplantation of TCD BM alone or TCD BM supplemented with GVH inocula. Following alloBMT, chimeras were monitored for clinical signs of GVHD using two previously established scoring systems that independently assess acute (20) and chronic GVHD (22). Although recipients of WT inocula developed clinical GVHD, no signs of acute or chronic GVHD were observed in chimeras injected with STAT3^KO inocula (Fig. 1A–C; p < 0.001, WT versus STAT3^KO). When TCD splenocytes were added to the GVH inocula, the same findings were observed as described above for studies with WT versus STAT3^KO T cells without additional TCD splenocytes (data not shown). This finding excluded the possibility that the lack of some other cellular component was responsible for the absence of GVHD in chimeras receiving STAT3^KO inocula. Routine histological sections of the ears of mice that received WT inocula revealed increased skin thickness and infiltration with mononuclear cells. In contrast, no histological changes were noted in the chimeras that received STAT3^KOCD4⁺ T cells (Fig. 1D).

During the course of chronic sclerodermatous GVHD, inflammatory leukocytes invade the skin and produce a range of inflammatory mediators (16). Flow cytometric analysis of single-skin cell preparations revealed that animals receiving STAT3^KOCD4⁺ T cells had reduced inflammatory Gr-1⁺CD11b⁺ monocyte infiltration and decreased IL-6 production in comparison with chimeras receiving WT CD4⁺ T cells (Fig. 1E, 1F). These observations were corroborated by gene expression analysis of inflammatory markers known to be highly expressed in the skin of chimeras with sclerodermatous GVHD (Fig. 1G) (16, 29). Taken together, these findings suggest that STAT3 ablation in graft-derived donor CD4⁺ T cells disables the inflammatory response that is crucial to the development of chronic sclerodermatous GVHD in this CD4⁺ T cell-dependent model.

We next focused on exploring how abrogation of STAT3 signaling in CD4⁺ T cells influences their in vivo fate and function after alloBMT. To assess cell fate, we monitored CD4⁺ and CD8⁺ T cell expansion and accumulation in secondary lymphoid (spleen) and epithelial (liver and skin) tissues. In conducting these analyses, we exploited the differential expression of CD90 allele on donor graft-derived T cells (CD90.1) and on T cells developing in vivo from transplanted donor HSCs (CD90.2). We found that STAT3 ablation resulted in a 20-fold lower accumulation of STAT3^KOCD4⁺ T cells in the spleen and liver on day 5 after alloBMT (Fig. 2A; p < 0.01). On day 14, the difference was still marked, but of smaller magnitude (>5-fold; Fig. 2A; p < 0.05). A similar pattern of CD90.1⁺CD4⁺ T cell accumulation was observed in the skin (Fig. 2B). The lower number of STAT3^KOCD4⁺ T cells observed at day 14 was maintained in the spleens and livers of chimeras over time. The expansion of CD90.1⁺CD8⁺ T cells only differed in the spleen on day 14 (Fig. 2A), a finding consistent with the dominant role of CD4⁺ T cells in this model. Next, we compared the GVH reactivity of STAT3^KO and WT GVH inocula. Despite lower expansion and accumulation of graft-derived T cells in spleen and epithelial tissues, recipients of STAT3^KO GVH inocula rapidly converted to full donor chimerism, with kinetics that were indistinguishable from those observed in WT GVH inoculum recipients (Fig. 2C and data not shown; p = NS for WT versus STAT3^KO for all time points assessed).

There are three potential explanations for the observed effect of STAT3 ablation on the in vivo fate of donor T cells: STAT3^KO T cell proliferation is reduced, alloantigen recognition is altered, or their in vivo survival is impaired. To examine the effect of STAT3 signaling on in vivo proliferation, graft-derived WT and STAT3^KOCD4⁺ T cells were labeled with CFSE and assayed for dye dilution. At day 5 after alloBMT, the overall percentage of proliferated splenic CFSE^low STAT3^KOCD4⁺ T cells was markedly lower than that of WT CD4⁺ T cells (p < 0.001; Fig. 3A). By day 14, the CFSE dilution of STAT3^KOCD4⁺ T cells was complete and indistinguishable from that of WT CD4⁺ T cells (data not shown).

To determine whether STAT3 signaling influences in vivo alloantigen-induced T cell activation, we examined changes in the expression of characteristic activation markers. We found no difference in the expression of CD25, CD69, CD11a, CD18, and CD29 on splenic and hepatic CD4⁺ T cells between the two sets of chimeras on day 14 after alloBMT (Fig. 3B and data not shown). At this time point, WT and STAT3^KOCD4⁺ T cells in the spleen and liver were predominantly of an effector memory phenotype (CD44^highCD62L^low/−; Fig. 3C; p = NS).

We next characterized the effect of STAT3 ablation on CD4⁺ T cell survival. Given that IL-6 expression increases in the posttransplant inflammatory milieu and that IL-6–induced STAT3 activation is indispensable for the prevention of apoptosis (29, 30), we expected that the in vivo survival of STAT3^KO T cells would be impaired. Surprisingly, we found a comparable fraction of Annexin V⁺STAT3^KO and WT CD4⁺ T cells in the spleen and liver of chimeras on day 5 posttransplant (Fig. 3D; p = NS). A control in vitro experiment confirmed that IL-6 could prevent apoptosis of optimally stimulated WT, but not STAT3^KO, T cells (data not shown). One possible explanation for our in vivo finding is that the posttransplant inflammatory milieu may provide other survival factors for STAT3^KOCD4⁺ T cells. TGF-β mediates antiapoptotic effects via multiple mechanisms, enhances survival of TCR-stimulated CD4⁺ T cells, and is abundant posttransplant in this model (16, 31, 32). We found that TGF-β, alone or in combination with IL-6, prevents apoptosis and rescues stimulated STAT3^KOCD4⁺ T cells from death in vitro (Supplemental Fig. 1). Thus, multiple signaling pathways influence the survival and apoptosis of STAT3^KOCD4⁺ T cells after transplantation.

We sought to further determine the impact of STAT3 signaling on donor T cell effector differentiation and cytokine secretion after alloBMT. The percentages of IFN-γ–secreting STAT3^KO and WT CD4⁺ T cells were similar in the skin of chimeras at day 14 after alloBMT (35% versus 33%; Fig. 4A), just prior to the development of cutaneous GVHD. We also found similar percentages of IFN-γ–secreting STAT3^KO and WT CD4⁺ T cells in the spleens and livers of chimeras at all time points analyzed (Supplemental Fig. 2A). Consistent with the reduction in the number of STAT3^KO T cells depicted in Fig. 2B, the absolute number of IFN-γ–secreting CD4⁺ T cells was reduced on day 14 in the spleen and liver (Fig. 4B; p < 0.01). This reduction in the absolute number of Th1 cells in the spleen and liver was also noted at later time points (Fig. 4B; p < 0.01).

We were unable to detect any IL-17–secreting cells in the skin of chimeras that received WT or STAT3^KO T cells at any time point examined, a result consistent with unpublished microarray data in this model (A.C. Gilliam, unpublished data). We were also unable to detect any significant IL-17 production by graft-derived WT CD4⁺ T cells in the spleens of chimeras at all time points analyzed (data not shown). However, serial monitoring revealed that the percentage and absolute number of Th17 cells in the livers of chimeras receiving WT T cells, but not STAT3^KO T cells, increased over time after alloBMT (Fig. 4C, 4D, Supplemental Fig. 2B). In addition, a substantial percentage of graft-derived WT CD4⁺ T cells coexpressed IL-17 and IFN-γ. These data indicate that donor graft-derived Th17 cells, as well as IL17⁺IFN-γ⁺CD4⁺ T cells, are generated in chimeras that develop chronic GVHD but that their emergence is delayed and organ dependent.

In addition to the Th1 and Th17 subsets, CD4⁺ T cells can differentiate into a distinct subset characterized by the expression of the forkhead box transcription factor FoxP3, considered to represent Tregs (33). Several groups reported marked conversion of STAT3^KO T cells into CD4⁺Foxp3⁺ Tregs upon in vitro stimulation in the presence of IL-6 and TGF-β (17, 26, 34). TGF-β and IL-6 are present in abundance in allogenic chimeras (29, 31). We hypothesized that this posttransplant milieu skews the fate of naive graft-derived STAT3^KO T cells toward a regulatory phenotype. To test this hypothesis, we sorted splenic- and liver-infiltrating graft-derived CD4⁺ T cells and analyzed the mRNA expression of lineage-specific transcription factors, representative cytokines, chemokines, and their receptors. We found that increased Foxp3 mRNA expression is a predominant characteristic of STAT3^KO versus WT CD4⁺ T cells, especially in the liver (Fig. 5A). Determining whether increased expression of Foxp3 and GATA3 mRNA detected in STAT3^KO CD4⁺ T cells retrieved from spleen indicates their differentiation toward Th2 and Foxp3⁺ cells or whether these cells represent previously described Tregs known to express both canonical transcription factors (35) will require further studies. Consistent with the reduced expression of IFN-γ protein as assayed by intracellular cytokine staining, STAT3^KOCD4⁺ T cells also exhibited decreased expression of IFN-γ (Fig. 5A). The expression of mRNA encoding proinflammatory chemokines CCL2, CCL5, and CCL20 and the CCR5 receptor was also reduced in STAT3-deficient CD4⁺ T cells compared with that in WT CD4⁺ T cells.

Next, we compared the percentage of Foxp3⁺ cells among STAT3^KO and WT CD4⁺ T cells retrieved from the spleens and livers of chimeras at multiple time points after alloBMT. The proportion of CD4⁺Foxp3⁺ T cells among all GVH inoculum-derived CD4⁺ T cells, although similar in both sets of chimeras at day 14, was substantially increased starting on day 28 in the group that received STAT3^KOCD4⁺ T cells (Fig. 5B; p < 0.05 for all time points after day 14). This prompted us to assess whether the increased percentage of STAT3^KOCD4⁺Foxp3⁺ T cells represents expansion of naturally occurring, thymus-derived CD4⁺CD25⁺ Tregs or adaptive CD4⁺CD25⁺ Tregs that are induced from CD25⁻ precursors in peripheral organs (induced Tregs [iTregs]). To address this, we constructed GVH inocula using CD4⁺CD25⁻ T cells that were sorted from WT or STAT3^KO CD4⁺ T cells to a purity exceeding 98%. The CD4⁺CD25⁻ fraction was then combined with donor CD8⁺ T cells and TCD BM. Twenty-eight days after alloBMT, spleens and livers were analyzed for the presence of graft-derived CD4⁺Foxp3⁺ Tregs. Recipients of STAT3^KO versus WT CD4⁺CD25⁻ T cells had a significantly higher frequency of CD90.1⁺Foxp3⁺ cells (9.5% ± 2.3% for STAT3^KO versus 2.7% ± 0.5% for WT; Fig. 5C; p < 0.05). A similar pattern was observed in the liver (1.1% ± 0.2% for STAT3 versus 0.4% ± 0.1% for WT; Fig. 5C; p < 0.05). The observed percentages of CD90.1⁺CD4⁺Foxp3⁺ T cells in recipients of STAT3^KO CD4⁺CD25⁻ inocula in spleen and liver were lower than those seen with intact STAT3^KO inocula on day 28 after alloBMT (Fig. 5B, 5C). These results indicate that the in vivo fate of STAT3^KO CD4⁺ T cells in inflammatory posttransplant milieu recapitulates previous in vitro observations (26, 36) and that the observed skewing of graft-derived T cells toward a CD4⁺CD25⁺Foxp3⁺ Treg phenotype is a result of predominantly naturally occurring, thymus-derived CD4⁺CD25⁺ Treg expansion rather than iTreg conversion.

We examined the role of STAT3 signaling in graft-derived CD4⁺ T cells on the overall quantitative reconstitution of the peripheral CD4⁺Foxp3⁺ Treg pool posttransplant. Cohorts of chimeras that received WT or STAT3^KO GVH inocula were serially analyzed, and the absolute numbers of CD4⁺Foxp3⁺ T cells were quantified. As shown in Fig. 6A i, the absolute numbers of CD4⁺Foxp3⁺ T cells were higher in animals that received STAT3^KO inocula than in recipients of WT GVH inocula from day 28 onward; this was especially evident on day 60 (Fig. 6Ai).

The peripheral Foxp3⁺ Treg pool can be restored posttransplant by residual host Tregs, homeostatically expanded graft Tregs, and new Tregs generated from donor HSC-derived progenitors by thymopoiesis. To evaluate these possibilities, we monitored the kinetics of Treg reconstitution and the relative contribution of each pathway in cohorts of chimeras that received WT or STAT3^KO GVH inocula. TCD BM recipients served as a control. The radio-resistant host CD4⁺Foxp3⁺ T cells were not detectable beyond day 7 posttransplant in either type of chimera receiving mature donor T cells with the graft (data not shown). When analyzed separately, the absolute number of graft-derived WT CD4⁺Foxp3⁺ T cells rapidly increased in comparison with STAT3^KOCD4⁺Foxp3⁺ Tregs (Fig. 6Aii; p < 0.05 for days 14 and 28). This early expansion of WT CD4⁺Foxp3⁺ Tregs is consistent with the overall proliferative advantage of T cells with intact STAT3 signaling (Figs. 2, 3). From day 28 onward, the absolute numbers of graft-derived CD4⁺Foxp3⁺ T cells in both groups declined; by day 60, these numbers were significantly reduced in both sets of chimeras. Interestingly, in chimeras that received STAT3^KO GVH inocula, thymus-derived CD90.2⁺CD4⁺Foxp3⁺ T cells rapidly increased after day 28 and represented the majority of Tregs in the peripheral Treg pool by day 60. In contrast, Treg reconstitution by thymopoiesis in chimeras that received WT GVH inocula was minimal (Fig. 6Aiii; p < 0.05 from day 28 onward).

We next investigated whether the apparent failure in central pathway reconstitution of HSC-derived CD4⁺Foxp3⁺ Tregs in recipients of WT GVH inocula was due to thymic damage. We observed that the thymuses of chimeras receiving WT CD4⁺ T cells were severely atrophic, an observation confirmed by histopathologic analysis and decreased overall thymic cellularity (Fig. 6Bi, 6Bii; p < 0.01 WT versus STAT3^KO). Flow cytometric analysis also revealed a decrease in the percentage of double-positive CD4⁺/CD8⁺ thymocytes in WT CD4⁺ T cell recipients (Fig. 6Biii; 38.6% versus 80.2%; p < 0.001 WT versus STAT3^KO). In contrast, the appearance and cellularity of thymuses and the phenotypic characteristics of T cell populations in STAT3^KOCD4⁺ T cell recipients were indistinguishable from those of chimeras that received only TCD BM. Moreover, the corresponding total numbers of naive T cells and CD90.2⁺ Tregs were significantly higher in STAT3^KO recipients than in WT CD4⁺ T cell recipients (data not shown). These results were not due to an inability of STAT3^KO T cells to infiltrate the host thymus, because comparable numbers of WT and STAT3^KO CD90.1⁺CD4⁺ T cells were found in thymus of both groups of chimeras (Fig. 6Biv; p = NS).

Taken together, our results suggest that the ablation of STAT3 signaling in graft-derived CD4⁺ T cells results in robust thymic-dependent Treg production. Furthermore, these results provide evidence that, consistent with other models (11, 37), the development of chronic GVHD in this model is also associated with dysregulated Treg reconstitution. However, here, it is primarily due to failed donor HSC-derived, thymus-dependent immunoreconstitution.

We next investigated the contribution of graft-derived Tregs and thymic output to chronic GVHD prevention in chimeras receiving STAT3^KO GVH inocula. To study the role of graft-derived Tregs, we depleted CD25⁺ cells in vivo using the anti-CD25 mAb PC61 (38). This strategy was used previously in multiple models to assess Treg contributions (39, 40). Anti-CD25 mAb treatment resulted in profound depletion of CD4⁺CD25⁺Foxp3⁺ T cells in all organs examined (data not shown). Consistent with previous studies, we observed that PC61 administration did not increase the severity of GVHD in animals receiving WT CD4⁺ T cells (19). In contrast, animals receiving STAT3^KO inocula followed by PC61 did not differ from counterparts that received control IgG₁ isotype until day 30, when they started to develop systemic GVHD. This condition was characterized by diarrhea, weight loss, hunched posture, and ruffled fur but interestingly not by cutaneous GVHD (Fig. 7A). Meanwhile, control chimeras reconstituted with only donor TCD BM that received PC61 remained healthy throughout the course of the experiment, confirming the role of adoptively transferred T cells in the development of GVHD in this setting. Histopathological examination of tissues obtained from WT inoculum recipients and STAT3^KO inoculum recipients treated with PC61 showed signs of extensive intestinal and hepatic GVHD. Epidermal hyperplasia and dermal and subdermal inflammation with destruction of the fatty layer were seen only in mice receiving WT CD4⁺ T cells (Fig. 7B). Dysregulated inflammatory response caused by CD25 depletion was also reflected in the serum levels of proinflammatory cytokines (Supplemental Fig. 3). Thus, a lack of Tregs early after transplantation does not directly exacerbate clinical GVHD, but their absence is critical for the induction of late de novo GVHD.

To examine the role of thymic output in preventing chronic GVHD, we conducted experiments using thymectomized BALB/c mice. Thymectomized BALB/c recipients of WT GVH inocula developed GVHD with similar penetrance, although their disease was more lethal in comparison with that of irradiated euthymic mice (data not shown). Interestingly, irradiated thymectomized BALB/c recipients of STAT3^KO GVH inocula developed GVHD, including characteristic skin manifestations, although with delayed kinetics (Fig. 7C). These findings suggest that STAT3^KOCD4⁺ T cells are capable of inducing tissue injury and that their inability to trigger GVHD is dependent on an intact thymus.

We hypothesized that the delayed alloimmune injury in chimeras that received STAT3^KO GVH inocula and were treated with PC61 was the result of an exaggerated graft-derived effector T cell response in the absence of Tregs, resulting in the de novo generation of pathogenic HSC-derived T cells. Consistent with our prediction, we found increased IFN-γ production in donor splenic graft-derived CD90.1⁺ and HSC-derived CD90.2⁺CD4⁺ T cells in chimeras that received STAT3^KOCD4⁺ T cells and were treated with PC61 (Fig. 8A). Because GVHD was mostly focused on the gastrointestinal system, we also examined the effector phenotype of CD4⁺ T cells in the MLNs. Macroscopically, STAT3^KO GVH inoculum recipients that received PC61 showed enlarged MLNs, a finding that was reflected by increased total cellularity. STAT3^KOCD4⁺ T cells retrieved from PC61-treated animals secreted significantly higher amounts of IFN-γ in comparison with their isotype-treated counterparts (Fig. 8B; p < 0.05). The observed increase in the percentage of IFN-γ–secreting STAT3^KOCD4⁺ T cells in MLNs was also reflected in their significantly higher absolute numbers, although PC61 treatment did not change the overall number of CD90.1⁺CD4⁺ T cells (Supplemental Fig. 4). Examination of HSC-derived CD90.2⁺ cells revealed a significantly increased percentage of CD4⁺ T cells secreting IFN-γ and IL-17, including those secreting both cytokines (Fig. 8B). The observed increase in double-positive (IFN-γ⁺IL-17⁺) CD4⁺ T cells is reminiscent of the population described in animals with autoimmune disease and those with autoimmune manifestations of chronic GVHD (11). These results strongly suggest that STAT3^KO Tregs play a critical role in preventing the dysregulation of alloreactive graft-derived T cells, which can offset the in vivo generation of Th1 and Th17 HSC-derived T cells.

In this study, we used a CD4⁺ T cell-dependent, MHC-matched model of alloBMT to examine the relevance of the STAT3 pathway in CD4⁺ T cells to chronic sclerodermatous GVHD development. Our studies revealed that STAT3 signaling in donor CD4⁺ T cells influences several key events in the pathogenesis of chronic GVHD. We found that abrogation of STAT3 signaling limits the in vivo proliferation and expansion of alloreactive CD4⁺ T cells. Decreased accumulation of these and other inflammatory cells in GVHD target tissues was also noted. Although STAT3 did not impact the capacity of naive CD4⁺ T cells to differentiate into Th1 effectors, it did promote CD4⁺Foxp3⁺ T cell reconstitution via thymus-dependent and -independent pathways, whereas WT counterparts experienced a profound loss of Tregs.

Our study provides new insight into the role of STAT3 signaling in naive T cell differentiation, which extends beyond its effect on Th17 generation. First, our results reveal that in an alloBMT setting, STAT3 ablation in alloreactive CD4⁺ T cells limits their in vivo proliferation and expansion in the secondary lymphoid tissues. A high T cell proliferative rate is known to correlate with acute GVHD, and the control of alloreactive T cell proliferation is the mechanism behind the function of pharmacological agents used for GVHD prophylaxis after alloBMT. Although STAT3^KOCD4⁺ T cells appeared to infiltrate the chimeras’ tissues in proportion to their level of expansion, their reduced CCR5, CCL2, CCL5, and CCL20 expression suggests that the ability of these cells to accumulate in GVHD target organs and orchestrate an inflammatory response is diminished. This is consistent with the known importance of IL-6–driven STAT3 signaling in the chemokine regulation of T cell recruitment to inflamed tissues (41). Despite these findings, the use of STAT3^KO GVH inocula still enabled conversion to complete donor chimerism, a reliable measure of lymphohematopoietic GVH reactivity (42) and a desirable outcome after alloBMT.

Second, our data show that Th1 cells are a dominant effector phenotype during the induction of chronic sclerodermatous GVHD. IL-17–secreting graft-derived T cells did accumulate in the liver and gut of chimeras that received WT GVH inocula, but their emergence was delayed. Donor HSC-derived Th17 cells that emerged in chimeras that had received STAT3^KO GVH inocula and anti-CD25 mAb also accumulated in the same organs. Interestingly, Zhang et al. (5) reported that in another MHC-matched model, donor HSC-derived, in vivo-generated pathogenic CD4⁺ T cells adoptively transferred into irradiated donor mice can selectively induce gut and liver, but not cutaneous, GVHD. The exact role of Th17 cells originating from donor mature T cells and HSC-derived, in vivo-generated Th1- and Th17-cell subsets in the pathogenesis of liver and gut GVHD needs to be examined further.

Finally, our most relevant observation was that STAT3 signaling in donor mature T cells provides a direct link between alloreactive T cell-mediated tissue damage and favorable Treg reconstitution posttransplant. There are several mechanisms behind this observation. Clearly, STAT3 abrogation limits the generation of alloreactive effector cells and damage to the host thymus, thus allowing early reconstitution of HSC-derived Tregs via the central pathway. The best confirmation for this is seen in the development of GVHD in thymectomized recipients of STAT3^KO GVH inocula. In addition, the role of STAT3 in thymus-independent Treg reconstitution is supported by the emergence of pathogenic donor HSC-derived T cells in chimeras reconstituted with STAT3^KO inocula and treated with anti-CD25 mAb. The absence of STAT3 signaling in donor graft-derived CD4⁺CD25⁻ T cells was also associated with increased conversion of CD4⁺CD25⁻ precursors to CD4⁺CD25⁺Foxp3⁺ Tregs in vivo. However, the overall quantitative contribution of iTregs to the regenerated peripheral Treg pool seems to be minor. Ultimately, a lack of GVHD is correlated with a renewal of thymic activity.

The early in vivo fate of mature graft-derived T cells in the current model differs from that reported for the MHC-mismatched model used to study chronic GVHD pathogenesis (11). Although the mechanisms behind these disparities are unknown, the differential kinetics of effector and Treg reconstitution noted in our study are especially relevant to the minor histocompatibility Ag-driven response that typically drives GVHD generation in the clinic. In addition, our observations provide insight into several published contradictory findings related to chronic GVHD pathophysiology. For example, thymus-dependent T cells play a role in mediating chronic GVHD in some, but not all, models (5, 9, 10), and in its sclerodermatous form, GVHD can occur after the administration of mature donor-derived, postthymic T cells in the form of donor lymphocyte infusions (43). Based on our findings, mature donor T cells administered with a graft are dominant and sufficient inducers of sclerodermatous GVHD and can induce GVHD in the absence of a thymus. We also found that pathogenic donor HSC-derived T cells can appear late posttransplant and that their emergence is not necessarily preceded by clinical signs of acute GVHD if donor Tregs are depleted early after alloBMT.

Several groups showed that in vitro stimulation of naive STAT3^KOCD4⁺ T cells in the presence of TGF-β and IL-6 results in more Foxp3-expressing CD4⁺ cells (17, 26, 34). Our results showed that the in vivo fate of CD4⁺ T cells with targeted deletion of STAT3 recapitulates previous in vitro observations. In addition, we provide unequivocal evidence for the role of the STAT3 pathway in GVHD, whose relevance was previously examined only by ex vivo treatment of donor T cells using small molecule inhibitors (44). However, our findings are contradictory to a report by Pallandre et al. (45), who showed that in vivo STAT3 neutralization results in exacerbation of acute GVHD. The most likely explanation for this contradiction is that in the referenced study, CD4⁺ T cells were retrieved from donor mice in which STAT3 was ablated using small interfering RNA in HSCs, whereas in our studies, STAT3 was ablated only in T cells. STAT3 abrogation in HSCs activates innate immunity, resulting in much stronger T cell responses, and causes autoimmunity (46, 47). This is not the case when STAT3 is ablated only in CD4⁺ T cells, because these animals possess a normal phenotype and numbers of Tregs at steady state (17). In that regard, it is important to emphasize that all observations on the effects of abrogating STAT3 signaling need to be considered in the context of the effects on other cell populations and the host cytokine milieu.

In summary, STAT3 signaling in donor graft-derived CD4⁺ T cells promotes their expansion and trafficking to epithelial organs and, in the setting of alloBMT, hinders their differentiation into Tregs. The resulting increase in Treg levels directly overrides the Th1 effector-dominated alloimmune response and tissue damage, preventing the loss of self-tolerance and the emergence of donor HSC-derived pathogenic Th1 and Th17 cells. Therefore, STAT3 signaling in donor graft CD4⁺ T cells plays an important role in chronic GVHD pathogenesis; abrogation of this signaling may have a critical effect on chronic GVHD modulation after alloBMT.

Disclosures The authors have no financial conflicts of interest.