Induction of a Novel Population of CD8⁺ Foxp3⁺ Regulatory T Cells During Graft Versus Host Disease

Abstract 821

Regulatory T cells defined as CD4⁺ and expressing the transcription factor Foxp3 have been shown to play a pivotal role in mitigating the severity of graft versus host disease (GVHD). In the course of studies designed to define the functional role of various CD4⁺ Treg populations in GVHD biology, we identified a novel population of CD8⁺ T cells that expressed Foxp3 and were induced early during this disease. While this population has been reported in patients with autoimmune disorders, the role of CD8⁺ Foxp3⁺ T cells in GVHD is unknown. To delineate the significance of this observation, we performed studies in which lethally irradiated Balb/c [H-2^d] mice were transplanted with bone marrow and spleen cells from C57BL/6J [H-2^b] mice that carried an EGFP reporter gene linked to Foxp3 (Foxp3^EGFP). Tissues (spleen, lung, liver and colon) were harvested 5, 7, 10, 14 and 21 days post transplantation to define the temporal kinetics and absolute numbers of CD8⁺ Tregs during acute GVHD. We observed that CD8⁺ Foxp3⁺ T cells were detectable as early as five days post transplantation and persisted for up to three weeks in all GVHD target tissues. This cell population was present in similar percentages and absolute numbers to CD4⁺ Tregs in these tissue sites which is noteworthy given that the CD4⁺ Treg pool is comprised of two populations (natural Tregs and induced Tregs) whereas the CD8 pool is made up almost exclusively of Tregs that are induced, since only a very small percentage of CD8⁺ T cells from normal mice (<1.0%) constitutively express Foxp3. To determine whether the induction of CD8⁺ Tregs was a function of MHC disparity, we performed similar transplant studies using murine models with varying degrees of MHC incompatibility. Notably, the relative and absolute number of CD8⁺ Tregs were much lower in an MHC-matched, minor antigen mismatched model of GVHD [B6→Balb.B], and were absent in a model where only three amino acids distinguish donor and recipient [B6→bm1], indicating a correlation between CD8⁺ iTreg generation and MHC disparity between donor and host. To confirm that in vivo-induced CD8⁺ Tregs were suppressive, CD8⁺ Foxp3⁺ and CD4⁺ Foxp3⁺ T cells were sorted from the spleen and liver of B6→Balb/c GVHD mice six days post transplantation and examined in standard MLC suppression assays. These studies revealed that in vivo-derived CD8⁺ and CD4⁺ Tregs equally suppressed alloreactive T cell responses. Phenotypic analysis of in vivo-differentiated CD8 iTregs revealed that these cells expressed many of the same cell surface molecules as CD4⁺ Tregs (e.g. GITR, CD25, CD103, CTLA-4). To determine if CD8⁺ Foxp3⁺ T cells could be induced in vitro and used as adoptive therapy for GVHD prevention, purified CD8⁺ Foxp3^EGFP– T cells were cultured with anti-CD3/CD28 antibodies, TGF-β and IL-2 for 3 days. Under these conditions, ∼30% of cells are induced to become Foxp3⁺. Addition of in vitro-differentiated CD8⁺ iTregs to a standard MLC resulted in potent suppression which was equivalent to that observed with in vitro-differentiated CD4⁺ Tregs. To determine whether these cells were suppressive in vivo, in vitro-differentiated CD8⁺ iTregs were adoptively transferred at a 1:1 Treg: effector cell ratio into lethally irradiated Balb/c mice that also received B6.PL BM and spleen cells to induce GVHD. In vitro-derived CD8⁺ iTregs failed to protect mice from GVHD in comparison to animals transplanted without CD8⁺ iTregs. This was attributable to reduced survival and the loss of Foxp3 expression in vivo. Furthermore, approximately 30–50% of these cells reverted to a proinflammatory phenotype characterized by IFN-γ secretion, similar to what has been described for in vitro-differentiated CD4⁺ iTregs (Beres et al, Clin Can Res, 2011). Finally, microarray studies were performed to compare the gene signatures of in vitro versus in vivo-induced CD8⁺ Tregs. Ontological analysis revealed that there was a 3–16 fold increase in the transcription of cytokine (e.g. IL-10) and cytotoxic (granzyme A, perforin, granzyme B) pathway genes in in vivo versus in vitro-induced CD8⁺ Tregs, suggesting that the former Treg population may employ similar mechanisms of suppression as has been reported for CD4⁺ Tregs. In summary, these studies have identified a novel population of CD8⁺ Foxp3⁺ cells that are induced early during GVHD, are able to suppress alloreactive T cell responses, and constitute another regulatory T cell population that is operative in GVHD biology.