Leukemia relapse represents a failure of graft-versus-leukemia (GVL) and remains the major limitation of allogeneic stem cell/bone marrow transplantation (BMT). Graft-versus-host disease (GVHD) within the gastrointestinal (GI) tract is the principal determinant of transplant-related mortality and is initiated by a network of alloantigen presentation by professional and non-professional APC that prime donor T cells in the GI tract and related lymphoid structures. Since GVL and lethal GVHD are mediated by donor T cells at spatially distinct sites; bone marrow (BM) and the GI tract respectively, we sought tractable approaches to spatially separate alloreactive responses at these two locations. The administration of high dose steroids in the peri-transplant period is permissive of T cell replete HLA-haploidentical BMT and significant GVL effects (Ogawa H, et al. BBMT. 2006). We utilized murine haploidentical BMT models (B6D2F1 → B6C3F1, B6 → B6D2F1) with recipient background MLL/AF9 primary acute myeloid leukemia (AML), with or without dexamethasone (Dex) administration (5 mg/kg/day i.p., days -1 to +5). Dex-treatment improved transplant survival (from 25% to 68% at day 100, P=0.0012) with significant reductions in GVHD histopathology specifically in the colon (histopathology scores 8.7±1.0 vs 4.6±0.8, P< 0.05), despite excellent leukemia control. To understand this paradox, we analyzed the kinetics of donor T cell expansion after BMT. In the mesenteric lymph node (mLN), Dex treatment significantly suppressed the expansion of both CD4 and CD8 T cells (3.3±0.3 x 105 vs 1.4±0.3 x 105, P< 0.001 and 4.2±0.4 x 105 vs 2.1±0.4 x 105, P< 0.01 respectively) and the activation of CD4 T cells (CD25 MFI: 2021±146 vs 1056±102, P< 0.01). In contrast, donor effector/memory CD44+ CD8 T cells were expanded in the BM of Dex treated recipients (1.9±0.3 x 105 vs 3.1±0.4 x 105, P< 0.05) that demonstrated high per cell cytolytic activity against leukemia (specific lysis: 65±2.4 % vs 62±2.6 % in untreated vs Dex-treated, P> 0.05). Surprisingly, there was no difference in proliferation (cell tracking dye dilution: 63±5.5 % vs 57±5.5 % in untreated vs Dex-treated, P> 0.05) or apoptosis (caspase-3: 6.6±0.4 % vs 6.1±0.6 %, caspase-8: 20±1.6 % vs 17±3.3 % in untreated vs Dex-treated, respectively, P> 0.05) of CD4 T cells in the mLN between the two groups. We undertook experiments with luciferase expressing T cells and noted that Dex-treatment preferentially inhibited T cell accumulation in the GI tract, but not marrow after BMT. Thus, it appeared that Dex treatment preferentially re-distributed donor T cells from the GI tract to the bone marrow. We next determined if Dex exerted effects via direct signaling to the donor T cell. We thus transplanted glucocorticoid receptor (GR)-deficient or intact T cells (GRfl/fl lck-Cre mice). Dex-treatment reduced donor CD4 T cell expansion in the mLN independent of their expression of the GR (untreated vs Dex-treated: 2.8±0.6 x 105 vs 1.2±0.3 x 105, lckCREGRfl/fl and 2.4±0.3 x 105 vs 1.4±0.4 x 105, GRfl/fl littermates, P< 0.05 both groups). Thus steroid effects were mediated indirectly, putatively via effects on recipient alloantigen presentation. There was a marked reduction in recipient dendritic cells (DC) and macrophages expressing the Ea peptide within MHC class II in the GI tract of Dex-treated recipients (terminal Ileum YAe+ DC number 896±93 vs 356±40, P< 0.01, YAe+ macrophage number 1035±136 vs 355±97, P< 0.01). In conjunction with this, expression of the gut homing integrin a4b7 expression was reduced in CD4 T cells from Dex treated recipient mLN (25±1.6 % vs 17±1.7 %, P< 0.01), while the marrow homing integrin VLA-4 (a4b1) was increased (a4: 62±2.2 % vs 75±1.6 %, P< 0.001, b1: 52±2.5 % vs 61±1.6 %, P< 0.05) in donor CD8 T cells from Dex treated recipient BM. Finally, Dex treatment enhanced GVL against a second primary AML (BCR/ABL-NUP98/HOXA9) relative to untreated recipients and those receiving post-transplant cyclophosphamide (PT-Cy) (relapse rate: 0% vs 40% vs 100% at day 35 in Dex vs untreated vs PT-Cy, PT-Cy vs Dex-treated, P< 0.0001; untreated vs Dex-treated, P=0.029). These data suggest a potential therapeutic strategy to modulate antigen presentation in the GI tract and consequent integrin imprinting that minimizes GVHD lethality whilst enhancing GVL within BM. Disclosures No relevant conflicts of interest to declare.