Spontaneous Memory CD4⁺ T Cells Preserve Graft-Versus-Leukemia without Causing Graft-Versus-Host Disease.

We have recently shown in a major histocompatibility complex (MHC) compatible, multiple minor histocompatibility antigen mismatched CD4-dependent murine model of allogeneic stem cell transplantation (alloSCT) that donor spontaneous effector memory (EM) CD4 cells depleted of regulatory (Treg) CD25⁺ cells (CD4⁺CD44⁺CD62L⁻CD25⁻) do not cause graft-vs.-host disease (GVHD), but engraft and mount recall responses to a model antigen (Anderson et al., JCI, 2003). Thus EM cells might allow immune reconstitution with less GVHD. However, as alloSCT is most frequently used to treat malignant diseases, it is important to determine whether spontaneous EM cells can mediate graft-vs.-leukemia (GVL). Chen et al (Blood, 2004) found that memory T cells from donors immunized to host antigens can mediate GVL in an MHC disparate model; however Tregs were not depleted and whether spontaneous memory T cells can mediate GVL, which would be a likely clinical scenario, was not addressed. Here, we report that donor CD25⁻ EM CD4 cells do not cause GVHD in the MHCII-disparate B6^bm12→B6 model, but mediate GVL against a murine model of chronic phase CML (mCP-CML) induced by retroviral insertion into bone marrow (BM) cells of the bcr-abl (p210) fusion cDNA, the defining genetic abnormality in human CML (Matte et al., Blood, 2004). mCP-CML is manifest by a high WBC count and splenomegaly dominated by maturing myeloid cells. The retroviral construct also expresses the human nerve growth factor receptor (NGFR) which allows the tracking of transduced cells by FACS. We first addressed GVHD. Lethally irradiated B6 mice were reconstituted with T cell depleted B6^bm12 BM with no T cells or with 5x10⁵ CD4⁺CD25⁻CD62L⁺CD44⁻ naïve (N) or CD4⁺CD25⁻CD62L⁻CD44⁺ EM B6^bm12 T cells. In 2 independent experiments EM cells induced neither clinical nor histologic GVHD. In contrast, N CD4 cells induced GVHD manifest by weight loss (P<0.007; EM vs. N), hunched posture and ruffled fur. GVHD was confirmed histologically in the liver, intestine and skin (P<0.0002 for liver and bowel, P=0.0553 for skin; EM vs. N). To determine whether EM CD4 cells mediate GVL, lethally irradiated B6 mice were reconstituted with p210-infected B6 BM, donor B6^bm12 BM, with or without a source of donor T cells (2.5x10⁵ or 5x10⁵ N CD4 cells or 10⁶ EM CD4 cells). 5/5 mice that did not receive donor T cells died from mCP-CML between days 18-21 post BMT. N CD4 recipients (n=5 for both 2.5 and 5x10⁵groups) cleared blood NGFR⁺ cells by day 22, but suffered from GVHD as expected. Strikingly, EM CD4 cells also mediated GVL as measured by improved survival and a 10-fold reduction in NGFR⁺ WBCs as compared to mice that received no T cells. Only 1/6 recipients died of mCP-CML; 5/6 recipients were still alive with a low WBC count at day 28 (P<0.0001; BM vs. BM/EM), although some NGFR⁺ cells were still present. Importantly, none of the EM recipients developed GVHD. EM cells were 99% pure and it is unlikely that less that 10⁴ contaminating N cells were responsible for the GVL we observed. Taken together, our prior published work and this new data suggest that donor EM cells can contribute to both immune reconstitution and GVL and thus selective infusion of EM cells may be a successful strategy in clinical alloSCT.