CD4⁺ Effector Memory and Naive T Cells Mediate Graft-Versus-Leukemia Via Direct Recognition of MHCII on Leukemic Cells.

One of the major challenges in allogeneic stem cell transplantation (alloSCT) is to separate graft-versus-host-disease (GVHD) from graft-versus-leukemia (GVL). We and others have previously demonstrated, in both major histocompatibility complex (MHC)-compatible/multiple minor histocompatibility antigen-mismatched and MHC-mismatched murine models of alloSCT, that spontaneous effector memory (EM) CD4⁺T cells depleted of regulatory CD25⁺ cells (CD4⁺CD44⁺CD62L^-CD25^-) do not cause GVHD. We have also shown that these EM CD4⁺ T cells can mediate GVL against a model of murine chronic phase of CML (mCP-CML) induced via retroviral transduction of BM cells with the bcr-abl fusion cDNA without causing GVHD (Zheng, et al ASH meeting 2004). In the present study we analyzed the effector mechanisms of these EM CD4⁺ cells in the B6bm12 → B6 MHCII disparate bone marrow transplantation (BMT) model. First, we demonstrated that the GVL activity of both EM and naïve CD4⁺ T cells required cognate interactions with CML targets as GVL was ineffective against mCP-CML induced in bone marrow from B6.I-Ab^−/− (MHCII⁻) mice. Recipients of MHCII⁻ mCP-CML died from mCP-CML between day 15-20 post BMT, regardless of whether they received EM or naïve CD4⁺ cells or no T cells at all. In light of data in the same model that parenchymal MHCII expression is not required for GVHD (Teshima et al, 2002), these data demonstrate distinct mechanisms for the cytotoxicity by CD4⁺ cells in GVL and GVHD—direct in the former and indirect in the latter. To further investigate the specific mechanisms of T cell killing, we tested the effectiveness of EM CD4⁺ cells in eradicating mCP-CML induced in bone marrow cells from Fas^−/− and TNFR1/R2^−/− mice. Both EM and naïve CD4⁺ cells mediated GVL against these gene deficient leukemias that was similar to that against wild type mCP-CML. In summary, these results suggest that EM and naive CD4⁺ cells mediate GVL via direct cognate engagement with targets. Their killing, however, does not depend on either FasL or TNF-α which suggests a dominant role for perforin, TRAIL, or both. Interestingly, although the mechanisms of recognition and killing of mCP-CML by either naïve or EM CD4⁺ T cells are so far indistinguishable, whereas only the naïve cells cause GVHD. Whereas a number of investigators have been able to separate mechanisms of killing in GVHD vs. GVL, this is to our knowledge the first clear demonstration of a difference in the mechanism of recognition between GVHD and GVL.