Mechanisms of GVL Against a Murine Blast Crisis CML.

Donor T cells mediate a graft-versus-leukemia effect that is responsible for much of the efficacy of allogeneic hematopoietic stem cell transplantation (alloSCT) in treatment of hematologic malignancies. Chronic phase chronic myeloid leukemia (CP-CML) is the most GVL-sensitive neoplasm. Unfortunately, most other malignancies are relatively GVL-resistant. A striking example is blast crisis CML (BC-CML) which, although sharing its genetic etiology with CP-CML, is nearly refractory to alloimmune T cells. A detailed understanding of GVL-resistance has been hindered by the absence of GVL-sensitive and GVL-resistant murine leukemias that are similar to their human counterparts and are inducible on different mouse strains. In particular, generating gene-deficient leukemias is important for mechanistic experiments. To address these limitations, we have adopted murine models of CP-CML (mCP-CML) and BC-CML (mBC-CML) that share pathology and genetic etiology with their human counterparts. mCP-CML is generated by retroviral transduction of murine bone marrow (BM) with the bcr-abl fusion cDNA (p210), the defining genetic abnormality in human CP-CML. As is the case with human CP-CML, mCP-CML is extremely GVL-sensitive at least in part due to the redundant immune mechanisms sufficient for GVL (Matte et al, Blood 2004). mBC-CML is induced by the retroviral transduction of BM with both p210 and the fusion cDNA NUP98/HOXA9 (Dash, PNAS, 2002), a translocation found in human BC-CML and AML. Relative to mCP-CML, mBC-CML is GVL-resistant. In the MHC-matched C3H.SW→B6 (H-2b) strain pairing, 30–40% of recipients of 4–6 million donor CD4 or CD8 cells die from mBC-CML. This dose is nearly 10-fold higher than required for a similar survival from mCP-CML, even though recipients of mBC-CML and no donor T cells die nearly a week later than recipients of only mCP-CML. Having established that mBC-CML is GVL-resistant, we investigated mechanisms of T cell killing and the roles of donor and recipient antigen presenting cells (APCs). Direct T cell:mBC-CML cognate interactions were required as MHCI⁻ and MHCII⁻ mBC-CML cells (generated in β₂microglobulin (β₂M) or IA^b β chain knockout (KO) BM) were completely insensitive to CD8 and CD4-mediated GVL, respectively. In contrast, neither CD8 nor CD4-mediated GVL was impaired against mBC-CML generated from TNF-receptor1/2 double KO or Fas^lpr BM. These are the same basic mechanisms of cytotoxicity we observed in GVL against mCP-CML. CD8-mediated GVL against mBC-CML required functional recipient APCs as we observed no GVL when recipients were MHCI⁻ β₂M KOs. As was the case with GVL against mCP-CML (Matte, N.Med. 2004), donor APCs were not required as GVL was equivalent in recipients of wild type and β₂M KO C3H.SW donor BM. We observed no GVL in MHCII⁻ recipients demonstrating that CD4-mediated GVL also requires functional recipient APCs. In sum, the basic rules of immunogenicity for GVL against mCP-CML and mBC-CML are similar, suggesting that other pathways are responsible for GVL-resistance. One possibility is differential sensitivity to TRAIL-mediated killing and we are currently generating TRAILR-deficient mBC-CML. Another candidate is PD-L1, a B7 family member that can suppress T cell responses. PD-L1 is highly expressed on mBC-CML relative to mCP-CML. We have already generated PD-L1-deficient mBC-CML and GVL experiments with it are underway.