VISTA As a Mediator of Immune Suppression in AML Initiated By DNMT3A and NPM1 mutations

In an immunocompetent host, cancer cells must overcome the immune response to develop into a clinically recognizable tumor. Although "tumor immune escape" mechanisms have been extensively studied in solid tumors (and successful immunotherapies have been developed), the mechanisms by which Acute Myeloid Leukemia (AML) cells escape the immune system are not well understood, and immunotherapies have largely been ineffective.

We have recently identified immune escape mechanisms relevant for AML, including the downregulation of MHC Class II in patients who relapse after allogeneic stem cell transplantation (Christopher et al., NEJM 2018), and an immunosuppressive phenotype in normal karyotype AML patients that is strongly associated with early relapse after chemotherapy (Ferraro et al., PNAS 2021), demonstrating the importance of the immune system in the response against AML. Therefore, we surveyed the TCGA AML RNA-seq dataset to identify dysregulated immune checkpoint genes, and identified VSIR overexpression (encoding VISTA, a checkpoint shown to be relevant in AML cell lines; Kim et al., JCI 2024) to be associated with DNMT3A and NPM1 mutations, two of the most common initiating mutations in AML. We also confirmed VISTA protein overexpression in patient AML blasts with these mutations. This association has also recently been independently reported (Bordeleau et al., Cell Reports 2024). Therefore, we hypothesize that suppression of T cell activity by VISTA expression on AML blasts may represent a novel mechanism of immune suppression in AML initiated by DNMT3A and NPM1 mutations.

In this study, we used a mouse model carrying the most common mutations for Dnmt3a (R878H, equivalent to R882H in humans) and Npm1 (NPM1c). While neither of these mutations independently causes highly penetrant AML in mice, nearly all mice with both mutations develop primary AML with a latency of 8-15 months. Using this model, we performed secondary transplants of AML cells (C57BL/6, Ly5.2+) into congenic Rag1^-/- (Ly5.1+, lacking B and T cells) and WT (Ly5.1+) recipients. Transplantation of 500,000 AML cells into non-irradiated recipients led to a median survival of 6.9 weeks in Rag1^-/- recipients (n=3), but median survival was not reached at 12 weeks of follow up in WT recipients (n=4, p=0.01). We also demonstrated an activated/exhausted T cell phenotype in WT recipients by flow cytometry and single cell RNA sequencing. These results suggest that immunocompetent mice can reject AML cells from Dnmt3a^R878H/+ x Npm1^cA mice, and that these tumors must employ mechanisms to evade T cell immunity.

Using an in vitro murine T cell suppression assay (splenic T cells are activated with anti-CD3/28 beads, co-cultured with different dosages of AML cells, and assessed for T cell activation by flow at 48 hours), we found that cells from 4 independent Dnmt3a^R878H x Npm1^cA AMLs were highly immunosuppressive (10.2% T cells activated at a 2:1 AML to T cell ratio, compared to 89.6% activated without AML cells, p<0.0001). This phenotype was not detected using murine AML cells initiated by fusion oncogenes (e.g. MLL::AF9, PML::RARA), nor was it detected in preleukemic hematopoietic progenitors from Dnmt3a^R878H x Npm1^cA mice, suggesting that it develops at progression to overt AML in this model. Single cell RNA sequencing of preleukemic vs. leukemic progenitor cells from these mice also showed that Vsir is overexpressed only in fully transformed cells. Using a CRISPR/Cas9 mediated approach to inactivate Vsir, we found that AML cells deficient for Vsir had reduced immunosuppressive activity (at a 1:1 ratio of T cells to AML cells, 59% of CD8+ T cells were activated in the Vsir KO sample, compared to 30% in the Rosa26 KO control, p=0.0001), strongly implicating VISTA as an immune checkpoint relevant for this model.

To confirm this hypothesis, we are performing transplants of secondary AMLs from this model into WT and Rag1^-/- recipients to determine whether inactivation of Vsir in AML cells will cause increased leukemia-free survival in immunocompetent recipients only. Additionally, we will overexpress Vsir in preleukemic progenitor cells in this model, to determine whether this is able to accelerate AML development in immunocompetent hosts. If successful, these studies will clarify how AML initiating and cooperating mutations may contribute to immune escape at AML progression, and may nominate VISTA blockade as a therapeutic strategy for AML patients.