SETDB1 Suppresses Interferon Responses and NK Cell-Mediated Immunosurveillance Specifically in Monocytic AML

Acute myeloid leukemia (AML) with monocytic differentiation (monocytic AML) responds poorly to current treatments, including venetoclax-based therapy. To identify novel therapeutic targets in monocytic AML, we conducted in vivo and in vitro CRISPR/Cas9 library screens using a pooled sgRNA library targeting epigenetic regulators with a mouse monocytic AML model driven by SETBP1 and ASXL1 mutations (cSAM: combined expression of SETBP1 and ASXL1 Mutations). We ranked all the epigenetic genes according to the in vivo specific essentiality, and selected several candidates which were more depleted in mice rather than in cell cultures. The top-ranked gene was Setdb1, which encodes a histone H3 lysine 9 (H3K9) methyltransferase. In addition, several H3K9 trimethylation (H3K9me3) regulators, ATF7IP and TRIM33, were also more important in vivo than in vitro in our screens. ATF7IP is a cofactor of SETDB1 to enhance its enzymatic activity. TRIM33 is also involved in H3K9me3 modification. These data suggest the critical role of the SETDB1-ATF7IP-TRIM33 H3K9 methyltransferase complex in the development of AML in vivo.

To validate the results of our screenings and to investigate the role of immune cells in our transplantation assay, we next transplanted control and Setdb1-knockout (KO) cSAM cells into three types of mice with different immune systems: the immunocompetent C57BL/6 mice, the immunodeficient NSG mice lacking T, B and natural killer (NK) cells, and the C57BL/6J- Rag2^-/- mice lacking mature T and B cells but possessing NK cells. Although Setdb1 depletion did not inhibit the growth of cSAM cells in NSG mice, it showed the strong growth-inhibitory effect in C57BL/6 mice and Rag2^-/- mice (Fig 1A). Since both C57BL/6 mice and Rag2^-/- mice have NK cells, these data suggest that SETDB1 promotes leukemogenesis by suppressing NK cell-mediated cytotoxicity in vivo. Mechanistically, loss of Setdb1 induced upregulation of NKG2D ligands, interferon-stimulated genes (ISGs), and endogenous retrovirus (ERV) sequences through H3K9me3 reduction, thereby rendering cSAM cells more susceptible to NK cell-mediated cytotoxicity. Setdb1 depletion also induced myeloid maturation of cSAM cells towards macrophages and downregulation of major histocompatibility complex class I (MHC-I), which contributed to the increased susceptibility of Setdb1-depleted cSAM cells to NK cell-mediated cytotoxicity.

Next, we investigated whether SETDB1 inactivation also sensitizes human AML cells to NK cells using the co-culture assay with an IL-2 dependent NK cell line NK-92. Loss of SETDB1 rendered monocytic AML cell lines, NOMO-1, THP-1, OCI-AML3, U937, and NB4, more susceptible to NK-92 cell-mediated cytotoxicity. SETDB1 depletion also induced overactivation of interferon signaling and intrinsic apoptosis in these monocytic AML cells. In contrast, non-monocytic AML cells, such as K562, TF-1, and Kasumi-1, were not sensitive to SETDB1 depletion. SETDB1 depletion neither increased the susceptibility to NK cell-mediated cytotoxicity nor induced apoptosis in non-monocytic AML cells. Subsequent analyses revealed that SETDB1 depletion reduced H3K9me3 at the monocyte-specific enhancer regions with ERV sequences to upregulate NK cell-activating ligands and ISGs. Because the “ERV enhancers” are active only in monocyte lineage, SETDB1 is essential specifically in monocytic AML.

Finally, we identified MNDA and its mouse counterpart Ifi203 as biomarkers to predict the sensitivity of each AML to SETDB1 depletion. MNDA is one of the ISGs and is involved in myeloid cell differentiation. Human monocytic AML cells that are dependent on SETDB1 expressed higher MNDA at both RNA and protein levels compared to other subtypes of AML. We also found that Ifi203-high mouse AML cells transformed by MLL-AF9 were sensitive to Setdb1 depletion, whereas Setdb1 was dispensable in Ifi203-low mouse AML cells transformed by RUNX1-RUNX1T1.

In summary, our findings reveal the critical and selective role of SETDB1 in monocytic AML. SETDB1 mediates H3K9me3 at the monocyte-specific ERV enhancers to repress the expression of NK cell-activating ligands and ISGs, thereby suppressing immunogenicity and apoptosis in monocytic AML (Fig 1B). SETDB1 as well as its binding partners could be promising therapeutic targets to treat monocytic AMLs that are relatively resistant to the current standard therapies.