OA-46: Epigenetic regulation of CD38/48 by KDM6A mediates NK cell response in multiple myeloma

Introduction: Daratumumab (Dara) is a therapeutic antibody against CD38, which triggers both direct and immune-mediated cytotoxicity in multiple myeloma (MM) cells. Although Dara combination therapies have high efficacy, relapse due to Dara resistance is common.

Methods: To identify tumor-intrinsic genes involved in resistance to Dara, we performed two CRISPR knockout (KO) screens, and RNA-seq in KDM6A KO cells.

Results: At the first screen, we found 433 genes that were positively correlated with MM cytotoxicity triggered by Dara and primary NK cells. 16 of these genes were also correlated with cytolytic activity in most of the 36 cancer types we analyzed in the TCGA database.
Dara efficacy is closely associated with the expression of CD38, with nonresponders having lower CD38 expression. Therefore, our second screen used CRISPR to find genes whose KO caused the lowest expression of CD38 (bottom 5%). KDM6A was the top-ranked gene out of the set of overlapping genes from these two screens. We confirmed that KDM6A KO significantly decreased CD38 expression (>2 times, p< 0.01) and that re-introducing KDM6A into KDM6A KO cells restored the expression of CD38 and Dara-induced cytotoxicity.
KDM6A encodes lysine-specific demethylase 6A, which demethylates lysine 27 of histone H3 (H3K27) to promote gene expression. In fact, when we compared the CD38 promoter area between KDM6A KO and control cells using ChIP-seq and ChIP qPCR, we found that the H3K27me3 level was higher in KDM6A KO cells. Conversely, the re-introduction of KDM6A into KDM6A-KO cells decreased the H3K27me3 level. Concordantly, Dara-resistant cell lines had lower CD38 expression and higher levels of H3K27me3 at the CD38 promoter.
Interestingly, we found that CD38 overexpression only partially restored the sensitivity of KDM6A KO cells to Dara treatment, suggesting other mechanisms for KDM6A. We performed RNA-seq in KDM6A KO cells and found that CD48, an NK-activating ligand, was downregulated. Overexpression of CD48 in KDM6A KO cells significantly increased secretion of Granzyme B and Perforin by NK cells, restoring the activity of NK cells and re-sensitizing KDM6A KO cells to Dara treatment. These data suggest that KDM6A not only regulates CD38 expression but also NK cell activity by CD48 regulation, thereby modulating the efficacy of Dara.
KDM6A acts to oppose the EZH2/PRC2 complex by demethylating H3K27me3, and we found that Tazemetostat (Taze), an FDA-approved EZH2 inhibitor, increased the expression of CD38 and CD48 at the protein, mRNA, and surface expression levels, especially in KDM6A-KO cells. Importantly, Taze also restored the sensitivity of KDM6A KO cells to Dara-mediated NK cell cytotoxicity.

Conclusions: Taken together, our data reveal that KDM6A assists in Dara targeting by upregulating CD38 and in Dara cytotoxicity by modulating NK activity through CD48 regulation. By imitating its function using an FDA-approved inhibitor, we may overcome Dara resistance and improve patient outcomes in MM.