Posttranslational Changes in Histone Modification Driven By USP48 Loss Increase Sensitivity to Hypomethylating Agents in Acute Myeloid Leukemia

Despite advances in therapeutic strategies for patients with AML not eligible for intensive chemotherapy, survival rates for this subset of patients remain particularly poor. Among this group are older adult patients who receive the hypomethylating agents (HMAs) azacitidine or decitabine as frontline therapy. Despite their status as FDA-approved drugs, HMA monotherapy only achieves low response rates in patients with AML and is not a curative treatment. Discovery of novel sensitizers that can be used in combination with HMAs are needed to improve patient outcome. One challenge is that the mechanism by which HMAs are active in AML has been elusive.

We deployed genome-wide CRISPR-Cas9 screens using the two standard HMAs (decitabine and azacitidine) and a selective DNMT1 inhibitor (GSK-3680532) to identify novel response and resistance mechanisms to HMAs. Our screens identified well-characterized, as well as unexplored targets, with a strong overlap among the three different drugs. We discovered an enrichment of genes involved in chromatin remodeling and transcriptional regulation, with knockout (KO) of the gene USP48, which encodes a deubiquitinating enzyme, as the top sensitizer in AML cells. We validated USP48 KO in competitive growth and toxicity assays as a selective sensitizer for HMAs in multiple AML cell lines. In contrast, no increased toxicity was seen with other nucleoside analogues or DNA damage inducing agents, such as cytarabine or doxorubicin, suggesting that loss of DNMT1 and the demethylation of DNA is responsible for the observed synergy.

To begin to understand which mechanism of HMAs is enhanced through USP48 KO, we utilized cell differentiation and cell death assays. While the loss of USP48 alone did not affect cell growth or viability, the addition of low dose HMA treatment led to a decrease in cell viability with increased DNA damage and induction of apoptosis. In contrast, no differentiation was observed. In vivo, in two independent studies using an orthotopic mouse model of human AML, we found that the AML model with USP48 KO and 7-day treatment with the DNMT1 inhibitor GSK-3680532 had ablated disease burden in the bone marrow (< 0.5% hCD45+ blasts) and prolonged survival. Mice with non-targeting guides or USP48 KO treated with the vehicle control had ~25% hCD45+ blasts, while the DNMT1 inhibitor in the non-targeting control mice showed some antitumor effect (9% hCD45+ blasts).

To date, little is known about the direct substrates of USP48. To gain better insight into possible targets of USP48 we used an omics-based approach, generating ubiquitinome, proteome, interactome IP-mass spectrometry and RNA-Seq data. The ubiquitinome data revealed a significant enrichment of H2A variant ubiquitination and enhanced ubiquitination of key proteins involved in the recognition of DNA double-strand breaks (DSB) (Ku70/Ku80 and PRKDC) upon USP48 KO. These findings were confirmed in the interactome data set, where the H2AX core histone complex was among the most enriched CORUM protein complexes. RNA-Seq data showed increased expression of gene sets involved in DNA damage repair and chromatin organization upon USP48 KO, consistent with a cellular response to maintain chromatin stability.

Additional genome-scale CRISPR-Cas9 screening of USP48 KO cells with HMA treatment revealed the KO of RNF138, UBE2K and UBE2D3 as rendering resistance to all three drugs. While previous reports showed that USP48 erases BRCA1 ubiquitin sites, ours is the first work linking the RNF138-UBE2K/UBE2D3 ubiquitination machinery to USP48. RNF138, an E3 ubiquitin ligase, is also described to ubiquitinate and displace repair proteins from DSB sites, connecting RNF138 itself to DNA repair pathway choice. Low-throughput validation confirms rescue of HMA + USP48 KO toxicity with the knockout of either RNF138 or its E2 ligases UBE2K/UBE2D3 in AML cells.

Taken together our data confirm a role for USP48 as a histone deubiquitinase, influencing the modification of histones and other proteins important for DNA damage repair pathway choice. While USP48 KO alone does not affect AML cell viability, combining USP48 KO with HMA treatment is highly lethal to AML suggesting a synergistic interplay between DNA hypomethylation, ubiquitin posttranslational modifications, DNA damage and AML cell death. USP48 is thus a new target for combination therapy with HMAs for AML.