NOXA modulates tipifarnib sensitivity in myeloid leukemia cells.

Abstract

Abstract Tipifarnib is a potent small-molecule farnesyltransferase inhibitor (FTI) that has demonstrated single-agent activity in acute myeloid leukemia (AML). Although not all patients responded to tipifarnib, those who did sustained durable, clinically meaningful responses. Given ongoing efforts to develop next-generation FTIs, there is a critical need to elucidate their mechanism of action in AML to identify biomarkers that can predict therapeutic response.To address this, we performed a genome-wide CRISPR screen in U937 leukemia cells treated with tipifarnib or vehicle. DNA was collected at Days 0, 7 and 14 and sequenced to determine sgRNA enrichment present in tipifarnib-treated cells. The most enriched sgRNA targeted PMAIP1, showing 1.5- and 2-fold increases at days 7 and 14, respectively, compared to control. PMAIP1 encodes for NOXA, a BH3-only member of the BCL-2 family that promotes apoptosis by allowing pro-apoptotic proteins to permeabilize the mitochondrial membrane. To validate these results, we generated NOXA KO clones in U937 and TF-1 cells. Cell viability assays showed that NOXA KO cells are significantly less sensitive to tipifarnib. Next, we performed competition assays by co-culturing GFP-labeled NOXA KO U937 cells with RFP-labeled WT cells, treating with tipifarnib for 7 days, and monitoring daily by flow cytometry. While no changes were seen with vehicle treatment, the ratio of KO/WT cells increased to 2.5 after 7 days, indicating that NOXA loss confers a context-specific fitness advantage in the presence of tipifarnib. Lastly, we generated tipifarnib-resistant U937 cells, which showed a decreased expression of NOXA, suggesting that NOXA is necessary for the sensitivity to tipifarnib treatment in leukemia cells.To assess whether NOXA's non-canonical CAAX box mediates its response to tipifarnib via direct farnesylation, we overexpressed WT and CAAX-mutant NOXA in NOXA KO cells. Both showed similar sensitivity to tipifarnib, suggesting the phenotype is not due to direct farnesylation. We then tested the stability of NOXA protein in the presence of tipifarnib by treating cells with either vehicle or tipifarnib and the translation inhibitor cycloheximide. The rate of NOXA degradation in tipifarnib-treated cells was significantly faster than in vehicle-treated cells, suggesting that tipifarnib is regulating NOXA protein degradation. To identify proteins involved in NOXA degradation, we analyzed negatively enriched CRISPR screen hits with CAAX motifs and found HDJ2 as a top candidate. HDJ2 is a chaperone protein, involved in protein folding and proteolysis, and is a canonical target of farnesyltransferase. To determine if NOXA and HDJ2 interact, we performed co-immunoprecipitation experiments in U937 cells treated with tipifarnib and observed an interaction that was enhanced in the tipifarnib-treated cells, suggesting that tipifarnib promotes HDJ2 binding to NOXA. We then analyzed mitochondrial localization of both NOXA and HDJ2 after tipifarnib treatment. NOXA and HDJ2 localization to the mitochondria was increased after tipifarnib treatment, suggesting that NOXA degradation depends on HDJ2 chaperone activity.Given these mechanistic data, we hypothesized that low NOXA levels may be associated with resistance to tipifarnib, while high levels with clinical response in AML. To identify patients most likely to respond to tipifarnib, we analyzed the TCGA AML dataset for subsets with elevated NOXA expression. Interestingly, we found that NOXA expression was increased in patients with IDH2 mutations. To confirm this in vitro, we analyzed NOXA mRNA expression in TF-1 cells with an isogenic IDH2 R140Q mutation and its WT control and demonstrated that expression of NOXA was increased in IDH2 mutant cells compared to WT. We then treated WT TF-1 cells with 2-hydroxyglutarate and observed an increase in NOXA expression compared to nontreated cells. Additionally, we observed that TF-1 IDH2 R140Q cells were more sensitive to tipifarnib than WT TF-1 cells, suggesting that patients with an IDH2 mutation may benefit from tipifarnib treatment. This is consistent with clinical data in angioimmunoblastic T-cell lymphoma that suggested responses were enriched in IDH2 mutated cases. Ongoing studies will include sequencing samples from patients treated with tipifarnib in the cooperative group phase 3 study (NCI 2009-00535) to determine the incidence of IDH2 and other mutations in responders vs. non-responders.