Targeting PRMT9 Overcomes Venetoclax Resistance in AML By Modulating Splicing and Inhibiting Translation

For older AML populations ineligible for intensive chemotherapy and allogeneic stem cell transplantation, combining the BH3 mimetic venetoclax (VEN) with hypomethylating agents has emerged as first-line therapy. Despite early responses, resistance to VEN, marked by decreased mitochondrial-apoptotic-priming (“BH3 priming”), emerges over time. Protein arginine N-methyltransferases (PRMTs), which frequently exhibit aberrant activity in malignant hematopoiesis, reportedly regulate RNA splicing. Recent reports reveal that disruption of the splicing machinery increases AML cell sensitivity to VEN. Hence, we asked if modulating PRMT activity would enhance VEN responses. Indeed, datamining of published CRISPR screen results revealed synthetic lethal interactions between loss of individual PRMTs (2,6,7,8, or 9) and VEN treatment.

To precisely identify PRMTs whose loss enhances sensitivity to mitochondrial apoptosis, we performed loss-of-function studies (based on pharmacological inhibitors or shRNAs) targeting individual PRMTs in VEN-resistant AML patient-derived-xenograft (PDX) cells and Molm13 (Molm13/R) cells using dynamic BH3 priming as readout. In these models, cells were rendered VEN-resistant via in-vivo or ex-vivo administration. The top hit in the BH3 priming assay was PRMT9, a recently defined PRMT (PMID: 38413714). Similar results were obtained using an isogenic Molm13 line resistant to VEN with TP53 R248Q heterozygosity (a gift from PMID: 35026842). Next, we examined PRMT9 function in Molm13/R cells transduced with a doxycycline (dox)-inducible PRMT9-shRNA. PRMT9 KD by dox treatment induced modest apoptosis in resistant cells (shCtrl 4.5%, shPRMT9 13.3%, p=0.02), while combining PRMT9 KD with VEN at clinically achievable levels (IC25 1µM) induced robust apoptosis (VEN monotherapy 14.6%, combination 70.7%, p<0.001). Viability assessment showed that PRMT9 KD decreased VEN IC50 from 3.9µM to 0.2µM. We then verified PRMT9 function in an MLL-AF9/FLT3-ITD double-hit Mx1-Cre/Prmt9^f/f (MA9/FLT3-ITD/Prmt9-cKO) mouse model. Unlike the MA9 single-hit model, the double-hit model exhibited poor responsiveness to VEN plus azacitidine (V/A) treatment, due to MCL1 upregulation. We next transplanted double-hit Prmt9-cKO or control leukemic cells into NSG mice and treated them with PIPC to induce PRMT9-KO or with V/A simultaneously. Prmt9 targeting combined with V/A treatment almost completely eliminated leukemia burden, and mice treated with both survived the entire observation period.

We then tested anti-AML effects of our in-house PRMT9 inhibitor (P9i) combined with VEN in VEN-relapsed samples (n=4). P9i alone induced modest apoptosis, while the P9i/VEN combination (at their IC25s) significantly increased apoptosis. We evaluated P9i as part of combination therapy in vivo using the double-hit model. Leukemic mice were divided into four groups and treated for 3 weeks with: 1) vehicle, 2) P9i (100mg/kg, IP, twice a day, 5 times/week), 3) VEN/AZA (V/A: VEN, 100 mg/kg, orally, daily; AZA, 3 mg/kg, IP, 3 times/week) or 4) the P9i/VEN/AZA combination. Notably, combination treatment significantly extended survival relative to control or monotherapy groups.

Mechanistically, Molm13/R cell transcriptome analysis revealed that PRMT9-KD significantly increased aberrant splicing of some transcripts whose loss exhibits synthetic lethality with VEN treatment, such as ALG13. ALG13 encodes the protein-forming uridine diphosphate-N-acetylglucosamine (UDP-GlcNAc) transferase, which is crucial for catalyzing protein asparagine (N)-linked glycosylation and maturation of proteins such as ABCC1 (PMID: 20535133) Interestingly, targeting ABCC1 reportedly reverses VEN resistance (PMID: 37726279). PRMT9 inhibition remarkably decreased ALG13 protein levels; ALG13 knockout (KO) phenocopied PRMT9 targeting-induced sensitivity to mitochondrial apoptosis. PRMT9 inhibition also significantly repressed protein translation, decreasing levels of short-lived proteins. Notably, MCL1 protein expression, which contributes to VEN resistance, remarkably decreased after PRMT9 inhibition, as evidenced by shifts in polysome profiling seen following PRMT9 KD.

In summary, our study indicates that targeting PRMT9 can overcome VEN resistance in AML, likely through splicing modulation and translation inhibition.