CAAP1 loss uncovers vulnerability of MTAP-deficient NSCLC to PRMT5 inhibition with AZD3470

Introduction: This study aimed to identify genes that sensitize NSCLC cells to AZD3470, an MTA-cooperative PRMT5 inhibitor that targets MTAP-deficient tumors while sparing normal tissues. Early clinical data for this class of drugs as monotherapy has shown objective response rates (ORRs) of approximately 20% across multiple tumor indications. Encouragingly, the duration of response and disease control rates are substantial, suggesting that PRMT5 inhibitors may play a significant role in disease stabilization. While these findings are promising, they also highlight the need to further improve clinical efficacy. Therefore, we aimed to identify genetic alterations-beyond MTAP deficiency-that contribute to sensitivity to PRMT5 inhibition. Experimental procedures: To identify genes associated with sensitivity to AZD3470 we performed a genome-wide CRISPR screen across six NSCLC cell lines. Hit validation was conducted using two NSCLC CRISPR/Cas9 generated isogenic pairs. RNAseq, cell cycle analysis, apoptosis assay and gammaH2AX staining were used to provide mechanistic insight for AZD3470 sensitization. Finally, findings were validated in 29 NSCLC PDX models. Results: The genome-wide CRISPR screen identified CAAP1 loss as one of the top sensitizers to AZD3470 treatment. The CAAP1 (Caspase activity and apoptosis inhibitor 1) gene is located on chromosome 9p21, in close proximity to MTAP, and was found to be co-deleted with MTAP in approximately 20% of tumors. CRISPR screen results were further validated using HCC15 and NCI-H838 CAAP1 KO/WT isogenic pairs in long term proliferation assays, where CAAP1 null cells showed increased sensitivity to AZD3470 compared to CAAP1 WT cells. In addition, cell cycle analysis of the NCI-H838 and HCC15 CAAP1 KO cells after 96h of treatment with 1µM AZD3470 showed a higher percentage of cells in G2/M phase compared to CAAP1 WT cells. Moreover, CAAP1 deficiency potentiated both DNA damage and apoptosis induced by AZD3470. RNAseq analysis of HCC15 CAAP1 KO/WT isogenic cell line pairs revealed significant number of differentially spliced events in CAAP1 KO versus CAAP1 WT cells that increased with AZD3470 dose. Notably, the most abundant changes were detected in the transcripts with skipped exons that were enriched for pathways such as DNA repair or cell cycle consistent with the increased DNA damage and cell cycle alterations detected in CAAP1 null cells treated with AZD3470. Finally, our findings were confirmed using PDX models, where 71% of NSCLC PDXs in which AZD3470 induced regression were CAAP1 null. Conclusions: In summary, CAAP1 deletion in MTAP-null tumors increases sensitivity to AZD3470, both in vivo and in vitro, by potentiating AZD3470 induced cell cycle alterations and DNA damage, potentially through the regulation of alternative splicing.