Abstract B023: Deep mutational scanning of KEAP1 to identify mutations resistant to radiation therapy and oxidative stress

Abstract

Listen

Translate article

Abstract Introduction: Mutations in KEAP1 and NFE2L2 (NFR2) occur in approximately 20% of non-small cell lung cancer (NSCLC) patients and are associated with resistance to radiotherapy and systemic treatments. In previous research, we found that not all KEAP1 mutations found in NSCLC patients are pathogenic. These benign mutations exhibit a wild-type phenotype that does not confer therapy resistance. With only roughly 100 of 13,000 (<1%) potential single amino acid substitution mutations in KEAP1 characterized thus far, here we comprehensively evaluated all possible point mutations in KEAP1 using deep mutational scanning (DMS). Methods: The High-throughput Mutagenesis by Integrated TilEs (MITE) method was used to generate the KEAP1 DMS library. In short, the KEAP1 sequence was segmented into 21 variable regions (tiles), each comprising a 90-base pair variable region flanked by a 30-base pair constant region. Each codon position included the 20 possible natural amino acids plus a stop codon, for a total of 13,076 KEAP1 variants. The tiles were then assembled to ensure that each full-length KEAP1 construct contained only a single amino acid substitution. The KEAP1 DMS lentivirus library was transduced into H1299-KEAP1NULLcells, followed by selective pressure with H2O2, radiation, or vehicle control. The KEAP1 ORF was then amplified from surviving cell genomic DNA, and mutations were quantified using next-generation sequencing. Functional annotations were derived through mutation enrichment analysis. Results: Both H2O2 and radiation selective pressure of the KEAP1 DMS library induced a bimodal distribution of KEAP1 mutations, with cells harboring KEAP1 nonsense mutations being enriched and cells with silent mutations being depleted. By setting a threshold of 95% specificity for silent mutations to differentiate between benign and pathogenic mutations, we identified 94% of nonsense mutations and 88% of previously characterized pathogenic mutations as pathogenic. While most KEAP1 mutations were categorized as benign (64%), analysis of mutations in TCGA NSCLC patients showed pathogenic mutations were enriched (72%). Mapping mutation phenotypes onto a KEAP1 structural model revealed an enrichment of pathogenic mutations at critical functional sites, including the NRF2 binding pocket of the Kelch domain, the BTB homodimerization interface, and the hydrophobic core of the oxidative stress-sensing IVR domain. Conclusions: Through deep mutational scanning experiments, we generated a comprehensive list of pathogenic KEAP1 single amino acid substitutions. These results align with previously known biological functions and prior single amino acid substitutions characterized in the literature. These findings will facilitate the identification of benign and pathogenic KEAP1 mutations in clinical settings, enabling personalized radiotherapeutic and systemic therapy strategies for patients harboring KEAP1 mutations. Citation Format: Noah Kastelowitz, Shashank Shrishrimal, Soyeong Jun, Anni MY Zhang, Rui Wang, Ilayda Ilerten, Maximilian Diehn. Deep mutational scanning of KEAP1 to identify mutations resistant to radiation therapy and oxidative stress. [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Targeted Therapies in Combination with Radiotherapy; 2025 Jan 26-29; San Diego, CA. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(2_Suppl):Abstract nr BO23.