In vitro and in silico modelling of ROS1 ‐positive non‐small cell lung cancer reveals fusion‐dependent tyrosine kinase inhibi

The emergence of drug resistance in ROS1‐rearranged non‐small cell lung cancer (NSCLC) represents a major therapeutic challenge. Although several resistance mutations have been described in patients, preclinical models derived from patient material remain limited, thereby hindering mechanistic insight into this malignancy. In this study, we investigated three clinically relevant ROS1 variants (G2032R, L2026M, and S1986Y) using CRISPR/Cas9‐edited patient‐derived cell lines and complemented these analyses with molecular docking and molecular dynamics simulations. The efficacy of tyrosine kinase inhibitors (TKIs) crizotinib, ceritinib, lorlatinib, entrectinib, and repotrectinib was systematically evaluated. Dose–response assays with CUTO‐28 (TPM3–ROS1) and CUTO‐37 (CD74–ROS1) lines reproduced clinical drug responses, revealing fusion‐dependent differences in resistance. The G2032R mutation led to reduced sensitivity to entrectinib and lorlatinib, with lower area‐over‐the‐curve values compared with ROS1 wild‐type (WT) cells. Similar reductions were observed for L2026M and S1986Y, while complete resistance was only seen in CUTO‐37 G2032R. Immunoblotting confirmed impaired inhibition of p‐ROS1 (Tyr2274) in resistant models. Structural modelling revealed alterations in kinase active site pocket volume, activation helix rotation, and activation loop dynamics in ROS1 mutants, providing a mechanistic basis for the observed drug responses. Molecular dynamics simulations validated the type I inhibitor binding mode across ROS1 WT and mutant complexes, while highlighting conformational effects extending beyond direct ligand interactions. Our findings underscore that although G2032R and L2026M mutations reside within the kinase active site, their impact extends far beyond steric hindrance, altering overall kinase domain dynamics. Collectively, these data establish a robust panel of patient‐derived ROS1 cell lines that recapitulate clinical resistance patterns and, together with complementary computational modeling, provide a valuable framework to dissect ROS1 tumor biology and support rational design of next‐generation inhibitors.