Abstract
Mutation-intolerant genes (MIGs), which are constrained in tumors yet variable in normal tissues, are critical for cancer survival. Herein, we developed miDriver, a computational framework using pancancer-normal mutation contrasts to identify 1,020 MIGs across 8,096 tumors of 13 cancer types. Strikingly, MIGs are highly associated with synthetic lethality, cell-cycle progression, and clinical outcome. CRISPR screening reveals MIGs, especially CHEK1, as cancer-specific vulnerabilities, whose suppression impairs tumor proliferation and migration. Single-cell transcriptomics reveals a CHEK1-high subpopulation exhibiting stem-like and immune-suppressed features, linking tumor-intrinsic fitness to microenvironment remodeling. Multiplexed immunofluorescence revealed that CHEK1 and MIF are co-expressed in tumor cells, and CHEK1-high tumor cells exhibit closer spatial proximity to M2-like macrophages. Mechanistically, CHEK1 promotes p53 phosphorylation to upregulate MIF expression and secretion, thereby driving M2-like macrophage polarization via the MIF-CD74 axis. In vivo, targeting the CHEK1-MIF axis (particularly CHEK1) broadly reverses immunosuppression. Clinically, higher tumor CHEK1 levels are associated with poorer response to anti-PD-1 therapy. Exemplified by CHEK1, these findings establish MIGs as dual therapeutic targets capable of simultaneously disrupting tumor-intrinsic fitness and remodeling the immunosuppressive niche. This work proposes a novel paradigm for selectively targeting MIGs to eliminate aggressive tumor subclones while minimizing toxicity to normal cells.