Abstract
The DNA damage checkpoint system ensures genomic integrity by preventing the division of damaged cells, primarily through the G1/S and G2/M checkpoints. How these checkpoints collectively minimize error remains unclear. Here, we exposed non-cancerous human cells to DNA damage and used single-cell imaging to monitor spontaneous arrest failure. Under intermediate damage, cells divided with error via two major routes: one involving mitotic skipping (extensive engagement of the G2/M checkpoint) followed by endoreplication (escape from G1/S), and the second involving escape from the G2/M checkpoint. These pathways produced distinct ploidy, nuclear morphology, and micronuclei composition. Simulations and experiments showed that strengthening one checkpoint reduced one failure mode but exacerbated the other, revealing an inherent tradeoff. Notably, our findings suggest that the DNA damage checkpoint system minimizes total error not by fully preventing either route but by permitting both to occur at sub-optimal frequencies.