Fig. 4. Error-prone TLS induces the majority of mutations observed upon genotoxin exposure.

a Estimated mean fractions of mutations caused or prevented by three C. elegans TLS polymerases (REV-3, POLH-1 and POLQ-1) across genotoxin exposures along with 95% confidence intervals. Bars with no fill colour reflect non-significant contributions (chi-squared test FDR 5% in each mutation category). Layout as in Fig. 3c. b Aristolochic acid-induced signatures in wild-type and TLS polymerases η (polh-1) and θ (polq-1) deficient C. elegans. Layout as in Fig. 3b. c Interactions of APOBEC-induced DNA damage with deficiency in human REV1- or UNG-dependent BER. APOBEC mutational signatures of base substitutions in REV1 and UNG wild-type (top) and deficient (centre) samples and signature fold changes per mutation class (bottom). Layout as in Fig. 3d. Right: Observed difference between the fractions of C > G and C > T mutations in a TCN context in samples with (mut) and without (wt) REV1 or UNG defects with sample sizes indicated in parentheses. Boxes denote the interquartile range (IQR, 25–75% percentile), thick lines the median, whiskers 1.5× the IQR below the first quartile and above the third quartile. Grey dots represent data from individual samples. Difference in the fraction of C > T mutations was significant between classes as assessed by a two-sided binomial test (p value ~5 × 10⁻⁷).