An underlying mechanism for the increased mutagenesis of lagging-strand genes in Bacillus subtilis

Abstract

We previously reported that lagging-strand genes accumulate mutations faster than those encoded on the leading strand in Bacillus subtilis. Although we proposed that orientation-specific encounters between replication and transcription underlie this phenomenon, the mechanism leading to the increased mutagenesis of lagging-strand genes remained unknown. Here, we report that the transcription-dependent and orientation-specific differences in mutation rates of genes require the B. subtilis Y-family polymerase, PolY1 (yqjH). We find that without PolY1, association of the replicative helicase, DnaC, and the recombination protein, RecA, with lagging-strand genes increases in a transcription-dependent manner. These data suggest that PolY1 promotes efficient replisome progression through lagging-strand genes, thereby reducing potentially detrimental breaks and single-stranded DNA at these loci. Y-family polymerases can alleviate potential obstacles to replisome progression by facilitating DNA lesion bypass, extension of D-loops, or excision repair. We find that the nucleotide excision repair (NER) proteins UvrA, UvrB, and UvrC, but not RecA, are required for transcription-dependent asymmetry in mutation rates of genes in the two orientations. Furthermore, we find that the transcription-coupling repair factor Mfd functions in the same pathway as PolY1 and is also required for increased mutagenesis of lagging-strand genes. Experimental and SNP analyses of B. subtilis genomes show mutational footprints consistent with these findings. We propose that the interplay between replication and transcription increases lesion susceptibility of, specifically, lagging-strand genes, activating an Mfd-dependent error-prone NER mechanism. We propose that this process, at least partially, underlies the accelerated evolution of lagging-strand genes.

Keywords: TCR; Y-family polymerases; excision repair; replication conflicts; transcription orientation.

Fig. 1.

The transcription-dependent mutation asymmetry requires PolY1. (A) Schematic of the amyE locus harboring mutation reporters oriented head-on (HM419, HM421, HM632, HM634, HM417, HM724) or codirectionally (HM420, HM422, HM633, HM635, HM418, HM725) to replication. (B) Wild-type and mutant stop codons for the hisC952 (HM419, HM420, HM421, HM422), metB5 (HM632, HM633, HM634, HM635), and leuC427 (HM417, HM418, HM724, HM725) mutation reporters. (C) Mutation rates for the hisC952 reporter in the presence (HM419 and HM420; gray) or absence of PolY1 (ΔyqjH) (HM421 and HM422; black), for the head-on and codirectional orientations, with (Trx+) and without (Trx−) IPTG. (D) Same as C but for the metB5 reporter strains (HM 632 and 633; gray) (HM634 and 635; black). (E) Same as C but for the leuC427 reporter strains (HM417 and HM418; gray) (HM724 and HM725; black). Mutation rates were estimated based on C = 36–48 for each strain and condition. Error bars are 95% confidence intervals (*P < 0.05). Significance was determined by Student’s t test.

Fig. 2.

Association of the replicative helicase with lagging, but not leading-strand genes increases without PolY1. (A) Relative association of DnaC was analyzed by ChIP–qPCR with the region harboring the P_spank(hy)-hisC952 reporter gene, compared with the control region yhaX, in the head-on (HO) or codirectional (CD) orientations (HM449 and HM450; gray bars), with and without PolY1 (HM611 and HM612; black bars), with (Trx+) and without (Trx−) IPTG. (B) Relative association of DnaC by ChIP–qPCR with the region harboring the P_xis-lacZ reporter gene, compared with the control region yhaX, in the HO orientation, with [HM352 (Trx+) and HM630 (Trx−); gray bars] and without PolY1 (ΔyqjH) [HM629 (Trx+) and HM631 (Trx−); black bars] as well as the relative association of DnaC with the P_xis-lacZ reporter gene in the CD orientation, with [HM664 (Trx+) and HM663 (Trx−); gray bars] and without PolY1 (ΔyqjH) [HM685 (Trx+) and HM684 (Trx−); black bars]. (C) Relative association of DnaC with the genomic loci rpsD and rplGB with (wt; gray bars) and without PolY1 (ΔyqjH; black bars). Error bars represent SE (n = 12) (**P < 0.01, ***P < 0.005). Significance was determined by Student’s t test.

Fig. 3.

Without PolY1, RecA localization to lagging-, but not leading-strand genes increases. (A) Representative images of cells expressing P_xis-lacZ in the head-on orientation with (HM352) and without PolY1 (HM629), and the codirectional orientation with (HM664) and without PolY1 (HM685). (B) Quantification of RecA-GFP focus formation in cells harboring the P_xis-lacZ reporter in the HO orientation with [HM352 (Trx+) and HM630 (Trx−); gray] and without PolY1 [HM629 (Trx+) and HM631 (Trx−); black] and the CD orientation, with [HM664 (Trx+) and HM663 (Trx−); gray] and without PolY1 [HM685 (Trx+) and HM684 (Trx−); black]. (C) Relative association of RecA by ChIP–qPCR with the region harboring the P_xis-lacZ reporter gene, compared with the control region yhaX, in the HO orientation, with [HM352 (Trx+) and HM630 (Trx−); gray] and without PolY1 (ΔyqjH) [HM629 (Trx+) and HM631 (Trx−); black] as well as the relative association of RecA with the P_xis-lacZ reporter gene in the CD orientation, with [HM664 (Trx+) and HM663 (Trx−); gray] and without PolY1 (ΔyqjH) [HM685 (Trx+) and HM684 (Trx−); black]. (D) Same as C but for strains harboring the P_spank(hy)-hisC952 reporter genes with (HM594 and HM595; gray) and without PolY1 (ΔyqjH) (HM596 and HM597; black), with (Trx+) or without (Trx−) IPTG. (E) Relative association of RecA with the genomic loci rpsD and rplGB with (wt; gray) and without PolY1 (ΔyqjH; black). Error bars represent SE (n = 6–12) (**P < 0.01, ***P < 0.005, ****P < 0.001). Significance was determined by Student’s t test.

Fig. 4.

PolY1 functions independently of RecA and AddAB. (A) Mutation rates were estimated for the hisC952 reporter in strains lacking AddAB (HM720 and HM721, ΔaddAB; gray) or AddAB and PolY1 (HM740 and HM741, ΔaddAB ΔyqjH; black), for the head-on (HO) and codirectional (CD) orientations with (Trx+) and without (Trx−) IPTG. (B) Same as A but for strains lacking either RecA (HM726 and HM727, ΔrecA; gray) or RecA and PolY1 (HM746 and HM747, ΔrecA ΔyqjH; black), with (Trx−; black bars) or without (Trx+; gray bars) IPTG. Mutation rates were estimated based on C = 30–36, for each strain and condition. Error bars represent 95% confidence intervals (*P < 0.05).

Fig. 5.

PolY1 acts in the same pathway as UvrA and Mfd. (A) Mutation rates were estimated for the hisC952 reporter in control backgrounds (HM419 and 420, ctrl; gray), strains lacking PolY1 (HM421 and HM422, ΔyqjH; black), strains lacking UvrA (HM735 and HM736, ΔuvrA; white) or UvrA and PolY1 (HM1102 and HM1103, ΔuvrA ΔyqjH; light gray), for the head-on (HO) and codirectional (CD) orientations, with (Trx−) or without (Trx) IPTG. (B) Same as A but for strains lacking either Mfd (HM444 and HM445, Δmfd; stripes) or Mfd and PolY1 (HM874 and HM875, Δmfd ΔyqjH; argyle), with (Trx−) or without (Trx+) IPTG. Mutation rates were estimated based on C = 30–48, for each strain and condition. Error bars represent 95% confidence intervals (*P < 0.05).

Fig. 6.

SNP counts show a stronger positive correlation with gene length on the lagging strand. The number of SNPs for each gene on the lagging strand (blue dots) and the leading strand (red dots) is plotted as a function of its (translated) protein length. The SNPs AG and GA (A), CT and TC (B), AC and CA (C), GT and TG (D), AT and TA (E), and CG and GC (F), were pooled because direction of the mutations was not distinguished. Negative binomial regression with a log link function was used to model the gene length dependence of the number of SNPs for the lagging strand (blue lines) and the leading strand (red lines). The mutation rate (slope) is greater for the GA/AG and TC/CT mutations but increases nonlinearly with gene length (with P < 0.0001) for all six mutations. The difference in the rate of increase on the lagging strand compared with the leading strand of 4.5% per kb is significantly greater (P = 0.005) for the TC/CT mutation (SI Appendix, Table S1 and Fig. S1).

Fig. 7.

Models of PolY1 function at regions of lagging-strand transcription. Replication of the lagging-strand–encoded genes generates increased DNA lesions, either due to the discontinuous nature of its synthesis and/or head-on collisions between the replication and transcription machineries. RNA polymerases stalled at lesions (as in the postreplication repair model) or at the site of the collision itself (as in the collision model) trigger transcription-coupled recruitment of the NER machinery. Upon excision of the lesion-containing patch by NER, PolY1 fills in the resulting gap, causing mutations.