The nature of mutations induced by replication–transcription collisions

Abstract

The DNA replication and transcription machineries share a common DNA template and thus can collide with each other co-directionally or head-on. Replication–transcription collisions can cause replication fork arrest, premature transcription termination, DNA breaks, and recombination intermediates threatening genome integrity. Collisions may also trigger mutations, which are major contributors to genetic disease and evolution. However, the nature and mechanisms of collision-induced mutagenesis remain poorly understood. Here we reveal the genetic consequences of replication–transcription collisions in actively dividing bacteria to be two classes of mutations: duplications/deletions and base substitutions in promoters. Both signatures are highly deleterious but are distinct from the previously well-characterized base substitutions in the coding sequence. Duplications/deletions are probably caused by replication stalling events that are triggered by collisions; their distribution patterns are consistent with where the fork first encounters a transcription complex upon entering a transcription unit. Promoter substitutions result mostly from head-on collisions and frequently occur at a nucleotide that is conserved in promoters recognized by the major σ factor in bacteria. This substitution is generated via adenine deamination on the template strand in the promoter open complex, as a consequence of head-on replication perturbing transcription initiation. We conclude that replication–transcription collisions induce distinct mutation signatures by antagonizing replication and transcription, not only in coding sequences but also in gene regulatory elements.

Extended Data Figure 1. Development of a forward mutation assay that detects loss-of-function mutations in B. subtilis

a, Simplified diagram of thymidine monophosphate (dTMP) synthesis. The phage-encoded thyP3 encodes thymidylate synthetase, which synthesizes dTMP and dihydrofolate (DHF) from dUMP and tetrahydrofolate (THF). DHF is recycled back to THF by dihydrofolate reductase (DHFR). Trimethoprim inhibits DHFR, thus blocking recycling of the essential cofactor THF and available THF is depleted by active thymidylate synthetase and cell growth is inhibited. Because cells with active thymidylate synthetase rely solely on endogenous dTMP synthesis, thyP3⁺ cells are sensitive to trimethoprim and loss-of-function mutations in thyP3 lead to trimethoprim resistance, which is the basis for the forward mutation assay. Viabilities of wild-type (thyA⁺ thyB^ts), thyP3⁺ (ΔthyA thyB^ts thyP3⁺) and thyP3⁻ (ΔthyA thyB^ts thyP3⁻) cells are shown in the table and representative colonies (at 45 °C) are shown on the right. b, Competition between strains carrying wild-type (wt) and mutant thyP3 in ΔthyA thyB^ts background to determine if there is any selective pressure on different mutants during growth phase at permissive temperature (37 °C). Relative fitness (mean±s.d) of six replicates is shown. c, Shifting the temperature to 45 °C does not affect plating efficiency during selection for trimethoprim resistance. Wild type and mutant thyP3 cells were grown at 37 °C and plated on solid medium supplemented with IPTG+thymine at 37 °C and 45 °C, and CFU/mL/OD was determined. Mean±s.d of 3 replicates is shown. d, thyP3 mutants have growth defects without thyB^ts. The doubling times of thyP3 mutant (a deletion and a frame-shift mutant) in the ΔthyA ΔthyB background at 37 °C are longer, indicative of growth defects in the absence of the backup gene thyB. Mean±s.d of 3 replicates is shown. For b and d, the mutant strains are listed in Extended Data Table 1.

Extended Data Figure 2. Expression level and mutation rate of thyP3

a, thyP3 expression in co-directional and head-on orientations. Using real-time quantitative PCR, mRNA level of thyP3 in the co-directional and head-on strains under induced (+IPTG) condition was measured and normalized to the reference gene accA. Since level of expression is similar between the strains, the observed difference in thyP3 mutation rate between co-directional and head-on orientations (Fig. 1d) is not caused by intrinsic differences in the expression level of thyP3. b, The orientation-specific difference in thyP3 mutation rate is not due to global increase of mutagenesis in the head-on strain. As a control to show that the increase in mutation rate is local to thyP3 reporter, we measured the mutation rate for resistance to nalidixic acid (Nal^R, conferred by mutations in gyrA gene) in co-directional and head-on strains. Since the Nal^R mutation rates in the two strains were similar, we conclude that the observed increase in head-on mutation rate is specific to thyP3 gene. c, Schematics of the co-directional and head-on thyP3 constructs (left) and an additional control to examine the effect of the genomic context on thyP3 mutagenesis, the neighboring genes were swapped (right). In each construct, the thyP3 gene is flanked by the lacI gene and the spectinomycin-resistance gene. The reporter constructs were integrated into the chromosome at the amyE locus by double crossover. The direction of replication is shown at the top. The co-directional-swapped strain was created by inverting the lacI-thyP3-spc from the head-on strain and the head-on-swapped construct was created by inverting the same from co-directional strain. Thus the swapped constructs switch the neighboring transcription units. The dotted lines in each construct show the swapping boundary. d, The mutation rate of swapped head-on strain is still higher than swapped co-directional strain when transcription is induced (+IPTG), indicating that the difference in mutation rate between reporter strains is not due to the direction of thyP3 relative to its neighboring genes. e, Mutation rate of co-directional and head-on thyP3 under uninduced (−IPTG) and induced (+IPTG) transcription. The rate of each class of mutations obtained under each condition is also depicted within each bar. For b, d and e mean±s.e.m of n≥3 independent experiments is shown. (**P<0.01, Student’s t-test).

Extended Data Figure 3. Mutation spectra of thyP3 under induced transcription

Illustrations of the mutation spectra of the thyP3 mutants obtained from fluctuation tests of: a, co-directional (n=214) and b, head-on (n=232) strains when transcription is induced (+IPTG). The thyP3 coding sequence with its promoter is shown. Sequence coordinates are indicated with reference to +1 transcription start site. The symbols used to represent different mutations are shown at the bottom, and base substitutions are shown in blue color above the sequence. The numbers marked in orange next to a mutation denote the frequency. The promoter elements, Shine-Dalgarno (SD) sequence, and start and stop codons are highlighted in each spectrum.

Extended Data Figure 4. Mutation spectra of thyP3 under un-induced transcription

Illustrations of the mutation spectra of the thyP3 mutants obtained from fluctuation tests of: a, co-directional (n=163) and b, head-on (n=178) strains when transcription is not induced (−IPTG). The thyP3 coding sequence with its promoter is shown. Sequence coordinates are indicated with reference to +1 transcription start site. The symbols used to represent different mutations are shown at the bottom, and base substitutions are shown in blue color above the sequence. The numbers marked in orange next to a mutation denote the frequency. The promoter elements, Shine-Dalgarno (SD) sequence, and start and stop codons are highlighted in each spectrum.

Extended Data Figure 5. Absence of selection bias in thyP3 forward mutation assay

Growth competition experiments were performed between the C_-7 promoter mutant against the following mutants: a, missense mutant, b, nonsense mutant and c, frameshift mutant. Each mutant was competed against the C_-7 promoter mutant to check if there is a competitive disadvantage for a mutant that has a mutation within the coding sequence, which may explain the high frequency of C_-7 mutation compared to other mutations. The results show no fitness disadvantage for any of the mutants tested suggesting that the high frequency of C_-7 promoter mutation is not due to a selection bias. For a–c mean±s.d of six replicates is shown and mutants competed are indicated within the plot. d, Plating efficiency of different thyP3 mutants. Plating efficiency was determined to check whether different classes of thyP3 mutants have differences in their plating efficiency on trimethoprim selection plates at 45 °C, which may explain the variation in the mutation rates and spectrum. The result shows similar plating efficiency among the different mutants, suggesting that plating efficiency does not underlie the variation in the observed mutation rates. The different mutants tested are indicated on the X-axis. Mean±s.d of 3 replicates is shown. The mutant strains are listed in the Extended Data Table 1.

Extended Data Figure 6. Mechanism of indel generation

a, Representative deletion and duplication events in thyP3. A high frequency deletion and duplication event observed in thyP3 gene in co-directional and head-on strains. The sequence coordinates are denoted and repeat sequence is underlined. b, Table showing the mutation rate of indels (≥3 bp) in intragenic region and promoter normalized by the length of the region suggests that the localized rate of indels is higher in the promoter than the intragenic region. c, First encounter between replication and transcription machineries generates indels. Model describing the first-encounter hypothesis proposed based on results presented in Fig. 2a–f. In co-directional orientation under induced transcription (+IPTG), when an array of RNA polymerase (RNAP) transcribe the gene the replisome is likely to collide with the first transcription complex at the promoter or promoter-proximal regions. On the contrary, when transcription is induced in head-on orientation, the replisome encounters the first transcription complex from the 3′ end. In support of this first encounter model, when transcription is not induced (basal level) the density of RNAP is sparse along the gene, hence the site of collisions are altered. In addition, it is possible that under basal transcription, collision can occur between replisome and RNAP complex arrested at the promoter or with the Lac repressor, which may explain the relatively high frequency of deletion at the promoter. Thus the “first-encounter” model of replication-transcription collisions supports that collisions stall replisome progression triggering indel mutations. d, Mutation rate of insertions and deletions (≥3 bp) within the intragenic region plotted individually. Frequency of insertions is increased by transcription in both co-directional and head-on orientations, whereas deletion frequency is specifically increased in head-on orientation. Mean±s.e.m of n≥3 experiments is shown. e, Models illustrating the different pathways that can lead to generation of indels following head-on collision-induced replication stalling: slippage, fork-reversal or template switching. f, Illustration of a complex mutation observed in thyP3 that is likely generated via Microhomology-Mediated Break Induced Replication (MMBIR). The complex mutation encompassing a deletion and insertion of an inverted region was observed under induced transcription in head-on orientation. The sequence coordinates are marked on the top with reference to the transcription start site (+1).

Extended Data Figure 7. Role of recombination protein RecA in collision-induced mutations

a, Mutation rates of co-directional and head-on thyP3 strains for trimethoprim resistance in ΔrecA background. Similar to wild-type the mutation rate of head-on is higher than the co-directional strain, although the total rate of mutation is decreased in ΔrecA background. b, Rate of ≥3 bp indels at the promoter in co-directional orientation is strongly decreased in ΔrecA cells suggesting that indels at the promoter are mostly RecA-dependent. c, Intragenic distribution of ≥3 bp indels in ΔrecA is similar to the distribution observed for wild-type (Fig. 2f), thus suggesting that RecA is not necessary for the collision-induced indels within coding region. d, Mutation rate of base substitutions in ΔrecA cells is higher in head-on than co-directional orientation. e, The rate of T_-7→C_-7 mutation is higher in head-on relative to co-directional orientation in ΔrecA cells, thus promoter substitutions can occur at a higher rate independent of recombination-mediated repair. All the fluctuation tests in ΔrecA background were performed under inducing conditions (+IPTG). For a–e mean±s.e.m of n≥3 experiments is shown. (**P<0.01; ***P<0.001; Student’s t-test)

Extended Data Figure 8. Base substitutions and the role of mismatch repair and enzymatic adenine deamination

a, IPTG-induction does not affect the base substitution rate in coding region of thyP3 when considering identical target sites, indicating that collisions may not be a major source of these mutations. In yeast, it was shown that transcription-associated mutagenesis is proportional to level of transcription. In B. subtilis, the total rate of base substitutions in the coding region significantly decreases upon IPTG induction, which could be due to an unidentified transcription dependent mutation-correction mechanism, or due to increase of target size of base substitutions in the coding sequence in un-induced (basal) transcription. b, Table showing the rates of base substitutions in coding region and promoter of thyP3 normalized by length of the region. Localized substitution rates are higher in the promoter than coding sequence, thus suggesting that collision has more drastic effect on promoter substitutions. c, Comparative genomic analysis of mutation rates of promoters with and without repressor binding. Nucleotide diversity per site (Theta) was calculated for each promoter across different strains of Bacillus subtilis. The comparison shows no significant difference in nucleotide diversity between repressor-bound promoters and rest of the promoters, indicating that repressor binding may not affect the substitution rate of a promoter. Whole genomes and the repressors analyzed are listed in Extended Data Table 2. (ns-not significant P>0.05; Mann-Whitney U test). d, The mutation frequency of T_-7→C_-7 mutation is higher in head-on than co-directional orientation in E. coli. The mutation frequency was calculated here from the plasmid-based forward mutation assay data reported by Yoshiyama et al., (2001). e, The restriction digestion-based assay to screen for T_-7→C_-7 mutation. Wild-type promoter sequence does not have an AflIII restriction site, whereas the promoter T_-7→C_-7 mutation will be digested by AflIII, which is illustrated by a representative agarose gel. f, Mismatch repair mutant (mutSL::kan) shows an expected increase (~60-fold) in total mutation rate of thyP3 in both co-directional and head-on orientation compared to wild-type. The mutation rates of the wild-type strains are presented before in Fig. 1d. g, Mismatch repair mutant shows a drastic ~1000 fold increase in mutation rate of T→C substitution hotspots within the coding sequence of head-on thyP3, indicating that mismatch repair corrects T→C substitution within coding sequence. h, Deletion of adeC gene encoding adenine deaminase modestly reduces the mutation rate of T_-7→C_-7 substitution in both co-directional and head-on orientation compared to wild-type. For f–h mean±s.e.m of n≥3 experiments is shown. (*P<0.05; ***P<0.001; Student’s t-test).

Figure 1. Transcription directionality affects spontaneous mutation rates and spectra in B. subtilis

a, thyP3 gene with an IPTG-inducible promoter is integrated into the chromosome either co-directionally or head-on to replication. Purple arrow: replication direction, oriC: replication origin, terC: replication terminus. b, Modified fluctuation test to measure the rate of spontaneous mutations conferring trimethoprim resistance. c, Distribution of mutants: number of mutants per culture (r) plotted against proportion of cultures with ≥r mutants (P(r)). d, The mutation rates in co-directional and head-on thyP3 (subdivided by mutation spectra) when transcription is induced with IPTG. Mutation rates are expressed as mean±s.e.m here and all figures.

Figure 2. Distributions of indels (insertions/deletions) are strongly dependent on transcription directionality and strength

a–d, Positional distribution of indels of ≥3 bp in co-directional and head-on thyP3 under induced (+IPTG) and un-induced (−IPTG) conditions. Each bar represents an insertion (black) or deletion (red). e, The rates of ≥3 bp indels at the promoter. f, The rates of ≥3 bp indels within 5′ (1–420 bp) and 3′ (421–840 bp) of coding region. Rates of insertions and deletions are plotted separately in Extended Data Fig. 6d. g, Model illustrating the mechanism of generation of indels in the vicinity of collision site in co-directional and head-on orientations, via fork slippage (shown here), template switch or fork reversal (Extended Data Fig. 6e).

Figure 3. Head-on transcription induces base substitutions at the promoter

a–d, Positional distribution of base substitutions in co-directional and head-on thyP3 under induced (+IPTG) and un-induced (−IPTG) conditions. Each dot records a base substitution mapped in 50 nt window. e, Promoter base substitution rate is strongly increased in head-on orientation upon IPTG-induction. f-, Distribution of mean nucleotide substitutions per site of promoters, each estimated pairwise among Bacillus strains. Lagging strand promoters (n=32) show increased substitutions than leading strand promoters (n=147). Nucleotide substitutions are comparable between promoters bound and not bound by transcriptional repressors (Extended Data Fig. 8c). Central mark of box-plot represents median, edges are 25^th and 75^th centiles, notches are 95% CI of median, and whiskers represent extreme data points within range. (n.s-not significant; *P<0.05, **P<0.01, ***P<0.0001; Student’s t-test, Mann-Whitney U test).

Figure 4. Promoter T_-7→C_-7 is a mutation hotspot generated via deamination

a, The consensus −10 element of B. subtilis SigA-dependent promoters (n=358). The strongly conserved T_-7 is frequently mutated to a C. b, Fitness of head-on T_-7→C_-7 mutant relative to head-on wild-type thyP3 cells under induced transcription (mean±s.d). c, Mutation rate of T_-7→C_-7 in yqjH mutant (error-prone polymerase PolIV). d, Model illustrating the mechanism of generation of T_-7→C_-7. During transcription initiation, the −10 element is single-stranded, creating solvent accessibility for A_-7 on the template strand (TS), allowing it to be deaminated to hypoxanthine (HX). HX basepairs with C during replication, resulting in T_-7→C_-7. e, T_-7→C_-7 frequencies in head-on thyP3 upon nitrous acid treatment of wild-type and ΔyxlJ (hypoxanthine-DNA glycosylase) strains. (*-P<0.05; Student’s t-test).