High fidelity DNA ligation prevents single base insertions in the yeast genome

Abstract

Finalization of eukaryotic nuclear DNA replication relies on DNA ligase 1 (LIG1) to seal DNA nicks generated during Okazaki Fragment Maturation (OFM). Using a mutational reporter in Saccharomyces cerevisiae, we previously showed that mutation of the high-fidelity magnesium binding site of LIG1^Cdc9 strongly increases the rate of single-base insertions. Here we show that this rate is increased across the nuclear genome, that it is synergistically increased by concomitant loss of DNA mismatch repair (MMR), and that the additions occur in highly specific sequence contexts. These discoveries are all consistent with incorporation of an extra base into the nascent lagging DNA strand that can be corrected by MMR following mutagenic ligation by the Cdc9-EEAA variant. There is a strong preference for insertion of either dGTP or dTTP into 3-5 base pair mononucleotide sequences with stringent flanking nucleotide requirements. The results reveal unique LIG1^Cdc9-dependent mutational motifs where high fidelity DNA ligation of a subset of OFs is critical for preventing mutagenesis across the genome.

Fig. 1. High-fidelity DNA ligation is critical for avoiding +1 mutations across the genome.

a The mutation rate ratios of the various mutation classes were calculated as the rate in a strain expressing Cdc9-EEAA divided by the rate in a strain expressing wild-type CDC9 (+/− MMR). The dotted vertical line indicates a ratio of 1. b The specificity of 1 bp insertions was determined for +G•C or +A•T insertions. The rate ratios were calculated as in (a). Source data are provided as a Source Data file.

Fig. 2. Low-fidelity ligation selectively increases insertion rates in short homopolymers.

a Insertion rates in the wild-type (wt) strain (CDC9⁺). Insertion rates (circles) in G•C (orange) and A•T (red) homopolymers are shown with their respective detection limits (dashed lines). Detection limits are the rates that would be implied by one observed insertion with a given base content and homopolymer length. The minimum detectable rate for a given context is the rate that would be calculated if one mutation were to be observed in one instance of that content. b As per (a), but for msh2Δ. c As per (a), but for cdc9-EEAA. d As per (a), but for cdc9-EEAA msh2Δ. e The sequence context-dependence of mutations caused by expression of the cdc9-EEAA variant in the presence of MMR. The ratio reported depends on the number of insertions observed in the cdc9-EEAA (N) and wt (CDC9⁺) (M) strains. If N > 0 and M > 0, the ratio uses both rates (cdc9-EEAA/CDC9⁺). If M = 0, the ratio is a lower bound estimate using the CDC9⁺ detection limit as the CDC9⁺ rate. If N = 0, the ratio is reported as 0. If there are no homopolymers of a given type and length, no ratio is reported. f As per (e), but for cdc9-EEAA msh2Δ. g Heatmap of G•C insertions in 3 bp homopolymers in the cdc9-EEAA msh2Δ strain across the 16 S. cerevisiae chromosomes (10 kbp bins). Unmapped bins are colored white. All others are colored by the number of observed mutations. Bins with no mutations are gray, bins with one mutation are blue, bins approaching the significance threshold are red, and intermediate counts transition between blue and red. Bins with counts exceeding the significance threshold have a black border (Šidák correction; each bin counts as a hypothesis tested; family-wise error rate = 0.05). Bins containing centromeres are crosshatched black. Source data are provided as a Source Data file.

Fig. 3. Lagging strand G insertions are favored over C insertions in the low-fidelity DNA ligase mutant lacking MMR.

All measures are from the cdc9-EEAA msh2Δ strain. a A diagram of G insertions preferentially made during synthesis of nascent lagging strands. Thick lines denote parental (black) and nascent strands (arrows; leading in blue; lagging in green). b Insertion loops are converted into mutations when new DNA strands (gray) are synthesized in the subsequent round of replication. After sequencing, mutations are reported in the top strand reference frame. A bias for lagging strand G insertion loops would thus cause C insertions to be preferentially called to the right of origins and G insertions to be preferentially called to the left. The opposite would be true if C insertions were preferred in the nascent lagging strand. c When the fraction of G (orange) and C (blue) insertions is mapped between all adjacent origins, the resulting X-pattern indicates a preference for G insertion in the nascent lagging strand. d A better proxy for bottom lagging strand synthesis is the fraction of the bottom strand synthesized by Pols α and δ (f_BSαδ; a from ribonucleotide mapping). As f_BSαδ approaches 1, the top strand G insertion rate decreases, and the top strand C insertion rate increases. e The top strand G insertion fraction (f_TS+G) decreases linearly as f_BSαδ increases (f_TS+G = −0.751 × f_BSαδ + 0.881; R² = 0.994). The y-intercept suggests a 7.4-fold preference for G insertions over C insertions during lagging strand synthesis in the cdc9-EEAA msh2Δ strain. Source data are provided as a Source Data file.

Fig. 4. In the low-fidelity cdc9-EEAA mutant, lagging strand G insertions have a template motif of 5′-C_nA-3′ that disappears, alongside bias, as homopolymer length increases.

a The preference for G insertions (orange) over C insertions (blue) during lagging strand synthesis is decreased in the presence of MMR (2.2x for cdc9-EEAA vs. 7.4x in cdc9-EEAA msh2Δ; see Fig. 3e). This suggests that MMR must have an opposite bias, preferentially repairing looped out G bases. b–d In the cdc9-EEAA msh2Δ strain, as homopolymer length increases from 3 to 5 bp, both the preference for G insertions and the correlations between strandedness and insertion bias both decrease (bias from 19x to 5.6x and R² from 0.969 to 0.748; Supplementary Table 4). e–h Sequence logos indicate preferred motifs for G insertions. e A sequence logo for all G insertions in the cdc9-EEAA stain. Using origin proximity to estimate strandedness, 98% of inferred lagging strand G insertions in 3 bp C-runs (n = 64) are found in runs that are followed by a template A, and 89% are then followed by template pyrimidines (C or T). These values drop to 80% and 68% in 4 bp template C-runs (n = 25; Supplementary Fig. 3). f Drawing from only the first and last 25% of each inter-origin space (see Fig. 4c), a nearly identical pattern was found in the cdc9-EEAA msh2Δ strain (98% A followed by 93% pyrimidine; n = 1488). g, h This motif disappeared with increasing C-run length (n = 697 and 347, respectively). i An illustration of the inferred motifs in the context of replication. Green arrows denote the direction of primer strand synthesis. Source data are provided as a Source Data file.

Fig. 5. In the low-fidelity cdc9-EEAA mutant, lagging strand T insertions have a template motif of 5’-CA_m-3’ that disappears, alongside bias, as homopolymer length increases.

a In the presence of MMR, the preference for T insertions (red) over A insertions (green) during lagging strand synthesis is greater than the preference for G relative to C insertions (3.9x for T vs. 2.2x for G; see Fig. 4a). b–d In the cdc9-EEAA msh2Δ strain, as homopolymer length increases from 4 to 6 bp, both the preference for T insertions and the correlations between strandedness and insertion bias both decrease (bias from 8.3x to 1.3x and R² from 0.948 to 0.045). e–h Sequence logos indicate preferred motifs for T insertions. Green arrows denote the direction of primer strand synthesis. e A sequence logo illustrates the lack of a motif for T insertions in the cdc9-EEAA strain. f There is a notable motif for lagging strand T insertions across from 4 bp template A-runs in the cdc9-EEAA msh2Δ strain. Drawing from only the first and last 25% of each inter-origin space (see Fig. 5c), 63% of inferred lagging strand T insertions in 4 bp A-runs (n = 64) are found in runs that are immediately preceded by a template C (n = 599). g, h This motif disappeared with increasing A-run length (n = 207 and 129, respectively). i An illustration of the inferred motifs in the context of replication. Source data are provided as a Source Data file.

Fig. 6. Loss of high-fidelity ligation causes 1 bp lagging strand insertions in contexts with a super-motif of 5′-C_nA_m-3′.

Proposed mechanisms for the creation of single-base insertions due to mutagenic ligation during lagging strand OFM. DNA strands are shown as lines with arrowheads indicating the direction of synthesis. Bold letters indicate DNA bases involved with the motifs for single-base insertions due to the low-fidelity ligase mutant. Parental DNA strands are shown in black with letters colored to match motifs from Figs. 4, 5, nascent lagging strands in green with black letters, and terminal phosphates as gray circles.