Roles for DNA polymerase δ in initiating and terminating leading strand DNA replication

Abstract

Most current evidence indicates that DNA polymerases ε and δ, respectively, perform the bulk of leading and lagging strand replication of the eukaryotic nuclear genome. Given that ribonucleotide and mismatch incorporation rates by these replicases influence somatic and germline patterns of variation, it is important to understand the details and exceptions to this overall division of labor. Using an improved method to map where these replicases incorporate ribonucleotides during replication, here we present evidence that DNA polymerase δ universally participates in initiating leading strand synthesis and that nascent leading strand synthesis switches from Pol ε to Pol δ during replication termination. Ribonucleotide maps from both the budding and fission yeast reveal conservation of these processes. These observations of replisome dynamics provide important insight into the mechanisms of eukaryotic replication and genome maintenance.

Fig. 1

Models of canonical polymerase division of labor and exceptions at replication origins and termination zones. a Replisome components and canonical polymerase division of labor. Red, green, and blue denote Polymerases α, δ, and ε, respectively, or the nascent DNA tracts they synthesize. DNA strands (colored bars) and proteins are not shown to scale. Other replisome components are omitted for simplicity. b A model of replication initiation. Replication initiates with Pol α priming on both strands. On the lagging strand, priming is repeated with Pol α passing the 3′ terminus to Pol δ for Okazaki fragment synthesis. On the leading strand, Pol α passes the 3′ terminus to Pol δ, which then catches the receding helicase complex and passes the 3′ terminus to Pol ε. c A model of replication termination wherein Pol ε disengages from the 3′ terminus and Pol δ assumes responsibility for the remainder of leading strand synthesis

Fig. 2

Polymerase usage across S. cerevisiae Chromosome X indicates deviations from a canonical division of labor that occurs at replication origins and termination zones. Data presented are averages of at least three replicates of strains with wild type polymerase or each of the polymerase variants. Points represent values in 50 bp bins. a, b Curves are 1 kb moving averages. a Fractional strand-specific Pol δ, α, and ε usage (green, red, and blue/yellow, respectively). Noise in RHII-HydEn-seq data push curves slightly outside of the 0–1 range. Steep transitions indicate either active origin positions or low coverage regions (e.g. transposons, telomeres, etc.). b Fractional Pol ε synthesis (f_ε) of top (blue) and bottom (yellow) strands (1.14x linear rescale to 100% maximum). c and d The DDAF (orange points) the measurement of division of labor between Pol ε and Pols δ and α. Green bars represent origin positions, and green diamonds indicate those with established firing times (not inferred herein). Red bars indicate predicted collision positions assuming optimal global fork rates and 100% origin efficiency. c Exemplar DDAF peaks at four early-firing, relatively efficient replication origins. Curves are unsmoothed. d The Chr X DDAF and Monte Carlo simulated fork collision density fit thereto (black curve; 1000 simulations; see the “Methods” section and Fig. S4 for parameters). Note noise around non-unique positions like sub/telomeres and transposons (purple). Orange bars below the horizontal axis indicate inter-origin tracts where simulation and observation deviate. The red diamond indicates an origin where simulated collision peaks are closer than DDAF peaks in both directions, suggesting later firing than expected. Gray backgrounds indicate tracts adjacent to origins with firing times inferred herein, rather than previously measured. e Comparing DDAF peak positions (orange with black border) with replication termination peak positions measured from BrdU ChIP-chip (light/dark purple for left/right terminus) or calculated from Okazaki fragment sequencing (red/pink for replicate A/B) or sort-seq (gray; high confidence, i.e. score > 0.004). f–h Mutation rates as orthogonal confirmation of f_ε calculations. Mutations accumulated over 3840 generations in mismatch repair-deficient S. cerevisiae with mutator Pol ε variant M644G. f Fractions of the S. cerevisiae genome partitioned (bin) by top strand f_ε. g The fraction of G to T substitutions increases linearly with top strand f_ε. The opposite holds for C to A substitutions. h Rates of Pol ε characteristic mutations increase with f_ε. Strandedness was assigned given the preference of M644G Pol ε for making C-dT (template C; incoming dTTP) vs. G-dA mispairs

Fig. 3

DDAF peaks at replication origins and termination zones in S. cerevisiae. a A DDAF heatmap in 50 bp bins for one kilobase on either side of 289 replication origins (green bars in Figs. 1d and S2). Origin motifs (ACS motifs) on the plus (top) strand are grouped as “plus strand ACS” and minus (bottom) strand oriented origins are gouped as “minus strand ACS”. Origins are ranked by efficiency. Blue indicates a high DDAF value (less Pol ε usage), red denotes the opposite. b Averaged DDAF value at inefficient origins (efficiency < 0.5). Orange curves are for plus strand origins and purple curves are for minus strand origins. The number of origins used is indicated. c Averaged DDAF value at efficient origins (efficiency > 0.9). d DDAF peak areas (after baseline subtraction) increase linearly with origin efficiency (R² = 0.986). The negative value at origin efficiency of zero is due to aggressive background subtraction (see the “Methods” section). e As per a but in 1000 bp bins for 20 kb on either side of predicted collision points (red bars in Figs. 1d and S2) for forks proceeding from 259 well-separated adjacent origins (distance ≥ 20 kb). The heatmap is ranked by the efficiency of the lesser ones of the pairs of adjacent origins. On average, origin pairs with a lesser member of both f moderate and g high efficiency have broad DDAF peaks centered at predicted termination zones, both if the data is truncated at the member origins (orange curves; heatmap in Fig. S4) or not (brown dotted curves). h The areas under truncated curves are independent of flanking origin efficiency. Slight negative areas for some inefficient origins suggest that true baselines are somewhat lower

Fig. 4

Exceptions to a canonical division of labor are conserved in S. pombe. Data are an average of at least three replicates for each genotype. a Fractional Pol ε synthesis (f_{Pol ε}) of top (blue) and bottom (yellow) strands. Points represent 50 bp bins. Curves are 2 kb moving averages. b, c The DDAF profile (orange points): the total fraction of synthesis by Pols α and δ across both strands. Green bars represent origin positions. Green diamonds indicate origin efficiencies over 0.4. Red bars indicate predicted collision positions assuming uniform global fork rates, uniform firing times and 100% origin efficiency. Points represent 50 bp bins. Curves are 3 kb moving averages. b DDAF peaks at four relatively efficient replication origins. Curves are 1.5 kb moving averages. c The DDAF across a section of Chromosome 2. Curves are 4 kb moving averages. Non-unique regions are excluded (purple). d A DDAF heatmap in 50 bp bins (3-bin moving average) for 1.2 kb on either side of 283 replication origins (green bars in a). Blue indicates a high DDAF (less Pol ε usage), red denotes the opposite. e Averaged DDAF curve at inefficient (efficiency < 0.5). The number of origins used is indicated below the graph. f Averaged DDAF peak at efficient (>0.7) origins. g DDAF peak areas (after baseline subtraction) increase linearly with origin efficiency (R² = 0.990). h Early firing origin pairs (efficiencies > 0.2, intervening origin efficiency < 0.1, separation > 20 kb) have broad DDAF peaks centered near the inter-origin midpoint (orange curve; truncated at member origins). These peaks resemble collision point-centered S. cerevisiae DDAF peaks (brown dotted curve). i Comparison of tract length inferred from DDAF peak area in S. cerevisiae and S. pombe