Izumi Miyabe

Mapping the usage of replicative DNA polymerases has previously proved to be technically challenging. By exploiting mutant polymerases that incorporate ribonucleotides into the DNA with a significantly higher proficiency than their wild-type counterparts, we and others have developed methods that can identify what proportion of each DNA strand (i.e., the Watson and Crick strands) is replicated by a specific DNA polymerase. The incorporation of excess ribonucleotides by a mutated polymerase effectively marks, in each individual cells, the DNA strand that is replicated by that specific mutated polymerase. Changes to DNA polymerase usage can be examined at specific loci by Southern blot analysis while a global analysis of polymerase usage can be achieved by applying next-generation sequencing. This genome-wide data also provides a direct measure of replication origin efficiency and can be used to indirectly calculate replication timing.

PCNA ubiquitylation on lysine 164 is required for DNA damage tolerance. In many organisms PCNA is also ubiquitylated in unchallenged S phase but the significance of this has not been established. Using Schizosaccharomyces pombe, we demonstrate that lysine 164 ubiquitylation of PCNA contributes to efficient DNA replication in the absence of DNA damage. Loss of PCNA ubiquitylation manifests most strongly at late replicating regions and increases the frequency of replication gaps. We show that PCNA ubiquitylation increases the proportion of chromatin associated PCNA and the co-immunoprecipitation of Polymerase δ with PCNA during unperturbed replication and propose that ubiquitylation acts to prolong the chromatin association of these replication proteins to allow the efficient completion of Okazaki fragment synthesis by mediating gap filling. PCNA is a homotrimeric complex that clamps around the DNA to provide a sliding platform for DNA polymerases and other replication and repair enzymes. The covalent modification of PCNA by ubiquitin on lysine reside 164 has been extensively studied in the context of DNA repair: it is required to mediate the bypass of damaged template bases during DNA replication. Previous work has shown that PCNA is modified by ubiquitin during normal S phase in the absence of DNA damage, but the significance of this modification has not been explored. Here we show that, in addition to regulating bypass of damaged bases, lysine 164 ubiquitylation plays a role in ensuring the completion of unperturbed DNA replication.

Pu-seq (polymerase usage sequencing) is an approach for identifying ribonucleotides by high-throughput sequencing to allow the mapping of replicative polymerase usage genome wide. This has been used to define origin efficiencies and replication timing in yeast. Ribonucleotides are frequently misincorporated into DNA during replication, and they are rapidly repaired by ribonucleotide excision repair (RER). Although ribonucleotides in template DNA perturb replicative polymerases and can be considered as DNA damage, they also serve positive biological functions, including directing the orientation of mismatch repair. Here we describe a method for ribonucleotide identification by high-throughput sequencing that allows mapping of the location of ribonucleotides across the genome. When combined with specific mutations in the replicative polymerases that incorporate ribonucleotides at elevated frequencies, our ribonucleotide identification method was adapted to map polymerase usage across the genome. Polymerase usage sequencing (Pu-seq) has been used to define, in unprecedented detail, replication dynamics in yeasts. Although other methods that examine replication dynamics provide direct measures of replication timing and indirect estimates of origin efficiency, Pu-seq directly ascertains origin efficiency. The Pu-seq protocol can be completed in 12–14 d.

New analyses in S. pombe show that both the leading and lagging strands are replicated by DNA polymerase δ when stalled replication forks use homologous recombination to restart semiconservative DNA synthesis. To maintain genetic stability, DNA must be replicated only once per cell cycle, and replication must be completed even when individual replication forks are inactivated. Because fork inactivation is common, passive convergence of an adjacent fork is insufficient to rescue all inactive forks. Thus, eukaryotic cells have evolved homologous recombination–dependent mechanisms to restart persistent inactive forks. Completing DNA synthesis via homologous recombination–restarted replication (HoRReR) ensures cell survival, but at a cost. One such cost is increased mutagenesis because HoRReR is more error prone than canonical replication. This increased error rate implies the HoRReR mechanism is distinct from that of a canonical fork. Here we demonstrate, in Schizosaccharomyces pombe, that a DNA sequence duplicated by HoRReR during S phase is replicated semiconservatively, but both the leading and lagging strands are synthesized by DNA polymerase δ.

Genome-wide DNA polymerase usage maps determined in fission yeast, using a new sequencing strategy based on ribonucleotide misincorporation, track the division of labor between replicative polymerases and reveal locations and efficiencies of replication origins. Three eukaryotic DNA polymerases are essential for genome replication. Polymerase (Pol) α–primase initiates each synthesis event and is rapidly replaced by processive DNA polymerases: Polɛ replicates the leading strand, whereas Polδ performs lagging-strand synthesis. However, it is not known whether this division of labor is maintained across the whole genome or how uniform it is within single replicons. Using Schizosaccharomyces pombe, we have developed a polymerase usage sequencing (Pu-seq) strategy to map polymerase usage genome wide. Pu-seq provides direct replication-origin location and efficiency data and indirect estimates of replication timing. We confirm that the division of labor is broadly maintained across an entire genome. However, our data suggest a subtle variability in the usage of the two polymerases within individual replicons. We propose that this results from occasional leading-strand initiation by Polδ followed by exchange for Polɛ.

Genetic instability, a hallmark of cancer, can occur when the replication machinery encounters a barrier. The intra-S-phase checkpoint maintains stalled replication forks in a replication-competent configuration by phosphorylating replisome components and DNA repair proteins to prevent forks from catastrophically collapsing. Here, we report a novel function of the core Schizosaccharomyces pombe checkpoint sensor kinase, Rad3 (an ATR orthologue), that is independent of Chk1 and Cds1 (a CHK2 orthologue); Rad3^ATR regulates the association of recombination factors with collapsed forks, thus limiting their genetic instability. We further reveal antagonistic roles for Rad3^ATR and the 9-1-1 clamp – Rad3^ATR restrains MRN- and Exo1-dependent resection, whereas the 9-1-1 complex promotes Exo1 activity. Interestingly, the MRN complex, but not its nuclease activity, promotes resection and the subsequent association of recombination factors at collapsed forks. The biological significance of this regulation is revealed by the observation that Rad3^ATR prevents Exo1-dependent genome instability upstream of a collapsed fork without affecting the efficiency of recombination-mediated replication restart. We propose that the interplay between Rad3^ATR and the 9-1-1 clamp functions to fine-tune the balance between the need for the recovery of replication through recombination and the risk of increased genome instability.

A new mechanism of chromosomal rearrangement is identified through the observation that broken or collapsed DNA replication forks restarted by homologous recombination have a high propensity for U-turns at short inverted repeats; the error-prone nature of this mechanism is suggested to contribute to gross chromosomal rearrangements and copy-number variations present in cancer and other genomic disorders. Stress that causes broken or collapsed DNA replication forks can induce copy-number variations (CNVs) and gross chromosomal rearrangements (GCRs), both of which are commonly found in cancer cells. Antony Carr and colleagues find that a collapsed fork that has been restarted by homologous recombination is more error prone than a fork that initiates at origins of replication. These restarted forks frequently perform U-turns at short inverted repeats, a possible cause of CNVs and GCRs. The authors suggest that the error-prone nature of restarted forks could be a source of cancer-causing gene abnormalities and spontaneous genetic disorders in the absence of a double-strand break. Impediments to DNA replication are known to induce gross chromosomal rearrangements (GCRs) and copy-number variations (CNVs). GCRs and CNVs underlie human genomic disorders1 and are a feature of cancer2. During cancer development, environmental factors and oncogene-driven proliferation promote replication stress. Resulting GCRs and CNVs are proposed to contribute to cancer development and therapy resistance3. When stress arrests replication, the replisome remains associated with the fork DNA (stalled fork) and is protected by the inter-S-phase checkpoint. Stalled forks efficiently resume when the stress is relieved. However, if the polymerases dissociate from the fork (fork collapse) or the fork structure breaks (broken fork), replication restart can proceed either by homologous recombination or microhomology-primed re-initiation4,5. Here we ascertain the consequences of replication with a fork restarted by homologous recombination in fission yeast. We identify a new mechanism of chromosomal rearrangement through the observation that recombination-restarted forks have a considerably high propensity to execute a U-turn at small inverted repeats (up to 1 in 40 replication events). We propose that the error-prone nature of restarted forks contributes to the generation of GCRs and gene amplification in cancer, and to non-recurrent CNVs in genomic disorders.

DNA damage checkpoint activation can be subdivided in two steps: initial activation and signal amplification. The events distinguishing these two phases and their genetic determinants remain obscure. TopBP1, a mediator protein containing multiple BRCT domains, binds to and activates the ATR/ATRIP complex through its ATR-Activation Domain (AAD). We show that Schizosaccharomyces pombe Rad4^TopBP1 AAD–defective strains are DNA damage sensitive during G1/S-phase, but not during G2. Using lacO-LacI tethering, we developed a DNA damage–independent assay for checkpoint activation that is Rad4^TopBP1 AAD–dependent. In this assay, checkpoint activation requires histone H2A phosphorylation, the interaction between TopBP1 and the 9-1-1 complex, and is mediated by the phospho-binding activity of Crb2^53BP1. Consistent with a model where Rad4^TopBP1 AAD–dependent checkpoint activation is ssDNA/RPA–independent and functions to amplify otherwise weak checkpoint signals, we demonstrate that the Rad4^TopBP1 AAD is important for Chk1 phosphorylation when resection is limited in G2 by ablation of the resecting nuclease, Exo1. We also show that the Rad4^TopBP1 AAD acts additively with a Rad9 AAD in G1/S phase but not G2. We propose that AAD–dependent Rad3^ATR checkpoint amplification is particularly important when DNA resection is limiting. In S. pombe, this manifests in G1/S phase and relies on protein–chromatin interactions. DNA structure–dependent checkpoint activation and the amplification of checkpoint signals are carefully modulated to allow the checkpoint kinases to delay mitosis and regulate DNA metabolism. While much work has gone into understanding how this checkpoint functions, the mechanism by which the checkpoint signal is amplified is less clear. We have characterised a conserved domain in the Schizosaccharomyces pombe TopBP1 homolog, Rad4^TopBP1 (also known as Cut5) that is capable of activating the ATR homolog Rad3^ATR. We demonstrate that this domain is not required for initial checkpoint activation, but functions to amplify the checkpoint signal, likely when the presence of single-stranded DNA is limiting. Our data suggest that the function of the Rad4^TopBP1 ATR-Activation Domain (AAD) is mediated by interactions between checkpoint proteins and phosphorylated histone H2A, which is itself promoted by Rad3^ATR. We propose that the resulting amplification of the checkpoint signal is particularly important in G1-S phase, when resection is limited.

Coordinated replication of eukaryotic genomes is intrinsically asymmetric, with continuous leading strand synthesis preceding discontinuous lagging strand synthesis. Here we provide two types of evidence indicating that, in fission yeast, these two biosynthetic tasks are performed by two different replicases. First, in Schizosaccharomyces pombe strains encoding a polδ-L591M mutator allele, base substitutions in reporter genes placed in opposite orientations relative to a well-characterized replication origin are strand-specific and distributed in patterns implying that Polδ is primarily involved in lagging strand replication. Second, in strains encoding a polε-M630F allele and lacking the ability to repair rNMPs in DNA due to a defect in RNase H2, rNMPs are selectively observed in nascent leading strand DNA. The latter observation demonstrates that abundant rNMP incorporation during replication can be tolerated and that they are normally removed in an RNase H2-dependent manner. This provides strong physical evidence that Polε is the primary leading strand replicase. Collectively, these data and earlier results in budding yeast indicate that the major roles of Polδ and Polε at the eukaryotic replication fork are evolutionarily conserved. It is important to understand the architecture of the DNA replication machinery and whether this is common to all organisms. Recent work in Saccharomyces cerevisiae has genetically assigned specific DNA polymerases to leading and lagging strand DNA synthesis, Polε and Polε respectively. In this manuscript, we use a similar genetic assay to demonstrate that, in the highly evolutionarily diverged yeast Schizosaccharomyces pombe, Polδ is similarly responsible for lagging strand synthesis. Importantly, we establish a novel physical assay, the incorporation of rNMPs into newly replicated DNA, which demonstrates that Polε is responsible for leading strand synthesis and does not contribute significantly to lagging strand replication. These data strongly support and consolidate the interpretation of previous genetic data and suggest that the division of labour between polymerases is conserved through evolution.