Processing of eukaryotic Okazaki fragments by redundant nucleases can be uncoupled from ongoing DNA replication in vivo

Abstract Prior to ligation, each Okazaki fragment synthesized on the lagging strand in eukaryotes must be nucleolytically processed. Nuclease cleavage takes place in the context of 5′ flap structures generated via strand-displacement synthesis by DNA polymerase delta. At least three DNA nucleases: Rad27 (Fen1), Dna2 and Exo1, have been implicated in processing Okazaki fragment flaps. However, neither the contributions of individual nucleases to lagging-strand synthesis nor the structure of the DNA intermediates formed in their absence have been fully defined in vivo. By conditionally depleting lagging-strand nucleases and directly analyzing Okazaki fragments synthesized in vivo in Saccharomyces cerevisiae, we conduct a systematic evaluation of the impact of Rad27, Dna2 and Exo1 on lagging-strand synthesis. We find that Rad27 processes the majority of lagging-strand flaps, with a significant additional contribution from Exo1 but not from Dna2. When nuclease cleavage is impaired, we observe a reduction in strand-displacement synthesis as opposed to the widespread generation of long Okazaki fragment 5′ flaps, as predicted by some models. Further, using cell cycle-restricted constructs, we demonstrate that both the nucleolytic processing and the ligation of Okazaki fragments can be uncoupled from DNA replication and delayed until after synthesis of the majority of the genome is complete.

fragment flaps. However, neither the contributions of individual nucleases to lagging-strand 23 synthesis nor the structure of the DNA intermediates formed in their absence have been clearly 24 defined in vivo. By conditionally depleting lagging-strand nucleases and directly analyzing 25 Okazaki fragments synthesized in vivo in S. cerevisiae, we conduct a systematic evaluation of 26 the impact of Rad27, Dna2 and Exo1 on lagging-strand synthesis. We find that Rad27 27 processes the majority of lagging-strand flaps, with a significant additional contribution from 28 Exo1 but not from Dna2. When nuclease cleavage is impaired, we observe a reduction in 29 strand-displacement synthesis as opposed to the widespread generation of long Okazaki 30 fragment 5' flaps, as predicted by some models. Further, using cell cycle-restricted constructs, 31 we demonstrate that both the nucleolytic processing and the ligation of Okazaki fragments can 32 be uncoupled from DNA replication and delayed until after synthesis of the majority of the 33 genome is complete. 34 35 INTRODUCTION maintain a ligatable nick that persists until it is sealed by DNA ligase I, encoded by the CDC9 48 gene in S. cerevisiae (Johnston and Nasmyth, 1978). Nuclease cleavage during Okazaki 49 fragment biogenesis represents an extremely abundant DNA transaction -one that must occur 50 tens of thousands of times during each S-phase in S. cerevisiae and millions of times per 51 human cell division. 52

53
The nucleases Rad27 and Dna2 have been proposed to cleave the majority of flaps during 54 Okazaki fragment maturation. Genetic and biochemical work has given rise to a 'two-nuclease' 55 model (Balakrishnan et al., 2010;Burgers, 2009). According to this model, RNase H2 first 56 removes most of the RNA primer (Qiu et al., 1999). Subsequently, iterative extension by Pol d is 57 followed by immediate cleavage of short DNA flaps by Rad27. If Pol d extension outpaces 58 Rad27 cleavage, Dna2 is required to process the resulting long flap. Rad27 and Dna2 show 59 distinct substrate requirements in vitro. Rad27 readily cleaves short flaps; longer flaps are 60 competent to bind RPA and thereby become refractory to Rad27 cleavage (Rossi and Bambara, 61 2006). Long, RPA-coated flaps are optimal substrates for Dna2 in vitro. Dna2 cleaves to leave a 62 short flap such that RPA dissociates and Rad27 can cleave (Ayyagari et al., 2003;Gloor et al., 63 2012). However, recent reports indicate that Dna2 activity is sufficient to process 5' flaps into 64 ligatable nicks in vitro (Levikova and Cejka, 2015). Genetic data suggest additional redundancy 65 in lagging-strand processing, and point to the likely involvement of Exo1 as a third Okazaki 66 nuclease. In vivo in S. cerevisiae, neither RAD27 nor DNA2 is strictly essential for replication or 67 viability (Budd et al., 2011), and the temperature-sensitive phenotypes of both rad27D and 68  Spot tests of indicated strains on YPD ± 2 υg/ml rapamycin at 30˚C or 37˚C, as indicated, for single and double mutant strains analyzed in this study. Addition of rapamycin depletes FRB-tagged proteins from the nucleus. B. End-labeled Okazaki fragments obtained after 1h rapamycin treatment to deplete DNA ligase I and FRB-tagged nucleases. Where indicated, DNA was treated with T4 DNA ligase after purification but before labeling to assess the proportion of Okazaki fragments poised for ligation. WT denotes an otherwise wild-type CDC9-FRB strain. C. End-labeled Okazaki fragments, as in (B), for WT, RAD27-FRB and RAD27-FRB DNA2-FRB; exo1Δ strains. DNA was treated or mock-treated with Taq polymerase and DNA ligase after purification, as indicated. In nucleosome-free regions, Okazaki fragment termini are normally enriched immediately 177 upstream of binding sites for the transcription factors Abf1, Reb1 and Rap1. These transcription 178 factors re-associate quickly following DNA replication, and appear to represent 'hard' barriers to 179 Pol d in contrast to the 'soft' barrier activity of nucleosomes (Smith and Whitehouse, 2012). We 180 analyzed the distribution of Okazaki fragment termini around these transcription factor binding 181 sites, as a means to investigate strand-displacement and processing on non-nucleosomal 182 templates (Fig. 3). Data were normalized to the median signal within the range in order to 183 assess the propensity of Pol d to terminate specifically at transcription-factor binding sites as 184 opposed to nearby sequences. 185

186
In both wild-type and single nuclease depletions, we observed a prominent peak of Okazaki 187 fragment 5' and 3' termini immediately upstream of a meta-binding site for Abf1/Reb1/Rap1 188 ( Fig. 3 A&B). However, the RAD27-FRB exo1D strain showed significantly reduced termination 189 on the replication-fork proximal edge of the transcription-factor binding sites relative to the wild-190 type strain. The RAD27-FRB DNA2-FRB; exo1D strain generated slightly fewer fragments 191 terminating at transcription-factor binding sites than the RAD27-FRB exo1D double mutant, 192 suggesting that the decreased strand-displacement on non-nucleosomal DNA in the absence of 193 both Rad27 and Exo1 can be further reduced by removal of Dna2. These data are consistent 194 with a globally similar distribution of nuclease activity on non-nucleosomal DNA to that observed 195 within nucleosomes -i.e. major redundant roles for Rad27 and Exo1 and a limited role for 196 Dna2. However, the distribution of Okazaki fragment termini around TF binding sites in both 197 RAD27-FRB and exo1D strains closely resembles that of a wild-type strain (Fig. 3A&B). 198 Therefore, our data suggest more extensive redundancy of Rad27 and Exo1 outside 199 nucleosomes than on nucleosomal templates. sufficient to support viability (Fig 4A). Consistent with delayed ligation of Okazaki fragments 215 when CDC9 expression is restricted, the CLB2-CDC9 strain transiently accumulated Okazaki 216 fragments following release from G1 arrest which disappeared when ligation was enabled by 217 CLB2-CDC9 expression in late S/G2 (Fig. 4B). Sporulation of a CLB2-RAD27/RAD27 CLB2-218 EXO1/EXO1 strain also generated four viable spores per tetrad, although CLB2-RAD27 CLB2-219 EXO1 double mutant haploids grew slowly (Fig. 4C). CLB2-RAD27 cells phenocopied the 220 temperature-sensitivity and slow growth of rad27D cells, suggesting that these phenotypes arise 221 due to a lack of Rad27 during S-phase (Fig. S2G). The observation that Rad27 and Exo1 222 together process the majority of Okazaki fragments in S. cereivisiae (Figs 2-3), but can maintain 223 viability when expressed only after the bulk of replication has been completed, suggests that 224 Okazaki fragment processing need not occur concurrently with replication. Tetrads from sporulation of a diploid CLB2-CDC9/CDC9 strain were dissected onto YPD and allowed to grow for 72h. B. CLB2-CDC9 or wild-type cells were synchronized using alpha-factor, and released into fresh YPD for the indicated time. DNA was purified and end-labeled as in Fig. 1B. A sample from an asynchronous culture carrying a doxycycline repressible allele of CDC9 is shown for comparison. Repression was for 2.5h as previously described (Smith and Whitehouse, 2012). C&D. Tetrads from sporulation of a diploid strain heterozygous for (C) CLB2-RAD27 and CLB2-EXO1 or (D) CLB2-RAD27 and CLB2-DNA2 were dissected onto YPD and allowed to grow for 72h. E. Tetrads from a diploid strain heterozygous for CLB2-RAD27, CLB2-DNA2 and GAL1-SSB were dissected onto either YPD or YPGal plates, as indicated, and allowed to grow for 72h.
Sporulation of a CLB2-RAD27/RAD27 CLB2-DNA2/DNA2 strain produced a mixture of viable 227 and inviable spores (Fig. 4D). Colonies formed from CLB2-DNA2 spores have an essentially 228 wild-type growth phenotype (Fig. 4D). Thus, the essential contribution of Dna2 to viability can 229 apparently be executed after the majority of the genome has been synthesized. In contrast to 230 CLB2-RAD27 CLB2-EXO1 double mutants, CLB2-RAD27 CLB2-DNA2 haploids are inviable 231 (Fig. 4D). Therefore, although Exo1 apparently contributes to the overall processing of Okazaki 232 fragments to a greater extent than Dna2 (Fig. 2) when grown on YPD to repress SSB expression, but viable when SSB was induced by growth 243 on 2% galactose (Fig. 4E). Thus, expression of a generic single-strand binding protein restores 244 viability when neither Rad27 nor Dna2 is expressed during early or mid S-phase, indicating that 245 endonucleolytic processing of the lagging strand can be deferred until after the majority of 246 replication has been completed. Because we observed that Okazaki fragment ligation can be 247 similarly deferred (Fig 4A-B), we conclude that both steps of lagging-strand processing can be 248 Okazaki fragments are generated even in the absence of all three nucleases (Fig. 1), and that

Cell-cycle synchronization and western blotting
200ml of exponential phase cell cultures at OD 0.3-0.4 was synchronized in G1 using 5µg/ml alpha factor. Cells were released into S-phase and harvested at different time points. Growth was immediately stopped by addition of ice cold water and cells were centrifuged at 4ºC. Cells were resuspended for 5 minutes in 2M lithium acetate at 4ºC, pelleted, resuspended for 5 minutes in 400mM sodium acetate at 4ºC, pelleted again and finally resuspended in 1X Laemmli buffer plus 5% beta-mercaptoethanol. After 5min of boiling at 100ºC, the lysates were centrifuged for 5 minutes at top speed and transferred to new tubes prior to loading on a 10% SDS-Page gel. After migration at 100V, samples were transferred to a PVDF membrane, blocked with 5%milk in TBS-0,1% Tween and probed with a C-Myc antibody (Genscript 346 A00173-100). Loading controls were done by Coomassie staining of gels loaded with identical 347 amounts of sample and run alongside gels for Western blots. Okazaki fragment purification and sequencing was carried out as previously described (Smith 371 and Whitehouse, 2012). Paired-end sequencing (2 × 75 bp) was carried out on an Illumina 372 Next-seq 500 platform. 373 374 Figure S1. Sequencing data replicate comparisons A. Distribution of Okazaki fragment 5' termini around nucleosome dyads, as in Fig. 2A, for a second biological replicate of each strain. B. Distribution of Okazaki fragment 3' termini around nucleosome dyads, as in Fig. 2B, for a second biological replicate of each strain. C. Distribution of Okazaki fragment 5' termini around Abf1/Reb1/Rap1 sites, as in Fig. 3A, for a second biological replicate of each strain. D. Distribution of Okazaki fragment 3' termini around Abf1/Reb1/Rap1 sites, as in Fig. 3B, for a second biological replicate of each strain. 24