br Results To understand how Rad functions to maintain
Results To understand how Rad53 functions to maintain replisome integrity, we first analyzed newly synthesized leading- and lagging-strand DNA in wild-type (WT) and rad53-1 mutant cells using BrdU-IP-ssSeq, a method that detects synthesis of both leading and lagging strands (Yu et al., 2014). Briefly, yeast cells were synchronized in G1 phase using α factor and released into fresh medium containing the nucleotide analog BrdU and 0.2 M hydroxyurea (HU) (Figure 1A). HU depletes dNTP pools through inhibition of RNR and activates the DNA replication checkpoint. However, HU has no apparent effect on the initiation of DNA replication from early replication origins. Newly synthesized DNA marked by BrdU was immuno-precipitated using anti-BrdU auda mg and subjected to strand-specific sequencing (BrdU-IP-ssSeq). Sequence reads were mapped to both Watson and Crick strands of the yeast reference genome (Figure 1B). Inspection of BrdU-IP-ssSeq peaks at early replication origins in WT cells revealed that these peaks were largely symmetrical around replication origins (Figure 1B). These results suggest that DNA synthesis progressed bi-directionally from the origin for similar distances on both the leading and lagging strand, consistent with the idea that synthesis of leading and lagging strand is coupled. To analyze BrdU-IP-ssSeq quantitatively, we used a sliding window of 200 bp and calculated the average log2 ratio of sequence reads of Watson over Crick strands surrounding 134 early replication origins fired in the presence of HU. We observed that BrdU-IP-ssSeq peaks exhibited a small but consistent bias toward the leading strand (Figure 1C and Figure S1). The leading-strand bias indicates that lagging-strand synthesis is slower relative to leading-strand synthesis in WT cells. Based on the analysis of seven independent BrdU-IP-ssSeq datasets in WT cells (Figures S1A and S1B), nascent lagging-strand DNA length was 359 nt shorter than nascent leading strand, with the average 5.32 Kb BrdU peak length of 134 origins. This may be slightly underestimated, because there are slightly more Ts where BrdU is incorporated at the lagging strand (226,146 Ts) than the leading strand (220,433 Ts) in the replicated regions. This result suggests that a short stretch of single-stranded lagging template is likely exposed in WT cells (see Figure 2). Together, these results are consistent with the idea that synthesis of leading and lagging strands is coupled in WT cells, with a short ssDNA gap on the lagging-strand template (Figure 1D). BrdU-IP-ssSeq revealed that late replication origins fired in the presence of HU in rad53-1 mutant cells (Figure 1B), consistent with published studies (Santocanale and Diffley, 1998). Remarkably, BrdU-IP-ssSeq peaks in rad53-1 cells were asymmetrically distributed and showed a strong bias toward lagging strands at both early and late replication origins, with a stronger bias at late replication origins (Figures 1B, 1E, and 1F). This strong lagging-strand bias contrasts the small leading-strand bias of BrdU-IP-ssSeq peaks in WT cells. Importantly, the strong lagging-strand bias was not detected in rad53-1 mutant cells without HU treatment (Figure S1C), suggesting that compromised Rad53 checkpoint function under replication stress is responsible for the observed strong lagging-strand bias. The dramatic lagging-strand bias suggests that lagging-strand synthesis progresses further than leading-strand synthesis at forks in rad53-1 mutant cells under replication stress (Figure 1G). We calculated that lagging-strand synthesis exceeds leading-strand synthesis by 1.52 Kb with an average of 5.44 Kb replicated DNA at the 134 early replication origins and by 1 Kb with an average of 2.35 Kb replicated DNA at 176 late replication origins in rad53-1 mutant cells under our experimental conditions. These observations predict the presence of long stretches of single-stranded leading-strand template at HU-stalled forks in rad53-1 mutant cells (Figure 1G). To test this idea, we analyzed the distribution of RPA, the major ssDNA binding protein in eukaryotic cells, at replication forks using RPA ChIP-ssSeq (Figure 2A). In this method, both template and newly synthesized DNA cross-linked to RPA are being sequenced. We have shown that ChIP-ssSeq peaks of Rfa1, a subunit of RPA complex, exhibited a (+) strand bias based on calculation of the average log2 ratio of Watson over Crick sequence reads surrounding Rfa1 ChIP-ssSeq peaks (Yu et al., 2017) (Figures 2B and 2C), indicating that more RPA binds to the lagging-strand template than to the leading-strand template at early origins in WT cells under replication stress. These results are consistent with the observation that there is excess single-stranded lagging-strand template based on BrdU-IP-ssSeq analysis (Figure 1). In striking contrast, Rfa1 ChIP-ssSeq peaks in rad53-1 mutant cells show a strong (−) strand-biased distribution at all replication origins in rad53-1 cells (Figures 2B, 2C, and 2F). The (−) strand bias indicates that more RPA molecules bind to the leading-strand template than to the lagging-strand template under these conditions. Supporting this interpretation, Rfa1 eSPAN (enrichment and sequencing protein-associated nascent DNA) (Yu et al., 2014), which detects how RPA is cross-linked to nascent DNA at DNA replication forks, revealed that more RPA bound to newly synthesized lagging-strand DNA than leading-strand DNA under the same conditions in rad53-1 mutant cells (Figure 2D; see Figure 2 of Yu et al., 2017 for detailed explanation of the apparent opposite bias pattern of Rfa1 eSPAN and ChIP-ssSeq peaks that supports the same conclusion on the association of RPA with DNA). Furthermore, RPA also bound long stretches of single-stranded leading-strand template in rad53-1 mutant cells without BrdU incorporation into replicating DNA (Figure S1D), ruling out any potential artifacts due to BrdU incorporation on the detection of long stretches of ssDNA in rad53-1 mutant cells using either BrdU-IP-ssSeq or Rfa1 ChIP-ssSeq. Taken together, these results indicate that long stretches of single-stranded DNA coated with RPA on the leading-strand template are generated in rad53-1 cells under replication stress (Figure 2F).