Exceptional origin activation revealed by comparative analysis in two laboratory yeast strains

We performed a comparative analysis of replication origin activation by genome-wide single-stranded DNA mapping in two yeast strains challenged by hydroxyurea, an inhibitor of the ribonucleotide reductase. We gained understanding of the impact on origin activation by three factors: S-phase checkpoint control, DNA sequence polymorphisms, and relative positioning of origin and transcription unit. Wild type W303 showed a significant reduction of fork progression accompanied by an elevated level of Rad53 phosphorylation as well as physical presence at origins compared to A364a. Moreover, a rad53K227A mutant in W303 activated more origins, accompanied by global reduction of ssDNA across all origins, compared to A364a. Sequence polymorphism in the consensus motifs of origins plays a minor role in determining strain-specific activity. Finally, we identified a new class of origins only active in checkpoint-proficient cells, which we named “Rad53-dependent origins”. Our study presents a comprehensive list of differentially used origins and provide new insights into the mechanisms of origin activation.

5. The authors should also use to DeOri database. Indeed, ARSseq was curated at DeOri. We have now referenced the database.
6. The authors need to point out that autophosphorylation is inferred rather than directly shown or provide alternative data demonstrating autophosphorylation. We refrained from suggesting that the phosphorylation we observed was due to autophosphorylation by Rad53.
7. The Remus studies refer to an in vitro analysis. Is it appropriate to apply these data to an in vivo situation? Furthermore, the authors need to modify their hypothesis for a negative regulation of fork movement given the possibility of alternative interpretations. We think so. Much of the molecular events during replication were first characterized in vitro, as exemplified by the ordered assembly of the pre-RC.
8. Provide a rationale for the use of the Vma1 protein for Western blot normalization. We added this information in the figure legend for figure 2.
9. The authors need generally to discuss their results and speculations to a greater degree in the text as discussed by the Reviewers. We have expanded both the result and discussion sections as directed (line# 394-423).
10. In Discussion, the authors need to differentiate between the dNTP effect on fork stalling vs. fork collapse. See previous point.
11. Studies were not conducted in the absence of HU to my understanding. Please eliminate indications to the contrary or provide additional data. Corrected.
12. Authors should compare the analysis from the Bielinsky lab who compared the genome-wide replication profiles of rad53-1 mutant are described in as much as the differing strain backgrounds permit. The study by Raveendranathan, Bielinsky, et al. utilized a microarray containing 424 previously identified origins (not truly genome-wide) to compare wild type vs. checkpoint mutants for origin activation. It by and large recapitulated the finding that checkpoint mutation allows late origins to prematurely activate in HU. And indeed, as the reviewer pointed out, due to differences in strain background and methodology we found a direct comparison to be tenuous.
13. Please state the reference sequence in Figure 4. Done. We stated in the Methods section, and now also in Figure 4 (the new Figure 5) legend, that the S288C reference genome R64.2.1 was used.
14. Please edit the text carefully to eliminate grammatical errors. Done.
15. Please address all other issues raised by the Reviewers in a point-by-point response. See below.

Response to reviewers
Reviewer #1: In this manuscript, the authors used a published single-stranded DNA mapping protocol to identify active replication origins in two different laboratory strains of S. cerevisiae (W303 and A364a) after their release from G1 into hydroxyurea in the presence or absence of an active Rad53 S-phase checkpoint factor. The majority of active origins were shared between the two strains and these origins also shared the same overall response to the status of the Rad53dependent S phase checkpoint. However, the level of ssDNA at origins and the distance traveled by the fork were significantly more reduced in the W303 rad53 mutant strain and correlated with a more robust phosphorylation of Rad53 in the W303 RAD53 strain. A number (71) of origins were only active in W303 or A364a. Comparison of published genome DNA sequencing data between the two strains found that sequence polymorphisms in the consensus ACS element were unlikely to account for the differences. The authors believed that origin activation likely occurred because of the presence of multiple mismatched ACS elements in the DNA sequences. Finally, the authors identified 6 origins that they characterized as Rad53-dependent, in that they were only active in RAD53 but not in rad53 strains. Interestingly, 5 of these origins also overlapped with an ORF, in contrast to the intergenic location of most origins. The lack of origin activation in rad53 cells was not a consequence of transcriptional interference, as published data indicated that the genes in which the origins were located were transcribed at a lower level in rad53 cells compared to RAD53 cells.
Overall, this is an interesting and well documented study, with many of the conclusions from genomic data validated by 2D agarose gel analysis. One of the most remarkable findings was the observation that despite the similar number of origins activated in rad53 between the two strains, W303 was significantly more sensitive to the reduced checkpoint than A364a, showing reduced levels of ssDNA at all origins and reduced spreading of ssDNA from origins. This is shown most dramatically in Figure 1C. The authors speculated that this might be due to the lower level of Rad53 in the W303 rad53 mutant, which in turn led to lower levels of Rad53-P. The striking differences between the two strains need to be much more fully discussed in the text. What could account for the lower activation of all origins in the W303 rad53 mutant? Is this a consequence of the role of Rad53 in the initiation of DNA replication that is independent of checkpoint regulation, and if so, by what mechanism? Is the loading of initiation factors altered at origins, and is Rad53 binding to origins affected? Is there an effect on nucleotide pools in this particular rad53 mutant? Some of these questions could be addressed using ChIP-qPCR of initiation factors at several of the well defined origins on chromosome III in the RAD53 and rad53 strains of both W303 and A364a. This would help to address questions on the differential effects of the rad53 mutation in the two strains, and perhaps more generally uncover some new insights into the relationship of the levels of Rad53 to replication initiation.
We thank the reviewer for these suggestions. Please see our response to point #1 in AE's comments.
Other comments: 1. The identification of RAD53-dependent origins in 5 ORFs represents a very interesting finding. The authors should confirm the published genomic data on these genes by performing RT-qPCR on RNA extracted from HU treated RAD53 and rad53 cells. Some speculation on the possible function of RAD53 at these genes would also be beneficial. Please see response to point #2 in AE's comments.
2. The text needs to be carefully edited as there were some missing sentence parts. Done.
Reviewer #2: Identification and characterization of replication origins are essential for a better understanding of the molecular mechanism of DNA replication. The authors conducted experimental procedures to perform the differential dynamics of origin activation in the A364a and W303 Saccharomyces cerevisiae strains. The authors also found the groups of "Rad53unchecked" and "Rad53-checked" origins by the cooperation of origin usage in wild type and the rad53 mutant cells. The identification of replication origins from the genome-wide and the analysis of a new class of origins would provide new insights into the replication mechanism of S. cerevisiae.
Major comments: 1.Authors should describe the detailed reasons for choosing the strains used in the study. What's the phylogenetic distance of A364a and W303 S. cerevisiae strains? The authors mentioned that "Genetic variation in diverse laboratory strains can manifest in distinct physiological properties". What's the phenotypic difference between these two strains? What's the significant phenotypic difference associated with DNA replication between these two strains? Done. Please see response to point #3 in AE's comments.
2.The authors identified the "Rad53-dependent origins", however, the author needs to clarify that these "Rad53-dependent origins" are not strain-specific. We state in the text that some of the Rad53-dependent origins are in fact strain-specific, though the two origins we characterized by 2D gel and gene expression are not.
Minor comments: 1.The authors used the ARS records of the OriDB database. OriDB is one of the well-known yeast replication origin databases. I recommended the authors also use the DeOri (DOI: 10.1093/bioinformatics/bts151), a database for eukaryotic replication origins, to verify and support your experimental results. Please see response to point #5 in AE's comments.
2.The authors performed the identification of replication origins of two yeast strains by genomewide level. I'd like to suggest the authors attach the detailed information of identified replication origins, including chromosomal position and replication origin sequences, to the supplementary file, which will help the authors and other researchers to further explore the mechanism of DNA replication. Please see response to point #4 in AE's comments.
3.I'd like to recommend some of recent papers related to DNA replication origins in Saccharomyces cerevisiae genome for your kind reference. We thank the reviewer for bringing to our attention these publications. We will certainly use them for our future studies.
Reviewer #3: Comments on the manuscript PONE-D-21-08460 (Peng et al. entitled 'Exceptional origin activation revealed by comparative analysis in two laboratory yeast strains').
In this manuscript, authors performed a comparative analysis of replication origin activation in two yeast strains W303 and A364a when cells were exposed with hydroxyurea. They also analyzed the effect on origin activation of the checkpoint kinase, Rad53, by combining the checkpoint-deficient Rad53-K227A mutant. They found that the activation of replication origins are similar in both strains. However, they find strain-specific origin usage. Although some of strain-specific origins have SNPs, they suggested that the SNP is not the reason of the strainspecificity. They also suggested that the difference of origin usage is partly depends on the activity of Rad53. Finally, they identified a new class of origins that are active only when Rad53 is functional. Although the study is descriptive rather than analytical, some of their findings are interesting and are worth for the publication in the PLoS ONE. However, I feel there are some points that should be clarified before acceptance.
Major points 1. Authors insist that the Rad53 levels are different between W303 and A364a (p. 10). It does not seem that there is a big difference between wildtypes. I also wonder why authors use Vma1 as a control rather than ponceau staining. Moreover, it is known that Rad53 autophosphorylates when it is activated. If authors want to indicate the difference of Rad53 activity in this situation, kinase assay would be the best way. The way simply compare band intensities might be inappropriate, because some of them are not autophosphorylated, it is difficult to extract only autophosphorylated ones, and the extent of autophosphorylation must be considered if you want to say the activity. Please see response to points #6 and #8 in AE's comments. We agree with the reviewer that the difference between Rad53 levels in WT strains is moderate. Yet they were reproducibly seen. We further corroborate with the ChIP-PCR data of Rad53 at origins.
In discussion, they suggested that Rad53 has the function on fork progression by referring the recent Remus Lab's results (ref #35). However, this is an in vitro results obtained by adding the aphidicolin to inhibit DNA polymerase function. I wonder this situation can be applied directly to cells treated with HU. Moreover, in W303 rad53-K227A cells, which has lower levels of Rad53 protein than corresponding A364a cells, fork did not travel longer than wild type nor A364a rad53-K227A (Fig. 2D). Therefore, their interpretation (p10 middle part) and discussion (p17 middle part) seem incorrect for me. Please see response to points #7 and #9. We now provide a more in-depth discussion on the replication fork phenotype in the mutant cells, which we believe clarified the questions the reviewer had.
2. Authors say, 'This is consistent with the notion that checkpoint failure causes replication forks to collapse shortly after initiation from the origin.' (p9, bottom). I think replication forks stall rather than collapse by dNTP shortage in HU-treated cells even in checkpoint-deficient cells. If there are evidences that support authors description, they should be clearly shown here. Shorter fork travel could be caused by more origin firing. Simply because less dNTPs are available when more origins fire. We agree with the reviewer that dNTP pool level should be considered here and we added that in our discussion. We have previously demonstrated that replication forks collapse in the checkpoint-deficient cells after HU treatment using density transfer as a readout (Feng et al. 2009). We now added this reference for clarification (line # 194).