Tup1 is critical for transcriptional repression in Quiescence in S. cerevisiae

  1. Thomas B. Bailey
  2. Phaedra A. Whitty
  3. Eric U. Selker
  4. Jeffrey. N. McKnight
  5. Laura E. McKnight

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Upon glucose starvation, S . cerevisiae shows a dramatic alteration in transcription, resulting in wide-scale repression of most genes and activation of some others. This coincides with an arrest of cellular proliferation. A subset of such cells enters quiescence, a reversible non-dividing state. Here, we demonstrate that the conserved transcriptional corepressor Tup1 is critical for transcriptional repression after glucose depletion. We show that Tup1-Ssn6 binds new targets upon glucose depletion, where it remains as the cells enter the G0 phase of the cell cycle. In addition, we show that Tup1 represses a variety of glucose metabolism and transport genes. We explored how Tup1 mediated repression is accomplished and demonstrated that Tup1 coordinates with the Rpd3L complex to deacetylate H3K23. We found that Tup1 coordinates with Isw2 to affect nucleosome positions at glucose transporter HXT family genes during G0. Finally, microscopy revealed that a quarter of cells with a Tup1 deletion contain multiple DAPI puncta. Taken together, these findings demonstrate the role of Tup1 in transcriptional reprogramming in response to environmental cues leading to the quiescent state.

Article activity feed

  1. Version published to 10.1371/journal.pgen.1010559
  2. Version published to 10.1101/2022.08.10.503497v3 on bioRxiv
  7. Review Commons

    Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    Thank you for giving us the opportunity to submit a revised draft of the manuscript “Tup1 is Required for Transcriptional Repression Necessary in Quiescence in S. cerevisiae” to Review Commons. We appreciate the time and effort that you and the other reviewers dedicated to providing feedback on our manuscript and are grateful for the insightful comments on and valuable improvements to our paper. We believe that the experiments suggested in these comments would bring clarity to the manuscript, and wish that we had the ability to perform them all. Unfortunately our lab is closing, and the only remaining lab member is the PI, so we are only able to perform limited experiments to address some of the concerns raised during review. We have incorporated several changes in response to comments from the reviewers. Those changes are highlighted within the manuscript. Please see below, in blue, for a point-by-point response to the reviewers’ comments and concerns. All page numbers refer to the revised manuscript Word file with tracked changes.

    Of particular note is the discussion of cellular morphology of the tup1∆ and sds3∆ strains. We realize that our findings are purely descriptive, and are not surprised that all three reviewers had comments on this data. This was the source of much discussion among the authors and consultation with other labs; we debated even including these observations in the manuscript, since we were unable to figure out the underlying mechanism. Ultimately we decided that it was worth reporting in case other labs may benefit from the knowledge, and we have altered the language in the manuscript (page 6) to better reflect this. However, if the reviewers feel that this observation would be better left out of the manuscript, we would be willing to remove Figure 6 and any discussion of these images.

    Reviewer #1 Comments:1. The authors chose to examine 3-day SP cells to interrogate quiescence because tup1∆ cells are highly flocculant, interfering with the isolation of purified quiescent cells. These cells are a mixture of both nonquiescent and quiescent cells, so it is not correct to state that they represent a quiescent cell population. The addition of EDTA to the gradients used to isolate quiescent cells could eliminate flocculation and permit the isolation of quiescent cells. EDTA is also often added to media in low amounts to reduce flocculation. The authors need to indicate the proportion of quiescent cells in their SP cultures by applying these tools.

    We appreciate the suggestion, but the phenotype of this strain is not typical flocculation (see photo below, also added to the paper as Supplementary Figure 1). We did add EDTA (pH 8.0) to a final concentration of 10 mM to two separate tup1∆ and it did not visibly affect the clumping of cells. Furthermore, changes to the cell wall are a distinct feature of quiescent S. cerevisiae and contribute to the ability to separate different cell types by density-gradient centrifugation, so it is difficult to anticipate how EDTA would affect our ability to isolate Q cells. We have provided more explanation in the manuscript to better explain this (page 3).

    1. The authors reported that while Xpb1 and Tup1 share many overlapping binding sites, but that Xbp1 does not regulate Tup1's binding. What other factors might be responsible for their shared binding? Could histone deacetylation play a role? This could be addressed by a Tup1 ChIP in an sds3∆ mutant.

    This is a good thought; histone acetylation levels may have a role in regulating Tup1 localization and we would have liked to address this if we had more time. Unfortunately, we had some difficulty performing ChIP of Tup1, because initially we used a FLAG tag which caused a phenotype similar to deletion of Tup1, and had to switch to making myc-tagged strains. This delay meant we did not have time to pursue creating myc-tagged Tup1 in an sds3∆ strain, and now we do not have the ability to follow up on this for revisions.

    1. Has PolII occupancy been examined in Log vs SP cells of tup1∆ to determine if Tup1 inhibits PolII association with its genes that are repressed ?

    We did not look at PolII occupancy in our Tup1 deletion, and could not find any existing datasets with this information. It is our hope that another lab is able to carry out this experiment, because it could be very enlightening, but it is beyond the scope of this work.

    1. The observation that tup1∆ cells have several nuclear puncta is intriguing, although the cytological images need to be improved.

    The nuclear puncta we see in the tup1 deletion are definitely a puzzle. We had limited time to investigate this phenomena, and in discussing the matter with some other labs it seemed doubtful that more advanced imaging would yield anything of use to us. We realized that we accidentally omitted important details for this figure and have updated the manuscript to add them. We imaged 2 biological replicates for each strain and imaged many yeast samples for each strain (which has been added to the caption for Figure 6) and found that our findings were statistically significant (p

    Reviewer #2 Comments1. The authors acknowledge that it would be better to work with purified quiescent cells but couldn't isolate pure populations. As a result, a mixture of quiescent and nonquiescent cells are analyzed in stationary phase. They say this is because Tup1 deletion strains are flocculent. But they performed ChIP-Seq on Myc-tagged Tup1 strain. Don't these cells express Tup1? If not, could this be performed in wild-type yeast with Myc-tagged Tup1? It seems important to separate quiescent from nonquiescent yeast for the authors' conclusions.

    It is true that we could have done ChIP-seq for Tup1 in purified Q cells. We considered it, but decided to look at the mixed population so that we could directly compare our RNA-seq results from the tup1∆ strain. It’s a balance between having some results that are specific to quiescence, versus being able to directly compare the effects of deletion of Tup1 at the sites where it binds. We are now unable to perform this experiment, but we have updated the language in the manuscript (page 3) to better reflect this choice.

    1. The Chipseq data in Fig 1B do not have a y axis and it is consequently not clear whether these data are normalized and shown with the same axis.

    Thank you for pointing this out - these data are normalized to RPKM during processing, and we have updated the caption for figure 1 and the methods on page 10 to reflect this information. Normalizing the data in IGB itself, however, causes an adjustment in the y-axis that makes the tracks appear to be inconsistent. In any case, we are not making claims about the relative amount of signal, and as it is common in the field to not include y-axes on IGB tracks, we have opted to keep the y-axis for Figure 1B as-is.

    1. In Fig 2, it seems important to determine how many genes are different between WT and Tup1 deletion strains in log phase. Are just as many genes different? Or is Tup1 more important in diauxic shift and stationary phase than log phase?

    We did intend to focus only on diauxic shift and stationary phase data for this paper, since there has already been so much work on the role of Tup1 in log phase. As mentioned above, comparisons of RNA between log and DS/Q is difficult. We attempted to find a publicly available dataset to perform some analysis for revisions, but unfortunately most previous work on the effect of Tup1 on transcription was performed via tiling arrays, which is not comparable.

    1. Are the genes that are regulated by Tup1 normally regulated during diauxic shift or stationary phase compared with log growth?

    Because there is a massive global decrease in the level of total RNA in diauxic shift and quiescence (McKnight, Boerma, et al., 2015) it is impossible to directly compare transcript levels between these states in our experiments. If there was time, we could have attempted to repeat these experiments with an external spike-in control; this is potentially something another lab could do to follow up on our findings.

    1. What fraction of the genes that are differentially expressed in Tup1 knockout yeast have Tup1 binding at the promoter? Enhancer? What fraction can be explained by Tup1, Hap1, Nrg1, Mig1 individually and together?

    We have added the number of genes that are differentially expressed in Tup1 knockout yeast during DS to the manuscript (page 3). Regarding enhancers, the genome of S. cerevisiae is very compact, and there is not evidence of long-distance activation of genes as seen in metazoans (Dujon, 1996; Dobi & Winston, 2007; Spiegel and Arnone, 2021). Upstream activating sequences (UASs) are generally considered the closest equivalent to enhancers in cerevisiae, and they tend to function within a few hundred base pairs of the promoter. Our analysis only identifies the nearest gene; it would be difficult to parse out locations in the promoter versus a UAS without a more advanced analysis that is beyond our capabilities now.

    As for the effect of Hap1, Nrg1, and Mig1, we were able to look for their motifs in the genes that are differentially expressed in the Tup1 knockout but we do not have binding data for these factors in quiescence or stationary phase so it is impossible to conclusively state what role those TFs play. This would be a very interesting followup to our work, but is outside the scope of this manuscript.

    References:

    Dujon, B. 1996. The Yeast Genome Project: What did we learn? Trends Genet. 12, 263-270.

    Dobi, K.C.; Winston, F. 2007. Analysis of Transcriptional Activation at a Distance in Saccharomyces cerevisiae. Mol Cell Biol. 27(15), 5575-5586. https://doi.org/10.1128/MCB.00459-007

    Spiegel, JA; Arnone, J.T. 2021.Transcription at a Distance in the Budding Yeast Saccharomyces cerevisiae. Appl. Microbiol. 1(1), 142-149. https://doi.org/10.3390/applmicrobiol1010011

    1. The methodology used to generate the gene ontology enrichments should be described in the methods.

    Thank you for noticing this omission; we have added the relevant information to the manuscript (page 10) and have also added the related citation (page 11).

    1. The authors should provide genomewide data to support the statement that Tup1 and Rpd3 ChIP datasets have substantial overlap. They should also provide genomewide data to support the statement that there is substantial overlap between Rpd3 and Tup1. How much overlap is observed and how much is expected by chance?

    We have compared the existing ChIP data for Rpd3 binding in quiescent cells to our ChIP data for Tup1 in 3-day cultures and included this in the manuscript (page 4, Supplementary Figure 2B), along with a p-value.

    1. For Sds3, similar to Tup1 inactivation, it would be helpful to know how many genes change in with Sds3 inactivation in log phase in addition to diauxic shift and stationary phase.

    As with our response to comment #3, we focused only on diauxic shift and stationary phase data for this paper, and analysis of this data would be difficult without a spike-in control. While there are some existing datasets for RNA-seq of Rpd3 knockouts, this would include both Rpd3L and Rpd3S activity, rather than just Rpd3L, which is our focus with the Sds3 deletion strains. As such, we did not perform RNA-seq of sds3∆ in log phase.

    1. If the argument is that Sds3 and Xbp1 cooperate with Tup1 to affect gene expression, testing the gene expression changes that are associated with Tup1 in Sds3 or Xbp1 knockout strains would help the authors make this point.

    We do not have tup1∆/sds3∆ or tup1∆/xbp1∆ double knockout strains. We attempted to make these strains but could not, which may indicate that these double deletions are synthetic lethal. Deletion of sds3 alone causes a significant reduction in growth rate, so it is perhaps not surprising that we could not create the double knockouts.

    1. The final phenotype of extra DAPI positive blobs in the nucleus is not very specific or clear.

    We agree, please see our comments at the top of this letter.

    Reviewer #3 (Major comments):Did tup1∆/sds3∆ double mutant show the same phenotype with tup1∆ (or sds3∆) single mutant in G0? If Tup1 actually plays role in tandem with Sds3 in the gene regulation during G0, the epistatic relationship might be estimated.

    We do not have tup1∆/sds3∆ or tup1∆/xbp1∆ double knockout strains. We attempted to make these strains but could not, which may indicate that these double deletions are synthetic lethal. Deletion of sds3 alone causes a significant reduction in growth rate, so it is perhaps not surprising that we could not create the double knockouts.

    The histone acetylation was not synergistically augmented in the above double mutant?

    Please see the response above.

    The authors showed that tup1∆ but not sds3∆ cells contain multiple DAPI signals but sds3∆ cells show abnormal cell shape in G0 phase. These phenotypic abnormalities in these mutants suggest a potential mitotic defect. Both mutants showed very similar abnormalities in H3K23 acetylation and gene expressions in quiescent state. Why these showed distinctly different abnormality in cell morphology during G0?

    Unfortunately we were unable to investigate this further.

    Did iswi2∆ cells also show abnormality in G0 phase?

    No, they did not; thank you for asking, this is a good question. We have added this information to the manuscript (page 6).

    (Minor comments)Supplementary figure1. This data seems to be very important. I recommend to use this data in the main figure with statistical analysis (p-values) to show the significant overlap of Tup1 and Rdp3 distribution.

    We have compared the existing ChIP data for Rpd3 binding in quiescent cells to our ChIP data for Tup1 in 3-day cultures and included this in the manuscript (page 4, Supplementary Figure 2B) . We do feel that this data belong in the supplement, however, because the data is not exactly equivalent to our studies: quiescent cells and 3-day cultures are not the same, and knockout of Rpd3 eliminates function of both Rpd3L and Rpd3S complexes, while knocking out Sds3 targets only the Rpd3L complex.

    Figure 4. Histone acetylation level data in Figure 4A and the data for gene repressions by Tup1 and Sds3 in Figure 4C seem to be very important. However, statistical analysis data (p-values) was not presented. Please show the statistical analysis data (p-values) as in figure 3 to show that the Tup1 and Sds3 contribute similarly in histone deacetylation and repression. The author did not find the significant changes of histone deacetylation in xbp1∆ cells but said that when filtered in Xbp1 binding motif Xbp1 depletion has similar effect on the acetylation level. Please show this data.

    We have added language in the manuscript comparing genes with altered acetylation levels to those that are differentially expressed in our RNA-seq datasets, along with a p-value, to page 5.

  8. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    In this study, the authors found that Tup1 corepressor coordinates with Rpd3L HDAC complex in the deacetylation of histone H3K23 during quiescence entry. They also found that Tup1 coordinates with ISWI2 to regulate +1 nucleosome at HXT family genes for the gene repression during G0. Finally, they showed that loss of Tup1 results in abnormal shape of nuclei during G0. These data suggest a critical role of Tup1 corepressor in the proper gene regulations in tandem with Rpd3 HDAC and ISWI2 remodeler upon quiescence entry. This study seems to be a critical progress from their previous study on the roles of Rpd3 complex in the quiescence entry (McKnight et al. 2015 Mol. Cell). My comments were listed bellow.

    Major comments

    Did tup1∆/sds3∆ double mutant show the same phenotype with tup1∆ (or sds3∆) single mutant in G0? If Tup1 actually plays role in tandem with Sds3 in the gene regulation during G0, the epistatic relationship might be estimated.

    The histone acetylation was not synergistically augmented in the above double mutant?

    The authors showed that tup1∆ but not sds3∆ cells contain multiple DAPI signals but sds3∆ cells show abnormal cell shape in G0 phase. These phenotypic abnormalities in these mutants suggest a potential mitotic defect. Both mutants showed very similar abnormalities in H3K23 acetylation and gene expressions in quiescent state. Why these showed distinctly different abnormality in cell morphology during G0?

    Did iswi2∆ cells also show abnormality in G0 phase?

    Minor comments

    Supplementary figure1. This data seems to be very important. I recommend to use this data in the main figure with statistical analysis (p-values) to show the significant overlap of Tup1 and Rdp3 distribution.

    Figure 4. Histone acetylation level data in Figure 4A and the data for gene repressions by Tup1 and Sds3 in Figure 4C seem to be very important. However, statistical analysis data (p-values) was not presented. Please show the statistical analysis data (p-values) as in figure 3 to show that the Tup1 and Sds3 contribute similarly in histone deacetylation and repression. The author did not find the significant changes of histone deacetylation in xbp1∆ cells but said that when filtered in Xbp1 binding motif Xbp1 depletion has similar effect on the acetylation level. Please show this data.

    Significance

    The current work by the authors significantly progressed their work (McKnight et al. 2015 Mol. Cell) showing important role of Rpd3 deacetylase complex in the quiescence entry. This study will significantly contribute to understanding the gene regulation mechanisms in the G0 entry.

  9. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    In this paper, the authors investigate the role of the Tup1 transcriptional co-repressor in transcriptional repression after glucose depletion in quiescent S. cerevisiae. Tup1 N terminal helix facilitates oligomerization and interacts with histone tails and a beta propeller domain interacts with histone deacetylases and DNA binding factors. Tup1 interacts with nucleosomes by binding histone H3 and H4 tails, binds to hypoacetylated tails and stabilizes +1 nucleosomes repositioned by Isw2. Tup1 had been identified as a gene essential for viability of cells in G0 and implicated in glucose repression in yeast. The authors found that Tup1-Ssn6 binds new targets when S cerevisiae are glucose deprived and as they enter G0. Tup1 was found to repress the expression of genes involved in glucose metabolism and glucose transport. Tup1 was discovered to coordinate with the Rpd3L complex to deacetylate H3K23. Tup1 was also found to coordinate with Isw2 to affect nucleosome positions at hexose transporter family genes. Some cells with Tup1 deletion had multiple DAPI puncta in stationary phase, suggesting a possible role for Tup1 in mitosis.

    Overall Comments

    In Fig 1, the shift in Tup1 binding with diauxic shift and stationary phase is clear. The data clearly show an effect of Tup1 on histone acetylation and nucleosome positioning. However, whether Tup1 has an important functional role in quiescence is not clear.

    Comments

    1. The authors acknowledge that it would be better to work with purified quiescent cells but couldn't isolate pure populations. As a result, a mixture of quiescent and nonquiescent cells are analyzed in stationary phase. They say this is because Tup1 deletion strains are flocculent. But they performed ChIP-Seq on Myc-tagged Tup1 strain. Don't these cells express Tup1? If not, could this be performed in wild-type yeast with Myc-tagged Tup1? It seems important to separate quiescent from nonquiescent yeast for the authors' conclusions.

    2. The Chipseq data in Fig 1B do not have a y axis and it is consequently not clear whether these data are normalized and shown with the same axis.

    3. In Fig 2, it seems important to determine how many genes are different between WT and Tup1 deletion strains in log phase. Are just as many genes different? Or is Tup1 more important in diauxic shift and stationary phase than log phase?

    4. Are the genes that are regulated by Tup1 normally regulated during diauxic shift or stationary phase compared with log growth?

    5. What fraction of the genes that are differentially expressed in Tup1 knockout yeast have Tup1 binding at the promoter? Enhancer? What fraction can be explained by Tup1, Hap1, Nrg1, Mig1 individually and together?

    6. The methodology used to generate the gene ontology enrichments should be described in the methods.

    7. The authors should provide genomewide data to support the statement that Tup1 and Rpd3 ChIP datasets have substantial overlap. They should also provide genomewide data to support the statement that there is substantial overlap between Rpd3 and Tup1. How much overlap is observed and how much is expected by chance?

    8. For Sds3, similar to Tup1 inactivation, it would be helpful to know how many genes change in with Sds3 inactivation in log phase in addition to diauxic shift and stationary phase.

    9. If the argument is that Sds3 and Xbp1 cooperate with Tup1 to affect gene expression, testing the gene expression changes that are associated with Tup1 in Sds3 or Xbp1 knockout strains would help the authors make this point.

    10. The final phenotype of extra DAPI positive blobs in the nucleus is not very specific or clear.

    Significance

    Discovering transcription factors that are critical for the entry into and maintenance of quiescence is an important area of discovery as relatively little is known about gene regulation during quiescence and understanding this process is fundamentally important for our understanding of cell biology. Previous studies had implicated Tup1 in stationary phase in yeast. These studies provide additional insight into the mechanism.

  10. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    In this manuscript, the authors have investigated the role of the Tup1 corepressor in transcriptional repression that occurs during the establishment of quiescence in S. cerevisiae. Using ChIP-seq, they found that Tup1 is present at many sites during log phase growth, and that it localizes to new targets upon glucose exhaustion at a point called diauxie; post-diuaxie, Tup 1 remains associated with these sites in stationary phase (SP), which they define as growth of yeast cells for 3 days after inoculation in glucose-rich medium. To understand the significance of Tup1 relocalization during quiescence establishment, they performed differential expression (DE) analysis following RNA-seq in WT and tup1∆ cells at the diauxic shift and 3-day SP. The large number of DE genes in both cell states were consistent with a role for Tup1 in the regulation of key targets associated with the initiation of quiescence, although there was not a strict correlation between Tup1 occupancy at these DE genes. Next, they investigated the relationship of Tup1 to Xbp1 and Rpd3, factors that play a role in transcriptional repression during quiescence. They noted significant overlap between the binding sites for the three factors in either isolated quiescent cells or 3-day SP cells using published and newly acquired ChIP-seq data. However, the absence of Xbp1, a transcription factor that recruits the Rpd3 histone deacetylase in quiescent cells, did not alter Tup1 binding to its targets in these cells. RNA-seq analysis of xbp1∆ and tup1∆ SP cells found a significant overlap between the genes that were repressed in the absence of the two transcription factors. Moreover, Tup1, Xbp1, and Rpd3 were noted to be required for H3K23 deacetylation at repressed genes in SP. Finally, the authors asked if the Tup1 affects the position of nucleosomes at TSSs in SP cells by performing MNase-seq. They found that Tup1 and the chromatin remodeler Isw2, which interacts with Tup1, are required to position nucleosomes at the promoters of a family of glucose transporter genes, targets of Tup1 during the initiation of quiescence.

    Comments:

    1. The authors chose to examine 3-day SP cells to interrogate quiescence because tup1∆ cells are highly flocculant, interfering with the isolation of purified quiescent cells. These cells are a mixture of both nonquiescent and quiescent cells, so it is not correct to state that they represent a quiescent cell population. The addition of EDTA to the gradients used to isolate quiescent cells could eliminate flocculation and permit the isolation of quiescent cells. EDTA is also often added to media in low amounts to reduce flocculation. The authors need to indicate the proportion of quiescent cells in their SP cultures by applying these tools.

    2. The authors reported that while Xpb1 and Tup1 share many overlapping binding sites, but that Xbp1 does not regulate Tup1's binding. What other factors might be responsible for their shared binding? Could histone deacetylation play a role? This could be addressed by a Tup1 ChIP in an sds3∆ mutant.

    3. Has PolII occupancy been examined in Log vs SP cells of tup1∆ to determine if Tup1 inhibits PolII association with its genes that are repressed ?

    4. The observation that tup1∆ cells have several nuclear puncta is intriguing, although the cytological images need to be improved.

    Significance

    This manuscript presents some new information that links Tup1, a multifacteted transcriptional co-repressor, to the repression of transcription that occurs during the establishment of yeast quiescence. The McKnight lab has previously shown that the Rpd3 HDAC plays an important role in this process, in part through its recruitment by the TF Xbp1. Tup1 also binds to Rpd3, and the overlap between the sites where Tup1 and Xpb1 bind and the targets that they repress, suggests a shared function in establishing transcriptional reprogramming. Whether this shared function is based only on the recruitment of the Rpd3 histone deacetylation is unclear, and, more importantly, there is also some question if deacetylation is in fact the main factor driving transcriptional repression during the initiation of quiescence. The finding that Tup1 is responsible for repositioning nucleosomes at a class of glucose transporters to repress the transcription of these genes suggests a specific function for Tup1 during the establishment of quiescence; however, this is apparently not a broad function. Thus, we are left with a lot of nice descriptive information as a result of well-done experiments that has not revealed much about the mechanism by which Tup1 regulates the global transcriptional reprogramming that occurs during quiescence.

  11. Version published to 10.1101/2022.08.10.503497v2 on bioRxiv
  12. Version published to 10.1101/2022.08.10.503497v1 on bioRxiv