Loss-of-function mutation survey revealed that genes with background-dependent fitness are rare and functionally related in yeast

  1. Elodie Caudal
  2. Anne Friedrich
  3. Arthur Jallet
  4. Marion Garin
  5. Jing Hou
  6. Joseph Schacherer

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

In natural populations, the same mutation can lead to different phenotypic outcomes due to the genetic variation that exists among individuals. Such genetic background effects are commonly observed, including in the context of many human diseases. However, systematic characterization of these effects at the species level is still lacking to date. Here, we sought to comprehensively survey background-dependent traits associated with gene loss-of-function (LoF) mutations in 39 natural isolates of Saccharomyces cerevisiae using a transposon saturation strategy. By analyzing the modeled fitness variability of a total of 4,469 genes, we found that 15% of them, when impacted by a LoF mutation, exhibited a significant gain- or loss-of-fitness phenotype in certain natural isolates compared with the reference strain S288C. Out of these 632 genes with predicted background-dependent fitness effects, around 2/3 impact multiple backgrounds with a gradient of predicted fitness change while 1/3 are specific to a single genetic background. Genes related to mitochondrial function are significantly overrepresented in the set of genes showing a continuous variation and display a potential functional rewiring with other genes involved in transcription and chromatin remodeling as well as in nuclear–cytoplasmic transport. Such rewiring effects are likely modulated by both the genetic background and the environment. While background-specific cases are rare and span diverse cellular processes, they can be functionally related at the individual level. All genes with background-dependent fitness effects tend to have an intermediate connectivity in the global genetic interaction network and have shown relaxed selection pressure at the population level, highlighting their potential evolutionary characteristics.

Article activity feed

  1. Version published to 10.1073/pnas.2204206119
  2. 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

    Reviewer #1 (Evidence, reproducibility and clarity):

    Thank you for the opportunity to review "Population-level survey of loss-of-function mutations revealed that background dependent fitness genes are rare and functionally related in yeast" by Caudal et al. This manuscript reports on the genetic background-dependent traits resulting from natural variation. Authors use 39 natural isolates of the budding yeast (S. cerevisiae) and apply transposon saturation mutagenesis approach to analyze fitness due to loss of function mutations. They identified background and environment dependent genes. They estimate that background specific rewiring is rare and represents instances of bridging between bioprocesses as well as connecting functional related genes.

    Major comments

    Authors filtered strains based on whole chromosome aneuploidies, but what about chromosome arm aneuploidies. Were they detected and if so how were they handled? This should be discussed.

    We did not detect any chromosome arm aneuploidies. In fact, if any significant segmental duplication were present in any of the tested strains, we would have observed changes of gene essentiality for multiple successive ORFs, which was not the case.

    How does chromatin structure variation across different genetic backgrounds affect the results of the screen? Is this a confounding variable? This should be discussed.

    We thank the author for raising this interesting point. There are two aspects to take into consideration. First, transposon insertion is biased by nucleosome occupation, as is more or less expected. In previous screens and in our data, this bias is translated by the lower insertion density in the promoter in addition to the ORF for essential genes. If the nucleosome occupancy were conserved across different genetic background, this insertion bias won’t be a confounding factor as the same gene will share the same bias across different genetic backgrounds. Second, if the nucleosome occupancy is variable across different genetic backgrounds, it could potentially lead to some background-specific insertion biases, however it is difficult to know whether it would be the cause or the consequence of the mutation. In any case, currently there is no chromatin structure data across different genetic backgrounds available and this could be a direction for future research.

    On page 7 authors discuss the involvement of other biological processes in addition to respiration and mitochondrial function. It is not clear what they are referring to. This should be clarified in the main text.

    We clarified this point in the modified ms.

    It would be useful to annotate the functional information discussed in the text directly on the network in Fig. 4 A and B.

    We included annotations on the networks (see Fig. 4 and Fig. S4) as suggested in the modified ms.

    On page 9, authors should comment on the origin of ACP and CLG strain that would result in the similarity of their fitness profile to S288C which they note as an exception.

    ACP is an isolate from Russian wine and CLG is a clinical isolate from UK. In terms of the overall genetic diversity, these two strains are not closely related to the reference strain S288C. As for other profiles, no correlations were observed between the background-dependent mutant fitness variation and their genetic origins.

    On page 10 authors discuss that background-specific fitness genes can belong to protein complexes. Can authors test this formally by looking at the overlap with the protein complex standard or protein interaction standard? This would strengthen this statement.

    Due to the low number of cases, it is impossible to test this using protein complex standards as the size of the terms are too small as well as the sample size. However, the enriched SAFE terms are in general representative of biological processes which includes multiple protein complexes with similar functions. The genes enriched for each SAFE term is further broken down to specific GO terms, as indicated in Table S4.

    Authors should discuss the reasons why transcription & chromatin remodeling and nuclearcytoplasmic transport, are anticorrelated with genes involved in mitochondrial translation in terms of their fitness profiles and the implications for the evolution of environment-dependent fitness genes.

    These observations were new and we are currently looking for potential explanations to this effect. Unfortunately, there is no obvious explanation we can think of and discuss at this point. More data and further experiments are needed to have some clues about this observation.

    Authors discuss the limitation of the Hermes system however couldn't they test this system with a different inducible promoter such as estradiol regulated promoter to remove the effect of galactose metabolism?

    For the Hermes system to work effectively, we need a highly expressed promoter system that is also inducible and *GAL1 *is the strongest available. As for the estradiol system, first it requires the induction machinery to be integrated in the strain and that will significantly limit the scaling of the project, and second, the maximum induction level is significantly lower than that from the *GAL1 *system, as is recently shown in Arita et al., MSB 2021. For these reasons, the effect of galactose metabolism is inevitable using any transposon system at present.

    Minor comments All figures should contain the appropriate colour bars and legends. For example, Figure S5B relies on the colour bar in Figure 5C but it should have its own colour bar.

    We modified the figures as suggested.

    Reviewer #1 (Significance):

    This work provides a comprehensive survey of the variation in natural isolates of yeast and would be interesting to a broad audience studying the genotype-to-phenotype relationship. It is the first study that systematically assessed the fitness effect of loss of function mutations across a large panel of natural isolates providing novel insight into the background specific and environment dependent genes. This represents a valuable resource for the community to ask questions about natural variation in yeast. My expertise is in complex genetic networks in yeast and genome evolution.

    Reviewer #2 (Evidence, reproducibility and clarity):

    For decades, geneticists have used loss of function (LoF) mutations to unravel the molecular bases of phenotypic variability. However, a common concern is to what extent the phenotypes observed in a strain or accession recapitulates what happens at the species level. In not few cases, anecdotal evidence show that an observed mutant phenotype is not recapitulated in another strain, presumably due to the "strain background". Recent efforts using different strains of Saccharomyces cerevisiae have addressed the problem, but the number has been limited. Here, Elodie Caudal et al. use an ingenious transposon-saturation strategy to carry out a large-scale, genome-wide screen of LoF mutations in 39 strains. Based on a competitive-pooling strategy, authors estimate the probability of 4,469 genes being essential, compared to the reference S288C laboratory strain. Background-specific effects were in general rare. Around 15% of these genes show an essential phenotype which is dependent on the strain background, most of them showing continuous variation across all backgrounds and one third being specific of only one strain. Such background specific genes are functionally related and are under relaxed purifying selection and show "intermediate" integration in genetic-interaction networks compared to essential and non-essential genes. The manuscript is very easy to follow, and the experimental/statistical procedures are transparent and in general well described.

    Major comments

    1. A limitation of the transposon saturation strategy is the need of galactose as the carbon source, which confounds scoring of genetic background effects. The study would highly benefit from any kind of orthogonal validation or phenotype predictions, beyond the BMH1 case presented (Fig S5). Few options would be direct testing of lethal/sick phenotypes of clean gene knockouts for discussed hits (Fig 5) in several strains and conditions including galactose, testing few of the transposon libraries under different conditions to validate the environment nature of the continuous behavior, or testing the predictive power of the method using data or strains used in Galardini 2019 (ref. 25).

    As all three reviewers suggested that validation of our predicted probability score should be supported by experimental data, we performed orthogonal validations for 8 genes across 17 backgrounds. We have included the new results in the revised ms.

    Showing the degree of replicability of the entire procedure would also help, form transposon insertion to phenotypic comparisons. If we understood correctly, this was indeed done for isolate AKE. What is the correlation of their probability scores?

    The AKE strain was done twice due to the mixed haploid/diploid profile, as mentioned in the text. In this case the reproducibility in terms of probability scores is expected to be lower. We plotted the predicted probability values for the two reps (attached below) and calculated the Pearson’s correlation. The correlation coefficient is 0.86 (P-value

    The use of "fitness genes" is confusing, since the main phenotypic output here scored for each gene is LoF lethality, or more specifically the probability of being lethal or non-essential. Lethality or essentiality would be a more appropriate concept throughout. A next step would indeed be to quantify the phenotypic effects in a more quantitative manner (which is generally used while referring to a gene's fitness effect).

    We clarified this point in the revised version and use “predicted fitness variation” instead of “fitness genes”.

    Some minor comments

    -Considering that part of the signal is coming from the specific environment tested, one would expect some degree of clustering among related strains based on their gene-essentiality probability (Fig2), given that growth phenotypes correlate well with strain origins when tested under different environments (Warringer et al., 2011). Please discuss.

    In Warringer et al. 2011, the correlation was more pronounced between species (*S. paradoxus *vs. S. cerevisiae) than intraspecifically. Moreover, it was based on a very small sample size. In fact, multiple more recent studies have shown that the growth phenotypes across a large number of conditions between strains in *S. cerevisiae *is not correlated with their genetic origins (Peter et al. Nature 2018). Indeed, it is not unexpected that the gene-essentiality probability profiles are not correlated with their origins.

    -Galactose is not a non-fermentable carbon source (pg 11, pg12). It is true that flux trough the fermentative pathways is lower and that the respiratory pathways are induced in galactose, when compared to growth on glucose, but galactose is readily fermented under low oxygen conditions. Indeed, variation in the regulation of these pathways could explain the environmental effects detected.

    The reviewer raised a good point. While galactose is not a non-fermentable carbon source, the entry of galactose into glycolysis requires the respiration pathways and rho-/rho0 yeast mutants are unable to grow on galactose as the sole carbon source. We clarified this in the new version of the ms.

    -Examples on FigS3 were useful for a better intuition of how the actual data looks like. Perhaps some of this belongs in Fig1.

    Schematic presentation of the insertion profiles is already shown in Figure 1C. Due to the limited size of Figure 1, we kept Fig S3 as it is in the new version of the ms.

    Fig2, restrict the #insertions label to the actual limits for the set of 39 strains. Currently, it seems there are strains with fewer than 100K and no strain with 300K insertions.

    We thank the reviewer for pointing this out, it was a scaling problem and we fixed it.

    -pg5 paragraph 2, a line on how representative is the set of 106 isolates and again later for the final data set of 39. Which main clades are missing or perhaps overrepresented?

    Compared to the original 106 isolates, the final 39 isolates are still broadly representative of the species diversity, albeit some of the most divergent clusters, such as isolates from the French Guiana and from China, are underrepresented. We included this comment in the revised version.

    -pg6 paragraph 1, should be 106 or 107?

    It was 106 plus the reference strain. This point is clarified now in the new ms.

    -pg14 line2, is OD of 0.5 correct or was also 0.05 as in galactose? This is relevant, since it would change the competitive selection regime under galactose or glucose (more generations under glucose in the latter case). For clarity, authors could here state an approximate number of cell divisions in each medium.

    The OD of 0.5 was correct as this step was only intended as a “recovery phase” and was used to increase the mutant pool for sequencing. We also clarified this point in the text.

    -pg14 line 2, correct wording "to enrich for cells the transposon.."

    We clarified this point in the revised version.

    Reviewer #2 (Significance):

    While recent previous studies have measured genetic background dependent effects of gene mutations at the genome-wide level, this is the first study addressing the problem at the broader population level. Confirming that such effects are in general rare, even at this broad level, is a significant advance in the field. It is limited in the number of environmental conditions and subsequent insights (as in Galardini 2019, ref #25) and in more mechanistic views of specific allele interactions (as in Mullis 2018, ref #5). We feel, however, that these directions would already be out of the scope of the well-framed question here addressed.

    Because of the problem addressed and tackled in an ingenious and comprehensive manner, this manuscript will attract the attention of a broad audience of geneticists, genome and systems biologists. Our main expertise is in yeast genetics and functional genomics.

    **Referee Cross-commenting**

    Reviewer #1 commented the possibility that insertion density could be determined by local chromatin status instead of gene essentiality, given that transposon insertion occurs more often at nucleosome free sites (point 2). While the insertion pattern around the essential gene's vicinity is convincing, we agree that it would help to show that these phenomena are independent from one another, or that this issue must at least be discussed.

    The seeming need of further experimental or analytical validation was raised by reviewers #1 and #3.

    As mentioned above, we performed orthogonal validations for 8 genes across 17 backgrounds and we included the results in the revised ms.

    __Reviewer #3 (Evidence, reproducibility and clarity): __ In this manuscript, Caudal et al tested differences in gene knockout phenotypes across genetically diverse yeast strains using a transposon system. After initially querying 106 strains, most tested strains were removed from further consideration due to low transposon insertion numbers, aneuploidies, or other issues. The authors used the remaining 39 strains to identify a set of 632 genes that are required for normal growth in some genetic backgrounds but not in others. These context-dependent fitness genes are enriched for genes with a role in respiration, which could be because the experiment is performed using galactose as carbon source. Further analysis of potential environment-dependent fitness genes revealed two separate groups of genes that were anti-correlated in their fitness profiles.

    I found this an interesting paper, that explores differential gene essentiality (fitness) across diverse yeast strains. The authors give a detailed description of their findings, thereby differentiating between "environmental" and "genetic background" factors. The paper is well-written and the results are clearly presented. I have only two main concerns, both regarding the quality of the produced data:

    Major comments:

    • Looking at the differential fitness scores in the supplementary data, none of the 57 genes that are known to show differential essentiality between S288C and Sigma1278b (Dowell et al., 2010, Science) appear to be identified as having differential fitness in the transposon screen. The authors mention that some of these genes have a severe fitness defect when deleted in the nonessential background and that some are only partially essential. Although this is certainly true for specific cases, deletion mutants of most of these 57 genes show a large difference in fitness between S288C and Sigma, and this thus doesn't sufficiently explain the complete lack of validation of 57 known positive cases. I think the authors need to further clarify why these known positive controls are not identified in their screen.

    In Dowell et al. 2010, the essentiality was determined by tetrad dissection comparing S288C and Sigma, and as shown in the supplemental data, ~1/3 out of the 57 are in fact extremely sick in one background and non-viable in the other. This strong fitness defect cannot be distinguished using the transposon method. More recently in Hou et al. PNAS 2019, it has been shown that ~15 out of the 57 original cases were due to chromosomal genetic modifiers, which again, mainly concerned the “domain essential” effect that we also captured in our data. An addition, 8 hits out of the 57 were shown to be related to mitochondrial genomes in Edward et al. PNAS 2014, and due the galactose condition we used, these cases were not detected. Other undetected cases were due to the low coverage in the corresponding regions in either one or both backgrounds.

    • Related to the previous point, the authors perform no secondary validation of identified context-dependent essential genes. They show that they can recapitulate known sets of essential and nonessential genes in S288c, but given my previous point, it is not clear how well their logistic model works for predicting differential gene essentiality/fitness. In my opinion, experimental validation of a subset of the identified differential fitness genes is needed to be able to be confident about the results.

    As already mentioned above, new experiments were performed in order to validate a subset of the identified differential fitness genes. The results were included in the revised version of the ms.

    Minor comments:

    • The authors provide lots of data spread over many columns in the supplementary tables. However, a description of what is in each column is missing, and without it, it is not always possible to understand the data.

    We added column annotations in the spread sheets as suggested.

    • I didn't understand the sentence at the bottom of page 5: "the number of insertion drops from -100 bp prior to CDS and extends to - 100 bp until the terminator region". Perhaps the authors can rephrase.

    We clarified this point in the revised version.

    __Reviewer #3 (Significance): __

    To my knowledge, this is the first paper exploring gene essentiality across a large number of genetically diverse yeast strains. This paper will be of interest to a broad range of geneticists.

  3. 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 manuscript, Caudal et al tested differences in gene knockout phenotypes across genetically diverse yeast strains using a transposon system. After initially querying 106 strains, most tested strains were removed from further consideration due to low transposon insertion numbers, aneuploidies, or other issues. The authors used the remaining 39 strains to identify a set of 632 genes that are required for normal growth in some genetic backgrounds but not in others. These context-dependent fitness genes are enriched for genes with a role in respiration, which could be because the experiment is performed using galactose as carbon source. Further analysis of potential environment-dependent fitness genes revealed two separate groups of genes that were anti-correlated in their fitness profiles.

    I found this an interesting paper, that explores differential gene essentiality (fitness) across diverse yeast strains. The authors give a detailed description of their findings, thereby differentiating between "environmental" and "genetic background" factors. The paper is well-written and the results are clearly presented. I have only two main concerns, both regarding the quality of the produced data:

    Major comments:

    • Looking at the differential fitness scores in the supplementary data, none of the 57 genes that are known to show differential essentiality between S288C and Sigma1278b (Dowell et al., 2010, Science) appear to be identified as having differential fitness in the transposon screen. The authors mention that some of these genes have a severe fitness defect when deleted in the nonessential background and that some are only partially essential. Although this is certainly true for specific cases, deletion mutants of most of these 57 genes show a large difference in fitness between S288C and Sigma, and this thus doesn't sufficiently explain the complete lack of validation of 57 known positive cases. I think the authors need to further clarify why these known positive controls are not identified in their screen.
    • Related to the previous point, the authors perform no secondary validation of identified context-dependent essential genes. They show that they can recapitulate known sets of essential and nonessential genes in S288c, but given my previous point, it is not clear how well their logistic model works for predicting differential gene essentiality/fitness. In my opinion, experimental validation of a subset of the identified differential fitness genes is needed to be able to be confident about the results.

    Minor comments:

    • The authors provide lots of data spread over many columns in the supplementary tables. However, a description of what is in each column is missing, and without it, it is not always possible to understand the data.
    • I didn't understand the sentence at the bottom of page 5: "the number of insertion drops from -100 bp prior to CDS and extends to - 100 bp until the terminator region". Perhaps the authors can rephrase.

    Significance

    To my knowledge, this is the first paper exploring gene essentiality across a large number of genetically diverse yeast strains. This paper will be of interest to a broad range of geneticists.

  4. 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

    For decades, geneticists have used loss of function (LoF) mutations to unravel the molecular bases of phenotypic variability. However, a common concern is to what extent the phenotypes observed in a strain or accession recapitulates what happens at the species level. In not few cases, anecdotal evidence show that an observed mutant phenotype is not recapitulated in another strain, presumably due to the "strain background". Recent efforts using different strains of Saccharomyces cerevisiae have addressed the problem, but the number has been limited. Here, Elodie Caudal et al. use an ingenious transposon-saturation strategy to carry out a large-scale, genome-wide screen of LoF mutations in 39 strains. Based on a competitive-pooling strategy, authors estimate the probability of 4,469 genes being essential, compared to the reference S288C laboratory strain. Background-specific effects were in general rare. Around 15% of these genes show an essential phenotype which is dependent on the strain background, most of them showing continuous variation across all backgrounds and one third being specific of only one strain. Such background specific genes are functionally related and are under relaxed purifying selection and show "intermediate" integration in genetic-interaction networks compared to essential and non-essential genes. The manuscript is very easy to follow, and the experimental/statistical procedures are transparent and in general well described.

    Major comments

    1. A limitation of the transposon saturation strategy is the need of galactose as the carbon source, which confounds scoring of genetic background effects. The study would highly benefit from any kind of orthogonal validation or phenotype predictions, beyond the BMH1 case presented (Fig S5). Few options would be direct testing of lethal/sick phenotypes of clean gene knockouts for discussed hits (Fig 5) in several strains and conditions including galactose, testing few of the transposon libraries under different conditions to validate the environment nature of the continuous behavior, or testing the predictive power of the method using data or strains used in Galardini 2019 (ref. 25).
    2. Showing the degree of replicability of the entire procedure would also help, form transposon insertion to phenotypic comparisons. If we understood correctly, this was indeed done for isolate AKE. What is the correlation of their probability scores?
    3. The use of "fitness genes" is confusing, since the main phenotypic output here scored for each gene is LoF lethality, or more specifically the probability of being lethal or non-essential. Lethality or essentiality would be a more appropriate concept throughout. A next step would indeed be to quantify the phenotypic effects in a more quantitative manner (which is generally used while referring to a gene's fitness effect).

    Some minor comments

    -Considering that part of the signal is coming from the specific environment tested, one would expect some degree of clustering among related strains based on their gene-essentiality probability (Fig2), given that growth phenotypes correlate well with strain origins when tested under different environments (Warringer et al., 2011). Please discuss.

    -Galactose is not a non-fermentable carbon source (pg 11, pg12). It is true that flux trough the fermentative pathways is lower and that the respiratory pathways are induced in galactose, when compared to growth on glucose, but galactose is readily fermented under low oxygen conditions. Indeed, variation in the regulation of these pathways could explain the environmental effects detected.

    -Examples on FigS3 were useful for a better intuition of how the actual data looks like. Perhaps some of this belongs in Fig1.

    Fig2, restrict the #insertions label to the actual limits for the set of 39 strains. Currently, it seems there are strains with fewer than 100K and no strain with 300K insertions.

    -pg5 paragraph 2, a line on how representative is the set of 106 isolates and again later for the final data set of 39. Which main clades are missing or perhaps overrepresented?

    -pg6 paragraph 1, should be 106 or 107?

    -pg14 line2, is OD of 0.5 correct or was also 0.05 as in galactose? This is relevant, since it would change the competitive selection regime under galactose or glucose (more generations under glucose in the latter case). For clarity, authors could here state an approximate number of cell divisions in each medium.

    -pg14 line 2, correct wording "to enrich for cells the transposon.."

    Significance

    While recent previous studies have measured genetic background dependent effects of gene mutations at the genome-wide level, this is the first study addressing the problem at the broader population level. Confirming that such effects are in general rare, even at this broad level, is a significant advance in the field. It is limited in the number of environmental conditions and subsequent insights (as in Galardini 2019, ref #25) and in more mechanistic views of specific allele interactions (as in Mullis 2018, ref #5). We feel, however, that these directions would already be out of the scope of the well-framed question here addressed.

    Because of the problem addressed and tackled in an ingenious and comprehensive manner, this manuscript will attract the attention of a broad audience of geneticists, genome and systems biologists. Our main expertise is in yeast genetics and functional genomics.

    Referee Cross-commenting

    Reviewer #1 commented the possibility that insertion density could be determined by local chromatin status instead of gene essentiality, given that transposon insertion occurs more often at nucleosome free sites (point 2). While the insertion pattern around the essential gene's vicinity is convincing, we agree that it would help to show that these phenomena are independent from one another, or that this issue must at least be discussed.

    The seeming need of further experimental or analytical validation was raised by reviewers #1 and #3.

  5. 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

    Thank you for the opportunity to review "Population-level survey of loss-of-function mutations revealed that background dependent fitness genes are rare and functionally related in yeast" by Caudal et al. This manuscript reports on the genetic background-dependent traits resulting from natural variation. Authors use 39 natural isolates of the budding yeast (S. cerevisiae) and apply transposon saturation mutagenesis approach to analyze fitness due to loss of function mutations. They identified background and environment dependent genes. They estimate that background specific rewiring is rare and represents instances of bridging between bioprocesses as well as connecting functional related genes.

    Major comments

    1. Authors filtered strains based on whole chromosome aneuploidies, but what about chromosome arm aneuploidies. Were they detected and if so how were they handled? This should be discussed.
    2. How does chromatin structure variation across different genetic backgrounds affect the results of the screen? Is this a confounding variable? This should be discussed.
    3. On page 7 authors discuss the involvement of other biological processes in addition to respiration and mitochondrial function. It is not clear what they are referring to. This should be clarified in the main text.
    4. It would be useful to annotate the functional information discussed in the text directly on the network in Fig. 4 A and B.
    5. On page 9, authors should comment on the origin of ACP and CLG strain that would result in the similarity of their fitness profile to S288C which they note as an exception.
    6. On page 10 authors discuss that background-specific fitness genes can belong to protein complexes. Can authors test this formally by looking at the overlap with the protein complex standard or protein interaction standard? This would strengthen this statement.
    7. Authors should discuss the reasons why transcription & chromatin remodeling and nuclearcytoplasmic transport, are anticorrelated with genes involved in mitochondrial translation in terms of their fitness profiles and the implications for the evolution of environment-dependent fitness genes.
    8. Authors discuss the limitation of the Hermes system however couldn't they test this system with a different inducible promoter such as estradiol regulated promoter to remove the effect of galactose metabolism?

    Minor comments

    All figures should contain the appropriate colour bars and legends. For example, Figure S5B relies on the colour bar in Figure 5C but it should have its own colour bar.

    Significance

    Significance

    This work provides a comprehensive survey of the variation in natural isolates of yeast and would be interesting to a broad audience studying the genotype-to-phenotype relationship. It is the first study that systematically assessed the fitness effect of loss of function mutations across a large panel of natural isolates providing novel insight into the background specific and environment dependent genes. This represents a valuable resource for the community to ask questions about natural variation in yeast. My expertise is in complex genetic networks in yeast and genome evolution.

  6. Version published to 10.1101/2021.08.25.457624v2 on bioRxiv
  7. Version published to 10.1101/2021.08.25.457624v1 on bioRxiv