A paternal bias in germline mutation is widespread in amniotes and can arise independently of cell division numbers

  1. Marc de Manuel
  2. Felix L Wu
  3. Molly Przeworski

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    This paper challenges a fundamental view concerning why males of most animals have a higher germline mutation rate than females. Evidence is provided to show that it is not simply the fact that males have more cell divisions in the germline, but instead, most of the mutations arise from a different balance of DNA damage vs. DNA repair. The case is supported by data from multiple species, from de novo mutation rate estimates from pedigrees, and from fits to a simple heuristic model. This work will be of interest to the broad field of DNA mutations and DNA repair, as well as evolutionary and phylogenomics researchers.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 and Reviewer #2 agreed to share their names with the authors.)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

In humans and other mammals, germline mutations are more likely to arise in fathers than in mothers. Although this sex bias has long been attributed to DNA replication errors in spermatogenesis, recent evidence from humans points to the importance of mutagenic processes that do not depend on cell division, calling into question our understanding of this basic phenomenon. Here, we infer the ratio of paternal-to-maternal mutations, α , in 42 species of amniotes, from putatively neutral substitution rates of sex chromosomes and autosomes. Despite marked differences in gametogenesis, physiologies and environments across species, fathers consistently contribute more mutations than mothers in all the species examined, including mammals, birds, and reptiles. In mammals, α is as high as 4 and correlates with generation times; in birds and snakes, α appears more stable around 2. These observations are consistent with a simple model, in which mutations accrue at equal rates in both sexes during early development and at a higher rate in the male germline after sexual differentiation, with a conserved paternal-to-maternal ratio across species. Thus, α may reflect the relative contributions of two or more developmental phases to total germline mutations, and is expected to depend on generation time even if mutations do not track cell divisions.

Article activity feed

  1. eLife

    Evaluation Summary:

    This paper challenges a fundamental view concerning why males of most animals have a higher germline mutation rate than females. Evidence is provided to show that it is not simply the fact that males have more cell divisions in the germline, but instead, most of the mutations arise from a different balance of DNA damage vs. DNA repair. The case is supported by data from multiple species, from de novo mutation rate estimates from pedigrees, and from fits to a simple heuristic model. This work will be of interest to the broad field of DNA mutations and DNA repair, as well as evolutionary and phylogenomics researchers.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 and Reviewer #2 agreed to share their names with the authors.)

  2. eLife

    Reviewer #1 (Public Review):

    This manuscript attempts to explain the well-known difference in DNA mutation rates between father vs. mother (paternal mutation is 4 times higher than maternal mutation in humans). Although the mutation rate difference was believed to arrive from the number of cell divisions (male germ cells undergo many more divisions compared to female germ cells), recent studies suggested that most mutations arise from DNA damage (which will be proportional to the absolute time) rather than DNA replication-induced mutations (which will be proportional to the number of cell divisions). The authors thus revisited the question as to why the paternal mutation rate is higher (if absolute time is more important than the number of cell divisions in causing mutations). They used 'taxonomic approaches' comparing paternal/maternal mutation rates of mammals, birds, and reptiles, correlating them to specifics of reproductive mode in these species. To measure paternal vs. maternal mutation rate, they compared the mutation rates of neutrally evolving DNA sequences between the X chromosome vs. autosomes, as well as the Z chromosome (utilizing the fact that the X chromosome will spend twice more generations in females than males, while autosomes spend equal time. Likewise, the Z chromosome will spend twice more time in males than in females, while autosomes spend equal time).

    They first confirm the paternal bias across a broad range of species (amniotes), eliminating many species-specific parameters (longevity, sex chromosome karyotype (XY vs. ZW), etc) as a contributor to the paternal bias. This implies that something common in males in these broad species causes paternal bias. They show that in mammals, the paternal bias correlates with a generation time. They propose that the total mutation is determined by the combination of the mutation rate during early embryogenesis (when both male and female have the same mutation rate) and the later mutation rate when two sexes exhibit different mutation rates. This model seems to explain why generation time correlates well with the extent of paternal bias in mammals. However, this does not explain at all why birds do not exhibit any correlation with a generation time. The speculation on this feels rather weak (although there is nothing they can do about this. Fact is fact).

    The logic behind their analysis is well laid out and seems mostly sound. Their finding is of broad interest in the field.

    - I am confused by this statement (the last sentence in the result section): 'If indeed the developmental window when both sexes have a similar mutation rate is short in birds then, under our model, generation times are expected to have little to no influence on α." Based on their model, if the early period is gone, when the mutation rates are similar between sexes are similar, intuitively it feels that generation time influences α even more. Am I missing something? (if the period with the same mutation rate is gone, then females and males are mutating at different rates the whole time).

    - The authors state that this paper provides a simple explanation as to why paternal biases arise without relying on the number of cell divisions. However, it seems to me that the entire paper relies on the recent findings that mutation arises based on absolute time (instead of cell division number), and the novelty in this paper is the idea of 'two-phase mutation rates' to explain the observed numbers of paternal bias in various species. Yet it fails to explain the mutation rate difference in birds. There is not enough speculation or explanation as to what determines different mutation rates in males of various species. Although the modeling seems to be sound and there is nothing that can be done experimentally, I felt somewhat unsatisfied at the end of the manuscript.

  3. eLife

    Reviewer #2 (Public Review):

    The primary goal of this paper is to re-assess the cause for the excess of male over female germline mutations seen in many animals. By re-analyzing X (Z) and autosomal substitution rates across 42 species of mammals, birds, and snakes, and fitting a model that allows for a constant and equal-sex embryonic mutation rate, along with a mutation rate that increases with age, the authors show that there is no need to invoke the model that assumes mutation rate depends strictly on numbers of cell divisions.

    Strengths
    1. The paper challenges a dogma in evolutionary genomics, which states that males have a higher germline mutation rate than females. It establishes convincingly that the count of pre-meiotic mitotic divisions is NOT the primary driver of the excess male mutations, but instead, it is the intrinsic mutation rate in males (balance of DNA damage vs DNA repair) that accumulates over time.

    2. The authors establish a simple model where the number of mutations that accumulate each generation depends on the embryonic mutation rate (which is shown empirically to not differ between the sexes) and a post-maturity mutation rate, which has elevated male mutation (driven presumably by a shift in the balance between DNA damage and DNA repair). The model is very clear and intuitive described.

    3. The paper is extremely carefully thought-out, planned, and executed. Criteria for inclusion and exclusion of species in the phylogenetic work are clearly laid out. Similarly, decisions about filtering genomic regions (avoiding repeats, etc.) are well done and exhaustively documented. The standard of scholarship is very high - for example, the analysis of de novo mutation rates in mammals pulled in data from no fewer than 15 published studies.

    Weaknesses
    1. The method of estimating alpha relies on the assumption that the mutation process (and rates) are the same in autosomes and sex chromosomes. There is an attempt to control for GC content and replication timing, but it is easy to imagine other factors at play, including the inactivation of one X in females, the extensive differences in chromatin modifications, especially of the X, that differ in males vs. females. The case of the cat X chromosome, with its 50 Mb of recombination cold spot and corresponding oddly slow substitution rate, might be just one example of features in other species that cause other perturbations in the substitution rate of the X. This does not seriously erode confidence in the results, but there is more potential for intrinsic mutation rates of sex chromosomes and autosomes to differ than is suggested by the authors.

    2. The authors point out that the human mutations in spermatogonia are due to mutation signatures SBS5/40 ( which are known not to be correlated with cell division rates). The work on the nonhuman species could be greatly extended with this mutation spectrum approach. For each species, one could ask: Are the mutation spectra of the embryonic mutations consistent between males and females? What about the mutation spectra for the post-puberty individuals? Is alpha consistent across mutation signatures? Does the GC bias correction impact these inferences?

    3. While the data do not suggest reasons WHY males display a higher mutation rate, it is fair to ask whether the evolutionary drive for a higher mutation rate might shape the mechanism whereby it happens. There is a certain amount of speculation in the paper as it is, and it is done in a way that is often well supported by data after the fact. Speculation about why males have an elevated mutation rate would not erode the overall quality of the paper, and I would expect that many readers would be eager to see what the authors have to say on the subject.

    Overall the paper achieves its intended goal of toppling the dogma that the excess male mutation rate is driven by number of rounds of cell division in spermatogenesis (compared to oogenesis).

  4. eLife

    Reviewer #3 (Public Review):

    The authors critically assessed a widespread assumption that paternal biases in the number of germline mutations passed to offspring and the number of germline cell divisions have a causal link. They gather a diverse set of previously published findings that are inconsistent with this assumption, including the accumulation of maternal DNMs with age, the consistent ratio of paternal-to-maternal germline mutation (α) in humans, the range of α in mammals, and the dominance of mutational processes that are uncorrelated to cell division in human germline and somatic tissues. They then generate estimates of α based on evolutionary rates at sex chromosomes vs autosomes. They find αevo of 1-4 across the species considered, which are robust to changes/exclusion of a number of potentially confounding factors. They find an increase in αevo with generation time in mammals but not in birds. The authors consider and evaluate a model with a fixed number of early mutations for both sexes followed by post sexual differentiation stage with a paternal mutation bias.

  5. Version published to 10.7554/elife.80008 on eLife
  6. eLife

    Author Response

    Reviewer 1

    This manuscript attempts to explain the well-known difference in DNA mutation rates between father vs. mother (paternal mutation is 4 times higher than maternal mutation in humans). Although the mutation rate difference was believed to arrive from the number of cell divisions (male germ cells undergo many more divisions compared to female germ cells), recent studies suggested that most mutations arise from DNA damage (which will be proportional to the absolute time) rather than DNA replication-induced mutations (which will be proportional to the number of cell divisions). The authors thus revisited the question as to why the paternal mutation rate is higher (if absolute time is more important than the number of cell divisions in causing mutations). They used 'taxonomic approaches' comparing paternal/maternal mutation rates of mammals, birds, and reptiles, correlating them to specifics of reproductive mode in these species. To measure paternal vs. maternal mutation rate, they compared the mutation rates of neutrally evolving DNA sequences between the X chromosome vs. autosomes, as well as the Z chromosome (utilizing the fact that the X chromosome will spend twice more generations in females than males, while autosomes spend equal time. Likewise, the Z chromosome will spend twice more time in males than in females, while autosomes spend equal time).

    They first confirm the paternal bias across a broad range of species (amniotes), eliminating many species-specific parameters (longevity, sex chromosome karyotype (XY vs. ZW), etc) as a contributor to the paternal bias. This implies that something common in males in these broad species causes paternal bias. They show that in mammals, the paternal bias correlates with a generation time. They propose that the total mutation is determined by the combination of the mutation rate during early embryogenesis (when both male and female have the same mutation rate) and the later mutation rate when two sexes exhibit different mutation rates. This model seems to explain why generation time correlates well with the extent of paternal bias in mammals. However, this does not explain at all why birds do not exhibit any correlation with a generation time. The speculation on this feels rather weak (although there is nothing they can do about this. Fact is fact).

    The logic behind their analysis is well laid out and seems mostly sound. Their finding is of broad interest in the field.

    • I am confused by this statement (the last sentence in the result section): 'If indeed the developmental window when both sexes have a similar mutation rate is short in birds then, under our model, generation times are expected to have little to no influence on α." Based on their model, if the early period is gone, when the mutation rates are similar between sexes are similar, intuitively it feels that generation time influences α even more. Am I missing something? (if the period with the same mutation rate is gone, then females and males are mutating at different rates the whole time).

    We apologize for the lack of clarity, as we should have made clear that here we are assuming a fixed ratio of paternal to maternal generation times. Under that assumption, if female and male germ cells are accumulating mutations as a fixed rate over time, then for each sex, the number of mutations accumulated with time is a line that goes through the origin, and the ratio of the paternal-to-maternal slopes (α) will be constant regardless of the age of reproduction. In other words, if Me=0 in equation 1, then α would be constant for any fixed ratio Gm/Gf. We have revised this sentence to be clearer; lines 334-338 now read:

    “If indeed the mutation rate in the two bird sexes differs from very early on in development (i.e., if term Me ≈ 0 in equation 1), then assuming a fixed ratio of paternal-to-maternal generation times, our model predicts the sex-averaged age of reproduction will have little to no influence on α.”

    • The authors state that this paper provides a simple explanation as to why paternal biases arise without relying on the number of cell divisions. However, it seems to me that the entire paper relies on the recent findings that mutation arises based on absolute time (instead of cell division number), and the novelty in this paper is the idea of 'two-phase mutation rates' to explain the observed numbers of paternal bias in various species. Yet it fails to explain the mutation rate difference in birds. There is not enough speculation or explanation as to what determines different mutation rates in males of various species. Although the modeling seems to be sound and there is nothing that can be done experimentally, I felt somewhat unsatisfied at the end of the manuscript.

    We agree with the reviewer that our paper does not address why the ratio of paternal-tomaternal mutation rates is lower in birds than mammals, and had stated so explicitly (lines 358360): “Another question raised by our findings is why, after sexual differentiation of the germline, mutation appears to be more paternally-biased in mammals (∼4:1) than in birds and snakes (∼2:1).”

    To try to gain more insight into this question, we are now analyzing mutations in a set of three generation pedigrees from birds and reptiles, which should allow us to obtain a direct estimate of α and characterize sex differences in the mutation spectra, which we can then compare to what is seen in mammals. While this analysis is beyond the scope of this manuscript, we now note how this question might be pursued (lines 360-362):

    “In that regard, it will be of interest to collect pedigree data from these taxa, with which to compare mutation signatures to those typically seen in mammals.”

    Reviewer 2 The primary goal of this paper is to re-assess the cause for the excess of male over female germline mutations seen in many animals. By re-analyzing X (Z) and autosomal substitution rates across 42 species of mammals, birds, and snakes, and fitting a model that allows for a constant and equal-sex embryonic mutation rate, along with a mutation rate that increases with age, the authors show that there is no need to invoke the model that assumes mutation rate depends strictly on numbers of cell divisions.

    Strengths

    1. The paper challenges a dogma in evolutionary genomics, which states that males have a higher germline mutation rate than females. It establishes convincingly that the count of premeiotic mitotic divisions is NOT the primary driver of the excess male mutations, but instead, it is the intrinsic mutation rate in males (balance of DNA damage vs DNA repair) that accumulates over time.
    1. The authors establish a simple model where the number of mutations that accumulate each generation depends on the embryonic mutation rate (which is shown empirically to not differ between the sexes) and a post-maturity mutation rate, which has elevated male mutation (driven presumably by a shift in the balance between DNA damage and DNA repair). The model is very clear and intuitive described.
    1. The paper is extremely carefully thought-out, planned, and executed. Criteria for inclusion and exclusion of species in the phylogenetic work are clearly laid out. Similarly, decisions about filtering genomic regions (avoiding repeats, etc.) are well done and exhaustively documented. The standard of scholarship is very high - for example, the analysis of de novo mutation rates in mammals pulled in data from no fewer than 15 published studies.

    Weaknesses

    1. The method of estimating alpha relies on the assumption that the mutation process (and rates) are the same in autosomes and sex chromosomes. There is an attempt to control for GC content and replication timing, but it is easy to imagine other factors at play, including the inactivation of one X in females, the extensive differences in chromatin modifications, especially of the X, that differ in males vs. females. The case of the cat X chromosome, with its 50 Mb of recombination cold spot and corresponding oddly slow substitution rate, might be just one example of features in other species that cause other perturbations in the substitution rate of the X. This does not seriously erode confidence in the results, but there is more potential for intrinsic mutation rates of sex chromosomes and autosomes to differ than is suggested by the authors.

    We agree with the reviewer that despite our attempts, we do not control for all factors that distinguish X and autosomes beyond exposure to sex. We had written that “while our pipeline may not account for all the differences between autosomes and X (Z) chromosomes unrelated to sex differences in mutation, the qualitative patterns are reliable.” and have now included a sentence to make this limitation clearer (lines 165-167):

    “Nonetheless, it is unlikely that our regression model perfectly accounts for all the genomic features that differ between sex chromosomes and autosomes other than exposure to sex.”

    In turn, the assumption that mutation rates in X (Z) and autosomes differ only with regard to their exposure to sex (after accounting for base composition and other genomic features) is unproven; we now state this assumption explicitly in the Methods (lines 678-681). Nonetheless, it seems warranted by the high concordance of evolutionary- and pedigree-based estimates of alpha in humans, mice and cattle. With regard to the specific factors mentioned by the reviewer, excluding CpG sites has little effect on our qualitative conclusions for mammals (see Fig S1E), suggesting that DNA methylation differences between X and autosomes are not having a major influence on our findings. Moreover, X-inactivation in the germline of mammals (as distinct from the soma) is likely quite short-lived, given that it lasts around three days in early development of mice (Chuva de Sousa Lopes et al. 2008) and at most four weeks in humans (Guo et al. 2015). Thus, it is unlikely to be an important mutation rate modifier. We have now reworked three paragraphs in the main text to make the limitations above clearer (lines 127-175).

    1. The authors point out that the human mutations in spermatogonia are due to mutation signatures SBS5/40 ( which are known not to be correlated with cell division rates). The work on the nonhuman species could be greatly extended with this mutation spectrum approach. For each species, one could ask: Are the mutation spectra of the embryonic mutations consistent between males and females? What about the mutation spectra for the post-puberty individuals? Is alpha consistent across mutation signatures? Does the GC bias correction impact these inferences?

    Unfortunately, there is not enough de novo data to address this question outside of humans. In turn, the analysis of substitution data is unreliable, because of the differential impact of repeated substitutions at a site and the effects of GC-biased gene conversion.

    1. While the data do not suggest reasons WHY males display a higher mutation rate, it is fair to ask whether the evolutionary drive for a higher mutation rate might shape the mechanism whereby it happens. There is a certain amount of speculation in the paper as it is, and it is done in a way that is often well supported by data after the fact. Speculation about why males have an elevated mutation rate would not erode the overall quality of the paper, and I would expect that many readers would be eager to see what the authors have to say on the subject.

    As we envisage it, along the lines of Lynch’s models for the evolution of germline mutation (Lynch 2010), there is likely selection to keep the mutation rate as low as possible, subject to the constraints of the need to replicate DNA, repair damage, etc. efficiently. Why the attainable lower limit would be higher in males than in females is unclear to us, both mechanistically and in terms of evolutionary selection pressures. As we now note lines 353-355, a potential proximal cause is a greater effect of reactive oxygen species, a major source of DNA damage, in male germ cells than in oocytes (Smith et al. 2013; Rodríguez-Nuevo et al. 2022). Potential evolutionary causes are even less clear to us, but could be related to the greater competition among sperm vs. oocytes (added in lines 354-357).

    Another way to think about these results is as shifting the question somewhat, broadening it from the long-standing puzzle of the selection pressures shaping sex differences to asking what determines the relative mutation rates of different cell types, including oocytes and spermatagonia but also somatic cell types/tissues. We had previously written that “our results recast long standing questions about the source of sex bias in germline mutations as part of a larger puzzle about why certain cell types (here, spermatogonia versus oocytes) accrue more mutations than others.” We have revised the final paragraph of the Discussion to try to emphasize this point.

    Overall the paper achieves its intended goal of toppling the dogma that the excess male mutation rate is driven by number of rounds of cell division in spermatogenesis (compared to oogenesis).

  11. Version published to 10.1101/2022.02.07.479417v2 on bioRxiv
  12. Version published to 10.1101/2022.02.07.479417v1 on bioRxiv