Evolutionary transcriptomics implicates new genes and pathways in human pregnancy and adverse pregnancy outcomes
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
The study by Mika and colleagues uses a comparative transcriptomics approach to identify changes in the expression of genes that specifically occurred during the evolution of the human endometrium. The authors find that hundreds of genes gained or lost endometrial expression in the human lineage and that several of these genes are potentially implicated in the pathophysiology of human pregnancy. The study contributes to ongoing interest in the effect of human evolution on the pathophysiology of human pregnancy, and has the potential to serve as a model of how to study the evolution of pregnancy-associated genomic changes in particular species and tissues.
(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, Reviewer #2 and Reviewer #3 agreed to share their names with the authors.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
- Evolutionary Biology (eLife)
Abstract
Evolutionary changes in the anatomy and physiology of the female reproductive system underlie the origins and diversification of pregnancy in Eutherian (‘placental’) mammals. This developmental and evolutionary history constrains normal physiological functions and biases the ways in which dysfunction contributes to reproductive trait diseases and adverse pregnancy outcomes. Here, we show that gene expression changes in the human endometrium during pregnancy are associated with the evolution of human-specific traits and pathologies of pregnancy. We found that hundreds of genes gained or lost endometrial expression in the human lineage. Among these are genes that may contribute to human-specific maternal–fetal communication ( HTR2B ) and maternal–fetal immunotolerance ( PDCD1LG2 ) systems, as well as vascular remodeling and deep placental invasion ( CORIN ). These data suggest that explicit evolutionary studies of anatomical systems complement traditional methods for characterizing the genetic architecture of disease. We also anticipate our results will advance the emerging synthesis of evolution and medicine (‘evolutionary medicine’) and be a starting point for more sophisticated studies of the maternal–fetal interface. Furthermore, the gene expression changes we identified may contribute to the development of diagnostics and interventions for adverse pregnancy outcomes.
Article activity feed
-
Author Response:
Reviewer #1 (Public Review):
The authors have conducted an investigation in to the impact of evolution on the endometrium.
A major strength of this work is the use of published single cell RNA seq and ChIPseq data sets to support their findings. The major weakness is the use of an algorithmic approach to deconvolve a specific endometrial signal.
The authors have broadly achieved their aims and the results support the conclusions. The weakness with the algorithmic approach to determine specific transcriptomic signal have been carefully addressed and the data presented in figure 1 is persuasive. I would be still slightly concerned that issues with the comparison of a receptive and non-receptive endometrium have not been fully accounted for. It would be nice to see this. It would also be of interest to understand …
Author Response:
Reviewer #1 (Public Review):
The authors have conducted an investigation in to the impact of evolution on the endometrium.
A major strength of this work is the use of published single cell RNA seq and ChIPseq data sets to support their findings. The major weakness is the use of an algorithmic approach to deconvolve a specific endometrial signal.
The authors have broadly achieved their aims and the results support the conclusions. The weakness with the algorithmic approach to determine specific transcriptomic signal have been carefully addressed and the data presented in figure 1 is persuasive. I would be still slightly concerned that issues with the comparison of a receptive and non-receptive endometrium have not been fully accounted for. It would be nice to see this. It would also be of interest to understand the impact of diapause on this analysis. Can the authors comment?
The evolutionary impact on the developing endometrium is of major importance to translational investigation of adverse events during human pregnancy. The methods presented are well described and straight forward for a computational group to follow.
We note that all RNA-Seq datasets used in the evolutionary analyses were from pregnant endometria, thus there is no need to account for or way to compare receptive and non-receptive endometrium samples or diapause (but this is a very interesting question for future studies!).
Reviewer #3 (Public Review):
Strengths:
Mika and colleagues used a comparative transcriptomics approach to identify genes (based on binary {plus minus} expression calls) that were recruited or eliminated in the evolutionary biology of the human endometrium. The recruited genes were then analyzed for potential roles in pregnancy pathophysiology using bioinformatic approaches. The study contributes to ongoing interest in the effect of human evolution on the pathophysiology of human pregnancy, and it is proposed that evolutionary studies of this kind, in combination with traditional methods, can be used to better characterize the genetic architecture of disease.
Weaknesses:
The conclusions of the paper are mostly supported by the analyses. However, it is unclear how the evolution of endometrial cell gene expression would contribute to adverse pregnancy outcomes since such conditions would compromise reproduction and therefore be selected against.'
We apologize if this reasoning was unclear. Our hypothesis is that genes that gained (or lost) endometrial expression in the human lineage will be important for the establishment, maintenance, and cessation of pregnancy. Their contribution to adverse pregnancy outcomes would be through mis-expression leading to dysfunction of the pathways they regulate. Variants that lead to mis-expression will then be selected against, although the efficacy of that negative selection will depend on numerous factors such as population size, drift, penetrance, and the effect size of the mutation.
It is stated that hundreds of genes that gained or lost endometrial expression in the human lineage were identified but these are not listed.
Genes that gained and lost expression were given in Figure 1 – Source data 2. However, we have added the files Figure 2 – Source data 2, which lists the HUGO gene names for genes that gained expression (BPP ≥ 0.80) in the Hominoid (human) lineage, and Figure 2 – Source data 3, which lists the HUGO gene names for genes that lost expression (BPP ≥ 0.80) in the Hominoid (human) lineage. We hope this makes these results more accessible to a broad audience.
Three genes were examined in detail for their roles in pregnancy and human-specific maternal-fetal communication but the rationale for selecting these genes is lacking.
We selected HTR2B and PDCD1LG2 because they had not previously been implicated in pregnancy, demonstrating the importance of explicit evolutionary studies for discovery of genes important for tissue and organ function, and CORIN because it had previously been shown to be important for pre-eclampsia but has restricted endometrial expression across species, demonstrating the importance of evolutionary studies for understanding the conservation of gene expression in tissue and organ systems across species.
The uncertain quality of the source transcriptome data is a weakness. The level of transcriptome "noise" in the data sets is unclear. It appears that the transcriptome data from most species was from bulk tissue total RNA and stage of pregnancy and anatomical site (e.g., over the placenta or at the fetal membranes) is not specified. Dissecting and isolating pregnancy endometrium is not trivial and as such this is a likely source of significant variation. Data on placenta-specific gene expression is provided to demonstrated lack of trophoblast cell contamination, however, this does not mean that the RNA was exclusively from endometrial cells since numerous non-endometrial cells are present a the maternal-fetal interface. Consequently, binary gene expression as on/off based on a 2 TPM threshold is problematic since it may be affected by the proportion of endometrial cells in the sample rather than gene expression in endometrial cells. In addition, although the application of binary encoding is understandable, important biology may be missed because gene function extends beyond on/off state.
We agree that there is likely significant “noise” in the cross-species gene expression data for all of the reasons the reviewer indicates. Indeed, the MDS plot shown for gene expression levels (expressed as TPMs) in Figure 1 – Figure 1A indicates there is significant noise in the data that overwhelms phylogenetic signal. However, binary encoding reduces the noise and unmasks phylogenetic signal as shown in Figure 1 – Figure 1B. Thus, binary encoding at least partly alleviates the noise problem as well as variation from different sampling locations.
We also agree that binary encoding will be affected by the proportion of different cell-types in each sample and that it is almost certain that the proportion of different cell-types will vary between species. Again, however, this is an argument in favor of binary encoding because it will reduce noise in quantitative gene expression estimates that arise from proportional cell-type differences between species. We note, however, that we do not claim all or even most of the gene expression changes we have identified are in endometrial cells. The data does, however, suggest that gene expression changes is enriched in endometrial cells, which guided our focus in gene expression in this cell-type. This is also one of the reasons we followed up with scRNA-Seq data, to identify which recruited genes were expressed in which cell-types, and confirmed expression of HTR2B, PDCD1LG2, and CORIN, in endometrial cells.
Finally, we also agree that binary encoding is guaranteed to miss quantitative gene expression differences between species that are important for the biology of pregnancy. However, our goal in this study was to focus on more gross gene expression changes which may have larger effect sizes than quantitative gene expression changes (although that is an assumption that itself requires validation).
Use of the Vento-Tormo scRNAseq data set (Nature 2018, 563:347-353) to establish the first trimester endometrial cell transcriptome is a strength. The study would be improved, however, if those data were compared with the term maternal-fetal interface scRNAseq data set produced by the Gomez-Lopez group (Pique-Regi et al. eLife 2019;8:e52004).
We have attempted several different methods to directly compare the VentoTormo and Pique-Regi scRNA-Seq datasets, and agree that this would be a potentially interesting comparison. Unfortunately, however, while the VentoTormo dataset is publicly available without restriction, the Pique-Regi dataset is only available through dbGAP. The data restrictions imposed by dbGAP require requesting (and being approved for) data access, which was not granted before submission of our revised manuscript. Therefore, we were unable to compare/contrast these datasets.
In Caveats and Limitation, the authors admit that they are unable to identify truly human-specific gene expression changes in pregnancy endometrium, yet sweeping conclusions are made about the changing transcriptome of the human endometrium and how the presumed changes in gene expression contribute to extant pathophysiology. The claim that the comparative transcriptomic approach (based on binary gene expression) provides and insight into human pathophysiology is therefore questionable.
While we agree with the reviewer that we interchangeably using human and human lineage specific when the taxon sampling only allows us to identify changes in apes (Hominoidea), we were writing for a more general audience than evolutionary biologists who might be unfamiliar with primate taxonomic nomenclature. Indeed, this is why we included the statement in the Caveats and Limitations section.
-
-
Evaluation Summary:
The study by Mika and colleagues uses a comparative transcriptomics approach to identify changes in the expression of genes that specifically occurred during the evolution of the human endometrium. The authors find that hundreds of genes gained or lost endometrial expression in the human lineage and that several of these genes are potentially implicated in the pathophysiology of human pregnancy. The study contributes to ongoing interest in the effect of human evolution on the pathophysiology of human pregnancy, and has the potential to serve as a model of how to study the evolution of pregnancy-associated genomic changes in particular species and tissues.
(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 …
Evaluation Summary:
The study by Mika and colleagues uses a comparative transcriptomics approach to identify changes in the expression of genes that specifically occurred during the evolution of the human endometrium. The authors find that hundreds of genes gained or lost endometrial expression in the human lineage and that several of these genes are potentially implicated in the pathophysiology of human pregnancy. The study contributes to ongoing interest in the effect of human evolution on the pathophysiology of human pregnancy, and has the potential to serve as a model of how to study the evolution of pregnancy-associated genomic changes in particular species and tissues.
(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, Reviewer #2 and Reviewer #3 agreed to share their names with the authors.)
-
Reviewer #1 (Public Review):
The authors have conducted an investigation in to the impact of evolution on the endometrium.
A major strength of this work is the use of published single cell RNA seq and ChIPseq data sets to support their findings. The major weakness is the use of an algorithmic approach to deconvolve a specific endometrial signal.
The authors have broadly acheived their aims and the results support the conclusions. The weakness with the alogrithmic approach to determine specific transcriptomic signal have been carefully addressed and the data presented in figure 1 is persuassive. I would be still slightly concerned that issues with the comparison of a receptive and non-receptive endometrium have not been fully accounted for. It would be nice to see this. It would also be of interest to understand the impact of diapause on …
Reviewer #1 (Public Review):
The authors have conducted an investigation in to the impact of evolution on the endometrium.
A major strength of this work is the use of published single cell RNA seq and ChIPseq data sets to support their findings. The major weakness is the use of an algorithmic approach to deconvolve a specific endometrial signal.
The authors have broadly acheived their aims and the results support the conclusions. The weakness with the alogrithmic approach to determine specific transcriptomic signal have been carefully addressed and the data presented in figure 1 is persuassive. I would be still slightly concerned that issues with the comparison of a receptive and non-receptive endometrium have not been fully accounted for. It would be nice to see this. It would also be of interest to understand the impact of diapause on this analysis. Can the authors comment?
The evolutionary impact on the devloping endometrium is of major importance to translational investigation of advserse events during human pregnancy. The methods presented are well described and straight forward fpor a computional group to follow.
-
Reviewer #2 (Public Review):
The authors of this manuscript did a wonderful job explaining how gene expression patterns have evolved during recent mammalian evolution, particularly within primates. They used a wide variety of approaches including tissue-based RNA-Seq data, single cell omics, and cell and culture approaches to make the case that a number of changes have occurred during the evolutionary history of eutherian mammals. Most interesting was the finding that the serotonin system, normally associated with neuronal function, appears to play a large role in the pregnant endometrium. Additionally, the roles of genes involved in maternal fetal immune-tolerance and tissue remodeling were confirmed. The methods used for data analysis are rigorous and reproducible and the conclusions of the study are warranted. All data are publicly …
Reviewer #2 (Public Review):
The authors of this manuscript did a wonderful job explaining how gene expression patterns have evolved during recent mammalian evolution, particularly within primates. They used a wide variety of approaches including tissue-based RNA-Seq data, single cell omics, and cell and culture approaches to make the case that a number of changes have occurred during the evolutionary history of eutherian mammals. Most interesting was the finding that the serotonin system, normally associated with neuronal function, appears to play a large role in the pregnant endometrium. Additionally, the roles of genes involved in maternal fetal immune-tolerance and tissue remodeling were confirmed. The methods used for data analysis are rigorous and reproducible and the conclusions of the study are warranted. All data are publicly available furthering the transparency of the author's approach. For those interested in the evolution of pregnancy this is an excellent model of how to study particular species and tissues.
-
Reviewer #3 (Public Review):
Strengths:
Mika and colleagues used a comparative transcriptomics approach to identify genes (based on binary {plus minus} expression calls) that were recruited or eliminated in the evolutionary biology of the human endometrium. The recruited genes were then analyzed for potential roles in pregnancy pathophysiology using bioinformatic approaches. The study contributes to ongoing interest in the effect of human evolution on the pathophysiology of human pregnancy, and it is proposed that evolutionary studies of this kind, in combination with traditional methods, can be used to better characterize the genetic architecture of disease.
Weaknesses:
The conclusions of the paper are mostly supported by the analyses. However, it is unclear how the evolution of endometrial cell gene expression would contribute to …
Reviewer #3 (Public Review):
Strengths:
Mika and colleagues used a comparative transcriptomics approach to identify genes (based on binary {plus minus} expression calls) that were recruited or eliminated in the evolutionary biology of the human endometrium. The recruited genes were then analyzed for potential roles in pregnancy pathophysiology using bioinformatic approaches. The study contributes to ongoing interest in the effect of human evolution on the pathophysiology of human pregnancy, and it is proposed that evolutionary studies of this kind, in combination with traditional methods, can be used to better characterize the genetic architecture of disease.
Weaknesses:
The conclusions of the paper are mostly supported by the analyses. However, it is unclear how the evolution of endometrial cell gene expression would contribute to adverse pregnancy outcomes since such conditions would compromise reproduction and therefore be selected against.'
It is stated that hundreds of genes that gained or lost endometrial expression in the human lineage were identified but these are not listed. Three genes were examined in detail for their roles in pregnancy and human-specific maternal-fetal communication but the rationale for selecting these genes is lacking.
The uncertain quality of the source transcriptome data is a weakness. The level of transcriptome "noise" in the data sets is unclear. It appears that the transcriptome data from most species was from bulk tissue total RNA and stage of pregnancy and anatomical site (e.g., over the placenta or at the fetal membranes) is not specified. Dissecting and isolating pregnancy endometrium is not trivial and as such this is a likely source of significant variation. Data on placenta-specific gene expression is provided to demonstrated lack of trophoblast cell contamination, however, this does not mean that the RNA was exclusively from endometrial cells since numerous non-endometrial cells are present a the maternal-fetal interface. Consequently, binary gene expression as on/off based on a 2 TPM threshold is problematic since it may be affected by the proportion of endometrial cells in the sample rather than gene expression in endometrial cells. In addition, although the application of binary encoding is understandable, important biology may be missed because gene function extends beyond on/off state.
Use of the Vento-Tormo scRNAseq data set (Nature 2018, 563:347-353) to establish the first trimester endometrial cell transcriptome is a strength. The study would be improved, however, if those data were compared with the term maternal-fetal interface scRNAseq data set produced by the Gomez-Lopez group (Pique-Regi et al. eLife 2019;8:e52004).
In Caveats and Limitation, the authors admit that they are unable to identify truly human-specific gene expression changes in pregnancy endometrium, yet sweeping conclusions are made about the changing transcriptome of the human endometrium and how the presumed changes in gene expression contribute to extant pathophysiology. The claim that the comparative transcriptomic approach (based on binary gene expression) provides and insight into human pathophysiology is therefore questionable.
-
-