The relationship between phenotypic variation arising through individual development and phenotypic variation arising through diversification of species has long been a central question in evolutionary biology. Among humans, reduced placental invasion into endometrial tissues is associated with diseases of pregnancy, especially pre-eclampsia, and reduced placental invasiveness has also evolved, convergently, in at least 10 lineages of eutherian mammals. We tested the hypothesis that a common genetic basis underlies both reduced placental invasion arising through a developmental process in human placental disease and reduced placental invasion found as a derived trait in the diversification of Euarchontoglires (rodents, lagomorphs, tree shrews, colugos and primates). Based on whole-genome analyses across 18 taxa, we identified 1254 genes as having evolved adaptively across all three lineages exhibiting independent evolutionary transitions towards reduced placental invasion. These genes showed strong evidence of enrichment for associations with pre-eclampsia, based on genetic-association studies, gene-expression analyses and gene ontology. We further used in silico prediction to identify a subset of 199 genes that are likely targets of natural selection during transitions in placental invasiveness and which are predicted to also underlie human placental disorders. Our results indicate that abnormal ontogenies can recapitulate major phylogenetic shifts in mammalian evolution, identify new candidate genes for involvement in pre-eclampsia, imply that study of species with less-invasive placentation will provide useful insights into the regulation of placental invasion and pre-eclampsia, and recommend a novel comparative functional-evolutionary approach to the study of genetically based human disease and mammalian diversification.
Phenotypic diversification between taxa may arise from the evolution of developmental programmes such that the genetic systems underlying homologous traits differ across taxa . Alternatively, variation in the regulation and ontological organization of a developmental system may give rise to divergent phenotypic outcomes that have a shared underlying genetic basis. Within species, this situation is best exemplified by polyphenism [2,3], but ontogenetic shifts in complex genetic systems are also likely to be important in generating interspecific phenotypic variation [4,5]. A number of recent studies indicate that evolutionary variation of a stable underlying genetic system can give rise to repeated convergent evolution in complex traits such as Drosophila wing pigmentation patterns, bee eusociality and stickleback body armour [6–8]. Establishing a common genetic basis to phenotypically divergent outcomes in human and non-human species is especially important in medical research that makes use of animal models. Here, we show that three independent convergent evolutionary transitions towards reduced placental invasiveness—in rodents, primates and tree shrews—share a common genetic basis with human diseases of placental invasion, suggesting that the ontogeny of human diseases can be recapitulated in large-scale mammalian morphological evolution.
The notable diversity found in mammalian placental development provides outstanding opportunities to analyse the relationship between phylogeny and ontogeny in this context. Phylogenetic reconstruction indicates that the earliest members of the superorder Euarchontoglires (comprising rodents, lagomorphs, tree shrews, colugos and primates) bore an invasive hemochorial placenta in which the fetal trophoblast is bathed directly in maternal blood, a situation found in a majority of extant Euarchontogliran mammals including humans . The phylogeny of this group includes three independent evolutionary transitions towards less-invasive endotheliochorial or epitheliochorial placentation—in tree shrews, strepsirhine primates and heteromyid rodents—in which maternal blood is physically separated from fetal tissues during gestation (figure 1). Reduced invasion of maternal vessels by fetal trophoblast tissue is characteristic not just of these derived forms of placentation but is also a central feature of pre-eclampsia in humans. The pre-eclamptic placenta, while remaining hemochorial, is characterized by markedly reduced invasion of the placental bed and reduced remodelling of the maternal vasculature by fetal tissues, which in healthy placentation supports nutrient supply to the fetus during pregnancy [10–14]. Although the precise causes of pre-eclampsia remain uncertain, symptoms accompanying reduced placental invasion include inappropriate expression of an immunologically pro-inflammatory and anti-angiogenic cytokine profile, misregulation of cell adhesion and apoptotic processes at the fetal–maternal interface, kidney damage and proteinuria, and maternal hypertension resulting in vascular damage, convulsions and sometimes death . Hypertensive disorders of pregnancy are a leading cause of maternal and fetal morbidity and mortality, accounting for 9.1% of maternal deaths in Africa and Asia, 25.7% in Latin America and 16.1% in developed countries . Pre-eclampsia is often regarded as a condition unique to humans (and perhaps great apes: [17,18]), and the apparent absence of naturally occurring analogues in non-human model species has hampered the study of its aetiology and the development of effective treatments .
Previous studies on the genetic basis of interspecific placental variation have attempted to identify lineage-specific genes or transcript variants that may underlie the development of placental structures unique to taxa including elephants , carnivores , cattle [21,22] and haplorhine primates [23,24]. Based on similarities in the changes to physiological systems that underlie reduced invasiveness of non-hemochorial placentation in Euarchontoglires and also reduced placental invasion in pathological, human hemochorial placentation—especially modifications to uterine vascular patterning, reduced motility of fetally derived trophoblast in the endometrium and altered immunological relations between mother and fetus—we here address the hypothesis that variation across and within species in placental invasiveness depends not merely on lineage-specific adaptation but also on variation within a core, overlapping set of genetic systems that regulate placental invasion.
To identify a set of genes undergoing adaptive evolution during the losses of hemochorial placentation in Euarchontoglires, we conducted genome-wide phylogenetic tests for statistical signals of positive selection specifically within the three lineages associated with the origin of endotheliochorial or epitheliochorial placentation. We then tested for involvement of these genes in the ontogeny of placental invasiveness, by determining whether they show genetic or gene-expression associations with pre-eclampsia.
2. Material and methods
(a) Sequence alignments
The Ensembl Perl application programming interface  was used to generate, for each human gene identifier, a file containing collected unaligned one-to-one orthologous protein-coding sequences from the 18 Euarchontogliran taxa available: pika (Ochotona princeps), rabbit (Oryctolagus cuniculus), ground squirrel (Spermophilus tridecemlineatus), guinea pig (Cavia porcellus), kangaroo rat (Dipodomys ordii), mouse (Mus musculus), rat (Rattus norvegicus), tree shrew (Tupaia belangeri), lemur (Microcebus murinus), galago (Otolemur garnettii), tarsier (Tarsius syrichta), marmoset (Callithrix jacchus), macaque (Macaca mulatta), gibbon (Nomascus leucogenys), orangutan (Pongo abelii), gorilla (Gorilla gorilla), human (Homo sapiens) and chimpanzee (Pan troglodytes). Each file was scanned for the presence of non-hemochorial species (kangaroo rat, tree shrew, lemur and galago), and files were discarded from further analysis if they contained no sequence data for these species. The remaining files were each aligned using a codon model in the probabilistic alignment application PRANK . Tests for positive selection are sensitive to alignment quality. A recent review  found PRANK's codon model (which aligns codons rather than nucleotides or amino acids, thus making use of information from both the primary nucleotide sequence and also the translation) to yield the most consistently high-quality alignments, with respect to success in inference of sitewise positive selection, of a number of methods assessed. Resultant alignments were manually checked for quality, and low-quality regions for which 50% or more species exhibited gaps were excised from the alignments prior to further analysis.
(b) Detection of selection
For each gene, a base phylogenetic tree (figure 1) was pruned according to the sequence data available. Focal branches upon which transitions from hemochorial to non-hemochorial placentation occurred (the branch leading to kangaroo rat, the branch leading to tree shrew and the branch leading to strepsirhine primates) were labelled as foreground branches for analysis in PAML (phylogenetic analysis by maximum likelihood) . Owing to the genomic data and/or inference of one-to-one othologues by Ensembl being incomplete, the resultant phylogeny varied across genes in the number and identity of terminal taxa. For 2093 genes, a phylogeny bearing one transition from hemochorial to non-hemochorial placentation was available, for 4293 genes a phylogeny bearing two such transitions was available and for 10 192 genes a phylogeny bearing all three possible transitions was available.
PAML was used to fit a branch-sites positive selection model (in PAML, model = 2 and NSsites = 2). The sites component of the model allows gamma-distributed variation in evolutionary rate across sites of a gene, modelling the fact that diversifying selection is likely to be rare and arising only within portions of the entire coding sequence. The branch component of the model allows the ratio of non-synonymous to synonymous substitutions (ω) to increase above unity—signifying positive, diversifying selection—only on branches of the phylogeny exhibiting transitions from invasive hemochorial placentation to less-invasive non-hemochorial forms. A likelihood ratio test was used to identify genes for which the likelihood of the branch-sites positive selection model is significantly higher than the likelihood of its neutral counterpart in which focal branches are restricted such that ω ≤ 1 (in PAML, fix–omega = 1). It is expected that some genes, for example those involved in antagonistic coevolution with genes expressed by parasites or disease vectors, may be under positive selection on all branches of the phylogeny. These genes may fit the branch-sites model better than the neutral model even though selection is not specifically restricted to the branches of interest that bear evolutionary transitions in placental invasiveness. In order to exclude such genes from our list of positively selected genes, we also fitted a third model in PAML in which a single evolutionary regime of positive selection was applied to all branches of the phylogeny (in PAML, model = 0 and NSsites = 2). Genes for which this global selection model fitted better than the branch-sites model were excluded from the list. As this model is not nested within the branch-sites model as a special case, Akaike's information criterion  was used to assess model fit rather than the likelihood ratio test. Finally, we excluded any genes for which the branch-sites model appeared to be the best fit but which exhibited signs of failure of convergence (i.e. genes for which the inferred positively selected ω = 1 or for which the proportion of sites assigned to the positive selection class = 0). All models were fitted to the data multiple times and the maximum-likelihood inferred over the course of these multiple trials was used in the likelihood tests.
(c) Gene set enrichment
Enrichment analysis was used to test the set of positively selected genes for functional signatures associated with placentation and placental disease. Enrichment analysis of gene ontology biological processes and Reactome canonical pathways was accomplished using ClueGo  (see the electronic supplementary material, tables, for parameters). Disease associations of positively selected genes were identified by extracting gene sets associated with disease MeSH terms from the Genetic Association Database  (which includes pre-eclampsia and other reproductive disorders within a comprehensive list of human diseases) and testing for enrichment using Fisher's exact test. Enrichment of human and mouse tissue types was tested using Enrichr . All multiple tests were subject to Benjamini–Hochberg false discovery rate correction.
(d) Detection of selected substitutions of large phenotypic effect
Genes bearing positively selected amino acid substitutions of large phenotypic effect may constitute novel candidate genes for involvement in the evolution of placental invasiveness and, potentially, in the pathogenesis of pre-eclampsia and other diseases of human placental invasion. PAML can be made to output, for each site of an alignment, a Bayes Empirical Bayes probability that the given site is subject to diversifying selection. Based on the alignments, it is possible to calculate, for each site with Bayes Empirical Bayes p > 0.95, the amino acid(s) present in non-hemochorial species that are non-synonymous with the amino acid present in human beings. The software application PROVEAN  was used to predict, computationally, which of these apparently positively selected amino acid substitutions would be associated with a major phenotypic effect, were they to arise in human beings.
(a) Identification of positively selected genes
Based on alignments of 16 578 protein-coding genes across 18 taxa (four of which have non-hemochorial placentas, figure 1), we identified a subset of 1254 genes which are inferred to have evolved adaptively in the three independent focal lineages undergoing evolutionary transitions towards reduced placental invasion, but neutrally in the remaining lineages undergoing no such evolutionary transition. These genes are presumed to be involved, potentially, in the macroevolution of less-invasive placentation.
(b) Significant overlap of positively selected genes with genes known to be involved in human reproductive disorders
Functional overlap between genes subject to adaptive evolution during independent losses of hemochorial placentation in tree shrews, strepsirhine primates and kangaroo rats, and genes associated with human disorders of placental invasion, was assessed using tests for enrichment of disease-association gene sets. Fishers exact tests based on the Genetic Association Database , comprising 167 130 gene–disease associations derived from human studies, indicate that the adaptively evolving gene set is significantly enriched with genes that underlie pregnancy-related disorders (including premature birth, chorioamnionitis, pre-eclampsia and cardiovascular complications of pregnancy) and circulatory disorders involving blood clotting and atherosclerosis (table 1). Pre-eclampsia is both a placental and a vascular disorder, giving rise to maternal vascular lesions that are highly similar to those found in atherosclerosis  and predisposing towards hypertension, ischaemic heart disease, cerebrovascular disease and thromboembolism . The enrichment results are thus consistent with positive selection on genes involved in pathology of the placenta and uterine vascular system, especially pre-eclampsia.
(c) Significant overlap between positively selected genes and genes involved in pre-eclampsia
Studies on the genetic basis of pre-eclampsia are dominated by two broad classes of approach. First, candidate gene and genome-wide association studies have been used to identify single nucleotide polymorphisms (SNPs) associated with pre-eclampsia. A comprehensive database of pre-eclampsia-associated SNPs  identifies within the set of 16 578 studied genes, 149 bearing putatively pre-eclampsia-associated SNPs in their coding sequence, introns or enhancer regions. These genes were significantly enriched within the set of adaptively evolving genes (22 genes, p = 0.002). Of the 149 SNP-bearing genes, 92 are significantly associated with pre-eclampsia with p < 0.05 in at least one study, and this more restricted set of loci is also enriched within the set of adaptively evolving genes (14 genes, p = 0.009).
Second, studies of differential gene expression in pre-eclamptic versus normal placenta and endometrium have been used to identify genes significantly upregulated or downregulated in the disease condition. A recent review of such studies  identifies, within the set of 16 578 studied genes, 84 genes that are found to be differentially expressed in at least one placental or endometrial tissue during pre-eclamptic pregnancies. Of these genes, 12 exhibit the statistical signal of adaptive evolution, constituting significant enrichment (p = 0.024). It has been noted that there is only a modest consensus in results from gene association and differential expression studies . Nevertheless, of the pre-eclampsia-associated genes present in our set of 16 578 tested for adaptive evolution, 12 bear pre-eclampsia-associated SNPs and are also differentially expressed in pre-eclampsia. To account for possible non-independence of results from genetic-association and gene-expression studies, we also calculated enrichment statistics for the set of unique pre-eclampsia-associated genes combined from both approaches. Of 221 such genes, 33 exhibit the statistical signal of adaptive evolution (p < 0.001).
(d) Functional enrichment
Genes subject to adaptive evolution were also significantly enriched with biological processes involved in placental function (the electronic supplementary material, table S2). A cluster of genes involved in angiogenesis and blood vessel development was the most strongly enriched (especially involving interleukin 6, an angiogenic cytokine possibly involved in trophoblast invasion into the endometrium [38,39]), consistent with selection on vascular patterning during the evolution of placental invasiveness and also consistent with the centrality of misregulated angiogenesis to the pathogenesis of preeclampsia [40,41]. Significant enrichment was also identified in a cluster of genes involving cytokine and chemokine regulation of the immune system and inflammatory response, and in clusters of genes related to cell migration and apoptosis, all central to fetal–maternal interactions in utero [42–44]. Analysis of enriched canonical pathways (electronic supplementary material, table S3) indicates especially strong selection on G-protein-coupled receptors, integrin and NOTCH receptors as well as some of their downstream intracellular signalling cascades PI3K and AKT, all known for involvement in promoting trophoblast motility , as well as numerous other signalling pathways associated with placentation and placental disease including fibroblast growth factor and interferon gamma [46,47]. Statistical identification of gene clusters enriched in common between adaptively evolving genes and combined pre-eclampsia-associated genes yielded similar results (electronic supplementary material, tables S6 and S8) confirming the pre-eclampsia associations of these processes and pathways.
(e) Identification of novel evolutionarily informed candidate genes for involvement in pre-eclampsia
One implication of the enrichment of genes associated with placental pathology within the set of adaptively evolving genes is that the latter may contain novel candidate genes and pathways involved in human placental disorders. To formally prioritize the set of genes subject to adaptive evolution for possible involvement in pre-eclampsia and related conditions, we used PROVEAN, a software tool for predicting the phenotypic impact of nucleotide substitutions in humans based on evolutionary conservation of amino acid sites . We identified 289 genes that evolved adaptively during loss of hemochorial placentation in Euarchontoglires, for which positive selection can be ascribed to specific amino acid sites with high probability (Bayes Empirical Bayes p > 0.95) and which also exhibit, in non-hemochorial Euarchontoglirans, substitutions at those sites that are predicted to be of major phenotypic effect relative to the normal human allele (PROVEAN score < −2.5). These genes are significantly enriched with putatively pre-eclampsia-associated SNPs (eight genes, p = 0.005, table 2; electronic supplementary material, tables S4 and S5) though not with transcripts differentially expressed in pre-eclampsia. Genes already known to be associated with pre-eclampsia were excluded and, through comparison with the combined set of pre-eclampsia-associated genes, those remaining were categorized into three subsets. First, we identified members of pathways and biological processes enriched in common between pre-eclampsia-associated and adaptively evolving genes (58 genes with robust functional support for involvement in pre-eclampsia; electronic supplementary material, tables S10 and S12). Second, we identified members of pathways and biological processes enriched uniquely in the adaptively evolving gene set (48 genes that may be involved in pre-eclampsia or may be involved in adaptations allowing low levels of placental invasion to be compatible with healthy mothers and offspring in kangaroo rats, tree shrews and strepsirhine primates; electronic supplementary material, tables S11 and S13). Finally, we identified a subset of 111 adaptively evolving genes with large predicted phenotypic effects involved in direct physical or genetic interactions with known pre-eclampsia-associated genes . The combination of these subsets constitutes a list of 199 novel, evolutionarily informed candidate genes for involvement in the evolution of placental invasiveness and in the pathogenesis of pre-eclampsia and other diseases of placental invasion, listed in full in the electronic supplementary material, table S14; those genes with known placental functions are described in table 3.
Gene Ontology biological processes, Reactome pathways, KEGG pathways and Wikipathways enriched among the 199 candidate genes are illustrated in figure 2. Overlapping gene sets are clustered and identified using ClueGo  under default parameters. A number of disease-associated modules are identified—including tuberculosis, leishmaniasis, cholera infection and amyotrophic lateral sclerosis. We suspect that the enrichment of such categories is a consequence of high levels of research effort into identification of disease-associated genes and arises from enrichment of genes generally involved in immunity and pathological processes. This view is supported by the fact that highly specific disease-associated gene sets are tightly clustered with more generic immune and cytokine processes. Hence, the enriched tuberculosis and amyotrophic lateral sclerosis (ALS) gene sets are tightly clustered with a number of pathways and processes involved in tumour necrosis factor signalling and inflammation. Tumour necrosis factor plays a major role in the response of the human body to tuberculosis infection  but additionally regulates trophoblast migration, invasion and apoptosis in early pregnancy [50–52], and proteins associated with the tumour necrosis factor pathway are specifically involved in remodelling of spiral arteries . Similarly, the leishmaniasis gene set is tightly clustered with biological functions and processes involved in angiogenesis and endothelial cell migration. Progressive angiogenesis and tissue remodelling in the spleen is characteristic of visceral leishmaniasis in mammals  and angiogenesis inhibitors are a potential treatment for the parasite . As noted above, pathological angiogenesis is a fundamental outcome in pre-eclampsia [40,41]. Several other enriched clusters broadly overlap with aspects of pre-eclampsia, including genes involved in NOTCH signalling, known to be involved in early angiogenesis [56–58] and the nitric oxide pathway, involved in the control of vasoconstriction and misregulated in pre-eclampsia [59–61].
The enrichment among the set of genes targeted by natural selection of distinct but overlapping human placental disorders ranging from premature membrane rupture to pre-eclampsia (table 1), along with significant overlap with the set of genes known to be differentially expressed in pre-eclampsia or bearing pre-eclampsia-asssociated SNPs (table 2), constitute robust support from multiple, independent approaches for the existence of a core set of genes underlying both pre-eclampsia in humans and the adaptive evolution of reduced placental invasion in Euarchontogliran mammals, and evidence against the notion that the pathogenesis of pre-eclampsia involves placental processes unique to humans. We have used a novel evolutionary computational approach to identify 199 genes (table 3), not currently known to underlie the pathogenesis of pre-eclampsia, as candidates for involvement with pre-eclampsia and other diseases of human pregnancy.
Evolutionary analyses have the potential to shed light on outstanding controversies in the study of placental disease. In particular, we find selection on genes involved in hypertension and blood pressure regulation as well as placental tissue remodelling and angiogenesis, supporting the controversial view that gestational hypertension and pre-eclampsia may actually overlap to some extent in their pathogenesis beyond mere superimposition in at-risk pregnancies . More generally, the overlap between the set of genes targeted by natural selection and genes involved in multiple placental pathologies supports the view that placental disorders are highly multifactorial and that there may exist different genetic routes to common symptomatic states . Controversy has also been raised over the relative contribution to pre-eclampsia of the maternal and fetal components of the placenta, especially the role of the maternal immune system in the disease . This is especially pertinent given the apparent association of disease risk with maternal exposure to partner-specific paternal alloantigens, with nulliparous women more than twice as likely to develop pre-eclampsia as multiparous women , and the risk of pre-eclampsia and/or pregnancy-induced hypertension inversely correlated with the duration of sexual cohabitation prior to conception [66–69] presumably as a result of tolerance arising from maternal exposure to paternal seminal antigens [70–72]. Mapping of the 199 genes bearing positively selected amino acid substitutions of large phenotypic effect to tissue-specific expression data  using Enrichr  indicates weak enrichment of proteins localized to CD14 + monocytes (enrichment score = 4.15; FMNL1, BID, MANBA, TNFRSF1B, POU2F2, PSTPIP1 and JAK2). These immune cells are characteristic of maternal uterine tissues during pregnancy  and are differentially expressed in the decidua in pre-eclampsia ; in the context of maternal immunity, they are known to be differentially regulated in the placenta depending on maternal lifestyle and possibly allergen exposure . A similar analysis using mouse gene atlas data [73,77] indicates enrichment for genes localized to IgE-bound mast cells (enrichment score = 4.42), which have a central role in allergy and inflammation and which are also active in the maternal reproductive tract throughout pregnancy  and are involved in the defective vascular remodelling of pre-eclampsia . Hence, components of the maternal immune response involved in pregnancy and pre-eclampsia are targets of natural selection during evolutionary transitions in placental invasiveness, supporting a role for genetic modules underlying maternal immunity, inflammation and allergy response in evolutionary transitions between placental types and in the pathogenesis of disorders of placental invasion.
The importance of Darwinian evolution of protein-coding genes in generating phenotypic diversity remains much debated, and it has long been argued that selection on transcriptional regulation may be of equal or greater importance than selection on protein-coding sequences [80,81]. Furthermore, both protein sequence and regulatory evolution appear to be dominated by neutral drift and purifying selection rather than diversifying selection [82–84]. A strong association with placentation and diseases of pregnancy in the gene lists described above provides strong support for the role of natural selection on protein-coding sequences in placental evolution. This is not inconsistent, however, with a major role for regulatory evolution including drift. Ten of the 74 genes listed in table 3 are themselves placentally expressed transcription factors (PKD1, PRKCB, GATAD1, CREB3L1, SP5, HDAC1, FIGLA, JAK2, NFYA and LZTFL1). But, more pertinently, an ad hoc test using Enrichr  indicates that 41—more than half—of the genes listed in table 3 are collectively regulated by eight transcription factors (SNAI1, SNAI2, TCF3, ZNF148, USF2, EGR1, E2F1 and JUN). At least six of these eight transcription factors whose targets are significantly enriched in table 3 (q < 0.005) have known roles in regulating blastocyst implantation, vascular inflammation or trophoblast differentiation and proliferation during pregnancy [85–90]. Hence, while it is possible that positive selection on the gene lists described above is directly involved in the generation of phenotypic diversity in mammalian placentation, it is also possible that the signal of adaptive evolution identified in this study represents the accumulation of mutations in protein-coding sequences that are compensatory to drift or adaptive evolution of the broader regulatory network in which protein-coding genes are embedded.
One such regulatory evolutionary process that has been implicated in the evolution of reproductive mode in reptiles  and at the time of the division between eutherian and metatherian mammals [92–94] is heterochrony. Deep placental invasion in species with hemochorial placentation naturally progresses through a process of increased invasiveness throughout gestation: at the moment prior to implantation, the uterus is of course not invaded at all, and the first trimester of human pregnancy is supported primarily by maternal uterine secretions and yolk sac placentation until at least the tenth week of pregnancy, when the vascularization of the chorionic villi supports establishment of true hemochorial placentation with direct fetal access to maternal blood . It is notable that species with derived, less-invasive forms of placentation exhibit prolonged yolk sac placentation and histotrophic nutrition, a phenomenon most pronounced in the epitheliochorial horses  but also observed in strepsirhine primates . In haplorhine primates, early placental phenotypes appear to be more conserved than phenotypes arising later in gestation , and in farm animals, gene-expression patterns of early pregnancy are more conserved than those in late pregnancy, which are characterized by species-specific divergence . These observations suggest a heterochronic model of placental evolution in which early stages of placentation are prolonged in order to accomplish reduced placental invasion, along with taxon-specific terminal addition of novel adaptations to support late pregnancy, such as the unique hemophagic regions of carnivores  and the areaolae or chorionic vesicles of strepsirhines . King  has argued that the degree of placental invasiveness is primarily dependent upon maternal endometrial reactions to trophoblast, a view supported by the fact that placental tissue from species with minimal invasiveness, such as pigs, expresses a highly invasive phenotype when transplanted into an ectopic site [101,102]. A third possible interpretation of our findings, then, is that the signal of adaptive evolution identified in this study reflects compensatory mutations in proteins expressed by the fetus and/or mother in response to maternal regulatory evolution, as part of a process of parent–offspring conflict over the degree of placental invasion [103–106]. We anticipate that combining the results presented above with future data on maternal and fetal transcriptomic evolution during transitions in placental invasiveness will help to tease apart the role of adaptively evolving proteins in generating phenotypic variation, adapting to regulatory evolution of the transcriptome and participating in parent–offspring conflict.
These findings support the hypothesis that a core set of genes and pathways underlying eutherian placental invasiveness are associated with both the pathogenesis of human pre-eclampsia and the convergent evolution of less-invasive (endotheliochorial and epitheliochorial) placentation. These results, derived from the study of three independent phylogenetic replicates of evolutionary transitions towards reduced placental invasion, support and complement previous work that examines a less (taxonomically and placentally) diverse set of taxa in which a single branch of the phylogenetic tree is associated with increased invasiveness through the evolution of spiral arteries . Establishing that the raw genetic basis of human placental disorders is of ancient lineage provides a fundamental empirical grounding for evolutionary theories of human placentation that are based on notions of parent–offspring or intragenomic conflict in mammals and viviparous vertebrates in general [103–106]. Furthermore, the view that endotheliochorial and epitheliochorial placentation represent extreme points along a common underlying genetic axis of variation suggests that studies of species bearing less-invasive forms of placentation, such as kangaroo rats, tree shrews and some insectivores, may provide useful models of the molecular control of maternal–fetal interactions and could yield important insights into the mechanisms underlying pre-eclampsia and other human disorders of placentation. More generally, these results suggest that disease-related ontogenetic changes can genetically recapitulate large-scale phylogenetic shifts in mammalian morphology, such that development and diversification share a common genetic basis.
B.J.C. was supported by NSERC. M.G.E. was supported by the Human Evolutionary Studies Program at Simon Fraser University.
Competing of interests
We have no competing interests.
This research was enabled in part by support provided by WestGrid (www.westgrid.ca) and Compute Canada Calcul Canada (www.computecanada.ca). The authors thank Graham Burton, Ashley Moffett, Eric Barrington and the Royal Society for their organization of the symposium giving rise to this special issue. M.G.E. and B.J.C. conceived of and designed the study; M.G.E. acquired and analysed the data; M.G.E. and B.J.C. drafted the manuscript. All authors gave final approval for publication.
One contribution of 13 to a discussion meeting issue ‘Human evolution: brain, birthweight and the immune system’.
- © 2015 The Author(s) Published by the Royal Society. All rights reserved.