Abstract
In mammalian evolutionary history, Cetacea (whales, dolphins, and porpoises) achieved astonishing success by adapting to an aquatic environment. One unique characteristic of Cetacea, contributing to this adaptive success, is efficient lipid utilization. Here we report comparative genetic analysis of aquatic and terrestrial Cetartiodactyla using 144 genes associated with lipid metabolism. We analysed genetic mutation rate, amino acid substitution, and metabolic pathways using genetic data publicly available. Our test detected 18 positively selected genes in Cetacea compared to 13 in Bovidae with little overlap between the lineages. There were lineage specific amino acid substitutions within functional domain regions of these genes. Moreover, a pathway analysis showed that the identified genes in Cetacea were associated with lipid digestion, lipid storage, and energy producing pathways. This study emphasizes the evolutionary context of lipid metabolism modification of Cetacea since the divergence from terrestrial ancestor. Our results provide a foundation for future studies of elucidating the adapted biological mechanisms of Cetacea lipid metabolism and a framework for incorporating ecological context into studies aimed at investigating adaptive evolution.
1. Introduction
The evolution of Cetacea represents one of the most striking adaptations for a habitat transition in mammalian evolutionary history [1, 2]. Although mammals are generally under greater constraints with regard to anatomical, physiological, and behavioural changes than other classes of vertebrates, Cetacea has achieved a remarkable macroevolutionary transition since their divergence from an artiodactyl ancestor approximately 50 million years ago [1]. The past two decades of paleontological and phylogenetic research have thoroughly characterized the evolutionary history of Cetacea [1-3], and there continues to be considerable attention on the molecular evolution behind the dramatic changes in their morphology and ecology upon aquatic adaptation [4]. Understanding the molecular basis underlying the adaptive traits will shed light on this unique evolutionary event from land to water.
Cetacea has developed unique characteristics that are different from those of terrestrial mammals but are common within the lineage [5]. For example, all species of Cetacea are carnivores and have streamline body morphology. Cetacea consumes lipids not only as one of their primary energy resources, but also as an important component of the thick blubber that provides thermo insulation and supports their streamlined morphology [6]. The evolution of lipid usage may be a key factor for the adaptive characteristics enabling Cetacea’s transition to an aquatic environment. Recent genomic studies have reported signatures of positive selection on diverse processes, including those of lipid metabolism [7-11]. However, evolutionary analyses of cetaceans have been with distantly related lineages, rather than contrasting them with closely related terrestrial species using the same analytical procedures. Furthermore, given the potential adaptive specificity of each Cetacea species to its own environment and ecology, analyses at the taxonomic-group level are necessary to interpret the detected genetic signatures as aquatic adaptations.
In this study, we identified the genetic signatures of positive selection on lipid metabolism in Cetacea and Bovidae (figure 1a). The goals of our study were to (1) collect genes associated with lipid metabolism from Cetartiodactyla species (even-toed ungulates, including Cetacea), (2) identify the positively selected genes (PSGs) for each lineage, Cetacea and Bovidae, (3) determine the amino acid substitutions that occurred in functional domain regions of the identified PSG, and (4) describe the biological functions and metabolic pathways that have been modified during aquatic adaptation (figure 1b). Our comparative analyses revealed unique and a greater degree of signatures for positive selection in the Cetacea lipid metabolism pathways compared with those of Bovidae. This study provides the genetic basis for the evolution of lipid metabolism developed by Cetacea in the context of aquatic adaptation and emphasizes the importance of comparative analyses based on the phylogenetic and ecological perspectives of the study organisms.
2. Materials and Methods
(a) Data sampling and preparation
All the genetic data in this study were downloaded from NCBI (https://www.ncbi.nlm.nih.gov). Genomes of Artiodactyla species that were publicly available and annotated were used. Orthologous cording DNA sequences (CDSs) were identified by blastn search with human genes as references. Genes that were not annotated in any of the study species were excluded from later analyses. CDSs were aligned with PRANK v.140603 [12] and manually edited with Mesquite v.3.2 [13]. To remove potentially unreliable sequences, the aligned CDSs were filtered with GUIDANCE2 using the default settings [14]. In addition, codon triplets with gaps in more than 50% of the sequences were discarded.
(b) Identification of positively selected genes
PSGs were identified using branch-site models implemented in the CODEML program in the PAML v. 4.8 software package [15]. The branch-site test detects positive selection on a priori specified branches. To detect selection in Cetacea and Bovidae separately, we conducted the tests twice for all genes evaluating one lineage at a time. A one-ratio model was employed as a null hypothesis in which all branches were under neutral evolution. A two-ratio model was conducted as an alternative hypothesis in which a specified lineage was under positive selection, and this model estimated the dN/dS ratio (ω: nonsynonymous to synonymous substitution ratio). Because the branch-site test estimates the dN/dS ratio on a subset of codons, the ω values are not reliable measure of the strength of positive selection [16]. Therefore, the likelihood ratio test was used to determine whether a gene was under positive selection. The likelihood ratio test was performed to compare 2Δl of the two models to the χ2 distribution for p-value evaluation.
(c) Domain annotation and amino acid substitutions
The identified PSGs were scanned for domains with the pfam_scan utility and HMMER3.1 [17] against the Pfam-A from Pfam v. 31.0 [18]. The domain arrangements were visualized with DoMosaics v. 0.95 [19]. Domain similarity was estimated by using amino acid sequences of Orcinus orca and Bos taurus as criteria for Cetacea and Bovidae PSGs, respectively, using DoMosaics, and the dot plots were generated using the R package ggplot2 [20]. The lineage specific amino acid substitutions were identified manually by looking at a conversion of residue that occurred in only one lineage and where all the species in that lineage had identical residue. Those instances were included when one of the outgroup species had a mutation at the same position.
(d) Function and pathway analysis of PSGs
Gene ontology (GO) terms of the identified PSGs were assigned according to DAVID Functional Annotation tool [21, 22] under the biological process domain. The pathways in which the identified PSGs participate were identified using KEGG Mapper [23]. Because a gene can be involved in multiple pathways, which can be categorized in multiple hierarchy, the pathways that was more representative and directly related with lipid metabolism were focused.
3. Results and Discussion
(a) Positively selected genes in Cetacea and Bovidae
To detect signatures of positive selection, we analysed the alignments of lipid metabolism-related genes collected from genomes of Cetartiodactyla. To assess the evolutionary transitions since their divergence from a terrestrial ancestor [1], we categorized the study species by their phylogeny and ecology in the target group (Cetacea), a control group (Bovidae), and an outgroup (Suina and Tylopoda). Among Ruminantia, the family Bovidae was used as a control group. Bovidae is a fully terrestrial and close clade of Cetacea [24]. The ecological and geographical diversity of Bovidae provides an unbiased representation of terrestrial adaptation. We opted to use only one family, because there are few differences in the adaptive fitness of Bovidae to an aquatic environment. Thus, positive selection detected in Cetacea can suggest the adaptations required to return to an aquatic environment. To gain insight into the evolution of lipid usage, we investigated the genes associated with lipid metabolism. Taken together, the results from these datasets will elucidate plausible molecular adaptations underlying Cetacea macroevolutionary transitions from a terrestrial to an aquatic environment.
A total of 14 Cetartiodactyla species were used in this study: five Cetacea, including the minke whale (Balaenoptera acutorostrata), sperm whale (Physeter catodon), killer whale (Orcinus orca), bottlenose dolphin (Tursiops truncatus), and Yangtze River dolphin (Lipotes vexillifer); five Bovidae, including cattle (Bos taurus), wild yak (Bos mutus), American bison (Bison bison), domestic goat (Capra hircus), and domestic sheep (Ovis aries); and Suina and Tylopoda, including pig (Sus scrofa), alpaca (Vicugna pacos), Bactrian camel (Camelus bactrianus), and Arabian camel (Camelus dromedaries) (figure 1c). For each species we searched orthologous of the 156 genes assigned to functions associated with lipid metabolism in humans. Our filtering strategy retained 144 genes covering the major lipid metabolic pathways (electronic supplementary material, table S1).
To identify genes under positive selection, we performed likelihood ratio tests using branch-site models implemented in PAML [15]. This analysis estimates dN/dS ratio (ω) for the lineage of interest, and the PSGs were determined by the likelihood ratio test. Of the 144 lipid metabolism genes, 14 genes were identified as PSG in the lineage of Cetacea, as well as 5 genes in the lineage of Bovidae at p < 0.05 (figure 2a, table 1, electronic supplementary material, table S2 and S3). Due to the low number of PSGs for assessing the selection landscape of metabolic pathways, we relaxed the significance level to 0.05 ≤ p < 0.1 and found an additional 4 and 8 PSGs in Cetacea and Bovidae, respectively. Among the identified PSGs, only one gene, GPAM (also known as GPAT1, glycerol-3-phosphate acyltransferase) was found in common at p < 0.05 (figure 2b).
To assess whether the detected amino acid substitutions have influence on functions of the identified PSGs, we examined the functional regions of the protein sequences, namely domains, as described in the Method section (figure 3 for EHHADH; for other PSGs see electronic supplementary material, figure S1-S31). Whereas the domain arrangements were largely conserved among Cetartiodactyla, the comparisons of the sequences of the domain region showed inter-lineage divergence between Cetacea and Bovidae. Furthermore, we explored amino acid substitutions that occurred only in Cetacea and remain identical within the lineage. Such lineage specific substitutions in the functional domain of genes, evolving under positive selection, would putatively be associated with adaptive phenotype in an aquatic environment. We found lineage specific substitutions in at least one domain of 14 Cetacea PSGs and 11 Bovidae PSGs (table 2 and figure 3c). For example, one Cetacea PSG, EHHADH, is composed of four domains, except for the domain arrangements of Tursiops truncates and Sus scrofa, probably because of isoforms (figure 3a). The amino acid sequence of each domain differs between Cetacea and Bovidae while they are relatively similar within each lineage (figure 4b). EHHADH encodes two enzymes, enoyl-CoA hydratase/isomerase (ECH) and 3-hydroxyacyl-CoA dehydrogenase (HCDH), involved in â-oxidation [25]. The amino acid sequences of both of these enzymes contains lineage specific substitutions (figure 4c), even though none of the substitutions were found in active sites [26]. We did not find lineage specific substitutions in four Cetacea PSGs (AGPAT, FBP1, GNPAT, and GPAM) nor in two Bovidae PSGs (GPAM and PGAM5). However, substitutions outside of domains may still have functional influence.
Our comparison of the lipid metabolism genes highlighted greater and unique selective pressures in Cetacea lipid metabolism compared with Bovidae. The results showed that both lineages have modified their lipid metabolism since divergence from a common ancestor, suggesting the adaptive importance of the usage of lipids in both aquatic and terrestrial environments. The relatively higher number of PSGs with p < 0.05 Cetacea, however, indicates that evolutionary changes in lipid metabolism were more substantial for the aquatic than the terrestrial lineage. Moreover, the slight overlap between the Cetacea and Bovidae PSGs indicated the differing selection regimes in lipid metabolism of aquatic and terrestrial lineages. Our results (1) provide support for the adaptive evolution of lipid metabolism in the context of aquatic adaptation and (2) highlighted that these adaptations are not species specific for T. truncatus [7, 27] or other Cetacea species [8].
(b) Functional analysis of Cetacea PSGs
To gain insights into the functional consequences of the lipid metabolism behind the aquatic adaptation of Cetacea, we assessed the GO terms and metabolic pathways of the identified PSGs (figure 4).
Our functional analyses revealed that six Cetacea PSGs (APOA2, APOB, CYP8B1, LIPF, PNLIP, and PNLIPRP2) were involved in the fat digestion and absorption pathway associated with the lipid catabolic process (GO:0016042) and lipoprotein biosynthetic process (GO: 0042158). LIPF, PNLIP, and PNLIPRP2 encode pancreatic lipase, which hydrolyses dietary lipids in the digestive system [28, 29]. CYP8B1 encodes a member of the cytochrome P450 superfamily of enzymes which catalyses the synthesis of primary bile acids. Bile acids play important roles not only in absorption of lipids but also in the regulation of lipid metabolism. Additionally, they are believed to function as a therapeutic approach for metabolic syndrome [30]. Lastly, APOA2 and APOB encode apolipoproteins, which aid in transporting hydrophobic lipids through the circulation system and are associated with atherosclerotic cardiovascular diseases [31], implying enhanced tolerance for high lipid concentration in their circulatory systems. Previous studies have also reported some genes in this pathway under selection in Cetacea [7, 8, 10, 32]; moreover, owing to the explicit comparison with their herbivorous relatives, our results provide further support of genetic signatures in Cetacea adaptation in their shift in diet.
The storage of fat is an essential part of lipid metabolism, whereby triacylglycerol (TAG) is synthesized to serve as a major component in several biological functions. Synthesis of TAG takes place in glycerophospholipid metabolism via two distinct major pathways, the phosphatidic acid pathway and the monoacylglycerol pathway [33]. We found five Cetacea PSGs (AGPAT1, GPAM, LIPF, PNLIP, and PNLIPRP2) involved in storage of fat, including the triglyceride metabolic process (GO: 0006641) and phosphatidic acid biosynthetic process (GO: 0006654). Three Bovidae PSGs (GPAM, MOGAT1, and MOGAT3) were also found in this category, including the glycerol metabolic process (GO: 0006071), in which TAG is synthesized from monoacylglycerol. These findings suggest enhanced TAG syntheses in both lineages, but under different molecular mechanisms in aquatic and terrestrial Cetartiodactyla. Functional evolution of the monoacylglycerol pathway associated with rumen evolution, for digesting C4 grasses that emerged approximately 40 million years ago, has been proposed in Bovidae [34]. Conversely, an evolutionary analysis of Cetacea reported positive selection on genes involved in both pathways of TAG synthesis for blubber thickening [8]. In these studies, one cannot rule out that changes in the monoacylglycerol pathway may have been present in a common ancestor (i.e. contemporaneous evolution) since close relatives were not investigated. Our analyses do not suffer from this shortcoming because both Cetacea and Bovidae were analysed in an equivalent manner. Therefore, our analyses support the evolution of the monoacylglycerol pathway in Bovidae and the phosphatidic acid pathway in Cetacea. The only common PSG between Cetacea and Bovidae, GPAM, encodes a mitochondrial enzyme that regulates and synthesizes TAG and other lipids [35]. More research, such as in vitro functional experiments, is necessary to determine whether the mutations in this gene produced a different phenotypic outcome between the lineages.
We also inspected evolutionary changes in energy producing pathways, including fatty acid degradation and glycolysis/gluconeogenesis, in which fatty acids and glucose serve as the main energy resource, respectively. We found three Cetacea PSGs (ACADVL, EHHADH, and HADHB) in â-oxidation (GO: 0006635). These three PSGs encode four enzymes including acyl-CoA dehydrogenase, enoyl-CoA hydratase, 3-hydroxyacyl-CoA dehydrogenase, and thiolase which drives the â-oxidation [25]. Our domain analysis revealed that Cetacea specific amino acid substitutions occurred in the functional domains of all the four enzymes (figure 3, table 2, and electronic supplementary material, figure S1, S9, and S13). Given the signatures of selection in the four main steps of this process, the â-oxidation of Cetacea may have been extensively modified during aquatic adaptation, potentially increasing the efficiency of producing energy from fatty acids. To our knowledge, this is the first report of an implication for evolution of Cetacea â-oxidation, including genomic and physiological studies.
The Cetacea PSGs exhibited three genes (ACSS2, FBP1, and PDHB) that were associated with the glycolysis/gluconeogenesis pathway (GO: 0006096 and GO: 0006094). Interestingly, gene expression of FBP1 is regulated by bile acids [36], implying a possible functional connection with the other identified PSG, CYP8B1. FBP1 encodes an enzyme at the rate limiting step of gluconeogenesis. Although metabolic studies of carnivores have primarily used domestic cats [37], the selection of the Cetacea gluconeogenesis may provide a new insight into the mechanisms for glucose biosynthesis by carnivorous animals. The remaining two PSGs encodes enzymes involved in Acetyl-CoA synthesis: PDHB is a part of the pyruvate dehydrogenase complex that converts pyruvate into acetyl-CoA [38], and ACSS2 is a part of acetyl-CoA synthetase that catalyses the formation of acetyl-CoA from acetate [39]. Acetyl-CoA is an important molecule that participates in the citrate cycle. It is noteworthy that we identified multiple signatures of positive selection involved in the synthesis of acetyl-CoA and selective pressure on â-oxidation. This emphasizes the importance of energy production for aquatic adaptation.
4. Conclusion.
The comparative analyses between aquatic and terrestrial Cetartiodactyla provided genetic signatures of the evolution of lipid metabolism in Cetacea. This study is unique in that it used a sufficient number of both Cetacea and Bovidae samples in an equal manner that enabled the identification of molecular changes putatively reflecting their ecologies. The lineage specific amino acid substitutions in domain regions implied functional modifications of multiple genes that are involved in important biological processes for aquatic adaptation. These genes and the associated pathways will be plausible targets for future investigations of Cetacea lipid metabolism.
Data accessibility
Additional results supporting this article have been uploaded as part of the online electronic supplementary material.
Author’s contributions
Y.E. conceived of the study, designed the study, and carried out the data collection, data analysis, bioinformatics works, and drafted the manuscript; K.K. and M.M. coordinated the study and participated in the design of the study, and helped draft the manuscript. All authors gave final approval for publication.
Competing interests
The authors declare no potential conflict of interests for this study.
Funding
This work was supported by the Japan Society for the Promotion of Science (JSPS; 16K14660 to K.K. and 25118005 to M.I-M.). This study was also partially supported by Kyoto University Supporting program for interaction-based initiative team studies (SPIRITS) to M.I-M. The WPI-iCeMS was supported by the World Premier International Research Centre Initiative (WPI), MEXT, Japan.
Acknowledgements.
The authors wish to thank Kishida Takushi and Daisuke Muramatsu for assisting with the data analysis and interpretation, and Maegan Fitzgerald for English correction.