‘Obesity’ is healthy for cetaceans? Evidence from pervasive positive selection in genes related to triacylglycerol metabolism

Cetaceans are a group of secondarily adapted marine mammals with an enigmatic history of transition from terrestrial to fully aquatic habitat and subsequent adaptive radiation in waters around the world. Numerous physiological and morphological cetacean characteristics have been acquired in response to this drastic habitat transition; for example, the thickened blubber is one of the most striking changes that increases their buoyancy, supports locomotion, and provides thermal insulation. However, the genetic basis underlying the blubber thickening in cetaceans remains poorly explored. Here, 88 candidate genes associated with triacylglycerol metabolism were investigated in representative cetaceans and other mammals to test whether the thickened blubber matched adaptive evolution of triacylglycerol metabolism-related genes. Positive selection was detected in 41 of the 88 candidate genes, and functional characterization of these genes indicated that these are involved mainly in triacylglycerol synthesis and lipolysis processes. In addition, some essential regulatory genes underwent significant positive selection in cetacean-specific lineages, whereas no selection signal was detected in the counterpart terrestrial mammals. The extensive occurrence of positive selection in triacylglycerol metabolism-related genes is suggestive of their essential role in secondary adaptation to an aquatic life, and further implying that ‘obesity’ might be an indicator of good health for cetaceans.

action of lipases [desnutrin/adipose triglyceride lipase (ATGL); hormone-sensitive lipase (HSL); and monoglyceride lipase (MGLL)] that releases free fatty acids, which in turn are utilized by other organs to meet the body's energy requirements 7,8 . A series of genes [e.g., Liver X-activated receptor (LXR), perilipin (PLIN), and phosphodiesterase 3B (PDE3B)] have been demonstrated to play an important role in regulating lipogenic and lipolytic processes 4,9 . In addition, some genes [e.g., cell death-inducing DNA fragmentation factor-α -like effector A (CIDEA) and apolipoprotein B (APOB)] have also been associated with TAG metabolism 10,11 . Therefore, these genes are widely recognized as candidates for controlling TAG storage conditions by changing the expression pattern and protein sequence, and a number of genes closely related to lipid metabolism have been identified to be under adaptive evolution at the genome level 12,13 .
Cetaceans (whales, dolphins, and porpoises) are a highly specialized group of mammals that evolutionarily transformed from a fully terrestrial quadruped to an obligate aquatic from approximately 53-56 million years ago (Ma) 14 . During this evolutionary transition, energy reserves and the maintenance of body temperature were the most critical challenges that these species encountered 15,16 . For example, the thickness of the blubber (the specialized hypodermis) in whales is approximately 20 cm, which is 10-fold greater than that of other artiodactyls species 17 . The blubber of cetaceans, which comprises TAG as its most important component, is dynamic and multifunctional, and acts as a metabolic energy storage site 18 , contributes to positive buoyancy 19,20 , provides thermal insulation 21 , supports locomotion, and increases swimming efficiency by streamlining the body surface 22,23 . However, the evolutionary mechanisms and driving force for the formation and maintenance of the thickened blubber in cetaceans have not been well explored to date.
In the present study, the coding regions of 88 genes that represent nearly all of the signal pathway members involved in TAG metabolism in the various cetacean lineages, were investigated and compared to orthologous sequences from terrestrial mammals using both gene-and protein-level approaches. The goal of the present study was to test whether evolutionary changes in these TAG metabolism-related genes were associated with their transition from land to water, and to determine the molecular mechanism underlying the cetacean blubber thickening during this adaptive process.
Positive selection of TAG metabolism-related genes in cetaceans. All-mammals dataset. To investigate the impact of positive Darwinian selection in the 88 TAG metabolism-related genes, we used likelihood models of coding sequence evolution 24 implemented in Codeml of the PAML package 25 . The branch-site model was used to test for positive selection in individual codons for the lineage leading to the common ancestor of each marine mammal groups (cetaceans: branch a; pinniped: branch v; polar bear: branch w; manatee: branch aa in Supplementary Fig. S1), the branch of combined marine mammals and the lineages of other groups (i.e., cetartiodactyls, carnivora, chiroptera, primates, and rodentia) across the mammalian phylogeny ( Supplementary Fig. S1). Interestingly, evidence for positive selection was detected in 9 (APOB, ACSS1, AGPAT5, DGAT1, HSL1, MLXIPL, PLCB2, PLCE1, and PLIN3) of the 88 TAG metabolism-related genes examined of the combined marine mammal branches, 5 (APOB, DGAT1, MGLL, MOGAT1, and SERTAD2) genes in cetaceans, 3 (APOB, FAS, and GNPAT) genes in pinniped, 10 (APOB, DGKB, DGKZ, DGKI, FATP4, HSL, PLCB2, PLCB3, PLCH2, and PPARD) genes in polar bear and 11 (APOB, ACSL1, FAS, FATP2, MOGAT1, MOGAT2, LPL, PLCB2, PLCE1, PLIN4, and PNLIP) genes in manatee ( Fig. 1 & Supplementary Table S2), which suggested the convergent evolution of TAG metabolism-related genes for the marine mammals during their adaptation to the aquatic environment. However, we also found some genes in the terrestrial groups to be positively selected, i.e. 4 genes in cetartiodactyla, 1 gene in carnivora, 9 genes in chiroptera, 2 genes in primates, and 4 genes in rodentia ( Fig. 1 & Supplementary Table S2).
To further test if similar patterns of evolution occurred to the marine mammal groups, we reconstructed ancestral nodes and mapped amino acid changes along four marine mammal branches within 28 positively selected genes totally identified. Sixty-eight statistically significant (P < 0.05) parallel/convergent nonsynonymous amino acid substitutions were identified in the 14 of these 28 genes across two of the marine mammal lineages (Supplementary Table S3). In addition, 10, 21, 7, 2, 13, and 17 parallel/ convergent mutations were found between branch pairs a vs v, a vs aa, w vs a, w vs aa, w vs v, and v vs aa in 4, 8, 3, 1, 2, and 8 genes, respectively. More importantly, the lineages leading to the common ancestor of pinniped, polar bear and manatee shared two amino acid changes (L316H & E331Q) in ACSL1.
Cetaceans-only dataset. The cetaceans-only dataset consisting of 88 TAG metabolism-related genes was further used to determine the selection pattern in the interior nodes of cetaceans. Of these, 35 genes were determined to be under significant positive selection along branches b-s in Fig. 2 using the branch-site   Table S4). A pair of site models (M8a vs. M8) 26,27 were also used to test whether specific codons in the TAG metabolism-related genes underwent positive selection, and 23 genes were determined to have undergone positive selection in the cetaceans-only dataset, where the LRTs of the site model were statistically significant (Table 1 and Supplementary Table S5). In combination with results from above branch-site and site models, it showed that 43 genes were totally detected to be under positive selection in cetaceans by PAML.
Furthermore, the fixed-effects likelihood (FEL) and random-effects likelihood (REL) models were employed to confirm the selection pattern of TAG metabolism-related genes in cetaceans, and 26 genes were determined to be positively selected in cetaceans by the Datamonkey web server (Table 2 and  Supplementary Table S6). The protein-level approach implemented in TreeSAAP 28 identified a series of putative positively selected sites from 39 genes in cetaceans ( Table 2 and Supplementary Table S6) (Table 2 and Supplementary Table S6).
To summarize the above results from the analyses of two datasets using different methods, 44 genes were identified to have undergone positive selection in cetaceans (Figs 2 and 3; Tables 1 and 2), and 5 (APOB, DGAT1, MGLL, MOGAT1, and SERTAD2) of these were subjected to strong positive selection in the common ancestor of the cetacean, whereas no selection was observed in terrestrial mammalian lineages such as cetartiodactyls, carnivores, and primates. It is generally accepted that a positively selected site is more reliable if it can be supported by two or more different methods. Of the positively selected genes, 41 were validated by at least two methods, thus they were used in the subsequent analyses.  Table S7). In 5 of these genes (i.e., ACSL1, CD36, DGAT2, GPAT2, and MOGAT1), 20 positively selected sites were located within the protein transmembrane domain of the corresponding genes. In addition, most of the positively selected sites were localized in the topological domain of ACSL1, ACSL5, ACSL6, CD36, DGAT1, DGAT2, FATP2, GPAT2, and SREBF2. For CD36, DGKA, DGKG, DGKH, DGKQ, LXRb, PLCD3, PLCD4, PLCE1, PLCH2, and PNLIP, some positively selected amino acids were located in the disulfide bond, glycosylation, zinc finger domain, substrate binding domain, ligand-binding domain, and so on, respectively (Supplementary Table S7). Furthermore, one positively selected site (29) was located in the nuclear localization signal motif of the FABP4 gene, and positively selected sites 440, 651, and 727 in the PDE3B gene were located in the catalytic regions, respectively (Supplementary Table S7).

Discussion
Cetaceans have a thick layer of blubber with a mean thickness of 98.4 ± 18.4 mm 29 . The thick blubber layer surrounding the cetacean body can comprise more than 30% of their body mass, and it is far greater than the 4-8% dissectible adipose found in the general healthy wild animal 30 . However, the genetic basis underlying the blubber thickening remains poorly explored. The present study therefore presents the first systematic investigation of TAG metabolism-related genes of representative cetaceans and closely related terrestrial mammals. Wide and strong signals of positive selection were detected in genes related to TAG synthesis or regulation of TAG synthesis, which could in turn provide novel insights into the evolution of blubber thickening in cetaceans.
The positive selection in cetacean DGAT1 and DGAT2 genes detected in the present study supports the morphological evidence that whales have thicker blubbers than those of other artiodactyl species 17 . The DGAT-catalyzed synthesis of TAG is the final and rate-limiting step in TAG formation (Fig. 3), and it is believed that DGAT is a key factor in controlling the production of triglycerides and fatty acids, as well as plays a key modulatory role in animal fat deposition 31 . Therefore, the observed positive selection in the DGAT1 and DGAT2 genes in cetaceans is suggestive of an enhanced capability for TAG formation. In addition, evidence of positive selection has been shown for other genes involved in the remaining pathways of TAG synthesis (Fig. 3). Effective uptake of free fatty acids can accelerate the expansion of adipocyte dimensions when lipids accumulate 32 . Acetyl-coA carboxylase 1 (ACC1), which play important roles in the de novo synthesis of FAs 6,33 , underwent positive selection in cetaceans (Table 2 and Fig. 3). Furthermore, PNLIP, the primary pancreatic TAG lipase 34 , was also subjected to positive selection across cetaceans, suggesting the essential role of FAs from the hydrolysis and absorption of long-chain triglyceride FAs from food during TAG synthesis in cetaceans. CD36, FABP2 (adipocytes lipid binding protein), FABP4, FATP2, FATP3, and GOT2, which are used in facilitating and regulating the transport of FAs across the plasma membrane, were determined to be under significant positive selection in cetaceans, which suggests that cetaceans might have acquired an enhanced capacity for FAs transport to maintain fat deposition. In additional, a series of essential enzymes involved in different steps of TAG biosynthesis pathways, i.e., ACSL1, ACSL5, ACSL6, ACSS1, AGPAT5, AGPAT6, DGKA, DGKG, DGKH, DGKQ, GNPAT, GPAT2, MOGAT1, MOGAT2, PLCB2, PLCB4, PLCD3, PLCD4, PLCE1, PLCG2, PLCH1, PLCH2, PLCZ1, and PPAP2C 18 , were determined to have undergone positive selection in cetaceans (Table 2 and Fig. 3), which in turn suggests that cetaceans have possess an effective ability to enhance their TAG synthesis during their adaptation to a fully aquatic life. Remarkably, some regulatory genes related to TAG synthesis were also determined to be under positive selection (Fig. 3), which may then imply a complex molecular mechanism of cetacean blubber thickening. Liver X-activated receptor b (LXRb), an important protein that controls the amount of cellular SREBP-1c, underwent significant positive selection in cetaceans ( Table 2 and Fig. 3). The expression of FAS and mtGPAT can be greatly increased once SREBP-1c is overexpressed, which in turn results in an increase in FA synthesis and TAG deposition 10,11 . Furthermore, CIDEA, which enhances lipid droplet size when ectopically expressed in preadipocytes and in turn favors cellular lipid accumulation 35 , also presented evidence of positive selection in cetaceans. Positive selection of these genes might therefore play an important role in promoting cetacean blubber thickening. Positive selection of different enzymes involved in cetacean TAG synthesis might directly explain the molecular basis of cetacean blubber thickening. Interestingly, besides from the TAG synthesis-related genes, MGLL, an important major lipases involved in lipolysis 36 , was also under strong positive selection, as indicated by the results of nucleotide-and protein-level analyses ( Table 2 and Fig. 3). These findings therefore suggest that lipolysis was advanced to a certain degree in cetaceans, and the metabolic rate of cetaceans might have been increased to compensate for the energy shortage.
Cetaceans exhibit physiological and anatomical adaptations that allow them to rely on the lipids stored in their blubber as a source of energy during annual fasting periods 37 . During times of energy shortage, TAGs stored in lipid droplets are hydrolyzed to FAs and glycerol via lipolysis for subsequent use by other organs, and dolphins had more rapid release of non-esterified FAs than other mammals 38 . In addition, the water derived from metabolism, particularly lipolysis, is considered to be the primary sources of fresh water for cetaceans 39 . MGLL hydrolyzes MAG, thereby producing glycerol and FAs 6 . Therefore, positive selection of MGLL might be helpful for rapid release FAs to producing energy and to more efficiently extracting water in cetaceans, which in turn allows them to survive annual fasting periods with limited amounts of TAG stored in body tissues.
Cetacean blubber was vertically stratified and each blubber layer performs a different function, with the stable outer layer used for structural support and the more variable inner layer used for energy storage 40 . A suitable blubber thickness is essential for cetaceans to adapt to the aquatic environment. However, when in an emaciated state during the fasting period, the dual roles of the blubber in providing insulation and storing metabolic energy are in direct conflict 20 . The absence of a mechanism to regulate or control the utilization of stored TAGs in the blubber as an energy resource can result in the excessive use of TAGs during the fasting period, which in turn might greatly decrease the thickness of blubber and further weaken the activities of thermoregulation, buoyancy control, streamlining, and locomotion that are essential for aquatic life. Amazingly, evidence of positive selection in two genes involved in the regulation of lipolysis (i.e., PDE3B and PLIN1) was detected in cetacean-specific lineages, which maybe a mechanism for avoiding excessive lipolysis. PDE3B is a very powerful regulator of adipocyte lipolysis that is triggered by a decrease in cAMP levels 41 . The importance of PDE3B in suppressing adipocyte lipolysis has been demonstrated in PDE3B null mice 42 . PLIN1 is a key member of the PLIN family that encode for proteins that cover the lipid droplets in adipocytes, as well as regulate the coordination of lipid storage and utilization in various cell types 36 . It has been suggested that PLINs play an important role in inhibiting lipolysis when it is unphosphorylated 43 . Considering the function of these genes in regulating and inhibiting TAG lipolysis, positive selection of these genes suggests that cetaceans have evolved an enhanced capacity for inhibiting unrestricted lipolysis and finely control the fatty acid content of blubber, and therefore is important for the maintenance of a suitable thickness of the blubber layer.
Blubber's thickness and lipid content extensively varies across cetaceans. For example, the lipid content of harbor porpoise blubber ranges from 76% to 88%, whereas that of minke whales ranges from 42% to 96% 20 . The thermal conductance capacity of the blubber is highly dependent on both its conductive quality and quantity (i.e., thickness), and the difference in the thickness and lipid content of the blubber among cetaceans might be the result of their adaptation to different aquatic habitats. Therefore, positively selected genes are distributed throughout almost the entire cetacean phylogeny, from the most common ancestral branch of cetaceans to the terminal branches (Fig. 2). A series of positively selected sites observed in 24 genes were localized in or near the functional regions on the crystal structure of the corresponding genes ( Supplementary Fig. S2 and Supplementary Table S7), which indicates the ongoing adaptive evolution of cetacean TAG metabolism-related genes. It is also reasonable to assume that different cetaceans require a fine blubber layer, which has driven the relevant genes to evolve in response to continuous changes in the aquatic environment since their origin and subsequent diversification in waters across the globe.
Remarkably, APOB, the primary lipid-binding protein of chylomicrons and low-density lipoproteins (LDL) 44 , was determined to have undergone positive selection in the lineage leading to the common ancestor of each marine mammalian groups (i.e., cetaceans, pinnipeds, polar bear, manatee) and the branch of the combined marine mammals (Supplementary Table S2), which is consistent with the previous findings that strong adaptive selection was detected in the APOB gene of polar bears 45 . In cetaceans, blood-based LDL levels are much lower than humans 46 , and APOB may play an important role in the LDL redistribution in the blubber of cetaceans. In addition, blubber is a critical component of mammalian adaptation to the aquatic environment, and cetaceans and other aquatic animals show remarkable similarities in blubber structure and function 37 . Comparative genomic analyses have found that convergent amino acid substitutions were widespread throughout the genome, and PNLIP and PLIN4 genes were identified to be positively selected respective along the combined marine mammal branch and the cetacean branch in Foote et al. (2015) 47 . In addition, 68 parallel/convergent nonsynonymous amino acid substitutions were identified in 14 positively selected genes detected. These results further supported the adaptive evolution of cetacean TAG metabolism-related genes, and suggest that cetaceans and other marine mammals have apparently been under similar pressure for adipose tissue development and fatty acid metabolism during their adaptation to the aquatic environments. However, further investigation on the functional verification of TAG metabolism-related genes in cetaceans and other marine mammals is necessary in the future to determine their role in aquatic adaptations. Moreover, some positively selected genes were also detected in the terrestrial groups, suggesting the importance of TAG metabolism during their adaptation to various environments, and further research should be focused on this interesting phenomenon to interpret the roles of TAG metabolism in terrestrial mammals. The present study is the first comprehensive and systematic analysis of the molecular genetic basis of blubber thickening in cetaceans. Wide and strong positive selection was detected in several genes involved in TAG synthesis, lipolysis, and regulation, which is concordant with the important functions of cetacean blubber in thermoregulation, buoyancy control, streamlining, metabolic energy storage, and locomotion. Interestingly, some regulation genes that inhibit lipolysis also showed significant evidence of positive selection, which suggests that cetaceans have evolved an enhanced capacity for inhibiting unrestricted lipolysis, particularly during fasting. This study provides novel insights into the effective and complex mechanism of maintaining a suitable blubber layer thickness in cetaceans, and also implies that 'obesity' might be an indicator of good health for cetaceans, compared to that determined in humans, which has been strongly associated to various chronic diseases.

Methods
Samples and DNA sequencing. Five cetacean species (two mysticetes and three odontocetes): common minke whale, Omura's whale, Beluga, Finless porpoise, and long-beaked common dolphin were sequenced during this study. All 5 cetacean samples used in the present study were collected from dead individuals in the wild, and sampling was conducted systematically in accordance with all ethical guidelines and legal requirements in China. The protocol of this study was approved by the Institutional Review Board of Nanjing Normal University (NNU). Voucher specimens were preserved at Jiangsu Key Laboratory for Biodiversity and Biotechnology, College of Life Sciences, Nanjing Normal University (NNU), China.
Total genomic DNA was extracted from muscle with a standard phenol/chloroform procedure followed by ethanol precipitation 48 . The DNA integrity was checked by 1% agarose gel electrophoresis. Primers were designed for the conserved regions based on an alignment of genomic data from the cow (Bos taurus) (http://asia.ensembl.org/Bos_taurus/Info/Index) and bottlenose dolphin (Tursiops truncatus) (http://asia.ensembl.org/Tursiops_truncatus/Info/Index). All PCR amplification were conducted using a BioRAD PTC-200 with 2 × EasyTaq PCR SuperMix (TransGen Biotech) and the following profile: 34 cycles at 94 °C for 5min, 94 °C for 30 s, 53 °C− 59 °C for 30 s, and 72 °C for 30 s, followed by a 10 min extension at 72 °C. The amplified PCR products were purified and sequenced in both directions using an ABI 3730 automated genetic analyzer. Three to five repeated amplification for each gene were conducted and resequenced to confirm its sequence. The specificity of these newly generated sequences was examined by comparison with the published nucleotide database at GenBank by BLAST (NCBI). Table S1) were sequenced in the five cetacean species mentioned above and the newly sequences were deposited in GenBank under accession numbers KR135543-KR135734. The exons of each gene were sequenced and concatenated before being analyzed together. Only high-quality and high-integrity sequences were used in the analysis.  Table S1) and 23-50 species from representatives mammalian lineages (i.e., Cetaceans, Artiodactyla, Chiroptera, Rodentia, Carnivora, and Primates) were analyzed in the present study (Supplementary Table S8 and Supplementary Table S9). Nucleotide sequences of each gene examined and their deduced amino acid sequences were aligned separately using MUSCLE 3.8 50 and MEGA 5.0 51 , and manually adjusted with GeneDoc.

Molecular evolutionary analyses.
The codon-based maximum likelihood models implemented in CODEML program in PAML 4.7 25 were applied to estimate the rates of synonymous (dS) and nonsynonymous substitutions (dN), as well as dN/dS ratio (omega, ω ). The non-synonymous to synonymous rate ratio ω indicates changes in selective pressures, where ω = 1, ω < 1, and ω > 1 correspond to neutral evolution, purifying, and positive selection, respectively. The well-supported phylogeny of Laurasiatheria 52 and Primates 53 was used as the input tree in all analyses (Tree file: Supplementary Fig. S1).
Positive selection was detected using branch-site model A, in which ω can vary among sites along specific lineages 54 . Modified branch-site model A (test 2) was performed for every gene in each foreground lineage, which facilitated the analysis of datasets, including all mammals (branches a, t-aa in Supplementary Fig. S1) or cetaceans only (branches b-s in Supplementary Fig. S1). To identify the probabilities of sites under positive selection in each gene for the cetacean species examined, site models were implemented where ω could vary among sites. All the positively selected sites in site models were identified by using Bayes Empirical Bayes (BEB) analysis 25 with posterior probabilities of ≥ 0.80. The likelihood ratio test (LRT) statistic (2Δ L) approximates to a Chi-square distribution and was used to compare nested likelihood models. In addition, the improved statistical methods in Datamonkey web server 55 , which computed nonsynonymous and synonymous substitutions at each codon position, was used to further evaluate the selection. Sequences of each gene in cetaceans-only dataset were analyzed by using two distinct models, namely, fixed-effect likelihood (FEL) and random effect likelihood (REL).The FEL model estimates the ratio of dN/dS on a site-by-site basis, without assuming an a priori distribution across sites. The REL model first fits a distribution of rates across sites and then infers the substitution rate for individual sites. Sites with P values < 0.1 for FEL, and Bayes factor > 50 for REL were considered as candidates under positive selection 55 .
Some studies have suggested that the ML method of evaluating positive selection might produce false-positive results even when no positively selected sites exist 56 or when positively selected sites and negatively selected sites are mixed 57 . Further support for the PAML results were obtained using a complementary protein-level approach implemented in TreeSAAP 28 . TreeSAAP compares the magnitude of property changes of non-synonymous residues across a phylogeny and identifies specific amino acid properties that have likely been affected by positive destabilizing selection during evolutionary 58 . Identification of parallel/convergent sites among marine mammals. The parallel/convergent sites among marine mammals were identified according to the methods previously described 47 . In detail, we reconstructed the ancestral sequences for 26 positive selection genes using the Codeml program in PAMLv4.7. For each of the four marine mammal groups (cetaceans, pinnipeds, polar bear and manatee) the extant sequences at each position were compared to the ancestral sequence at the node corresponding to the most recent ancestor. We used the software CONVERG 2 59 to test whether the number of observed parallel/convergent amino acid substitutions was significantly higher than that expected by chance, given the total numbers of amino acid replacements in the two evolutionary lineages under investigation. The positions of the parallel/convergent nonsynonymous amino acid substitutions that were found in positively selected genes are shown in Supplementary Table S3. Mapping of positively selected sites onto protein structures. To gain insights into the functional significance of the putatively selected sites, we mapped the sites under positive selection to crystal structures. The 3D structures of genes under positive selection were predicted by using the homology modeling software provided by the I-TASSER server 60 . The protein sequences of positively selected genes were derived from the common bottlenose dolphin (Tursiops truncutus) genome, which were obtained from the Ensembl genome database (http://www.ensembl.org/index.html). In addition, the functional information of genes identified as being under positive selection was derived from the uniprot (http:// www.uniprot.org/).