Background

Evolutionary transitions, understood as the acquisition of a novel lifestyle within a lineage (e.g. [1]), encompass drastic modifications in several physiological systems (e.g. [2, 3]), to sustain the impacts of a novel environment in physiological homeostasis. Cetacea, an iconic group of aquatic mammals composed of baleen whales (Mysticeti), toothed whales, dolphins and porpoises (Odontoceti), are a “poster child for macroevolution” and typify such an evolutionary process [4]. The number of phenotypic modifications and adaptive traits associated with this land-to-water evolutionary transition is astonishing and implied a profound reorganization in different organ systems (e.g. [4,5,6]; Fig. 1). In fact, many cetacean physio-anatomical traits set them apart as unique among mammals. One remarkable example is their circadian rhythmicity, which was adapted by abolishing neurochemical signals like the hormone melatonin and the neuropeptide cortistatin [5, 7,8,9]. This alteration was accompanied by anatomical and physiological rearrangements, as evidenced by their lower cortical orexinergic bouton density, when compared to closely related terrestrial artiodactyls [10]. Yet, the precise molecular events underpinning the coalescence of these interconnected systems into novel physiological states remain largely unexplored. Among the various chemical messengers modulating physiological functions, neuropeptides are central in regulating various physiological mechanisms [11]. They are defined as short sequences of amino acids synthesized and released by neurons or glia. Moreover, they are responsible for slow-onset, long-lasting modulation of synaptic transmission, acting on neighbouring cells via G protein-coupled receptors (GPCRs) [12].

Fig. 1
figure 1

Evolutionary transitions and pleiotropic molecules. Land-to-water habitat transition experienced by marine mammals such as Cetacea was accompanied by remarkable anatomical, physiological and behavioural adaptations. In this study, we aim to assess the coding status of several neuropeptides with distinct physiological roles. Fossil timeline (grey bars) depicts the estimated ages of the appearance and extinction of key cetacean ancestors. Dating of these events was obtained in the following references [100,101,102,103,104,105,106,107]. In the lower panel, references for each reported role associated with neuropeptides are provided by [14, 18, 26, 47, 53, 56, 58, 60, 66, 108,109,110,111,112,113,114,115,116,117,118]

Full size image

In this study, we delve into the evolution of several neuropeptide and receptors, with an important role in the coordination of different physiological compartments that have been clearly modified in Cetacea: including the liver, lungs, heart and brain (Fig. 1; [13]). It is worth noting that neuropeptides and their receptors often exhibit pleiotropic effects, implying a potential wide range of physiological functions. As such, neuropeptide FF-amide peptide precursor (NPFF) seems a major player in pain modulation [14, 15], but also has a role in cardiovascular regulation (e.g. [16]). Pyroglutamylated RFamide peptide (QRFP) and its receptor (QRFPR) play a role in the control of feeding behaviour and also participate in sleep regulation [17,18,19]. Neuropeptide B (NPB) and one of its receptors (NPBWR2) are also important for the control of feeding behaviour, as well as pain modulation and sleep regulation [20,21,22,23]. Neuropeptide S (NPS) and its receptor (NPSR1) are central to sleep regulation and are linked to the control of feeding behaviour [24,25,26]. Neuromedin B receptor (NMBR) has been linked with both hormonal regulation and immunity control [27, 28]. Somatostatin receptor 4 (SSTR4) plays a role in hormonal regulation and in stress responses [29]. Finally, neuropeptide Y6 receptor (Npy6r) was associated with hormonal regulation and control of feeding behaviour [30]. Such pleiotropy raises the critical question on how these regulators evolved in such a drastic ecological transition as that experienced in Cetacea evolution. Strikingly, by expanding the analysis to non-Cetacea 161 mammalian species (22 taxonomic orders; Additional file 1: Table S1), we also revealed a complex mutational landscape, possibly indicative of distinct adaptations to multiple environments.

Results

Variable patterns of sequence alterations are found in two RFamide neuropeptides and their receptors

Comparative analysis of the neuropeptide FF-amide peptide precursor (NPFF) exonic sequences revealed localized events of gene pseudogenization across the mammalian tree. More specifically, 38 mammal species presented a PseudoIndex higher than 2 (Additional file 1: Table S2). This built-in assistant metric incorporated in PseudoChecker [31] assesses the sequence erosion status of the tested genes on a discrete scale ranging from 0 (coding) to 5 (pseudogenized).

Subsequent manual annotation allowed the identification of non-conserved open reading frame (ORF) disruptive mutations in both Mysticeti and Odontoceti species, including loss of start codon or indels in exons 1 and 2 (Additional file 1: Table S3). When expanding our search to other mammalian lineages, deleterious mutations were found in kangaroo rats (Dipodomys spp.), Hawaiian monk seal (Neomonachus schauinslandi) and flying foxes (Pteropus spp.) (Fig. 3); in the latter a conserved mutation was validated by SRA searches (Additional file 2: Figs. S1–S4). Also, although no inactivating mutations were discovered, exons 1 and 2 in Canidae (Canis spp. and Vulpes spp.) were not found (Additional file 1: Table S3).

Next, we sought to characterize the coding status of both NPFF receptors. PseudoChecker analyses for neuropeptide FF receptor 1 (NPFFR1) resulted in 127 species presenting a PseudoIndex > 2, contrasting with neuropeptide FF receptor 2 (NPFFR2) where only 48 species seem to yield eroded genes (Additional file 1: Tables S4 and S5). Annotation of collected cetacean genomic sequences revealed extension of the NPFFR1 gene across all analysed species, due to a conserved 11 nucleotide deletion in exon 4 (Fig. 2; Additional file 1: Table S6). For NPFFR2, most of the cetaceans displayed several disrupting mutations including a conserved 2 nucleotide deletion in exon 3 within Delphinoidea (oceanic dolphins, porpoises, beluga and narwhal) and within Physeteroidea (sperm whales) (Fig. 2; Additional file 1: Table S7). In both these receptors, the majority of the ORF-disrupting mutations are found in the last exon, which translates in ~ 20% of the total gene content completely modified (for NPFFR1) and for NPFFR2 more than 25% of the total gene length is not transcribed given the change of frame due to a frameshift mutation.

Fig. 2
figure 2

Pseudogene annotation in Cetacea. Examples of ORF-disrupting mutations in Cetacea for the study genes and respective sequence alignment. Each box represents an exon and lines represent intronic regions. An exhaustive description of the reported disruptive mutations can be seen in Additional file 1: Tables S6, S7, S9, S11, S13, S15, S17, S19 and S21

Full size image

In other mammalian species, a highly eroded NPFFR1 was also found within Chiroptera (bats), where the majority of the analysed species exhibited several conserved ORF-abolishing mutations and/or lack of exons (Fig. 3; Additional file 1: Table S6). Moreover, ORF-disrupting mutations were also identified in other mammals; however, data from independent SRAs only validated mutations in Canis spp., red fox (Vulpes vulpes) and naked mole-rat (Heterocephalus glaber) (Additional file 1: Table S7). SRA validations for NPFFR1 and NPFFR2 are presented in Additional file 3 (Figs. S1–S10) and Additional file 4 (Figs. S1–S11).

Fig. 3
figure 3

Schematic representation of gene loss events along the mammalian tree. Phylogenetic relationships were adapted from Vazquez [119]. For each gene, we represented in green the species where no ORF-disrupting mutations (frameshift mutations, in-frame premature stop codons, loss of canonical splicing site) were found. On the other hand, species presenting ORF-disrupting mutations are highlighted in red. In orange are symbolized cases where there is some mutational evidence, however the coding status cannot be entirely deduced (e.g. inactivating mutations that occur at the beginning or end of a gene). True absence of exons in the study genes is illustrated in blue. Species presenting (at least) one exon with poor alignment identity with the corresponding reference orthologue are represented in grey. Finally, question marks represent the cases where the absence of a given exon either likely results from the fragmentation (Ns) or incompleteness of the scaffold in the genomic region containing the target gene. An exhaustive description of the reported disruptive mutations can be seen in Additional file 1: Tables S3, S6, S7, S9, S11, S13, S15, S17, S19, S21, S23 and S25

Full size image

Within Cetacea, PseudoChecker analyses for pyroglutamylated RFamide peptide (QRFP) resulted in 20 species presenting a PseudoIndex higher than 2 (Additional file 1: Table S8). In cetaceans, most of the species showed either an eroded or sequence-modified QRFP with conserved mutations within Mesoplodon spp. (beaked whales) and Delphinidae (Fig. 2; Additional file 1: Table S9). Specifically for baleen whales and oceanic dolphins presenting only mutations at the beginning of the gene, we may not rule out the possibility of having a functional QRPF in these species, as the peptide does not suffer any change of frame while transcribed. Investigation of QRFP inactivation in other mammalian lineages revealed episodes of gene loss (Fig. 3). In particular, QRFP was found to be inactivated and subsequently validated in species such as Pteropus spp.—with the presence of a conserved 1 nucleotide deletion among members of this group—lemurs (Lemuroidea) and echidna (Tachyglossus aculeatus) (Additional file 1: Table S9 and Additional file 5: Figs. S1–S15). In the case of pyroglutamylated RFamide peptide receptor (QRFPR), 64 species presented signs of loss of the in-study gene (Additional file 1: Table S10), comprising mostly cetaceans and chiropterans (Figs. 2 and 3). Accordingly, in all analysed cetaceans the orthologous exon 2 of QRFPR was not found and the exon 5 was either not found or exhibited a conserved in-frame premature stop codon (Additional file 1: Table S11). Moreover, flying foxes showed, once more, evidence of inactivation with several conserved mutations in their coding reading frame (Additional file 1: Table S11). Other examples of species presenting a sequence-altered QRFPR include the California sea lion (Zalophus californianus) and echidna (Fig. 3; Additional file 1: Table S11 and Additional file 6: Figs. S1–S12).

Convergent gene loss among marine mammals of the neuropeptide S and receptor

We next examined the coding status of neuropeptide S (NPS) and neuropeptide S receptor 1 (NPSR1) in mammals to investigate independent gene loss events across the mammalian phylogeny. Although 96 species presented a PseudoIndex higher than 2 for NPS (Additional file 1: Table S12), only odontocetes (toothed cetaceans) except sperm whale (Physeter macrocephalus) and Indus River dolphin (Platanista minor), and West Indian manatee (Trichechus manatus latirostris) exhibited valid truncating mutations (Fig. 3; Additional file 1: Table S13 and Additional file 7: Figs. S1–S5). Regarding NPSR1, PseudoChecker analyses reported a total of 85 species labelled with a PseudoIndex higher than 2 (Additional file 1: Table S14). Annotation of collected genomic sequences revealed NPSR1 gene erosion mainly in two groups: in toothed cetaceans and aquatic carnivores (pinnipeds and otters) (Figs. 2 and 3). While in the first group pseudogenization was confirmed either by the absence of several exons and/or the presence of conserved deleterious mutations, for the latter different ORF-disrupting mutations were retrieved across species (Additional file 1: Table S15). In addition to these two groups, we also found evidence of NPSR1 inactivation in Spanish mole (Talpa occidentalis), Damaraland mole-rat (Fukomys damarensis) and Chinese tree shrew (Tupaia chinensis) (Fig. 3; Additional file 1: Table S15 and Additional file 8: Figs. S1–S14).

Inactivation of NPB and NPBWR2 genes is found in several mammalian lineages

Sequence search and analysis for neuropeptide B (NPB) in 202 mammalian genomes returned a total of 46 species with PseudoIndex higher than 2 (Additional file 1: Table S16), the majority due to fragmentation of the genomic region (presence of Ns), true absence of exons, poor alignment identity or incompleteness of the scaffold in the NPB genomic region (Additional file 1: Table S17). From these species, members of Delphinoidea presented conserved (and validated) frameshift mutations in exon 1 (Fig. 2; Additional file 1: Table S17 and Additional file 9: Figs. S1–S7). With respect to neuropeptides B and W receptor 2 (NPBWR2), 107 species displayed a PseudoIndex higher than 2 (Additional file 1: Table S18). However, in 52 species with NPBWR2 putatively pseudogenized, such coding status was predicted due to a true absence of exons, poor alignment identity or fragmentation of the genomic region (presence of Ns) (Fig. 3; Additional file 1: Table S19). Notwithstanding, ORF-disrupting mutations were identified in 58 species mostly affecting Cetacea and Carnivora orders, but also members of Lagomorpha, Perissodactyla, Proboscidea, Primates and Rodentia orders (Fig. 3; Additional file 1: Table S19 and Additional file 10: Figs. S1–S15).

Recurrent erosion of a neuromedin, a somatostatin and a neuropeptide Y receptors in mammals

We additionally investigated the coding condition of three neuropeptide receptors given their “low-quality” tag in Cetacea genome annotations: neuromedin B receptor (NMBR), somatostatin receptor 4 (SSTR4) and neuropeptide Y receptor Y6 (Npy6r). For NMBR, 35 species displayed a PseudoIndex higher than 2 (Additional file 1: Table S20); however, validated ORF-disrupting mutations were only determined in toothed whales and golden snub-nosed monkey (Rhinopithecus roxellana) (Fig. 3; Additional file 1: Table S21 and Additional file 11: Figs. S1–S8). On the other hand, a detailed analysis of SSTR4 occurrence in mammalian genomes revealed an extensive level of gene loss in mammals, with 104 species presenting a PseudoIndex higher than 2 (Additional file 1: Table S22). In general, no ORF was detected in the genomes of cetaceans, even-toed ungulates, marine and dog-like carnivorans, bats, pangolins, horses, mole-rats, manatee and aardvark (Fig. 3; Additional file 1: Table S23 and Additional file 12: Figs. S1–S20). For Npy6r, a total of 133 species exhibited a PseudoIndex higher than 2 (Additional file 1: Table S24), with gene lesion events found and validated in most of cetaceans, even-toed ungulates, primates, carnivores, eulipotyphlans and finally in some rodents (Fig. 3; Additional file 1: Table S25 and Additional file 13: Figs. S1–S22).

RNA-seq expression

Further examination of several transcriptome datasets of four different cetacean species (Additional file 1: Table S26) allowed the assessment of the functional condition of several neuropeptides and receptors reported in this study. Priorly, assessment of the quality of the transcriptomic datasets was performed by checking transcript abundance in genes reported as being highly expressed in a specific tissue, using the Human Protein Atlas as a proxy (http://www.proteinatlas.org). Overall, we could confirm high levels of transcript abundance in the expected tissues, except in testes and brain where no/low expression was found in the expected genes (Additional file 1: Table S27).

Across all species, there was a low number of mRNA reads of our target genes, except in NPFF (for all species), NPB (for Balaenoptera acutorostrata), QRFPR (for Balaena mysticetus) and NPFFR1/NPFFR2 (for Monodon monoceros) (Additional file 1: Table S28).

After counting the number of spliced reads, exon–intron reads and exonic reads in genes presenting some gene expression, transcriptomic data presented a substantially high proportion of exon–intron reads versus spliced reads in NPFF, in stark contrast to the pattern found for QRFPR (for B. mysticetus), NPFFR1 (for Delphinapterus leucas) and NPFFR2 (for M. monoceros) (Additional file 1: Table S28). Moreover, further verification of the presence of ORF disruptive mutations in the produced transcripts allowed the detection of at least one premature stop codon (or lack of terminal stop codon as observed in NPB and NPFFR1 for Monodontidae) in the transcripts of the analysed cetacean species (Additional file 14: Figs. S1–S2; Additional file 15: Figs. S1–S2; Additional file 16: Figs. S1–S3; Additional file 17: Figs. S1–S3; Additional file 18: Fig. S1; Additional file 19: Figs. S1–S2 and Additional file 20: Figs. S1–S2) revealing either the production of truncated or extended proteins. This is especially relevant in cases where we observed a considerably distinct ratio of exon–intron reads/spliced reads among the remaining species (QRFPR for B. mysticetus, NPFFR1 for D. leucas and NPFFR2 for M. monoceros) (Additional file 1: Table S28).

Selection analyses indicates relaxation of purifying selection in NMBR of Mysticeti

To examine whether genes that are inactivated in odontocetes but not in mysticetes (QRFP, NPS, NPSR1, NPB and NMBR) exhibit a relaxation in the intensity of selection (both positive and negative), we employed the RELAX method, which explicitly incorporates a selection intensity parameter (k). With the exception of NMBR, for which the RELAX test for selection relaxation (k = 0.17) yielded significance (p = 0.01, LR = 10.38), we did not find any evidence for relaxation in the selection intensity for the other genes (Additional file 1: Table S29). Further analyses using aBSREL and BUSTED did not reveal any evidence for positive selection in the Mysticeti clade for the NMBR gene (Additional file 1: Table S30). Therefore, the observed rate of acceleration in mysticetes likely indicates a trend of relaxed purifying selection rather than an episode of positive selection.

Discussion

From an evolutionary perspective, how the neuronal control of physiological systems modifies and adjusts to radical habitat transitions is mostly unknown. Importantly, these modifications translate into considerable alterations at the anatomical (i.e. pineal gland loss [32]), physiological (i.e. hypoxia-induced adaptations [33]) and behavioural levels (i.e. sleep [34]). Still, the molecular basis of such changes—in perception, motor control, sleep, spatial control, motivation and learning/memory—is challenging to assess. Here, we used as proxy the plethora of neuropeptides known to regulate a vast network of functions: including sleep, feeding, pain, stress, immunity (Fig. 1). Some of the reported receptors have previously been identified as lost in early vertebrate lineages, such as SSTR4 in ray-finned fish [35], or in the ancestors of mammals and birds, like QRFPR [36]. Additionally, certain receptors, such as NPFFR2 [37], QRFPR [5], SSTR4 [35] and Npy6r [38], have been reported as lost in other mammals.

Comparative analysis of 40 cetacean genomes uncovered an extensive loss pattern of several neuropeptides and receptors (Figs. 2 and 3). The marked gene loss pattern of pleiotropic actors (e.g. [39]), acting on different tissues, anticipates the drastic modification of distinct gene regulatory networks as a result of adaptation to a fully aquatic lifestyle. However, regarding some neuropeptides and receptors (e.g. NPB), Mysticeti (baleen whales) seem to retain functional genes (Fig. 2). A similar outcome was previously observed in other organ systems (e.g. [40]), thus implying the existence of critical changes in selective pressures over the time in these two related lineages. Nevertheless, our results also suggest that the intensity of purifying selection acting on Mysticeti NMBR has been relaxed, which might indicate earlier stages of gene inactivation. Therefore, at least for NMBR, it is possible that loss of function has occurred on the stem cetacean branch, rather than on the branch leading to the radiation of odontocetes.

Circadian rhythmicity is one of the most prominent behavioural adaptations in Cetacea. A strong association between the loss of the full melatonin-related gene hub and the existence of unique bio-rhythmicity in cetaceans has been described [5, 7, 9]. Moreover, disruption of neuropeptides and receptors, such as the neuropeptide cortistatin and the dopamine receptor D5 (DRD5), has been correlated with changes in daily activity patterns and energy metabolism [8, 41]. Thus, the absence of a subset of neuropeptides and receptors may have consequences in terms of the neuroendocrine regulation of Cetacea circadian rhythmicity and sleeping behaviour. In agreement, some of the analysed receptors are expressed in hypocretin/orexin neurons (i.e. NPFFRs) [42] and QRFP was suggested to partially co-localize with orexin [43]. The orexinergic system plays a crucial role in regulating sleep/wake rhythms and thermoregulation, but also in cardiovascular responses, feeding behaviour, spontaneous physical activity and control of energy metabolism [44]. Given the differences in terms of orexinergic bouton density [10] and number of hypothalamic orexinergic neurons [45] found in Cetacea in comparison with their closest relatives (even-toed ungulates), it is expected that disruption of these endocrine messengers may influence cetacean’s unique behaviour. Strikingly, we report the loss of sleep inducing (QRFP, NPB) [17, 19, 22] or arousal (NPS) [24, 26] neuropeptides, and related receptors, in most Cetacea. Yet, sleep-modulators somatostatin (SST) and neuropeptide Y (NPY) show no signs of erosion [46, 47]. In Cetacea, the loss of multiple neuropeptide sleep regulation seems to have paired with a novel rearrangement of sleep: with the loss of REM sleep and the transition from bihemispheric into unihemispheric sleep [34]. Such transition likely entailed novel regulatory networks acting on a single brain hemisphere rather than the brain as a whole. In agreement, the emergence of Cetacea unihemispheric sleep was accompanied by local thermoregulatory adjustments: uncoupling the brain hemispheric temperature and stabilizing brainstem temperature to prevent sleep-inducing cooling and maintaining motor and sensorial functions (Fig. 4) [48, 49]. In addition, hemisphere communication was also adjusted, with the occurrence of interhemispheric fascicle projections, directly into dorsal thalamus nuclei instead of the brainstem projections observed in other mammals [34]. Such functional and anatomical changes possibly allowed the simultaneous vigilance and sleep needed for a mammal to inhabit the high sea.

Fig. 4
figure 4

Loss of neuropeptides and receptors contributed to adaptation in an aquatic environment. Impairment of neuropeptides and receptors led to drastic modifications in terms of response to physiological stress, biological rhythmicity and metabolism

Full size image

A signalling shift seems to have also occurred in the context of the Cetacea diving response: during which the heart rate is reduced (bradycardia) and the consequent drop in arterial blood pressure is restrained through a strategy of peripheral vasoconstriction, to maintain blood distribution to heart and brain during diving apnoea [50]. Previous reports associated events of gene loss and adaptive gene evolution (positive selection) to the modulation of Cetacea blood volume, pressure and peripheral vasoconstriction [41, 51, 52]. The loss of the hypertensive and heart rate-modulating neuropeptides and receptors (NPFF, QRFP, NPB; cognate receptors and NMBR) seems to further contribute to the blood pressure and heart rate regulation [16, 17, 53,54,55,56]. NPFF in particular was shown to be responsible, in rat (Rattus norvegicus), for up to 50% of the cardiac component of the baroreflex, participating in the maintenance of the blood pressure levels, and to modulate the chemoreflex through action on ASIC channels, which elicits hyperventilation [16, 53, 57,58,59,60]. SSTR4 was also suggested to modulate ventilation, with airway contraction observed in knock-out mice [61]. On the other hand, the intact neuropeptides SST and NPY were suggested to modulate breathing, notably under hypoxic conditions, with SST attenuating ventilation and inducing apnoea responses [62, 63]. NPY is also known for its prominent effects on the cardiovascular system [64]; thus, conservation of NPY may importantly contribute to blood pressure regulation in this group.

Metabolism is yet another major target of neuropeptide regulation through feeding behaviour and energy expenditure (NPFF; NPB/NPW; QRFP; NMBR) [17, 20, 54, 65,66,67] or by controlling insulin secretion (QRFP; NPB/NPW) [18, 68, 69]. Accordingly, cetaceans are known to present unique metabolic needs to adapt to a low glucose diet and to the metabolic constraints of oxygen restriction (Fig. 4) [70, 71]. Thus, loss of these feeding suppression neuropeptides and metabolic modulators could contribute to fat signalling and deposition, as well as to the peripheral insulin resistance and high circulating glucose levels under fasting, as observed in odontocete cetaceans [72, 73]. These conditions seem to ensure a steady glucose supply to the central nervous system in the absence of alternative fuel sources—i.e. ketogenesis impairment resulting from genomic loss of hydroxymethyl-CoA synthase—in the context of their low carbohydrate, high protein and fat diet (Fig. 4) [74,75,76]. In agreement, NPY, an inducer of obesity and fat storage, was found intact, possibly contributing to the overall fat profile in Cetacea [77]. Although fat deposition and obesity trigger adipose tissue inflammation in most mammals, Cetacea exhibit a healthy fat phenotype [70, 78]. This could be concurrent with the loss of inflammatory modulation by NPFF, NPS or NPB/NPW [15, 79, 80]. In addition to adipose tissue inflammation, the loss of these neuropeptides and receptors, notably the SST receptor SSTR4 or NPBWR2 [23, 61], might have further helped cetaceans extend their diving capacities by attenuating lung pain and inflammation, resulting from lung collapse as well as the formation of gas bubbles during deeper dives [52, 80,81,82,83].

Interestingly, we observed a lack of conserved founding mutations common to all analysed Cetacea in most study genes, thus indicating that pseudogenization events took place after the diversification of modern Cetacea lineages (Fig. 2). Given the role of neuropeptides as master regulators of a different set of physiological processes, it is expected that, in the course of evolution, molecules with a pleiotropic action should suffer a progressive decay, in parallel with the modifications of the physiological compartments in which they are associated. Adaptation to specific ecological niches might have triggered the occurrence of disrupting mutations in different genomic positions/timescales, thus promoting independent pseudogenization events.

Our approach also identified evolutionary convergence in patterns of gene loss across marine mammals, NPS and NPSR1 (Fig. 3). Together, these results suggest that these convergent genomic variations possibly contributed to the transition from land to water in marine mammal lineages. This is a pattern previously reported for other genes [5, 84, 85] and reinforces the idea of gene loss as a remarkable molecular signature of adaptive evolution in habitat shifts. We also have found other convergent patterns of disruption less easy to explain from an evolutionary perspective. As an example, convergent inactivation of the NPFF and receptors (NPFF and NPFFR2 as in Cetacea results on NPFFR1 coding status were not completely clear) and QRFP and receptor in flying foxes (Pteropus sp.) and Cetaceans (Fig. 3) might be a genomic signature of adaptation to novel niches—an evolutionary convergence previously reported for other genes (e.g. [86]).

Conclusions

Our findings suggest that alteration/loss of neuropeptide and receptors parallel adaptive shifts in behavioural processes such as bio-rhythmicity, diving and feeding behaviour. On another level, the patterns of gene modification observed in this study reinforce the idea of gene loss as a major evolutionary driver in the land-to-water habitat transition experienced by the Cetacea ancestor.

Methods

Genomic sequence compilation

An exhaustive literature review was used to identify a list of key genes that encode neuropeptides and respective receptors with pleiotropic action in physiological compartments that have suffered drastic adjustments in the land to water transition. Consequently, NPFF, NPFFR1, NPFFR2, QRFP, QRFPR, NPS, NPSR1, NPB, NPBWR2, NMBR, SSTR4 and Npy6r corresponding genomic regions from 172 mammals with the genome annotated were collected from NCBI (Additional file 1: Table S1). Retrieval was performed using Genomic Sequence Downloader (https://github.com/luisqalves/genomic_sequence_downloader.py), using human synteny for each target gene as input—downstream and upstream flanking genes considered were the ones presenting a gene type as “protein coding”. Whenever Genomic Sequence Downloader failed to obtain a genomic region containing the target gene, there was a tempt to collect directly the corresponding genomic sequences from the reference genome assemblies available at NCBI—either using the target gene or the corresponding flanking genes to tell us the genomic region comprising the physical location of the gene. In addition to the species previously referred, 29 genomes from cetaceans with no annotated genome available and Hippopotamus amphibius (hippopotamus)—either from NCBI, DNA Zoo [87] or Bowhead Whale Genome Resource (http://www.bowhead-whale.org)—were also inspected. In this case, the available genome assemblies were searched through blastn using Bos taurus target gene corresponding orthologue coding sequence (CDS), as well as the CDSs of the flanking genes in the same species. Exception made for SSTR4 where Homo sapiens CDS for target gene and flanking genes were used as query. The best matching genome scaffold was retrieved. When no consensual blast hit was obtained, all hits corresponding to the query were inspected, the aligning regions submitted to a back-blast search against the nucleotide (nt) database of NCBI, with the matching genomic sequence(s) corresponding to the gene of interest being the one(s) selected for annotation (when existing).

Inference of the coding status

Gene annotation was firstly performed using a pseudogene inference pipeline, PseudoChecker [31]. For each run, the Homo sapiens (human) orthologous gene and genomic sequences were used as reference (NCBI Accession ID regarding human NPFF: NM_003717.4; NPFFR1: NM_022146.5; NPFFR2: NM_004885.3; QRPF: NM_198180.3; QRPFR: NM_198179.3; NPS: NM_001030013.2; NPSR1: NM_207172.2; NPB: NM_148896.5; NPBWR2: NM_005286.4; NMBR: NM_002511.4; SSTR4: NM_001052.4; default parameters were maintained). For gene annotation in cetaceans and hippopotamus, B. taurus CDS were USED as reference, whenever it presented a curated transcript (NCBI Accession ID regarding cattle NPFF: NM_174123.3; QRFP: NM_198222.1; QRFPR: NM_001192681.1; NPSR1: NM_001192977.2; NPB: NM_173944.1; NPBWR2: NM_174075.1; NMBR: NM_001205710.1). Finally, for Npy6r annotation, Mus musculus (house mouse) orthologous gene was used as reference (NCBI Accession ID: NM_010935.4). Estimation of the erosion condition of the tested genes was performed through PseudoIndex, a user assistant metric built into the software PseudoChecker, that ranges on a discrete scale from 0 (coding) to 5 (pseudogenized) [31]. PseudoIndex considers three key components: absent exons, shifted codons and truncated sequences, each quantifying various aspects of mutational evidence. The predicted sequences were classified into “functional” if PseudoIndex was between 0 and 2, or “putatively pseudogenized” when PseudoIndex was higher than 2. Manual gene annotation and validation was performed for species presenting PseudoIndex higher than 2, with the collected genomic sequences being uploaded into Geneious Prime® 2021.2.2 and the gene sequence manually predicted as described in Lopes-Marques [88] using the previous coding sequences from the same species as reference. Briefly, reference exons were mapped to the corresponding genomic sequences and subsequently aligned regions were manually inspected to find putatively ORF disrupting mutations (frameshifts (indels), premature stop codon, loss of canonical splice sites). Given the small size of the first exon in human NPFFR1 and NPS (only 7 and 8 base pairs, respectively), putative inactivating mutations in these exons were not considered. The identified ORF disrupting mutations (one per species) were validated by searching at least two independent Sequence Read Archive (SRA) projects (when possible) of the corresponding species. Conserved deleterious mutations were validated in only one species.

Assessment of gene expression in multiple tissues

The global analyses of relative gene expression in Cetacea species were performed in four cetacean species, using the approach described previously [89]. The genome and annotations of D. leucas (Accession number: GCF_002288925.2), M. monoceros (Accession number: GCF_005190385.1) and B. acutorostrata (Accession number: GCF_000493695.1) were downloaded from NCBI genome browser. On the other hand, the genome and annotation of B. mysticetus were retrieved from http://www.bowhead-whale.org/. Next, we collected the RNA-seq datasets from NCBI (Additional file 1: Table S26). When more than one dataset was available for a specific tissue, both were concatenated. Before proceeding with relative gene expression analyses, the genomes and annotations were checked for errors with (agat_convert_sp_gxf2gxf.pl) script and converted from.gff to.gtf format with (agat_convert_sp_gff2gtf.pl) script of AGAT tool [90]. The RNA-seq datasets were mapped against each genome with the Hisat2 v.2.2.1 aligner [91, 92], and the relative gene expression determined with the StingTie v.2.2.1 software [93], following the authors protocol [94]. In the end, the gene expression quantifications were used in transcript per million (TPM).

To assess the quality of each transcriptomic dataset, we calculated gene expression in a highly expressed gene according to the Human Protein Atlas (http://www.proteinatlas.org). Genes included in this analysis were the following: arginine vasopressin (AVP; for brain), phosducin (PDC; for retina), uromodulin (UMOD; for kidney), transition protein-1 (TNP1; for testes), nicotinamide riboside kinase 2 (NMRK2; for muscle), acrosin (ACR; for heart), secreted phosphoprotein 2 (SPP2; for liver), advanced glycosylation end-product specific receptor (AGER; for lung), C–C motif chemokine ligand 25 (CCL25; for lymph) and leptin (LEP; for blubber).

To inspect the occurrence of expression reads from our target genes, mRNA reads previously collected were mapped to corresponding annotated genes using the “map to reference” tool available in Geneious Prime® 2021.2.2 (maximum read gap size adjusted to the length of the maximum distance between exons with a maximum mismatch rate of 1%). The mapped reads were then classified as either spliced reads (spanning over two exons), exon–intron reads or exonic reads, based on the genomic region they mapped to. After classification, the reads of each class were counted, and aligning regions covering mutated regions were screened for mutated transcripts.

Molecular evolutionary analyses

Given the slow rate of evolution in Mysticeti [95], evidence for pseudogenization does not accumulate rapidly. For neuropeptide and receptor genes that undergo knockout in odontocetes, yet remain intact in mysticetes (QRFP, NPS, NPSR1, NPB, NMBR), estimation of the selection intensity acting on these genes (through dN/dS analyses) within Mysticeti would provide valuable insights. For this purpose, we translation-aligned the predicted sequences of Mysticeti (retrieved from PseudoChecker) and several mammals from different lineages without any inactivating mutations (see Additional file 1: Table S31) using the Blosum62 substitution matrix in Geneious Prime® 2021.2.2. To investigate potential relaxation of purifying selection, often linked with gene pseudogenization, we conducted RELAX analysis using the HyPhy package [96, 97]. This analysis involves comparing a foreground set of species (Mysticeti in this case) with a background set (all non-cetacean species) within a hypothesis-testing framework. In cases where significant evidence of relaxation was found, we additionally run aBSREL [98] and BUSTED [99] to ascertain whether positive selection has manifested across a subset of branches or to assess gene-wide (rather than site-specific) positive selection. The idea behind these analyses is that if there is no evidence of positive selection, the accelerated rate in a particular clade probably indicates a pattern of relaxed purifying selection.