Degenerated vision, altered lipid metabolism, and expanded chemoreceptor repertoires enable Lindaspio polybranchiata to thrive in deep-sea cold seeps

Abstract

Background

Lindaspio polybranchiata, a member of the Spionidae family, has been reported at the Lingshui Cold Seep, where it formed a dense population around this nascent methane vent. We sequenced and assembled the genome of L. polybranchiata and performed comparative genomic analyses to investigate the genetic basis of adaptation to the deep sea. Supporting this, transcriptomic and fatty acid data further corroborate our findings.

Results

We report the first genome of a deep-sea spionid, L. polybranchiata. Over long-term adaptive evolution, genes associated with vision and biological rhythmicity were lost, which may indirectly benefit oligotrophy by eliminating energetically costly processes. Compared to its shallow-sea relatives, L. polybranchiata has a significantly higher proportion of polyunsaturated fatty acids (PUFAs) and expanded gene families involved in the biosynthesis of unsaturated fatty acids and chromatin stabilization, possibly in response to high hydrostatic pressure. Additionally, L. polybranchiata has broad digestive scope, allowing it to fully utilize the limited food resources in the deep sea to sustain a large population. As a pioneer species, L. polybranchiata has an expanded repertoire of genes encoding potential chemoreceptor proteins, including ionotropic receptors (IRs) and gustatory receptor-like receptors (GRLs). These proteins, characterized by their conserved 3D structures, may enhance the organism’s ability to detect chemical cues in chemosynthetic ecosystems, facilitating rapid settlement in suitable environments.

Conclusions

Our results shed light on the adaptation of Lindaspio to the darkness, high hydrostatic pressure, and food deprivation in the deep sea, providing insights into the molecular basis for L. polybranchiata becoming a pioneer species.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12915-025-02112-2.

Background

Advances in ocean exploration have revealed that numerous animals, ranging from invertebrates (e.g., corals, sponges, worms, and mussels) to vertebrates (e.g., cartilaginous and bony fishes), are found in the deep sea—waters that constitute approximately 95% of the ocean’s volume [1, 2]. Spionidae is one of the largest and most diverse families of polychaetes, annelid worms found in almost all ocean habitats, from shallow waters to the deep sea [3, 4]. Spionids are considered opportunistic species characterized by early reproductive maturity, high fecundity, and short life spans, which enabled them to quickly colonize new habitats and emerge as dominant species in various benthic communities [5, 6]. Recently, many new species of the family Spionidae have been discovered in the deep sea [7–9]. The genus Lindaspio is an intriguing group of spionids endemic to deep-sea chemosynthetic ecosystems such as hydrothermal vents, methane seeps, and whale falls [10–12].

Lingshui Cold Seep was found to have erupted after 2019, located off Hainan Island in the South China Sea at a depth of 1700 m. This seep is characterized by typical deep-sea features, including complete darkness, high hydrostatic pressure, and high levels of reductive compounds like methane that characterize this seep. A continuous survey of the seep has revealed that the microbial communities have gradually changed over the past 5 years since its first recorded eruption [13]. Remarkably, rapid community succession has also been observed, with the cold seep sediment being colonized by L. polybranchiata. This species has emerged as the dominant species, developing a dense population with 30,000 individuals per square meter [14]. Within just 1 year, L. polybranchiata established a high-abundance population, making it a pioneer species in the early cold seep stage [14, 15]. Furthermore, L. polybranchiata is endemic to deep-sea chemosynthetic ecosystems and we discovered this species in Haima Cold Seep, around 150 km away from Lingshui Cold Seep. Its ability to become a pioneer species in deep-sea chemosynthetic ecosystems suggests the existence of unique adaptive mechanisms that allow it to thrive in extreme deep-sea environments and rapidly form large populations.

Morphological innovations of L. polybranchiata include enlarged gills (ostensibly an adaptation to hypoxia), an enlarged caruncle, and loss of eyes [14], but their genetic basis and molecular mechanisms are unknown. The enlarged caruncle, a sensory organ, in these worms may play an important role in detecting chemical clues and colonizing chemosynthetic ecosystems [14]. The loss of eyes and enlargement of sensory organs are common evolutionary responses to perpetual darkness in the deep sea [16–18]. Many deep-sea invertebrates and fish have evolved specialized chemosensory systems to detect and respond to chemical gradients in the water, ensuring survival in lightless conditions [19]. In contrast, some deep-sea organisms, including the silver spinyfin, have evolved a unique visual system with an expanded repertoire of rod opsin genes, enabling color vision in the dark and the ability to perceive bioluminescent signals [20]. Thus, this contrasting evolutionary path in deep-sea organisms warrants further investigation.

How largely undescribed deep-sea species evolved and adapted to their extreme environments can now be explored using genomic tools. Here, we assembled and annotated the genome of L. polybranchiata and juxtaposed it with its shallow-water Spionidae member Streblospio benedicti and other polychaete worm species to reveal candidate deep-sea adaptation genes.

Results and Discussion

Characteristics of the Lindaspio polybranchiata genome

We sequenced and assembled a 1.66 Gb L. polybranchiata genome with 8977 contigs and a contig N50 length of 303,282 bp using PacBio HiFi long reads with Hifiasm (Table 1). The assembly included 59.15% (0.98 Gb) repetitive elements, and 21,462 (77.3%) of the protein-coding gene models were functionally annotated (Additional file 1: Fig. S1). The BUSCO metazoan completeness score of the gene models was 90.1% (84.5% single copy and 5.6% duplicated genes) (Table 1). The average gene length was 12,902 bp, and the mean intron length was 2242 bp, similar to other polychaete species.

Table 1.

Statistics of the genome characteristics of Lindaspio polybranchiata

Genome size	1,664,211,210 bp
No. of contig	8977
N50 length	303,282 bp
GC content	38.73%
BUSCO score (genome)	C: 89.1% (S: 82.7%, D: 6.4%, F: 6.0%, M: 4.9%)
Repeat sequence	59.15%
No. of protein-coding genes	27,746
BUSCO score (gene set)	C: 90.1% (S: 84.5%, D: 5.6%, F: 3.7%, M: 6.2%)
Average gene length	12,902 bp
Average intron length	2242 bp
Average exon length	236 bp
Average intergenic length	26,935 bp

The L. polybranchiata assembly (1.66 Gb) was much larger than that of Streblospio benedicti (701.4 Mb) collected from the intertidal zone, which was also generated using long-read sequencing reads [21]. A proliferation of transposable elements often underlies large genome sizes across Metazoa [22–24]. There was a higher proportion (59.15% vs. 40.36%) of repetitive and other non-coding sequences in the L. polybranchiata assembly (Table 2). Many repetitive sequences (38.00%, average length 413 bp) were unclassified, hinting that further research is needed to classify annelid repetitive sequences.

Table 2.

Statistics of the repeat sequences in the genomes of Lindaspio polybranchiata and Streblospio benedicti

	L. polybranchiata	S. benedicti
DNA	6.58%	5.02%
LINE	5.70%	4.44%
SINE	2.21%	0.99%
LTR	1.23%	2.68%
Unclassified	38.00%	26.16%
Satellites	0.12%	0.31%
Simple repeats	4.36%	1.12%
Low complexity	0.17%	0.12%

Comparative genomics reveals L. polybranchiata deep-sea and pioneer species adaptations

Deep-sea conditions, including a reduced temperature and high hydrostatic pressure, may promote genome size increases that provide genetic material for evolutionary innovation [25, 26]. The genomes of 11 other polychaete worm species were obtained to provide a framework for comparative analyses [21, 27–35]. We identified 11,055 orthologous groups (OGs) with OrthoFinder, and a phylogenetic tree based on 134,940 amino acids of 528 single-copy OGs showed a topology consistent with a recent study that employed mitochondrial genomes in phylogenetic analyses [14]. L. polybranchiata showed a sister group relationship to S. benedicti with high support (bootstrap value: 100) (Fig. 1A). Although the clustering of two Spionidae species and Siboglinidae showed relatively low support (bootstrap value: 60), our results agree with the monophyly of clade Canalipalpata [36]. The divergence time between L. polybranchiata and S. benedicti was estimated at around 244.40 Ma (95% confidence interval of 150.16–346.29 Ma). During this time, magmatism of the Central Atlantic Magmatic Province (CAMP) and the Karoo-Ferrar Large Igneous Province (KFLIP) triggered mass extinctions [37]. Furthermore, the supercontinent Pangea had begun to rift, likely facilitating the colonization of the deep sea by coastal species [38]. These events may have promoted the divergence of deep-sea and shallow-water organisms.

Fig. 1.

Comparative genomics analyses of Lindaspio polybranchiata and other annelids. A Phylogenetic tree and divergence times of annelids. Gray lines indicate the divergence time with the highest posterior density (HPD); red dots, fossil calibrations. B Venn diagram of spionid and siboglinid gene families. C The top 20 enriched gene ontology (GO) terms among expanded L. polybranchiata gene families

A total of 1398 gene families were unique to L. polybranchiata compared to the four other Canalipalpata species in our dataset (Fig. 1B). These gene families are involved in glycosaminoglycan (GAG) and glycosphingolipid (GSL) biosynthesis, and signal transduction (Additional file 1: Fig. S2). Deep-sea cold seeps are rich in hydrogen sulfide (H₂S), a gas used by resident bacteria to fix carbon and provide essential sugars and amino acids, but this process also generates sulfur that is toxic at high levels [39, 40]. The unique L. polybranchiata GAG genes are associated with sulfur metabolism and include a homolog of chondroitin 4-sulfotransferase 11 (CHST11) hypothesized to play a role in the detoxification of sulfate to the cell matrix polysaccharide chondroitin 4′-sulfate by a population of the deep-sea vent-inhabiting lobster Shinkaia crosnieri [41]. The unique GSL genes (e.g., FUT1_2) could contribute to a more hydrostatic pressure-resistant cell membrane, a feature observed in studies of deep-sea organisms in the last 50 years [42] (and see “ High hydrostatic pressure tolerance”). Our analysis revealed that 95 shared gene families expanded in the L. polybranchiata genome, while 39 families contracted (Additional file 1: Fig. S3 and Additional file 2: Table S1). We performed GO and KEGG enrichment analyses to predict the functional outcomes of these events. The expanded gene families were primarily associated with processes such as unsaturated fatty acid metabolism, chitin binding, olfactory receptor activity, and chromatin stabilization (e.g., chromatin assembly or disassembly, DNA replication-dependent nucleosome assembly, and nucleosome assembly) (Fig. 1C, Additional file 2: Table S2). Contracted gene families included opsins and cryptochromes, genes associated with the sensory perception of light. Finally, we found that 135 genes with positive selection in L. polybranchiata showed enrichment for gene ontology terms associated with signal transduction, hormone secretion, and endocytosis (Additional file 1: Fig. S4). These genomic changes may reflect evolutionary innovations of L. polybranchiata, some of which are expanded upon below.

Evolution of eye loss

Waters below ~ 100 m lack sunlight and many sedentary deep-sea animals exhibit degenerated or modified eye structure [16, 43, 44]. Species within the Canalipalpata clade Sedentaria are no exception (Fig. 2A). Little research exists on how their eyes were lost, and the extent of genetic convergence that underlies such loss in marine worm lineages that diverged ~ 400 million years ago is unknown.

Fig. 2.

Gene families related to vision in marine sedentarians (spionids and siboglinids). A The anterior end of L. polybranchiata, dorsal view. ca, caruncle; pa, palp. B Diagram showing key transcription factors in eye development. Red crosses indicate gene loss in L. polybranchiata, disrupting Mitf and Otx pathways (dashed lines), essential for pigment cell specification. These interruptions may explain the lack of eyes. C The number of genes encoding eye transcription factors. D The number of genes involved in the vision and biological rhythm

All species in Siboglinidae, which inhabit ocean depths from 100 to 10,000 m, also lack eyes [45]. Previous studies have shown that transgenic expression of Ridgeia piscesae Pax6 (RpPax6; a transcription factor) in Drosophila leads to abnormal eye development, by repressing downstream transcription factors (e.g., sine oculis, so; a homolog of the vertebrate SIX homeobox genes) [46]. These findings suggest that similar disruptions in Pax6 function could underlie the loss of eyes in Siboglinidae. Our dataset revealed the amino acid changes in RpPax6 of the three other Siboglinidae species but not in Spionidae or Errantia (Additional file 1: Fig. S5). Next, we examined eye development transcription factors (the majority of which show functional conservation across vertebrates and invertebrates) of L. polybranchiata, the sighted S. benedicti, and the three blind siboglinids in our dataset. Two genes, Otx and Mitf, are only lost in L. polybranchiata (Fig. 2B, C). Additionally, a search for these genes in de novo-assembled transcriptome sequences and PacBio HiFi reads with tblastn yielded no hits, further supporting the loss of Otx and Mitf. Otx (orthodenticle-related homeobox) is essential for the formation of rhabdomeres, light-guiding rods in invertebrate microvillar photoreceptors, and is expressed earlier than any other eye field marker gene in the anterior neuroectoderm [47, 48]. Mitf (microphthalmia-associated transcription factor) acts in concert with Otx to regulate eye formation by supporting peripodial epithelium (PE), functionally analogous to vertebrate retinal pigmented epithelium [49, 50]. The lack of these transcription factor genes was likely essential for the loss of eyes by L. polybranchiata.

Adaptive convergent evolution between L. polybranchiata and Siboglinidae is therefore evident at a higher level, the eye development pathway: essential transcription factors are lost (L. polybranchiata Oxt and Mitf) or show loss-of-function (Siboglinidae Pax6) in distantly related, darkness-adapted species. We also identified a loss of all opsin and cryptochrome (CRY) photoreceptor genes in L. polybranchiata and the three Siboglinidae species in our dataset (Fig. 2D). The downstream photoreceptor genes likely became redundant in deep sea canalipalpates once their eyes were lost. Our results support the theory of convergent gene loss in species with similar ecological pressures (see [51] for review).

Biological rhythms in the deep sea

In L. polybranchiata and siboglinids, we observed loss of the genes encoding the photoreceptor CRY, which CHRONO regulates, and the essential master transcription factor CLOCK [52] (Fig. 2D). In agreement with the loss of their eyes, we speculate that blind, deep-sea canalipalpates do not show light-mediated biological rhythms. However, alternative biological activity mechanisms, including cues like tides and their transmitters, cannot be ruled out. In the amphipod Parhyale hawaiensis and the isopod Eurydice pulchra, the core circadian clock gene BMAL1 is essential for circatidal rhythms [53, 54]. The presence of BMAL1 in L. polybranchiata and three siboglinids suggests that these deep-sea worms may also retain a biological rhythm uncoupled from sunlight. Interestingly, in addition to eye loss, eliminating biological rhythm by the Mexican cavefish Astyanax mexicanus reduces energy consumption by a third compared to sighted surface fish [55]. Similarly, loss of visual and circadian system genes in blind canalipalpate worms could be an adaptation to a nutrient-poor environment.

High hydrostatic pressure tolerance

Hydrostatic pressure increases by ~ 100 kPa every 10 m, and L. polybranchiata is constantly exposed to ~ 17 MPa pressure. Such pressure should affect the cells of deep-sea organisms at all levels, including the cytoskeleton and cell membranes, protein stability, DNA structure, and the process of cell division [56, 57]. In addition to the unique GLS and GAG gene families described above, families associated with cytoskeleton maintenance (i.e., actin cytoskeleton, cortical cytoskeleton, and microtubule cytoskeleton) expanded in the L. polybranchiata genome (Additional file 2: Table S2). We also observed increased gene copy numbers of histone proteins (H1, H2B, H3, and H4), which may serve to sustain chromatin architecture at high pressure [58].

In deep-sea organisms—from bacteria to invertebrates and vertebrates—the composition of the cell membrane, in particular unsaturated fatty acids, appears to be critical to maintaining its structure and function at high hydrostatic pressure [59–61]. Compared to the shallow-water S. benedicti, L. polybranchiata has additional copies of fatty acid synthase (FASN) (three vs. seven copies). FASN plays a pivotal role in fatty acid biosynthesis and catalyzes a series of reactions, synthesizing long-chain fatty acids from acetyl-CoA and malonyl-CoA (Fig. 3A) [62]. The resulting products, myristate (C14:0) and palmitate (C16:0), serve as substrates for chain elongation and desaturation, ultimately producing unsaturated fatty acids [63, 64]. Intriguingly, the FASN copy number also increased in the genomes of other deep-sea inhabitants, including the Mariana Trench snailfish (Pseudoliparis swirei) and the hydrothermal sea anemone (Alvinactis idsseensis) [60, 65]. These findings suggest that extra copies of FASN represent a convergent evolutionary response in deep-sea organisms. Downstream genes, including SCD1 (two copies), FADS1 (one copy), and FADS2 (four copies), are essential for the synthesis of the omega-3 fatty acid eicosapentaenoic acid (EPA) and the balance between marine fatty acid intake and circulating levels of long-chain omega-3 fatty acids (Fig. 3A). Eicosapentaenoic acid (EPA) stabilizes deep-sea marine bacteria under high pressure [66]. Notably, all FADS genes and one SCD1 gene copy were highly expressed in the L. polybranchiata midgut (PB in Fig. 3B). The acyl-CoA oxidase (ACOX) gene family also expanded in L. polybranchiata (p = 0.030). ACOX1 and ACOX3 are involved in β-oxidation, the last step in the synthesis of the omega-3 fatty acid docosahexaenoic acid (DHA) downstream of EPA (Fig. 3A) [67].

Fig. 3.

Fatty acid analysis between deep-sea and shallow-water spionids. A Overview of unsaturated fatty acid synthesis. Expanded gene families of L. polybranchiata marked in pink. B Heatmap showing the expression of fatty acid synthesis genes in L. polybranchiata. CE denotes cephalosome (chaetigers 1–4); AB, anterior body without parapodia (chaetigers 5–20); PB, posterior body without parapodia (chaetigers 40 to end); VB, parapodia with both ventral and dorsal branchiae (chaetigers 40 to end); and DB, parapodia with only dorsal branchiae (chaetigers 5–20). C Fatty acid composition for two sea spionids. SFA, saturated fatty acids; MUFA, monosaturated fatty acids; PUFA, polysaturated fatty acids. D Polyunsaturated fatty acids with significantly different content in the two spionids in C. The data in C and D represent means ± standard deviations from six biological replicates, and asterisks indicate a significant difference (one-way ANOVA, p < 0.05) between the two species

Next, we compared the fatty acid composition of L. polybranchiata and the closely related shallow-sea species Rhynchospio aff. asiatica. The results aligned with the comparative genomic analyses. L. polybranchiata had a greater level of polyunsaturated fatty acids (PUFAs) (p = 0.004, one-way ANOVA; Fig. 3C), and EPA constituted the most abundant fatty acid in L. polybranchiata—accounting for 36.41–45.42% of the total fatty acid content (Fig. 3D). Our results support that PUFAs play a pivotal role in membrane function of deep-sea organisms and signify an evolutionary adaptation to high hydrostatic pressure in polychaetes.

Adaptions for deep sea oligotrophy

The organic input in pelagic oceans decreases by 90% at a depth of 1000 m [68]. The deep-sea heterotrophic macrofauna must adapt their nutrient absorption and utilization strategies to thrive in this environment [22, 69, 70]. L. polybranchiata is found in large numbers in the Lingshui Cold Seep, further limiting food availability. The peptide hormone cholecystokinin (CCK) stimulates the digestion of fat and protein and modulates appetite and energy balance via its receptor CCKR [71], and the expansion of the CCKR (15 gene copies) may optimize L. polybranchiata food utilization and energy storage. Several digestive enzyme gene families have undergone expansion in L. polybranchiata, including the pivotal proteolytic enzymes trypsin and chymotrypsin [72, 73] (Fig. 4A). These genes were predominantly expressed in the L. polybranchiata midgut (PB region) (Fig. 4B). Their expansion may allow efficient digestion of diverse food sources such as marine snow (biological debris from the top layers of the ocean that includes microorganisms and other nutrient-rich debris [74]) and resident chemosynthetic microorganisms [75, 76]. Bacteria and archaea are primary producers in deep-sea ecosystems and can serve as a crucial nutrient source for macrofauna [77, 78]. Compared with the shallow-water S. benedicti, cell wall-digesting lysozyme genes were expanded in L. polybranchiata (one vs. five gene copies) and deep-sea siboglinids (Fig. 4A). The number of chitinase genes—which digests hard-to-degrade macromolecules of bacteria, fungi, animals, and plants—was also increased in L. polybranchiata. In addition to maximizing nutrient uptake, the digestive enzyme gene duplications may also reflect compensation for reduced kinetic efficiency at low water temperatures (~ 2.5 °C in the case of L. polybranchiata), as proposed for Antarctic fish species [79, 80].

Fig. 4.

Expanded genes associated with digestion in L. polybranchiata. A Heatmap of the number of digestion genes in L. polybranchiata and four other sedentarians. Red stars indicate that the number of genes in L. polybranchiata significantly differs from the shallow-water S. benedicti (one-tailed Fisher’s exact tests, *p < 0.05, **p < 0.01). B Heatmap showing the expression of L. polybranchiata digestive enzyme genes. CRTB denotes chymotrypsin; PRSS, trypsin; CHIA, chitinase; LYZL, lysozyme. Body parts are designated CE (cephalosome: chaetigers 1–4), AB (anterior body without parapodia: chaetigers 5–20), PB (posterior body without parapodia: chaetigers 40 to end), VB (parapodia with both ventral and dorsal branchiae: chaetigers 40 to end), and DB (parapodia with only dorsal branchiae: chaetigers 5–20)

Genomic evidence for L. polybranchiata as a pioneer species

Pioneer species are the first to colonize a newly created or disturbed environment. Infaunal polychaetes are often the first colonizers in early stages of cold seep ecosystems. For instance, in the Hikurangi Margin cold seep, heterotrophic ampharetid polychaetes initially colonized white bacterial mats and sulfide-rich patches [81]. Similarly, in the Congo lobe complex, opportunistic, motile, and sulfide-tolerant taxa such as dorvilleid and hesionid polychaetes led the first wave of colonization. L. polybranchiata is highly abundant in its cold seep habitat and classified as the pioneer [14]. Other spionids were likely pioneers in nearshore ecosystems, including Streblospio gynobranchiata, closely related to S. benedicti in our dataset [82, 83].

These polychaetes are capable of tolerating low-oxygen and high-hydrogen sulfide environments, which enables them to become dominant species. Although we lack genomic data for the mentioned polychaetes, all polychaetes in our data set possess genes related to hydrogen sulfide tolerance and detoxification (Additional file 2: Table S3). In addition to these shared characteristics, our comparative genomic analyses revealed a dynamic gene evolution by spionids (here: L. polybranchiata and S. benedicti), including 59 expanded and ten contracted gene families (Additional file 1: Fig. S4). The expanded gene families showed enrichment for sensory systems, protein digestion and absorption, and endocrine systems (e.g., thyroid hormone signaling pathway, oxytocin signaling pathway, and insulin secretion) gene ontology terms (Additional file 2: Table S4). Some of these are highlighted below.

Rapid growth and reproduction characterize pioneer species. Reproduction of all animals is mediated by steroid and peptide hormones and their receptors (i.e., the endocrine system). Little is known about endocrine signaling pathways in invertebrates beyond model species (e.g., Drosophila), but hormone receptors and their ligands associated with growth and development are considered a universal feature of this large and diverse group [84]. In the genomes of spionids, the gene families encoding hormone receptors associated with the regulation of development and reproduction have expanded, including the growth hormone secretagogue receptor (GHSR), somatostatin receptor (SSTR), and thyrotropin-releasing hormone receptor (TRHR) (Additional file 2: Table S4). Furthermore, endocrine signaling pathways for estrogen, gonadotropin-releasing hormone, and oxytocin are under positive selection in spionids (Additional file 2: Table S5). Because all extant spionids were likely pioneer species, these genomic changes may contribute to the success of these marine worms in their diverse habitats.

The settlement and recruitment of pioneer species of marine polychaetes rely on environmental sensing [85], particularly of microbial communities [86–88]. Chemoreception should be crucial for larval settlement and recruitment (in addition to reproduction and foraging) [89, 90]. Gustatory receptor-like receptors (GRLs) and ionotropic receptors (IRs) are important metazoan chemoreceptor proteins [91, 92]. Homologs of GRL (four genes) and IR (28 genes) were found in spionids genomes. Invertebrate GRLs contain seven transmembrane domains (TMDs), like G protein-coupled receptors but with an inverted membrane topology [93]. L. polybranchiata and S. benedicti harbor two GRL homologs, while none was identified in the three siboglinid genomes examined (Fig. 5A). The GRLs were expressed by the L. polybranchiata cephalosome (which includes its chemosensory antennae) (Fig. 5B). Spionid GRL genes form a single lineage, distributed across two subgroups along with the terrestrial polychaete Capitella teleta (Fig. 5A). Specifically, members of subgroup 1 exhibit a conservative motif order of 5–4–7–2–8–3-1, while subgroup 2 follows the motif order 5–4–6–2–8–3-1 (Fig. 5A). Further supporting their annotation, spionid GRLs harbor the gustatory receptor family signature motif (“TYhhhhhQF,” where “h” represents a hydrophobic amino acid) is present as well as tyrosine (Y) residues TM5–TM7 regions (Additional file 1: Fig. S6) crucial for olfactory receptor ion channel function [94]. The retainment of GRL genes could reflect a central chemoreceptive role by the pioneer species family Spionidae.

Fig. 5.

Chemoreceptors in marine sedentarians (spionids and siboglinids). A Phylogenetic tree of gustatory receptor-like receptors. Motifs were classified using MEME Suite. B Heatmap showing the expression of L. polybranchiata chemoreceptor genes. The prefix GRL denotes gustatory receptor-like receptors; IR, ionotropic receptors. C 3D structure-based tree of candidate IRs from spionids generated using AlphaFold2. Blue molecules indicate the ligand-binding domain of every type, with ligand-binding residues (R, T, D/E) in red. The accession numbers of the reference sequences are in Table S2

Most invertebrates rely more on IRs than GRLs to sense chemicals [95]. The IR gene family exhibited significant expansion in spionids compared with siboglinids (p = 0.022). The IR-like sequences grouped into two clades. Clade I proteins have three domains: an amino-terminal domain (ATD), a ligand-binding domain (LBD), and a transmembrane domain (TMD). Clade II proteins lack an ATD. A phylogenetic tree branched the spionid IRs into three main groups (Additional file 1: Fig. S7). To gain deeper insight into the structure and function of these IR candidates, we employed AlphaFold2 [96] to predict their tertiary structures and generate a structure tree (Fig. 5C). The proteins were classified into six types: types I and IV are specific to the spionids (L. polybranchiata + S. benedicti), and type II is abundant in S. benedicti. The ligand-binding pockets are primarily formed by β-sheets and α-helices, creating a cavity for signal molecule binding. The configurations of these pockets differ in pocket size, shape, and ligand-binding residues (i.e., different ligands) (Fig. 5C). Notably, type VI is a conserved IR25a subclade in the five canalipalpates in our dataset (Fig. 5C). The predicted 3D structures of lipolIR25a align well with canonical models, featuring three conserved domains (Fig. 6A–C). The structure of the LBD in lipolIR25a was highly similar to that of Drosophila melanogaster (root mean square deviation, RMSD, of 0.899 Å) (Fig. 6D). In D. melanogaster, IR25a mediates multiple sensory modalities, including chemical sensing, thermal sensing, and humidity sensing [97, 98]. These similarities suggest that IR25a is essential for the sensory behavior of L. polybranchiata and other marine worms. L. polybranchiata IR genes show distinct expression (Fig. 5B), suggesting they mediate different signals and functions. We propose that the diverse IRs in spionids facilitate the detection of a broad range of chemical signals to trigger responses from settlement to metamorphosis, a prerequisite for spionids to be pioneer species in either deep-sea or shallow ecosystems. The exclusive distribution of L. polybranchiata in chemosynthetic ecosystems hints at a preference for specific microbial communities or bacterial metabolites within sediments. These chemoreceptors may compensate for degraded vision and aid L. polybranchiata in perceiving external chemical stimuli essential for early-stage colonization and survival at the Lingshui Cold Seep.

Fig. 6.

The chemosensory ionotropic receptor IR25a is conserved in marine sedentarians (spionids and siboglinids). A Alignment of the S1 and S2 ligand-binding domains from seven candidate IR25a proteins. Three key ligand-binding residues (R, T, and D/E) are marked with an asterisk; S1 and S2 domains by horizontal orange lines. B Schematic representation of IR25a protein is shown. C AlphaFold2 prediction of L. polybranchiata IR25a. The structure shows distinct domains, including the amino-terminal domain (ATD), the ligand-binding domain (LBD), and the transmembrane domain (TMD). D Alignment of LBD structure of IR25a of L. polybranchiata (red) and Drosophila melanogaster (turquoise). Accession number: phauc_IR25a (XP_059149035), drmel_IR25a (NP_001260049)

Conclusions

The Lindaspio genus represents a fascinating group of spionids endemic to chemosynthetic ecosystems such as hydrothermal vents, methane seeps, and whale falls. Here, we assembled and annotated the genome of Lindaspio polybranchiata from the Lingshui Cold Seep, marking the first genome of a deep-sea Spionidae. Our study also offers a valuable genomic resource and new insights into the ecology and evolution of L. polybranchiata and other deep-sea spionids. However, it is important to acknowledge that the genome assembly quality of L. polybranchiata is not optimal. These limitations can be attributed to several challenges commonly faced in assembling genomes of deep-sea organisms. The extreme conditions of the deep sea, such as high hydrostatic pressure and low temperature, often result in DNA degradation during sampling and transport [99, 100]. Moreover, the inherent genomic complexity of L. polybranchiata, including higher heterozygosity (2.12%), elevated repetitive sequence proportion (59.15%), and expanded gene families, adds to the difficulty of achieving a high-quality assembly [101]. Future efforts are needed to further improve the quality of the genome.

Based on our comparative genomic analyses, we propose a conceptual model illustrating how L. polybranchiata evolved as a pioneer species in the deep sea (Fig. 7). Compared to their shallow-water counterparts, L. polybranchiata shows evidence of adaptation tailored to a deep-sea environment. This includes a reduction in genes related to vision and biological rhythms likely driven by life in perfect darkness but possibly also reducing energy requirements, similar to the Mexican cavefish [55, 102]. In this context, such gene loss may be considered an adaptation to nutrient-poor environments (oligotrophy). Future studies should consider the adaptive benefits of losing ostensibly disused genes in deep-sea animals. L. polybranchiata also shows evidence of an expanded digestive capability, possibly to utilize a broader range of substrates as food, advantageous for survival in a nutrient-limited environment. There is also a modification in membrane lipid composition, specifically an increase in PUFAs, fatty acids that enhance membrane fluidity under high hydrostatic pressure.

Fig. 7.

A conceptual model illustrating how L. polybranchiata evolved as a pioneer species in the deep sea. The red crosses indicate gene loss, while red highlights denote expanded gene families. A The sun with arrows illustrates light activating circadian rhythm genes and triggering light transduction via opsins, while darkness leads to vision degeneration. The pie with an hourglass symbolizes circadian rhythm and the core circadian clock genes were lost. B Diverse digestive enzymes break down macromolecules into absorbable nutrients, supporting adaptation to oligotrophic conditions. C The synthesis pathways of unsaturated fatty acids, GSLs, and GAGs, involving unique and expanded gene families, enhance membrane fluidity in response to high hydrostatic pressure. D The worm model represents the larval stage, where expanded IRs and GRLs facilitate the perception of chemical cues for colonization in the sediment. E The gene families encoding hormone receptors, regulating development and reproduction have expanded

L. polybranchiata is a pioneer species in its nascent cold seep, but most spionids are considered pioneer species. Their pioneer status is strengthened by shared genomic features, such as an expanded chemoreceptor repertoire that likely aids in detecting various chemical cues, facilitating the colonization of new habitats. Additionally, an increase in spionid hormone receptors may regulate development and reproduction, supporting population propagation. Our findings provide a molecular foundation for understanding how spionids perceive chemical signals in their environment, settle in suitable habitats, and establish abundant populations.

Methods

Sampling and sequencing

Lindaspio polybranchiata were collected using a remotely operated vehicle (ROV) in the Lingshui Cold Seep at 1700 m depth (Additional file 1: Fig. S8). The specimen (n = 1) was rinsed with sterile water before being flash-frozen in liquid nitrogen and then stored at − 80 °C for subsequent analysis. For whole-genome resequencing, genomic DNA was extracted by phenol–chloroform extraction, which was used to construct DNA fragment libraries with a TruSeq DNA PCR-Free library Prep kit and then sequenced on the Illumina NovaSeq 6000 platforms. For PacBio HiFi sequencing, DNA was extracted with DNeasy Blood & Tissue Kit (QIAGEN). After obtaining high-quality genomic DNA, a PCR-free SMRT bell library was constructed, and sequencing was performed using the PacBio Sequel II sequencing platform by Annoroad Gene Technology, Beijing, to generate 97.4 Gb raw bases and 869.7 Gb subread bases.

Genome assembly

We used FastQC v0.11.5 [103] to assess the quality of raw Illumina data and Trimmomatic v0.36 [104] to filter the raw Illumina data and obtain clean data with the parameter as “LEADING:5 TRAILING:5 SLIDINGWINDOW:5:20 MINLEN:50 TOPHRED33.” K-mer frequency-based method was used to determine genome size, heterozygosity ratio, and repeat content. In detail, the K-mers were counted using Jellyfish v2.2.3 [105] with the parameters “-m 17,” “-m 19,” and “-m 21.” The resulting k-mer frequency histogram was then used to estimate genome size in GenomeScope [106] with the parameters “-k 17 -p 2,” “-k 19 -p 2,” and “-k 21 -p 2.” The results from all three evaluations were similar (Additional file 1: Fig. S9). Next, CCS v. 4.2.0 (https://github.com/PacificBiosciences/ccs) was used to filter subreads data, retaining 46.92 gigabase pairs (Gbp) PacBio HiFi reads. We assembled the L. polybranchiata using Hifiasm v0.19.4 with HiFi reads [107] and used Purge Haplotigs v1.1.2 [108] to remove redundant sequences and generate a draft genome. Benchmarking Universal Single-Copy Orthologs (BUSCO) v5.2.2 with the metazoan_odb10 dataset was used to assess the completeness of the genome draft [109].

Genome prediction and functional annotation

Before gene model prediction, repetitive sequences were identified by RepeatModeler v2.0.1 and RepeatMasker v4.1.2 [110] based on the Repbase v21.01 and Dfam v3.5 database [111, 112]. The repetitive regions were soft-masked for later gene model prediction. For ab initio gene prediction, the clean transcriptome (RNA-seq) reads were mapped to the genomes with HISAT2 v2.2.1 [113], and the mapping results and homologous protein sequences were used to train Augustus v3.4.0 in Braker2 v3.0.3 [114, 115]. Moreover, the EST evidence assembled with Trinity v2.8.5 [116] with default parameters and protein homology evidence were used for genome model prediction with MAKER v3.1.4 [117]. For transcript-based prediction, we used Stringtie v2.1.4 [118] to obtain the gene annotation files based on RNA-seq alignments generated by HISAT2 v2.2.1 and then used Transdecoder v5.5.0 to predict gene structure information [113]. The final version of protein-coding genes was generated by integrating all evidence in EVidenceModeler v2.1.0 [119]. Gene functional annotation was performed by searching the following databases: National Center for Biotechnology Information (NCBI) NR v202304, Swiss-Prot v202304, eggNOG v5.0, Kyoto Encyclopedia of Genes and Genomes (KEGG) v87.0, and HMMER v3.3.1 (Additional file 1: Fig. S1).

Molecular phylogeny and gene family analysis

To determine the phylogenetic position of L. polybranchiata, we downloaded the genomes of 11 annelids from NCBI (Additional file 2: Table S6). We used OrthoFinder v2.5.4 to identify orthologs, using BLASTP program with a threshold value of e-value ≤ 1e − 9 and query-coverage ≥ 0.5 [120]. According to the results of gene family clustering, 528 single-copy orthologous genes were used for phylogenetic analyses. For each ortholog group, the amino acid sequences were aligned with “linsi” of MAFFT v7.520 [121]. Then, we used Gblocks 0.91b to extract conserved blocks with default settings [122]. These conserved blocks were concatenated to construct a phylogenetic tree with the maximum likelihood method under the PROTGAMMAGTRX model using RAxML v8.2.12 applying 1000 bootstrap replicates [123]. Furthermore, the divergence time was calculated with the MCMCtree program in PAML, with settings nburn-in = 10,000,000, nsamfreq = 100, and nsample = 500,000 [124]. Four fossil correction points were used: Oweniidae appeared about 514 million years ago (Ma); Eisenia andrei and Helobdella robusta diverged at 295.9–345.0 Ma; Sedentaria and Errantia diverged 207.5–513.5 Ma; Siboglinidae appeared about 50–126 Ma (http://www.timetree.org/). Fossil calibrations are marked in Fig. 1A.

CAFE (Computational Analysis of gene Family Evolution) v4.2.1 was used for gene family evolution analysis based on the OrtherFinder results, and only orthologous groups with a p value < 0.05 were considered significantly expanded or contracted compared to the ancestor in each node [125]. Expanded and contracted gene families in L. polybranchiata compared to the shallow-water Streblospio benedicti were tested for gene ontology (GO) and KEGG enrichment with OmicShare tools (https://www.omicshare.com/tools).

Positive selection analysis

Positive selection analysis was conducted using the codeml package in PAML v4.9 [124]. The phylogenomic results were used as the species tree to guide the analysis. To identify candidate positively selected genes (PSGs), we compared the likelihood of selected genes using the YN00 model and the branch-site model, with L. polybranchiata designated as the foreground and other annelids as the background. To pinpoint potential positive selection on the foreground branch, a likelihood ratio test (LRT) was performed to compare two proposed positive selection sites. p values were calculated using chi-square (χ²) statistics (df = 1; FDR corrected) and genes with a p value < 0.05 and a Bayesian probability > 90% were classified as PSGs. GO and KEGG enrichment analyses were performed following the same protocols as those used for the expansion of gene families.

Chemoreceptor and photoreceptor gene annotation

The chemoreceptor sequences of the superphylum Lophotrochozoa (including Annelida) were downloaded from NCBI and UniProtKB databases. All sequences were aligned to the genomes using tblastn v2.14.1 [126] with an e-value of 1e − 8. We extracted the candidate gene regions and their 20-kb flanking sequences and used GeneWise v2.4.1 [127] to predict the complete gene structure. HMMSCAN (HMMER v3.3.2) was applied to identify Pfam domains in protein-coding genes. Finally, all candidate protein sequences were aligned with MAFFT v7.505 [121], and the maximum likelihood tree was constructed with FastTree v2.1.11 with default setting [128]. The information of the reference sequences used for tree construction is provided in Additional file 2: Table S7. The motifs of these candidate proteins were identified using meme in MEME Suite v5.5.7 [129]. The gene family expansion or contraction analyses in the chemoreceptor gene family were conducted using Fisher’s exact test (one-tailed) in R 3.6.2 with the stats package [130]. To better understand the structure and function of these candidate proteins, we used AlphaFold2 v2.0.1 to predict their 3D structures and constructed a structural tree [96].

The photoreceptor gene family was searched following the same pipeline. Moreover, we search the lost photoreceptor genes in de novo-assembled transcriptome sequences and PacBio HiFi reads using tblastn v2.14.1 [126] at the thresholds of ≥ 30% amino acid identity, ≤ 1e − 5 e-value, and ≥ 60% query cover per subject. Candidate transcript ORFs were predicted and translated into amino acid sequences using TBtools [131]. For candidate HiFi reads, GeneWise v2.4.1 [122] was used to predict complete gene structures. Then, all candidate proteins were analyzed for Pfam domains using HMMSCAN (HMMER v3.3.2) and aligned with MAFFT v7.505 [116] to examine key residues.

Transcriptome sequencing and analysis

The body of Spionidae is elongated, with no distinct segmentation, composed of a trunk and parapodial appendages. L. polybranchiata is characterized by a distinct caruncle (a sensory organ extending to chaetiger 2), modified notopodial spines (chaetigers 2–4), and more neuropodial branchiae in the parapodia (from chaetiger 20) (Additional file 1: Fig. S10) [15]. To facilitate genome annotation and provide a transcriptional profile along the body axis of L. polybranchiata, worms (n = 6) were sectioned into five parts [15]: CE (cephalosome, chaetigers 1–4), AB (anterior body without parapodia, chaetigers 5–20), PB (posterior body without parapodia, chaetigers 40 to end), VB (parapodia with both ventral and dorsal branchiae, chaetigers 40 to end), and DB (parapodia with only dorsal branchiae, chaetigers 5–20). According to the standard manufacturer’s protocol, total RNA was extracted using TRIzol reagent (Invitrogen). RNA quality was determined by 1% agarose gel electrophoresis and Agilent 5400 (Agilent, USA), while RNA concentration was measured with a Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, USA). Transcriptome libraries were generated using NEBNext Ultra RNA Library Prep Kit (NEB, USA) and then sequenced on the Illumina platform by Novogene Bioinformatics Technology Company (Beijing, China) to produce 150 bp reads, yielding approximately 10 Gb of raw data per sample. The RNA-seq data was filtered with fastp v0.23.4 with the following parameters: “–detect_adapter_for_pe –cut_mean_quality 20 –length_required 50 −5 −3 -W 4 -e 20” [132] and aligned to the reference genome with HISAT2 v2.2.1 with default parameters [113]. Then, transcripts per kilobase million (TPM) were calculated using Cufflinks v2.1.1 [133]. The heatmap of expression levels across different tissues was generated with the pheatmap package in R 3.6.2 [134].

Fatty acid analysis

To compare the difference in fatty acid (FA) composition between L. polybranchiata and shallow-water spionids, we collected a closely related species, Rhynchospio aff. asiatica, in the intertidal region of Qingdao Bay. The specimens of L. polybranchiata (n = 6) and R. aff. asiatica (n = 6) were dried in a freeze dryer for 24 h. We weighed 2–5 mg of dried tissue for FAs analysis. FAs were extracted at 20 °C for 12 h in 3 ml extraction solution with mixed solution of dichloromethane, chloroform, and methanol in equal volumes. At the beginning of the extraction, 100 μl of C19:0 and C21:0 was added to each sample as an internal standard. We washed the extract with a 2.25 ml 1 M solution of potassium chloride, collected the lower liquid phase, and blew out the residual water using a nitrogen blower. Next, 100 μl toluene and 200 µl methanol (supplemented with 1% concentrated sulfuric acid) were added to every sample. The esterification reaction was continued at 50 °C for 12 h. The fatty acid methyl esters (FAMEs) were then washed with 600 µl 5% NaCl solution and 200 µl hexane. The upper layer of clarified liquid was collected into a new glass centrifuge tube and 100 μl of hexane was added to redissolve the FAMEs. Finally, the FAMEs were analyzed by gas chromatography on a mass spectroscopy ISQ 7000 instrument (Thermo Fischer Scientific, Waltham, USA) using hydrogen as carrier gas.

Abbreviations

CRY

Cryptochrome

PER

Period

CLK

CLOCK

TIM

Timeless

VRI

Vrille

PDP1

Par domain protein 1

OTX

Orthodenticle-related homeobox

MITF

Microphthalmia-associated transcription factor

CRTB

Chymotrypsin

PRSS

Trypsin

CHIA

Chitinase

LYZL

Lysozyme

DPP4

Dipeptidyl-peptidase 4

EGase

β-1,4-Endoglucanase

MANB

Mannosidase

BGL

β-Glucosidase

CELA2

Pancreatic elastase II

GAG

Glycosaminoglycan

GSL

Glycosphingolipid

CHST11

Chondroitin 4-sulfotransferase 1

CHST15

N-acetylgalactosamine 4-sulfate 6-O-sulfotransferase

FASN

Fatty acid synthase

ACOX

Acyl-CoA oxidase

FUT1_2

Galactoside 2-L-fucosyltransferase

PRDM9

PR/SET domain containing protein 9

IR

Ionotropic receptors

GRL

Gustatory receptor-like receptors

SSTR

Somatostatin receptor

TRHR

Thyrotropin-releasing hormone receptor

GHSR

Growth hormone secretagogue receptor

Authors' contributions

Y. Y., M. W., and C. L. conceived the idea. Y. Y., H. W., M. L., Z. Z., and H. Z. collected the sample. Y. Y. prepared DNA sequencing and performed genomic and transcriptomic analyses. Y. G. performed genome assembly. D. W. performed genome structure analyses. Y. Y wrote the manuscript and additional supplementary files. C. L., M. W., I. S., and Y. G. revised the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (42030407 and 42076091), the NSFC Innovative Group Grant (42221005), the National Key R&D Program of China (2022YFC2804003), the National Key R&D Program of China (2023YFC2811501), Marine S&T Fund of Shandong Province for Pilot National Laboratory for Marine Science and Technology (Qingdao) (2022QNLM030004), and State Key Laboratory of Microbial Technology Open Projects Fund (Project NO. M2023-10).

Data availability

The assemblies and protein sequences of published genome used in this study were from the NCBI repository: Streblospio benedicti (GCA_019095985.1) [21], Paraescarpia echinospica (GCA_020002185.1) [28], Lamellibrachia luymesi (GCA_009193005.1) [29], Lamellibrachia satsuma (GCA_022478865.1) [33], Branchipolynoe longqiensis (GCA_030323885.1) [34], Harmothoe impar (GCA_947462335.1) [35], Capitella teleta (GCA_000328365.1) [31], Helobdella robusta (GCA_000326865.1) [31], Owenia fusiformis (GCA_903813345.2) [27], Dimorphilus gyrociliatus (GCA_904063045.1) [27], and from the National Genomics Data Center: Eisenia andrei (GWHACBE00000000) [32]. Genome assembly and annotation of the Lindaspio polybranchiata are available at Figshare [135]. All sequencing data for the genome and transcriptome are also available at the CNGB Sequence Archive (CNSA) of the China National GeneBank DataBase (CNGBdb) under project CNP0006250 [136] and National Center for Biotechnology Information (NCBI) under project PRJNA1188207 [137]. The computer commands were shared on GitHub [138].

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.