Whole Genome Duplication and Gene Evolution in the Hyperdiverse Venomous Gastropods

Abstract

The diversity of venomous organisms and the toxins they produce have been increasingly investigated, but taxonomic bias remains important. Neogastropods, a group of marine predators representing almost 22% of the known gastropod diversity, evolved a wide range of feeding strategies, including the production of toxins to subdue their preys. However, whether the diversity of these compounds is at the origin of the hyperdiversification of the group and how genome evolution may correlate with both the compounds and species diversities remain understudied. Among the available gastropods genomes, only eight, with uneven quality assemblies, belong to neogastropods. Here, we generated chromosome-level assemblies of two species belonging to the Tonnoidea and Muricoidea superfamilies (Monoplex corrugatus and Stramonita haemastoma). The two obtained high-quality genomes had 3 and 2.2 Gb, respectively, and 92–89% of the total assembly conformed 35 pseudochromosomes in each species. Through the analysis of syntenic blocks, Hox gene cluster duplication, and synonymous substitutions distribution pattern, we inferred the occurrence of a whole genome duplication event in both genomes. As these species are known to release venom, toxins were annotated in both genomes, but few of them were found in homologous chromosomes. A comparison of the expression of ohnolog genes (using transcriptomes from osphradium and salivary glands in S. haemastoma), where both copies were differentially expressed, showed that most of them had similar expression profiles. The high quality of these genomes makes them valuable reference in their respective taxa, facilitating the identification of genome-level processes at the origin of their evolutionary success.

Introduction

Venomous organisms display a high level of diversification, both at the taxonomic and the functional levels (Casewell et al. 2013; Fry et al. 2015; Zancolli and Casewell 2020). The production of venom has been reported across at least 101 independent lineages in animals (Schendel et al. 2019), among which some well-known examples are snakes, spiders, and scorpions, with multiple instances of convergent evolution (King et al. 2008; Barua and Mikheyev 2019; Forde et al. 2022). Venom is a mix of toxins (mostly proteins and peptides) and other molecules, produced and stored in specialized anatomical structures of an organism and injected in another organism to disrupt its normal physiological processes (Fry et al. 2009). Venomous animals thus developed complex systems adapted to their target organisms, comprising not only toxic molecules but also the self-resistance mechanism needed to deal with their production, storage, and delivery, as well as specialized morphological structures including secretory glands and delivery weapons (Schendel et al. 2019). Venoms can be used for multiple purposes, including predation, defense, or intraspecific competition (Casewell et al. 2013). The specific functions of the toxins that compose the venoms are extremely diversified, and in most cases, toxins act synergistically to incapacitate the target organism by inducing paralysis through targeting neuromuscular receptors, causing blood loss by injecting anticoagulant, or causing intense pain by triggering pain receptors, among other mechanisms (Bohlen et al. 2011; Zancolli and Casewell 2020). Given the adaptive value of venom, understanding the origin and evolution of such diversified molecules and the organisms that produce them have become crucial in evolutionary biology.

Neogastropoda is a group of hyperdiverse predatory marine mollusks. More than 15,000 species were described so far, representing 22% of the gastropod diversity (Lemarcis et al. 2022). Beside their high phylogenetic and taxonomic diversity, they are also characterized by a remarkable disparity in feeding strategies which include—but are not restricted to—envenomation, asphyxia, and vampirism (Taylor et al. 1980; Modica and Holford 2010; Olivera et al. 2017). Indeed, several neogastropod species were found to produce molecular compounds to subdue their preys (Modica et al. 2018; Turner et al. 2018), with Conus species undeniably representing the most studied lineage of neogastropods, and one of the gold standards in venom research. Cone snails represent around 1,000 species (WoRMS Editorial Board 2022), are widely distributed, and have different hunting targets (Zhao and Antunes 2022). Thus, more and more toxins (“conotoxins”) were found (∼200 toxins produced per species) in Conus species (Balaji et al. 2000; Buczek et al. 2005; Akondi et al. 2014), some of which entered clinical trials that, in one case, led to the market: the synthetic analog of a conotoxin, Ziconotide, is currently commercialized as an anesthetic (Layer and McIntosh 2006; Fedosov et al. 2012; Safavi-Hemami et al. 2019; Reynaud et al. 2020; Zhao and Antunes 2022).

Conotoxins are small peptides (usually shorter than 40aa) and possess a conserved gene structure composed of an N-terminal signal sequence, a pro-region, and a mature peptide (Kaas et al. 2010; Puillandre et al. 2012; Fedosov et al. 2021). The majority of them are rich in cysteine residues, arranged in specific patterns (‘cysteine frameworks’), resulting in disulfide bounds that determine the precise 3D conformation and high stability of the toxins themselves (Jin et al. 2019). Conotoxins were shown to have diverged, at least partly, through the onset of gene variants, especially due to insertions and deletions (Yang and Zhou 2020). Toxins closely related to conotoxins or sharing the same structural conformation were also found in other Conoidea (e.g., in Turridae, with longer mature peptide regions [Watkins et al. 2006], in Terebridae [Gorson et al. 2015], and in Drilliidae [Lu et al. 2020]). Beside conotoxins, a large variety of toxins were described in the Neogastropoda, acting as pore forming (e.g., colutoxins [Gerdol et al. 2018], hemolytic [e.g., echotoxins; Shiomi et al. 2002]) or anesthetic toxins like ShKT (Gerdol et al. 2019), to cite only few of their putative functions. Toxins were found in species belonging to multiple neogastropod families, including Babyloniidae, Muricidae, Buccinidae, Colubrariidae (Turner et al. 2018), Mitridae (Reynaud et al. 2020), and Costellaridae (Kuznetsova et al. 2022) but were reported in some of the closely related Caenogastropoda as well, namely the Tonnoidea families, Cassidae, Ranellidae, and Charoniidae (Hwang et al. 2021; Jia et al. 2022). However, the diversity of these toxins, the genomic mechanisms that may have triggered their evolution, and their potential adaptive and evolutionary implications remain understudied in non-Conoidea neogastropods. In particular, for a comprehensive evaluation of the molecular mechanisms of venom toxins evolution, a genomic perspective is indeed required. However, only a few comparative genomic studies are available to date on venomous organisms (Drukewitz and von Reumont 2019; Zancolli and Casewell 2020; Almeida et al. 2021; Barua and Mikheyev 2021; Koludarov et al. 2022), suggesting that duplication and neofunctionalization events constitute crucial steps in the onset of toxins from ancestral genes (Wong and Belov 2012; Giorgianni et al. 2020). If gene duplication can lead to evolutionary novelties even when restricted to a single gene (Magadum et al. 2013), large-scale duplication events indeed can dramatically increase the adaptative potential of species (Ren et al. 2018).

Polyploidy or whole genome duplication (WGD) is a large-scale duplication event that generates additional copies of an entire genome within a species. This phenomenon is widespread in plant genomes (del Pozo and Ramirez-Parra 2015) but has also been recognized through multiple other eukaryotic lineages (Sankoff 2001). WGD is commonly followed by massive gene loss or duplication and other genomic rearrangements (Sémon and Wolfe 2007), which would imply the emergence of evolutionary novelties in such species. A hypothesis was suggested in a prior study that WGD may have taken place in Neogastropoda and some closely related superfamilies of Caenogastropoda. This was determined through a predictive model based on karyological data, which employed a likelihood method that modeled the background rate of chromosome number evolution as a birth-death process (Hallinan and Lindberg 2011). A recent publication confirmed WGD in Conus ventricosus genome, where 35 pseudochromosomes with a total genome size of 3.5 Gb were described, evidenced by the presence of syntenic blocks with Pomacea canaliculata, an Ampullarioidea species, where one pseudochromosome of P. canaliculata represented two to four pseudochromosomes of C. ventricosus, a duplicated cluster of Hox genes, and a Ks (synonymous substitutions) distribution peak (Pardos-Blas et al. 2021).

Given the high potential of WGD to foster diversification (Moriyama and Koshiba-Takeuchi 2018), it is crucial to understand whether this particular event is restricted to Conus or if it occurred in other lineages across the Neogastropoda, possibly related to the hyperdiversity of the entire group. To answer this question, high-quality genomes, assembled at the chromosomal level, are required.

As sequencing cost is decreasing and sequencing quality is improving, it became easier to obtain chromosome-level assemblies in nonmodel organisms with the use of long-read technologies (PacBio or Nanopore) combined with Hi-C or Omni-C links (Ghurye et al. 2019). Thus, gastropod genomes are increasingly being sequenced, summing up to ∼45 genomes, eight of which being of neogastropods (Stockwell et al. 2010; Hu et al. 2011; Barghi et al. 2016; Pardos-Blas et al. 2021; Peng et al. 2021); however, their taxonomic coverage is mostly restricted to Conoidea species, and their quality is uneven. The only available chromosome-level genome to date is from C. ventricosus (Pardos-Blas et al. 2021), even though the BUSCOs values for the annotation were low. Outside Neogastropoda, four Ampullarioidea species have good-quality genome assemblies (Sun et al. 2019), among which P. canaliculata is the only one with chromosome-level assembly, which makes it a good candidate for comparative analysis (Liu et al. 2018).

Here, we generated chromosome-level assemblies of two additional gastropod species: Stramonita haemastoma (Linnaeus, 1767) from the neogastropod superfamily Muricoidea and Monoplex corrugatus (Lamarck, 1816), belonging to the Tonnoidea superfamily, closely related to the Neogastropoda (Lemarcis et al. 2022) and also including species able to produce venom (Bose et al. 2017). Both species are known to be venomous marine predators and are good candidates to unveil the molecular diversity of venomous gastropods, as they are taxonomically distant both from each other and from Conus (Lemarcis et al. 2022). We first investigated WGD in both genomes by analyzing synteny, hox gene clusters, and Ks distribution to confirm the taxonomic span of such event in Neogastropoda but also in Tonnoidea. Then, we looked for potential toxin-related genes in both genomes using toxin databases as Conoserver (Kaas et al. 2012). We also performed gene expression analysis of S. haemastoma genes in order to obtain insights about their evolutionary patterns.

Results

Chromosome-Level Genomes

Respectively, 359 (120 ×) and 195 Gb (89 ×) long reads (PacBio) were generated to assemble a 3 and 2.2 Gb draft genomes of M. corrugatus and S. haemastoma, falling within the size range of other neogastropod genomes (table 1) and significantly higher than other Caenogastropoda, Patellogastropoda, Neomphalina, and Vetigastropoda genomes available to date (supplementary table S1, Supplementary Material online). We used the Dovetail Omni-C method to achieve chromosome-level assemblies for both species, resulting in 35 pseudochromosomes each. No previous report was available in the literature on the karyotype of Tonnoidea species, including M. corrugatus, whereas only 36 karyotypes have been described for neogastropod species, none of them being S. haemastoma (supplementary table S2, Supplementary Material online). Among the species belonging to the Muricidae family (which includes S. haemastoma), most have 35 chromosomes (12 species over a total of 21, see supplementary table S2, Supplementary Material online). Actually, most of the neogastropods possess between 30 and 43 chromosomes, with the exception of Strigatella litterata, Nucella lapillus, and Myurella nebulosa (n = 19, 14–16, and 17, respectively). The single chromosome-level neogastropod genome (of the 7 available; see supplementary table S1, Supplementary Material online), from C. ventricosus (Pardos-Blas et al. 2021), also presented 35 assembled pseudochromosomes, congruently with our results (supplementary table S2, Supplementary Material online).

Table 1.

Assembly and Annotation Statistics of Monoplex corrugatus, Stramonita haemastoma (from This Study) Compared with Conus ventricosus, Pomacea canaliculata, and Achatina fulica.

Species	Assembly Length (Gb)	N50 Mb (L50)	BUSCO Score on Assembly	GC Content (%)	Number of Protein-Coding Genes	Exons/cds	Introns/CDS	CDS Length	Exon Length	Intron Length	BUSCO Score on Annotation
Monoplex corrugatus	3.075	86.8 (15)	C: 91.4% (S: 84.6%, D: 6.8%)	41	36,924	4.1	3.1	1,128	281	1,621	C: 83.6% (S: 83.3%, D: 0.3%)
Stramonita haemastoma	2.236	58.4 (16)	C: 89.1% (S: 83.8%, D: 5.3%)	42	34,865	4.8	3.8	1,141	243	1,229	C: 78.7% (S: 77.1%, D: 1.6%)
Conus ventricosus	3.592	93.5 (16)	C: 84.7% (S: 81.8%, D: 2.9%)	43	31,591	4.6	3.6	1,126	246	1,603	C: 32.2% (S: 31.6%, D: 0.6%)
Pomacea canaliculata	0.447	32.6 (6)	C: 97.3% (S: 96.5%, D: 0.8%)	39	18,265	9.3	8.3	1,671	263	1,542	C: 89.1% (S: 88.4%, D: 0.7%)
Achatina fulica	1.856	59.6 (13)	C: 93.4% (S: 87.7%, D: 5.7%)	39	23,726	7.1	6.1	1,559	271	3,793	C: 89.8% (S: 82.9%, D: 6.9%)

Note.—N50 values are in Mega base, BUSCO score column reports complete BUSCO sequences (C) number, that is the sum of single BUSCO sequences (S) and duplicated BUSCO sequences (D).

We retrieved an N50 value of 87 Mb for M. corrugatus and 58 Mb for S. haemastoma (table 1) indicating a high contiguity of the assemblies when compared for example with C. tribblei with a N50 value of 2.7 Kb. In total, 91% and 89% metazoan BUSCO genes, respectively, were found in these assemblies, indicating a valuable completeness of the genomes, even higher than that obtained for C. ventricosus (84.7%). When looking at single genes only (supplementary table S1, Supplementary Material online), found in the assemblies using BUSCO, the values are similar to C. ventricosus (82%, 84%, and 85% in C. ventricosus, S. haemastoma, and M. corrugatus, respectively), whereas duplicated genes number doubles in M. corrugatus (7%) and S. haemastoma (5%) compared with C. ventricosus (3%). This might indicate a higher purge of genes after the WGD in C. ventricosus than in M. corrugatus and S. haemastoma. After masking repeated elements, representing 61% and 40% of each genome, respectively, we predicted 36,924 and 34,865 protein-coding genes (table 1 and supplementary table S1, Supplementary Material online for more details). The global statistics of the annotations were similar to C. ventricosus, including the number of predicted genes which is, however, slightly higher than what has been reported in the other caenogastropods (e.g., in C. ventricosus or Batillaria attramentaria) and significantly higher than in P. canaliculata (2 and 1.8 times higher for M. corrugatus and S. haemastoma, respectively). The average number of exons per coding sequence (CDS), number of introns per CDS, and CDS length were lower in M. corrugatus and S. haemastoma compared with P. canaliculata and other Ampullarioidea (table 1). We also observed a higher proportion of CDS in our genomes, with CDS BUSCO scores of 84% and 79%, respectively, compared with the reported score of 32% for C. ventricosus, which could be attributed in part to sequencing errors for the C. ventricosus genome, as the sequencing depth was of 54X only (table 1; Pardos-Blas et al. 2021).

Duplication Events in Neogastropoda and Tonnoidea

We detected paralogous genes in M. corrugatus, S. haemastoma, C. ventricosus, and P. canaliculata using the best reciprocal hit (BRH) method, as well as orthologs shared between species pairs (fig. 1). We plotted the distribution of the identity percent for each comparison of the orthologs between the four species and found that orthologs between M. corrugatus and S. haemastoma had higher median and average identity percent than orthologs between C. ventricosus and the two other species M. corrugatus and S. haemastoma (supplementary fig. S1, Supplementary Material online).

Fig. 1.

Schematic representation of the phylogenetic relationships between the species compared in this study. Simplified tree of the four Caenogastropoda species studied in this work. Numbers for each pair correspond to the number of orthologs detected between the two species (top) and the number of syntenic regions detected between the two genomes, all chromosomes considered (bottom). The three first pictures credits: MNHN CC BY 4.0. The picture of P. canaliculata was taken from Ng et al. 2017.

To further detect duplication events, we searched for conserved synteny among the chromosome-level assemblies and compared them with P. canaliculata (fig. 2). MCScanX detected collinear orthologs (called segmental) in each pairwise comparison (see numbers on fig. 1). Among these species, almost all pseudochromosomes of each species had homologous regions detected in the two other species; for example, all pseudochromosomes numbered 1 (largest pseudochromosomes of M. corrugatus, S. haemastoma and C. ventricosus) matched each other's, with some translocation events in pseudochromosomes 2 or 3 (fig. 2). More interestingly, each of P. canaliculata pseudochromosomes displayed synteny with more than one pseudochromosome in C. ventricosus, S. haemastoma, and M. corrugatus but the chromosome 14. For example, syntenic blocks of P. canaliculata pseudochromosome 7 were retrieved in pseudochromosomes 1 and 2 in M. corrugatus, 1 and 3 in S. haemastoma, and 1 and 2 in C. ventricosus; pseudochromosome 3 had syntenic blocks occurring in pseudochromosomes 6, 10, 19, and 24 in M. corrugatus, 6, 11, 21, and 22 in S. haemastoma and 7, 8, 20, and 22 in C. ventricosus. In summary, ten pseudochromosomes of P. canaliculata had homologous regions in exactly two pseudochromosomes of C. ventricosus, S. haemastoma, and M. corrugatus, and three pseudochromosomes had homologous regions in exactly four pseudochromosomes of C. ventricosus, S. haemastoma, and M. corrugatus. Genes found in chromosome 14 of P. canaliculata (990 annotated genes) had 136, 140, and 122 orthologs in M. corrugatus, S. haemastoma, and C. ventricosus, respectively. Although syntenic blocks were not conserved in this chromosome, most of the orthologs were found in the pseudochromosomes 29 and 34 of M. corrugatus, in the pseudochromosomes 24 and 31 of S. haemastoma, and in the pseudochromosomes 32 and 34 of C. ventricosus, all pseudochromosomes which had none or very few syntenic blocks (fig. 2). Although a strong synteny was found between each of these species' pseudochromosomes (fig. 2), very few syntenic blocks were detected among the homologous pseudochromosomes within a species (supplementary fig. S2, Supplementary Material online). This is in contrast to Achatina fulica (Liu et al. 2021), where synteny is strongly conserved between homologous pseudochromosomes within the species.

Fig. 2.

Syntenic blocks in Caenogastropoda. Representation of the pseudochromosomes (black and gray rectangles with their pseudochromosome numbers) of the four Caenogastropoda generated using Synvisio (Bandi and Gutwin 2020). Each line represents a syntenic block in reverse or in forward orientation. Only links between side-by-side species are represented by full lines, whereas the dotted lines represent links between not juxtaposed species (no synteny detected between the two juxtaposed pseudochromosomes).

To further validate the presence of a putative WGD as suggested by our results, showing a higher number of pseudochromosomes in Neogastropoda and in closely related Caenogastropoda, and in agreement with reported chromosome counts retrieved from the previously published in the relevant literature (supplementary table S2, Supplementary Material online), we employed an independent approach based on the analysis of the number of Ks (synonymous substitutions) between paralogs within a genome. This approach is used to estimate the timing of gene duplication events by comparing the number of synonymous substitutions between paralogs (synonymous divergence) to estimate the time since the gene duplication event occurred (Van de Peer et al. 2009). Synonymous substitutions, which are changes in the DNA sequence that do not result in a change in the amino acid sequence of the protein, are used because they are less likely to be affected by natural selection than nonsynonymous substitutions, which do change the amino acid sequence of the protein. Here, the distribution of Ks between all paralogous genes in each of the neogastropods and M. corrugatus (fig. 3B) showed a first high peak corresponding to recent events of duplication as very few substitutions were detected. As very few genes were conserved between two putative homologous pseudochromosomes (ohnologs, see fig. 3A) in M. corrugatus and S. haemastoma, we plotted the distribution of Ks between orthologs in a pairwise comparison (fig. 3B) to identify in which order the events occurred. Three peaks were detected, where the first peak appears for paralogous genes in each of the three species, the second one for the paralog and ortholog genes within and between the three species, and the third one for the orthologs between P. canaliculata and each of the three species.

Fig. 3.

(A) Orthologs, paralogs, and ohnologs. The rectangle on a line represents a gene on a chromosome. Lines from species 1 to species 2 identify orthologs; the upper curved line defines paralogs (in tandem on the same chromosome); the lower curved line identify ohnologs (in two homologous chromosomes within a single species). (B) Ks density plot. Distribution of synonymous divergence between pairs of paralogs (first three lines, Conus, Monoplex and Stramonita) and orthologs (for all remaining lines). The first peak represents the newest events of duplication of each species (Conus ventricosus, Monoplex corrugatus, and Stramonita haemastoma); the middle peak corresponds to the whole genome duplication event; and the third peak represents the point of divergence between P. canaliculata and the three other species.

Homeobox Genes

Homeobox genes are involved in the early stages of embryonic development (Gehring et al. 1994). For our target species, no RNAseq data of development stages from previous studies were available yet, and thus the automatic gene annotation failed to detect them. Therefore, we manually annotated them, not only in the two genomes generated in this study but also in other genomes (P. canaliculata, Chrysomallon squamiferum, Haliotis rubra, Lottia gigantea, and Pecten maximus) for which homeobox genes were not annotated yet. These new annotations showed that the early diverged gastropods Ch. squamiferum, H. rubra, L. gigantea, and the bivalve P. maximus had exactly one cluster of homeobox genes with the same order on specific superscaffolds/pseudochromosomes. Anterior group genes (Ant, Lox) are placed on the same superscaffold/pseudochromosome and are followed by posterior group genes, whereas parahox genes (Xlox, Gsx, and Cdx genes) are found on a separate superscaffold/pseudochromosome (fig. 4). Neither A. fulica nor C. ventricosus homeobox genes were reannotated here (since already described in Liu et al. 2021 and Pardos-Blas et al. 2021), but both show a duplication of homeobox components on two different pseudochromosomes. Although A. fulica shows a duplication of a part of the anterior group genes (Lox5, Antp, Lox4, and Lox2), Hox1 and 5 in addition to Lox2 and Lox4 were also found as duplicated in C. ventricosus. Beside this duplication, C. ventricosus genes are arranged in a similar order as in P. canaliculata and opposed to A. fulica: posterior group genes are in between two types of anterior group genes in C. ventricosus, whereas anterior group genes are clustered together and followed by the posterior group genes in A. fulica. In M. corrugatus, most genes were found duplicated (we could not detect Hox4 and Post1 in Scaffold_30) and arranged in a different order when comparing each hox cluster in this species genome. Fewer genes were found duplicated in hox cluster in S. haemastoma (we could not find Lox5, Antp, Lox4, and Lox2 in Scaffold_28), but we found three tandem duplication events (for Hox4 in scaffold_23 and Hox1 and Post2 in Scaffold_28). Interestingly, in addition to the two copies of Hox5, we also detected a homeobox gene closely related to Hox5 in both M. corrugatus and S. haemastoma (see fig. 4A, “Hox5-like” in S. haemastoma and M. corrugatus). Our annotated homeobox genes were further verified with a phylogenetic analysis, in which orthologous genes of a given group are clustered together (fig. 4B).

Fig. 4.

Hox genes clusters. (A) Representation of the gene order of Hox genes; colored boxes with the arrow indicating the gene direction on the pseudochromosome (Chr) or superscaffold (Sc) represented by a line, as detected during the annotation. Star marks species for which data were retrieved from literature (Liu et al. 2020; Pardos-Blas et al. 2021). (B) Tree representing the phylogenetic relationships between all the sequences of the analysis; color dots correspond to the species as listed on the (A) part of the figure.

Putative Toxins Found within Syntenic Blocks

To target toxin-coding genes potentially resulting from a post-WGD sub or neofunctionalization, we focused on the genes found in syntenic blocks with P. canaliculata. From the 166 and 130 putative toxin genes annotated using the in-house toxin database (see Materials and Methods and supplementary table S3, Supplementary Material online) in M. corrugatus and S. haemastoma, respectively, 16 and 5 genes were found in syntenic blocks in M. corrugatus and S. haemastoma, respectively (table 2). From the 16 M. corrugatus genes, two were discarded as they could not be reliably identified as toxins or toxin-related genes. Then, for each of the 14 remaining genes, we searched for their putative orthologs in S. haemastoma (if not already found in the toxin annotation results) and in P. canaliculata, as well as for the possible paralogs in the homologous pseudochromosome of M. corrugatus (as defined by the P. canaliculata synteny, fig. 2; see Materials and Methods). Finally, 12 groups of genes related to toxins were analyzed in more details (table 2).

Table 2.

Toxins and Related Genes Detected in Syntenic Blocks in Monoplex corrugatus, Stramonita haemastoma, and Pomacea canaliculata.

Group	Nb Genes Per Species	Description of the Best Match against NR	Putative Function
1	2 M. corrugatus 2 S. haemastoma	Allatotropin-1 precursor (Charonia tritonis)	Conotoxin
	1 P. canaliculata	Hypothetical protein (P. canaliculata)
2	2 M. corrugatus 2 S. haemastoma	Hypothetical protein BaRGS_000210 (Batillaria attramentaria)	Enzyme toxin
	1 P. canaliculata	Zinc metalloproteinase nas-13-like (P. canaliculata)
3	8 M. corrugatus	Chitinase-like lectin, partial (Crepidula fornicata)	Chitinase: enzyme/toxin
	3 S. haemastoma	Probable chitinase 10 (P. canaliculata)
	3 P. canaliculata	Probable chitinase 10 (P. canaliculata)
4	2 M. corrugatus	Conotoxin precursor conkunitzin (Conus ebraeus)	Toxin
	2 S. haemastoma	Conotoxin superfamily conkunitzin-8 (C. magus)
	1 P. canaliculata	Hypothetical protein (P. canaliculata)
5	1 M. corrugatus	Kunitz-type peptide 1 (Bitis atropos)	Toxin
	1 S. haemastoma	Carboxypeptidase inhibitor SmCI-like isoform X2 (Limulus polyphemus)
6	2 M. corrugatus 2 S. haemastoma 1 P. canaliculata	FMRFamide precursor (Ch. tritonis)	Neurotrasmitter/toxin
7	1 M. corrugatus 1 S. haemastoma 1 P. canaliculata	Lysosomal alpha-mannosidase-like (P. canaliculata)	Enzyme for PTM: N-linked glycosylation, often present in venom
8	4 M. corrugatus 3 S. haemastoma	Protein disulfide isomerase (C. vexillum)	Enzyme involved in toxin folding, often present in venom
	2 P. canaliculata	Protein disulfide-isomerase-like isoform X1 (P. canaliculata)
9	1 M. corrugatus 1 S. haemastoma 1 P. canaliculata	conopressin/conophysin (Haliclona caerulea)	Neurotransmitter/toxin
10	2 M. corrugatus 1 S. haemastoma	CCAP-1 precursor (Ch. tritonis)	Toxin
	1 P. canaliculata	ConoCAP-like (P. canaliculata)
11	3 M. corrugatus 2 S. haemastoma 1 P. canaliculata	Probable serine carboxypeptidase CPVL (P. canaliculata)	Enzyme toxin
12	2 M. corrugatus 2 S. haemastoma	Achatin-2 precursor (Ch. tritonis)	Neuropeptide, maybe a toxin?
	1 P. canaliculata	Uncharacterized protein (P. canaliculata)

Note.—See supplementary table S3, Supplementary Material online for more details.

In most cases, we found at least one copy in each species but for the group 5, where the P. canaliculata orthologous gene was not detected (table 2). The groups 5, 7, and 9 had exactly one copy in each genome: we could not detect a second copy in the other pseudochromosomes. Genes from group 5 are probable Kunitz domain toxins; group 7 are probable enzyme responsible for posttranscriptional modification (PTM enzyme) known to be related to toxins (Soares and Oliveira 2009), whereas group 9 are probable neurotransmitters. The groups 1, 2, 4, 6, and 12 had all exactly two copies in M. corrugatus and S. haemastoma genomes, in homologous pseudochromosomes. All genes from group 1 had the best matches with allatotropin precursor genes (Gruber and Muttenthaler 2012) except for P. canaliculata gene, which had the best match with itself as it was already been annotated in NR as “hypothetical protein C0Q70_00675.” Genes from group 2 had best matches with portions of three genes, which are functionally annotated differently, but all had an astacin domain. This domain is common in venom metalloprotease toxins from different organisms (Trevisan-Silva et al. 2010; Ponce et al. 2016; Gerdol et al. 2019). All genes from group 4 had matches with conkunitzin-M16 that encode for Kv channels-blocking conotoxins (Finol-Urdaneta et al. 2020). However, the sequence from P. canaliculata was annotated as “hypothetical protein C0Q70_05870”. Genes from group 6 had all their best matches with a FMRFamide precursor from Charonia tritonis, a neurotransmitter whose function is unknown but is highly conserved through molluscs (Klein et al. 2021). Finally, genes from group 12 were annotated as neuropeptide as they had all their best matches with sequences annotated as precursor of achatin, a neuromodulator described in A. fulica (Liu and Takeuchi 1993), except for P. canaliculata in which the function is unknown (“uncharacterized protein LOC112557935”). For the group 10, we found two gene copies in M. corrugatus on the respective homologous pseudochromosomes, whereas only one copy was found in S. haemastoma. Group 10 genes had hits with precursors of ConoCAP, a cardioactive neuropeptide from cone snails (Möller et al. 2010), where the copies in the homologous pseudochromosomes in M. corrugatus are similar, but the only copy found in S. haemastoma did not conserve an open reading frame in the genome and had no hits in NR database. The group 11 included duplicated genes in the pseudochromosome 3 of M. corrugatus, whereas no other copies were detected in the other pseudochromosomes, similarly to S. haemastoma having two copies in the same pseudochromosome 2. They were all annotated as probable serine carboxypeptidase CPVL, known as main enzyme acting in blood digestion (Motobu et al. 2007). Finally, groups 3 and 8 had more than two copies in each of the three species where group 3 is a member of the chitinase gene family and group 8 of the thioredoxin gene family. As an illustration, we choose to highlight group 4 sequences, closely related to conotoxin, and group 1 by representing the syntenic blocks where it was found (fig. 5A) and the multiple alignment of their sequences with conotoxin sequences from C. betulinus and C. magus for group 4 sequences (fig. 5B).

Fig. 5.

Synteny block and multiple alignment of toxins similar to conotoxins. (A) Representation of syntenic blocks (light shadow) on each superscaffold/pseudochromosome and each species with putative toxins from group 4 (A1) and 1 (A2). Each horizontal line of a superscaffold/pseudochromosome represents one species. Red blocks are the detected toxins, black blocks are the genes found in the syntenic region, and green blocks are ortholog genes found in synteny (indicated by green lines). A toxin from group 1 found in Monoplex corrugatus (scaffold 2) was found in synteny with Pomacea canaliculata in a region that also shares a syntenic block with scaffold 1 of M. corrugatus close to the region of the paralog toxin. (B) Multiple alignment of group 4 (B1) and group 1 (B2) sequences (supplementary table S3, Supplementary Material online) manually annotated and found in a syntenic block with the addition of two conotoxins from Conus betulinus and C. magus, generated using MUSCLE v.3.8 (Edgar 2004).

Differential Gene Expression in the Neogastropoda

To assess the evolutionary origin of Neogastropoda genes, we performed differential gene expression on S. haemastoma transcriptomes. We sequenced three replicates for two tissues, the osphradium and the salivary glands, retrieving a total of 6,412 differentially expressed genes (DEGs) between the two tissues, 2,439 of which were up-regulated in the salivary gland. We also realigned the DEGs on the genome of S. haemastoma to detect potential ohnolog genes not previously found in our annotation. For most of the genes (66% and 61%) that were up- and down-regulated, we did not detect a paralog in the whole genome (table 3). For others, we found duplicates in the homologous chromosome (ohnologs). Although the majority of ohnologs had no differential expression, most of the ones found to be differentially expressed had similar expression pattern, that is both copies were either up- or downregulated. Few of the ohnolog copies had opposite expression profiles.

Table 3.

Numbers of Differentially Expressed Genes and Their Copies in the Genome.

	Upregulated Genes	Downregulated Genes
Total number of DEG	2,439	3,973
Number of copies found per DEG	1,611	2,410
Total copies found as ohnologs Expression of the copies: up | down | not DEG	267 62 (1) | 26 | 179	376 19 | 102 (2) | 255

Note.—Up and downregulated in salivary gland compared with osphradium tissues. Numbers between parenthesis in table line “Expression of the copies” are the predicted toxin-coding genes.

Among the toxins found in the genomes and described in table 2, only three pairs of ohnologs were found differentially expressed (supplementary table S3, Supplementary Material online). The two copies of the genes of S. haemastoma in groups 1 and 6 were both downregulated, and the two copies of genes from group 8 were both upregulated in the salivary glands.

Discussion

First Assembled Genomes of Tonnoidea and Muricoidea

The present work is a significant addition to the genome data available to date for Neogastropoda and Caenogastropoda. These previously included only eight neogastropod species, six of which belonging to Conoidea and two to Buccinoidea superfamily, and six caenogastropod species, four of which from a same family, the Ampullariidae (supplementary table S1, Supplementary Material online). Here, we added two genomes, one from the Muricidae family (Neogastropoda) and one for the Tonnoidea superfamily (Littorinimorpha), closely related to neogastropods (Lemarcis et al. 2022), thus broadening the taxonomic span of the available genomes. With the further addition of the recently sequenced genome of C. ventricosus (Pardos-Blas et al. 2021), we can now rely on reference genomes that effectively represent the diversity of predatory gastropods and remarkably expand the available molecular databases. Furthermore, these genomes were assembled at a chromosomal level with high BUSCO scores and thus constitute valuable assets to reconstruct evolutionary processes and patterns at the genomic level.

In addition to the high genome assembly quality, gene annotations provide a reliable nearly complete dataset of genes, whose completeness is comparable with available data on the Ampullarioidea superfamily (see annotation BUSCO score, table 1 and supplementary table S1, Supplementary Material online). General features of these two new genomes do not significantly differ from previously published neogastropods genomes, even though more genes were found, likely as a consequence of the higher quality of the sequencing (fewer base errors). Although gene numbers were 1.8–2 times higher in our two genomes than in Ampullarioidea genomes, almost twice less exons were found per CDS, with a length 1.5 times shorter, indicating a tendency to favor smaller genes in Caenogastropoda, a feature potentially associated to an increased transcription and expression frequency (Lopes et al. 2021). The overall sizes of our newly sequenced genomes were five to six times larger than Ampullarioidea genomes; this observation, coupled with the fact that more than twice pseudochromosomes were detected, strongly implies the occurrence of a WGD event. Together with the proliferation of repeated elements observed in the newly sequenced genomes (from 11% in P. canaliculata to 60% and 40% in M. corrugatus and S. haemastoma), these processes may have considerably increased genetic variation and diversity of gene expression regulation patterns in these species.

WGD in Neogastropoda and Tonnoidea

WGD is an important evolutionary event where all chromosomes are doubled in a genome. It provides increased opportunities for rearrangement, deletion, and insertion events that could contribute to evolvability, the capacity of species to adapt to different environments and diversify (Phuong et al. 2019), ultimately leading to the emergence of adaptive evolutionary novelties (Flagel and Wendel 2009; Van de Peer et al. 2009). Indeed, WGDs have been detected in many species-rich lineages, including, for example teleost fishes (Glasauer and Neuhauss 2014) and orchids (Moriyama and Koshiba-Takeuchi 2018), to name only few of them. In caenogastropods, karyotype analyses have been carried out on several individual species, providing data for a subsequent study that predicted a WGD event in Neogastropoda, Tonnoidea, Cypraeidae, and Capulidae based on genome size and chromosome counts (Hallinan and Lindberg 2011). WGD has been subsequently confirmed in C. ventricosus (Pardos-Blas et al. 2021), based on the detection of synteny and colinear blocks, the distribution of Ks peak, and the duplication of Hox gene clusters. Other WGD events were described in gastropods within the Heterobranchia subclass, as for example in A. immaculata and A. fulica (Liu et al. 2021), each possessing 31 chromosomes. In this work, we detected synteny between the different species (fig. 2), and the comparison with P. canaliculata showed a doubling (or sometimes quadrupling) of the pseudochromosome number in each of our target species. Having more than two copies of a chromosome raises the question of multiple WGD, but it can also be explained by duplication events limited to some chromosomes only. To clarify this issue, genome comparison between more species within neogastropods and in closely related caenogastropods like, for example Tonnoidea, Calyptraeidae, or Cypraeidae families (Lemarcis et al. 2022), would be needed. In addition, the exact timing of the WGD event remains to be determined. Our data suggest that the WGD event took place at least prior to the divergence of the Tonnoidea + Neogastropoda lineage (Lemarcis et al 2022) from the rest of the Caenogastropoda. However, according to the hypothesis proposed by Hallinan and Lindberg (2011), the WGD event may have also affected Cypraeidae and Capulidae, implying that it could have occurred earlier in the evolutionary history of Caenogastropoda, if confirmed in other lineages potentially closely related to neogastropods such as Calyptraeoidea and Velutinoidea. Although sequencing the genomes for the missing lineages would provide valuable insights going well beyond the scope of this work, such an endeavor would be highly costly and time consuming, given the high species diversity of the group; an estimation of their chromosome number and/or genome size could be a more affordable approach to determine the origin and nature of this WGD (Hallinan and Lindberg 2011).

The occurrence of syntenic blocks between the analyzed species (M. corrugatus and S. haemastoma) and P. canaliculata, the Ks distribution, and the duplication of Hox genes found within the species we analyzed strongly suggest WGD, with the same evidences found in Achatina species and C. ventricosus (Liu et al. 2020; Pardos-Blas et al. 2021). Although the first higher peak of the Ks distribution would correspond to the newly duplicated genes within each species, the second smaller peak would correspond to a WGD in neogastropods and tonnoideans, whereas the third peak would correspond to divergence between neogastropods, Tonnoidea, and the other caenogastropods. Hox clusters were found in both M. corrugatus and S. haemastoma, and both showed a duplication in the pseudochromosomes found to be homologs with those of P. canaliculata. In those clusters, we can identify some differences, possibly due to rearrangements, deletions, and/or insertions. All these evidences confirm the WGD event previously suggested (Bose et al. 2017; Pardos-Blas et al. 2021) in neogastropods and in the Tonnoidea, for which data concerning karyotypes or genome sequencing were completely lacking until the present study. The presence of only few colinear blocks within M. corrugatus and S. haemastoma (supplementary fig. S2, Supplementary Material online), in opposition of what was found in Achatina, strongly suggests a major role for other evolutionary events after the WGD. Indeed, genomic doubling would imply failure of cell division; thus, interchromosomal rearrangements are suggested to be the most frequent events after a WGD (Otto and Whitton 2000). Additionally, a WGD implies the retention of two copies of a gene or a repertoire of genes (Sémon and Wolfe 2007) which could lead to neofunctionalization and subfunctionalization events, as it has been reported in plants (He and Zhang 2005).

When a gene is duplicated, one of the copies can acquire a new function (neofunctionalization), retain a subset of the original gene’s functions (subfunctionalization), or degenerate to the point of losing its function (nonfunctionalization) (Force et al. 1999; Kuzmin et al. 2022). Gene expression analysis (Pasquier et al. 2017), structural analysis (Voordeckers et al. 2012), or experimental manipulation (Hittinger and Carroll 2007) can be applied to determine whether gene duplicates have undergone subfunctionalization, neofunctionalization, or nonfunctionalization. Since it is still a challenge to maintain neogastropods alive in the laboratory and available data are too limited to reliably predict protein structure in such nonmodel organisms, we focused on gene expression in two available tissues from S. haemastoma to gain insights on the fate of duplicated genes, the salivary glands (which are often involved in venom secretion, (Ponte and Modica 2017), and the osphradium, a sensory organ located in the molluscan mantle (Lindberg and Sigwart 2015). However, as synteny was rarely conserved between the homologous chromosomes within a given species, in most cases we were not able to detect the second copy of a gene, suggesting that the copy diverged enough to be detectable or has been depleted and lost its function, which is the most common event after WGD (Pasquier et al. 2017). Nonetheless, when the two copies were detected, most of the ohnologs that were found differentially expressed in the salivary gland compared with the osphradium had similar expression patterns. Whether these similar expression profiles correspond to sub- or neofunctionalization would require expression data from species that diverged before the WGD event, as described in snakes, whose venom evolution occurred through gene duplication and subfunctionalization of genes encoding salivary proteins (Hargreaves et al. 2014). Furthermore, our gene expression analyses were limited to two tissues, whereas expression profiles in a wide range of tissues and/or different developmental stages would be needed to fully understand how expression patterns of ohnologs are correlated. Future studies would ideally include structural characterization of proteins produced by each gene copy and functional assays to test for differences in their biological activity. Such a multidisciplinary approach would provide a more comprehensive understanding of the roles and functions of the duplicated genes in the organism.

Toxin Evolution

Toxins are peptides or proteins, produced or stored by an organism, which interfere with physiological or biochemical processes in another organism (Olivera et al. 2015; Arbuckle 2018). They were found to be produced by multiple species across the eukaryote kingdom (Casewell et al. 2013; Fry et al. 2015) and were known to serve multiple functions. In this work, we were attentive to find putative toxin candidates in both M. corrugatus and S. haemastoma. Multiple hits of toxins or related genes were found in both genomes. Among these genes, at least one group of genes (group 1) was found closely related to a conotoxin (fig. 5B), in both M. corrugatus and S. haemastoma; its occurrence was also reported in another neogastropod species belonging to the genus Vexillum (Costellariidae; Kuznetsova et al. 2022). This finding might indeed suggest that the appearance of conotoxin-related peptide could be a shared evolutionary novelty of predatory gastropods, predating the recruitment of such toxins before the origin of Conoidea, where they might indeed have gone through multiple duplication-neofunctionalization rounds to attain the observed diversity. More studies are needed also in other neogastropods (and related predatory marine snails) to confirm that the capacity to produce toxins is not a key innovation limited to cone snails or conoideans, as traditionally seen, but can be extended to the whole neogastropods (and close relatives) making them one of the most, if not the most, diverse group of venomous animals on Earth.

As we focused only on genes found in syntenic blocks, it should be noted that our findings constitute a limited and preliminary assessment of toxin repertoire in M. corrugatus and S. haemastoma. Moreover, our analysis was based on sequence similarity of toxin databases against the genomes and gene expression in S. haemastoma, but gene expression studies and proteomic analyses are needed in order to confirm their expression in secretory organs used in prey capture. Our DEG analysis actually suggests that many genes are up or down-regulated in the salivary glands that are involved in toxin production (Ponte and Modica 2017). Furthermore, the functions of these genes related to conotoxins, toxins, and related genes found in M. corrugatus and S. haemastoma are yet to be confirmed by testing their potential functions in vitro or in vivo. Nonetheless, identifying such sequences in the genome helps in understanding their evolutionary origin and history. Interestingly, the toxin genes on which we focused all had sequence similarities with P. canaliculata, the apple snail. This freshwater snail is extremely polyphagous, feeding mostly on vegetal material (Estebenet and Martin 2002) but no evidences indicate a possible toxin production. Thus, it is likely that these genes encode for proteins having other functions in P. canaliculata, and their duplication during the WGD followed by neofunctionalization events led to the apparition of new functions, including toxins. Duplication events, followed by sub and/or neofunctionalization, has been proposed as the main mechanism driving the evolution of venom genes in different clades, including spiders (Haney et al. 2016), snakes (Hargreaves et al. 2014; Giorgianni et al. 2020), and cone snails (Duda and Palumbi 1999; Chang and Duda 2012), especially in presence of positive selective forces (Kordiš and Gubenšek 2000). Growing evidence support the hypothesis of a relationship between WGD events and the emergence of evolutionary novelties, despite other factors, such as local duplications (Zhang et al. 1998) and co-option of single genes (Martinson et al. 2017), might be involved in the onset of adaptive traits. It has been argued that the ability of WGD to convey selective advantages and drive the appearance of evolutionary novelties would only emerge in presence of drastic environmental changes, remodeling ecological communities and increasing the availability of new niches (Moriyama and Koshiba-Takeuchi 2018). Indeed, in amphibian genomes, a potential correlation between WGD, adaptive radiation, and the well-known mass extinction event at the K-Pg boundary has been reported (Mable et al. 2011). In fishes and plants genomes, WGD that could have predated adaptive radiation could have been triggered by subsequent environmental changes, as proposed in the “WGD radiation lag-time model” (Eric Schranz et al. 2012). Interestingly, fossil data strongly suggest that the emergence of Neogastropoda and carnivorous caenogastropods, and their rapid radiation, could be placed in the context of the major reorganization of the benthic marine communities known as the Mesozoic marine revolution (Vermeij 1977; Bouchet et al. 2002; Harper 2003), where the evolutionary potential offered by the WGD could have been adaptive to fully exploit novel niches. Expanding genomic characterization to a broader array of taxa within the caenogastropods could help shedding light into the role of WGD and major genomic rearrangements in the origin of hyperdiverse groups.

Materials and Methods

DNA and RNA Samples

One female (MNHN-IM-2019-21838) and one male (MNHN-IM-2019-21839) specimen of M. corrugatus were collected in Corsica, France, whereas one female (MNHN-IM-2019-21840) and two male (MNHN-M-2019-21841 and MNHN-IM-2019-21842) specimens of S. haemastoma were collected in Capo Linaro, Italy. The foot (muscular tissue only) and the columellar muscle associated with the cephalopodium (pooled samples 1020.1 and 1020.5) of the M. corrugatus MNHN-IM-2019-21838 were used for DNA sequencing, whereas the heart (samples 1020.2 and 1020.15), kidney (samples 1020.3 and 1020.16), and ctenidium (samples 1020.4 and 1020.17) of each of the two specimens, respectively, were kept for RNA sequencing. For S. haemastoma, the heart, the capsule gland, the gland of Leiblein, the kidney, and ctenidium samples (pooled samples 0620.59, 0620.61, 0620.65, 0620.70, and 0620.72) from the specimen MNHN-IM-2019-21840 were used for DNA sequencing, and the osphradium (samples 0620.38, 0620.51 and 0620.71) and salivary gland (samples 0620.43, 0620.54, and 0620.68) of each of the three specimens, respectively, were kept for RNA sequencing. Samples were sent to Dovetail (Scotts Valley, California, USA) for library preparation, sequencing, assembly, and annotation.

PacBio Library Preparation, Sequencing, and Assembly

Both M. corrugatus and S. haemastoma DNA samples were generated following extraction protocol performed at Dovetail (Grohme et al. 2018) and were quantified using Qubit 2.0 Fluorometer (Life Technologies, Carlsbad, CA, USA). The PacBio SMRTbell library (∼20 kb) for PacBio Sequel was constructed using SMRTbell Express Template Prep Kit 2.0 (PacBio, Menlo Park, CA, USA) using the manufacturer-recommended protocol. The library was bound to polymerase using the Sequel II Binding Kit 2.0 (PacBio) and loaded onto PacBio Sequel II. Sequencing was performed on three PacBio Sequel II 8 M SMRT cells generating 359 and 195 Gb of data (M. corrugatus and S. haemastoma, respectively). The reads were used as an input to Wtdbg2 v2.5 (Ruan and Li 2020) with a genome size of 3.0gb, a minimum read length of 20 Kb, and a minimum alignment length of 8,192. Additionally, purge_dups v1.2.3 (Guan et al. 2020) was used to remove haplotigs and contig overlaps in the resulted assembly.

Omni-C Library Preparation, Sequencing, and Scaffolding

For each Dovetail Omni-C library, chromatin was fixed in formaldehyde in the nucleus and then extracted. Fixed chromatin was digested with DNAse I, chromatin ends were repaired and ligated to a biotinylated bridge adapter followed by proximity ligation of adapter containing ends. After proximity ligation, crosslinks were reversed and the DNA purified. Purified DNA was treated to remove biotin that was not internal to ligated fragments. Sequencing libraries were generated using NEBNext Ultra enzymes and Illumina-compatible adapters. Biotin-containing fragments were isolated using streptavidin beads before PCR enrichment of each library. The libraries were sequenced on an Illumina HiSeqX platform to produce an approximately 30 × sequence coverage. The input de novo assemblies and Dovetail OmniC libraries reads were used as input data for HiRise with MQ > 50 as parameter, a software pipeline designed specifically for using proximity ligation data to scaffold genome assemblies (Putnam et al. 2016). Dovetail OmniC library sequences were aligned to the draft input assemblies using BWA (Li and Durbin 2009). The separations of Dovetail OmniC read pairs mapped within draft scaffolds were analyzed by HiRise to produce a likelihood model for genomic distance between read pairs, and the model was used to identify and break putative misjoins, to score prospective joins, and to make joins above a threshold.

RNA Sequencing

Total RNA extraction was done using the QIAGEN RNeasy Plus Kit following manufacturer protocols. Total RNA was quantified using Qubit RNA Assay and TapeStation 4200. Prior to library prep, a DNase treatment was performed, followed by AMPure bead cleanup and QIAGEN FastSelect HMR rRNA depletion. Library preparation was done with the NEBNext Ultra II RNA Library Prep Kit following manufacturer protocols. Then, these libraries were run on the NovaSeq6000 platform in 2 × 150 bp configuration.

Gene Annotation Pipeline

Repeat families found in the genome assemblies of M. corrugatus and S. haemastoma were identified de novo and classified using the software package RepeatModeler, v.2.0.1 (Flynn et al. 2020). The custom repeat libraries obtained from RepeatModeler, v.2.0.1 (Flynn et al. 2020), were used to discover, identify, and mask the repeats in the assembly files using RepeatMasker, v.4.1.0 (Smit et al. 2013). CDSs from published annotations (Aplysia californica, Biomphalaria glabrata, Crassostrea virginica, L. gigantea, and Octopus bimaculoides) were used to train the initial ab initio model for both species using the AUGUSTUS software, v.2.5.5 (Stanke et al. 2006). Six rounds of prediction optimization were done with the software package provided by AUGUSTUS, v.2.5.5 (Stanke et al. 2006). The same CDSs were also used to train a separate ab initio model for each genome using SNAP, v.2006-07-28 (Korf 2004). RNAseq reads were mapped onto the genome using the STAR aligner software, v.2.7 (Dobin et al. 2013), and intron hints generated with the bam2hints tool within the AUGUSTUS software, v.2.5.5. MAKER, v3.01.03 (Cantarel et al. 2008), SNAP and AUGUSTUS (with intron-exon boundary hints provided from RNA-Seq) were then used to predict genes in the repeat-masked reference genome. To help guide the prediction process, Swiss-Prot peptide sequences from the UniProt database were downloaded and used in conjunction with the published protein sequences listed above to generate peptide evidence in the Maker pipeline. Only genes that were predicted by both SNAP and AUGUSTUS were retained in the final gene sets.

Duplications, Syntenic Regions, and Ks Distribution

Paralogs and orthologs were detected using the BRH method. Briefly, using BLASTp v2.13.0 (Camacho et al. 2009), all proteomes of M. corrugatus, S. haemastoma, C. ventricosus, and P. canaliculata were aligned to each other's with a maximal evalue of 1e−10 and at least 50% of one of the two sequences length aligned. Duplications within a species genome (M. corrugatus, S. haemastoma, C. ventricosus, P. canaliculata) and syntenic regions between two different species (pairwise comparison) were assessed using MCScanX v.2 (Wang et al. 2012). First, alignments between proteins from the same species and proteins from two different species were computed using BLASTp v2.13.0 (Camacho et al. 2009) with the parameters “number of alignments” equal to 5 and with a “maximal e-value” of 1e−10. The resulted BLAST output and GFF files were given to MCScanX v2 (Wang et al. 2012) scripts (McscanX, duplicate_gene_classifier, detect_collinear_tandem_arrays and dot_plotter) with default parameters. To visualize syntenic regions, online synvisio program (Bandi and Gutwin 2020) was used to generate the figures using the collinearity file resulted from MCScanX v.2. We defined homologous pseudochromosomes within C. ventricosus, M. corrugatus, and S. haemastoma resulting from the syntenic regions shared with P. canaliculata. All orthologs defined by the BRH methods and found in homologous pseudochromosomes had their protein sequences aligned between them using MAFFT v.7.471 with default parameters (Katoh and Frith 2012). Then, the protein multiple alignments were associated to the corresponding DNA sequence into a codon alignment using pal2nal with -nogap option. The number of synonymous substitutions per synonymous sites (Ks) for each BRH were calculated using KaKs_Calculator, v2.0 (Wang et al. 2010). The Ks distribution was plotted using ggplot package from R v3.6.3 (R Core Team 2022) with density function.

Homeobox Gene Annotations

Hox genes were specifically annotated in M. corrugatus, S. haemastoma, and in four species representatives of different gastropod subclasses, having their genomes sequenced and available: P. canaliculata (Caenogastropods; Architaenioglossa), Ch. squamiferum (Neomphalina), H. rubra (Vetigastropoda), and L. gigantea (Patellogastropoda). In addition, we annotated Homeobox genes in a Bivalvia species, P. maximus. First of all, we created a database of homeobox genes from different species including Cr. gigas (Zhang et al. 2012), L. goshimai (Huan et al. 2020), Acanthochitona rubrolineata (Huan et al. 2020), C. ventricosus (Pardos-Blas et al. 2021), Elysia marginata (Maeda et al. 2021), and P. canaliculata (full NCBI annotation). Then, the database was aligned against all predicted proteomes considered in this study. From the five best matches, we assigned the predominant function (minimum of three same functions over five). This constituted a second database of homeobox genes, and was aligned again on the same proteomes. Proteins not found in the first round and having more than 60% of identity and aligned on more than 60% of the sequence length were kept. Then, all the proteins detected were mapped on the seven genomes of interest using exonerate v2.2 (Slater and Birney 2005) and all matches were given to Gmove v2 (Dubarry et al. 2016) for automatic annotation. Each predicted homeobox gene was further annotated manually. Briefly, we verified the presence of homeobox domains using CDDsearch (online version, (Marchler-Bauer et al. 2015) and manually annotated the genes regarding the mapping of proteins from the reference database using IGV v2.5.3 (Thorvaldsdottir et al. 2013). The phylogeny of homeobox domains predicted was performed on their multiple alignment using the MAFFT webserver tool with default parameters (Katoh and Frith 2012) and IQ-tree v1.6.12 (Minh et al. 2020).

Putative Toxins Identification

All sequences from ConoServer (Kaas et al. 2012) and the sequences annotated with the key words “toxin” or “venom” from uniprotKB were mapped on both genomes using BLASTx v2.13.0 (Camacho et al. 2009) with a maximal e-value of 0.001. Next, Exonerate v.2.2 (Slater and Birney 2005) was launched to better align the exons boundaries of the hits identified using BLAST v2.13.0. We then used Gmove (Dubarry et al. 2016) in order to detect possible complete ORF in both genomes. We discarded reverse transcriptases from the analysis after running BLASTp v2.13.0 against NR. Only the putative toxin-coding genes detected in M. corrugatus and S. haemastoma, respectively, and found within syntenic blocks between each species and P. canaliculata were manually annotated using the combination of the online Introproscan domain search (Jones et al. 2014), ConoPrec (Kaas et al. 2012), and the online BLASTp v2.13.0 against NR database (Sayers et al. 2022). From the detected M. corrugatus' putative toxin genes, we further mapped them on the genomes of S. haemastoma, P. canaliculata, and itself to span all possible ortholog and paralogous genes using BLASTx v2.13.0 with an e-value of 1E−5 followed by exonerate with default parameters.

Differential Gene Expression in S. haemastoma

The analysis of differential gene expression was conducted on the available RNAseq data from S. haemastoma as three replicates were sequenced for each tissue (osphradium and salivary gland). First quality control was performed on each replicates raw reads using fastqc v.0.11.9 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and quality filtered and trimmed using fastp v.0.23.1with default parameters (Chen et al. 2018). Then, cleaned reads were mapped onto the genome sequences of S. haemastoma using STAR v.2.7.10b (Dobin et al. 2013) with the following parameters: –alignIntronMax 1,000 –alignMatesGapMax 10,000, and the addition of the gene annotation file. Each bam file was indexed using samtools v.1.15.1 (Danecek et al. 2021). Read counts were assessed using featureCounts from subread package v.2.0.1 (Liao et al. 2014). Differential gene expression analysis was performed using the R software v.4.2.2 (R Core Team 2022), Bioconductor (Gentleman et al. 2004) packages including DESeq2 v.1.38.1 (Anders and Huber 2010; Love et al. 2014), and the SARTools package v.1.8.1 (Varet et al. 2016). The parameters used for this analysis were the following: fitType parametric, cooksCutoff TRUE, independentFiltering TRUE, alpha 0.05, pAdjustMethod BH, typeTrans VST, and locfunc median. All genes detected as differentially expressed were then realigned on the genome using exonerate v.2.4.0 (Slater and Birney 2005) to check for potential undetected copies in the genomes and compared with our available annotation.

Supplementary Material

Supplementary data are available at Molecular Biology and Evolution online.

Data Availability

Monoplex corrugatus and Stramonita haemastoma genome and transcriptome raw sequences and assembly data are available on SRA under the BioProject PRJNA912994 and PRJNA913004, respectively. Gene annotations and toxins for both M. corrugatus and S. haemastoma as well as Hox genes annotation and predicted proteins of all seven species are available at https://doi.org/10.48579/PRO/CEWICN.