Interspecies comparison of sea star adhesive proteins

Abstract

Sea stars use adhesive secretions to attach their numerous tube feet strongly and temporarily to diverse surfaces. After detachment of the tube feet, the adhesive material stays bound to the substrate as so-called ‘footprints’. In the common sea star species Asterias rubens, the adhesive material has been studied extensively and the first sea star footprint protein (Sfp1) has been characterized. We identified Sfp1-like sequences in 17 additional sea star species, representing different taxa and tube foot morphologies, and analysed the evolutionary conservation of this protein. In A. rubens, we confirmed the expression of 34 footprint proteins in the tube foot adhesive epidermis, with 22 being exclusively expressed in secretory cells of the adhesive epidermis and 12 showing an additional expression in the stem epidermis. The sequences were used for BLAST searches in seven asteroid transcriptomes providing a first insight in the conservation of footprint proteins among sea stars. Our results highlighted a high conservation of the large proteins making up the structural core of the footprints, whereas smaller, potential surface-binding proteins might be more variable among sea star species.

This article is part of the theme issue ‘Transdisciplinary approaches to the study of adhesion and adhesives in biological systems’.

1. Introduction

Sea stars (Echinodermata, Asteroidea) are capable of adhering strongly and reversibly to a large variety of surfaces. As most echinoderms, they use extensible tube feet (podia) for locomotion and maintenance of position. Tube feet were long described as suction-based; a misassumption that is still widely spread. It has been shown that tube foot temporary adhesion is mainly based on adhesive and de-adhesive secretions [1,2]. When a tube foot detaches, the adhesive material stays behind as a so-called footprint, which consists of a surface-coupling layer topped by a structural meshwork [3,4]. In recent years, the adhesive of the common sea star Asterias rubens has been intensively studied, revealing its proteinaceous nature [5]. In a combined transcriptomic and proteomic approach, 34 footprint-specific proteins have been identified [6]. The second most abundant footprint protein has been fully characterized and named Sea star footprint protein 1 (Sfp1) [7]. Sfp1 is 3853 amino acid (aa) long and contains various functional domains known to be involved in protein–protein, protein–carbohydrate and protein–metal interactions. Sfp1 is auto-catalytically cleaved into four subunits (named Sfp1 alpha to Sfp1 delta) before its secretion. Peptide antibodies directed against Sfp1 alpha and Sfp1 beta have revealed the presence of these subunits within the adhesive cell secretory granules and within the meshwork of the footprints. The high abundance and localization of Sfp1 in footprints, and its multi-modular structure based on functional domains indicate that Sfp1 might function as a cohesion protein [7].

Sea stars inhabit diverse marine habitats from the intertidal to the abyssal zones. They are classified into five orders: Forcipulatida, Velatida, Spinulosida, Paxillosida and Valvatida [8]. Influenced by adaption to different environments and by evolutionary lineage, three tube foot morphologies have been described: simple disc-ending, reinforced disc-ending and knob-ending [9]. Disc-ending tube feet are mainly used for attachment and locomotion. Reinforced disc-ending tube feet differ from simple disc-ending by the presence of numerous collagen bundles in the disc that presumably allow stronger attachment. Simple disc-ending tube feet are common among Valvatida, whereas reinforced disc-ending tube feet are found in all sea star orders except Paxillosida [9,10]. Knob-ending tube feet are prevalent in the order Paxillosida and are used by the animals to locomote and to bury themselves in soft substrates. Despite the high diversity of tube foot morphologies and functions among sea stars, adhesive protein sequences have been exclusively investigated in the forcipulatid species Asterias rubens. This species is commonly found in exposed intertidal areas and has reinforced disc-ending tube feet allowing strong attachment [11]. Antibodies directed against the footprint material of A. rubens led to a cross-immunoreactivity in the adhesive epidermis of 12 other asteroid species [9], indicating a conservation of some adhesive components among all major orders and tube foot morphologies.

In this study, we investigated if adhesive proteins identified in A. rubens are conserved among sea stars representative from different taxa and with different tube foot morphologies. The protein sequence of Sfp1 was used to identify homologous sequences in asteroid transcriptomes and to unravel the evolutionary history and relationships of this protein among asteroids. The expression of 34 footprint-specific proteins within the tube foot adhesive epidermis of A. rubens was confirmed with in situ hybridization. These sequences were used for BLAST searches within seven asteroid transcriptomes and gave a first indication of the level of conversation of adhesive proteins among distantly related species. The expression of four conserved transcripts was additionally analysed in the tube feet of the distantly related species Asterina gibbosa and was identical to the expression pattern observed in A. rubens, further indicating a conserved function of the encoded proteins.

2. Material and methods

(a). Tube foot transcriptome of Asterina gibbosa

A de novo tube foot transcriptome was sequenced and assembled by the Beijing Genomics Institute (BGI, Hong Kong) using a Illumina HiSeq 2500 platform and Trinity assembler. For details, see the electronic supplementary material.

(b). Data collection of Sfps sequences

The assembled A. rubens tube foot transcriptome has been deposited as a Transcriptome Shotgun Assembly project at DDBJ/EMBL/GenBank under the accession GHKZ00000000. The version described in this paper is the first version, GHKZ01000000 [6]. All Sfps sequences and their specific accession numbers can be found in the electronic supplementary material, table S2. The Sfp1 protein full-length sequence (GenBank: AHN92641.1) and delta region (comprising aa 3257–3853), as well as the translated protein sequences of the other Sfps, were used as starting query sequences for BLASTp and tBLASTn searches in publicly available transcriptomic datasets: http://echinodb.uncc.edu/ [12], www.echinobase.org/ [13], http://marinegenomics.oist.jp/ [14] and in the newly generated transcriptome of Asterina gibbosa. For transcriptomes without a publicly available BLAST server [15], a local BLAST search system was used [16]. Sequences corresponding to BLAST hits with an e-value of e-85 or lower were used for further phylogenetic analyses (electronic supplementary material, table S1). The presence of structural domains was analysed with Interpro [17]. Predicted functional domains were only considered if they were consistently recognized in the full-length sequence and sequences of the Sfp1 subunits.

(c). Alignment and phylogeny

The phylogeny was based on 21 Sfp1 delta sequences from 18 species (electronic supplementary material, table S1). Additional outgroup sequences (the closest homologous sequences to Sfp1 delta subunit) were collected from non-Asteroidea genome and transcriptome data [18–20] and included in the analysis. A multiple alignment was performed with the MAFFT algorithm using consistency-based iterative refinement method E-INS-i implemented in Geneious 2 (v11.1.2). A strict trimming was then performed using TrimAL algorithm (automated heuristic) implemented in Metapiga 3.1 [21]. Maximum-likelihood (ML) phylogenetic analysis was performed using PhyML 3.0. with SPR tree searching and five random starting trees [22]. Prior to the ML analysis, automatic model selection was performed using Smart Model Selection implemented in the PhyML environment and based on the Akaike information criterion. The Whelan and Goldman (WAG) model of amino acids substitution was selected. Bootstrap analysis (1000) was performed. The analyses were performed with and without outgroup rooting.

(d). Immunohistochemistry and histological staining on tube foot sections

Tube feet were either fixed in 4% paraformaldehyde in phosphate-buffered saline for 2 h at room temperature and subsequently dehydrated in graded ethanol or directly fixed with 95% ethanol. Tube feet were then embedded in paraffin wax and cut longitudinally into 5 µm thick sections. After dewaxing and rehydration, either Heidenhain azan trichrome staining or immunohistochemistry was performed. For immunohistochemistry, antigen retrieval was achieved by incubation in 0.05% (w/v) trypsin (Sigma) and 0.1% (w/v) CaCl₂ for 15 min at 37°C. Samples were blocked in 50 mM Tris-buffered saline containing 3% (w/v) bovine serum albumin for at least 1 h at 4°C. The polyclonal anti-Asterias rubens Sfp1 beta antibodies [7] were diluted 1 : 50 in blocking solution and added to samples overnight at 4°C. Alexa Fluor 594-conjugated goat-anti-rabbit immunoglobulins (Invitrogen) were diluted 1 : 200 in blocking solution and applied for 1 h at room temperature. Samples were mounted in Vectashield^® (Vector) and analysed with a Zeiss Axioscope A1 epifluorescence microscope.

(e). In situ hybridization

RNA probe synthesis and whole-mount in situ hybridization were performed as previously described [23]. Paraffin section in situ hybridization was performed with some alterations to the previously published protocol for A. rubens [24] (see the electronic supplementary material for the detailed protocols). Sequences and primers are listed in the electronic supplementary material, table S2. Briefly, transcript-specific primers were designed and a T7/SP6 promoter region was added at their 5′ end. The purified polymerase chain reaction (PCR) product was then used to produce single-stranded digoxigenin-labelled RNA probes. RNA probes were used at a concentration of 0.1–0.2 ng µl⁻¹ and detected with anti-digoxigenin-AP Fab fragments (Roche) at a dilution of 1 : 2000. The signal was developed using the NBT/BCIP system (Roche) at 37°C. Images were taken with a Zeiss Axioscope A1 microscope.

(f). Rapid amplification of cDNA ends-polymerase chain reaction

Total RNA was extracted from tube feet with TRI reagent (Applied Biosystems) and reverse transcription 3′ and 5′ rapid amplification of cDNA ends (RACE) was carried out by using the FirstChoice RLM-RACE kit (Ambion). Obtained PCR products were cloned into a pCR2.1-TOPO vector using the TOPO TA cloning kit (Invitrogen) and Sanger-sequenced by StarSeq (Germany). For sequences and primers, see the electronic supplementary material, table S2.

3. Results

(a). Occurrence of Sfp1 in sea stars from different asteroid orders

In A. rubens, the protein Sfp1 contains three auto-cleaving sites (GDPH), mediating the cleavage of the large precursor protein into four subunits before its secretion (figure 1a) [7]. In the original paper characterizing Sfp1 [7], 14 functional domains were identified based on specific hits from the Conserved Domain Database of the National Center for Biotechnology Information [25]. We updated the proposed protein model by the inclusion of all functional domains predicted by Interpro [17] (figure 1a).

Figure 1.

Sfp1 structural domains and presence of Sfp1-like proteins in other sea star species, representing different orders and tube foot morphotypes. (a) Sfp1 subunits and domains in A. rubens, modified after [7]. EGF, EGF-like domain; FAMet, farnesoic acid O-methyl transferase domain; FIB, fibrinogen-like C-terminal domain; GAL, d-galactose/l-rhamnose binding lectin domain; FA58C, coagulation factor 5/8 C-terminal domain; PIL, protease inhibitor-like domain; TIL, trypsin inhibitor-like domain; TILa, trypsin inhibitor-like domain a; vWF, von Willebrand factor type D domain. (b) Sfp1-like proteins identified in sea star families. For transcriptomic data, black and grey symbols indicate complete and incomplete sequences, respectively. The tube foot morphology appears to be consistent throughout families, except in the Pterasteridae where species with reinforced disc-ending tube feet and species with knob-ending tube feet co-occur (electronic supplementary material, figure S2; [9,10]). (c–h) Exemplary scanning electron microscopy and histological sections of the three tube foot morphologies with (c,d) reinforced disc-ending (Asterina gibbosa), (e,f) knob-ending (e) Astropecten irregularis and (f) Luidia ciliaris, and (g,h) simple disc-ending (g) Protoreaster lincki and (h) Protoreaster nodosus.

The primary sequence of Sfp1 was used to identify homologous sequences in publicly available transcriptomic datasets and in the newly generated tube foot-specific transcriptome of Asterina gibbosa. Sfp1-like sequences were found in 17 sea star species, representative of 10 families from four out of the five sea star orders (figure 1b; electronic supplementary material, table S1). Presumably owing to the unusual large size of Sfp1 (3853 aa, encoded by a 11 559 bp mRNA) and to mRNA degradability, most detected sequences were partial. Only in three species—Pisaster ochraceus, Pteraster tesselatus and Patiria pectinifera—was the full-length coding sequence of Sfp1 present as one assembled transcript. The amino acid identity/similarity between the protein sequences translated from these transcripts and A. rubens Sfp1 was the highest in the other forcipulatid Pi. ochraceus, with 76.7/83.3%, followed by the valvatid Pa. pectinifera with 51.46/62.31%, and the velatid Pt. tesselatus, with 45.6/56.5%. In two other species—Acanthaster planci and Labidiaster annulatus—an almost complete Sfp1 sequence was assembled as one transcript, but lacked the N-terminal part, including the secretory signal peptide and the first two or three domains, respectively. In all five translated proteins the main functional domains of A. rubens Sfp1 and the three auto-cleaving sites were conserved. A variation in the number of galactose-binding lectin (GAL) domains was observed, however, in the Sfp1 delta subunit (see next section). In the fully annotated sea star genome of Ac. planci [14], the Sfp1 gene is present on a single scaffold and comprises 66 exons and 65 introns, overall spanning an area of 61 915 bp.

Sfp1-like sequences were discovered in transcriptomes generated from different sea star tissues, including arms, ovaries and tube feet. For instance, in the ovary-specific transcriptome from A. rubens [13], a 1386 bp long cDNA sequence with 99% identity (1386/1389 bp) to the Sfp1 cDNA sequence was present. However, it is unknown if Sfp1 mRNA in tissues other than the tube foot adhesive epidermis is transcribed into protein and what the protein function might be.

Polyclonal antibodies directed against peptides of the A. rubens Sfp1 subunits were raised in a previous study [7]. The anti-Sfp1 beta antibodies were used in immunohistochemistry on tube foot sections from members of all five orders (electronic supplementary material, table S3). Only in two (Marthasterias glacialis and Henricia sp.) out of 24 tested species, was immunoreactivity within the adhesive cell secretory granules observed (electronic supplementary material, figure S1). However, the high sequence variability between the species in the Sfp1 region where this peptide occurs probably explains this low number. Noteworthy, the anti-Sfp1 beta antibodies reacted with the adhesive cell secretory granules of Henricia sp., a species of the order Spinulosida for which transcriptomic data are sparse and no Sfp1-like sequences could be identified in our database search.

Histological staining was performed on tube foot sections from 21 species to determine their tube foot morphology (electronic supplementary material, figure S2). Exemplary scanning electron microscopy and histological staining of the three tube foot morphotypes is shown in figure 1c–h. The complete list of the staining results, including sample fixation and species full names, can be found in the electronic supplementary material, table S3.

(b). Phylogeny of Sfp1

Next, we analysed the phylogenetic relationships between Sfp1 proteins among asteroid species. As most sequences we obtained were partial, we restricted the analysis to the Sfp1 delta subunit. This C-terminal subunit was the best represented in the different transcriptomes, most likely owing to the common polyA-tail enrichment in sequencing protocols. Overall, we used sequences from 18 species for the analysis (electronic supplementary material, table S1). In 13 species, the Sfp1 delta subunits were retrieved as complete sequences and, in one additional species (Remaster gourdoni), only the first three amino acids were missing. In all these 14 species, the main functional domains of Sfp1 delta were conserved. However, in four species—Pt. tesselatus, Psilaster charcoti, Ac. planci and Glabraster antarctica—only one GAL domain was recognized by Interpro [17]. Another variation was observed in Astropecten duplicatus, in which the trypsin inhibitor-like (TILa) domain was absent. In three species (Asterias forbesi, Luidia clathrata and Patiria miniata), the retrieved sequences showed high similarity to Sfp1 delta and contained at least two of the conserved domains, but were incomplete.

As an outgroup, sequences with the highest similarities to Sfp1 from three other echinoderm species (the brittle star Amphiura filiformis, the sea cucumber Apostichopus japonicus and the sea urchin Strongylocentrotus purpuratus) were selected (electronic supplementary material, table S1). Maximum-likelihood phylogenetic analysis with 1000 bootstrap replicates resulted in the presented phylogenetic tree (figure 2). The species from each of the four represented sea star orders tended to cluster together, but the relationships between Sfp1 delta sequences from species of Paxillosida and Valvatida could not be clearly resolved. Valvatida and Paxillosida thus appeared paraphyletic in our analysis. The longer branches associated with the sequences from As. duplicatus and Lu. clathrata reflect a higher amino acid substitution rate in these species which, in turn, could be associated with a weaker selective pressure on the sequence of Sfp1 delta in Paxillosida. It is also in this order that sequence similarity between Sfp1 delta homologues and the protein from A. rubens is the lowest (median of 58.67%), followed by Velatida (60.03%), Valvatida (63.97%) and finally Forcipulatida (83.78%, including A. rubens; figure 2a). Our findings therefore indicate that Sfp1 delta is highly conserved among asteroids and sequence variations would be mainly influenced by evolutionary distance between the lineages.

Figure 2.

(a) Amino acid similarity (%) between Sfp1 delta from A. rubens and homologous proteins from 17 other sea star species. Similarity measurements were based on a local alignment after trimming. Each point represents one species (with grey points representing incomplete sequences after trimming) and the vertical lines indicate the median similarity for each considered sea star order. (b) Phylogeny of Sfp1 delta proteins within sea stars, based on maximum-likelihood analysis. The tree comprises three distinct clades corresponding to sea stars taxonomic orders: Velatida, Forcipulatida and Paxillosida + Valvatida. Both Valvatida and Paxillosida appear paraphyletic in our analysis. Bootstrap values are illustrated by shades of grey. The branch lengths represent the number of amino acid substitutions per site. Symbols after the names indicate the tube foot morphology with open squares simple disc-ending, filled squares reinforced-disc ending and open inverted triangles knob-ending.

(c). Expression of other Sfps in the tube feet of Asterias rubens

Sfp1 is the best characterized but, by far, not the only footprint protein in A. rubens. In a previous study, 34 footprint-specific proteins have been identified both at the mRNA and protein level and 41 additional proteins co-occurred in footprint and mucus samples [6]. Whereas the mucus may derive from tube foot and/or body wall secretions, the footprint proteins are supposed to be produced in the adhesive epidermis of tube feet. Sfp1 has been identified within both footprint and mucus samples, but showed a much higher abundance in footprints [6]. We performed whole mount and section in situ hybridization using specific Sfp1 antisense probe and confirmed its sole expression in the adhesive cells of tube feet (figure 3a,b). The negative control using Sfp1 sense probe did not lead to any staining (figure 3c,d). Next, in situ hybridization experiments were performed in order to localize the expression sites of the 34 footprint-specific proteins. As traces of Sfp1 were also found in mucus samples at the proteomic level [6], another protein (encoded by transcript Arub-21) sharing this distribution profile was added to our list. During sequence analyses, some transcripts were identified as overlapping sequences and treated as one transcript (Arub-7 and Arub-9). Therefore, our final list for the in situ hybridization screening consisted of 34 sequences, including Sfp1 (figure 3i, for sequences and accession numbers of the transcripts, see the electronic supplementary material, table S2). Out of the 34 sequences, 22 were exclusively expressed in secretory cells of the adhesive epidermis (figure 3e,f and electronic supplementary material, figure S3) and 12 within both the adhesive epidermis and the stem epidermis (figure 3g,h and electronic supplementary material, figure S4).

Figure 3.

Exemplary expression pattern of Sfps in A. rubens and level of conservation among sea star species. (a,c,e,g) Whole mount and (b,d,f,h) section in situ hybridization of selected footprint proteins. (a,b) Sfp1, (c,d) Sfp1 sense (negative control), proteins encoded by (e,f) Arub-12 and (g,h) Arub-22. (i) List of Sfps from A. rubens, indicating their site of expression in tube feet, their translated protein length and sequence completeness, and the Expect values (E) of the best hits from BLAST search results in transcriptomes from seven other species (ae, adhesive epidermis).

(d). Occurrence of Sfps in other asteroid species

Next, we asked if all Sfps from A. rubens are conserved among sea stars. As the publicly available transcriptomes were of varying quality and based on different tissues, we focused on the five transcriptomes in which Sfp1 was retrieved as complete or almost complete sequences. We additionally included the transcriptome from Ps. charcoti as a representative of the order Paxillosida, and the newly made tube foot-specific transcriptome of Asterina gibbosa. For many of the 34 A. rubens Sfps, only a partial protein/cDNA sequence is known (figure 3i). The cDNA sequences encoding for four of the most abundant Sfps were therefore elongated using 3′ and 5′ RACE PCR and thereby the full-length coding region of three more transcripts could be determined (sequences in the electronic supplementary material, table S2). We used the translated protein sequences of all A. rubens Sfps as queries for the BLAST searches. As very often partial protein sequences were used as query and/or retrieved during the BLAST search, evaluation of the similarity between different species is preliminary at this stage and the results should be interpreted with caution. Nevertheless, these results highlighted that proteins expressed in both the disc adhesive epidermis and the stem epidermis tend to be more conserved than proteins exclusively expressed in the adhesive epidermis (figure 3i). Additionally to Sfp1, only two large proteins solely expressed in the adhesive epidermis (Sfps encoded by transcripts Arub-10 and Arub-13) were highly conserved in all seven species. The protein encoded by Arub-10 shared many characteristics with Sfp1, including domain structure (two fibrinogen-like domains, three von Willebrand factor domains, two coagulation factor 5/8 C-terminal domains, three TIL domains, three GAL domains and five EGF-like domains) and the presence of two autocatalytic cleavage sites. The transcript Arub-13 encodes for an alpha-macroglobulin-like protein. For several other adhesive epidermis-specific Sfps (e.g. proteins encoded by Arub-6, Arub-12 or Arub-17), sequence similarity appeared to vary according to evolutionary lineage.

To validate the transcript expression pattern in another species, we performed in situ hybridization of four conserved sequences from the valvatid Asterina gibbosa (Agib-Sfp1, Agib-13, Agib-15 and Agib-25), homologous of three adhesive epidermis-specific transcripts (Sfp1, Arub-13 and Arub-15) and one with an additional expression in the stem (Arub-25) from A. rubens (electronic supplementary material, figure S5). All tested transcripts showed the same expression pattern in the tube feet of Asterina gibbosa as in those of A. rubens, further confirming a conserved function of these proteins between species.

4. Discussion

A growing research area is focusing on the characterization of biological and especially marine adhesives [26,27]. So far, most studies have been based on the adhesive characterization in single species. Interspecies comparison of adhesive proteins, however, allows identifying common protein characteristics like amino acid composition and functional domains and thereby facilitates the design of biomimetic glues. In sea stars, comparative studies on the adhesive material have been limited to morphological and immunohistochemical investigations, which hint at some conservation between species [2,9,28,29]. In this study, we used the footprint protein (Sfp) sequences from A. rubens to investigate sequence similarity within asteroid orders. To date, however, only one footprint protein, Sfp1, has been fully characterized and shown to have a cohesive function within the adhesive layer.

(a). Sfp1 localization and phylogeny

In Ac. planci and Pa. pectinifera, the expression level of mRNAs between tissues has been investigated [14,15] and, in both species, Sfp1 is specifically expressed in tube feet. Our in situ hybridization experiments confirmed previous immunohistochemical results [7] and highlighted that, in tube feet, Sfp1 is exclusively expressed within the secretory cells of the adhesive epidermis in A. rubens (figure 3a,b) as well as in Asterina gibbosa (electronic supplementary material, figure S5a). Nevertheless, within transcriptomic databases, Sfp1 mRNA sequences were found in various tissues, indicating that at least a low level of expression is present there. There is no information on protein levels of Sfp1 in tissues other than tube feet and on what its functional role might be.

The anti-Sfp1 beta antibodies were tested on tube foot sections from 24 species but did not lead to a staining in the majority of samples. However, as the antibodies were directed against a short peptide sequence that varies among species, a negative staining result does not indicate the absence of Sfp1. Indeed, in Asterina gibbosa the antibody showed no immunoreactivity [28], but Sfp1 was identified in the new tube foot-specific transcriptome and its expression within the adhesive glands was confirmed with in situ hybridization (electronic supplementary material, figure S5a). The only two species with a positive Sfp1 beta staining in adhesive secretory granules were M. glacialis and Henricia sp (electronic supplementary material, figure S1). The latter belongs to the order Spinulosida that is not well represented in transcriptomic databases. The positive Sfp1 beta immunoreactivity is an indication that Sfp1 might be also present and conserved in this fifth sea star order. Overall, our database searches and antibody staining revealed that Sfp1 is present in all sea star orders and in representatives of all three tube foot morphotypes, regardless of functional adaptations to habitat.

The evolutionary relationships within asteroids are highly debated, resulting in several, often contradicting, phylogenetic studies over the last years [8,12,30–33]. In the most recent phylogeny, based on transcriptomic data, the five asteroid orders are split into two major groups with Forcipulatida and Velatida on one side and Paxillosida, Spinulosida and Valvatida on the other [12]. In our Sfp1 delta phylogenetic tree, the Velatida was placed as a sister group to the other three represented orders and Valvatida and Paxillosida appeared to be paraphyletic (figure 2). Overall, this tree is in accordance with previously published general phylogenies of asteroids that resulted in the same placement of the different orders [30,33]. Therefore, we conclude that the sequence variation between the Sfp1 delta orthologues is mainly influenced by the evolutionary distance between lineages. The lowest sequence similarity to A. rubens Sfp1 delta was found in Paxillosida. Interestingly, paxillosid species possess knob-ending tube feet and mainly live on or in soft substrates. It seems possible that in these habitats tube foot tenacity is not as crucial for survival as on hard surfaces where disc-ending tube feet are common, which might indicate a weaker selective pressure on Sfps in general and Sfp1 in particular in members of this order. However, adhesive strength measurements of knob-ending tube feet are lacking and, therefore, this is speculative at the moment.

(b). Interspecies comparison of footprint proteins

Among Sfps, we assume that proteins responsible for tube foot temporary adhesion are exclusively expressed in secretory cells of the adhesive epidermis. In A. rubens, this specific expression pattern reduces the number of putative adhesive proteins to 22, although it must be kept in mind that some of these proteins could also be involved in tube foot detachment. The 12 proteins with an additional expression in the stem epidermis might therefore have a more general function like forming the glycocalyx-like cuticle for example. Interestingly, on average, the sequence conservation of the 22 adhesive epidermis-specific proteins among species was lower than that of the 12 non-specifically expressed proteins.

Within the list of adhesive protein candidates, only large proteins, like Sfp1 and the proteins encoded by transcripts Arub-10 and Arub-13, were highly conserved among species. Our findings might indicate that whereas large proteins making up the structural core of the footprints (i.e. bulk proteins) are conserved among sea stars, the sequences of proteins responsible for surface contact might vary more considerably. This trend was also observed in other marine adhesives. In barnacles, the protein sequence of the large bulk protein cp-100 k is highly conserved [34,35]. The sequence similarity among the smaller potential surface-binding proteins is lower, but these proteins display characteristic residue composition and biochemical properties [34,36,37]. Amino acid composition analyses cluster barnacle cement proteins into two or three groups but, in any case, separate large cohesive proteins from potential surface-binding proteins [36,37]. It seems that in large structural proteins the selection pressure is high for the conservation of functional domains, while in surface-binding proteins the relative amino acid composition is more important than the primary sequence.

Our findings indicate that large, predicted cohesive proteins are conserved among distantly related sea stars, regardless of inhabited environments and morphological adaptions of their tube feet. By contrast, the sequences of smaller, potential surface-binding proteins might be more variable and potentially more influenced by adaptations to the habitat and mode of living. However, it must be highlighted that, for many Sfps from A. rubens, the complete coding sequences are not available and that, for comparison among asteroid species, transcriptomes of varying origin and quality were used. Sequence conservation among adhesive proteins might therefore be underestimated. Additionally, functional conservation could be achieved through similar biochemical properties, like similar amino acid composition or post-translational modifications rather than sequences similarity. For a more precise analysis, genomic data with complete sequences would be required. Nevertheless, as this is, to our knowledge, the first study of this kind in sea stars, we consider the observed trends valuable information for upcoming studies.

Data accessibility

The assembled Asterias rubens tube foot transcriptome has been deposited as a Transcriptome Shotgun Assembly project at DDBJ/EMBL/GenBank under the accession GHKZ00000000. The version described in this paper is the first version, GHKZ01000000.

Authors' contributions

B.L. and P.F. conceived the study, interpreted results and wrote the paper. M.A., M.L., E.H. and B.L. performed in situ hybridization experiments. B.L. performed immunohistochemistry and histological staining. J.D. performed bioinformatical analyses including alignments and phylogeny. M.L. and E.H. performed RACE experiments. All authors revised and approved the final manuscript.

Competing interests

We declare we have no competing interests.

Funding

B.L. is funded by a Schrödinger Fellowship of the Austrian Science Fund (FWF): [J-4071]. P.F. is Research Director of the Fund for Scientific Research of Belgium (F.R.S.-FNRS). J.D. is supported by a WISD-PDR project from the F.R.S.-FNRS. Work supported by a FNRS CDR grant no J.0013.18, and by the ‘Communauté française de Belgique—Actions de Recherche Concertées’ (ARC-17/21 UMONS 3). B.L., E.H., M.A., M.L. and P.F. are members of the COST Action ‘European Network of Bioadhesion Expertise’ (CA15216).