Introduction

Mineralized tissues are a composite material of inorganic crystals and organic molecules. Although the organic molecules, collectively known as the organic matrix, are a minor component of mineralized tissues, they exert remarkable control over the mineralized tissues (Lowenstam 1981; Lowenstam and Weiner 1989; Belcher et al. 1996; Falini et al. 1996). Molluscan shells have provided one of the best studied examples of the organic matrices in mineralized tissues in invertebrates, and many shell matrix proteins have been identified to date (Miyamoto et al. 1996; Shen et al. 1997; Sudo et al. 1997; Sarashina and Endo 1998, 2001; Samata et al. 1999; Kono et al. 2000; Mann et al. 2000; Marin et al. 2000; Miyashita et al. 2000; Weiss et al. 2000, 2001; Marxen et al. 2003; Michenfelder et al. 2003; Zhang et al. 2003; Suzuki et al. 2004; Tsukamoto et al. 2004; Gotliv et al. 2005).

These shell matrix proteins in mollusks are made of different modules. In most of the domains of shell matrix proteins, the sequence similarities to other proteins seem to be due to the biased amino acid compositions. Thus the evolutionary relationships among them are not clear and the origins of most shell matrix proteins remain obscure. Some domains of the shell matrix proteins in mollusks, however, exhibit remarkable sequence similarities among metazoans or even more distantly related organisms, suggesting a Precambrian origin of the genes. In the case of nacrein and N66, encoding shell matrix proteins in the nacreous layer of the pearl oyster Pinctada species (Miyamoto et al. 1996, Kono et al. 2000), carbonic anhydrase domains of these proteins have probably been inherited from bacteria by exon shuffling (Bork 1991; László 1996, 1999; Marin and Luquet 2004). Two other proteins extracted from the nacreous layer of the abalone Haliotis laevigata, called Perlucin and Perlustrin, have highly conserved domains among metazoans (Mann et al. 2000; Weiss et al. 2001). The former, Perlucin, shows similarity to C-type (Ca2+-dependent) lectin-like domain proteins (CTLDs), which has been found in different calcifying systems such as in mammals ( De Caro et al. 1987; De Reggi and Gharib 2001), birds (Mann and Siedler 1999; Lakshminarayanan 2003; Mann and Siedler 2004), a shark (Neame et al. 1992), and sea urchins (Wilt et al. 2003). The latter, Perlustrin, shows similarity to the mammalian insulin-like growth factor binding proteins (IGFBPs). IGFBPs are involved in bone resorption and formation in mammals (Ueland 2004).

Dermatopontin, a molluscan shell matrix protein extracted from the crossed lamellar layer of the freshwater snail Biomphalaria glabrata (Marxen et al. 2003), is one such protein, considered to be of Precambrian origin. Dermatopontin, also called TRAMP (tyrosine-rich acidic matrix protein), has been found in mammals (Neame et al. 1989, Superti-Furga 1993, Forbes et al. 1994), an arthropod (Fujii et al. 1992), and a sponge (Schütze et al. 2001). This protein has a widespread tissue distribution in mammals, including skin, skeletal muscle, heart, lung, kidney, cartilage, and bone, with possible functions in cell-matrix interactions and matrix assembly (Forbes et al. 1994). In mollusks, Dermatopontin is considered to be a major component of the shell matrix proteins (Marxen and Becker 1997; Marxen et al. 2003).

The fossil record indicates that shell-bearing representatives of mollusks were present in the Lower Cambrian. Thus molluscan shell matrix proteins are likely to have been innovated or recruited in the Late Proterozoic or Early Cambrian (Marin and Luquet 2004). This argument, however, is in disagreement with the fact that, whereas Dermatopontin is a major shell matrix component in the freshwater snail B. glabrata (Marxen et al. 2003), this protein has not so far been found from the shells of the abalone and the pearl oyster, which have been best studied in molluscan biomineralization. At least in the case of dermatopontin, the recruitment for the function involved in biomineralization may perhaps have taken place more recently than in the Cambrian. In the present report, we identify several dermatopontin genes in species relatively closely related to B. glabrata, or within Pulmonata. We also perform a comparison of primary structures of Dermatopontin and expression analyses of dermatopontin. Based on these results, we characterize functionally important residues and discuss the molecular evolution of molluscan dermatopontin genes and the origin of Dermatopontin as a molluscan shell matrix protein.

Materials and Methods

Samples

Lymnaea stagnalis, originally collected in Neustadt, Donau, Germany, and Biomphalaria glabrata, originally collected in Costa Rica, were reared in deionized and calcium carbonate saturated tap water at about 23°C. Japanese land snails, Euhadra herklotsi nesiotica, E. brandtii, E. peliomphala, E. amaliae, Satsuma japonica, Mandarina aureola were collected from Ei Town, Kagoshima Prefecture; Tsukuba City, Ibaraki Prefecture; Heta Village, Shizuoka Prefecture; Heta Village, Shizuoka Prefecture; Kitaibaraki City, Ibaraki Prefecture; Hahajima, Bonin (Ogasawara) Islands, respectively.

RNA Purification and Complementary DNA Synthesis

We extracted the total RNA from the mantle tissues using Isogen (Nippon Gene, Toyama, Japan). The total RNA was applied as a template for reverse transcription to prepare complementary DNA (cDNA), primed with oligo(dT)20 (Toyobo, Osaka, Japan), and catalyzed by ReverTra Ace (Toyobo). For performing 3′ rapid amplification of cDNA ends (RACE), we primed the total RNA with Tt-1 primer (Table 1), rather than oligo(dT)20 primer.

Table 1 List of primers
Full size table
Figure 1
figure 1

Alignment of the amino acid sequences of Dermatopontin of molluscs, the demosponge Suberites domuncula (Porifera; Schütze et al. 2001; DDBJ Accession No. AJ299722), Atlantic horseshoe crab Limulus polyphemus (Arthropoda; Fujii et al. 1992; DDBJ Accession No. M96983) and human Homo sapiens (Chordata; Superti-Furga et al. 1993; DDBJ Accession No. AL049798). The species for which sequences determined in this work are shown in bold. The amino acid sequence of one of the two types of Biomphalaria glabrata Dermatopontin (BgDerm1) has already been reported by Marxen et al. (2003; DDBJ Accession No. AY161235). Potential glycosylation sites are indicated in bold and underlined. Potential phosphorylation sites are indicated in bold italic. Highly conserved motif of E-D-R-X-W/F/Y-X-F/Y/I/L/M-X1-2-C, where X is any amino acid, and conserved cysteine residues are boxed. N- and C-terminal amino acid residues are in bold. DDBJ Accession numbers of the genes identified in this work are AB210093 (Lsderm1), AB210094 (Lsderm2), AB210095 (Lsderm3), AB210096 (Bgderm1), AB210097 (Bgderm2), AB210098 (Ebderm1), AB210099 (Ebderm2), AB210100 (Ehderm1), AB210101 (Ehderm2), AB210102 (Eaderm1), AB210103 (Epderm2), AB210104 (Maderm1), AB210105 (Maderm1) and AB210106 (Sjderm2).

Full size image

Amplification and Sequencing of cDNA

The target cDNA sequences, encoding partial amino acid sequences of Dermatopontin, were amplified by reverse transcription-polymerase chain reaction (RT-PCR) from the cDNA using the degenerate antisense primer DgA1 (Table 1) and one of the three degenerate sense primers (DgS1, DgS2, and DgS3; Table 1). The four degenerate primers, DgS1, DgS2, DgS3, and DgA1, were designed based on the amino acid sequence of B. glabrata Dermatopontin (Marxen et al. 2003) (Fig. 1). Three-prime RACE was performed using the cDNA prepared for 3′ RACE and the primer pair of Tt-A2 primer (Table 1) and one of the three degenerate sense primers, DgS1, DgS2 and DgS3 (Table 1). The PCR products were subcloned into the pGEM-T Easy Vector (Promega, Madison, WI, USA). The inserts of the vectors were sequenced using T7 and SP6 primers as well as the two internal primers, LsS4 and LsS5 (Table 1).

Phylogenetic Analyses

The 14 sequences of Dermatopontins for mollusks (BgDerm1, BgDerm2, LsDerm1, LsDerm2, LsDerm3, EbDerm1, EbDerm2, EhDerm1, EhDerm2, EaDerm1, EpDerm2, MaDerm1, MaDerm2, SjDerm1) and the sequence of the outgroup S. domuncula (SdDerm) were manually aligned. The sequence of amino acid position 87–137 of BgDerm1 and the corresponding amino acid sequences of other molluscan and S. domuncula Dermatopontins were used for phylogenetic analyses. The sequence of S. domuncula served as an outgroup. We performed the Bayesian estimation of phylogeny (Huelsenbeck and Ronquist 2001) using the program MrBayes version 3.0 supplied on MrBayes homepage site (http://www.morphbank.ebc.uu.se/mrbayes). To approximate the posterior probabilities of the trees conditioned on the observed amino acid sequences, we used the likelihood model with a gamma-distributed rate variation and the average over 11 models for the rate matrix for amino acid data. Eight Markov chains were run in parallel for 106 generations. After the burn-in set at 2000 generations, Markov chains sampled trees every 100 generations of 106 generations. Results were compared among the eight chains and a 50% majority rule tree was constructed. Furthermore we performed the Maximum likelihood estimation of phylogeny using the program ProtML (Adachi and Hasegawa 1992) implemented in Molphy version 2.3 (Adachi and Hasegawa 1996). The relative substitution rate matrix was inferred using the Jones, Taylor, and Thornton (JTT) model (Jones et al. 1992) and local rearrangement search mode was adopted for topology searching.

RT-PCR for Gene Expression Analysis

Total RNA was isolated as described above from each of the mantle, foot, kidney, and hematopancreas tissues of one and the same adult individual of L. stagnalis and each of the mantle, kidney, and hematopancreas tissues of one and the same adult individual of E. brandtii. An amount of 2.5 μg each of total RNA, treated with RQ1 RNase-Free DNase (Promega), was subjected to reverse transcription, catalyzed by ReverTra Ace (Toyobo), and primed with Random Primer (Toyobo). PCR, catalyzed by Ex Taq Hot Start Version (Takara, Otsu, Japan), consisted of an initial step at 94°C for 2 min and 25 cycles (or 30 cycles only for Ebderm1 and 35 cycles only for Ebderm2) at 95°C for 30 s, an appropriate annealing temperature for 1 min, and 72°C for 1 min, followed by 1 min of incubation at 72°C. The following primer pairs were used for amplification of L. stagnalis type-1 dermatopontin (Lsderm-1), L. stagnalis type-2 dermatopontin (Lsderm-2), L. stagnalis type-3 dermatopontin (Lsderm-3), L. stagnalis EF-1α (LsEF-1α), E. brandtii type-1 dermatopontin (Ebderm-1), E. brandtii type-2 dermatopontin (Ebderm-2), and E. brandtii EF-1α (EbEF-1α) gene fragments, respectively (the annealing temperature used and the expected fragment size in brackets): LsS1 and LsA1 [62°C; 191 bp], LsS2 and LsA2 [62°C; 191 bp], LsS3 and LsA3 [57°C; 223 bp], LsEFS and LsEFA [59°C; 193 bp], EbS1 and EbA1 [56°C; 159 bp], EbS2 and EbA2 [46°C; 154 bp], and EbEFS and EbEFA [52°C; 217 bp] (Table 1). Size and amount of RT-PCR products were verified by 1.5% agarose gel electrophoresis.

Real-Time PCR Analysis

The following primer pairs were used for Real-time PCR amplification of Lsderm-1, Lsderm-2, and LsEF-1α gene fragments, respectively: Ls-S1 and Ls-A1, Ls-S2 and Ls-A2, and LsEF-S and LsEF-A (Table 1). We performed real-time PCR using the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems, Foster, CA, USA) with PCR amplification reaction mixtures containing 2 ng of cDNA already prepared for RT-PCR for gene expression analyses, 5 pmol of each primer, and 5 μl of SYBR Green PCR Master Mix (Applied Biosystems) in a total volume of 10 μl. PCR conditions consisted of HotStarTaq DNA polymerase thermal activation at 95°C for 10 min, then 40 cycles of denaturation at 95°C for 15 s and annealing/extension at 60°C for 1 min. Calculation of relative transcript levels was based on the comparative Ct method (Kenneth and Thomas 2001). The relative quantification value of the sample, normalized to the EF-1α gene, is expressed as 2−ΔΔCt, where threshold cycle Ct indicates the fractional PCR cycle number at which the amount of amplified target reaches a fixed threshold, ΔCt = Ct(dermatopontin) – Ct(EF-1α) and ΔΔCt = ΔCt(dermatopontin in the tissue examined) – ΔCt(dermatopontin in the hematopancreas tissue).

Sequence Analysis

Potential glycosylation and phosphorylation sites were analyzed using the prediction servers at the Center for Biological Sequence Analysis, BioCentrum-DTU Technical University of Denmark (CBS prediction servers; http://www.cbs.dtu.dk/).

Results

Amplification and Cloning of cDNA for Molluscan Dermatopontin

In L. stagnalis, fragments of approximately 170 and 150 bp were amplified by the PCR analyses using the primer pairs, DgS1-DgA1 and DgS2-DgA1, respectively. Fragments of approximately 1300, 800, and 900 bp were also amplified by the 3′ RACE procedures using the primer pairs, DgS1-TtA2, DgS2-TtA2, and DgS3-TtA2, respectively. A total of 10 clones were sequenced and the sequences were classified into three types based on the deduced amino acid sequences (Table 2), which we have designated Lsderm1, Lsderm2, and Lsderm3 (Fig. 1). Lsderm1, Lsderm2, and Lsderm3 have 3′-untranslated regions of 393, 354, and 803 bp, respectively. The nucleotide sequences of 3′-untranslated regions were completely different among the three types of L. stagnalis dermatopontin. Because of these differences of length and sequences of the 3′-untranslated regions, as well as the fact that all three types were found from one and the same individual, it is very likely that the three types of L. stagnalis dermatopontin are not alleles but represent paralogous genes resulted from gene duplications.

Table 2 Number of sequenced clones and identified types of Dermatopontin
Full size table

In B. glabrata, a total of 11 clones were sequenced, and the sequences fell into two types with respect to the deduced amino acid sequences (Table 2), which we have designated Bgderm1 and Bgderm2 (Fig. 1). cDNA for Bgderm1 encoded the amino acid sequence of 108 residues long, which completely matched the sequence of amino acid position 41–148 of Bionphalaria glabrata Dermatopontin reported by Marxen et al. (2003). Marxen et al. (2003) could not detect a signal at the amino acid position 44 by Edman degradation, and they inferred mainly from sugar analysis that the undetectable amino acid represented asparagine. We confirmed their inference more directly by sequencing the cDNA. The number of sequenced clones and revealed types of Dermatopontins of the remaining species are also indicated in Table 2.

Phylogenetic Analyses of Molluscan Dermatopontins

To reconstruct the evolutionary relationship of molluscan Dermatopontins, the deduced amino acid sequences have been analyzed using Bayesian and maximum likelihood methods. Dermatopontin sequences have been determined in mammals (Neame et al. 1989; Superti-Furga et al. 1993), an arthropod (Fujii et al. 1992), and a sponge (Schütze et al. 2001) as well as mollusks. When aligned with molluscan Dermatopontins, some amino acid indels were recognized in the mammal and arthropod Dermatopontins but not in the sponge Dermatopontin, at least in the conserved region used for the phylogenetic analyses. We thus carried out the phylogenetic analyses using the unambiguously aligned regions with the sponge Suberites domuncula (Schütze et al. 2001) as outgroup.

The Bayesian tree is presented in Fig. 2. Basommatophoran Dermatopontins (BgDerm1, BgDerm2, LsDerm1, LsDerm2, and LsDerm3) and stylommatophoran Dermatopontins (EbDerm1, EhDerm1, EhDerm2, EaDerm1, EpDerm2, MaDerm1, MaDerm2, and SjDerm2) were recovered as a monophyletic group, respectively. Basommatophoran type-1 Dermatopontins (BgDerm1 and LsDerm1), basommatophoran type-2 Dermatopontins (BgDerm2 and LsDerm2), stylommatophoran type-1 Dermatopontins (EbDerm1, EhDerm1, EaDerm1 and MaDerm1), and stylommatophoran type-2 Dermato pontins (EhDerm2, EpDerm2, MaDerm2, and SjDerm2) were also recovered as a monophyletic group, respectively. Basommatophoran type-3 Dermatopontin (LsDerm3) was placed outside the remaining basommatophoran Dermatopontins. The maximum likelihood inference yielded the same topology as the Bayesian tree.

Figure 2
figure 2

Bayesian and maximum likelihood phylogenetic reconstruction of molluscan Dermatopontins using the amino acid sequence data. Numerals at each node show posterior probabilities estimated by MrBayes and local bootstrap probability values estimated by ProtML. Branch lengths are proportionate to Bayesian estimates produced by MrBayes.

Full size image

Gene Expression Analyses of MolluscanDermatopontin

We performed RT-PCR to analyse expression patterns of dermatopontin in several tissues of adult L. stagnalis and E. brandtii (Fig. 3). In L. stagnalis, Lsderm1 transcript was expressed at an extremely high level only in the mantle tissue. Lsderm2 transcript was expressed at a high level in the kidney tissue rather than in the mantle tissue. Lsderm3 transcript was detected in all four tissues examined, and the transcript was more abundant in the foot and kidney tissues than in the mantle and hematopancreas tissues. LsEF-1α transcript, included as a positive control for cDNA preparations, was expressed at approximately the same level among all four tissues examined. In E. brandtii, Ebderm1 transcript was expressed at an extremely high level only in the mantle tissue. Ebderm2 transcript was detected in all three tissues examined. EbEF-1α transcript, included as a positive control for cDNA preparations, was expressed at approximately the same level between the mantle and the kidney tissues, and at a slightly higher level in the hematopancreas tissue.

Figure 3
figure 3

RT-PCR for analysis of dermatopontin expression in L. stagnalis and E. brandtii. Lanes: M, mantle; F, foot; K, kidney; H, hematopancreas. A housekeeping gene, EF-1α, is includedas a positive control.

Full size image

RT-PCR is a simple and powerful method, but not sufficiently quantitative for the measurement of amounts of the transcripts among tissues. Thus in order to quantify the expression levels of both Lsderm-1 and Lsderm-2 transcripts, we performed real-time PCR analysis for the cDNA prepared from the mantle, foot, kidney, and hematopancreas tissues (Table 3, Fig. 4). The results revealed that Lsderm-1 is almost exclusively expressed in the mantle tissue, while Lsderm-2 is more ubiquitously expressed and is most abundant in the kidney tissue, a pattern which harmonizes with the results of RT-PCR.

Table 3 Relative Quantification using the Comparative Ct Methods for L. stagnate Real-time PCR
Full size table
Figure 4
figure 4

Levels of Lsderm-1 and Lsderm-2 transcripts analyzed by real-time PCR and normalized to the EF-1α transcripts. The levels are expressed as an n-fold difference relative to the levels in the hematopancreas tissues. Error bars represent standard deviations.

Full size image

Deduced Amino Acid Sequences of MolluscanDermatopontins

Molluscan Dermatopontins contain eight cysteine residues, and a sequence motif of E-D-R-X-W/F/Y-X-F/Y/I/L/M-X1-2-C repeats itself three times in the entire sequence (Fig. 1). This sequence motif and the cystein residues are conserved in all Dermatopontin examined so far, whereas Dermatopontin of mammals and L. polyphemus have an additional pair of cysteines (Neame et al. 1989, Fujii et al. 1992, Superti-Furga 1993, Schütze et al. 2001, Marxen et al. 2003). Several potential phosphorylation sites were found in molluscan Dermatopontins, for example, in BgDerm1, a total of nine potential phosphorylation sites are contained (Fig. 1). Potential N-linked glycosylation sites were identified exclusively in Basommatophoran type-1 Dermatopontin (BgDerm1 and LsDerm1; Fig. 1). Potential O-linked glycosylation sites were fonud only in MaDerm1 and SjDerm1 (Fig. 1).

Discussion

Molluscan Dermatopontin as a Shell Matrix Protein

BgDerm1 was originally extracted from shells of B. glabrata to be identified as an N-glycosylated Dermatopontin by Marxen et al. (2003). There has been, therefore, no doubt that BgDerm1 is a shell matrix protein. Lsderm1 must be an orthologue of Bgderm1 according to the results of phylogenetic analysis (Fig. 2) and its transcript is strongly expressed only in the mantle tissue (Fig. 4). Thus LsDerm1 is also highly likely a shell matrix protein. Basommatophoran type-2 and type-3 Dermatopontins (Lsderm2 and Lsderm3) are expressed in all the tissues examined and may therefore have more general functions in the extracellular matrix as in the case of mammals (Forbes et al. 1994). In Stylommatophora, Ebderm1 is expressed only in the mantle tissue (Fig. 3), suggesting that Stylommatophoran type-1 Dermatopontins also represent shell matrix proteins. Unlike the relationship between Bgderm1 and Lsderm1, the Stylommatophoran type-1 dermatopontin, however, is not a direct relative of the Basommatophoran type-1 dermatopontin, because duplication of dermatopontin gene occurred independently in basommatophoran and stylommatophoran lineages as discussed below. Stylommatophoran type-2 Dermatopontin (Ebderm2) are expressed in all tissues examined and may have more general functions like Lsderm2 and Lsderm3.

Functionally Important Regions of Molluscan Dermatopontins

In bovine and L. polyphemus Dermatopontins, the presence of five disulfide bonds has been determined experimentally by digestion with chymotrypsin and thermolysin, respectively (Neame et al. 1989; Fujii et al. 1992). The positions of disulfide bonds are likely to be conserved among L. polyphemus, bovine, and mollusks because of the conserved nature of the cysteine positions in the primary structures of those Dermatopontins (Fig. 1). Four disulfide bonds, therefore, can be assigned to Cys14-Cys40, Cys49-Cys93, Cys66-Cys92, and Cys103-Cys147 in BgDerm1 (Fig. 5a). Referring to bovine and L. polyphemus Dermatopontins (Neame et al. 1989; Fujii et al. 1992), three of the four disulfide bonds probably defined three loop structures (Fig. 5a).

Figure 5
figure 5

Schematic representation of molluscan Dermatopontin. A circle indicates an ainino acid residue. (a) Schematic representation of BgDerm1. Black circles indicate cysteine residues and black bars indicate putative disulfide bonds. Numbers adjacent to the black circles represent amino acid positions of the cysteine residues. Three disulfide bonds in four define three loops shown as Loop-1, Loop-2 and Loop-3. Highly conserved motifs of E-D-R-X-W/F/Y-X-F/Y/I/L/M-X1-2-C, where X is any amino acid, are shown in gray except for the C residues. The potential glycosylation site is shown in gray, (b) Schematic representation of BgDerm1. Double circles indicate amino acid residues conserved among all Basommatophoran type-1 and type-2 Dermatopontins examined (BgDerm1, BgDerm2, LsDerm1 and LsDerm2). Gray circles indicate amino acid residues conserved within Basommatophoran type-1 Dermatopontins (BgDerm1 and LsDerm1) but not between type-1 and type-2 Dermatopontins in Basommatophora. Letters in circles denote one-letter abbreviation of amino acid residues. Black bars indicate putative disulfide bonds.

Full size image

The RGD (Arg-Gly-Asp) motif is a cell attachment site of a large number of adhesive extracellular matrix, blood, and cell surface proteins (Ruoslahti 1996). The EDR (Glu-Asp-Arg) motif, the first part of the conserved sequence motif of E-D-R-X-W/F/ Y-X-F/Y/I/L/M-X1-2-C in invertebrate Dermatopo- ntins, is similar to the RGD motif in that this motif is comprised of three amino acids residues and contains both acidic and basic amino acids. This EDR motif could be functionally important because it is conserved among all invertebrate Dermatopontins examined so far. While only speculative, the EDR motif is possibly a cell attachment site as suggested previously by Fujii et al. (1992).

The presumed disulfide bonds, some of which help to form the loop structures, and the sequence motif of E-D-R-X-W/F/Y-X-F/Y/I/L/M-X1-2-C are conserved in all Dermatopontins including both type-1 and type-2 Dermatopontins in Basommatophora. They must, therefore, be functionally important for all Dermatopontins in general, but not specifically for shell matrix Dermatopontins. In Fig. 5b, double circles indicate amino acid residues conserved among all Basommatophoran type-1 and type-2 Dermatopontins examined (BgDerm1, BgDerm2, LsDerm1, and LsDerm2). Gray circles indicate amino acid residues conserved within Basommatophoran type-1 Dermatopontins (the shell protein BgDerm1 and the likely shell protein LsDerm1) but not between type-1 and type-2 Dermatopontins in Basommatophora. The amino acid residues indicated by gray circles, therefore, are the candidates that are expected to be playing important roles in shell formation. The candidates include a dipeptide NV (Asn-Val), which is predicted to form an acceptor site for N-linked glycosylation together with an adjacent amino acid, Thr (in BgDerm1) or Ser (in LsDerm1). N-linked glycosylation is an important posttranslational modification, and the function is known to include protein folding and cell adhesion (Bhatia and Mukhopadhyay 1999). This N-linked glycosylation seems to be functionally important for the shell matrix Dermatopontins.

Stylommatophoran type-1 Dermatopontins (EbDerm1, EhDerm1, EaDerm1, MaDerm1) that are also considered as shell matrix proteins, however, do not have a potential N-linked glycosylation site, which as discussed above would be functionally important for Basommatophoran shell matrix Dermatopontins (BgDerm1 and LsDerm1). Thus between the two lineages, functions of Dermatopontins in the shell matrices could well be different.

Origin and Evolution of Dermatopontin as a Shell Matrix Protein

As shown in Fig. 2 and summarized in Fig. 6, molluscan dermatopontin genes have experienced at least three gene duplications (Dp1, Dp2, and Dp3 in Fig. 6) after the divergence of basommatophoran and stylommatophoran lineages (node A in Fig. 6). Our results indicated that two dermatopontin gene duplications (Dp1 and Dp2) precede the divergence of B. glabrata and L. stagnalis in basommatophoran lineage (node B in Fig. 6) and that another dermatopontin gene duplication (Dp3) preceded the divergence of Satsuma and Euhadra in stylommatophoran lineage (node C in Fig. 6). The last common ancestor of Biomphalaria and Lymnaea (node B in Fig. 6) is considered to have had three dermatopontin genes, one being a shell matrix protein and the other two being probably with more general functions as noted above. The last common ancestor of Satsuma, Euhadra, and Mandarina (node C) is considered to have had two dermatopontin genes, one being a shell matrix protein and another being probably with more general functions. In contrast, the last common ancestors of basommatophoran and stylommatophoran snails (node A) are considered to have had a single dermatopontin gene. We cannot decide whether the Dermatopontin at nodes A was a shell matrix protein or a protein with more general functions by the information obtained from the studied taxa alone. We can infer it, however, using the information external to the studied taxa and by applying the method of outgroup comparison (Maddison et al. 1984). The ancestors at node A are likely to have had Dermatopontin with general functions as in the mammal and sponge outgroups (Forbes et al. 1994; Schütze et al. 2001). Furthermore, given that the ancestor at nodes A had only one dermatopontin gene, it appears that the Dermatopontin they had must have had more general functions rather than a specialized shell matrix protein, because Dermatopontin itself must have a critical role in extracellular matrices with widespread distributions among tissues and among various animal taxa.

Figure 6
figure 6

Proposed scenario for the evolutionary history of molluscan Dermatopontin. Node A represents the last common ancestor of basommatophoran and stylommatophoran snails. It possessed one dermatopontin gene encoding a protein having a general function. During basommatophoran and stylommatophoran evolution, three gene duplications (Dp1, Dp2, and Dp3) resulted in three Dermatopontin proteins in basommatophoran snails (node C and the members derived from this node) and two Dermatopontin proteins in stylommatophoran snails (node D and the members derived from this node). The dermatopontin gene was co-opted for calcification independently in each lineage.

Full size image

The fact that homologues of dermatopontin have been found in a sponge, a crab, and mammals as well as in pulmonate mollusks suggests that a homologue of dermatopontin was present in the last common ancestor of mollusks. This, however, does not necessarily mean that the last common ancestor of mollusks already had Dermatopontin as a shell matrix protein. Our results suggest that following dermatopontin gene duplications, recruitment for a novel function involved in shell biomineralization occurred after the divergence of basommatophoran and stylommatophoran lineages (node A). This implies a more recent event than late Proterozoic or early Cambrian. In the fossil record, both basommatophoran and stylommatophoran snails first appeared during the Carboniferous, and snails belonging to the Lymnaeidae and Planorbidae, which includes L. stagnalis and B. glabrata, respectively, first appeared during the Jurassic (Tracey et al. 1993). It is thus plausible that the dermatopontin gene duplications in the basommatophoran lineage occurred between Carboniferous and Jurassic. Moreover, snails belonging to the Camaenidae, which includes S. japonica, and the Bradybaenidae, which embraces Euhadra, first appeared during Cretaceous and Tertiary, respectively (Tracey et al. 1993). Thus it is likely that the dermatopontin gene duplication in the stylommatophoran lineage occurred between Carboniferous and Tertiary. Although we would never exclude the possibility that some genes were co-opted for molluscan calcification at the time of the “Cambrian explosion” (Marin and Luquet 2004), the dermatopontin gene in particular was probably co-opted for molluscan calcification at least twice independently in more recent evolutionary time. This in turn suggests that, although mollusks have a number of extracellular matrix proteins, the potential candidates for shell matrix proteins might not have been so many. Dermatopontin was perhaps one of the few available proteins from which the functions involved in shell calcification could have evolved. We will need a systematic sampling of mollusks including marine species to further understand the mode of evolution of the shell matrix protein Dermatopontin, or whether or not co-option of dermatopontin for the function involved in shell calcification is common in mollusks. We report here evolutionary convergence of Dermatopontin in shell calcification. The precise functions of Dermatopontins, however, may well be different between basommatophoran and stylommatophoran lineages as discussed above. Origins and evolution of molluscan shell matrix proteins might be more complicated than we have expected.