Glassin, a histidine-rich protein from the siliceous skeletal system of the marine sponge Euplectella, directs silica polycondensation

Significance

Hexactinellid sponges of the genus Euplectella produce highly ordered and mechanically robust skeletal systems of amorphous hydrated silica. The high damage tolerance of their constituent skeletal elements and the environmentally benign conditions under which these sponges form have prompted additional investigations into the characterization of the proteins driving the synthesis of these materials. In the present report, we describe a previously unidentified protein, named “glassin,” extracted from the demineralized skeletal elements of Euplectella. Glassin is a histidine-, aspartic acid-, threonine-, and proline-rich protein and directs silica polycondensation at neutral pH and room temperature.

Abstract

The hexactinellids are a diverse group of predominantly deep sea sponges that synthesize elaborate fibrous skeletal systems of amorphous hydrated silica. As a representative example, members of the genus Euplectella have proved to be useful model systems for investigating structure–function relationships in these hierarchically ordered siliceous network-like composites. Despite recent advances in understanding the mechanistic origins of damage tolerance in these complex skeletal systems, the details of their synthesis have remained largely unexplored. Here, we describe a previously unidentified protein, named “glassin,” the main constituent in the water-soluble fraction of the demineralized skeletal elements of Euplectella. When combined with silicic acid solutions, glassin rapidly accelerates silica polycondensation over a pH range of 6–8. Glassin is characterized by high histidine content, and cDNA sequence analysis reveals that glassin shares no significant similarity with any other known proteins. The deduced amino acid sequence reveals that glassin consists of two similar histidine-rich domains and a connecting domain. Each of the histidine-rich domains is composed of three segments: an amino-terminal histidine and aspartic acid-rich sequence, a proline-rich sequence in the middle, and a histidine and threonine-rich sequence at the carboxyl terminus. Histidine always forms HX or HHX repeats, in which most of X positions are occupied by glycine, aspartic acid, or threonine. Recombinant glassin reproduces the silica precipitation activity observed in the native proteins. The highly modular composition of glassin, composed of imidazole, acidic, and hydroxyl residues, favors silica polycondensation and provides insights into the molecular mechanisms of skeletal formation in hexactinellid sponges.

The hexactinellids are a circumglobal group of predominantly deep sea sponges. Exhibiting a diverse array of morphologies, the skeletal systems of hexactinellids, which are composed of amorphous hydrated silica (the other silica-forming sponge class is the Demospongiae), range in structural complexity from loose aggregations of individual skeletal elements (spicules) to complex, hierarchically ordered lattices. Hexactinellids have evolved the ability to colonize either rocky substrates or soft sediments and exhibit remarkable skeletal modifications to accomplish these feats (1). For example, the sediment-dwelling hexactinellids produce bundles of long fibrillar anchor spicules that form robust holdfast structures. These anchor spicules exhibit surprising flexibility and damage tolerance and have served as useful model systems for investigating high performance silica-based organic-inorganic biological composites (2–4). In one representative genus, Euplectella (Fig. 1), the constituent spicules are assembled into a highly regular cylindrical lattice that exhibits multiple levels of structural hierarchy spanning from the nanoscale to the macroscale. The lattice is composed of two interpenetrating grid-like networks of nonplanar cross-like spicules that are reinforced with diagonal spicule bundles and consolidated by a laminated silica cement. Each of the constituent spicules contains a central axial filament of protein that directs the spicule's core geometry, and the main load-bearing spicules exhibit a laminated architecture consisting of concentric silica cylinders separated by thin organic interlayers (5).

Fig. 1.

The hexactinellid sponge Euplectella. (A and B) Dried silica skeleton of E. aspergillum (A) and live specimen of E. curvistellata (B). The fibrous holdfast structure of the live specimen is missing, which is likely a result of damage incurred during collection. (Scale bars: 1 cm.)

Recent investigations with individual hexactinellid spicules have demonstrated excellent mechanical performance (1, 2), exceeding those of man-made glass rods with similar dimensions, and exhibit intriguing optical wave-guiding properties (6). Although the ultrastructural and mechanical properties of spicules from hexactinellids like Euplectella have been well documented, characterization of the macromolecular components of the spicules and their roles in spicule formation has largely remained unexplored. Identification and characterization of these occluded organic molecules is an initial step in understanding the molecular mechanisms of spicule formation, and furthermore, the lessons learned from such studies could be applied toward the development of new synthesis routes to silica-based composite materials under environmentally benign conditions (7).

Previously, with Morse and colleagues, we demonstrated that the organic axial filaments from the siliceous spicules of the demosponge Tethya aurantia are composed of a group of closely related proteins called “silicateins” (8). The silicateins share high sequence similarity with cathepsin L and other members of the papain-like family of cysteine proteases and the larger superfamily of catalytic triad-containing hydrolases. In vitro, silicateins promote the hydrolysis and polycondensation of silicon alkoxides to yield silica at neutral pH through a similar mechanism to that of proteases in the hydrolysis of peptide bonds and the more general mechanism of the hydrolases (9–11). Because these initial early discoveries, silicateins have been identified in numerous other demosponge species (12–18).

More recently, silicatein-like proteins were also identified in the hexactinellid sponges. Müller et al. (19–21) reported that antibodies against silicatein α from the demosponge Suberites domuncula cross-reacted with a 27-kDa protein in the giant anchor spicules of Monorhaphis chuni and with a 24-kDa protein in the spicules of Crateromorpha meyeri. Mass spectrometric analysis of a fragment from the M. chuni 27-kDa protein demonstrated that it contained sequences corresponding to the silicateins from demosponges (22). Silicatein sequences deduced from cDNA clones from M. chuni and C. meyeri contain the characteristic catalytic amino acid triad, serine-histidine-asparagine (19, 21). The silicatein gene was also identified from Aulosaccus schulzei, although here, a catalytic cysteine was confirmed instead of the usual serine (23). A partial silicatein cDNA from Euplectella aspergillum has also been archived [(GenBank accession no. FR748156; Müller et al. (2011)], although a detailed description is not yet available.

In addition to the silicateins, other organic molecules proposed to be involved in spicule formation have been identified from hexactinellid sponges. Travis et al. (24) reported that the demineralized skeletons of Euplectella sp. contained collagen, cellulose-like filaments, hexosamine-containing organics, and proteins containing relatively large amounts of aspartic acid, glutamic acid, glycine and histidine. Ehrlich et al. (25) isolated a fibrous material from the anchoring spicules of the hexactinellid sponge Hyalonema sieboldi, and demonstrated that the material was composed of collagen with an unusual glycine-(3-hydroxyproline)-(4-hydroxyproline) motif, which in vitro, exhibited silica polycondensation and templating activities. Ehrlich and Worch (26) also reported chitin in hexactinellid sponges including Farrea occa and E. aspergillum as an organic component of their silicious skeletal systems.

Inspired by these previous studies, we investigated the spicule-associated proteins from the mineralized skeletal framework of Euplectella. From these analyses, we describe a previously unidentified silica-occluded protein, named “glassin,” as the main constituent in the water-soluble fraction extracted from demineralized spicules, and through a series of in vitro assays, investigate its potential role in biogenic silica formation.

Results

Following Euplectella skeletal dissolution with buffered hydrofluoric acid, the resulting organic material was centrifuged for separation into water-soluble and insoluble fractions. When 2 g of silica skeletal material (dry weight) was used as starting material, the water-soluble fraction yielded 0.2 mg of protein.

When the water-soluble protein fraction was dissolved in 100 mM sodium phosphate buffer at pH 6.0 and mixed with a freshly prepared metastable silicic acid solution, a silica precipitate could be collected by centrifugation, while even after 30 min, no silica was formed in the protein-free control under otherwise identical experimental conditions. The amount of precipitate formed was directly proportional to protein concentration (Fig. 2A) with silica precipitation occurring at pH values ranging from 6.0 to 8.0, with the maximum product yield occurring at pH 7.0 (Fig. 2B). Little or no precipitate was formed below pH 5.0 and heat denaturation of the protein solution did not affect the activity. No significant precipitate formation was observed when tetraethoxysilane (which requires hydrolysis before condensation) was used as a silica precursor. This latter finding is significant, because tetraethoxysilane and related conjugates of silicon and heavy metals are readily hydrolyzed by silicatein to form the corresponding silica, silsesquioxanes, or metal oxides (9, 11).

Fig. 2.

Silica precipitation activity by the water-soluble fraction obtained from the demineralized skeletons of Euplectella. (A) Silica yield increases as a function of protein concentrations. (B) Effect of pH on silica precipitation. In the presence of the water-soluble fraction (solid squares), precipitates were formed at pH values ranging from 6.0 to 8.0, with the maximum production of silica at pH 7.0. Little or no precipitate formed below pH 5.0, and no precipitate was observed at any pH in the absence of the water-soluble fraction (open squares). Heat denaturation of the protein solution did not affect the activity (a solid circle at pH 6.0). Each reported value is the average of three independent experiments and in all cases, the SD was less than 10% of the mean. (C) Scanning electron micrograph of the precipitated silica which exhibits a distinctly granular substructure. (D and E) Transmission electron micrographs of silica in the reaction mixtures incubated for 1 min (D) and 5 min (E). Silica nanoparticles form network-like aggregates and the size of the particles increases over time. (Scale bars in C–E: 50 nm.)

The silica precipitates formed by the water-soluble protein fraction were collected via centrifugation and observed by scanning electron microscopy. The resulting silica product exhibited a fine granular texture composed of silica nanoparticles measuring 20–30 nm in diameter, a morphology typical of silica formed via sol-gel processes (Fig. 2C). In addition, the process of silica polycondensation in the reaction mixture was followed using transmission electron microscopy, which revealed that the silica nanoparticles formed network-like aggregates and that the size of the particles increased over time (Fig. 2 D and E).

The proteins in the water-soluble fraction isolated from E. aspergillum were analyzed by SDS-containing polyacrylamide gel electrophoresis (SDS/PAGE (Fig. 3A). After staining with Coomassie Brilliant Blue R250 (CBB), the main constituent protein exhibited an apparent molecular mass of 23 kDa (Fig. 3A, lane 2), which we named “glassin.” Similar results were also obtained when the water-soluble fraction prepared from E. curvistellata was analyzed (Fig. 3A, lane 3). In contrast, the SDS-extractable protein yield from the insoluble fraction following silica demineralization was low and, as a result, was not further investigated as part of the present study.

Fig. 3.

SDS/PAGE analysis of the Euplectella skeletal proteins. (A) The water-soluble fractions (1 μg) were subjected to SDS/PAGE, and the proteins were visualized with CBB staining: lane 1, protein standards with molecular masses indicated at the left margin; lane 2, the water-soluble fraction from a skeleton of E. aspergillum; lane 3, the water-soluble fraction from a skeleton of E. curvistellata. An open triangle points to the dominant constituents of the water-soluble fractions. (B) The water-soluble fraction was incubated with peptide-N-glycosidase F or O-glycosidase and then analyzed by SDS/PAGE. Lane 1, protein molecular weight standards; lane 2, no glycosidase (control); lane 3, peptide-N-glycosidase; lane 4, O-glycosidase. The mobility of glassin increased following incubation with O-glycosidase. Incubation with peptide-N-glycosidase F did not significantly change the mobility of glassin. The open triangle indicates the position of intact glassin and the filled triangle indicates glassin after O-glycosidase treatment. Lettered arrows indicate peptide-N-glycosidase F (P) and O-glycosidase (O).

Protein posttranslational modification, specifically glycosylation, was tested with peptide-N-glycosidase and O-glycosidase, which remove N-linked and O-linked carbohydrate moieties linked to proteins at the amines of asparagine and hydroxyls of serine/threonine, respectively. After enzymatic treatment, the samples were subjected to SDS/PAGE (Fig. 3B). The electrophoretic mobility of glassin increased following treatment with O-glycosidase but not peptide-N-glycosidase F, suggesting that glassin is O-glycosylated.

Amino acid analysis of the water-soluble protein fraction from demineralized Euplectella skeletons (which is primarily glassin) is shown in Table 1. For comparison, the calculated amino acid composition from the sequence of the mature peptide of T. aurantia silicatein α (8) and glassin (as described below) are also included. The high compositional similarity between glassin and the water soluble protein fraction extracted from demineralized Euplectella skeletons combined with the relatively simple electrophoretic profile revealed from SDS/PAGE indicates that glassin is the dominant water-soluble spicule-associated protein from Euplectella.

Table 1.

Percentage of amino acid composition from the Euplectella skeleton soluble-protein fraction, Euplectella glassin, and T. aurantia silicatein α

Amino acid	Soluble fraction, %	Glassin, %*	Silicatein α, %*
Aspartic acid	10.34	11.54	11.21
Threonine	11.26	11.23	3.74
Serine	4.81	3.74	12.15
Glutamic acid	2.66	2.34	8.88
Glycine	8.16	7.96	12.62
Alanine	5.02	4.22	10.28
Valine	3.88	2.34	7.01
1/2-Cystine	0.00	0.00	2.80
Methionine	0.00	0.00	2.34
Isoleucine	1.04	0.94	4.21
Leucine	2.49	2.34	4.67
Tyrosine	0.60	0.47	7.94
Phenylalanine	0.55	0.94	2.34
Lysine	4.48	3.74	3.74
Histidine	30.17	35.56	0.93
Arginine	1.05	0.94	2.80
Proline	13.49	11.70	2.34
Total	100.00	100.00	100.00

^*Calculated from the amino acid sequences of Euplectella glassin (Fig. 4A) and T. aurantia silicatein α mature peptide (8).

The most striking compositional feature of glassin is its high histidine content, at more than 30% of the total amino acids. As confirmed through site-directed mutagenesis studies (10), histidine is an essential active site amino acid in the silica-precipitating activity of silicatein α, in which the histidine content is only 1%. Proline and threonine are also abundant in glassin, at 13% and 11%, respectively. Hydroxyls play important roles in the induction of polycondensation by silicatein α (9, 10, 27), which contains 16% hydroxy amino acids (which is similar to glassin). However, threonine is dominant in glassin (11% threonine and 5% serine), whereas serine is rich in silicatein α (12% serine and 4% threonine). Cysteine is not found in glassin, although cysteine forms disulfide bonds that constrain the 3D structure of silicatein α and represents 3% of their total amino acid content. The content of hydrophobic residues (alanine, valine, leucine, isoleucine, and phenylalanine) is also very low in glassin, in contrast to the high hydrophobic amino acid content in silicatein α. These results demonstrate that the amino acid composition of the Euplectella glassin is significantly different from that of silicatein α, suggesting that the two proteins are not closely related.

Partial amino acid sequences of glassin were analyzed with a peptide sequencer and the amino-terminal peptide was identified as R(H or P)GHHGHH. A fragment generated by digestion with trypsin had the sequence HDHHDHHHDHAPPXPPXVPP, where X could not be identified. The sequence should follow a lysine or arginine (trypsin cleavage sites), so glassin should contain the peptide (K or R)HDHHDHHHDHAPPXPPXVPP. Both the amino-terminal and trypsin-derived peptide sequences of glassin from freshly collected live specimens were identical to those of glassin derived from commercially available dry skeletons. These sequences were used to design degenerate oligonucleotide primers for amplification of the corresponding cDNA by reverse transcription–PCR, starting with RNA obtained from a live specimen. We obtained a 1338-bp cDNA clone, RT-PCR 1, encoding the 3′ region of the gene (Fig. S1). Then, rapid amplification of cDNA ends (RACE) was performed to obtain upstream sequence information. 5′-RACE produced three clones, 5′-RACE 1/clones 1–3, all of which contained one potential amino-terminal sequence of glassin, RPGHHGHH; 5′-RACE 1/clone 2 contained a stop codon in the coding region of RT-PCR 1. A second round of 5′-RACE was performed with gene-specific primers using sequence information from 5′-RACE 1/clones 1–3. The second round clone, 5′-RACE 2/clone 1, covered the other potential amino-terminal sequence, RHGHHGHH. When RT-PCR was performed with primers encoding the region upstream of RHGHHGHH and the region downstream of the stop codon of RT-PCR 1, a product encoding the entire sequence could not be obtained. Instead, a clone, RT-PCR 2, was obtained by RT-PCR with primers encoding upstream of RHGHHGHH and downstream of the stop codon of 5′-RACE 1/clone 2. The sequence of clone RT-PCR 2 was consistent with that of 5′-RACE 1/clone 2.

Fig. S1.

Schematic of glassin cDNA cloning. First, a 1,338-bp cDNA fragment (RT-PCR 1) was cloned by RT-PCR; 5′-RACE was then performed to obtain cDNA coding the upstream regions. The first 5′-RACE produced 3 clones (5′-RACE 1/clones 1–3), of which clone 2 contained a stop codon in the coding region of RT-PCR 1. The second 5′-RACE product (5′-RACE 2/clone 1) covered the amino-terminal sequence obtained from peptide sequencing. RT-PCR was then performed to obtain clones covering the entire coding sequence, but only one clone (RT-PCR 2) was obtained, and its sequence was similar to that of 5′-RACE 1/clone 2. The open boxes indicate primers for 5′RACE; the filled black boxes, oligo dT adaptor primers; the filled gray boxes, gene specific primers.

Fig. 4A shows the amino acid sequence of glassin deduced from clone RT-PCR 2. Both of the two candidates for the N-terminal peptide sequence, RHGHHGHH and RPGHHGHH, are contained in the deduced sequence, RHGHHGHH being located before RPGHHGHH and represents the N-terminal sequence of the mature glassin polypeptide. The sequence also contains KHDHHDHHHSHAPPSPPTVPP. The 10th amino acid was identified as aspartic acid by peptide sequencing, but as serine in the sequence deduced from the gene. Unidentified amino acids in the tryptic fragment are assigned as serine and threonine. Misidentified or unidentified amino acids in peptide sequencing were likely attributable to posttranslational modifications of serine and threonine as suggested by the SDS/PAGE experiment showing an apparent molecular weight reduction following O-glycosidase treatment of glassin (Fig. 3B).

Fig. 4.

Amino acid sequence of glassin. (A) The entire amino acid sequence of glassin deduced from the DNA sequence of clone RT-PCR 2 (Fig. S1). The sequence consists of 233 aa, abundant in histidine (cyan), aspartic acid (magenta), proline (tangerine), and threonine (green). The amino acid sequences with lines below the letters that match information obtained from direct peptide sequencing of the native glassin. The solid lines indicate potential amino terminal sequences R(H or P)GHHGHH. The sequence indicated by a dashed line corresponds to that obtained from tryptic peptides. (B) Repeat sequences in glassin. The glassin sequence contains two repeat sequences, amino acids 19–109 and 132–227. The sequences 19–65 of the first repeat and 132–183 of the second repeat, indicated by arrows marked HD-rich, are characterized by alternating histidine(s) and aspartic acid or lysine. The sequences 67–82 of the first repeat and 185–200 of the second repeat, indicated by arrows marked P-rich, are abundant in proline. The sequences 83–109 of the first repeat and 201–227 of the second repeat, indicated by arrows marked HT-rich, are characterized by repeats of histidine and threonine. Identical amino acids between the first and the second repeats are shaded. (C) Schematic of the domain structure of glassin. HD-rich, P-rich, and HT-rich domains are represented by blue, yellow, and green boxes. The open box represents the domain connecting the first and second repeats.

The sequence from the amino-terminal arginine to the stop codon contains 215 aa, and its molecular mass is calculated as 23 kDa, matching the apparent molecular mass of the protein observed by SDS/PAGE. The sequence is characterized by abundant histidine, reaching 35% of the total amino acids. Proline, threonine, and aspartic acid are also rich in the sequence (12%, 11%, and 10%, respectively), consistent with the amino acid analysis results.

The histidines always form H(G/D/H/S/A/T/P/N) or HH(G/D/T) repeats. The overall sequence consists of two histidine-rich repetitive sequences (91 and 96 aa long, respectively) connected by a 22-aa linker sequence. Alignment of the two repeated sequences reveals that 90% of the corresponding amino acids are identical (Fig. 4B). Each repetitive sequence can be divided into three domains: (i) a histidine and aspartic acid (HD)-rich domain, (ii) a histidine and threonine (HT)-rich domain, and (iii) a proline (P)-rich domain (Fig. 4C). In the HD-rich domain, a doublet HD and a triplet HHD is often seen and HHD is sometimes replaced by HHG or HGK. In the HT-rich domain, HTHATVP is repeated and followed by HTHATHT. The connecting sequence between the two repeats consists of acidic and hydrophobic amino acids with no regularity. Protein database analysis reveals that glassin shares no significant similarity with any known proteins. All of the above mentioned sequences have been deposited in the European Molecular Biology Laboratory database, the GenBank database, and the DNA Data Bank of Japan. Accession numbers are as follows: RT-PCR 1, accession no. LC012024; 5′-RACE 1/clone 1, accession no. LC012025; 5′-RACE 1/clone 2, accession no. LC012026; 5′-RACE 1/clone 3, accession no. LC012027; 5′-RACE 2/clone 1, accession no. LC012028; and RT-PCR 2, accession no. LC010923.

Recombinant glassin containing a His tag at the amino terminus was expressed in Escherichia coli, and purified using a HisGraviTrap column from cells extracted with Fast Break Cell Lysis reagent (Fig. S2). Expression and purification of recombinant glassin was confirmed by SDS/PAGE and Western blot analyses with an anti-glassin antibody (Fig. 5 A and B). Recombinant glassin (29 kDa as calculated from the deduced amino acid sequence) exhibited an apparent molecular mass of 35 kDa in SDS/PAGE and Western blots. When recombinant glassin was added to a metastable solution of silicic acid, silica promptly precipitated, as was seen with native glassin (Fig. 5C). These results demonstrate that the protein sequence of the recombinant glassin is sufficient to accelerate silica precipitation and that glycosylation is not required for the observed activity.

Fig. S2.

Schematic of the gene for recombinant glassin production. Glassin cDNA encoding 215 aa with molecular mass of 24,461 Da was inserted into the pET100 vector, which carries a T7 promoter (P_T7), a lac operator gene (lacO), a ribosome binding site (RBS), a codon for the start methionine, a 6 × His tag, and an Xpress epitope, in this order from 5′ to 3′. Accordingly, transformed bacteria produce a recombinant glassin with a molecular mass of 28,587 Da, containing the His tag and the Xpress epitope at the amino terminus. An asterisk indicates the stop codon derived from the glassin gene.

Fig. 5.

Production of recombinant glassin and its silica precipitating activity. (A) SDS/PAGE analysis of recombinant proteins. Each lane contained 10 μL of samples as follows: lane 1, bacterial lysate; lane 2, proteins that did not bind to the HisGraviTrap column; lane 3, the first fraction of proteins eluted from the column with 500 mM imidazole; lane 4, the second eluted fraction; and lane 5, the third eluted fraction. (B) Western blot analysis with a rabbit anti-glassin antibody. The contents of samples were the same as in A. (C) Silica precipitation in the presence of recombinant glassin. Silica precipitates were obtained by mixing metastable silicic acid with purified recombinant glassin [R, the second fraction (lane 4) in A and B] or the native water-soluble fraction from demineralized skeletal material from Euplectella (N). The experiments were performed at pH 6.0, and product yields were determined by the molybdate blue assay.

Discussion

In the present study, we identified a water-soluble protein, glassin, extracted from the demineralized siliceous skeleton of Euplectella. Glassin is a protein characterized by alternating histidine and aspartic acid or threonine sequences, shares no significant sequence similarity with any other known proteins, and accelerates silica polycondensation from silicic acid at near neutral pH.

Although glassin is the dominant constituent protein in the water soluble fraction from Euplectella skeletal material, we also obtained water-insoluble materials, which were not characterized as part of the present study due to their low yield of SDS-extractable proteins. Previously, the presence of collagen was suggested in this water-insoluble fraction based on amino acid analysis (24), whereas chitin was identified as the dominant nonprotein organic component (26). Further detailed characterization of this water-insoluble fraction and its effect on the induction of silica polycondensation when combined with glassin may provide new insights into the roles of these specific components in hexactinellid sponge spicule formation.

Glassin is quite different from silicatein α, which we and our colleagues previously identified from the siliceous spicules of the demosponge, T. aurantia (8, 9). The silicateins are assembled into a linear fiber (28, 29) that is found within the core of demosponge spicules and catalyze silica formation via a hydrolysis pathway from organically functionalized silicic acid precursors at neutral pH (8, 9, 11). In addition to their discovery in several other demosponge species (12–18), silicateins or silicatein-like proteins have also been reported from hexactinellid sponges such as M. chuni and C. meyeri (19–21). If silicatein exists in Euplectella, as is suggested based on partial cDNA data [GenBank accession no. FR748156; Müller et al. (2011)], then the two proteins (glassin and silicatein) may either act cooperatively during silica biosynthesis in hexactinellids or may play distinct functional roles. For example, silicatein may be involved in establishing spicule symmetry and primary mineralization and glassin may be associated with secondary spicule cementation and subsequent skeletal consolidation.

It is noteworthy that the band of native silica-purified glassin is broad and diffusive in SDS/PAGE, whereas the recombinant glassin exhibits a very well defined band. This could be explained by the fact that the silica-purified glassin may be a mixture of closely related polypeptides, an idea supported by the observation that glycine and proline can be substituted in the native glassin at position 2. Silafins, the group of polypeptides directing silica formation in diatoms, are also a mixture of related polypeptides, producing a similarly broad band pattern in SDS/PAGE (30).

The high histidine content in a protein involved in biogenic silica formation is not surprising considering that artificial histidine-rich peptides have often been used as catalysts for silica polycondensation. For example, 20-kDa histidine homopolypeptides and even much smaller decahistidine homopolypeptides possess the ability to form silica at near neutral pH from either organically substituted silanes or solutions of metastable silicic acid (31, 32). In addition, site-directed mutagenesis showed that the active site histidine and serine in silicatein α play essential roles in the in vitro hydrolysis and polycondensation of silicon alkoxides (10), and subsequent studies confirmed that the coexistence of hydroxyl or acidic amino acid residues with histidine residues increases their catalytic activity in silica polycondensation (32). Phage display selection of silica-precipitating peptide sequences independently confirmed that the presence of hydroxyl- and imidazole-containing amino acids were critical for the induction of silica polycondensation (33). Kuno et al. (32) recently demonstrated that a diblock copolypeptide of histidine and aspartic acid (H₅D₅) and an alternating polypeptide of histidine and aspartic acid ((HD)₅) were more effective in hydrolysis of trimethylethoxysilane than histidine homopolypeptide, and that (HD)₅ was more effective than H₅D₅ because of a “charge relay effect.” Accordingly, alternating histidine and aspartic acid/threonine motifs in glassin may play a similar functional role in its observed activity.

The induction of silica polycondensation from silicic acid is not a simple process because supersaturated silicic acid solution contains various silicic acid species and particles of different sizes, each of which behaves differently (34). As a result, the precise mechanism of action of glassin has yet to be fully elucidated. Perry and coworkers proposed that the imidazole group may accelerate silicic acid condensation from this medium by forming hydrogen bonds with silicic acid, increasing silica precipitation activity through electrostatic interaction with oligomeric (poly)silicic acids (35). Hydrogen bond formation and electrostatic interaction between imidazole groups in glassin and specific silicic acid species may be modified by adjacent acidic (aspartic acid) or hydroxyl (threonine) residues, as predicted through computational simulation (36).

The glassin protein exhibits a distinctive structural hierarchy. First, histidine residues always form doublets (HX) or triplets (HHX) with amino acids including aspartic acid, glycine, and threonine. Second, these doublets and triplets form two distinct histidine-rich domains, the HD and HT domains, which are separated by the P-domain. Third, the series of HD, P, and HT domains forms a tandem repeat in the overall glassin molecule.

Previous studies of other biosilica associated proteins like the silicateins and silafins (an unrelated group of proteins shown by Kroger and colleagues to mediate silica formation from silicic acid in diatoms) have shown that they require strict 3D structures or complex posttranslational modifications for their observed activities (9, 30). Glassin, in contrast, is thermally stable (which may be related to its largely random coil architecture revealed by Chou–Fasman secondary structure prediction) and does not require posttranslational modification for the induction of silica polycondensation.

In conclusion, we have identified a protein, glassin, with remarkably high histidine content in the siliceous skeleton of Euplectella, and have demonstrated its capacity to direct silica precipitation from metastable silicic acid solutions at neutral pH. Future in vivo studies investigating the localization and expression profiles of glassin in living sponges may ultimately provide critical insight into the processes of spicule synthesis and consolidation in Euplectella, and the development of environmentally benign biomimetic synthesis routes to silica and other inorganic materials.

Materials and Methods

Samples.

Dried skeletons of the hexactinellid sponge E. aspergillum were of Philippine origin and acquired from commercial sources (Fig. 1A). The top and bottom thirds of each skeleton were removed, so that the specimens consisted entirely of skeletal tubes, but not the sieve plate or holdfast apparatus. Live specimens of Euplectella curvistellata (37) were collected by beam trawling at a depth of 236 m at 32°30 N, 129°10 E in the East China Sea on March 4, 2012, during an expedition (Expedition 343) on the research vessel Nagasaki Maru of Nagasaki University (Fig. 1B).

Protein Extraction.

Dried skeletal material was soaked in a 5% (vol/vol) solution of sodium hypochlorite overnight to remove any residual external organic matter, rinsed with Milli-Q (Merck Millipore) purified water five times, and air dried at room temperature. The resulting skeletal material was treated with 2 M HF/8 M NH₄F to dissolve the silica and the remaining solution was dialyzed against 10 volumes of Milli-Q water at 4 °C for more than 4 h per dialysis, and the outer solution was changed seven times. The dialysate was centrifuged at 10,000 × g for 20 min at 4 °C to separate the supernatant from the insoluble material and the supernatant was concentrated with Amicon Ultra-15 Ultracel (nominal molecular weight limit, M_r 3,000) Centrifugal Filter Units (Merck Millipore).

Protein Analysis.

Protein concentration was determined using a DC protein assay kit (Bio-Rad), with γ-globulin as the standard, according to the manufacturer’s instructions. SDS/PAGE was performed on the soluble protein fraction from demineralized skeletons using NOVEX 10% Bis-Tris gels in the NuPAGE system (Thermo Fisher Scientific) with Mes buffer. Samples were dissolved in 1 × LDS sample buffer (Thermo Fisher Scientific) containing 1% DTT. After electrophoresis, the gels were stained with CBB (EzStain Aqua; ATTO) for protein visualization. Molecular masses of the proteins were estimated using Novex Sharp Prestained Protein Standards (Thermo Fisher Scientific). Protein glycosylation was analyzed enzymatically: 1 μg of protein from the water soluble fraction was incubated with 500 units of Peptide-N-Glycosidase F (New England Biolabs) to determine the existence of asparagine-linked carbohydrates (O-linked), or 40,000 units of O-glycosidase to determine the existence of serine/threonine-linked (O-linked) carbohydrates (New England Biolabs), for 4 h at 37 °C. After incubation, the proteins were analyzed by SDS/PAGE as described above. For amino acid analyses, the soluble fraction was hydrolyzed in 6 M HCl at 110 °C for 22 h under vacuum and analyzed with an automated amino acid analyzer (l-8500; Hitachi). Amino-terminal and internal peptide sequences were analyzed as follows. The water-soluble fraction with and without trypsin (Promega) treatment were separated by SDS/PAGE, and then electrophoretically transferred to polyvinylidene difluoride (PVDF) (ATTO) membranes. After CBB staining, the PVDF membranes containing the proteins of interest were processed on an automated Edman-degradation protein sequencer (PPSQ31A; Shimadzu).

Silica Precipitation Assay.

The assay follows a previous reported protocol (29). A solution of orthosilicic acid was freshly prepared by dissolving tetramethoxysilane in 1 mM HCl to a final concentration of 1 M. Proteins to be tested for silica precipitation activity were dissolved in 100 mM sodium phosphate buffer (pH 4–8) to a final volume of 10 μL. Subsequently, 1 μL of the 1 M orthosilicic acid solution was added and the samples were incubated for 5 min at room temperature. The samples were centrifuged for 5 min at 14,000 × g and the pellets were washed three times with Milli-Q water to remove free silicic acid and phosphate. Washed pellets were incubated in 10 μL of 0.1 M NaOH at 95 °C for 30 min to dissolve the precipitated silica. Concentrations of silicic acid were determined by the molybdate blue method (34). For scanning electron microscopy studies, washed silica precipitates were placed directly on aluminum pin mounts and air-dried at room temperature. The specimens were briefly sputter-coated with platinum/palladium and imaged with a field emission scanning electron microscope (SU8020; HITACHI, Tokyo, Japan). To monitor the silica formation process, the reaction mixture was spotted on Formvar-coated copper grids and after 1 or 5 min incubation, the grid was rinsed with Milli-Q water twice, blotted on filter paper to remove excess water, and air-dried at room temperature. The resulting reaction products were observed via transmission electron microscopy (JEM-1400; JEOL).

Cloning of cDNA.

After collection of living sponges, the cellular material was immediately immersed into TRIZOL solution (Thermo Fisher Scientific). Extracted RNA (2 μg) was converted to cDNA with the SuperScript preamplification system (Thermo Fisher Scientific) according to the manufacturer’s instructions, except that the oligo dT-3 sites adaptor primer (Takara Bio) was used instead of oligo dT₂₀. The degenerate oligonucleotides 5′-AR CAY GAY CAY CAY GAY CAY C-3′ and 5′-CAY CAY CAY GAY CAY GCN C-3′ were synthesized from the partial protein sequences KHDHHDHH and HHHDHAP, respectively, which were used for subsequent PCR amplification. PCR products containing the 5′ regions of the glassin cDNA were obtained by RACE-PCR (38). PCR products were ligated into pCR4 vector (Thermo Fisher Scientific), and both strands of the cloned inserts were sequenced. The Chou-Fasman secondary structure prediction was performed with GENETYX-MAC Network Version 17.0.4 (GENETYX).

Recombinant Protein Expression.

Glassin cDNA encoding the mature protein region was amplified by PCR with Platinum Pfx DNA polymerase (Thermo Fisher Scientific) with the primers 5′-CGT CAC GGT CAC CAT GGT C-3′ and 5′-TCA GGA AAG AGA CCA GGT GAT G-3′ and was subcloned into pET100 (Thermo Fisher Scientific). BL2 Star (DE3) cells (Thermo Fisher Scientific) were transformed with the plasmid and cultured according to the manufacturer’s protocol. Recombinant protein expression was induced by 0.1 mM isopropyl β-D-1-thiogalactopyranoside. After incubation for 4 h, the cells were harvested and then solubilized in Fastbreak cell lysis reagent (Promega). The recombinant protein was purified using a His tag protein affinity column (HisGraviTrap; GE). Expression and purification of recombinant glassin was confirmed by Western blot analysis using a WesternBreeze immunodetection kit (Life Technologies) and a rabbit anti-glassin antibody that was raised with a synthetic peptide HTHPLPPHTHATVPHTHA from the glassin sequence.