Expression of silicatein and collagen genes in the marine sponge Suberites domuncula is controlled by silicate and myotrophin


W. E. G. Müller, Institut für Physiologische Chemie, Abteilung für Angewandte Molekularbiologie, Universität Mainz, Duesbergweg 6, D-55099 Mainz, Germany. Tel.: + 49 6131 3925910, Fax:. + 49 6131 3925243, E-mail: WMUELLER@mail.UNI-Mainz.DE


The major skeletal elements in the (Porifera) sponges, are spicules formed from inorganic material. The spicules in the Demospongiae class are composed of hydrated, amorphous silica. Recently an enzyme, silicatein, which polymerizes alkoxide substrates to silica was described from the sponge Tethya aurantia. In the present study the cDNA encoding silicatein was isolated from the sponge Suberites domuncula. The deduced polypeptide comprises 331 amino acids and has a calculated size of Mr 36 306. This cDNA was used as a probe to study the potential role of silicate on the expression of the silicatein gene. For these studies, primmorphs, a special form of aggregates composed of proliferating cells, have been used. It was found that after increasing the concentration of soluble silicate in the seawater medium from around 1 µm to approximately 60 µm, this gene is strongly upregulated. Without additional silicate only a very weak expression could be measured. Because silica as well as collagen are required for the formation of spicules, the expression of the gene encoding collagen was measured in parallel. It was also found that the level of transcripts for collagen strongly increases in the presence of 60 µm soluble silicate. In addition, it is demonstrated that the expression of collagen is also upregulated in those primmorphs which were treated with recombinant myotrophin obtained from the same sponge. Myotrophin, however, had no effect on the expression of silicatein. From these data we conclude that silicate influences the expression of the enzyme silicatein and also the expression of collagen, (via the mediator myotrophin).



In living organisms four major groups of biominerals exist: ion compounds, which are restricted to Prokaryota; calcium phosphates, found in Metazoa; calcium carbonates, used by Prokaryota, Protozoa, Plantae, Fungi and Metazoa; and silica (opal) in sponges (Porifera), Plantae and a few Protozoa and Protophyta [1]. The element silicon [Si] contributes up to about 28% of the earth's crust and is, after oxygen, the second most abundant element on earth [2].

The transition from Protozoa to Metazoa is dated 600–1000 million years ago [3,4]. It has been proposed that oxygen level, temperature, and seawater chemistry played a major role in the evolution from Protozoa to Metazoa [5]. In particular, it was proposed that during the period of appearance of sponges the ocean was rich in sodium carbonate at the expense of sodium chloride [6]. It was assumed that such a ‘soda ocean’ had a pH of above 9. Most silicates are hardly soluble in water at neutral pH, whereas the soluble portion increases with alkalization. Therefore, it seems likely that the concentration of soluble silicate was considerably higher in seawater than it is today. These higher concentrations might have favored the formation of silica in the sponge spicules.

With the exception of the class of Calcarea, which possess spicules formed from calcium carbonate, the spicules of sponges grouped into the Demospongiae and Hexactinellida classes are composed of hydrated, amorphous silica [7]. Based on morphological criteria the phylogenetic relationship of Calcarea, Demospongiae and Hexactinellida is under dispute; in one hypothesis the Calcarea are grouped together with the Demospongiae [8] while in a second the Demospongiae are more closely related to the Hexactinellida [9]. Using informative proteins, deduced from cDNAs, from the three sponge groups it became evident that the Hexactinellida are phylogenetically older than the Demospongiae, while the Calcarea evolved later [10,11]. This finding implies that during evolution the calcareous spicules were formed independently from the siliceous spicules [12].

The secretion of spicules in Demospongiae occurs in intracellularly specialized cells, the sclerocytes, where silica is deposited around an organic filament [13,14]. It should be noted at this point that silica is not only deposited in spicules, but also as ribbons in the nucleus [15]. The latter observation is interesting insofar as siliceous spikes show a variable size and shape suggesting that silica in those structures is subjected to a high turnover. If the formation of siliceous spicules is inhibited the sponge body collapses [16]. The synthesis of spicules is a rapid process; in Ephydatia fluviatilis a 100-µm long megasclere is formed within 40 h [17].

Inhibition studies revealed that skeletogenesis of siliceous spicules is enzyme-mediated [18,19]. The siliceous spicules contain a definite axial filament and their synthesis is genetically controlled [20]. In Demospongiae the spicules are glued together by a collagenous ‘cement’ made of microfibrils [13,21]. Experimental evidence suggested that deposition of silica particles induces biological reactions, especially that of collagen fibrillogenesis [22].

It is interesting to note that the silicification machinery in sponges, like in diatoms, is effectively inhibited by the element germanium [Ge][23–25] and a detailed study has been performed with the sponge Suberites domuncula[26]. It was reported that at a Ge/Si ratio of 1 hatching of gemmules is inhibited in this species. Moreover, spicule formation is inhibited at a Ge/Si ratio of 0.1 [26]. Germanium is also a constituent of seawater, but only at low concentrations of about 1 nm[27].

A major step towards elucidating the formation of siliceous spicules at the molecular level was the finding that the formerly termed ‘axial organic filament’ of siliceous spicules is in reality an enzyme, silicatein, which mediates the apposition of amorphous silica and hence the formation of spicules [28,29]. These studies have been performed with the demosponge T. aurantia.

In the present study it is shown that in the S. domuncula system the gene of silicatein is strictly dependent on extracellular soluble silicate. In addition, we revealed that the gene coding for collagen, a polymer which is required for the formation of the functional skeleton in sponges [7], is also upregulated in the presence of silicate. Furthermore experimental data are given which suggest that for the collagen expression myotrophin is required as an intermediate factor in the induction cascade.

Materials and methods

Chemicals and enzymes

The sources of chemicals and enzymes used were given previously [30,31]. In addition, sodium hexafluorosilicate, germanium (IV) oxide, and tetraethoxysilane (tetraethyl-orthosilicate) were obtained from Aldrich (Deisenhofen, Germany), natural Ca2+ and Mg2+-containing seawater from Sigma (Deisenhofen, Germany). DIG (digoxigenin) DNA labeling kit, DIG-11-dUTP, anti-DIG AP Fab fragments, CDP-Star (disodium 2-chloro-5-(4-methoxyspiro[1,2-dioxetane-3,2′-(5′-chloro)-tricyclo(,7)decan]-4-yl)phenyl phosphate) came from Boehringer Mannheim (Mannheim, Germany) and the ‘Spectroquant’ test for soluble silicate from Merck (Darmstadt, Germany).

Determination of silicate concentration

For determination of the water soluble portion of silicate (silicic acid) a ready to use kit ‘Spectroquant’ (Merck) was used. To adjust distinct concentrations of silicate sodium hexafluorosilicate was added.


Live specimens of S. domuncula (Porifera, Demospongiae, Hadromerida) were collected by SCUBA diving near Rovinj (Croatia) from depths between 20 and 35 m. The sponges were brought, under constant aeration and cooling to 17 °C, by car to Mainz (Germany) and there kept in 10 dm3 tanks at 17 °C under continuous aeration for more than one month before use in the experiments.


The myotrophin-like polypeptide from the sponge S. domuncula was produced in Escherichia coli[32]. Briefly, the cDNA (accession no. AJ252240) with a potential open reading frame of 360 nucleotides was isolated from the cDNA library. It was expressed as recombinant oligohistidine-rMYO fusion protein. The protein was purified by affinity chromatography using Ni–nitriloacetic acid–agarose resin. The resulting recombinant protein preparation was found to be almost completely pure. Its size, 13.8 kDa, corresponded to the calculated Mr.

Dissociation of cells and formation of primmorphs

The procedure was applied as described [33]. Primmorphs, the special form of aggregates, are formed from single cells after transferring them from Ca2+- and Mg2+-free artificial seawater, used for dissociation of cells, into natural seawater. This medium is supplemented with antibiotics and 0.1% (v/v) of Marine broth 2216 (Difco) [33]. Starting from single cells primmorphs of at least 1 mm in diameter, with an average of 2–5 mm are formed after two days. After 5 days, primmorphs were used for the experiments.

Where indicated the primmorphs were inspected with an Olympus AHBT3 microscope.

Incubation studies with primmorphs

Primmorphs, 5 days after reaggregation of single cells, were taken and incubated for different periods of time after addition of sodium hexafluorosilicate at a final silicate concentration of 60 µm. In one series of experiments the silicone alkoxide tetraethoxysilane was used at a concentration of 60 µm. In the incubation experiments with germanium (as GeO2), the final concentration of the compound was likewise 60 µm. Recombinant myotrophin was added to the primmorphs at a concentration of 1 µg·mL−1. Inspection of the primmorphs was performed at a magnification of 100-fold.

Library screening and isolation of the S. domuncula silicatein cDNA

The complete sponge cDNA, encoding the silicatein SILICA_SUBDO, termed SUBDOSILICA, was isolated from the S. domuncula cDNA library [30] by PCR [34]. The degenerate reverse primer, directed against the characteristic hydroxy amino-acid cluster of T. aurantia silicatein ([28]; amino acids 264 to 271; Fig. 1) 5′-GAA/GCAICGIGAIGAA/GTCA/GTAIAC-3′ was used in conjunction with the 5′-end vector-specific primer. The PCR was carried out at an initial denaturation at 95 °C for 3 min, followed by 30 amplification cycles at 95 °C for 30 s, 55 °C for 45 s, 74 °C for 1.5 min, and a final extension step at 74 °C for 10 min. The reaction mixture was as described earlier [35]. The longest fragment of ≈ 850 bp was used to isolate the cDNA from the library [34]. The longest insert obtained had a size of 1200 nucleotides [excluding the poly(A) tail]. The clone was termed SUBDOSILICA and was sequenced using an automatic DNA sequenator (Li-Cor 4200).

Figure 1.

Nucleotide sequence of the cloned silicatein cDNA, SUBDOSILICA and deduced amino-acid sequence SILICA_SUBDO. The nucleotides are numbered in the 5′ to 3′ direction, starting with nucleotide triplet encoding the start methionine; the putative start methionine is underlined and the stop codon is marked (★); the location of the primer is also underlined.

Isolation of the S. domuncula cDNA encoding the collagen-like polypeptide

The complete cDNA, SUBDOCOLL2, encoding the putative collagen-like polypeptide COLL2_SUBDO, was isolated from the cDNA library by screening with digoxygenin-11-dUTP labeled DNA-probes (DIG random primed DNA labeling kit, Boehringer Mannheim) SUBDOCOL1 (accession no. AJ252241[32]) from the same species [34]. Screening of the library was performed under low stringency hybridization conditions of plaque lifts from 3 × 105 p.f.u. on nitrocellulose. Filters were hybridized at 37 °C overnight in 35% formamide, 5 × NaCl/Cit, 0.02% SDS, 0.1% N-laurylsarcosine and 1% blocking reagent (Boehringer Mannheim) containing 10 ng·mL−1 of the DIG labeled probes. Filters were washed twice in 2 × NaCl/Cit, 0.1% SDS (5 min, room temperature), followed by two additional washes in 0.1 × NaCl/Cit, 0.1% SDS (15 min, 42 °C). Again, the positive clones were detected with an alkaline phosphatase conjugated antidigoxygenin antibody using 5-bromo-4-chloroindol-2-yl phosphate/Nitro Blue tetrazolium as substrate [36]. The cDNA encoding the collagen-like polypeptide COLL2_SUBDO was 1082 nucleotides long.

Sequence analysis

The sequences were analyzed using the computer program blast[37]. Multiple alignments were performed with clustal w Ver. 1.6 [38], and the graphic presentations were prepared with GeneDoc [39]. The graphical output of the bootstrap figure was produced by the program treeview (R. D. M. Page, University of Glasgow, UK;

Hydropathy plot was generated by the soap program [40]. The hydropathy values were calculated according to [41].

Northern blotting

RNA was extracted from liquid nitrogen pulverized sponge tissue with TRIzol Reagent. Five micrograms of total RNA was subjected to electrophoresis through 1% formaldehyde/agarose gel and blotted onto Hybond N+ membrane following the manufacturer's instructions (Amersham; Little Chalfont, UK) [35]. Hybridization was performed with the complete 1.2-kb SUBDOSILICA probe or the total cDNA insert of the S. domuncula collagen SUBDOCOL1 sequence (accession no. AJ252241). For the quantification of the Northern blot signals the chemiluminescence procedure was applied [42] and CDP-Star was used as substrate. The screen was scanned with a GS-525 Molecular Imager (Bio-Rad; Hercules, CA).


PCR cloning and sequencing of the cDNA encoding S. domuncula silicatein

The complete S. domuncula cDNA, SUBDOSILICA, was isolated and characterized. The 1169-bp long sequence has an open reading frame of 993 nucleotides (Fig. 1). Northern blot analysis revealed a single band of 1.4 kb, indicating that the clone is of full length (see below). The predicted translation product of 330 amino acids, named SILICA_SUBDO (Fig. 2A) has a calculated size of Mr 36 306 and a pI of 5.95 [40].

Figure 2.

.Silicatein polypeptide. (A) The deduced amino-acid sequence from the S. domuncula cDNA, SUBDOSILICA, termed SILICA_SUBDO, is aligned with homologous sequence from T. aurantia (SILICA_TETHYA; accession no. AAC23951) as well as the most similar protein the cathepsin l-like protease precursor from Artemia franciscana (CATH_ARTEMIA; AAD39513) and the sponge (G. cydonium) cathepsin (CATH_GEOCY, Y10527). Residues conserved (similar or related with respect to their similar physico-chemical properties) in all sequences are shown in white on black and those in at least two sequences in black on gray. The characteristic sites within the silicatein sequences are indicated: the three aa Cys, His and Asn form the catalytic triad (CT), the 17-residues long signal peptide (SP), the processing site for the conversion of the proenzyme to the mature enzyme, the cysteine residues potentially involved in the three putative disulfide bonds (●), as well as the serine cluster (Ser). (B) Phylogenetic relationships of the sponge silicatein sequences from S. domuncula (SILICA_SUBDO) and T. aurantia (SILICA_TETHYA), with cathepsin L sequences: from the fungus Dictyostelium discoideum (Dictyostelium; X03344) and the sponge G. cydonium (Geodia; Y10527); the invertebrates Sarcophaga peregrina (Sarcophaga; D16533) and A. franciscana (Artemia); and from the vertebrates Homo sapiens (Human; X12451), Mus musculus (Mouse; M20495) and Rattus norvegicus (Rat; Y00697). The two cathepsin related sequences from the Protozoa Tetrahymena thermophila (Tetrahymena; L03212) and Paramecium tetraurelia (Paramecium; X91754) were used as outgroup. The rooted tree was computed by neighbour-joining and distance matrix determinations as described under Experimental procedures. Scale bar indicates an evolutionary distance of 0.1 amino-acid substitutions per position in the sequence.

As already published [28], silicatein is a new member of the cathepsin L subfamily. The S. domuncula silicatein shares 70% identical and 79% similar (identical plus physico-chemical related amino acids) amino acids with the T. aurantia silicatein polypeptide; while an alignment with the corresponding Geodia cydonium cathepsin protein [43] shows an identity/similarity of 43%/59% (Fig. 2A). The three amino acids Cys, His and Asn which form the catalytic triad of cysteine proteases [44], are present in the sponge cathepsin at the characteristic sites, Cys125, His164 and Asn184 [43]. However, in both sponge silicatein sequences the cysteine residue is replaced by serine. In the sponge silicatein sequence the eukaryotic thiol (cysteine) proteases histidine active site is present within the amino acids 275 to 285 region and reads, LNHAMVVTGYG. The eukaryotic thiol (cysteine) proteases asparagine active site is found at amino acids 292 to 311, YWLAKNSWGTNWGNSGYVMM [45]. It is established that for the formation of active cathepsins processing is required either by autolysis [46] or by a second protease [47]. Based on comparisons with cathepsins, the characteristic cleavage site is also present in the sponge silicatein sequences, suggesting a processing of the proenzyme of a Mr of 36 306 to the mature form with an Mr of 23 125. Also the signal peptide is found in silicatein (Fig. 2A). The location of three putative disulfide bonds, known from other cysteine proteases [48] also exist in the sponge silicatein polypeptide between amino acids 135 to 178, 169 to 211 and 270 to 329, Fig. 2A.

In addition, the silicatein sequence shows two potential transmembrane helices, spanning from amino acids 2 to 17 and amino acids 135 to 150, respectively [40,49]. The transmembrane helices have been computed and the sum of the values for the ‘buried-helix parameter’ is plotted (Fig. 3A). In contrast to the silicatein sequence the cathepsin polypeptide does not contain any putative transmembrane region (Fig. 3B). As reported earlier for the T. aurantia silicatein, the cluster comprising the characteristic hydroxy amino acid (serine) is also present in the corresponding S. domuncula moleculeFig. 2A.

Figure 3.

Prediction of the potential transmembrane regions. The two deduced polypeptides, the silicatein from S. domuncula (A) as well as the cathepsin from G. cydonium (B), have been analyzed for potential transmembrane alpha helices [40]. The x-axis shows the number of aa of the respective sequence and the y-axis the values for the ‘buried-helix parameter’ (BHP). The stretches of the two potential transmembrane regions are indicated (TM-1 and TM-2).

Phylogenetic analysis

The databank search with the deduced S. domuncula silicatein revealed greatest similarity with the T. aurantia silicatein and with the cathepsin L sequences. A rooted phylogenetic tree was constructed. The relationship showed that the sponge sequences form the basis of the cathepsin L molecules, both from fungus Dictyostelium discoideum and the sponge G. cydonium. More distantly related are the cathepsins from Sarcophaga peregrina and Artemia franciscana and from the vertebrates Homo sapiens, Mus musculus and Rattus norvegicus (Fig. 2B). The two cathepsin related sequences from the Protozoa Tetrahymena thermophila and Paramecium tetraurelia form the outgroup. This rooted tree suggests that the split between the sponge silicateins and the cathepsins occurred early in evolution, very likely during the transition of the Protozoa to the Metazoa.

Expression of silicatein and collagen in primmorphs treated with silicate

The concentration of silicate in the natural seawater used for the experiments was determined to be ≈ 1 µm. Under this condition the expression of silicatein in primmorphs is low and almost undetectable, as measured by Northern blot experiments (Fig. 4A). If the concentration of silicate in the seawater was enhanced to a concentration of 60 µm, then the expression of silicatein strongly increased after an incubation period of 1 day. After 5 days the expression reached a maximum (Fig. 4A), longer incubations for 10 days resulted in a decrease of the expression level by ≈ 25%. The concentration of 60 µm silicate is the value which was previously found to be optimal for the spicule formation in S. domuncula[26].

Figure 4.

Level of transcripts for silicatein and collagen in primmorphs from S. domuncula. Primmorphs formed after 5 days from single cells were incubated for 0 (controls; Con) to 5 days either in the absence of exogenous silicate [minus (–) silicate; open bars] or presence of 60 µm sodium silicate [plus (+) silicate; closed bars]. Subsequently, RNA was extracted and 5 µg of total RNA was size separated; after blot transfer hybridization was performed either with the silicatein probe, SUBDOSILICA (A), or the collagen probe, SUBDOCOL1 (B), both isolated from S. domuncula. The intensities of the bands were quantified, as described under Materials and methods. The relative degree of expression was correlated with that seen for the maximal expression (after 5 days in the presence of silicate).

Recently the gene encoding collagen was isolated from S. domuncula[32]. Because this polypeptide is required for the formation of the functional skeleton in sponges [50], it was plausible to determine if the expression of this gene is also influenced by silicate. Again, Northern blot experiments were performed which revealed a drastic increase in the expression of the collagen gene (Fig. 4B). At the natural silicate concentration of the seawater used (≈ 1 µm) the collagen expression was relatively low and amounted to ≈ 15% of that seen in the presence of 60 µm exogenous silicate. The silicone alkoxide tetraethoxysilane, added at a concentration of 60 µm, caused no effect on collagen expression.

Spicule formation in primmorphs

In a semiquantitative approach the effect of silicate on the spicule formation in vitro was determined. If primmorphs, formed from single cells (Fig. 5B) after 5 days, were incubated further in the absence of exogenous silicate for 5 days no spicules, (neither the tylostyle with the ball-shapes head nor the needle-shaped tip) (Fig. 5A), could be seen by microscopic inspection (Fig. 5C). However, if the primmorphs were incubated in the presence of 60 µm silicate bundles of spicules are present (Fig. 5D).

Figure 5.

Formation of spicules in primmorphs. Primmorphs were obtained from single cells (B); five days after reaggregation the primmorphs were used for the experiments. They remained either in the absence of exogenous silicate for additional 5 days (C), or were incubated with 60 µm sodium silicate for the same period of time (D). In the sodium silicate treated primmorph (D) spicules can be observed. In (A) the tylostyles, which are characteristic for S. domuncula are shown. Magnification: A 500×; B 100×; C and D 300×.

Influence of myotrophin on the expression of silicatein

In a recent study it could be established that myotrophin from S. domuncula has the potency to induce collagen in the primmorph system [32]. This result is confirmed here (Fig. 6A). Recombinant myotrophin, rMYO, was added to the primmorphs at a concentration of 1 µg·mL−1. After incubation for 1–3 days the level of collagen expression significantly increased; after 5 days the level of collagen expression without myotrophin is only 12% of that seen in the presence of myotrophin (Fig. 6A). In contrast, this agent has no measurable influence on the expression of the silicatein gene during the same period of incubation time (Fig. 6B).

Figure 6.

Effect of recombinant myotrophin on the expression of collagen and silicatein in primmorphs. Primmorphs were incubated for 0 (controls) up to 5 days in the absence, (–) myotrophin; open bars, or presence of 1 µg·mL−1 of myotrophin, (+) myotrophin; closed bars. Then Northern blotting was performed with SUBDOCOL1 to determine the expression of collagen (A) or with the SUBDOSILICA probe to determine the level of expression of silicatein (B). Further details are given in the materials and methods section.

Effect of germanium on the expression of silicatein

Primmorphs were incubated with germanium for 5 days at a concentration of 60 µm. Under such conditions no expression of the gene encoding silicatein could be detected, as in the controls. If exogenous silicate was added at a concentration of 60 µm the strong upregulation of the expression was measured again. If exogenous silicate (60 µm) was added together with an equimolar concentration of germanium, the expression of silicatein was the same as that seen in the absence of germanium (data not shown).

The S. domuncula collagen-like polypeptide

Collagen is known to be the major structural protein in invertebrates in general and in sponges in particular [7,51]. Based on biochemical and electron microscopical data it was suggested that in sponges collagen is either dispersed as thin fibrils in the intercellular matrix or organized as bundles, termed spongin, in the mesohyl [51]. In order to clarify further, if more than one type of collagen gene is present in sponges, the cDNA library of S. domuncula was screened with the collagen probe SUBDOCOL1 from the same species.

Only one additional sequence which is different from SUBDOCOL1 could be identified. The SUBDOCOL1–like cDNA was termed SUBDOCOLL2. The 1082-nucleotide-long sequence encodes a deduced related molecule comprising 295 amino acids in length, the calculated Mr is 32 977. It was named collagen-like polypeptide COLL2_SUBDO, as the characteristic G-X-Y collagen triplets are missing (Fig. 7). The collagen-like polypeptide shares high similarity to the C-terminal segment of the full-length sponge collagen COL1_SUBDO; in this region 55% of the amino acids are of similar physico-chemcial properties. While COL1_SUBDO comprises three segments: the noncollagenous N-terminal domain (NC1); the collagenous internal domain (COL) with 24 G-X-Y collagen triplets; and the noncollagenous C-terminal domain (NC2), the new collagen-like polypeptide shows no characteristic signatures found in other collagens.

Figure 7.

The S. domuncula collagen-like polypeptide. The new, collagen related molecule, collagen-like polypeptide (COLL2_SUBDO), is compared with the collagen molecule COL1_SUBDO, recently identified. The collagen molecule shows the noncollagenous N-terminal domain (NC1), the collagenous internal domain (COL) with the collagen triplets (★) and the noncollagenous C-terminal domain (NC2). In contrast, the collagen-like polypeptide comprises only a noncollagenous stretch (NC). Similar amino-acid residues are shown in white on black.

Based on this result it appears to be less likely that a second collagen gene exists in sponges.


The formation of siliceous spicules in sponges is genetically controlled. The data presented here demonstrate that under suitable silicate concentration of ≈ 60 µm, silicate induces the genes encoding collagen and silicatein. Germanium, an element chemically related to silicon shows no such effect. However, the gene induction of silicatein appears to be a complex process, as the morphogen myotrophin induces only the collagen but not the silicatein gene. This finding supports earlier observations indicating that the formation of collagen fibrils can proceed independently of spicule formation as well as closely connected with this process [7]. Collagen fibrils are synthesized in several cell types [7] with the lophocytes as the most prominent cells [22]. In addition, spongocytes have been identified which secrete spongin fibers that contribute to the synthesis of the horny skeleton [22]. According to the chemical analysis, spongin should be considered as a collagen-like molecule [52]. As only one gene coding for collagen has been identified in sponges it remains open if spongin is only a processed form of collagen.

In contrast to the formation of collagen fibrils disconnected from the synthesis of spicules, those collagen polypeptides which are formed in connection with the silicification are produced in sclerocytes and exopinacocytes [17]. The sclerocytes secrete the axial filaments, while the exopinacocytes secrete the collagen-like spongin, which functions as an organic sheath around the spicules [53,54]. Collagen is required for the formation of the functional skeleton in sponges [50]. In this study it is reported that in newly formed tissue the spicules are not entrapped by these fibrils, while in older regions the spicules are encased in the organic matrix.

The elegant work by the group of Morse clarified that the axial filaments around which silica is deposited are composed of the enzyme silicatein [29]. In the present study the silicatein cDNA was isolated from the sponge S. domuncula and found to be closely related to the T. aurantia silicatein. As reported earlier [28] silicatein is closely related to the enzyme cathepsin L. The phylogenetic analysis of the two sponge silicateins shows that they branched off from a common ancestor with the cathepsin L molecules prior to the appearance of sponges. This finding could imply that animals which evolved earlier than sponges already contained silicatein. As silicatein has not been identified in other metazoan phyla it has to be concluded that this gene has been lost during the evolution from sponges to the invertebrate and vertebrate phyla.

Recently it has been shown that the recombinant silicatein catalyzes the reaction of tetraethoxysilane to silica and silicone [29]. As in the present study the induction of silicatein gene expression could only be observed with silicate but not with tetraethoxysilane it might be postulated that the organic silica substrate is formed intracellularly from silicate.

The data presented here indicate that together with the induction of the gene expression for the axial filament/silicatein which leads to the initial process of spicule formation, the collagen gene is expressed in parallel and, in a second cell [54]. Based on inhibition studies it could be demonstrated that the formation of siliceous spicules can occur in the absence of the synthesis of such collagen polypeptides which form the perispicular spongin sheath [18]. This finding suggests that the expression of collagen is not directly connected with the formation of the siliceous spicules.

In a second set of data it is shown that the myotrophin-related polypeptide causes an upregulation of collagen but not of silicatein. Therefore, we postulate at present that silicate causes, in addition to the expression of silicatein, the expression of myotrophin. Myotrophin is assumed to be released by the sclerocytes and to cause the expression of spongin/collagen in exopinacocytes (Fig. 8).

Figure 8.

Postulated formation of functional skeleton in siliceous sponges. Silicate is assumed to cause in sclerocytes the expression of silicatein and myotrophin. While silicatein catalyzes the apposition of amorphous silica under the formation of spicules within those cells, myotrophin might be released from sclerocytes and triggers exopinacocytes for the production of the collagen-like spongin. Spicules and collagen are cemented together under formation of the functional skeleton.

In an approach to clarify if sponges contain more than one collagen cDNA species the library was screened with the previously isolated clone, SUBDOCOL1 cDNA [32]. Until now, only one additional cDNA has been isolated. It comprises great similarity at the noncollagenous segment, but the characteristic collagen G-X-Y triplets are absent. Therefore, this molecule was termed collagen-like polypeptide, SUBDOCOLL2. This piece of evidence supports the assumption that the sponge collagens are not highly diverse as known from vertebrates [55]. In addition, this finding strengthens the notion that spongin might be a processed form of collagen.

Germanium is a known inhibitor of spicule formation [25], see introduction. Here it is shown that this element has (a) no effect on silicatein expression and (b) does not interfere with the silicate-induced silicatein expression. From this finding it might be concluded that germanium inhibits silicification on the level of silicatein-mediated enzymic silica deposition.

The result reported here demonstrates that the formation of siliceous spicules is controlled by extracellular silicate which results in an upregulation of the enzyme silicatein which forms the amorphous silica and of collagen which is likely to be involved in the formation of the collagenous sheath. Evidence is given which shows that the latter step involves the mediation of myotrophin, a potential morphogen recently cloned from S. domuncula[32]. Studies are now in progress to elucidate the detailed metabolism of silicon in sponge cells.


We thank Ms. Renate Steffen for technical assistance. This work was supported by grants from the Bundesministerium für Bildung und Forschung (Project ‘Cell Culture of Sponges’), the European Commission (Marine Science and Technology Programme “SPONGE”) and the International Human Frontier Science Program (RG-333/96-M).


  1. Note: the sequences of silicatein and collagen-like polypeptide reported here are deposited in the EMBL/GenBank data base under accession nos AJ272013 and accession no. AJ272012.