Introduction

It is now well established that all metazoan phyla, including sponges (phylum Porifera), evolved from one common ancestor, the Urmetazoa (Müller 2001). Data from molecular biological studies with the demosponge Suberites domuncula revealed that sponges have characteristic cell–cell and cell–matrix adhesion molecules, such as integrins (Pancer et al. 1997; Müller 1997) and collagens (Exposito et al. 1991; Schröder et al. 2000), in common with the “higher” metazoan phyla. The arrangement of cells within a sponge organism is genetically controlled by the same evolutionary conserved genes found also in higher metazoans; homeodomain proteins (reviewed by Wiens et al. 2003) and T-box transcription factors (Adell et al. 2003) have been also found to be responsible for the bodyplan formation in Porifera. The sponge body is surrounded by a pinacoderm, which has been described as an epithelium-like layer of cells; however, these cells are not connected by cell junctions according to the literature (Mackie 1984; reviewed by Nielsen 2001). The apical cell contacts observed in the ciliated cells of sponge larvae have been looked upon as cell junctions, but not related to metazoan cell junctions (Rieger 1994). “Septate” junctions, similar to those found in some invertebrate taxa (see Knust 2000), were visualized in some sponge species between choanocytes and sclerocytes (Ledger 1975). Until now, no gene presumably involved in the formation of cell junctions has been identified in sponges.

The knowledge on the degree of differentiation of the pinacoderm, the sealing cell layer in sponges, is crucial for its classification as a “porous” cell layer or as epithelium. Three forms of pinacoderm layers are distinguished in Calcarea and Demospongiae; the exopinacoderm facing the external milieu, the endopinacoderm surrounding the aquiferous canal system, and the basopinacoderm, which attaches the animal to the substratum (Garrone 1978). The first molecular evidence indicating that in some sponges a basement membrane exists was the discovery of type IV collagen in the demosponge Pseudocorticium jarrei (Boute et al. 1996). A confirmation that the pinacoderm is really composed of cells with a polarized phenotype in an asymmetric distribution of organelles and their polarized orientation of the cytoskeleton (Knust 2000) can be expected from the isolation of genes controlling this pattern. Polarization of epithelial cells occurs at the tight junctions by the assembly of protein scaffolds that cluster transmembrane proteins and anchor them to the underlying cytoskeleton (see: Yeaman et al. 1999). Especially in vertebrates tight junctions regulate the passage of water and ions as well as of organic molecules through the paracellular pathway.

Among the dominant transmembrane proteins of the tight junctions, the receptors occludin, claudin, and other junctional adhesion molecules should be mentioned; they are linked to characteristic classes of scaffold molecules, the intracellular effector molecules/receptors (Yeaman et al. 1999). Occludin and claudin form the backbone of the tight junction strands; like tetraspanin they have four transmembrane regions (González-Mariscal et al. 2003). The scaffold molecules are subdivided into proteins containing the PDZ motif and others that lack this motif (González-Mariscal et al. 2003). Among the PDZ-containing proteins, the role of the membrane-associated guanylate kinases (MAGUK) in the function of tight junctions is well understood (González-Mariscal et al. 2000). A homologue of a MAGUK protein, a zonula occludens homologue, was identified in the Cnidaria Hydra vulgaris (Fei et al. 2000). It is well established that Cnidaria possess functional epithelia that define distinct tissue layers (Sarras et al. 1991); therefore tight junction proteins might be predicted. In sponges, where the existence of true epithelia is still under dispute (see: Pechenik 2000), the proof of the existence of scaffold proteins characteristic for tight/cell junctions would change the view on the quality of the pinacoderm layer as epithelial layer. Since no approach to identify potential junctions in S. domuncula has been performed microscopically, we term the linkages between the cells operationally cell junctions.

In the present study the successful screening for a gene encoding a cell junction scaffold protein from a sponge, S. domuncula, is reported. After domain analyses of the 15,000 ESTs in our database from S. domuncula (Chevreux et al. 2004) a cell junction scaffold protein has been identified. The gene encodes the scaffold protein membrane-associated guanylate kinase with inverted arrangement (MAGI). MAGI proteins are characterized by a unique arrangement of the functional domains: six PDZ domains, two WW domains, and a truncated guanylate kinase motif that, in contrast to other MAGUK proteins, is located at the NH2-terminus rather than at the COOH-terminus (Dobrosotskaya et al. 1997). MAGI proteins function as scaffold molecules, linking the membrane-associated receptors, the signaling cortical proteins, and the actin cytoskeleton (Shoji et al. 2000; Dobrosotskaya and James 2000). Some isoforms from alternative splicing have been also found to localize in the nucleus, which supports their role in the intracellular communication pathways (Dobrosotskaya et al. 1997; Dobrosotskaya and James 2000). In this report, in situ hybridization studies demonstrate that the MAGI gene from S. domuncula, termed SDMAGI, is highly expressed at the surface of the animal and in cells of the pinacoderm lining the canals.

In addition we describe the existence of one tetraspan receptor, tetraspanin (SDTM4SF), in S. domuncula; this molecule has been described earlier (Müller et al. 1999b). The tetraspanins are a huge superfamily, with more than 37 members described alone in Drosophila melanogaster and numerous more in other organisms. Together with occludin and claudin, they belong to a group of hydrophobic proteins with four transmembrane domains and a series of conserved aa residues in the extracellular loops (Yunta and Lazo 2003; González-Mariscal et al. 2003). Tetraspanin forms complexes with other cell surface receptors, e.g., integrin, and is involved in regulation of proliferation and migration of cells (Tiwari-Woodruff et al. 2001).

The regeneration capacity of sponge tissue is high. After ablation of tissue from the body surface a series of genes is upregulated, i.e., the integrin receptor (Wimmer et al. 1999) and the allograft inflammatory factor in S. domuncula (Kruse et al. 1999). The primmorph system from sponges, which represents special 3D-cell aggregates composed of proliferating and differentiating cells, has successfully been applied to investigate the role of cell–cell and cell–matrix adhesion molecules as well as of transcription factors (Wiens et al. 2003; Perović et al. 2003). After approximately 5 days in culture the aggregates/primmorphs are surrounded by a surface layer composed of cells reminiscent of pinacocytes (Müller et al. 1999a). Expression of SDMAGI (encoding MAGI scaffold protein) as well as of SDTM4SF (S. domuncula tetraspanin [Müller et al. 1999b]) is highly upregulated after regeneration of the epithelial layer of the sponge and after formation of the pinacoderm layer in primmorphs.

Materials and Methods

Chemicals and Enzymes

The sources of chemicals and enzymes used were given previously (Kruse et al. 1997; Krasko et al. 2000).

Sponges

Specimens of S. domuncula (Demospongia, Hadromerida, Suberitidae) were collected from the Adriatic Sea, close to Rovinj (Croatia), and then kept in aquaria in Mainz for more than 3 years, as described by Le Pennec et al. (2003). For the regeneration studies the 2-mm-thick surface layer, including the pinacoderm, was ablated from the test animal. Then this surface sample and tissue from the interior of the specimen were taken and used for the Northern blot experiments.

Formation of Primmorphs

The procedure described for the formation of primmorphs from single cells was applied (Müller et al. 1999a). Starting from single cells, primmorphs of approximately 5 mm formed after 5 days. They were cultivated in natural seawater, supplemented with 0.2% of RPMI-1640 medium and with the optimal concentration of 60 μM silicate and 30 μM Fe3+ (added as ferric citrate) (Krasko et al. 2000).

cDNA for the Putative MAGI Scaffold Protein

A fragment of the cDNA encoding the putative MAGI protein was identified in the S. domuncula EST database after domain analysis (Chevreux et al., submitted); the cDNA library of S. domuncula had been constructed as described (Kruse et al. 1997). The fragments obtained were used to screen the cDNA library (Wiens et al. 2000); the cDNA, termed SDMAGI, was completed using the RACE technique (Invitrogen GeneRacer Kit; Invitrogen, Groningen, The Netherlands). DNA sequencing was performed with an automatic DNA sequenator (Li-Cor 4000S). As determined by Northern blotting the complete sequence SDMAGI was obtained; it comprises 3271 nucleotides (nt). The phylogenetic tree was constructed on the basis of aa sequence alignments by neighbor-joining; degree of support for internal branches was further assessed by bootstrapping (Felsenstein 1993).

In situ Localization Studies

The method applied bases on the procedure described by Polak and McGee (1998) with modifications described recently (Perović et al. 2003). In brief, sections were obtained at −30°C and fixed with paraformaldehyde. The cuts were incubated with proteinase K and subsequently fixed again with paraformaldehyde. To remove the sponge color the cuts were incubated with ethanol. After rehydration with 1 × PBS, digoxigenin (DIG)-labeled DNA probes were added to the hybridization solution. Hybridization was performed overnight in a glass chamber at 45°C; the subsequent washes were performed at 50°C as described (Perović et al. 2003). After blocking the sections were incubated with an anti-DIG antibody conjugated with alkaline phosphatase. NBT/X-phosphate was used for the visualization of the signals. In all cases reported here positive signals were obtained with the antisense probes, while the sense probes (controls) showed no staining.

In situ hybridization was performed with DIG-labeled ssDNA probes. The probes were labeled with the PCR DIG Probe synthesis Kit (Roche). DNA probes were constructed, basing on the S. domuncula cDNA sequences, and spanned the region of amino acid 131 (aa131) to aa363 of the deduced MAGI_SD sequence or aa86 to aa160 of TM4SF_SD. Both antisense and sense probes were prepared by PCR, using the linearized cDNA as a template. The sense probes were obtained by applying a forward primer in the 5′-to-3′ sense direction; the complementary antisense probes were obtained by using a reverse primer in the 3′ to 5′ orientation. The probes had lengths of 693 bp (SDMAGI) and 225 bp (SDTM4SF), respectively.

The sections were inspected with an Olympus VANOX AHBT3 microscope; the images were recorded with a ColorView 12 camera and were composed applying the Soft Image System analySIS 3.0 (Soft Image System GmbH, Münster; Germany).

Northern Blotting

Total RNA was extracted from liquid nitrogen-pulverized primmorphs with TRIzol reagent (Gibco-BRL, Grand Island, NY). An amount of 5 μg of total RNA was electrophoresed through a 1% formaldehyde/agarose gel and blotted onto a Hybond-N+ nylon membrane following the manufacturer’s instructions (Amersham; Little Chalfont, Buckinghamshire, UK) (Wiens et al. 1998). Hybridization was performed with the 693-nt-large part of the SDMAGI cDNA and the 225-nt-long SDTM4SF (tetraspanin; CAA77026.1) probe. The probes were labeled with the PCR-DIG-Probe-Synthesis Kit according to the instruction manual (Roche). After washing, DIG-labeled nucleic acid was detected with anti-DIG Fab fragments (conjugated to alkaline phosphatase; dilution of 1:10,000) and visualized by chemiluminescence technique using CDP, the chemiluminescence substrate alkaline phosphatase, according to the instructions of the manufacturer (Roche). For the Northern blot studies the cDNA of the housekeeping gene β-tubulin, SDTUB (AJ550806), of S. domuncula was used as an internal standard.

Results and Discussion

Identification of a MAGI-Related Gene inS. domuncula

The cDNA, encoding the complete MAGI scaffold protein, SDMAGI, was isolated. The nucleotide sequence is 3271 residues long, with the open reading frame ranging from nt7–9 to nt3241–3243(stop) (accession number AJ580406). The deduced amino acid (aa) sequence, MAGI_SD, with 1078 residues, represents a calculated 119,042-Da polypeptide (Fig. 1A). MAGI_SD shares the highest similarity to the related metazoan scaffold proteins with an “expect value” (e; Blast-NCBI [Coligan et al. 2000]) of ≍e −149. As shown in Fig. 1A, the sponge MAGI protein displays the six characteristic PDZ domains: at aa22 to aa105 (domain 0 [according to González-Mariscal et al. 2003]), aa422 to aa504 (domain 1), aa549 to aa612 (2), aa656 to aa738 (3), aa785 to aa868 (4), and aa937 to aa1019 (5). The PDZ domains are conserved structural elements of approximately 80 aa which are involved in the formation of protein–protein interactions (Ponting et al. 1997; Ranganathan and Ross 1997). Among the six PDZ domains in MAGI proteins, PDZ-2 has been shown to bind to the signaling molecule GEP (for MAGI-1 in mammals [Mino et al. 2000]) and to PTEN, a tumor suppressor gene (for MAGI-2 in humans [Wu et al. 2000b; Tolkacheva et al. 2001]). Human MAGI-3 binds through PDZ-2 to PTEN and in the synaptic junctions through the PDZ-5 to the NMDA receptor (Hirao et al. 1998; Wu et al. 2000a,b). MAGI proteins, including that from S. domuncula, possess additionally two WW domains that bind to proline-rich peptide motifs (Sudol et al. 1995; Ermekova et al. 1998) of those proteins that are involved in signal transduction processes (Bork and Sudol 1994). Finally, the conserved guanylate kinase (GuK) motif is found from aa109 to aa189 of the sponge MAGI protein (Fig. 1A). The conserved guanylate kinase (GuK) motif in MAGUK proteins obtained its name because of its sequence similarity to the yeast GuK, an enzyme which synthesizes GDP from GMP under consumption of ATP (Berger et al. 1989). However, none of the GuK domains in MAGUK are enzymatically active (Dobrosotskaya et al. 1997).

Figure 1
figure 1

S. domuncula MAGI-related scaffold protein in S. domuncula. A The deduced sponge MAGI protein was aligned with the human membrane-associated guanylate kinase 2 (MAGI2_HS; AAC05370 [Wood et al. 1998]). Amino acids, similar among the sequences, are in inverted type. The characteristic conserved regions of the MAGI scaffold proteins, the six PDZ domains (= PDZ =), the guanylate kinase motif (– GuK –) and the two WW domains (∼ WW ∼) are marked. B Rooted phylogenetic tree constructed from the above-mentioned sequences as well as the other human sequences, MAGI-1B alpha beta (MAGI1_HS; AAK94064 [Laura et al. 2002]) and MAGI-3 (MAGI3_HS; NP_690864 [Thomas et al. 2002]), as well as two distantly related sequences from Caenorhabditis elegans, interacting protein 1 (IAP1_CAEEL; NP_502221), Drosophila melanogaster, the hypothetical protein CG30388-PA (CG30388_DROME; NP_611551), and the yeast (Saccharomyces cerevisiae) protein involved in ubiquitin-mediated protein degradation, Rsp5p (Rsp5p_YEAST; NP_011051) together with the plant (Arabidopsis thaliana) guanylate kinase (GUK_ARATH; NP_566276). The tree was rooted using the plant sequence as an outgroup. The scale bar indicates an evolutionary distance of 0.1 aa substitution per position in the sequence.

Full size image

Phylogenetic Relationship

In vertebrates three members of the MAGI scaffold proteins have been identified, which localize in the tight junctions of epithelial cells and also in the synaptic junctions (Dobrosotskaya et al. 1997; Wood et al. 1998; Wu et al. 2000a; Laura et al. 2002; Thomas et al. 2002; reviewed by González-Mariscal et al. 2003). All three MAGI proteins represent different isoform products of alternative splicing. For the present alignment the longest forms of the proteins have been included. We selected the three human MAGI scaffold proteins, MAGI-1 (Laura et al. 2002), MAGI-2 (Wood et al. 1998), and MAGI-3 (Dobrosotskaya et al. 1997; Thomas et al. 2002), that displayed the highest sequence similarity to the S. domuncula protein. They share 45% similar aa residues and 28% identical residues with the S. domuncula protein. It is striking that the highest similarity between the sponge and the human molecules is seen within the conserved regions of the polypeptide (Fig. 1A). In contrast, the sequence similarity to the two MAGI sequences from invertebrates (from Drosophila melanogaster, the hypothetical protein CG30388-PA, and from Caenorhabditis elegans, the interacting protein 1 molecule) is considerably lower (31% similarity/19% identity to these invertebrates). Only distantly related are the molecules from the yeast Saccharomyces cerevisiae, a protein involved in ubiquitin-mediated protein degradation, and the plant Arabidopsis thaliana, a guanylate kinase (<10% similarity/<10% identity). These relationships are also reflected by the rooted tree, constructed from these sequences (Fig. 1B). The sponge MAGI protein forms the base of the branch, including the three human MAGI forms, while the insect and nematode polypeptides branch off earlier. The plant and yeast sequences show only a distant relationship.

Identification of Cells Expressing the MAGI Gene

Since the technique of in situ hybridization was established in sponges on the cellular level (Perović et al. 2003), this method became a powerful tool to learn more about the function of the expressed genes. Both antisense and sense SDMAGI probes were applied for the hybridization procedure, in tissue of S. domuncula specimens. While the sense probe did not show any signal (Fig. 2B), the antisense SDMAGI reacted with cells lining the pinacocyte layer, the pinacoderm, and some canals within the mesohyl, the internal part of the sponge (Fig. 2A). A higher magnification of the sections showed that especially the surface cell layer of the pinacoderm reacted positively with the antisense SDMAGI probe (Figs. 2C and D), likewise only one cell sheet, surrounding the canals reacted with the probe (Fig. 2D and E).

Figure 2
figure 2

Expression of MAGI in tissue from S. domuncula, determined by in situ hybridization. Tissue slices through the surface of a specimen were prepared and hybridized with antisense (A, C–G) as well as sense (B) SDMAGI probe. The antisense MAGI reacted with the pinacoderm layer (pd), which seals the animals (A, C, D), and also with the cells which surround the canals (ca) (D, E). F Low expression of MAGI at the surface (surf) of the sponge tissue immediately after ablation of the pinacoderm layer. G After a regeneration period of 6 days a strong signal is seen at the newly formed pinacoderm (pd). Original magnifications: A and B, ×40; C to E, ×400; F and G, ×100.

Full size image

From these data we conclude that the sponge SDMAGI gene is expressed in the surface/epithelial layers in sponges, primarily in the pinacoderm and also frequently around the canals.

The S. domuncula Tetraspanin

To further support the existence of MAGI and its selective expression in the surface cellular layers another putative receptor found in cell junctions of higher metazoans was analyzed in S. domuncula. Until now only a sequence for tetraspanin was identified; no other junctional proteins, e.g., occludin, claudins, or junctional adhesion molecules, have been identified yet. The identification of the cDNA encoding the S. domuncula tetraspanin (SDTM4SF; accession number CAA77026.1) was given previously (Müller et al. 1999b). The deduced protein of 248 aa has a calculated M r of 26,701 and shows four putative transmembrane segments as concluded from the hydrophobicity plot (Kyte and Doolittle 1982) (Fig. 3A). At the NH2-terminus, the first cytoplasmic domain, and at the COOH-terminus, the third domain, is found, while the second domain is located between the second and the third transmembrane domain (Fig. 3A); especially in these regions the similarity to the other metazoan molecules is high. The characteristic aa residues within the tetraspanins are also found in the S. domuncula molecule (Fig. 3A), especially the conserved residues in the second extracellular loop (see Yunta and Lazo 2003).

Figure 3
figure 3

S. domuncula tetraspanin. A The deduced protein TM4SF_SD was aligned with the human tetraspanin (TM4SF_HS; NP_003261 [Berditchevski 2001; Maeda et al. 1998; Todd et al. 1998]). The characteristic segments within the receptor, the four transmembrane regions (TM1–TM4) and the two extracellular loops (EC1 and EC2) are marked together with the conserved amino acids in EC2 (•). B The rooted phylogenetic tree was constructed together with the sequences from C. elegans, a tetraspanin family member (TSP8_CAEEL; NP_510445), D. melanogaster, tetraspanin 74F (TM4SF_DROME; NP_524132), S. cerevisiae, ZIPpering protein (ZIP_YEAST; NP_011265), and A. thaliana, the senescence-associated protein (SEN_ARATH; NP_176515). The plant sequence was used as outgroup.

Full size image

A comparison to the human tetraspanin (Berditchevski 2001; Maeda et al. 1998; Todd et al. 1998) reveals a strong relationship (Fig. 3A); the polypeptides share 42% similar aa and 25% identical aa with each other. Again, the sponge polypeptide clusters in the rooted phylogenetic tree with the human tetraspanin (grouped with the transmembrane 4 superfamily which interacts with integrins; [Berditchevski 2001]; NP_003261). The tetraspanins from the two selected invertebrates, D. melanogaster (a deduced protein obtained from random sequencing [Adams et al. 2000]; NP_524132) and C. elegans (a tetraspanin family member, tsp-8, that is orthologous to the human gene KANGAI 1 [NP_510445; sequence source, NM 078044.1]), form a separate branch (Fig. 3B).

The nonmetazoan polypeptides from S. cerevisiae, the ZIPpering protein (presumably involved in segregation of chromosomes [NP_011265]), and from A. thaliana, the senescence-associated protein (cDNA, obtained by random sequencing of chromosome 1 [Tettelin et al. 1997]; NP_176515]), have only a weak similarity.

Localization of Cells Expressing Tetraspanin Within Sponge Tissue

The in situ hybridization technique was applied to localize the cells in the sponge tissue which highly express tetraspanin. The sections through the outer surface of the specimens revealed that the tetraspanin gene is expressed especially in the pinacoderm layer (exopinacoderm) (Fig. 4A). The cells of the endopinacoderm which surround the canals are highlighted to a lower degree (Fig. 4B). From these data we conclude that the expression pattern of the tetraspanin gene is very similar to that seen for the MAGI gene.

Figure 4
figure 4

Localization of cells within sponge tissue, expressing tetraspanin. Cryosection through tissue was prepared and subjected to in situ hybridization using the SDTM4SF antisense probe. A Area of the exopinacoderm; the pinacoderm layer (pd) is marked. B Cut through the internal body of the sponge specimen; one canal is marked (ca). Original magnification: A, ×100; B, ×250.

Full size image

Expression of tetraspanin and MAGI inSponge Tissue and Primmorphs

Two biological systems were used to determine the expression of the two cell junction proteins semiquantitatively; (i) the regeneration of the pinacoderm layer in adult animals and (ii) the formation of primmorphs from cells of S. domuncula.

The surface layer of intact animals (Fig. 5A-a) was ablated (Fig. 5A-b). A sample each from the surface as well as from the interior tissue of the animal was immediately used for the preparation of the RNA. Six days after regeneration of the ablated tissue had started, another sample was taken from the regeneration zone (Fig. 5A-c). Northern blot analysis was performed using the labeled probes from SDMAGI (encoding MAGI), SDTM4SF (tetraspanin), and SDTUB (β-tubulin). The results revealed that the expression of the 3.3-kb SDMAGI transcript was detectable in the surface tissue which had been removed from the specimen, while the RNA from the interior part of the animal gave no signal (Fig. 6; left). After 6 days of regeneration the SDMAGI transcripts could be identified again in the RNA sample (Fig. 6; left). A similar expression pattern is seen for the tetraspanin (SDTM4SF) gene. The intensities of the tubulin gene were almost identical in the different samples, indicating that the same amount of RNA was loaded on the gels (Fig. 6; left).

Figure 5
figure 5

S. domuncula animals and primmorphs. A-a Sponge specimens. Magnification, ×1. A-b From the surface of a specimen a 2-mm-thick layer was ablated (surf), and the interior of the animals became visible (int); ×1. A-c After a period of 6 days the surface of the animal regenerated and almost no scar (sc) was seen; ×2. Primmorph formation. Dissociated cells (magnification, ×20; (B-a), after 3 days, form compact aggregates (×5; B-b), which develop further to primmorphs after 6 days (×3; B-c).

Full size image
Figure 6
figure 6

Expression of SDMAGI (the cell junction scaffold protein) and SDTM4SF (the tetraspanin receptor). Left: A surface sample (surf) as well as tissue from the interior of the individual (int) was collected immediately after cutting. In parallel, RNA was extracted from a tissue sample of the surface of the animal 6 days after regeneration (reg) and processed likewise. Five micrograms of total RNA per each slot were size-separated; after blot transfer hybridization was performed with either SDMAGI, SDTM4SF, or β-tubulin. Right: Detection of transcripts for the genes SDMAGI, SDTM4SF, and β-tubulin. RNA samples were obtained from single cells (time: 0 days), from aggregates (3 days after reaggregation was started), or from primmorphs (after 6 days).

Full size image

In dissociated cells (Fig. 5B-a) obtained from the sponge almost no transcripts could be resolved on the blots for SDMAGI and for SDTM4SF (Fig. 6; right). Expression started, however, already 3 days after reaggregation (Fig. 5B-b) and increased during the subsequent 3 days of incubation (Fig. 6; right), the period during which the primmorphs with their pinacoderm layer are formed (Fig. 5B-c). Again the levels of the tubulin transcripts remained the same (Fig. 6; right).

From these data we conclude that the expressions of the genes coding for S. domuncula MAGI and tetraspanin proteins are high in tissue samples which contain epithelial surface cells, the pinacoderm. Focusing on the primmorphs it became evident that MAGI and tetraspanin are expressed during formation of the smooth cell surface layer.

To monitor the expression of SDMAGI during restoration of the pinacoderm after ablation of a part of the surface tissue from an individual sponge, in situ hybridization analysis was performed with the antisense SDMAGI probe. The data prove that immediately after the removal of the surface tissue, the cells at the border did not show high expression levels of MAGI (Fig. 2F). However, if the surface of regenerating tissue is analyzed after 6 days a strong signal is seen in the newly formed cell layer, covering the specimen (Fig. 2G).

Conclusion

It was already assumed, but it could not be proven, that the cells of the outermost sealing layer of sponges (Porifera) are tightly attached to each other; therefore the epithelial layers were termed “ectoderme” and “endoderme,” and their cells “cellules conjonctives” (Lieberkühn 1857; DeLage 1892). Only a few electron microscopic studies showed that the cell–cell contact of the cells in the epithelial layer is firm; therefore the contact zones between the cells were cautiously termed membrane bridges in the demosponge Microciona prolifera (Evans and Bergquist 1974), attachment plaque in the demosponge Ephydatia muelleri (Pavans de Ceccatty 1981), desmosome-like in Ephydatia fluviatilis (Pottu-Boumendil 1975), and septate junctions in E. fluviatilis (De Vos 1977) and in the calcareous sponge Sycon ciliatum (Ledger 1975). In addition, it could be shown, also by electron microscopy, that the intracellular fibers are connected with intercellular bridges (Pavans de Ceccatty 1989). In spite of these data a convincing demonstration of the existence of junctions in sponges will remain incomplete and not generally accepted (Nielsen 2001), without the demonstration that also the junctional molecules connecting the cells are present. The first molecular data came from the demonstration that in sponges genes exist indicative of the presence of a basement membrane, like type IV collagen in the demosponge P. jarrei (Boute et al. 1996). Furthermore, it can be hypothesized that asymmetrically arranged cells might be present on the surface of sponge specimens. Polarized cells are required for the establishment of epithelia, sheets which haven been considered fundamental features of multicellular organisms (Perez-Moreno et al. 2003). Without the demonstration of the existence of the characteristic junctional proteins and the proof of their localization/function in sponges, their cellular contacts were termed, in the past, “almost” junctions (Westheide and Rieger 1996; Nielsen 2001).

The proof of the junctional molecules in S. domuncula by molecular biological methods gave us sufficient evidence that in demosponges the underlying receptors and their organizing scaffold proteins are present. The demonstration that the gene for a tetraspanin receptor in S. domuncula with its characteristic functional amino acids exists was not surprising, since this type of receptor is not exclusively found in tight/cell junctions but is functionally present also in other microdomains of metazoan cells, where they interact with integrin receptors (Yunta and Lazo 2003). However, the result that the gene coding for a scaffold protein, MAGI, exists in S. domuncula as well and is even expressed in the surface layers, both the exopinacoderm and the endopinacoderm, provides compelling evidence for the presence of cell junctions.

It cannot be ignored that other molecules related to the tight junctions, such as the zonula occludens proteins, already found in cnidarians (Fei et al. 2000), could be present also in sponges. A schematic model for the organization of the junctions in sponges can be proposed (Fig. 7), in which the S. domuncula tetraspanin interacts with SD_MAGI, through its PDZ modules. The complexes formed are able to cover spatial cues at the cell surface, originating functionally distinct membrane domains that are characteristic for (postulated polarized) epithelia.

Figure 7
figure 7

Schematic representation of the molecules involved in the formation of cell junctions in sponges, here from S. domuncula. Epithelial cells are sealed together using specific receptors/molecules clustered in cell junctions (cj). The adjacent plasma membranes are held together by strands of transmembrane cellular (junctional) proteins/receptors (TM-cj). In S. domuncula, data suggest that the cell membrane-spanning receptor tetraspanin associates with the scaffold protein, the PDZ protein MAGI. MAGI, as a scaffold protein, plays a crucial role in the organization of the membrane receptor molecules and the effector molecules. The latter ones compose the cytoskeleton and the signal transduction molecules.

Full size image

In summary, the identification of cell junction proteins, the tetraspanin receptor and the scaffold protein MAGI, confirms that during the evolutionary transition from the fungi, as the closest related kingdom to the Metazoa (Schütze et al. 1999), to the multicellular animals, with the sponges as the oldest phylum, epithelial layers with cell junctions had been formed.