Isolation and cloning of a C-type lectin from the hexactinellid sponge Aphrocallistes vastus: a putative aggregation factor
glycob.oxfordjournals.org
Dietmar Gundacker, Sally P. Leys2, Heinz C. Schröder, Isabel M. Müller and Werner E.G. Müller1
Institut für Physiologische Chemie, Abteilung Angewandte Molekularbiologie, Universität, Duesbergweg 6, D-55099 Mainz, Germany, and 2Department of Biology, University of Victoria, British Columbia V8W 3N5, Canada
Received on April 25, 2000; revised on July 20, 2000; accepted on September 12, 2000.
Abstract
Among the sponges (Porifera), the oldest group of metazoans in phylogenetic terms, the Hexactinellida is considered to have diverged earliest from the two other sponge classes, the Demospongiae and Calcarea. The Hexactinellida are unusual among all Metazoa in possessing mostly syncytial rather than cellular tissues. Here we describe the purification of a cell adhesion molecule with a size of 34 kDa (in its native form; 24 kDa after deglycosylation) from the hexactinellid sponge Aphrocallistes vastus. This adhesion molecule was previously found to agglutinate preserved cells and membranes in a non–species-specific manner (Müller, W. E. G., Zahn, R. K, Conrad, J., Kurelec, B., and Uhlenbruck, G. [1984] Cell adhesion molecules in the haxactinellid Aphrocallistes vastus: species-unspecific aggregationfactor. Differentiation, 26, 30–35). The fact that the aggregation process required Ca2+ and was inhibited by bird’s nest glycoprotein and D-galactose but not by D-mannose or N-acetyl-D-galactosamine suggests that this cell adhesion molecule is a C-type lectin. To test this assumption, two highly similar C-type lectins were cloned from A.vastus. The deduced polypeptides of the two cDNA species isolated classified these molecules as C-type lectins. The calculated Mr of the 191 aa long sequences were 22,022 and 22,064, respectively. The C-type lectins showed highest similarity to C-type lectins (type-II membrane proteins) from higher metazoan phyla; these molecules are absent in non-Metazoa. The two sponge C-type lectins contain the conserved domains known from other C-type lectins (e.g., disulfide bonds, the amino acids known to be involved in Ca2+-binding, as well as the amino acids involved in the specificity of binding to D-galactose) and a hydrophobic N-terminal region. The N-terminal part of the purified C-type lectin was identical with the corresponding region of the deduced polypeptide from the cDNA. It is proposed that the A.vastus lectins might bind to the cell membrane by their hydrophobic segment and might interact with carbohydrate units on the surface of the other cells/syncytia.
Key words: sponges/Hexactinellida/Aphrocallistes vastus/aggregation factor/lectin/C-type lectin/evolution/cell adhesion
Introduction
Multicellularity has arisen several times in evolution and in all major kingdoms (prokaryotes, plants, fungi, and animals; Schopf, 1993). It is generally agreed that multicellular plants, the red algae, the brown algae, the land plants and the fungi arose separately from unicellular ancestors (Devereux et al., 1990; Kirk, 1997). Molecular (Müller, 1995) and morphological (Müller, 1997; Wimmer et al., 1999) data indicate that the transition from the Protozoa to the Metazoa occurred only once in evolution. Sequence data from a variety of proteins involved in cell–cell interactions indicate that all animals, including sponges, are of monophyletic origin (Müller, 1995).
The phylum Porifera, the oldest group of multicellular animals, includes three classes: the Demospongiae, the Calcarea, and the Hexactinellida. The Hexactinellida differ substantially from the other two sponge classes in having largely syncytial tissues (Reiswig, 1979; Mackie and Singla, 1983; Leys, 1995, 1999). The majority (75%) of the sponge constitutes a multinucleated syncytium; the remaining portions of the sponge function independently as cells but are still connected to the whole by open or plugged cytoplasmic bridges (Leys, 1999). Knowledge of hexactinellid embryology is still very limited, as for the most part these sponges inhabit deep waters accessible only by submersible or dredge. Early descriptions of hexactinellid embryos obtained from dredged specimens (Ijima, 1901; Okada, 1928) and an examination of embryos from a population of hexactinellids recently discovered (Boury-Esnault et al., 1999) suggest that the embryo is “cellular” but that the larva is clearly syncytial. How the syncytial tissue arises remains unclear.
Two alternative hypotheses have been proposed to explain relationships between the major sponge classes. One suggests that the Demospongiae are more closely related to Hexactinellida based on presumed larval similarities (Böger, 1988). The other divides the Porifera into the adelphotaxa of the Hexactinellida and the Demospongiae/Calcarea based on the gross difference in tissue structure (Reiswig and Mackie, 1983; Leys, 1999). Recent molecular evidence supports the latter view (Koziol et al., 1997; Kruse et al., 1998; Skorokhod et al., 1999) suggesting that Calcarea are most closely related to other diploblasts and form a clade with the Demospongiae. The Hexactinellida are thought to have diverged first from a common ancestor of the Metazoa (Müller et al., 1998; Müller and Müller, 1999).
There is now a wealth of cellular and molecular evidence to suggest that the Demospongiae have many of the features characteristic of much evolutionarily younger metazoan taxa (e.g., the receptor tyrosine kinases (Müller and Schäcke, 1996), integrins (Pancer et al., 1997; Wimmer et al., 1999), collagen (Exposito et al., 1991), and the metabotropic glutamate/GABA-like receptor (Perovic et al., 1999)).
The role of the aggregation factor (AF) in sponges has been studied in detail (reviewed in Müller, 1982, and Fernà ndez-Busquets and Burger, 1999). The AF isolated from demosponges has been shown to function in a species-specific manner (Moscona, 1968; Müller et al., 1979a). Furthermore, in the Demospongiae a lectin belonging to the S-type lectins (Pfeifer et al., 1993) has been found to be involved in a species-specific aggregation complex (Wagner-Hülsmann et al., 1996). A similar “factor” isolated from the hexactinellid Aphrocallistes vastus aggregated cells in the presence of Ca2+ non-species-specifically (Müller et al., 1984). The preparation contained several fractions with one dominant protein species of 34,000 kDa. Sugar analysis revealed that the A. vastus AF was a glycoprotein consisting of 55% (w/w) protein, 40% neutral carbohydrate and 2% hexuronic acid (Müller et al., 1984). The experiments also showed that the A. vastus AF “agglutinated” the cells by interacting in a homo- or heterophilic manner of the second order.
Until now no cell adhesion molecules have been cloned from hexactinellid sponges. However, the syncytial nature of hexactinellid tissue, namely, its ability to fuse to form a syncytium after dissociation through fine mesh, and the evidence that hexactinellid sponges may have been the earliest multicellular animals to have evolved on earth, suggest that a study of cell adhesion molecules in hexactinellid sponges may reveal new mechanisms of cell–cell recognition within the Metazoa.
In the present study, the previously isolated AF was purified and shown to have a size of 34 kDa (24 kDa after deglycosylation). The factor caused aggregation in the presence of Ca2+ and this function was inhibited by D-galactose. These properties—Ca2+-dependency and sugar specificity—suggest that the previously termed AF is in fact a Ca2+-dependent lectin. To test this assumption, two highly similar Ca2+-dependent lectins (C-type lectin) have been cloned from A.vastus. Phylogenetic analysis revealed that the cloned putative C-type lectin represents the oldest (phylogenetic) member of this class within the Metazoa.
The sequences reported here are deposited in the EMBL/GenBank data base under the accession no. AJ276450 (APHRLECC1) and AJ276451 (APHRLECC2) as Aphrocallistes vastus C-type lectins.
Results
Aggregation-promoting activity of an extract from A.vastus
It was previously found that the partially enriched fraction of the A.vastus extract (formerly termed AF) caused aggregation of preserved cells and syncytia at Ca2+ concentrations greater than 1 mM (Müller et al., 1984). By applying the enrichment procedure described previously, aggregates larger than 500 µm diameter (Figure 1B,C) were obtained from single cells/membranes (Figure 1A). The size of the aggregates increased with increasing concentration of the extract: at 5 units of extract the diameter of aggregates was approximately 500 µm (Figure 1B); at 20 units of extract they were larger than 4 mm (Figure 1 C).
PAGE revealed several protein bands in the fraction described previously by Müller et al. (1984). In order to select an affinity resin suitable for final purification of the extract, inhibition was performed using bird’s nest glycoprotein, which was found to potently inhibit extract-mediated aggregation. If 5 µg/ml of this glycoprotein was added together with the purified extract (Fraction V, Table I; see Materials and methods) the size of the aggregates was greatly reduced (Figure 1D). Consequently this glycoprotein was used for further purification (see below). Other carbohydrates were tested for their inhibition potential; neither D-mannose nor N-acetyl-D-galactosamine (5 mM) had any effect on aggregation of A.vastus syncytia/cells, while D-galactose significantly inhibited cell–cell interactions at a concentration of >0.1 mM. In one series of experiments, the cells/membranes were pretreated with -galactosidase as described in Materials and methods. Addition of lectin to these galactosidase-pretreated cells/membranes reduced the aggregation-promoting activity of the lectin (data not shown). All cell agglutination studies were performed in ASW containing 10 mM of CaCl2.
Purification of the extract
In a previous study it was shown that the extract from A.vastus can be highly enriched by the following steps (Fractions): (1) crude extract preparation, (2) Ca2+ precipitation, (3) Biogel P300 gel filtration, (4) sucrose gradient centrifugation, and (5) CsCl gradient centrifugation (Müller et al., 1984; Table I , Fractions I–V). By applying this procedure an increase in the specific activity (units/mg protein) from 18 (crude extract; Fraction I) to 690 (CsCl gradient centrifugation; Fraction VI) was achieved. It is interesting to note that purification with CaCl2 (Fraction II) resulted in a more than 10-fold enrichment (Table I), which contained three major protein species with sizes of 130, 81, and 34 kDa, as shown by PAGE analysis (Figure 2A).
For further purification, Fraction V was loaded onto the affinity column (bird’s nest glycoprotein-Sepharose 4B), which had been equilibrated with Ca2+-containing ASW, and the purified extract was eluted from the column using CMFSW-E. This procedure increased the specific activity from 690 to 1230 (units/mg); Table I. After this purification step only a single protein band with a size of 34 kDa was detected (Figure 2B, lane a).
Our earlier study (Müller et al., 1984) showed that the A. vastus extract contained a high percentage of neutral carbohydrates. As other C-type lectins are known to contain N-glycosylated side chains (Arai et al., 1998), the purified A. vastus lectin was incubated with endoglycosidase H to remove oligosaccharide side chains. Two protocols for the digestion were used: the first without pretreatment of the lectin with 1 M NaCl and bird’s nest glycoprotein and the second after inclusion of these two components as described in Materials and methods. Without the preincubation step PAGE analysis of the extract revealed a band of 24 kDa in addition to the 34 kDa band (Figure 2B, lane c). This result already suggests that the purified extract from A. vastus is a glycoprotein possessing at least one 10 kDa carbohydrate side chain. However, an analysis of the deduced polypeptide sequence of the cDNA encoding the A.vastus lectin revealed only one N-glycosylation site. Complete deglycosylation with this endoglycosidase was achieved after preincubation of the lectin sample at higher ionic strength (1 M NaCl) in the presence of bird’s nest glycoprotein. The rational was to allow the endoglycosidase to reach all potential cleaving sites of the lectin. Again the digested sample was analyzed by PAGE; the data show that under those conditions the size of the lectin dropped from 34 kDa (Figure 2C, lane a) to 24 kDa (lane b), suggesting that the lectin was completely hydrolyzed. The glycosidase itself has a Mr of 29,000 (Trimble and Maley, 1984), which could not be identified on the gel by staining with Coomassie brilliant blue (Figure 2 B, lane b).
The final purified extract (Fraction VI) caused strong aggregation-promoting activity of A. vastus syncytia and cells. In the presence of 5 units or 20 units of A. vastus extract, the sizes of the aggregates were almost identical to those observed after incubation of the cells with Fraction V. In addition, after complete deglycosylation of the lectin, no aggregation-promoting activity of the lectin could be detected in assays with A. vastus syncytia/cells (data not shown).
Cloning of cDNA encoding the A. vastus C-type lectin
A degenerate primer, designed against a conserved pattern within the large conserved disulfide bond terminal of Hydra C-type lectin was used to screen an A. vastus cDNA library. Two different clones were obtained and termed APHRLECC1 (size 899 nt) and APHRLECC2 (814 nt). The potential open reading frames of both start with the Met residue, found between nt134 to nt136 (APHRLECC1)and nt49 to nt51 (APHRLECC2). Northern blot analysis performed with the sponge APHRLECC1 clone as a probe yielded one broad band of 1.0Â kb, confirming that full length cDNA was isolated (data not shown).
Analysis of the deduced polypeptides of the C-type lectins
The protein sequences deduced from the cDNA clones APHRLECC1 (LECC1_APHR) and APHRLECC2 (LECC2_APHR) are both 191 aa in length (Figure 3 B). The calculated molecular weight for LECC1_APHR is 22,022, and 22,064 for LECC2_APHR. In the alignment shown (Figure 3B) these two sponge sequences were compared with the protein sequence for C-type lectin from chicken (Bezouska et al., 1991). The N-terminus of the purified A. vastus lectin was sequenced; and the first 11 aa, ALLILIGGLAM, were found to be identical with the aa moieties, aa3 to aa13, in the polypeptide deduced from the cDNA.
The two sponge C-type lectins comprise the most conserved domain which was first characterized in some animal lectins and which seems to function as a calcium-dependent carbohydrate-recognition domain (Drickamer, 1993). This domain is known as the C-type lectin domain (CTL) or as the carbohydrate-recognition domain that comprises 130 residues (PC/GENE-Prosite, 1995); Figure 3B. There are four Cys residues that are perfectly conserved, and which are involved in two disulfide bonds (Drickamer, 1988, 1993). They are found in the A.vastus lectins at aa94 and aa185 forming the long-range disulfide bond, and at aa162 and aa177, which putatively form the internal disulfide bond (Figure 3B). The C-type lectin consensus pattern—C-[LIVMFATG]-x(5,12)-[WL]-x-[DNS]-x(2)-C-x(5,6)-[FYWLIVSTA]-[LIVAT]-C (the residues found in the A.vastus lectins are underlined)—is found between aa162 and aa185 (Figure 3B; PC/GENE-Prosite, 1995). This pattern encompasses three Cys residues that are involved in disulfide bond formation. In addition to the two conserved disulfide bonds, another disulfide bond is present in some C-type lectins, which spans aa69 to aa78 in the A.vastus lectins.
The amino acids involved in the Ca2+ -binding of C-type lectins have been identified by Weis et al. (1991) and are marked in Figure 3 B. It is noteworthy that there is an N-glycosylation site at Asn51 (Figure 3 B). Furthermore, the N-terminal end of the A. vastus lectins shows one putative transmembrane segment, as calculated by the method described by Kyte and Doolittle (1982), that ranges from aa2 to aa17 (Figure 3A,B).
Some predictions about the potential carbohydrate binding site of C-type lectins comes from crystal structure analysis and mutagenicity experiments (Weis et al., 1991; reviewed in Hansen and Holmskov, 1998), using the rat mannose-binding lectin A (Drickamer et al., 1986). If the CTL-domain contains a Glu residue at aa185 and Asn at aa187, the C-type lectin preferentially binds mannose/glucose. However, if aa185 has Gln and aa187 an Asp then the lectin becomes galactose specific (Hansen and Holmskov, 1998). In the A. vastus C-type lectins those positions are occupied by Gln (aa153) and Asp (aa155), which suggests they have galactose specificity. Interestingly, the purified sponge extract cell agglutinating activity is inhibited by galactose and not by mannose.
Phylogenetic analysis of A.vastus C-type lectins
The C-type lectin domain signature has been found in a series of molecules that have been grouped into the following classes based on the location of the functional domains (Drickamer, 1988, 1993): (1) the type-II membrane proteins where the CTL domain is located at the C-terminal extremity of the proteins—these include the asialoglycoprotein receptors (known as hepatic lectins; Spiess, 1990) and the Kupffer cell receptor (Hoyle and Hill, 1988); (2) proteins that consist of an N-terminal collagenous domain followed by a CTL-domain (Weis et al., 1991)—these proteins are also called collectins; (3) the cell adhesion molecules “selectins” (Lasky, 1991); (4) type-I membrane proteins, for example, the 180 kDa secretory phospholipase A2 receptor (Lambeau et al., 1994); and (5) some invertebrate soluble galactose-binding lectins, for example, echinoidin (Giga et al., 1987), a lectin from the coelomic fluid of a sea urchin, as well as BRA-2 and BRA-3, two lectins from the coelomic fluid of a barnacle.
A databank search with the two deduced hexactinellid C-type lectins, LECC1_APHR and LECC2_APHR, revealed highest identity (similarity) to the C-type lectin from chicken (hepatic lectin; Bezouska et al., 1991). The sponge sequences share 18% identical and 33% similar aa with this lectin. Since this lectin has the same overall structure with respect to the CTL domain, the sponge lectins might be grouped with the type-II membrane proteins.
A phylogenetic analysis of the two sponge C-type lectins was performed with the most closely related sequences listed in the databanks. A radial unrooted phylogenetic tree revealed that the A. vastus lectins group in the same branch as the mouse and human C-type lectins (type-II membrane proteins; Richard and Beaulieu, 1998; Bates et al., 1999; Figure 4A). However, as the significance of the branching order is low, as seen from the low bootstrap values, the tree was rooted (operatively) using the Geodia cydonium S-type lectin (Wagner-Hülsmann et al., 1996) as an outgroup; Figure 4B. The resulting tree suggests that the A. vastus C-type lectins diverged first from a common ancestor, while the type-II membrane lectins from the sea urchin (Giga et al., 1987) and the Kupffer cells receptor (Hoyle and Hill, 1988) are the next most closely related molecules.
Add comment November 14th, 2006