Báo cáo khoa học: A (1fi3)-b-D-glucan recognition protein from the sponge Suberites domuncula Mediated activation of fibrinogen-like protein and epidermal growth factor gene expression pot

A 1fi3-b- D -glucan recognition protein from the spongeSuberites domuncula Mediated activation of fibrinogen-like protein and epidermal growth factor gene expression Sanja Perovic´-Ottsta

A (1fi3)-b- D -glucan recognition protein from the sponge

Suberites domuncula

Mediated activation of fibrinogen-like protein and epidermal growth factor gene expression

Sanja Perovic´-Ottstadt1, Teresa Adell1, Peter Proksch2, Matthias Wiens1, Michael Korzhev1, Vera Gamulin3, Isabel M Mu¨ller1and Werner E G Mu¨ller1

1

Institut fu¨r Physiologische Chemie, Abteilung Angewandte Molekularbiologie, Universita¨t, Mainz, Germany;2Institut fu¨r

Pharmazeutische Biologie, Heinrich-Heine-Universita¨t, Du¨sseldorf, Germany;3Institute Rudjer Boskovic, Department of

Molecular Biology, Zagreb, Croatia

Sponges (phylum Porifera) live in a symbiotic relationship

with microorganisms, primarily bacteria Until now,

mole-cular proof for the capacity of sponges to recognize fungi in

the surrounding aqueous milieu has not been available Here

we demonstrate, for the demosponge Suberites domuncula

(Porifera, Demospongiae, Hadromerida), a cell surface

receptor that recognizes (1fi3)-b-D-glucans, e.g curdlan or

laminarin This receptor, the (1fi3)-b-D-glucan-binding

protein, was identified and its cDNA analysed The gene

coding for the 45 kDa protein was found to be upregulated

in tissue after incubation with carbohydrate Simultaneously

with the increased expression of this gene, two further genes

showed an elevated steady state level of expression; one

codes for a fibrinogen-like protein and the other for the

epidermal growth factor precursor Expression of the

(1fi3)-b-D-glucan-binding protein and the fibrinogen-like

protein occurred in cells on the sponge surface, in the pin-acoderm By Western blotting, the product of the fibrin-ogen-like protein gene was identified, the recombinant protein isolated, and antibodies raised to this protein Their application revealed that a 5 kDa factor is produced, which

is apparently processed from the 77 kDa epidermal growth factor precursor Finally, we provided evidence that a tyrosine kinase pathway is initiated in response to exposure

toD-glucan; its phosphorylation activity could be blocked

by aeroplysinin In turn, the increased expression of the downstream genes was suppressed We conclude that sponges possess a molecular mechanism for recognizing fungi via theD-glucan carbohydrates on their surfaces Keywords: D-glucan binding protein; epidermal growth factor; fungi; sponges; symbiosis

Sponges (phylum Porifera) are, among all metazoan taxa,

those animals which contain the widest range of specific and

very effective bioactive compounds [1,2] It has been

assumed that most of these secondary metabolites are produced by symbiotic microorganisms which are harbored

by the sponges [3] Among these microorganisms, bacteria [4] and fungi are the most potent producers of secondary metabolites in sponges Hence, sponges must be provided with mechanisms to distinguish between harmful (perhaps infectious) and symbiotic bacteria and fungi At a molecular level, most of the work carried out towards understanding this host–microorganism symbiotic relationship has been performed with the demosponge Suberites domuncula Sponges are provided with a very efficient immune system, reminiscent of that found in higher metazoan phyla, particularly deuterostomians [5] In addition, sponges produce the same proteinaceous defense molecules (e.g tachylectin) that are known to be induced in protostomians

as a defense against bacteria [6] It has also been found that

S domuncula recognizes the lipopolysaccharide (LPS) molecule on the surface of bacteria and responds by activation of the mitogen-activated protein kinase (MAPK) pathway [7] Until the present study was undertaken, nothing was known, at a molecular level, about the system

by which sponges recognize fungi In an approach to eluci-date this mechanism, we activated sponge cells by selected model glucan polymers, including the (1fi3)-b- -glucans,

Correspondence to W E G Mu¨ller, Institut fu¨r Physiologische

Chemie, Abteilung Angewandte Molekularbiologie, Universita¨t,

Duesbergweg 6, 55099 Mainz, Germany.

Fax: + 49 6131 39 25243, Tel.: + 49 6131 39 25910,

E-mail: wmueller@uni-mainz.de

Abbreviations: EGF, epidermal growth factor; LPS,

lipopolysaccha-ride; MAPK, mitogen-actived protein kinase; PoAb, polyclonal

antibody.

Note: This article is dedicated to Professor Zeeck (University of

Go¨ttingen) on the occasion of his 65th birthday.

Note: The cDNA sequences from Suberites domuncula have been

deposited in EMBL/GenBank as follows: the (1fi3)-b- D

-glucan-binding protein (GLUBPp_SUBDO) under the accession number

AJ606470, the fibrinogen-like molecule (FIBl_SUBDO) under the

accession number AJ606471, and the epidermal growth factor

precursor (EGFl-PREC_SUBDO) under the accession number

AJ606469.

(Received 23 January 2004, revised 9 March 2004,

accepted 22 March 2004)

which have been isolated from cell walls of plants, but

also from bacteria and fungi [8] Prominent purified

glu-can molecules of this group are (a) curdlan, a linear

polysaccharide from Alcaligenes faecalis [9,10] and (b)

lami-narin, a poly (1fi3)-b-D-glucan with some interstrand

(1fi6)-b-D-glucan branch points [11], isolated from the alga

Laminaria digitata (1fi3)-b-D-Glucans induce immune

res-ponses in protostomians [12,13] and deuterostomians [14]

In the first series of experiments, using the model

compound curdlan, we demonstrated that sponges (S

do-muncula) indeed react to incubation with (1fi3)-b-D-glucan

First, we analysed whether curdlan influences the

phos-phorylation of MAPKs by S domuncula; however, no

change in the phosphorylation level was seen (data not

shown) Subsequently, we determined whether treatment

with curdlan modulates the tyrosine kinase pathway of

sponges Using an antibody specific for phosphotyrosine we

showed that at least one protein species underwent

phos-phorylation after incubation with this glucan To determine

the specificity of this reaction, the known tyrosine kinase

inhibitor, aeroplysinin, isolated from the sponge Verongia

(syn: Aplysina) aerophoba [15,16] was used

After proving that S domuncula recognizes (1fi3)-b-D

-glucan, the respective (1fi3)-b-D-glucan-binding protein

had to be identified Such a molecule has previously been

isolated and cloned from a number of protostomians – from

crustaceans [13,17], earthworm [12] and insects [18], as well

as from sea urchins [19] After successfully cloning the

(1fi3)-b-D-glucan-binding protein from S domuncula, we

continued our search for other potential binding proteins

that might be involved in recognizing theD-glucan

Prom-ising candidates were molecules that display lectin

proper-ties, e.g the horseshoe crab acetyl group-recognizing lectin

[20] or lectin molecules with fibrinogen domains, e.g the

ficolins [21] This rationale led to the isolation of a

fibrinogen-like molecule from S domuncula

It is known that cells from deuterostomians react to

fungal cell wall polysaccharides by producing cytokines [22]

The epidermal growth factor (EGF) domain occurs very

frequently in cytokines; for sponges this domain has already

been described [23] Therefore, degenerate primers were

designed to identify genes which comprise this domain and

that are expressed by the stimulation of sponges with

D-glucans This approach resulted in the identification of a

cDNA whose deduced polypeptide, termed EGF precursor,

comprises three EGF domains

Our data provide, for the first time, an insight into the

response of sponges to stimulation with (1fi3)-b-D-glucan

We show that the polysaccharide binds to the (1fi3)-b-D

-glucan-binding protein; subsequently, a gene encoding a

fibrinogen-like protein, and also one for a cytokine, are

strongly expressed

Materials and methods

Chemicals and enzymes

The sources of chemicals and enzymes used were as given

previously [24,25] Laminarin from L digitata, curdlan

from A faecalis and LPS from Escherichia coli O55:B5, as

well as monoclonal antibody (mAb) against

phosphotyro-sine were purchased from Sigma-Aldrich (Deisenhofen,

Germany) Aeroplysinin was isolated from the sponge Aplysina aerophoba, as described previously [15,16] Curdlan was labeled with biotin according to Novotna

et al [26] The glucans were dissolved as described previ-ously [12,27]

Sponges Live specimens of S domuncula (Porifera, Demospongiae, Hadromerida) were collected near Rovinj (Croatia) and maintained in aquaria in Mainz(Germany) for more than

10 months prior to use

Exposure of tissue samples fromS domuncula

to curdlan and Western blotting Tissue samples (2 g) were maintained for 1–3 days in seawater in the presence or absence of curdlan (10 lgÆmL)1) and were then processed as described previously [28] Where indicated, the tissue was additionally treated with 1 lgÆmL)1

of aeroplysinin Samples were homogenized in lysis buffer [1· Tris-buffered saline (TBS), pH 7.5, 1 mMEDTA, 1% Nonidet-P40, 10 mMNaF, protease inhibitor cocktail (one tablet per 10 mL) and 1 mM sodium orthovanadate], centrifuged and the supernatants analysed by Western blot

To determine the phosphorylation of tyrosine, the tissue samples were treated for 6 h with polysaccharide

Total tissue extracts (20 lg per lane) were subjected to electrophoresis in 8% polyacrylamide gels containing 0.1% SDS, as described by Laemmli [29] Western blotting experiments were performed as described previously [30] The membranes were incubated with mouse mAb-anti-phosphotyrosine (mAb-aTyr) (1 : 2000 dilution) After washing, the blots were incubated with peroxidase-coupled goat antimouse IgG (1 : 2000 dilution) Detection of the immunocomplex was carried out using the BM Chemoluminescence Blotting Substrate kit from Roche (Mannheim, Germany)

Ligand-binding blot The assay was performed as described previously [12] Extracts from tissue were incubated for 1 day with

10 lgÆmL)1curdlan, then treated with 0.2% SDS, but not with 2-mercaptoethanol (the samples were not boiled prior

to separation) The samples were then size separated by SDS-PAGE (12% gel) After separation, the proteins were transferred to poly(vinylidene difluoride)-Immobilon After blocking with BSA (1%, w/v), the blots were incubated with biotin-labeled curdlan (5 lgÆmL)1) Visualization was per-formed with peroxidase-avidin, using 4-chloro-1-naphthol

as the substrate In competition experiments, after transfer

of the proteins, the blots were first incubated with either

10 lgÆmL)1laminarin or 2 lgÆmL)1LPS The blots were then washed and incubated with biotin-labeled curdlan followed by peroxidase-avidin/4-chloro-1-naphthol

Isolation of cDNA for the (1fi3)-b-D-glucan-binding protein

The cDNA encoding a potential (1fi3)-b-D-glucan binding protein (GLUBPp_SUBDO) was isolated from the

S domuncula cDNA library [24] by PCR The primers

were designed against the highly conserved region within

the (1fi3)-b-D-glucan-binding proteins; in the

b-1,3-glucan-binding protein from the black tiger shrimp, Penaeus

mon-odon (accession number AF368168-1) the stretch reads

MLWPAIWM (amino acids 160–167) The degenerate

primer, 5¢-TGGCTITGGCCIGCIATA/C/GTGGATG-3¢,

was used in the PCR reaction, together with the vector

primer The PCR was carried out as follows: initial

denaturation at 95C for 4 min, followed by 30

amplifi-cation cycles at 94C for 30 s, 62 C for 45 s and 70 C

for 1.5 min, and a final extension at 70C for 10 min The

reaction mixture was as described previously [31] The

fragments obtained were used to isolate the cDNA from

the library [32] and identified one clone with a 1327

nucleotide insert [excluding the poly(A) tail] The clone

was termed SDGLUBP; it was sequenced using an

automatic DNA sequencer (Li-Cor4200; MWG Biotech,

Ebersberg, Germany)

cDNA corresponding to the fibrinogen-like protein

Following the strategy described in the Introduction, a

conserved fibrinogen domain was selected for the design

of degenerate primers Aligning different

fibrinogen-domain containing proteins, fibrinogens, fibroleukins/

techylectins, angiopoietins, ficolins and tenascins, the

following consensus was deduced: FSTxDNDND It is

located in the human fibrinogen a/a-E chain precursor

(accession number P02671) between amino acids 785 and

793 The degenerate forward primer, 5¢-TTC/TTCIACI

TGGGAC/TACC/TGAC/TACC/TGAC/T-3¢, was used

in the PCR reaction The PCR conditions were as

described above, except that 65C were used during the

amplification cycles Only one species of insert was

obtained, with a size of 1079 nucleotides This clone

was termed SDFIBI

cDNA encoding the putative EGF-like precursor,

EGFI-PREC_SUBDO

The EGF precursor, EGFI-PREC_SUBDO, was cloned

from the cDNA library using degenerate primers that

were designed against the conserved domain, including

the first Cys residue of the EGF domain from the

human pro-epidermal growth factor precursor (P01133),

DVNECAF; 5¢-GAC/TGAIAAC/TGAA/GTGC/TGCITTC/

T-3¢, was used in the PCR The PCR conditions were as

described above, with the exception that a temperature of

57C was used during the amplification cycles Only one

species of insert was obtained; it was 2446 nucleotides in

size This clone was termed SDEGFI-PREC

Sequence analysis

The sequences were analyzed using the computer programs

BLAST [33] and FASTA [34] Multiple alignments were

performed usingCLUSTAL W, Version 1.6 [35] Phylogenetic

trees were constructed on the basis of amino acid sequence

alignments by neighbour-joining, as implemented in the

NEIGHBOR program from the PHYLIP package [36] The

distance matrices were calculated using the

matrix model, as described previously [37] The degree of support for internal branches was further assessed by bootstrapping [36] The graphic presentations were pre-pared usingGENEDOC[38]

Recombinant EGF precursor and production of antibodies The sponge SDEGFI-PREC sequence was isolated by PCR using the forward primer, f1 [5¢-CCATGGAGA AGATTCTAGCAACAGTCAATTCAAATGAC-3¢ (the NcoI restriction site is underlined), nucleotides 1060–1098], and the reverse primer, r1 [5¢-GCGGCCGCTG TATCTGAAGTTGGGGAATTACTGTGTTTGTTGTT-3¢ (the NotI restriction site is underlined); nucleotides 2206– 2241] The full-length PCR product (1143 bp) was expressed in E coli The cDNA was cloned into the bacterial glutathione-S-transferase/oligohistidine/S expres-sion vector, pET41a (Novagen, Madison WI, USA) via the mentioned restriction sites After transformation with this plasmid, expression of the fusion protein was induced in E coli strain BL21 for 6 h at 37C with

1 mM isopropyl thio-b-D-galactoside [32] Bacterial pellets were obtained from 500 mL cultures The fusion protein was extracted and purified first with the His-tag purifi-cation kit (Novagen) and subsequently with the glutathi-one-S-transferase-tag purification kit (Pharmacia, Freiburg, Germany), as described by the manufacturer Finally, the fusion protein was cleaved with enterokinase (5 U; Novagen), as recommended The recombinant EGF precursor, r-EGF_SUBDO, was obtained tag-free through purification in a batch procedure using the glutathione-S-transferase-tag purification kit; the recom-binant protein remained in the supernatant The purity

of the material was verified by electrophoresis through 10% polyacrylamide gels containing 0.1% SDS, accord-ing to Laemmli [29] The protein was dialyzed against

25 mM Tris/HCl buffer (pH 7.2), supplemented with

10 mM DL-dithiothreitol

Polyclonal antibodies (PoAb) were raised against the recombinant EGF protein in female rabbits (White New Zealand), as previously described [39] Animal experiments were registered and performed according to German law After three booster immunizations, the serum was collected; the PoAbs were termed PoAb-EGF protein In control experiments, 100 lL of the PoAb-EGF protein was adsor-bed to 20 lg of r-EGF_SUBDO (30 min; 4C) prior to use Western blotting of EGF

For the identification of EGF in extracts from sponge tissue, extracts were prepared, as described above, and subjected to electrophoresis through 15% polyacrylamide gels contain-ing 0.1% SDS, as described previously [29] The membranes were incubated with rabbit PoAb-EGF precursor (1 : 500 dilution); the immune complexes were visualized by incu-bation with alkaline phosphatase-conjugated antirabbit IgG, followed by staining with 4-chloro-1-naphthol To quantify a given signal on the blots, scanning with the GS-525 Molecular Imager (Bio-Rad) was performed The relative value, with respect to the signal seen in the nontreated extract, is given for the signal seen in extract from curdlan-treated tissue

RNA preparation and Northern blot analysis

RNA was extracted from liquid-nitrogen pulverized tissue

using TRIzol reagent (GibcoBRL, Grand Island, NY,

USA), as described previously [40] Then, 5 lg of total

RNA was electrophoresed and blotted onto Hybond-N+

nylon membrane (Amersham, Little Chalfont, Bucks, UK)

Hybridization was performed with a 550 nucleotide region

of the SDGLUBP cDNA, a 220 nucleotide region of the

SDFIBI cDNA and a 200 nucleotide region of the

SDEGFL-PRECcDNA Regions spanning the open

read-ing frames were selected The housekeepread-ing gene (b-tubulin)

of S domuncula, SDTUB (accession number AJ550806),

was used as an internal standard The probes were labeled

using the PCR-DIG-probe-synthesis kit (Roche) After

washing, DIG-labeled nucleic acid was detected with

anti-DIG Fab fragments and visualized by chemiluminescence

using CDP (Roche)

In situ localization studies

The method applied was based on the procedure described

by Polak & McGee [41], with modifications described

recently [42] Frozen sections of 8 lm were obtained, fixed

with paraformaldehyde, treated with Proteinase K and

subsequently fixed again with paraformaldehyde To

remove the sponge color, the sections were washed with

increasing concentrations of ethanol and finally

isopro-panol After rehydration, the sections were hybridized with

labeled probes, the 550 nucleotide SDGLUBP or the 200

nucleotide SDFIBL cDNA After blocking, the sections

were incubated overnight, at 45C, with an alkaline

phosphatase-conjugated antidigoxigenin immunoglobulin

The dye reagent, Nitro Blue tetrazolium/X-Phosphate,

was used for visualization of the signals Antisense and

sense single stranded DNA digoxigenin-labeled probes were

synthesized by PCR using the PCR DIG Probe synthesis

Kit (Roche) Sense probes were used, in parallel, as negative

controls in the experiments

Results

Effect of incubation with curdlan on the phosphorylation

of tyrosine in sponge tissue

It is known that (1fi3)-b-D-glucans are activators of gene

expression in mammalian cells [22] Therefore, we

investi-gated whether sponges react to curdlan with an increased

phosphorylation of tyrosine Tissue samples were incubated

in the presence or absence of 10 lgÆmL)1curdlan Extracts

were prepared and the proteins were size separated by

SDS/PAGE After transfer, the blot was incubated with

mAb-aTyr and then with a labeled secondary antibody The

results show that in the absence of curdlan no bands were

detected on the blots (Fig 1B; lane a); however, in the

extracts from curdlan-treated tissue a strongly staining band

of 32 kDa was observed (lane b) When the tissue was

treated with curdlan and the tyrosine kinase-inhibitor,

aeroplysinin (1 lgÆmL)1), no 32 kDa band was detected

(lane c) In parallel, the gels were stained with Coomassie

Brilliant Blue (Fig 1A) and no change in the banding

pattern and their intensities occurred

Detection of glucan-binding activity in extracts fromS domuncula

Tissue from S domuncula was incubated for 1 day with

10 lgÆmL)1 curdlan Then, extracts were prepared and subjected to PAGE in the presence of a low concentration of SDS and in the absence of b-mercaptoethanol After size separation (Fig 2; lane a), the proteins were transferred and – after blocking – probed with labeled curdlan A 43 kDa polypeptide was observed in the extract (lane b) When the blot was first preincubated with 10 lgÆmL)1laminarin and – after washing – probed with the labeled curdlan, no band was detected (lane c) However, when the blot was preincubated with 2 lgÆmL)1 LPS and subsequently with labeled curdlan, the intensity of the band was only slightly reduced (lane d) From these data we conclude that a

43 kDa protein is present in the extract from curdlan-treated tissue, and that this protein comprises a specificity for (1fi3)-b-D-glucans In parallel, incubation experiments with curdlan had been performed for only 6 h Under these conditions, the binding between labeled curdlan and the

43 kDa polypeptide was very low (data not shown)

Cloning of cDNA encoding theS domuncula (1fi3)-b-D-glucan-binding protein

Sequence The insert with SDGLUBP comprises one ORF, which ranges from nucleotides 46–48 to nucleo-tides 1252–1254(stop); the cDNA is of full length, as shown by Northern blot analysis (1.4 kb; see below) The deduced protein shows high sequence similarity to the

Fig 1 Phosphorylation of a 32 kDa protein after incubation of sponge tissue with curdlan Tissue samples were incubated for 6 h with or without 10 lgÆmL)1curdlan Protein extracts were then prepared and size-separated by PAGE (8% gel) (A) The gel was stained with Coomassie Brilliant Blue (B) Proteins were blot transferred and reacted with mouse antiphosphotyrosine mAb and then with labeled goat anti-mouse IgG Detection of the immunocomplex was carried out as described in the Materials and methods Protein extract from tissue incubated in the absence (lane a, – cur), or in the presence (lane

b, + cur) of curdlan In one series of experiments, the tissue was additionally treated with 1 lgÆmL)1aeroplysinin (lane c, + aero) M, protein size markers.

(1fi3)-b-D-glucan-binding proteins and was therefore

termed GLUBPp_SUBDO The protein comprised 402

amino acid residues, with a calculated size of 45 040 Da,

and possessed, between amino acid 49 and amino acid 296,

one characteristic domain for glycosyl hydrolases of the

family 16 (PFAM: PF00722) with a high significance value

(E-value) of 2e-05 Two transmembrane regions were

identified [43], which ranged from amino acids 2 to 23 and

from amino acids 361 to 401 (Fig 3A) From these data we

conclude that the 43 kDa protein identified in the

ligand-binding blot probably corresponds to the 45 kDa

(1fi3)-b-D-glucan-binding protein deduced from SDGLUBP

Phylogenetic analysis The sponge glucan-binding protein

shares highest sequence similarity with the (1fi3)-b-D

-glucan-binding proteins with average sizes 350–400 amino

acids The highest similarity was calculated with the

b-1,3-glucan-binding protein from the black tiger shrimp,

P monodon, having approximately 37% identical and

53% similar (with respect to the physico-chemical

prop-erties) amino acid residues to the sponge protein The

similarity of the sponge glucan-binding protein to related

insect and crustacean proteins (35% identity/50%

simi-larity) was only slightly lower No considerable similarity

was found to exist to the nonmetazoan and the

protostomian/nematode putative proteins present in the

database After alignment of all similar sequences, a radial

tree was constructed which shows that the sponge

glucan-binding protein forms the basis for the insect molecule on

one side and the molecule from crabs on the other

(Fig 3B)

cDNA encoding the fibrinogen-like protein Sequence One species of insert was identified – the ORF, which spanned nucleotides 31–33 to nucleotides 877–879(stop) The full size cDNA (SDFIBI; 1.1 kb by Northern blot analysis; see below) encoded the predicted protein, termed FIBI_SUBDO, comprising 282 amino acid residues (giving a calculated Mrof 31 997) Domain searches revealed that within the polypeptide, one fibrinogen domain for b- and c-chains (PFAM: PF00147) exists between amino acids 81–270 One conserved disulphide bond exists connecting Cys225 to Cys239 and one eukaryotic secretory signal sequence can be predicted [44] (Fig 4A) The highest similarity exists with vertebrate fibrinogens; therefore the sequence was named fibrinogen-like protein

Fibrinogens are the principal proteins of the vertebrate clotting system and form hexamers, composed of the three different chains: a, b and c [45] As outlined by Spraggon

et al [46], the b- and c-chains are homologous throughout the complete sequence, while the a-chain comprises the highest similarity only in the first 200 residues Alignment studies with the sponge and three mammalian fibrinogens showed that the sponge fibrinogen-like protein, even though the full-length sequence is available, shares similarity only within the middle segment of the a-, b- and c-chains Hence,

no further classification of the sponge protein to any of the three vertebrate chains can be made In the sponge sequence, besides the first disulfide bridge mentioned, the disulfide rings and the thrombin attack point (which exist in the human sequence) are lacking A potential arginine residue in the sponge fibrinogen at amino acid position 18 cannot be recognized by thrombin owing to a negatively charged glutamic acid residue at position P2 [47] The conserved central segments within the fibrinogen domain [46] are present in the sponge protein

Phylogenetic analysis The analysis was performed with the fibrinogen domain of the sponge fibrinogen-like protein The highest similarity, with approximately 35% identical and 45% similar amino acid residues, was found to the fibrinogens in the databases (Fig 4B); the human fibrinogen c-chain precursor (P02679) was used for the alignment (Fig 4A) The sequences were compiled and an unrooted (slanted) tree was constructed (Fig 4B) The trichotomous tree shows that the families of the fibroleukins, with the human member (Q14314) as an example, together with the techylectins from the horseshoe crab, Tachypleus tridentatus [20] (AB024737.1 and AB024738.1), and the angiopoietins from mammals, e.g humans (O15123), form the second branch The third branch is built by the ficolins, with the mouse ficolin B as an example [48] (AF063217), and the tenascins, including also the precursors from humans (dJ1141O19.1), as members (Fig 4B) The basis again is the sponge-deduced protein

cDNA of a potential EGF precursor Sequence One species of cDNA, which encodes a deduced protein containing EGF domains, and was therefore termed EGF precursor (SDEGFI-PREC) was isolated from the library The 2446 nucleotide contains an ORF, from nucleotides 100–102 to nucleotides 2242–2244(stop); the

Fig 2 Detection of a glucan-binding protein in extracts from

Suber-ites domuncula A protein sample was prepared from tissue that had

been incubated for 1 day with 10 lgÆmL)1curdlan, as described in the

Materials and methods Extract from curdlan-treated tissue was size

separated by SDS-PAGE (lane a) After separation, the protein

ex-tracts were transferred to poly(vinylidene difluoride)-Immobilon and

incubated with biotin-labeled curdlan (cur, 5 lgÆmL)1) (lane b).

Alternatively, the blots were first preincubated with 10 lgÆmL)1

lam-inarin (lam, lane c), or 2 lgÆmL)1lipopolysaccharide (LPS, lane d), for

5 h, and then washed and probed with biotin-curdlan (cur), as

des-cribed in the Materials and methods.

Fig 3 The Suberites domuncula potential beta-1,3-glucan-binding protein (GLUBPp_SUBDO) (A) The deduced sponge sequence (GLU-BPp_SUBDO) is aligned with the most related sequence, the b-1,3-glucan-binding protein from the black tiger shrimp Penaeus monodon (GLUBP_PENMO, AF368168-1) Identical amino acids are shown in white on black The positions of the two potential transmembrane regions (TM) and the glycosyl hydrolases-16 domain (glyco-hydr) are indicated The segment towards which the degenerate primers were designed is underlined by dashes (B) Phylogenetic analysis of these two sequences with the GLUBP from the blue shrimp, Litopenaeus stylirostris (GLUBP_LITSTY, AF473579-1), the putative Gram-negative bacteria-binding proteins from the Diptera Anopheles gambiae (ENSAN1_ANGA, XP_312118.1), (ENSAN5_ANGA, XP_312116.1) and (BACBP_ANGA, CAA04496.1), as well as the GLUBP from the lobster Homarus gam-marus (GLUBP_HOGAM, CAE47485.1) and the crayfish Pacifastacus leniusculus (GLUBP_PACLE, CAB65353.1) After alignment, the radial tree was constructed.

Fig 4 Suberites domuncula fibrinogen-like protein (A) The deduced sponge protein, FIBl_SUBDO, is aligned with the related fibrinogen c-B chain precursors from humans (FIBG_HUMAN; P02679) The conserved fibrinogen domain (FIBR) and one conserved disulfide bridge (C–C) are present in the sponge protein, while the second disulfide bridge found in the human sequence is absent; this is marked ([C]–[C]) The predicted eukaryotic secretory signal sequence terminates after amino acid 22 (SS) The thrombin attack point ({}) and the disulfide rings (underlined) in the human sequence are indicated The double underlined amino acids represent the regions towards which degenerate primers were designed at the nucleotide level The conserved central segments within the fibrinogen domain are marked (++) (B) A slanted cladogram was constructed using the conserved fibrinogen domains of the two sequences mentioned above and of the following sequences (i) Fibrinogens: fibrinogen a-2 chain precursor from the sea lamprey Petromyzon marinus (FIB2_PETMA; P33573), fibrinogen a/a-E chain precursors from chicken (FIBA_CHICK; P14448), human (FIBA_HUMAN; P02671) and rat fibrinogen (FIBA_RAT; P06399), and the fibrinogen c-B chain precursors from bovine (FIBG_BOVIN; P12799), rat (FIBG_RAT; P02680), frog (FIBG_XENLA; P17634), and sea lamprey (FIBG_PETMA; P04115) (ii) Fibroleukins and techylectins: the fibroleukin precursors from mouse (FGL2_MOUSE; P12804) and human (FGL2_HUMAN; Q14314), as well as the techylectins from the horseshoe crab Tachypleus tridentatus (TECL5A-TACTR; AB024737.1 and TECL5B_TACTR; AB024738.1) (iii) Angiopoietin: angiopoietin 1 and 2 precursors from mouse (AGP1_MOUSE; O08538 and AGP2_MOUSE; O35608), bovine (AGP1_BOVIN; O18920 and AGP2_BOVIN; O77802), and human (AGP1_HUMAN; Q15389 and AGP2_HUMAN; O15123) (iv) Ficolins: ficolin A and B from pig (FICOLA_PIG; L12344 and FICOLB_PIG; L12345), mouse (FICOLA_MOUSE; AB007813 and FICOLB_MOUSE; AF063217), rat (FICOLA_RAT; AB026057), and the echinoderm Parastichopus parvimensis (FIBA_PARPA; P19477) (v) Tenascins: the tenascin precursors from chicken (TENA_CHICK; P10039), human (TENA_HUMAN; P24821 – and – TENAl_HUMAN; dJ1141O19.1), fish Danio rerio (TENAC_DARE; CAA61489.1), and pig (TENA_PIG; Q29116 – and – TENAX_PIG; CAA60686.1), as well as the the microfibril-associated glycoprotein (MFA4_HUMAN; P55083) The numbers at the nodes are an indication of the level of confidence, given as a percentage, for the branches as determined by bootstrap analysis (1000 bootstrap replicates).

size of the transcript, based on Northern blotting, is 2.6 kb.

The deduced 714 amino acids have a calculated Mr of

77 901 (putative EGF precursor, EGFI-PREC_SUBDO)

By comparison with the Isrec-Server [49] domain database,

three EGF domains were identified in EGFI-PREC_SUB-DO; they span the regions amino acids 331–368 (EGF1), amino acids 364–410 (EGF2) and amino acids 407–455 (EGF3) Furthermore, three low-density lipoprotein

receptor repeats and one serine-rich segment were predicted

(Fig 5A) One strong transmembrane region is present

between amino acids 624 and 663 The EGF domains are

characterized by three typical intramolecular disulfide bonds [50], which are found in the sponge domains 2 and

3 with high similarity (Fig 5A) The EGF domains from

Fig 5 The putative epidermal growth factor (EGF) precursor, EGFl-PREC_SUBDO, from Suberites domuncula (A) The 714 amino acid poly-peptide comprises three potential low-density lipoprotein receptor repeats (LDL) and also three EGF-like domains (EGF) Within the EGF domains 1 and 2, the three characteristic intramolecular disulfide bonds (C ¼¼ C) are marked In addition, a Ser-rich segment (<Ser>) and a transmembrane region () exists The essential amino acid residues involved in binding to a receptor (#) and necessary for the biological function

of the factor (§) are marked in domain 2 (B) Unrooted tree constructed from the three sponge EGF domains (EGF_SUBDO) and the next similarity domains present in the pro-EGF precursor from human (EGF_HOMO, P01133), the bovine fibrillin 1 precursor (MP340) (FBN1_BOVIN, P98133; amino acids 2205–2229), the transforming growth factor-a precursor (TGF-a) from Ovis aries (sheep) (TGFA_SHEEP, P98135; amino acids 7–49), fibrillin 1 from the Cnidaria Podocoryne carnea (FBN1_PODCA, AAA91336; amino acids 443–487), rat cubilin (CUBIL_RAT, NP_445784; amino acids 2819–2864), frog (Xenopus laevis) neurogenic locus notch protein homolog precursor (NOTC_XENLA, P21783; amino acids 1735–1780), and the hypothetical protein ZC116.3 from Caenorhabditis elegans (ZC116_CAEEL, CAA98952; amino acids 3000–3054).

S domunculacan be grouped to the calcium-binding

EGF-like domains (NCBI:cd00054.2, EGF_CA); calcium is

crucial for protein–protein interactions Comprehensive

studies have been performed with mammalian EGF to

elucidate the characteristic sites [51,52] All amino acid

residues essential for ligand–receptor interaction and for

biological activity are present in domain 2 (Fig 5A)

Phylogenetic analysis The three EGF domains from

EGFI-PREC_SUBDO were aligned with the most closely

related EGF domains found in the proteins exclusively from

metazoans The unrooted tree shows that the two sponge

EGF domains (2 and 3) share highest similarity to the

domains present in the human EGF (P01133), the bovine

fibrillin 1 (P98133) and the Podocoryne carnea (Cnidaria)

fibrillin 1 (AAA91336) More closely related to the sponge

EGF domain 3 are the domains present in the transforming

growth factor-a from sheep (P98135), the notch/xotch

protein from frog (P21783) and a hypothetical protein

(ZC116.3) from Caenorhabditis elegans

Upregulation of gene expression for the described genes

in response to curdlan: Northern blotting

To assess the effect of curdlan and its subsequent binding to

the cell surface [probably to the (1fi3)-b-D-glucan-binding

protein] on the expression of the gene coding for this

receptor, Northern blot experiments were performed The

results revealed that the expression level of the

glucan-binding protein at the beginning of the experiments is low

However, after only 1 day of incubation in the presence of

10 lgÆmL)1curdlan, a strong upregulation of the expression

is seen, which increases during the following 2 days (Fig 6)

Parallel experiments were performed to determine the expression of the SDFIBI gene (fibrinogen-like protein) Again, in the absence of curdlan, no transcripts can be detected by this technique, while, in the presence of the polysaccharide, a strong increase in the expression of SDFIBIoccurs Likewise, a strong expression pattern was determined for the SDEGFI-PREC gene (Fig 6) The level

of expression of the housekeeping gene, tubulin, was not altered by the presence of curdlan

In view of the above finding, that curdlan causes phosphorylation of Tyr residue(s) in polypeptides of

S domuncula, which is prevented by aeroplysinin, it seemed necessary to determine whether this protein tyrosine kinase inhibitor also has an effect on the pronounced increase of expression of the three genes under study Therefore, the tissue was incubated with 10 lgÆmL)1 curdlan, together with 1 lgÆmL)1aeroplysinin In this co-incubation experi-ment it was evident that the inhibitor, aeroplysinin, completely prevented any upregulation of the expres-sion of SDGLUBP [(1fi3)-b-D-glucan-binding protein], SDFIBI(fibrinogen) or SDEGFI-PREC (EGF precursor) (Fig 6)

Identification of cells expressing glucan-binding protein and fibrinogen

Data from the literature [12,13,22], as well as the binding studies reported here, suggest that cells expressing the (1fi3)-b-D-glucan-binding protein are located in areas where the polysaccharide contacts the tissue, the pinaco-derm A similar localization, expression of the (1fi3)-b-D -glucan-binding protein in the cells of the pinacoderm, can be expected for fibrinogen or the related molecules, e.g ficolins, which are involved in host defense [48] Therefore, in situ hybridization studies were performed

Our studies revealed that the cryosections from tissue which had not been incubated with curdlan showed no cells which hybridized with the antisense probes, either for the sponge (1fi3)-b-D-glucan-binding protein (Fig 7A, a) or for fibrinogen (Fig 7B, a) However, after only 1 day of incubation with curdlan, the cells that are primarily located around the canals react with the antisense probes (Fig 7A, b; Fig 7B, b) After a longer incubation (for 3 days) the density of the hybridizing cells increased considerably (Fig 7A, c; Fig 7B, c) In contrast, no reaction was observed if the cells were treated with both sense probes (data not shown)

Level of low-molecular-weight EGF in tissue after treatment with curdlan

As a first step, antibodies were required that could identify the (mature) EGF product and the approximate level in tissue by Western blotting The recombinant protein was prepared in E coli using the cDNA (clone SDEGFI-PREC) spanning the three EGF domains (corresponding to amino acids 321–713) After induction with isopropyl thio-b-D -galactoside, the bacterial extract was isolated and purified

by affinity chromatography (Fig 8A; lanes a and b) The

68 kDa recombinant fusion protein (r-EGF_SUBDO) was used to raise PoAbs, as described in the Materials and methods After cleavage with enterokinase, the sponge

Fig 6 Steady-state expression of the SDGLUBP gene [(1fi3)-b- D

-glucan-binding protein], the SDFIBI gene (fibrinogen-like protein) and

the SDEGFI-PREC gene (epidermal growth factor EGF precursor) in

tissue from Suberites domuncula after exposure to 10 lgÆmL)1curdlan.

The housekeeping gene, b-tubulin, SDTUB (accession number

AJ550806), of S domuncula was used as an internal standard In one

series of experiments the tissue samples were co-incubated with

1 lgÆmL)1of aeroplysinin (aero) RNA extraction was performed 0, 1

or 3 days after incubation with curdlan Equal amounts were loaded

onto the gel The RNA was size separated, blot transferred and then

hybridized with the labeled probes, as described in the Materials and

methods.

recombinant protein, r-EGF_SUBDO, showed an expected

molecular weight of 41 kDa (data not shown)

This PoAb, PoAb-EGF precursor, was used for the

Western blot experiments Tissue extract from S domuncula

was prepared (Fig 8B; lane a) and used for the blotting

studies In nontreated, as well as in curdlan-treated

extracts from tissue, a strong band was detected using the

PoAb-EGF precursor, which corresponded to a molecular weight of 5 kDa This molecular weight predicts a poly-peptide of approximately 45–50 amino acids, which matches exactly a processed form of the EGF from the 78 kDa EGF precursor (Fig 8B; lanes b and c) No further major band, including the 78 kDa EGF precursor, was detected In the control experiment, using PoAb-EGF precursor which had been adsorbed with recombinant r-EGF_SUBDO, no band was seen (Fig 8B; lanes d), indicating that the immune reaction with the 5 kDa protein is specific The relative expression value for the band corresponding to the 5 kDa polypeptide was assessed This approach revealed that in curdlan-exposed tissue, a threefold higher level of the EGF exists

Discussion Based on the known symbiotic relationship between spon-ges and microorganisms, such as bacteria and fungi, it can

be deduced that these animals must have an efficient recognition system which is able to discriminate between self and symbiont On the next level of specificity, the sponges must be provided with pattern recognition molecules serving

as biosensors for the detection of invading pathogens (parasitic), of commensalic (of benefit only for one partner)

or of mutualistic organisms (benefit for both partners) It has been shown, in sponges, that some microbial secondary metabolites, e.g okadaic acid [4], are beneficial for the host However, the origin of most secondary metabolites identi-fied in sponges is not clear The understanding of the pathways which result in the synthesis of these compounds

in sponges is crucial for their sustainable production/ exploitation for human benefit [53,54]

It is now established that sponges can recognize bacteria and react to them with an increased phosphorylation of key kinases of the MAPK pathway [7,28] Furthermore, the first genes encoding antibacterial proteins, such as perforin [55]

or the lectin tachylectin [6], have been identified in the

Fig 8 Level of epidermal growth factor (EGF) in tissue after treatment

with curdlan (A) Antibodies against the EGF precursor were prepared

from the recombinant protein expressed in Escherichia coli (A, a;

stained with Coomassie Brilliant Blue) The purified recombinant

41 kDa polypeptide (A, b; stained with Coomassie Blue) was used for

immunization (B) The polyclonal antibody (PoAb)-EGF precursors

were used to identify the protein in extracts from sponge tissue The

extracts were size separated and the gels stained with Coomassie

Brilliant Blue (B, a) Then, the proteins were blot transferred and

reacted with PoAb-EGF precursor The level of cross-reacting proteins

was assessed by molecular imaging, as described in the Materials and

methods (B, b and c) In one series (B, d) the PoAb-EGF precursor was

adsorbed with recombinant r-EGF_SUBDO prior to the application.

The immunocomplex has been visualized by a labeled secondary

antibody For further data see the Materials and methods.

Fig 7 Spatial expression pattern of (1fi3)-b- D -glucan-binding protein and fibrinogen in sections of Suberites domuncula Cryosections were per-formed of S domuncula tissues, which were then hybridized with DIG-labeled SDGLUBP antisense DNA (A), or SDFIBl antisense DNA (B) Subsequently, the samples were incubated with antidigoxigenin/alkaline phosphatase and the signals detected with Nitro Blue tetrazolium/X-Phosphate, as described in the Materials and methods Tissue samples, which were untreated (a), or treated with 10 lgÆmL)1of curdlan for 1 day (b), or 3 days (c), were analyzed The canals (ca) of the aquiferous system are lined by an epithelial layer formed from pinacocytes Magnifications: A-a and B-a, · 25; A-b and B-b, · 50; A-c and B-c, · 100.