Evaluation of the ecological function of amphitoxin in the reef-dwelling sponge Callyspongia (Euplacella) biru (Haplosclerida: Callyspongiidae) at southwest Sulawesi, Indonesia

The ecological function of the secondary metabolite amphitoxin produced by Callyspongia (Euplacella) biru is evaluated by a forced confrontation with the free­living scleractinian coral Fungia fungites at different exposure times. Our major goal was to determine whether such a forced confrontation with a spatial com­petitor would cause a significant change in the concentra ­ tion of amphitoxin in the sponge tissue, indicating a regulatory mechanism of amphitoxin production. Firstly, the concentra­ tions of amphitoxin of sponge fragments submitted to forced confrontation with a mushroom coral did not differ between the exposure times. Secondly, all sponge fragments, including the controls had a consistently lower amphitoxin concentration dur­ ing the course of the experiment than the natural variation. Third­ ly, the concentration of amphitoxin varied substantially between specimens implying that other ecological factors besides spatial competition regulate the production of bioactive compounds of C. biru.


Introduction
In recent decades, marine invertebrates have attracted increasing interest as sources of bioactive metabolites from pharmaceutical companies and research institu tions. Although more than 14,000 distinct natural products have been discovered from marine inverte brates, only few marinederived compounds have passed pre and clinical trials (Proksch et al., 2002;Mendola, 2003). In most cases the compoundproduc ing marine organisms were haphazardly collected, without an understanding of the ecological function of these compounds. Recently, the question has arisen whether the function of these compounds could be deduced from field observations in order to collect only specimens of species that are expected to pro duce metabolites of pharmaceutical importance (Uriz et al., 1992;Proksch et al., 2002). Field observations can also be used to evaluate the biological aspects of the production of bioactive compounds for the fea sibility of eventual mass production (Duckworth and Battershill, 2003). For instance, the absence of foul ing organisms on softbodied invertebrates might indicate that the bioactive compounds in these inver tebrates probably have potential antibacterial and/or fouling properties (Amade et al., 1987;Kelman et al., 2001). In addition to antifouling properties, the bio active compounds may also act to deter predators, as shields for UVprotection, and as means for substrate acquisition (Aerts and van Soest, 1997;Assmann et al., 2000;de Voogd et al., 2004). Space is a highly restricted resource for sessile organisms when sub strate availability is limited, such as in coral reefs (Dayton, 1971;Beccero et al., 2003). Extensive stud ies have assessed competitive mechanisms employed by scleractinian corals (Lang, 1973;Lang and Chor nesky, 1990), corallimorpharians (Chadwick 1991;Chadwick and Adams, 1991), actiniarians (Bak and Borsboom, 1984), alcyonaceans (Benayahu and rarely been isolated or identified (Thompson et al., 1985;Thacker et al., 1998). Aerts and van Soest (1997) remarked that some sponges might be more aggressive towards corals in localities with high coral cover. Highly aggressive corals in the Caribbean are relatively minor reef components (Lang, 1973) while in the Indo-Pacific many of the most aggressive corals dominate (Sheppard, 1979;Dai, 1990). Nev ertheless, in the Pacific, Rützler and Muzik (1993) reported spatial dominance and aggression in over growing corals of the sponge Terpios hoshinota in Japan and Guam.
Ecological studies on sponges producing this com pound revealed deterrent properties towards several sponge predators and it was suggested that amphi toxin acts as an antifeeding agent. (Albrizio et al., 1995;Pawlik et al., 1995;Burns et al., 2003). The selective toxicity of amphitoxin produced by the Red Sea sponge Amphimedon viridis Duchassaing and Michelotti, 1864 against different bacterial strains may keep this sponge free from microbial pathogen esis (Kelman et al., 2001). However, de Voogd et al. (2004) proposed that amphitoxin might play a role in competition for space with scleractinian corals. Loya, 1981), zoanthids (Suchanek and Green, 1982), bryozoans (Buss, 1979;Turner and Todd, 1994), and algae (Dai, 1991). Some of these mechanisms in clude: sweeper tentacles and mesenterial filaments (Bak et al., 1982;van Veghel et al., 1996), overgrow ing and smothering (Cope 1982), and exudation of toxic substances (Rinkevich and Loya, 1983;Sam marco et al., 1983). Very few of these studies include sponges (Jackson and Buss, 1979;Buss, 1990). Sponges are a particularly rich source of secondary metabolites with antimitotic, cytotoxic, antibacterial, antifungal and/or antiviral properties (Faulkner, 1984;Munro et al., 1999). The ability of sponges to gain and hold space depends mainly on their com petitive ability (Aerts and van Soest, 1997). Spatial competition can be visibly determined as overgrowth, necrosis, and bleaching of the interacting neighbour (Aerts, 2000). A few studies have suggested that waterborne biologically active metabolites may deter particular substrate competitors from growing in di rect contact with sponges (Turon et al., 1996;Nishi yama and Bakus, 1999), but these chemicals have In the present study we evaluate the ecological function of amphitoxin produced by Callyspongia biru by a forced confrontation with a freeliving scle ractinian coral. Our major goal was to determine whether such a forced confrontation with a spatial competitor would cause a significant change in the concentration of amphitoxin in the sponge tissue, indicating a regulatory mechanism of amphitoxin production.

Experimental set-up and research site
This study took place at the northwest side of the fringing reef of Samalona island (05º07.326'S 119º20.410'E) in the Spermonde Archipelago, 7 km off Makassar, SW Sulawesi, Indonesia from Septem ber to November 2002 (Fig. 1).
The sponge Callyspongia (Euplacella) biru de Voogd, 2004 (Callyspongiidae; Haplosclerida) was selected because its bioactivity and secondary me tabolites are known. (Albrizio et al., 1995: unpublished 'Symbiosponge' data www.science.uva.nl/ZMA/ Invertebrates/Symbiosponge). Three types of ex periments were performed, prior to the current ex periment, to evaluate what environmental parameters might induce a change in bioactivity of the sponge crude extract. Sponge fragments were confronted with corals, injured repeatedly to simulate predator bites, and transplanted from deep to shallow water and vice versa to evaluate the impact of UVradiation. A change in bioactivity was detected with the Artemiatoxicity test (for a description of the Artemia toxicity test see de Voogd, 2005). This test is an ef fective assay method to detect cytotoxic activity of secondary metabolites from terrestrial and marine organisms (Meyer et al., 1982;RichelleMaurer et al., 2003), it has not been used to detect changes in bioactivity so far. Nevertheless, a change in bioactiv ity (mortality of Artemia nauplii ≥ 50% exposed to the crude extract of C. biru varying between 1 mg/L and 10 mg/L) was revealed in the experimental con frontation of the sponge with a coral (Bavinck, 2001;Becking, 2002); thus this experimental setup was chosen for quantification of amphitoxin. Twenty branch fragments of approximately 12 cm length of living sponge were collected at a depth of 12 m and attached to a frame with electric cable wire with a protective plastic sheath at the same depth. Frames were made of PVC tubes of a rectangular form (70  100 cm) on which nylon fishing net had been tightened. The freeliving coral Fungia fungites Lin naeus, 1758 (Scleractinia: Fungiidae; see Hoeksema, 1989) is known to be an aggressive competitor for space (Hildemann et al. 1975; Thomason and Brown, 1986). After the sponges were allowed to acclimatize for a period of four weeks, 16 individuals of the coral were collected and placed on the frames at a distance of 0.51 cm to an equal number of sponge fragments. In a similar manner to the sponges, the corals were tied to the frame with cable wire to keep the neighbouring coral and sponge in place (Fig. 2). Four sponge fragments were used as controls, and these were not confronted with the coral. Two of these fragments were collected at the beginning of the experiment and the remaining two fragments were left on the frame without coral confrontation, until their collection at the end of the experiment. Of the 16 sponge fragments confronted with the corals, two fragments were collected at eight different time intervals; T1 = ½ hour, T2 = 1 hour, T3 = 4 hours, T4 = 24 hours, and T5 = 72 hours, T6 = 8 days, T7 = 19 days and T8 = 28 days. For the natural varia tion, three sponge fragments of approximately 12 cm in length were collected from different depths at different locations at the reef surrounding the island of Samalona.

Extraction and isolation
The sponge fragments were measured (length in cm; growth in thickness was not observed) and extracted in ± 50 ml pure methanol for 24 hours at room temperature immediately upon collection. Prelimi nary investigation with this sponge showed that the methanolic extract was found to be active and con tained the 3alkylpiperidine alkaloids. We therefore only used pure methanol for extraction. The sponges were cut into smaller pieces and were washed twice in succession with the solvent to ensure an exhaustive extraction. All the filtrates were combined, and placed in a cooled evaporator. The crude extract was col lected and weighed after approximately 24 hours of evaporation and stored at 7°C. Detection of the 3 alkylpiperidine alkaloids was determined using Thin Layer Chromatography.

Thin Layer Chromatography
Although 3alkylpiperidine alkaloids have previ ously been isolated from C. biru, the presence of these alkaloids was verified for several sponge frag ments prior to HPLC-quantification. A capillary tube was held in the methanolic sponge extract and care fully spotted five times onto 0.25 mm Si-gel coated sheets (5  2 cm). The plates were eluted with a solvent mixture containing water / acetonitrile (7:3) + 0.1% trifluoroacetic acid. After the completion of the chromatography, the dried plates were sprayed with a solution of Dragendorff reagent. 3alkylpipe ridine alkaloids can be detected as one of the major compounds present in the extract and appears as a bright orange spot.

Quantification of amphitoxin
Analysis of amphitoxin was performed by highper formance liquid chromatography (HPLC) with a  ently lower mean amphitoxin concentration per dry weight than the natural variation (Fig. 5A). The per centage amphitoxin varied from 0.040.07% of the dry weight of the fragments, whereas the percentage for the natural variation varied from 0.060.20% of the dry weight. Also, the mean percentage crude extracts / dry weights were consistently lower than that of the natural variation. All sponge fragments, including the natural variation, exhibited a large in terindividual variability. The mean crude extract per dry weight of the natural variation varied between 4248%, whereas this was only 2637% for the frag ments of the experimental setup (Fig. 5B). The sponge fragments were measured after transplanta tion and after collection, and all sponge fragments showed at least 10% growth in a period of at least four weeks. White (bleached) spots in the area adjacent to the sponges were observed in some of the indi viduals of Fungia fungites after several days.

Discussion
In the present study we evaluated the ecological function of the major secondary metabolite, amphi toxin, produced by Callyspongia (Euplacella) biru by confronting this species with the space competitor Fungia fungites. Importantly, all sponge specimens analysed with HPLC in this study were found to contain amphitoxin. Amphitoxin is known to be Waters 600E system controller equipped with a Wa ters 717 plus autosampler and a Waters 848 Tunable Absorbance Detector. Crude extracts from the sponge fragments were dissolved in a mixture (70:30) of water and acetonitrile + 0.1% trifluoroacetic acid and 10 μl was injected by autosampling into an HPLC system. The column used was a Lichrospher RP18 (125 by 2.0 mm with a 3 µm pore size) and the mo bile phase consisted of acetonitrile (70%) and nano pure water (30%) with trifluoroacetic acid (0.1%) at a flow rate of 1 ml/min. Amphitoxin was detected at its UVmaximum of 266 nm. The quantitative analy sis was based on peak area calibration using earlier purified amphitoxin (Fig. 3). We compared our data with spectral data reported in the literature (Albrizio et al., 1995). To calculate the percent yield of amphi toxin, we calculated the mass amphitoxin divided by the mass of dried sponge, multiplied by 100. The percentage yield of the total crude extract was calcu lated as the mass of the extract divided by the mass of dried sponge, multiplied by 100.

Results
The concentrations of amphitoxin / dry weight of sponge fragments submitted to forced confrontation with a coral did not differ between the time intervals. All sponge fragments, including the controls (0d and C28d), attached to the PVCframe had a consist mainly responsible for the bioactivity of the crude extract of the Indonesian sponge C. biru (Dubut, 2000). The mean percentage amphitoxin concentra tion per dry weight was consistently lower for all sponge fragments, including the control fragments, during the course of the experiment than the amphi toxin concentration of the natural variation. In a similar experiment to ours, the Caribbean sponge Agelas conifera Schmidt, 1870 was experimentally confronted with the scleractinian coral Madracis sp., and no significant change in bioactive compound concentration was observed. In the same study, in flicted sponge damage induced a fourfold rise in concentration, and yielded detectable amounts of the Agelas alkaloids in the surrounding seawater, sug gesting allelopathic activity (RichelleMaurer et al., 2003). Although a detection analysis of the chemicals in the surrounding seawater was not performed in the present study, the lower concentrations of amphi toxin in the experimental sponge fragments might imply allelopathic activity. Several studies (Porter and Targett, 1988;Nishiyma and Bakus, 1999;Engel and Pawlik, 2000) suggested that water borne allelo chemicals might deter substrate competitors from growing in direct contact with sponges. Sullivan et al. (1983) isolated secondary metabolites excreted within mucus from an excavating sponge, which inhibits coral growth. Turon et al. (1996) suggested that the sponge Crambe crambe Schmidt, 1862 re leases waterborne chemicals and may cause tissue necrosis to adjacent organisms at small scales.
Certain tolerant coral species, however, may grow adjacent to sponges, whereas others are rarely or never found in close proximity. Nishiyama and Bakus (1999) demonstrated that three sponges released allelochemicals into the surrounding seawater, and were toxic to several scleractinian coral species. Although they did not quantify the secondary me tabolites in the sponge tissue, these results suggest the release of compounds into the seawater. It is pos sible that the release of allelochemicals into the sur rounding seawater of the confronted sponges in the present study might trigger nonconfronted proximate individuals to release their toxins as well, and might explain the lower concentrations of amphitoxin in the control sponges. Besides, it is known that certain green algae rapidly transform stored nonactive compounds into very reactive toxins upon mechani cal wounding (Paul and Van Alstyne, 1992;Jung and Pohnert, 2001). However, it is not known in sponges, whether such enzymatic transformations take place upon wounding or during other stress events. Not withstanding, the sponges exhibited a large interindi vidual variation, indicating a regulatory mechanism for the production of amphitoxin. Thacker et al. (1998) found no evidence that the production of secondary metabolites was induced by the presence of a spatial competitor, rather they suggested that the constant threat of predators may maintain high con centrations of the compound. They also observed a large variation in the production of secondary me tabolites, which according to them might be due to differences in nutrient availability or genetic differ ences. Thus an increase or decline in the production of secondary metabolites may not be so clearly at tributed to the presence of spatial competitors or predators. On the other hand, the ecological role and strength of pure metabolites, in this case amphitoxin, may be quite different from the total crude extract. From unpublished records, it was hypothesized that amphitoxin was responsible for the general toxicity of this particular species (Artemiatoxicitytest). Besides, the mean percentage crude extract per dry weight was consistently lower from the manipulated sponge fragments than those of the natural variation. However, it has been suggested that different me tabolites present in a species may interact and produce different bioactivities when tested together or sepa rately. Although no other major metabolites have been isolated from this sponge species, other chem ical products, even salts, may have a synergetic effect on its bioactivity (Marti et al., 2003).
The optimal defense theory assumes that the pro duction of defensive secondary metabolites is costly and consequently would repress growth and repro duction (Schupp et al., 1999). In the present study, all sponge fragments did grow at least 10% during the experimental period, and upon collection, larvae were observed in the tissue of several sponge frag ments. RichelleMaurer et al. (2003) suggested that there is always a minimal production of bioactive compounds possibly at the boundary of efficiency. Thus, it is probable, that wounded individuals invest their limited available energy in healing and growth rather than in the production of secondary metabo lites. However, some of the individuals of F. fungites were observed with white spots during the course of the experiment confronted with C. biru suggesting exudation of secondary metabolites and competitive dominance of the sponge, although other enviromen tal stressors may have induced bleaching of the cor als (cf. Hoeksema, 1991). Scleractinian corals are in general sedentary animals, however most mushroom corals (Fungiidae) have a freeliving phase in their life cycle (Hoeksema, 1989). Mobility may help mushroom corals to disperse and reach favourable habitats, to prevent burial in soft sediments or to function in avoiding competition for space (Hoek sema, 1988(Hoek sema, , 1993Chadwick, 1988;ChadwickFur man and Loya, 1992). Fungiid corals, however, have been shown to competitively dominate nonfungiid scleractinians during interactions in physical contact experiments (ChadwickForman and Loya, 1992). The thick mucus secreted by most unattached mush room corals may have a protective function in ag gressive reactions during accidental contact with other sedentary organisms (Hildemann et al., 1975;Chadwick, 1988). Competitive interactions with other sessile invertebrates though, are only known with octocorals, where the interacting organisms appear not to harm each other (ChadwickForman and Loya, 1992). Also, observations of natural inter actions of C. biru with scleractinian corals demon strated necrosis in more than 90% of overgrowth interactions . In conclusion, although a forced confrontation of the sponge C. biru with the scleractinian F. fungites showed a consist ently lower amphitoxin concentration compared to the natural variation, this may not be so straightfor ward be attributed to a competitive regulatory mechanism induced by the spatial competitor. Rather, the concentration of amphitoxin varies sub stantially between specimens implying that other ecological factors besides spatial competition interact to regulate the production of bioactive compounds of C. biru. More importantly, in scope of eventual mass production, manipulated sponges might produce consistently lower concentrations of the target bioac tive compound than sponges in their natural environ ment.