Metabolite comparisons and the identity of nutrients translocated from symbiotic algae to an animal host

Many corals, sea anemones and other benthic marine animals bear symbiotic dinoflagellate algae of the genus Symbiodinium, also known as zooxanthellae. These symbioses have a nutritional basis, involving the sustained and substantial bidirectional translocation of nutrients between living cells of the algal and animal partners. In particular, the algal cells photosynthesize at high rates and release much of their photosynthetically fixed carbon to the animal tissues, making a substantial contribution to the carbon and energy budgets of the animal partner (e.g.Muscatine et al., 1984;Edmunds and Davies, 1986).

The nutrients translocated between the partners in a symbiosis are referred to here as mobile compounds. This study concerns the identity of the mobile compounds derived from the photosynthetically fixed carbon pools of the algal cells. This topic is technically challenging for two reasons. Firstly, the Symbiodinium cells in most associations are intracellular,specifically located in the cytoplasm of animal cells and individually enclosed within an animal membrane termed the symbiosomal membrane(Wakefield et al., 2000). Secondly, it is expected that the mobile photosynthetic compounds are metabolized rapidly on receipt by the animal cytoplasm, thereby limiting their backflow to the algal cells (Douglas,1994). It is widely accepted that the mobile photosynthetic compounds include low-molecular-mass compounds such as glycerol, glucose and organic acids. The principal evidence is that these are the compounds released into the medium by Symbiodinium cells freshly isolated from certain symbioses and incubated with extract of the host animal(Trench, 1971a;Hinde, 1988). Exceptionally,in the tridacnid clam–Symbiodinium association, the mobile compounds can be identified relatively easily in the intact symbiosis because the algal cells are extracellular. In this system, the dominant photosynthetic compound released from Symbiodinium cells is glucose in the symbiosis but glycerol in isolated cells (Streamer et al., 1988; Rees et al.,1993; Ishikura et al.,1999). This discrepancy suggests that it may not be justified generally to equate the compounds released by isolated algal cells and the mobile compounds in the intact symbiosis. Compounding these difficulties, the Symbiodinium cells in some animal hosts, e.g. Zoanthus robustus and Anemonia viridis, release markedly more photosynthetic compounds to the animal tissues in the intact association than after isolation and incubation with host homogenate(Sutton and Hoegh-Guldberg,1990; L. Whitehead, J. T. Wang and A. E. Douglas,unpublished).

Various approaches have been adopted to study photosynthate release from Symbiodinium cells in the intact symbiosis, thereby circumventing the limitations of studies with isolated algae. These approaches include identification of photosynthetically fixed ¹⁴C- and ¹³C-labelled compounds in the animal partner(Trench, 1971b;Battey and Patton, 1984;Johnston et al., 1995),measurement of respiratory quotients(Gattuso and Jaubert, 1990) and identification of organism-specific compounds(Harland et al., 1991). Although these studies were not designed specifically to identify the mobile photosynthetic compounds, they collectively suggest that intact lipids,glycerol and fatty acids may be translocated from Symbiodinium cells to the animal tissues. These data are partly, but not completely, consistent with the candidate mobile compounds identified from studies with isolated algal cells (see above).

The purpose of the present study was to apply the approach of `metabolite comparisons', summarized in Fig. 1, to explore the identity of mobile photosynthetic compounds in an intact symbiosis. The two treatments are: (1) the `control animals'incubated with NaH¹⁴CO₃ under photosynthesizing conditions and (2) the `experimental animals' incubated with one of a panel of ¹⁴C-labelled organic substrates under non-photosynthesizing conditions. The radioactively labelled compounds in the animal fraction are the products of animal metabolism of the mobile compounds and of the exogenously applied organic compounds for the control and experimental animals, respectively. The exogenous organic substrates that generate the same pattern of labelled metabolites in the animal fraction as that obtained with the NaH¹⁴CO₃ incubations can, consequently, be interpreted as the mobile photosynthetic compound (or a compound metabolically closely allied to the mobile compound). By this approach, candidate mobile photosynthetic compounds have been identified in the symbiosis between the sea anemone Anemonia viridis and algae of the genus Symbiodinium. This association has been the subject of a number of physiological and metabolic studies (e.g.Taylor, 1969;Schlichter, 1978;Tytler and Davies, 1984;Stambler and Dubinsky, 1987;Harland et al., 1992;Davy et al., 1997;Furla et al., 1998;Roberts et al., 1999) but the mobile compounds have not been identified.

Anemonia viridis (Forskal), green colour morph, were collected from a tide pool at Morfa Nefyn, North Wales, UK (Ordinance Survey grid reference 274, 410) and maintained in aerated artificial seawater (Instant Ocean Salts; Aquarium Systems, Sarrebourg, France) of salinity 35‰ and pH 8.2 at 16°C under a 16 h:8 h light:dark regime at 10–30 μmol photons m^–2 s^–1 photosynthetically active radiation (PAR). Experiments were conducted 3–5 days after the animals had received their weekly feed of chopped squid and were performed under the same light and temperature conditions as for anemone culture.

Replicate groups of three freshly excised tentacles of A. viridis(taken from three separate individuals) were incubated in 1 ml seawater buffer, comprising 420 mmol l^–1 NaCl, 26 mmol l^–1 MgSO₄, 23 mmol l^–1MgCl₂, 9 mmol l^–1 KCl, 9 mmol l^–1CaCl₂, 2 mmol l^–1 NaHCO₃ and 10 mmol l^–1 Hepes, pH 8.2 (Wang and Douglas, 1997), for 30 min prior to an experiment, by which time the tentacles had relaxed. Where appropriate, algal photosynthesis was inhibited by 5 μmol l^–1 DCMU (dichlorophenyl dimethyl urea) added 15 min before the experiment. Experiments were initiated by adding radiolabelled substrate (details below) and, unless stated otherwise,comprised a 5-min time course, to minimize the extent of animal metabolism of the radioactive substrate whilst generating sufficient radioactivity for quantification. Samples without added radioactivity were included in all experiments as controls.

After the incubation with 0.37 MBq (10 μCi)NaH¹⁴CO₃ (ICN Pharmaceuticals, Costa Mesa, CA, USA), the tentacle samples were rinsed in ice-cold 0.5 mol l^–1 NaCl and homogenized in a hand-held glass homogenizer in 1 ml ice-cold 0.5 mol l^–1 NaCl. The protein content of each homogenate was quantified using the Bio-Rad protein micro-assay kit (Bio-Rad Laboratories GmbH, Munchen, Germany) according to the manufacturer's instructions, with bovine serum albumin as standard. The homogenate was centrifuged at 1000 g for 5 min at 4°C, and the pellet was washed twice by centrifugation and resuspension with 1 ml 0.5 mol l^–1 NaCl. The supernatants were combined to form the animal fraction, and the final suspension of algal cells was designated the algal fraction. The cell density in the algal fraction was quantified using an improved Neubauer haemocytometer(six replicate counts per sample). To quantify ¹⁴C incorporation,25 μl of each fraction was acidified with 4 mol l^–1 acetic acid and shaken for 60 min to remove unfixed ¹⁴C. Scintillation cocktail (Ultima-Gold™ XR; Packard Bioscience B.V., Gröningen, The Netherlands) was then added, and the radioactivity quantified in a scintillation counter (Packard Tri-Carb Scintillation Analyser) using pre-set ¹⁴C windows. The counts in the samples incubated without added radioactivity were subtracted from values for the experimental samples.

Samples of the algal and animal fractions were incubated in 5%trichloroacetic acid (TCA) for 15 min at room temperature and then centrifuged at 10 000 g for 15 min. The supernatant (TCA-soluble fraction)was extracted three times with ether to remove the TCA, evaporated to dryness in a Speed-Vac (SC110; Savant Instruments, Holbrook, NY, USA) and then resuspended in 80 μl and 200 μl of deionised water for algal cell and animal samples, respectively. (The animal samples required a larger volume to dilute the salt; ethanol extraction was not used to remove the salt because this procedure substantially reduced the yield of radioactivity.) The lipid fraction was extracted from the TCA-insoluble fraction with 1 ml methanol:chloroform (2:1) for 15 min at room temperature followed by centrifugation at 10 000 g for 15 min. The chloroform layer containing the lipid was separated from the methanol by the addition of 0.4 ml 0.1 mol l^–1 KCl. The final pellet that contained protein and nucleic acids was solubilized by incubation in 0.2 ml 1 mol l^–1 NaOH at 100°C for 10 min and then neutralized with 0.2 ml 1 mol l^–1 HCl. Samples of each fraction were removed for the quantification of radioactivity. The recovery of radioactive compounds after fractionation was, on average, 90% and 65% for the algal and animal samples, respectively. It is unclear why recovery was lower for the animal samples but it seems unlikely to be due to the salt content as similar recovery figures were achieved after salt removal by Streamer et al.(1988).

The TCA-soluble fraction was further separated into neutral, acidic and basic fractions using cation (Dowex 50 × 8 H⁺; Sigma-Aldrich,Dorset, UK) and anion (Dowex 1 × 8; Sigma-Aldrich) exchange columns,following the procedure of Quick et al.(1989). To identify individual radiolabelled compounds, samples of the TCA-soluble fraction were separated by thin layer chromatography/thin layer electrophoresis, as in Wang and Douglas(1997), for organic acids,phosphate esters, sugars and sugar alcohols and as in Wang and Douglas(1999) for amino acids. The chromatography/electrophoresis plates were exposed to Kodak X-ray film for 4–6 weeks, and radioactive spots were identified by comparison with authentic compounds. The identification of the sugars, sugar alcohols and amino acids was verified by reverse-phase HPLC using the procedures of Ashford et al. (2000) and Karley et al.(2002).

The analyses were performed as for NaH¹⁴CO₂incubations with the following modifications. All experiments were performed under non-photosynthesizing conditions (in the presence of 5 μmol l^–1 DCMU) in autoclaved seawater buffer supplemented with the antibiotics ampicillin (1 mg ml^–1) and streptomycin (50 μg ml^–1) to inhibit the metabolism of any contaminating bacteria(Wang and Douglas, 1998). Organic substrates {[U-¹⁴C]glucose, [U-¹⁴C]glycerol,[2,3-¹⁴C]succinate, [2,3-¹⁴C]fumarate (Sigma-Aldrich),[U-¹⁴C]malate and [1,5-¹⁴C]citrate (Amersham Pharmacia Biotech UK Ltd, Amersham)} were supplied at a concentration of 1 mmol l^–1 and 0.11 MBq ml^–1 for single substrate incubations and 0.5 mmol l^–1 and 0.06 MBq ml^–1 for double substrate incubations (i.e. giving a total concentration of 1 mmol l^–1 and 0.11 MBq ml^–1). The chemical purity of radiochemical compounds was>99%, as determined by thin layer chromatography. The experimental design was developed following preliminary experiments on the uptake of organic substrates by tentacle samples over time from 5 s to 5 min, which revealed minimal nonspecific binding of ¹⁴C with >98% recovery of radioactivity from the anemone fraction, i.e. <2% in the algal cells. Algal cells and animal tissue were therefore not separated in the definitive experiments. The experimental material was homogenized in ice-cold deionised water, and the TCA-soluble fractions, which were essentially salt-free, were resuspended in 80 μl volumes after evaporation.

The rate of NaH¹⁴CO₃ fixation by the excised tentacles of A. viridis in the light was 3.70.29×10⁵d.p.m. mg^–1 total protein h^–1 (mean ± s.e.m., N=19), equivalent to 0.09±0.008 d.p.m. algal cell^–1 h^–1, and fixation was linear in incubations of 5–60 min duration with 30±1.5% of ¹⁴C recovered from the animal tissue. Fixation in the presence of 5 μmol l^–1 DCMU was, on average, 6% of that in samples without DCMU.

Incorporation of radioactivity into the principal chemical fractions of the algal cells and animal tissue after 5-min incubations is shown inFig. 2. The distribution of radioactivity (expressed as d.p.m. mg^–1 protein) across the chemical fractions varied significantly between the algal cells and animal tissues and was significantly altered by the DCMU treatment (statistical analysis shown in legend to Fig. 2). The two principal factors contributing to this variation were:(1) the greater incorporation of radioactivity into the TCA-soluble fraction of animal tissues (94% of the total) compared with algal cells (59%) under photosynthesizing conditions and (2) the disproportionate reduction in ¹⁴C incorporation into the lipid fraction in the DCMU-treated samples relative to those under photosynthesizing conditions. The radioactivity in the neutral, acidic and basic fractions of the TCA-soluble fraction is shown in Fig. 3,with statistical analysis in the legend. For the samples under photosynthesizing conditions, the distribution of radioactivity differed significantly between the algal cells and animal tissues, with recovery from the neutral, acid and basic fractions in approximate ratios of 2:1:2 for algal cells and 1:2:1 for the animal tissues. The distribution of radioactivity was also significantly altered by DCMU treatment.

Various compounds were radioactively labelled in both the algal cells and animal tissue of samples incubated with NaH¹⁴CO₃ under photosynthesizing conditions (Table 1). Differences between the algal cells and the animal tissue included the detection of ¹⁴C-labelled glycerol in the algal cells but not in the animal tissue, and intense labelling of malate in the animal tissue as opposed to citrate in the algal cells. The ¹⁴C content of DCMU-incubated samples was so low that identification of individual radiolabelled compounds was not feasible.

Tentacles of A. viridis incubated under non-photosynthesizing conditions with each of the six organic substrates (glucose, glycerol, malate,citrate, succinate and fumarate) accumulated radioactivity equivalent to between 2.74 nmol substrate mg^–1 total protein and 4.20 nmol substrate mg^–1 total protein over the 5-min incubation period(Table 2). The pattern of incorporation of radioactivity into the principal chemical fractions was very similar for all the organic substrates tested. Radioactivity was recovered predominantly from the TCA-soluble fraction (99%), with a small amount in the protein/nucleic acid fraction (approximately 0.8%) and barely detectable amounts (0.2%) in the lipid fraction. For all the organic substrates, the distribution of radioactivity between the neutral, acidic and basic fractions of the TCA-soluble fraction was different to that recorded from the control experiments where the symbiosis was supplied with NaH¹⁴CO₃ under photosynthesizing conditions(Fig. 4). For every substrate,radioactivity was detected almost exclusively in just two of the three TCA-soluble fractions: little radioactivity was recovered from the neutral fraction (≤0.7% of total ¹⁴C in the TCA-soluble fraction) when an organic acid was supplied or from the basic fraction when glucose (4% of the total) or glycerol (5% of the total) was supplied. For most organic substrates tested, the most strongly radiolabelled fraction corresponded to that of the substrate supplied. Exceptionally, 58% of the total radioactivity derived from exogenous ¹⁴C-labelled malate was detected in the basic fraction.

Individual radioactively labelled compounds recovered from tentacles that had been incubated in the various organic substrates are shown inTable 3, which includes the control data obtained for tentacles incubated with NaH¹⁴CO₃ (Table 1) for comparison. For all substrates tested, four ¹⁴C-labelled compounds – malate and three sugar phosphates– were detected in common with the control. The most strongly labelled compounds were those corresponding to the substrate supplied and to glucose 6-phosphate. In general, when an organic acid was supplied, a greater number of ¹⁴C-labelled compounds were detected in common with the control compared with when glucose and glycerol were supplied. However, certain ¹⁴C-labelled compounds detected in the control samples (notably proline) were only detected in the experimental samples supplied with ¹⁴C-labelled glucose.

When the tentacle samples were incubated with two ¹⁴C-labelled substrates simultaneously, the total uptake varied between 4.25 nmol mg^–1 total protein and 5.21 nmol mg^–1 total protein (Table 2). Incorporation of radioactivity into the principal fractions was very similar to single substrate incubations, with 99% of the radioactivity recovered from the TCA-soluble fraction. The pattern of incorporation that most closely resembled the control incubations with NaH¹⁴CO₃ was obtained with glucose and either succinate or fumarate as substrates(Fig. 5).

The profile of labelled compounds in double substrate incubations(Table 4) was identical to the combined profile of compounds labelled in previous experiments using the same substrates applied singly (Table 3). For the incubations with glucose and succinate or glucose and fumarate, the radiolabelled compounds were very similar to those labelled in the control experiments incubated with NaH¹⁴CO₃. The products of double substrate incubations that included glycerol were also similar to the results from the control experiments, but two of the most strongly labelled compounds, glycerol and an unknown amino acid, in these double substrate incubations were not detected in the control incubations.

In summary, the profile of metabolites in the animal tissues derived from exogenous glucose and succinate/fumarate, but no other exogenous compounds tested, was closely similar to the equivalent profiles derived from photosynthetically fixed compounds translocated from the algal cells. This comparison of metabolite profiles (see Fig. 1) suggests that glucose and a dicarboxylic acid (probably succinate and/or fumarate), or metabolically allied compounds, are important mobile photosynthetic compounds in the Anemonia viridis–Symbiodinium symbiosis.

A fundamental feature of many symbiotic microorganisms is that they provide the host with nutrients that would otherwise enhance their own growth and proliferation rates. Our understanding of the processes mediating this co-operative trait of nutrient release is limited by uncertainty about the chemical identity of the mobile compounds. The metabolite comparison method outlined in Fig. 1 contributes to the resolution of this limitation by identifying candidate mobile compounds on the basis of their pattern of metabolism by the host. The method assumes that animal metabolism of exogenous compounds taken up into ectodermal cells is equivalent to animal metabolism of photosynthetic compounds translocated across the symbiosomal membrane into endodermal cells. Consistent with this assumption, Trench (1971b) has shown that the pattern of metabolism of photosynthetic compounds is similar in the ectodermal and endodermal cell layers of the sea anemone Anthopleura elegantissima. The present study has identified glucose and succinate/fumarate as candidate mobile compounds in Anemonia viridisusing short duration experiments. This probably does not represent the total complement of mobile photosynthetic compounds because other compounds,possibly including lipid and amino acids, may be translocated many minutes to hours after fixation (e.g. Battey and Patton, 1984; Wang and Douglas, 1999).

To our knowledge, the photosynthetic mobile compounds in the A. viridis symbiosis have not previously been identified. This may be because although a number of studies have demonstrated carbon translocation in the intact symbiosis (Stambler and Dubinsky, 1987; Davy et al.,1997), the algal cells isolated from A. viridis do not release photosynthetically fixed carbon, even in the presence of host homogenate (L. F. Whitehead and A. E. Douglas, unpublished results). The conclusion that glucose and succinate/fumarate are the likely photosynthetic mobile compounds in A. viridis is partly consistent with previous reports, in that these compounds have been cited as mobile compounds in a range of symbioses involving Symbiodinium (e.g.Trench, 1971a;Rees et al., 1993;Wang and Douglas, 1997). However, glycerol, which is accepted widely as a dominant mobile photosynthetic compound in many symbioses involving Symbiodinium, is apparently not translocated in the A. viridis system. If glycerol were mobile in A. viridis, it would be expected that ¹⁴C-labelled glycerol would be detected in the animal fraction after incubation in NaH¹⁴CO₃, because exogenous glycerol was not metabolized rapidly in the animal tissues [also noted by Ishikura et al. (1999) in tridacnid clams]. Contrary to this expectation, glycerol was not recovered as a radioactively labelled metabolite in the animal tissues in any of the many NaH¹⁴CO₃ incubation experiments conducted here. This result is open to at least three alternative explanations. First, as suggested by the study of Ishikura et al.(1999), glycerol may not generally be an important mobile compound in symbiosis. Its dominance among the photosynthetic compounds released from isolated algal cells incubated with host homogenate may reflect a general shift in metabolism associated with the physiological responses either to isolation from the symbiosis or to the host homogenate (Goiran et al.,1997). Second, the symbiosis in A. viridis may be metabolically different from many symbioses, linked to its exclusively high latitude distribution (Manuel,1988) or its genetically distinctive symbiotic algal partner(Savage et al., 2002). Finally, the identity of the mobile photosynthetic compounds may vary with environmental conditions. For example, Gattuso et al.(1993) concluded that glycerol is released from the algal cells to the animal tissues of the coral Stylophora pistillata only under high light conditions, whereas the experiments described here on A. viridis were conducted under relatively low light conditions. Further research is required to discriminate between these possibilities. For the present, we note that although glycerol is incorporated preferentially into lipid in other symbioses(Trench, 1971b;Schmitz and Kremer, 1977), we found no evidence for this in A. viridis and that the generally lower lipid content of temperate symbioses, including A. viridis, compared with tropical species (Harland et al.,1991) is consistent with glycerol not being an important mobile photosynthetic compound in the A. viridis symbiosis.

Other data obtained in this study support the possibility that the fate of photosynthetic carbon in the A. viridis symbiosis may differ from that in low-latitude symbioses. In particular, photosynthetic carbon is preferentially incorporated into acidic compounds in the TCA-soluble fraction of the animal tissue in A. viridis(Fig. 3B), but very little radioactivity is recovered from this fraction in low-latitude symbioses,including Zoanthus flos marinus and Condylactis gigantea(von Holt and von Holt, 1968). A second distinctive feature of the data obtained here is the low incorporation of photosynthetically fixed carbon into the lipid fraction of both the algal cells (16%) and the animal tissue (3%), relative to other symbioses (35–90% and 20–60%, respectively;von Holt and von Holt, 1968;Trench, 1971b;Gattuso et al., 1993). This may reflect the low total lipid content of A. viridis compared with tropical anemones and corals (11% and 50% of dry mass, respectively;Harland et al., 1992), which has been hypothesised to result from the lower light and lower temperature in temperate environments (Harland et al.,1992; Muller-Parker and Davy,2001). However, further research on the impact of season, light and temperature on the incorporation patterns of photosynthetic carbon is required to resolve this issue.

In conclusion, the metabolite comparison method developed here to study the identity of mobile photosynthetic compounds in symbioses involving Symbiodinium has potential value to investigate a range of issues that are currently not readily tractable to study. These include comparisons among symbioses from different habitats (e.g. temperate versustropical) involving different algal genotypes and maintained under various environmental conditions, e.g. irradiance and temperature. With appropriate modifications, the approach may also contribute to the study of nutritional interactions in other symbioses between microorganisms and plants or animals,especially where, as with the Symbiodinium systems, alternative methods to study nutrient translocation in the intact association are not fully developed.

We thank Kim Simpson and Alex Venn for help with the collection and maintenance of sea anemones, Dr David Ashford for advice and assistance with analysis of sugars, and the Leverhulme Foundation for financial support.