Pseudodiarrhoea in zebra mussels Dreissena polymorpha (Pallas)exposed to microcystins

Microcystins (MCs) are endotoxins produced by cyanobacteria, present inside the algal cells that are released into the environment after cell lysis(Sivonen and Jones, 1999). Cyanobacteria are considered as a nuisance because they form harmful algal blooms (HABs) in freshwater bodies subject to eutrophication(Codd, 1992; Nicholls et al., 2002; Mur et al., 1999). Microcystins are hepatotoxic, tumour promoting(Falconer, 1991) and induce liver cell apoptosis (Chen et al.,2005). They therefore cause serious, often lethal, health effects on fish, domestic animals and humans(World Health Organisation,2003; Azevedo et al.,2002; Carmichael,2001; Chorus and Bartram,1999).

The freshwater zebra mussel Dreissena polymorpha (Pallas) is invasive and established in water bodies throughout Europe and North America and has recently colonised Ireland. As they reproduce rapidly and foul a wide range of structures, they are of serious concern to industry and environmental managers. Efficient suspension feeders, they occur in extremely high densities and cause considerable changes in ecosystem composition/function such as local extirpation of native mussel populations(Strayer et al., 1999),increases in water clarity (Budd et al.,2001) and the removal of microalgae from water columns(Raikow et al., 2004), hence influencing HABs (Reeders et al.,1989; Roditi et al.,1996).

One of the most common species responsible for HABs is Microcystis aeruginosa (Kützing). The relationship between this cyanobacterium and the zebra mussel has already been studied but with conflicting results. Laboratory experiments have shown a decline in Microcystis aeruginosaabundance because mussels preferentially ingested cyanobacteria over diatoms(Baker et al., 1998), while others have reported enhanced intake of green algae in the presence of cyanobacteria (Dionisio Pires et al.,2004). Field studies have shown that, in the North American Great Lakes, mussels were capable of selective rejection of toxic Microcystis aeruginosa in pseudofaeces, promoting cyanobacterial HABs(Vanderploeg et al., 1996; Vanderploeg et al., 2001). Dionisio Pires and Van Donk (Dionisio Pires and Van Donk, 2002) found that lower clearance rates of phytoplankton by Dreissena were obtained with a mixture of toxic cyanobacteria and Chlamydomonas reinhardtii (Dangeard) than with a mixture containing non-toxic cyanobacteria. Their results also showed that mussels might finally promote a dominance of Chlamydomonas over Microcystis, especially when the cyanobacterial strain used was toxic.

Detailed knowledge of the capture and sorting of particles by the gills and labial palps of the zebra mussel (Baker et al., 1998; Baker et al.,2000) is available, but very little information has been collected concerning the effects of cyanobacterial toxins on food selection processes. Furthermore microcystins belong to a family of endotoxins that comprises different variants. Microcystin-LR (MC-LR) occurs most frequently(Dawson, 1998) and is considered the most toxic (Imanishi et al., 2005) but few data are available concerning the toxicity of the other variants. Our experiments were therefore designed to provide greater insight regarding the toxicity of microcystins by comparing the effects of Microcystis aeruginosa strains, of different toxicity levels and different toxic profiles, on the ecophysiology of zebra mussels.

All algal species were cultured under controlled conditions(20±1°C; 12 h:12 h light:dark cycle). Three strains were cultured and used in the exponential stable phase of their growth. The strains cultured were the diatom Asterionella formosa (Hassal) (SAG 8.95), obtained from Sammlung von Algenkulturen Göttingen (SAG), and two strains of the toxic cyanobacterium Microcystis aeruginosa (CCAP 1450/10 and CCAP 1450/06), obtained from the Culture Collection of Algae and Protozoa (CCAP). Asterionella formosa was cultured in Diatom + Vitamin Mix medium and the other species were cultured in Jaworsky medium (JM).

To obtain toxic profiles, the general procedure of Ortea et al.(Ortea et al., 2004) was followed. Briefly, algal samples (50 ml) were freeze-thawed and sonicated to burst the cells so that all of the toxin would be released. The samples were then filtered through GFB filter papers by vacuum filtration. The filtrates were then applied to optimised solid-phase extraction (SPE) columns (Bakerbond C₁₈ Polarplus cartridge; Phillipsburg, NJ, USA) using the method based on that developed by Lawton et al.(Lawton et al., 1994). The SPE cartridge was conditioned with methanol and water. The filtered algal sample was applied and the cartridge was washed with 10 ml portions of 25%methanol/water. Toxins were then eluted using methanol (6 ml) containing 0.1%trifluoroacetic acid (TFA) and the eluent was evaporated to dryness using a Turbo Vap LV Evaporator, and reconstituted in water (1 ml) for analysis.

Liquid chromatography–tandem mass spectrometry (LC-MS/MS) analyses of microcystins were carried out using a HP100 series LC system with an ultraviolet-photodiode array (UV-PDA) detector (Agilent, Ipswich, UK) linked with an LCQ ion-trap mass spectrometer (ThermoFinnegan, San Jose, USA). The mass spectrometer was equipped with an electrospray ionisation (ESI)interface. Using a flow injection rate of 3 μl min^–1, the mass spectrometer was tuned using MC-LR; the optimised temperature was 220°C with a voltage of 3.0 V. The optimum relative collision energy (%RCE) was determined for each of the four toxic variants studied: MC-LR, MC-RR,MC-YR and MC-LF. Separation of the microcystins was achieved using a Luna (2)(5 μm, 2.0×150 mm, Phenomenex, Macclesfield, UK) column at 40°C(5 μl injection). A gradient elution of acetonitrile (with 0.05% TFA) over 42 min was used at a flow rate of 0.2 ml min^–1.

Thirty mussels were collected from the submerged area of a quay in Ballina Marina, Killaloe, Tipperary, Ireland, in March 2004. Only adult mussels between 20 and 24 mm shell length were used to minimize size-related variation. They were brought back to the laboratory in lake water and acclimated to standard filtered (0.45 μm) freshwater maintained at 20±1°C for 48 h prior to the feeding experiments. Only active individuals showing valve and siphon movements were used in the experiments.

Each cultured strain was used in single cell suspensions to feed the mussels, corresponding to three separate diets. They were diluted in standard filtered (0.45 μm) freshwater to adjust each diet to comparable mass of suspended matter. Standard freshwater was prepared according to Sprung(Sprung, 1987). The fourth diet consisted of a mixture of the diatom Asterionella formosa and the toxic cyanobacterium Microcystis aeruginosa CCAP 1450/10. This was made up so that each of the two strains represented 50% of the total biomass of cells present in the mixture. The total cell mass per unit of volume was similar to that employed in the other three diets. Each diet was supplied to the mussels at bloom concentration (10⁶ cell ml^–1 for the two cyanobacterial strains and the mixed diet and 10⁴ cell ml^–1 for the non-toxic diatom). All algal cell concentrations were measured by Coulter Counter after determining their equivalent spherical diameter (ESD).

An experimental Perspex flow-through system, consisting of a chamber with baffles, was designed to ensure constant unidirectional flow at a steady microalgal concentration, as described(Barillé et al., 2003)(Fig. 1). The system was supplied with the algal diets from an agitated tank through a Masterflex L/S Economy Drive peristaltic pump (Cole Parmer Instrument Company, Vernon Hills,IL, USA) at a flow rate of around 100 ml h^–1.

The system was particularly designed to separate physically pseudofaeces from faeces produced by the mussels (Fig. 1). The mussels were placed close to the outflow and the baffle acted as a separation; the inhalant siphon opened in the middle part of the chamber (between the two baffles) and the exhalant siphon in the right hand part of the chamber (between the right baffle and the outflow). Pseudofaeces were consequently expelled in the middle part of the chamber and the faeces in the right hand section of the system, facilitating separate collection using a Pasteur pipette (Fig. 1).

Feeding experiments consisted of 90 min trials, performed for each diet with six individual replicates for the single cell suspensions diets and ten for the mixture trial. Individuals were filmed at 25 fields s^–1 from the side of the chamber, or from below as required. Video analysis was performed for two diets: Asterionella formosa and Microcystis aeruginosa CCAP 1450/10. For these diets, the mussels'movements were deconstructed into different actions and the frequency of each action evaluated (see Results for details). For the other diets, videos were screened to evaluate the general behaviour of the mussels.

At the end of 90 min trials with the mixture of non-toxic diatoms and highly toxic cyanobacteria, pseudofaecal ejecta were collected using a Pasteur pipette. Subsamples (each 1 ml) were analysed with a Coulter Counter to determine concentrations of the two cell types, following resuspension of ejecta in 10 ml of standard freshwater. The rest of the sample was volumetrically measured, then filtered and dried to determine dry weight. From the algal cultures, the individual dry weights of each of the two cell types were determined. These were then used to calculate the weight of each of the algal types present in the pseudofaecal materials from the cell concentrations. Assuming that the pseudofaecal ejecta are composed only of cells and mucus, we calculated the percentage of mucus by mass, by subtraction.

Each of the diets was supplied to the mussels at bloom concentration. For the two cyanobacterial strains, cell concentrations were adjusted to 10⁶ cell ml^–1 and 10⁴ cell ml^–1 for the non-toxic diatom by dilution from the culture. Similar concentrations were obtained with the mixture diet. Size spectra obtained with the Coulter Counter and expressed as equivalent spherical diameter (ESD) showed that the diatom Asterionella formosa was the largest of the three algae tested (ESD: 6.5–12.0 μm). The two cyanobacterial strains were smaller (CCAP 1450/10: ESD: 3.0–4.5 μm;CCAP 1450/06: ESD: 3.0–6.5 μm). These values are comparable to the size of the cells examined with a light microscope(Fig. 2). Microscopic observations showed that the cyanobacterial strains used consisted of single cell suspensions (rather than colonies) and the diatom Asterionella formosa appeared to form chains of 5–10 attached cells(Fig. 2).

Four microcystin variants were investigated in each of the two strains of Microcystis aeruginosa cultured. The strain CCAP 1450/06 retained only one variant: MC-LR in a low concentration (7.4 μg l^–1) whereas the strain CCAP 1450/10 was very toxic with 23.8μg l^–1 of MC-LR and 82.9 μg l^–1 of MC-LF (Table 1).

Analysis of videos showed that all six individuals tested per given diet showed the same feeding behavioural patterns in response to that diet. When fed the non-toxic diatom Asterionella formosa, all mussels behaved`normally' with expulsion of pseudofaeces through their inhalant siphons(Fig. 3A; see movie 1 in supplementary material) and ejection of faeces through their exhalant siphons(Fig. 3B; see movie 2 in supplementary material); both types of ejecta being produced in small quantities. We also observed that during `normal' behaviour, mussels periodically closed their shell valves and siphons (not illustrated).

When fed the very toxic cyanobacteria Microcystis aeruginosa CCAP 1450/10, a markedly different pattern was observed in all test mussels. Normal expulsion of pseudofaeces through the inhalant siphon and faeces through the exhalant siphon was observed as well as periodical valve closures, but quantities of pseudofaecal material were greatly enhanced, and appeared as jets of liquid green material (Fig. 3C; see movie 3 in supplementary material). In addition, we noticed expulsion of other pseudofaecal material as green clouds through the pedal gape (effectively an `extra' siphon in this species) of the mussels(Fig. 3C–F; see movies 4,5 and 6 in supplementary material); this we term `pseudodiarrhoea'. Pseudodiarrhoeal pulses through the inhalant siphon and pedal gape were accompanied by sharp adductions of the shell valves; this suggests that the mantle cavity was becoming full of the green material, which was being expelled hydraulically, rather than being ejected by the ciliary tracts used during normal pseudofaecal expulsion.

The same phenomenon was observed when we gave a mixture of the non-toxic Asterionella formosa and highly toxic Microcystis aeruginosaCCAP 1450/10 to mussels (not illustrated). However, when mussels were fed with low-toxicity cyanobacteria Microcystis aeruginosa strain CCAP 1450/06, only `normal' feeding behaviour (not illustrated) was observed, with small quantities of pseudofaeces being expelled through the inhalant siphon,faeces expelled intermittently through the exhalant siphon and mussels closing their shell valves and siphons periodically.

The frequency of each of the mussels' actions previously described was recorded over the 90-min feeding trial when the mussels were fed the non-toxic diatom and the very toxic cyanobacterial strain Microcystis aeruginosa CCAP 1450/10. The frequency values given hereafter are average values ± s.e.m.

Concerning pseudofaeces production, no statistical difference could be observed (Student's t-test; P=0.510) between the two diets even though the frequency of pseudofaecal expulsions was 10.4±1.5 h^–1 with mussels fed the diatom against 8.9±1.7 h^–1 when the mussels were fed the toxic cyanobacteria(Fig. 4). The same pattern was observed for faeces production (Mann–Whitney Rank Sum test, P=0.818) between the two diets even if the frequency of faeces production was 7.3±4.0 h^–1 for mussels fed Asterionella against 2.3±0.8 h^–1 when offered Microcystis aeruginosa CCAP 1450/10. However, significantly more valve closure events were observed in mussels fed the toxic cyanobacteria compared with those fed the non-toxic diatom (Student's t-test; P=0.008; 22.4±3.3 h^–1 and 10.3±1.9 h^–1, respectively; Fig. 4).

Pseudofaecal and pseudodiarrhoeal samples were analysed for algal type,algal concentration and mucus content when mussels were offered the mixture diet (Fig. 5). As shown earlier, Microcystis aeruginosa CCAP 1450/10 and Asterionella formosa cells did not overlap in size (no overlap in their ESD). It was therefore possible to determine their concentration within the same sample with the Coulter Counter. From the culture we determine that the specific mass of Microcystis aeruginosa CCAP 1450/10 cells was 1.4×10^–8 mg cell^–1. Asterionella formosa cells had a specific mass of 2.57×10^–7 mg cell^–1.

Back calculations from the cell densities showed that the two materials differed slightly in composition. Pseudofaeces expelled by the inhalant siphon were composed of 46.1±3.2% M. aeruginosa, 24.5±2.4%Asterionella formosa and 29.4±5.2% mucus by dry mass (mean± s.e.m.). Pseudodiarrhoea expelled through the pedal gape was 40.7±6.6% Microcystis aeruginosa, 14.0±1.04%Asterionella formosa and 45.2±7.00% mucus by mass (mean± s.e.m.). Statistical analysis showed no significant differences in the proportion of Microcystis aeruginosa between pseudofaeces and pseudodiarrhoea (one-way ANOVA on arcsine transformed data; P=0.558; N=10; data normally distributed). However, pseudodiarrhoeal material had significantly less Asterionella formosa than pseudofaeces (ANOVA on ranks on arcsine transformed data; P=0.002; N=10; data not normally distributed). Although pseudodiarrhoea appeared to contain more mucus than pseudofaeces, the difference was not statistically significant(one-way ANOVA on arcsine transformed data; P=0.078; N=10;data normally distributed).

Zebra mussels are suspension feeders that are able to filter a wide range of phytoplanktonic particles (Horgan and Mills, 1997). Particles are generally filtered and captured through the mussels' inhalant siphons and sorted on the gills and labial palps(Baker et al., 2000). Unwanted particles are embedded in strings of mucus and transported to the inhalant siphon for rejection as pseudofaeces, whereas preferred particles are carried to the mouth for ingestion and digestion. They are ultimately ejected as faeces through the exhalant siphon of the bivalve(Baker et al., 1998; Baker et al., 2000). The mechanisms underlying food selection are not well known but variations in particle clearance rates have been attributed to differences in food quality[related to size, polyunsaturated fatty acid (PUFAs) content or toxic content](Dioniso Pires and Van Donk, 2002; Dionisio Pires et al.,2004).

Our study demonstrates that toxic strains of Microcystis aeruginosa have a dramatic effect on processing of microalgae by zebra mussels, whereas nontoxic strains do not. Our data document the first sublethal adverse effect of microcystins on the ecophysiology of an invertebrate species; an acute irritant response to highly toxic Microcystis aeruginosa. This response could be considered as analogous to the acute diarrhoeal symptoms of microcystin toxicity in humans(World Health Organisation,2003; Dillenberg and Dehnel,1960) in that it causes substantial shedding of Microcystis aeruginosa in highly mucous pseudofaeces and pseudodiarrhoea. Furthermore, we noted that, in the presence of the very toxic cyanobacteria,mussels closed their shell valves and siphons significantly more often than they did when fed a non-toxic diatom species. Though mussels normally punctuate their active filtering sessions with periods of quiescence during which the valves close and the siphons withdraw(Morton, 1969; Horgan and Mills, 1997), these marked differences in valve closure rhythm are another sign of the mussels'irritation due to the toxicity of microcystins.

Analysis of the toxic profiles of the two cyanobacteria showed that the most abundant microcystins variant was MC-LF in the very toxic cyanobacterial strain whereas only a small amount of MC-LR was present in that same strain and in the low-toxicity strain. Previous studies investigating zebra mussels feeding on toxic and non-toxic Microcystis aeruginosa(Vanderploeg et al., 2001; Dionisio Pires and Van Donk,2002) were performed with strains containing only the most abundant variant in nature [MC-LR (Dawson,1998)]. MC-LR is also considered to be the most toxic variant among microcystins (Imanishi et al.,2005). However, no information is available on the relative toxicity of MC-LF, but we suggest that, the reaction of the mussels,manifested as the production of copious `pseudodiarrhoea', indicates that MC-LF is a highly toxic microcystins variant that merits further studies.

Analysis of the two types of pseudofaecal ejecta corresponding to the mixture diet showed that they were mostly composed of toxic cyanobacteria,particularly the `pseudodiarrhoea'. This confirms that Microcystis aeruginosa was preferentially being rejected in comparison with Asterionella formosa, since proportions of the two species in the diet were equal, a process being greatly enhanced by production of`pseudodiarrhoea'. Earlier work in the Great Lakes of Canada(Vanderploeg et al., 1996; Vanderploeg et al., 2001)demonstrated that zebra mussels could distinguish between toxic Microcystis aeruginosa and desirable food algae(Cryptomonas), removing the former in pseudofaeces released into the environment, and hence encouraging blooms of the toxic species. This response does not occur in the presence of low toxicity Microcystis aeruginosa, which are ingested. Furthermore, we noted that the pseudodiarrhoeal material was easily resuspended in water after collection with the Pasteur pipette, indicating that it could potentially enhance algal blooms depending on the water mixing regime. Microscopic observations of the`pseudodiarrhoea' stained with Trypan Blue also showed mucus aggregations(blue staining) and Microcystis cells that appeared green, therefore still viable (not illustrated). Selective rejection, accentuated by the pseudodiarrhoeal response, will therefore tend to enhance the presence of toxic Microcystis aeruginosa in mixed Microcystis aeruginosacyanobacterial blooms, as well as transferring toxins from the water column to the benthos. These results also imply that zebra mussels can distinguish between cyanobacterial strains even though the organisms are of indistinguishable size and morphology but of different toxic content.

The long-term effects of microcystins have been studied on zebra mussel larvae (Dionisio Pires et al.,2003) and other zooplanktonic species(DeMott, 1999; Lampert, 1981). These studies all showed that microcystins had an inhibitory effect, mostly on growth,feeding and generally survival of the animals. However, the impact of pseudodiarrhoea on zebra mussels exposed to high levels of microcystins for long periods is not known, but at the very least the response is likely to be energetically costly, since molluscan mucus is known to be expensive to produce (Davies and Hawkins,1998). Further investigations are therefore needed to evaluate the impact of microcystins on the physiological energetics of the zebra mussel.

As mentioned previously, microcystins are considered to have generally negative effects on aquatic animals but no detrimental effects have so far been observed in adult zebra mussels. The expulsion of toxic Microcystis aeruginosa in mucous pseudodiarrhoea (rather than their ingestion), may therefore perhaps be a positive resistance factor that contributes to the successful spread of zebra mussels throughout water bodies subject to eutrophication.

We acknowledge the Higher Education Funding Council of Ireland for funding supplied under the Programme for Research in Third Level Institutions, Cycle 2(National Development Plan), to the ZEBRATOX project. We also thank the two anonymous referees for their valuable comments on the manuscript and Ballina Marina for access to the sample site.