The Effects of Cyanobacteria and the Cyanobacterial Toxin Microcystin-Lr on Ca²⁺ Transport and Na⁺/K⁺-ATPase in Tilapia Gills

In the gills of freshwater fish, Ca²⁺ influx is facilitated by a Ca²⁺ channel in the apical membrane of the chloride cells (Verbost et al. 1989); extrusion of Ca²⁺ from the cell across the basolateral membrane is mediated by a high-affinity Ca²⁺-ATPase and possibly by a Na⁺/Ca²⁺ exchanger (Flik et al. 1993; Verbost et al. 1994). The intracellular Ca²⁺ concentration could be regulated by this extrusion process, by Ca²⁺-binding proteins, by active Ca²⁺ uptake into endoplasmic reticulum (ER) (Somlyo, 1985) or by sequestration by mitochondria via a uniporter (Carafoli, 1982; Gunter and Pfieffer, 1990; Gunter et al. 1994). Inhibition of the Ca²⁺-ATPase of the basolateral membrane of chloride cells by Cd²⁺ (Verbost et al. 1987a, 1988, 1989) causes hypocalcaemia (Giles, 1984), indicating the importance of this branchial Ca²⁺ pump in calcium homeostasis.

Eutrophication of water bodies has resulted in an increased frequency of cyanobacterial (blue-green algal) blooms that may produce neuro-and/or hepatotoxins (Codd et al. 1989; Carmichael, 1992). On senescence of a bloom, these toxins are released into the water and they may cause fish kills (e.g. Schwimmer and Schwimmer, 1968). The majority of kills have been linked with hypoxia caused by the high biological oxygen demand on senescence of cyanobacterial blooms and scums. However, oxygen levels were above 90 % saturation when moribund brown trout were retrieved after the lysis of a bloom of the toxic blue-green alga Anabaena flos-aqua at Loch Leven, Scotland (Rodger et al. 1994). Histopathology revealed that these fish had severe liver damage similar to that of fish administered extracts of Microcystis spp. containing the hepatotoxin microcystin-LR (MC-LR) or purified MC-LR injected intraperitoneally (Phillips et al. 1985; Sugaya et al. 1990; Råbergh et al. 1991) or after gavage experiments with microcystin-producing cyanobacteria (Tencalla et al. 1994). However, immersion trials using concentrations of extracts of the hepatotoxic cyanobacterial cells similar to those found in eutrophic environments have produced sublethal effects in fish, e.g. increased plasma cortisol and glucose levels, while plasma Na⁺ and Cl^-concentrations decreased over a 4 h exposure period (Bury et al. 1996). Long-term exposure, for 63 days, was detrimental to fish growth and resulted in disturbances to ionic regulation (Bury et al. 1995).

In vitro, MC-LR specifically inhibits protein phosphatases 1 and 2A (Eriksson et al. 1990; Honkanen et al. 1990; MacKintosh et al. 1990; Matsushima et al. 1990). There is also one report that MC-LR inhibits Na⁺/K⁺-ATPase activity in carp gill microsomes (Gaete et al. 1994), and these observations imply that aspartic acid–phosphatase hydrolase activity of P-type ATPases is also a target for MC-LR. In view of these results, the present study examines the effects of purified MC-LR and of extracts of cells from the cyanobacteria Microcystis aeruginosa 7820 (which produces a range of microcystins including MC-LR) or M. aeruginosa CYA 43 (which produces toxins containing very small quantities of MC-LR) on Ca²⁺ transport in tilapia, Oreochromis mossambicus. This investigation focuses on whole-body Ca²⁺ influxes, on Ca²⁺ transport in a gill basolateral membrane vesicle (BLMV) preparation and in endoplasmic reticulum (ER) and mitochondria and on Na⁺/K⁺-ATPase activity of the gill BLMVs.

Freshwater tilapia (Oreochromis mossambicus Peters) were obtained from laboratory stocks and were held in running Nijmegen tap water (in mmol l^₋1, 0.7 Ca²⁺, 0.5 Na⁺, 0.06 K⁺, pH 7.8, 27 °C) under a light regime of 12 h:12 h light:dark. Fish were fed Trouvit fish pellets (Trouw and Co., Putten, The Netherlands) at a ration of 1.5 % body mass per day.

A method for culturing Microcystis aeruginosa PCC 7820 has been described previously (Bury et al. 1995). Chlorophyll a levels were measured according to MacKinney (1941). Cells were harvested in the early stationary phase of batch culture by continuous centrifugation in a Sharples centrifuge (Sharples Ltd, Surrey), the cell pellet was collected, freeze-dried and microcystin-LR (MC-LR; molecular mass 994 kDa) extracted by high-performance liquid chromatography (HPLC) (Lawton et al. 1994). M. aeruginosa CYA 43 was similarly cultured, but does not yield MC-LR at levels detectable by HPLC (detection limit, 5 ng of microcystin on the HPLC column).

For in vitro Ca²⁺ transport and Na⁺/K⁺-ATPase assays, freeze-dried cyanobacterial cells were twice extracted in methanol (0.314 g per 50 ml methanol), centrifuged at 100 g for 15 min (Sorval RC-5B), the supernatants were combined, dried under vacuum, and the extract was resuspended in 2 ml of methanol and stored at -20 °C. Samples were dried and resuspended in sucrose buffer containing 250 mmol l^-1 sucrose and 10 mmol l^-1 Hepes/Tris, pH 7.4, to give a final concentration of 150 mg dry mass ml^-1, equivalent to 0.5 mg ml^-1 MC-LR for M. aeruginosa PCC 7820 extracts. Additions of the cyanobacterial extracts will be referred to as milligrams of freeze-dried cells per milligram of membrane protein. Purified microcystin-LR was resuspended in sucrose buffer to a concentration of 2 μg μl^-1 MC-LR.

The methanol extracts from M. aeruginosa 7820 and CYA 43 were fractionated by a Pharmacia ‘Smart’ system using the eluents acetonitrile/0.5 % trifluoroacetic acid (TFA) and milliQ water/0.05 % TFA. A 65 % acetonitrile (v/v) gradient was built up over a 25 min period with a flow of 150 μl min^-1, during which fractions of 100 μl were collected and pooled into six sequential groups, fraction 1 being the most hydrophilic and fraction 6 being the most hydrophobic. Fractions were resuspended in sucrose buffer in a volume equivalent to the volume of methanol injected onto the column to allow comparisons of toxin levels in fractions and cell extracts (see also toxin application).

Purified MC-LR (1 mg) was dissolved in 2 ml of ethanol and diluted in 100 ml of distilled water. Freeze-dried M. aeruginosa 7820 (0.314 g yielded 1 mg of MC-LR) was extracted twice with 50 ml of 4 % (v/v) ethanol in water and centrifuged at 1000 g (Sorval RC-5B) and the supernatants combined. A similar quantity of M. aeruginosa CYA 43 was similarly extracted.

Tilapia (13–35.9 g) were starved for 48 h and then six fish were held for 24 h in 4 l vessels containing Nijmegen tap water at 27 °C, to which 90 μg l⁻¹ MC-LR, freeze-dried extract of M. aeruginosa 7820 (86.4 μg l⁻¹ MC-LR, 27 mg dry mass l⁻¹) or M. aeruginosa CYA 43 (27 mg dry mass l⁻¹) was added. MC-LR levels were determined by HPLC. Fish exposed to 0.025 % (v/v) ethanol in water acted as controls.

Exposure was for 4 and 24 h. Whole-body Ca²⁺ influx was measured by the uptake of ⁴⁵Ca²⁺ (0.11 MBq l⁻¹, Dupont) over 4 h. Fish were killed after this time by an overdose of the anaesthetic 2-phenoxyethanol, blood was removed via the caudal vessel and the plasma was collected by centrifugation (1 min at 13 000 g).

Carcasses were digested for 48 h in 50 ml of H₂O₂ at 60 °C, 1 ml of digest was removed and 4 ml of Aqualumar added. ⁴⁵Ca²⁺ was measured in a Pharmacia Wallac 1410 liquid scintillation counter (all subsequent measurements of ⁴⁵Ca²⁺ levels are made with the same counter) and calculated on the basis of carcass total radioactivity (disints min⁻¹) and water specific activity (disints min⁻¹ ml⁻¹). Plasma osmolality was measured with a Vogel osmometer using distilled water and a 300 mosmol kg⁻¹ standard as references. Plasma total Ca²⁺ concentrations were measured using a calcium kit (Sigma Diagnostics), Na⁺ concentrations were measured by atomic absorption (Pye-Unicam SP9 atomic absorption spectrophotometer) and Cl⁻concentrations were determined with a chloride meter (Jenway PCLM3 chloride meter).

Methods for basolateral membrane vesicle preparations are described by Flik et al. (1985). Briefly, gills were excised from tilapia (250 g), washed in buffer containing 250 mmol l⁻¹ sucrose, 5 mmol l⁻¹ NaCl and 5 mmol l⁻¹ Hepes/Tris, pH 7.4, and kept on ice. All subsequent procedures were performed at 1–4 °C. The epithelium was scraped off with a glass microscope slide and homogenised in the above buffer for 2 min using a Polytron ultra-turrax homogeniser set at 20 % of its maximum speed. This procedure kept red blood cells intact while branchial epithelium was disrupted. In this respect, it is similar to a previously described technique using a Douncer homogenisation device instead of a Polytron ultra-turrax homogeniser (Flik and Verbost, 1994).

The red blood cells and cellular debris were removed from the homogenate by centrifugation at 550 g for 10 min. Membranes were collected by centrifugation at 30 000 g (Sorval RC-5B) and were resuspended in buffer using a Douncer-type homogenisation device (100 strokes). The resulting suspension was differentially centrifuged; 1000 g for 10 min, 10 000 g for 10 min and 30 000 g for 30 min, and the final pellet was resuspended in buffer containing 150 mmol l⁻¹ NaCl, 0.8 mmol l⁻¹ MgCl₂ and 20 mmol l⁻¹ Hepes/Tris, pH 7.4, for the Na⁺/K⁺-ATPase assay or 150 mmol l⁻¹ KCl, 0.8 mmol l⁻¹ MgCl₂ and 20 mmol l⁻¹ Hepes/Tris, pH 7.4, for Ca²⁺ transport measurements. The vesicles were resealed by 10 passages through a 23 gauge needle, giving 19.2–30 % inside-out (IOV) and 29–44 % rightside-out vesicles (ROV) (Flik et al. 1985; Verbost et al. 1994). The amount of membrane protein present was determined using a commercial kit (Bio-Rad) with bovine serum albumin (BSA) as a standard and was adjusted to 1.5 mg ml⁻¹. To obtain maximum Na⁺/K⁺-ATPase activity, vesicles were permeabilised with saponin (0.2 mg mg⁻¹ protein) to ensure optimal substrate accessibility.

The preparation of permeabilised gill cells was based on the methods of Verbost et al. (1994). Gills were excised and the epithelial scrapings incubated in lysis medium containing 9 parts of 0.17 mmol l⁻¹ NH₄Cl and 1 part of 0.17 mmol l⁻¹ Tris/HCl, pH 7.4, for 20 min. The cells and red cell material were resuspended at the beginning and end of lysis by passage through a 10 ml pipette, large cell clusters were removed by passage through a 100 μm nylon mesh. Cells were collected, and the lysed red cells removed by centrifugation at 150 g for 5 min. The resulting pellet was resuspended in buffer containing 150 mmol l⁻¹ NaCl, 0.8 mmol l⁻¹ MgCl₂ and 20 mmol l⁻¹ Hepes/Tris, pH 7.4, and incubated with 0.3 mg ml⁻¹ saponin at 37 °C for 5 min. The cells were centrifuged for 5 min at 150 g and the pellet washed twice and finally resuspended in assay buffer containing 120 mmol l⁻¹ KCl, 1.2 mmol l⁻¹ KH₂PO₄, 5 mmol l⁻¹ succinate, 5 mmol l⁻¹ pyruvate, 0.5 mmol l⁻¹ EGTA, 0.5 mmol l⁻¹ nitrilotriacetic acid (NTA) and 24 mmol l⁻¹ Hepes/KOH, pH 7.1. The protein content was adjusted to 2 mg ml⁻¹.

The BLMVs and permeabilised cell preparations were exposed to purified MC-LR at concentrations in the range 1.56–333 μg mg⁻¹ membrane protein (equivalent to 1.57–335 nmol mg⁻¹ membrane protein) and for experiments with methanol extracts of cyanobacteria in the range 0.12–25 mg of freeze-dried material per milligram membrane protein. For experiments with the six fractions obtained from the methanol extracts, a volume of 83.3 μl of cyanobacterial extract fraction per milligram membrane protein was applied, which was equivalent to 12.5 mg of freeze-dried material per milligram membrane protein. For permeabilised cell preparations, thapsigargin (10 μmol l⁻¹) was used to inhibit endoplasmic reticulum Ca²⁺ transport and Ruthenium Red (1 μmol l⁻¹) to inhibit mitochondrial Ca²⁺ transport. All preparations were incubated for 2 h on ice prior to assays.

K⁺-dependent p-nitrophenol phosphatase (pNPPase) activity, which reflects the dephosphorylation step (i.e. the phosphatase activity) of the Na⁺/K⁺-ATPase, was determined. Medium A contained 100 mmol l⁻¹ KCl, 75 mmol l⁻¹ MgCl₂, 300 mmol l⁻¹ imidazole, 10 mmol l⁻¹trans-1,2,-diaminocyclohexane-N,N,N’,N’-tetraacetic acid (CDTA) and 5 mmol l⁻¹p-nitrophenolphosphate (pNPP), pH 7.4, whilst medium E consisted of medium A to which 1 mmol l⁻¹ ouabain was added and from which KCl was omitted. Toxin-treated vesicles (10 μl) were mixed with 500 μl of either medium A or medium E and incubated for 15 min at 37 °C. The K⁺-dependent ouabain-sensitive pNPPase activity was defined as the difference in activity measured between medium A and medium E at 420 nm.

Toxin-treated vesicles (12.5 μl) were added to 50 μl of assay medium; 150 mmol l⁻¹ KCl, 1 μmol l⁻¹ ‘free’ Ca²⁺, 0.8 mmol l⁻¹ ‘free’ Mg²⁺, buffered with 0.5 mmol l⁻¹ EGTA, 0.5 mmol l⁻¹N-(2-hydroxyethyl)-ethylenediamine-N,N’,N’-triacetic acid (HEEDTA), 0.5 mmol l⁻¹ NTA, with or without 3 mmol l⁻¹ ATP and containing ⁴⁵Ca²⁺. Both solutions were pre-warmed to 37 °C and subsequent incubations performed at this temperature for 1 min, to determine initial velocities. The transport was stopped by the addition of 1 ml of ice-cold stop buffer containing 150 mmol l⁻¹ KCl, 1 mmol l⁻¹ EGTA and 10 mmol l⁻¹ Hepes/Tris, pH 7.4. Vesicles were collected by rapid filtration using Schleicher and Schuell ME25 mixed cellulose filters, pore size 0.45 μm, and were rinsed twice with ice-cold stop buffer. Filters were placed in scintillation vials and dissolved in 4 ml of Aqualuma for 30 min before activity was counted.

Permeabilised gill cell preparations were mixed thoroughly and 10 μl of the suspension was added to 50 μl of assay medium, both having been pre-warmed to 28 °C. Assay medium contained the same concentration of ligands and KCl as above, but with 5 mmol l⁻¹ ATP. ‘Free’ [Mg²⁺] was set at 0.8 mmol l⁻¹; for endoplasmic reticulum Ca²⁺ transport measurements, ‘free’ [Ca²⁺] was set at 0.1 μmol l⁻¹, and for mitochondrial Ca²⁺ transport measurements, ‘free’ [Ca²⁺] was 1 μmol l⁻¹ (Verbost et al. 1987b). Cells were incubated at 28 °C for 2 min and transport was stopped by the addition of 1 ml of stop buffer, as above. Cells were collected on Schleicher and Schuell (GF92 diameter 25 mm) glass filters, and rinsed and counted as above.

All results were compared with control values using a Student’s t-test on the Minitab computer package. ‘Free’ Ca²⁺ and Mg²⁺ concentrations were calculated using a matrix computer program developed by Schoenmakers et al. (1992).

Whole-body Ca²⁺ influx rates were inhibited by 40 % after 24 h of exposure to extracts of M. aeruginosa 7820 and by 31 % with M. aeruginosa CYA 43 extracts (Fig. 1). Table 1 shows that plasma Ca²⁺ concentrations were significantly lower in fish exposed for 24 h to M. aeruginosa CYA 43, while plasma Na⁺ and Cl⁻levels showed a small but significant increase in fish exposed to M. aeruginosa 7820. Purified microcystin-LR (MC-LR) had no effect on plasma electrolyte levels or whole-body Ca²⁺ influx when compared with controls at 24 h (Table 1; Fig. 1). None of the treatments had a significant effect on whole-body Ca²⁺ influx or body ion concentrations after 4 h of exposure.

BLMV K⁺-dependent pNPPase activity was not affected by purified MC-LR, but was inhibited by increasing doses of methanol extracts of both M. aeruginosa 7820 and CYA 43 (Fig. 2). Fraction 5 of the cyanobacterial extracts appeared to produce the greatest reduction in K⁺-dependent pNPPase activity (Table 2), but this was not significant when compared with the effects of other fractions.

Purified MC-LR had no effect on Ca²⁺ uptake in BLMVs, although methanol extracts of both M. aeruginosa 7820 and CYA 43 progressively inhibited Ca²⁺ uptake with increasing doses (Fig. 3). M. aeruginosa 7820 and CYA 43 extract fractions 4, 5 and 6 significantly inhibited Ca²⁺ uptake in BLMVs compared with controls (Table 3).

Ca²⁺ uptake specific to the endoplasmic reticulum (ER) of the permeabilised cells was verified using the inhibitor thapsigargin at 0.1 μmol l⁻¹ Ca²⁺, which produced an 82 % inhibition. Ca²⁺ uptake specific to mitochondria of the permeabilised cells was verified using Ruthenium Red at 1 μmol l⁻¹ Ca²⁺ as an inhibitor, which resulted in 80 % inhibition. Purified MC-LR had no effect on ER or mitochondrial uptake of Ca²⁺ (Figs 4, 5). The methanol extracts from both strains of M. aeruginosa inhibited Ca²⁺ uptake by ER (Fig. 4) and mitochondria (Fig. 5), most noticeably at higher concentrations. M. aeruginosa 7820 and CYA 43 extract fractions 4, 5 and 6 all inhibited Ca²⁺ transport in both systems compared with controls, with fraction 5 being the most potent (Table 4).

The toxins from cyanobacteria have been reported to impede Na⁺/K⁺-ATPase activity of carp gill microsomes (Gaete et al. 1994), and so we focused our studies on the ATPases that mediate Ca²⁺ transport in the gill of tilapia. Our results show that, in vitro, MC-LR did not inhibit Na⁺/K⁺-ATPase activity, measured as K⁺-dependent pNPPase (Fig. 2), or the ATP-driven Ca²⁺ transport of the gill basolateral plasma membrane (Fig. 3); in addition, there was no inhibition of SERCA-type ATPases of the endoplasmic reticulum (ER, Fig. 4) or of the Ca²⁺ uniporter of the mitochondria (Fig. 5). In contrast, extracts, particularly in the hydrophobic fractions, from the 7820 and CYA 43 strains of M. aeruginosa inhibited all the systems investigated. Na⁺/K⁺-ATPase activity appeared to be less sensitive than the Ca²⁺-dependent transporters to the toxic substances within the cyanobacteria.

The cyanobacterial toxin microcystin-LR (MC-LR) inhibits protein phosphatases 1 and 2A (PP1 and PP2A) (Eriksson et al. 1990; Honkanen et al. 1990; MacKintosh et al. 1990; Matsushima et al. 1990). If MC-LR also inhibits Na⁺/K⁺-ATPase activity of carp gill microsomes (Gaete et al. 1994), this would imply either that the inhibitory action of MC-LR is wider, i.e. it would also inhibit P-type ATPase activity (where the phosphatases show an aspartic acid–phosphate hydrolase activity) and/or that the fish ATPase activity is aberrant in having a PP1/PP2A-type phosphatase activity as part of its Na⁺/K⁺-ATPase cycle. However, in the present study, MC-LR did not inhibit any Na⁺/K⁺-ATPase activity in tilapia gill (Fig. 3), nor has MC-LR inhibition been found in carp gill, dog kidney or human erythrocytes (N. R. Bury and G. Flik, personal observations). Moreover, other Ca²⁺-ATPases in the plasma membranes and ER proved to be unaffected by the toxin, in line with the specificity of the toxin to PP1 and PP2A. We have no doubt of the potency of our MC-LR preparations, as acute hepatotoxicity had been ascertained by intraperitoneal injection in mice (Falconer et al. 1981). Moreover, the preparation appeared as a single homogeneous peak on HPLC with the same retention time as that previously published for MC-LR (Lawton et al. 1994).

Tilapia exposed for 24 h to extracts from either strain of M. aeruginosa, 7820 or CYA 43, showed inhibited Ca²⁺ influx (Fig. 1), which resulted in hypocalcaemia in fish exposed to M. aeruginosa CYA 43 (Table 1). The small but significant increase in plasma Na⁺ and Cl⁻concentrations in fish exposed to M. aeruginosa 7820 (Table 1) was unexpected, and there is no immediate explanation. A possible explanation for the lack of in vivo effects at 4 h is that the toxin has yet to exert an inhibitory effect at the basolateral membrane on Ca²⁺ transporting mechanisms (see Flik et al. 1985, 1993).

Toxicity tests with invertebrates and bacteria have verified the toxicity of the microcystins (Penaloza et al. 1990; Demott et al. 1991; Kiviranta et al. 1991), but have also illustrated the presence of additional cytotoxic substances, as yet unidentified, within cyanobacteria (Nizan et al. 1986; Jungmann et al. 1991; Jungmann, 1992; Jungmann and Benndorf, 1994; Campbell et al. 1994). Penaloza et al. (1990) found that cyanobacterial fractions toxic to zooplankton contained factors with a molecular mass similar to that of MC-LR, but toxicity was lost upon boiling, while microcystins are heat-stable to 160 °C (e.g. Jungmann and Benndorf, 1994). Cyanobacteria have been screened for compounds with therapeutic potential and are a rich source of novel bioactive substances; for example, antineoplastic compounds (Patterson et al. 1991), and lipophilic cyclic peptides (laxaphycins) that are antifungal and cytotoxic (Frankmolle et al. 1992). There is no account of these substances inhibiting ATPases, but fungi have been shown to possess citreoviridin, a polyene neurotoxin which inhibits mitochondrial ATPase (Sayood et al. 1989), and tentotoxin, a cyclic tetrapeptide which inhibits ATPase activity in cyanobacteria (Ohta et al. 1993).

In conclusion, we have shown that strains of the cyanobacterium Microcystis aeruginosa produce compound(s) that inhibit Ca²⁺ uptake and gill K⁺-dependent pNPPase activity in tilapia. We also present evidence that these inhibitory effects are not due to the protein phosphatase inhibitor MC-LR. However, further work is required to determine the exact structure and inhibitory mode of action of the compound(s) which affect gill Ca²⁺ transport and may inhibit physiological processes dependent upon Ca²⁺, thus contributing to the fish death that often accompanies senescence of a cyanobacterial bloom.

N.B. is funded by the NERC (GT/92/140/A). We also thank the European Science Foundation Ecotoxicology Short-Term Visiting Fellowship programme (SVF/94/37) for funding this study, F. A. T. Spanings for the husbandry of the tilapia and K. A. Beattie for help with the cyanobacterial cell culture.