ELSEVIER Environmental Toxicology and Pharmacology 2 (1996) 223-232

Enhancement of acetylcholine release by homoanatoxin-a from Oscillatoria formosa

P%l Aas a. * , Stig Eriksen ‘, Jgrgen Kolderup b, Paul Lundy ‘, John-E. Haugen d, Olav M. Skulberg ‘, Frode Fonnum a

Received 12 January 1996; revised 13 June 1996; accepted 28 June 1996

Abstract

The strain NIVA-CYA 92 of Oscillutoria fo~~osu Bory ex Gormont produces phycotoxins with neurotoxic properties. Chemical analysis by gas chromatography/mass spectrometry of a water extract of lyophilized material of the organism showed the presence of only homoanatoxin-a. The mechanism of action of homoanatoxin-a on peripheral cholinergic nerves is so far not known. The neurotoxicity of 0. formosu containing homoanatoxin-a was investigated in rat bronchi, rat brain synaptosomes and in GH,C, cells. The water extract of lyophilized material of the organism produced a concentration-dependent reversible increase in the release of [‘Hlacetylcholine from both Kt (51 mM) depolarised and non-depolarised cholinergic nerves of the rat bronchial smooth muscle. The Kf-evoked release of [‘H]acetylcholine was enhanced by about 75% by a water extract from 15-20 mg/ml of lyophilized algal material. The enhanced release of [ ‘H]acetylcholine was substantially reduced by the L-type Ca’+-channel blocker verapamil (I 00 /IM) and not by the N-type Ca*+-channel blocker w-conotoxin GVIA (I .O FM) or the P-type Ca’+-channel blocker w-agatoxin IV-A (0.2 FM). Chelation of intra-cellular Ca’ A by I ,2-bis-(aminofenoxi)etan-N, N, N’, N’-tetraacidic acid/acetoxymethyl (BAPTA/AM) (30 PM) had no effect on the phycotoxin-induced release of [‘H]acetylcholine. indicating that an extracellular pool of Ca2+ was important for the action of the phycotoxin on the release of [‘H]acetylcholine from peripheral cholinergic nerves. In rat brain synaptosomes the algal extract enhanced the influx of “Ca’+ in a tetrodotoxin (I .O KM) and o-conotoxin MVIIC (blocker of N-, P- and Q-type Cal+ channels) (I.0 PM) insensitive manner. Patch-clamp studies showed that the phycotoxin opened endogenous voltage dependent L-type Ca2+ channels in neuronal GH,C, cells. These Ca’+ channels and the effect of the toxin on the channels were blocked by the L-type Ca’+-channel antagonist gallopamil (200 PM). The present results suggest. therefore, that the investigated strain of 0. formosa contains homoanatoxin-a, which enhances the release of acetylcholine from peripheral cholinergic nerves through opening of endogenous voltage dependent neuronal L-type Ca’- channels.

Ke~vwrds: Acetylcholine; Phycotoxin: Ca’+ channel; Cqanobacterium; Oscillotoriu,f~rmosu; Patch clamp; Peripheral nervous system; Vesicle

1. Introduction species according to the current botanical taxonomy (Skul-

The blue-green algae (Cyanophyceae/Cyanobacteria) are a rather well circumscribed class of photosynthetic procaryotes. The class contains about I50 genera and about 2000 species (Fott, 197 I). Cyanophytes are the most diverse and widespread of the procaryotic microalgae. The present known toxigenic cyanophytes constitute about 40

Corresponding author. Tel.: +47-63 80 78 43: Fax: +17-63 80 78 I I.

berg et al., 1993). Several new toxigenic species are likely to be discovered as investigations are continued. Species of the genera Oscillutoria and Anabaena are among the most distributed toxin producing cyanophytes in eutrophicated freshwater (Berg et al.. 1986; Carmichael et al., 1990; Skulberg et al., 1993). Consequently, blooms of these organisms may have negative impact on water quality for water supply, public health, aqua culture and cattle rearing. Several cases of death of livestock, wildlife and pets following drinking of water during algal blooms have been

1382.6689/96/$15,00 Copyright 6 1996 El\evier Science B.V. All rights reserved. PII S 1382.6689(96)00059-2

reported (Smith and Lewis. 1987: Edwards et al., 1992; Falconer, 1993).

Four neurotoxins from Cyanobacteria have previously been studied in detail. Of these. anatoxin-a and anatoxin- a(s) seem unique to cyanobacteria. The other two, saxi- toxin and neo-saxitoxin, are present in certain marine phytoflagellates as well (Carmichael. 1994). Recently, ho- moanatoxin-a (2-(propan- I-oxo- 1 -yl)-9-azabicyclo- [4.2.l]non-2-ene), a low molecular weight bicyclic sec- ondary amine alkaloid, was isolated from a toxic oscillato- rialean strain. and the structure characterised and its toxic- ity investigated (Skulberg et al., 1992). Homoanatoxin-a (methylene-anatoxin-a) (see Fig. 4) is named according to its similarity in chemical structure to anatoxin-a, a toxin isolated from Anabaena ,flos aquae (Gorham et al., 1964; Huber. 1972; Devlin et al.. 1977; Carmichael and Gorham, 1978). Both anatoxin-a and homoanatoxin-a are potent nicotinic agonists (Spivak et al.. 1980. Wonnacott et al.. 1992) and have shown to cause complete, but reversible neuromuscular depolarising blockade of nicotinic choliner- gic receptors in striated skeletal muscle of mammals (Skul- berg et al., 1992). Anatoxin-a has previously also been shown to be a potent cholinergic agonist on both nicotinic and muscarinic receptors (Aronstam and Witkop, 1981). Homoanatoxin-a has an LD,, (i.p. mouse) of 250 pg/kg and leads to body paralysis and death by respiratory arrest in 7-12 min. These effects are similar and related to those observed for d-tubocurarine (Skulberg et al., 1992).

In addition to homoanatoxin-a, 0. ,fivmo.sa may contain other toxins of unknown chemical structures. Only a few have been identified in Cyanophyres. synthesised, or toxi- cologically described. The most intensively studied are the hepatotoxic cyclopeptides and the neurotoxic alkaloids produced by several cyanophyte species (Carmichael et al.. 1990; Codd and Poon, 1988).

In recent years toxins produced by the marine phytoflagellates P~ynnesium pntell(ferum have been re- sponsible for mortality of salmon and rainbow trout in fish farms in Norwegian coastal waters (Eikrem and Thrond- sen, 1993). Meldahl et al. (1994) have shown that the toxic mechanism of P. patelliferum could be due to an enhanced influx of Ca’+ in nerve terminals, and thereby a subse- quent massive increase in the release of neurotransmitters in the nervous system.

The aim of the present study was to examine the prejunctional mechanism of action of a water soluble toxin produced by 0. ,formosa on the peripheral cholinergic nervous system. We have used three well defined experi- mental preparations and techniques: (1) cholinergic nerves of the bronchial smooth muscle to study the effects of the alga toxin on the Ca”-dependent release of [3H]- acetylcholine. since the tissue is highly enriched with cholinergic nerves and is a ‘pure’ muscarinic preparation (Aas and Helle, 1982; Barnes, 1986): (2) brain synapto- somes to examine the effects of the toxin on the influx of 15Ca?f ions in nerve terminals (Nachshen and Blaustein,

1980): and (3) patch clamping (Neher and Sakmann, 1976) of neuronal GH,C, cells to study in detail the effects of the toxin on specific voltage dependent Ca’+ channels in neuronal cell membranes. We report here that the water- soluble toxin of the extract of lyophilized 0. formosa is homoanatoxin-a. The substance enhances the influx of Ca’+ ions in the cholinergic nerve terminals through open- ing of endogenous voltage dependent Ca2+ channels of the L-type.

2. Materials and methods

2. I. Animals

Male Wistar rats (200-300 g: from Mollegaard, Copen- hagen, Denmark) were used. The rats were given a stan- dard laboratory diet and water ad libitum. The animals were kept in standard laboratory cages, six in each, for approximately 2 weeks before the start of the experiments under constant photoperiod conditions (12 h : 12 h, L : D) at a temperature of 20°C and 60% relative humidity. The sawdust bedding was replaced daily to ensure that the concentration of ammonia was kept at a very low level. The rats had no signs of symptoms of respiratory tract infections.

2.2. Chemicals

4-Aminopyridine. D-600 (gallopamil), diltiazem hydro- chloride, trypsin inhibitor (from soya bean), verapamil hydrochloride and tetrodotoxin were purchased from Sigma Chemical Co. (MA. USA). w-Conotoxin GVIA and nifedipine were from Research Biochemicals (MA, USA). Trypsin (from bovine pancreas) was from Boehringer Mannheim (Germany). w-Agatoxin IV-A was from La- toxan (France) and w-conotoxin MVIIC from Bachem (Torrance, CA, USA). 1,2-Bis(aminofenoxi)etan-N, N, N’,N’-tetraacidic acid/acetoxymethyl (BAPTA/AM) was from Molecular Probes (OR, USA). Methyl- [ ‘H]choline chloride and “CaCl, was from New England Nuclear (Boston. MA, USA). Opti-fluor was from Packard Instr. (Groningen, Netherlands). All other chemicals were of analytical laboratory reagent grade.

2.3. Extraction of toxin from Oscillatoria formosa

The material of 0. .formosa Bory ex Gormont was cultivated from the strain NIVA-CYA 92 (Skulberg et al., 1992). The clone culture of this strain is kept in the NIVA Culture Collection of Algae (Skulberg and Skulberg, 1990). Mass cultivation was performed in a semi-continuous oper- ation under controlled laboratory conditions at the Norwe- gian Institute for Water Research (Skulberg et al., 1992). The biomass grown was harvested, and lyophilized mate- rial was used for the experimental investigations. The verification of acute toxicity was made using mouse bioas-

P. Aas et ul. / En~~irontttuntul Toxicology crnd Pharmacology 2 llYY61 223-232 225

says (Berg et al., 1987; Skulberg et al., 1992). The lyophilized material was dissolved in the physiological buffer (see Section 2.4), sonicated in l-2 min to disrupt the cells followed by centrifugation (12 000 rpm in 50 mitt). Different concentrations were obtained by dilution of the supernatant with buffer. The preparations from 0. formosa NIVA-CYA 92 are in the paper designated as Oscillatoria extracts.

2.4. Determination qf [~‘H~acetylcholine release

Following decapitation of the animals, the bronchi were removed and transferred to the physiological buffer (see below). Pieces of bronchial smooth muscle tissue (five tissue pieces X 1 mg wet wt.> were superfused in small chambers, as previously described by Aas and Fonnum (1986). Prior to the start of superfusion the tissue was incubated for 60 min in 1.1 PM [‘HIcholine chloride (370 GBq mmoll’) in the superfusion medium in a shaking water bath (25°C). The tissue was washed twice and superfused, by using a peristaltic pump with a flow rate of 200 ~1 mini’, for 60 min prior to the collection of samples. The release of [‘Hlacetylcholine was induced by raising the K+ concentration for 5 min. The superfusion media contained hemicholinium-3 (10 PM) to inhibit the high affinity uptake of choline (Yamamura and Snyder, 1973). The superfusion buffer (normal buffer) had the following composition (in mM1: NaCl 140.0, KC1 5.1, CaCl z 2.0, MgSO, 1.0, Na,HPO, 1.2, Tris-HCl 15.0, glucose 5.0. The depolarisation buffer (high K+) was as the normal buffer, but contained 51 mM KC], and the concentration of NaCl was reduced accordingly to keep the ionic strength constant. The media (pH 7.4) were oxy- genated and kept in a thermostatically controlled water bath at 25°C during the experiments. KC1 (51 mM) in- duced approximately 50% of the maximal release of [3H]acetylcholine. A total of two (Sl and S3) or three (Sl, S2 and S3) K+ stimulations (51 mM) for 5 min were separated by 35 min superfusion with normal buffer. The Oscillatoria extract and drugs were added prior to and during the second stimulation (S2) or instead of the second stimulation (S2) (time interval is given in figures and tables). Each experiment with drug and/or alga toxin had its own control experiment without drug and/or alga toxin. The collected fractions of the superfusion media were counted in an Opti-fluor scintillation cocktail for aqueous and non-aqueous samples. Release of ‘H during stimulation of tissues preincubated with [ “HIcholine has been shown to be a good measure for [‘H]acetylcholine release (Richardson and Szerb, 1974).

2.5. Determination Qf the ejffect of Oscillatoria extract on the coltage sensitice Ca2 f currents in GH,C, cells

The GH, cell line was chosen as preparation due to its well characterised electrophysiological properties. These cells are derived from the rat anterior pituitary. The cells

were grown as monolayer cultures under standard tissue culture conditions at 37°C. The cells used for recordings were seeded l-3 days before the experiments in 35 mm culture dishes. Prior to the experiments the culture medium was replaced with the following buffer (in mM): NaCl 150, KC1 5, CaCl, 10, MgC12 1.3, glucose 10, buffered by 10 mM Hepes-NaOH (pH 7.4). After initial control record- ings the medium was replaced with a corresponding buffer solution containing Oscillatoria extract (10 mg/ml). The electrode was filled with (in mM) Cs aspartate 120, CsCl 20. EGTA 0.25, MgATP 2, buffered by 10 mM Hepes- NaOH (pH 7.2). Cs’+ salts were used in the experiments to eliminate currents through the K+ channels. To facili- tate the Ca” currents, the Ca2+ concentration in the external buffer was 10 mM. Tetrodotoxin (1 PM) was added to the registration medium, in order to block any Na’ currents.

We have used the whole cell patch clamp method in the voltage clamp mode (Hamil et al., 1981). Patch electrodes were firepolished, and had a resistance of 5-15 M 0 when filled with buffer. The electrodes were connected to a List L/M-EPC 7 patch clamp amplifier and standard recording equipment, and the recordings were performed at room temperature.

2.6. Determination qf “‘Ca - in&x in rat brain synapto- somes

Cortical tissue from rats was homogenised in 0.32 mM sucrose by using a motor-driven Teflon-glass homogenise. The homogenate was centrifuged at 4°C 1000 X g for 10 min. The supematant was decanted and centrifuged at 12400 X g for 25 min and the pellet was resuspended in Hepes-buffered PSS containing (in mM): choline chloride 132, KC1 5.0, MgCl? 1.3. CaCl, 1.5, NaH,PO, 1.2, o-glucose 10 and Hepes 20 adjusted to pH 7.4 with Tris base. The protein concentration was adjusted to 1 .O-1.5 mg/ml with PSS. Ca’+ influx was carried out essentially as described previously (Lundy et al., 1989). Briefly, various concentrations of toxins were preincubated to- gether with the synaptosomes for 15 min (5 mM K+) at 30°C. At the end of the incubation period, a 100 ~1 aliquot of this synaptosomal suspension was injected into PSS or depolarising PSS (PSS in which 20 mM K+, final concen- tration, was substituted iso-osmotically for choline) con- taining 0.5 to 1 .O PCi of 15CaCI 2. The influx was allowed to proceed for various time periods up to 30 s, when the influx was stopped by the rapid dilution of the suspension with 4 ml of ice-cold quench buffer (Ca*+-free choline resting buffer containing 4 mM ethylene glycol-bis( p- aminoethyl ether)-l\r,N’-tetraacetic acid). The synapto- somes in the quench buffer were filtered rapidly under vacuum on 0.45 p.m Gelman GA-6-S filters on a Hoeffer filtration apparatus (Hoeffer Scientific, San Francisco, CA, USA) and washed twice with this same buffer. The filters were allowed to dry and the radioactivity trapped by the filters was determined by liquid scintillation spectroscopy.

2.7. Chemical analysis qf the Oscillatoria e.xtruct

A recently developed method for isolation and quantita- tive determination of alkaloid neurotoxins was used to obtain homoanatoxin-a extracts from algal cultures (Haugen et al.. 1994). Algal biomass was obtained by filtration of 0.3 liter water sample containing cyanophytes with a Whatman GF/C microfiber filter using a Millipore vac- uum stand. The biomass retained by the filter was lyophilized. 2 ml of 0.05 M acetic acid (Merck, analytical grade) was added to 13.3 mg of freeze-dried biomass and ultrasonicated for 5 min. The sample was centrifuged at 15 000 ‘pm for 30 min. The pH of the aqueous extract was adjusted to pH 2 11 with 0.5 M Na,CO, (Merck, analyti- cal grade) and passed through a Cl 8 cartridge of 1 ml volume (Sep-Pak, Waters) preconditioned with 4 ml of methanol (Rathburn, HPLC grade) followed by 8 ml of distilled water. The sample was applied to the cartridge which was washed with 8 ml of distilled water followed by 8 ml of methanol. The toxin-containing fraction was eluted with methanol. 20 ~1 of the toxin-containing fraction was evaporated to dryness in a reaction vial with a flow of nitrogen. In order to obtain a thermally more stable and less polar compound suitable for gas chromatography, the sample was derivatised as follows: The sample residue was dissolved in 150 ~1 of acetonitrile (Rathburn. HPLC grade) and heptafluorobutyric acid anhydride (Pierce) (3 : 1 by volume) was added as acylation reagent. Derivatisation was performed at 50 f 1°C for 20 min. The sample was evaporated to dryness with a flow of nitrogen and redis- solved in 100 ~1 cyclohexane (Rathburn, HPLC grade).

An HP-5980 gas chromatograph connected to an HP- 5989A quadrupole mass spectrometer was used for detec- tion of toxin. Separation was carried out on a 25 m X 0.2 mm i.d. fused silica column coated with 0.1 I ,um HP Ultra 2 stationary phase. He was used as carrier gas at a flow rate of 30 cm/s. 1 ~1 sample was injected splitless at an injector temperature of 250°C and the following tempera- ture program was employed: 100°C for 2 min, then lO”/min to 250°C isothermal for 5 min.

Racemic synthetic homoanatoxin-a was used for verifi- cation of the test compound. Full-scan spectra were recorded from m/z 50 to 400 at a scan rate of 600 amu/s. The mass spectrometer was operated in the negative ion chemical ionisation (NICI) mode with methane (Messer Griesheim, 99.95% purity) as reagent gas. The ion source pressure was 0.5 T, and the ion source temperature was 200°C. The electron energy of the primary electrons was in the order of 120 eV.

2.8. Statistics

Statistical analyses were done with Student’s t test (two-tailed). The fractional rate of evoked release of [“Hlacetylcholine, peak areas, as well as basal release before and after the depolarisation period. and ratios be- tween peak areas were calculated. K’-evoked [jH]-

acetylcholine release was calculated by subtracting the basal release from the evoked release. The K’-evoked release was always performed three times (Sl, S2 and S3) consecutively and the K+-evoked release of [3H]- acetylcholine was calculated as percent of that released in the first stimulation (Sl) in each experiment. Mean and standard error of the mean (S.E.M.) were calculated for all data and n equals the number of experiments. Significance for difference between a control group and an experimen- tal group were calculated by Student’s t test. The experi- ments included a control group and an experimental group and they were always performed in parallel at the same time. * * * P < 0.01; * * P < 0.02; * P < 0.05; llS. P > 0.05.

3. Results

3.1. The effect of Oscillatoria extract on the release of [ ‘Hlacetylcholine ,from cholinergic nerces

The water extract of lyophilised 0. &~;3yrnosu enhanced the spontaneous as well as the K’ (5 1 mM) evoked release of [3H]acetylcholine in a dose-dependent manner (Fig. IA and I B) in the concentration range studied (l-20 mg freeze-dried alga/ml buffer). The concentrations of toxin used in the experiments are comparable with levels ob- served in nature following lysis of water blooms of neuro- toxic cyanophytes (Skulberg, 1996). The increase in the K+-evoked release of [3H]acetylcholine in the presence of algae (20 mg freeze-dried algae/ml) was approximately 77% relative to control stimulation with Kf (51 mM) (Table 1, experiments A and C). The release of [3H]- acetylcholine was 46% relative to a control stimulation with K+ (51 mM) when the nerves were stimulated by the O.scillatnria extract (20 mg/ml) in a low concentration of K+ (5. I mM) (Table 1, experiment D). The effect of the Uscillatoria extract on the release of [ “Hlacetylcholine was in general reversible, and only a small effect was observed in the third consecutive stimulation (S3) by K+ without Oscillatoria extract present, compared to control (S 1) (Table 1).

A low concentration of Ca’+ ions (0 mM Ca2+ in the buffer), the presence of L-type Ca’+-channel blockers such as nifedipine (100 PM) and diltiazem (100 PM), or the N-type Ca’+-channel blocker w-conotoxin GVIA (1 .O PM), all which reduced the K+ (51 mM) evoked release of [‘H]acetylcholine significantly, did not inhibit the stim- ulation of the [ ‘H]acetylcholine release induced by the Oscillatoria extract (Table 1). The L-type Ca’+-channel blocker verapamil in a concentration of 100 PM, but not 1 .O PM. on the other hand, reduced both the K’-evoked release as well as the Oscillatoria extract induced release of [3H]acetylcholine (Table 1). The release of [jH]- acetylcholine induced by the Oscillatoria extract in the

P. Aas rf al. /t‘nr~ironmenfnl Toxicology and Pharmacology 2 (19961 223-232 221

.4. I x0

I 60

140

120

? 100 c * 8.

60

40 i

20

0 1

Sl

I Control

K’ (51 mM

mglml

1~ IO

mglml

s2

mglml

K+ 1 2.5 5 10 15 20 (51 &) my/ml mg/ml mgiml mg/ml mg/ml mgiml

Sl S2 _____~___- Fig. 1. The effect of a water extract of the cyanophyte 0. ,formosa on the release of [‘Hlacetylcholine from peripheral cholinergic nerves in airways in the presence (A) or absence (B) of high K* (51 mM). Bronchial smooth muscle was stimulated for 5 mitt three times consecutively with K+ (51 mM) (A) with a water extract of lyophilized 0. formosa (mg/ml buffer) present during S2, or (B) stimulated 5 min three times consecu- tively to release of [‘Hlacetylcholine; two times with potassium (51 mM) (S 1 and S3) and once with Oscillnroria extract (S2. 5 min). In experiment A the Oscillatoria extract was present 5 min before and during the second stimulation (S2). but not during the first (S 1) or the third stimulation (S3). The time interval between the three consecutive stimulations were 35 min. A control experiment without Oscillaroricl extract present during S2 was performed for each experiment (A). The evoked release of [iH]acetylcholine in S2 and S3 was calculated relatively to the first KC evoked release in each experiment. Each value is calculated from six experiments, and S.E.M. is calculated for all data. (100% = 112317 dpm/g tissue w.w./min, n = 14.)

low K+ buffer (5.1 mM) was reduced from 46% to 18% by verapamil (0.1 mM) and the effect was reversible in the absence of the Ca’+-channel blocker. w-Agatoxin IV-A (0.2 FM), on the other hand, potentiated both the K’ (51 mM) and Oscillatoria extract evoked release of [ 3H]acetylcholine (Table 1).

The presence of the intracellular Ca’+ chelator BAPTA/AM (30 PM) before (30 min) and during stimu- lation of the nerves with Oscillatoria extract had no effect on the release of [“H]acetylcholine neither in the presence nor in the absence of extracellular Ca’+ ions (Table 2). Stimulation of the release by high K+ (51 mM) (control stimulation) was not altered by BAPTA/AM (Table 2).

Tetrodotoxin (3.0 FM), which is an inhibitor of Naf channels, did not alter the K+ or the alga induced release of [ -?H]acetylcholine (Table 3). 4-Aminopyridine (1 .O mM), which mainly blocks voltage dependent Kf channels. on the other hand, almost completely blocked the K--evoked release of [‘H]acetylcholine. The alga induced release of [ “H]acetylcholine was also substantially reduced when 4- aminopyridine was present in a concentration of 1.0 mM, but not when present in a concentration of 30 PM (Table 3).

To eliminate any proteinaceous toxins from stimulating the cholinergic nerves, the Oscillatoria extract (20 mg/ml) was trypsinated with 100 and 300 pg trypsin/ml buffer and trypsin inhibitor (300 and 900 pg/ml respectively) was added before exposure of the cholinergic nerves. There was no alteration of the effect of the Oscillatoria extract on the release of [‘Hlacetylcholine (S2) following trypsination of the extract or on the subsequent stimulation (S3) with Kf (51 mM) (not shown).

3.2. The eflect of Oscillatoria extract on 4sCa’ ’ uptake in rat bruin synaptosomes

““Ca’+ influx was measured in rat brain synaptosomes during exposure to Oscillutoria extract (10 mg freeze-dried algae/ml buffer), a concentration giving approximately 50% of maximal alga induced release of [3H]acetylcholine from cholinergic nerves in the bronchial smooth muscle. The Oscillutoria extract enhanced the influx of 45Ca’+ significantly from 7.00 f 0.97 to 14.01 * 1.59 nmol/mg protein (P < 0.05) in a tetrodotoxin insensitive manner (Fig. 2). The influx of “Ca’+ upon exposure to the Oscillatoria extract was not altered by the N-, P- and Q-type Ca *+-channel blocker w-conotoxin MVIIC even in a high concentration (1 .O PM) (Fig. 2).

3.3. The eflect of Oscillatoria extract on the voltage sensi- tiL)e Cu2 + currents of GH,C, cells

The effect of Oscillatoria extract on the transmembrane Ca’- currents were examined using the whole cell patch clamp method in voltage clamp mode. The holding poten- tial was - 67 mV. and the 256 ms command pulse varied in 10 mV steps from - 87 mV to + 43 mV. Fig. 3A shows

228 P. Aas rt ul. / Enkwunentul Toxicology and Pharmacology 2 f 19961 223-232

Table 1 K+ (51 mM) and 0. formosu (20 mg/ml) induced release of [ ‘H]acetylcholine with Ca’+-channel blockers present before and during the second stimulation (S2)

Experiment s2 Stimulation X2 SI (5%) s2 (5%) Significance of S3(%) n difference

A Control 51 mM KC1 100 83+2 71+2 29 B Control, 0 M Ca’ + 51 mM KC1 100 13+1 (A) * * 72 f 7 3 C 20 mg/ml 0.f. 51 mM KC1 100 160f28 (A) * x 71+7 8 D 20 mg/ml 0.f. 100 46 +4 101 + 2 15 E 20 mg/ml O.f:, 0 M Ca’+ 100 35 i 2 (D) n.s. 14 f 4 3 F Control, 1.0 PM verapamil 51 mM KC1 100 83Ik I (A) n.s. 81 *3 4 G Control, 0. I mM verapamil 51 mM KC1 100 49+3 (A) * * * 74 f 8 6 H 20 mg/ml Olf., I .O FM verapamil loo 125 i 10 (D) n.s. 62 +4 6 I 20 mg/ml 0.f.. 0. I mM verapamil 100 18*3 (D) * * * 88 +6 9 J Control, 0. I mM nifedipine 51 mM KC1 100 66,l (A) * * * 79i4 9 K 20 mg/ml 0.f.. 0. I mM nifedipine 100 33 *5 (D) n.s. 85 +3 9 L Control. 0.1 mM diltiarem 51 mM KCI 100 41+2 (A) * * 70*3 6 M 20 mg/ml O.f., 0.1 mM diltiazem 100 57+12 (D) ns. 83 i 2 6 N Control, 1.0 PM c+conotoxin 51 mM KCI 100 67 f 3 (A) * * * 63 + 3 6 0 20 mg/ml O.f., 1.0 FM wconotoxin 100 96 + 8 (D) * - * 64 f 2 6

L

Control. 0.2 PM wagatoxin 51 mM KC1 100 99 *9 (Al * _ * 82+4 6 20 mg/ml O.f., 0.2 FM w-agatoxin 100 245 +75 (D) * 96+5 6

Bronchial smooth muscle was stimulated for 5 min three times (S 1, S2 and S3) consecutively with K+ (5 I mM) (A, B, F, G, J, L, N, P), K+ (5 1 mM) and 0. formosa (0.f.f.) extract (20 mg/mll (S2) (C) or with 0.f. extract (20 mg/ml) and voltage dependent ion channel blockers present in the perfusion buffer 62) (H, I, K. M. 0, Q). Ion channel blockers and Oscillatoria extracts were present 5 min before and during stimulation, except in experiment N and 0 were w-conotoxin GVIA was present and experiments P and Q where w-agatoxin IV-A was present 20 min before and during stimulation (S2). The results are mean + S.E.M. of n experiments. Significance for difference ( * * *, P < 0.01; * * , P < 0.02: * P < 0.05; ns., P > 0.05) from experiment in brackets are shown.

Table 2 The effect of the intracellular calcium chelator BAPTA/AM on the regulation of K+ (51 mM) or 0. furmosa (20 mg/ml) evoked release of [ ‘Hlacetylcholine

Experiment S2 Stimulation S2 Sl (So) SZ(%) Significance of S3(%) n difference

A Control 51 mM KC1 100 83 + 2 71+2 29 B 20 mg/ml 0.f: 100 46+4 101+2 I5 C Control, 30.0 PM BAPTA/AM 51 mM KCI 100 83+3 (A) n.s. 76+5 6 D 20 mg/ml O.f., 30.0 FM BAPTA/AM 100 44 + 2 (B) n.s. 77k2 6 E 20 mg/ml 0.f.. 0 mM Ca’+. 30.0 PM BAPTA/AM 100 3X+5 (B) n.s. 90F6 6

Bronchial smooth muscle was stimulated for 5 min three times (Sl. S2 and S3) consecutively with K’ (51 mM) (A, C) or only with 0. formosu t0.f.) extract (20 mg/ml) present in the perfusion buffer (S2) (B, D, E). Oscillatoria extract was present 5 min before and BAPTA/AM was present 30 min before and both were present during stimulation (S2). The results are mean + S.E.M. of n experiments. Significance for difference (n.s., I’ > 0.05) from experiment in brackets are shown.

Table 3 K+ (51 mM) or 0. ,fix-rnosz (20 mg/ml) induced release of [‘H]acetylcholine with the Na+-channel blocker tetrodotoxin or the K+-channel blocker 4-aminopyridine present 5 min before and during the second stimulation

Experiment S2 Stimulation S2 Sl(7c’c) S2CR) Significance S3(%) n

Control 20 mg/ml 0.f. Control. 3.0 FM tetrodotoxin 20 mg/ml Olf., 3.0 FM tetrodotoxin Control, 30 /*M 4-aminopyridine 20 mg/ml 0.f.. 30 FM 4-aminopyridine Control, I .O mM 4.aminopyridine 20 mg/ml O.j’., I .O mM 4.aminopyridine

51 mM KC1 100 loo

51 n&l KC1 100 100

51 mM KCI 100 100 100 100

83 &2 46i4 s2*4 34+5 80* 3 45 i 6

4+1 27 + 1

71+3 29 101 + 2 I5

(A) n.s. 82 +6 6 (B) n.s. 85 + 2 6 (A) n.s. 67 f 6 3 (B) ns. 91 +9 3 (A) * 1 * 88 +9 3 (Bl * * I 87 i 3 6

Bronchial smooth muscle was stimulated for 5 min three times (S I, S2 and S3) consecutively with K+ (5 1 mM) (A, C, E) or only with 0. forrnosa (0.f.) extract (20 mg/ml) present in the perfusion buffer (S2) (B, D, F, H). Drugs and Oscillatoriu extracts were present 5 min before and during stimulation (S2). The results are mean f S.E.M. of n experiments. Significance for difference (* * * , P < 0.01; ns., P > 0.05) from experiment in brackets are shown.

P. Aas et a/. / Enr~ironmental Toxicolop and Pharmacology 2 (19Y6) 223-232 229

*

Basal -

0. formosa 0. formoca 0. formosa

(Control) Tetrodotoxin w-Conotoxin

MVIIC

xxx xxx xxx

T- 7

i-

Fig. 2. The effect of a water extract of lyophilized 0. fi~mosa (10 mg/ml) on the of 45Ca’+ influx (nmol mg protein-’ ) in rat brain synaptosomes in the presence and absence of the Na+-channel blocker tetrodotoxin (1 .O FM) and the P/Q calcium channel blocker wconotoxin MVIIC (1 .O PM). The results are the mean f S.E.M. of 3-4 experiments. * * * P < 0.01.

the current traces at the 3 mV clamp potential averaged from ten cells in both the control solution and after lo-20 min exposure to the extract. Fig. 3B shows the correspond- ing I/V curves. Following lo-20 min incubation with Oscillutoriu extract (10 mg/ml) a 3-4-fold increase in the Ca*+ currents compared with control was observed (Fig. 3C). This current increase was observed in both the T-type Ca2+ channels (measured at the initial peak of the Ca*+ current) and the L-type Cal- channels (measured after 230 ms of the pulse. The voltage dependence of the inward current components was. however, not affected by the Oscillatoria extract. The current observed following expo- sure to Oscillatoria extract was almost completely blocked by the Ca’+-channel blocker gallopamil (200 PM) (Fig. 30

After 30-40 min exposure to the Oscillatoria extract,

Fig. 3. Recordings of voltage-dependent Ca’- currents in whole-cell voltage clamped clonal anterior pituitary cells in the absence and pres- ence of a water extract of lyophilized 0. formosa (20 mg/ml) and the Ca’+-channel blocker gallopamil. A: recordings of voltage-dependent Ca’+ current in whole-cell voltage clamped clonal anterior pituitary cells. Holding potential was -67 mV. The upper trace is the control. When the cells were exposed to O.scil/atoria extract (lower trace). the Ca’+ current increased 3-4 times the control. B: the current-voltage relation for the current through the T-type (0, before exposure to gallopamil (200 FM); n , after exposure to gallopamil (200 wM)) and the L-type (0, before exposure to gallopamil (200 FM); (0, after exposure to gallopamil (200 PM)) Ca’+ channela in control experiments. C: the same current-voltage relations and exposure to gallopamil as in B, but following exposure to Oscillutorirr extract. The outgoing currents at high voltage steps are likely K* currents through incompletely blocked K+ channels.

the effect on the Ca” currents was strongly reduced or disappeared completely. An outgoing current was observed at high voltages. The outward L-type current was enhanced in the presence of the Oscillatoria extract. The outward current components are probably due to incompletely blocked K’ channels by intracellular Cs*+, the low con- centration of internal EGTA and an enhancement of Ca*+- dependent K+ currents evoked by the Ca*’ influx. This is in accordance with the fact that GH,C, cells are shown to have Ca*+-dependent K+ channels (Sand et al., 1989). This explains the shift of the reversal potential of the

‘I Control .._ ..,---, _~ ---

‘_. Extract of 0 fortno.w . ..-_-- .__ ,. .- -.

__,_- .‘. _-

; ,I’

IO0 Ins

B. Control

C. Extract of 0 fornwsu

p 0 CH.CH,

FK, CY _”

Fig. 4. Full scan negative ion chemical ionisation spectrum of homoana- toxin-a standard (A) and water extract of lyophilized 0. ./orrnosa (B).

L-type Ca’ A current to the left. The reversal potential would as expected shift in a positive direction for a conduction increase for Ca’+ only. The observed reversal potential was, however, probably influenced by an outward K- current through Ca’+-dependent K+ channels evoked by the Cat influx.

3.4. Chemicul analysis of the nlga estrwt 0~ GC/MS

The negative ion chemical ionisation spectrum of the derivatised homoanatoxin-a standard shows the molecular ion at m/z 375 as well as abundant fragment ions at m/z 355, m/z 335, m/z 31.5 and m/z 295 (Fig. 4). The fragmentation is characteristic for the heptafluorobutyryl group which promotes a series of HF eliminations leading to [M-I-IF]-‘. [M-2HF]-‘, [M-3HF]-’ and [M-4HF]-’ ions. The dominating signal of the chromatogram of the test sample had identical retention time as the homoanatoxin-a standard. The spectrum of this peak shows the same mass fragmentation pattern. Quantification based on synthetic homoanatoxin-a as external standard gave a toxin concen- tration of the test material of 0.6 pg/mg dry weight.

4. Discussion

In the present work we have shown that water extracts of lyophilized 0. ,formosa contain a neuroactive substance

of non-proteinaceous nature, and by mass-spectrometric analysis it was shown that only one toxin could be identi- fied. This toxin was structurally equal to homoanatoxin-a, which is a secondary amine alkaloid identified as methy- lene-anatoxin-a (Fig. 4). The toxin-induced increase in the release of [ “H]acetylcholine from the cholinergic nerves was observed in both the K+-depolarised as well as in the non-depolarised experimental situation. This indicates that the toxin may induce release of [ ‘Hlacetylcholine either by opening of voltage-dependent Kf channels. or alterna- tively by opening of other voltage-dependent endogenous ion channels or by producing Ca*+ channels in the nerve terminal and thereby induce release of neurotransmitters. The toxin evoked release of [3H]acety1choline was sub- stantially reduced only by the use of the L-type Ca2’-chan- nel blocker verapamil (a phenylalkyl-amine) in high con- centrations. while other L-type channel blockers, such as nifedipine (a dihydropyridine) and diltiazem (a benzo- thiazepine), had no effect on the release. The reason for this difference in the efficacy among the L-type Ca2+- channel blockers is not clear, but could be due to a diversity among presynaptic Ca’+ channels (Wheeler et al., 1994) and thereby a difference in the sensitivity to Ca’+-channel blockers. Such a difference among L-type Ca’+ channels is, however, shown in the central nervous system, but there is no good evidence for the existence of such L-type Cal+ channels in peripheral nerves (Lundy and Few, 1993). Several studies have also been unable to demonstrate dihydropyridine attenuation of Ca” influx into synaptosomes (for review see Miller and Freedman, 1984).

Although high concentrations of verapamil had to be used to block the effect of the toxin on the release of [‘H]acetylcholine, the effect of the toxin was probably still a specific effect on L-type Ca’+ channels and not an effect on endogenous Na’ channels or K- channels, since tetrodotoxin (3.0 ,uM) and 4-aminopyridine (30 PM) did not alter the effect of the toxin induced release of [‘H]acetylcholine under the low K’ (5.1 mM) condition. Only following exposure to high concentrations of 4- aminopyridine (1.0 mM) was the toxin-evoked release of [ ‘H]acetylcholine reduced and not enhanced. Furthermore, an inhibition of the release of [ 3 H]acetylcholine by dilti- azem (0.1 mM) and nifedipine (0.1 mMl, as was seen for 4-aminopyridine, would have been expected if K+ chan- nels play a role in the action of the toxin, due to a non-specific blockage of the Kf channels at the high concentration of the Ca’--channel blockers used in the present experiments.

Naturally occurring toxins which have been shown to be blockers of N- and P-type Ca’+ channels were not effective in reducing the effects of the 0. ,furmosa toxin on the release of [3H]acety1choline. o-Conotoxin GVIA and w-agatoxin IV-A both enhanced rather than decreased the effect of 0. formosa toxin on the release of [‘HI- acetylcholine. The fact that w-Conotoxin GVIA and o-

agatoxin IV-A enhanced rather than decreased the effect of the toxin suggests that w-conotoxin GVIA and w-agatoxin IV-A might have other effects. such as binding to phosphe lipid bilayers as has previously been shown for other Cazf-ion channel blockers (Thomas and Verkleij, 1990). The two naturally occurring toxins could combine with the toxin of 0. ,formosa and thereby mediating its interaction with the plasma membrane, resulting in increased effects of the 0. formosa toxin.

Evidence was also obtained to imply that extracellular Ca’+ was important for the effect of homoanatoxin-a from 0. formosa on the cholinergic nerves. Chelating intra- cellular Cal + by using the intracellular Ca2+ chelator BAPTA/AM had no effect on the Kf or homoanatoxin-a evoked release of [3H]acetylcholine. This indicates, there- fore, that homoanatoxin-a in the Oscillatoria extract in- duce opening of voltage-dependent Ca” channels and that only extracellular Ca’+ plays an important role in the mechanism of action of this toxin on the cholinergic nerves. It is, however, difficult to explain why the omis- sion of Ca’+ in the perfusion buffer had only a small and insignificant effect on the Oscillatoria-toxin evoked re- lease of [ 3H]acetylcholine, but extraneuronal tissue Ca’ ’ might play a role.

Further evidence for the involvement of Ca’+ channels was obtained in the experiments on a synaptosomal prepa- ration from rat brain. A significant, tetrodotoxin (I .O PM) insensitive increase in the influx of -‘“Ca’+ was observed following exposure to the Oscilhtoria extract (I 0 mg/ml), which was not inhibited by w-conotoxin MVIIC, an in- hibitor of N- and P-, and at high concentrations such as used in the present experiments. Q-type Ca’- channels (Hillyard et al., 1992). This indicates that homoanatoxin-a in the algal extract may neither induce opening of endoge- nous Ca’+ channels, generate new N-, P- or Q-type ion channels permeable to Cal’ or that it might be due to a pore forming property of the toxin. By using a third technique, patch clamping of the well characterised neu- ronal GH,C, cell type isolated from the rat pituitary gland. it was shown that homoanatoxin-a in the Oscillutoria extract opened both L- and T-type Ca’- channels in the membranes of these neuronal cells. This channel opening effect was shown to be sensitive to the L-type Ca’+-chan- nel blocker gallopamil (200 FM). Although T-type Ca’+ channels were opened, it is not likely that these type of Ca’+ channels play any role in the peripheral cholinergic nerves. T-type Ca2+ channels have not been implicated in neurotransmitter release in any preparations (Spedding and Paoletti, 1992) and the biophysical properties of this T-type channel suggest that it carries too little current and inacti- vates too rapidly to be involved in neurotransmitter release (Spedding and Paoletti, 1992).

It is, therefore, most likely that it is an endogenous L-type Ca’+ channel which is opened by the algal toxin. It is highly unlikely that the toxin induced new Ca” chan- nels with exactly the same voltage dependency as the

endogenous Ca’+ channels. Since Ihe voltage dependency: shown in the present experiments on GH,C, cells, was the same both before and after addition of the Oscillatoria toxin, the present results provide evidence for the involve- ment of opening of only existing endogenous Ca” chan- nels and not opening of new types of voltage-dependent Ca’+ channels.

In conclusion, it is reason to suggest that the strain NIVA-CYA 92 of 0. formosa contains homoanatoxin-a. Homoanatoxin-a induces a concentration-dependent open- ing of endogenous voltage dependent Ca2+ channels in the cholinergic nerve terminals, and that these voltage-depen- dent ion channels probably are similar to L-type Ca’+ channels. The findings of the present experiments, by using three different biological methods described, all support this conclusion.

Acknowledgements

The authors are grateful to Ms. Randi Skulberg (Norwegian Institute for Water Research) for providing the cultured cyanophyte material, and to Ms. Rita Tans@ (Norwegian Defence Research Establishment) for her ex- cellent technical assistance.

References

Aas, P. and K.B. Helle, 1982, Neurotensin receptors in the rat bronchi. Regul. Pept. 3, 405.

Aaq. P. and F. Fonnum, 1986. Presynaptic inhibition of acetylcholine release, Acta Physiol. &and. 127. 33.5.

Aronstam. R.S. and B. Witkop, 1981, Anatoxin-a interactions with cholinergic synaptic molecules, Proc. Natl. Acad. Sci. USA 78. 4639.

Barnes, P., 1986, Neural control of human airways in health and disease, Am. Rev. Respir. Dis. 134, 1289.

Berg, K.. O.M. Skulberg. R. Skulberg, B. Underdal and T. WillCn, 1986. Observations of toxvz blue-green algae (Cyanobacteria) in some scan- dinavian lakes, Acta Vet. Stand. 27. 440.

Berg, K., W.W. Carmichael, O.M. Skulberg. C. Benestad and B. Under- dal. 1987, Investigation of a toxic water bloom of Microcyds

treruginosu (Cyanophyceae) in Lake Akersvatn, Norway Hydrobiol. 144. 07.

Carmichael. W.W., 1994. The toxins of Cyanobacteria, Sci. Am. 1, 64. Carmichael. W.W. and P.R. Gorham. 1978. Anatoxin from clones of

Anabawrr jlos-ayutle isolated from lakes of western Canada, Mitt. Int. Verein. Limnol. 21, 285.

Carmichael, W.W., N.A. Mahmood and E.G. Hyde, 1990. Natural toxins from cyanobacteria (blue-green algae), in: Marine Toxins: Origin, Structure and Molecular Pharmacology, ACS-Symposium Series 418, eds. S. Hall and G. Strichartz (American Chemical Society, Washing- ton, DC) p. 87.

Codd. G.A. and G.K. Poon, 1988, Cyanobacterial toxins, in: Biochem- istry of the Algae and Cyanobacteria, eds. L.I. Rogers and J.G. Gallon (Oxford Science Publications, Clarendon Press, Oxford, UK) pp. 283.

Deulin, J.P.. O.E. Edwards, P.R. Gorham. N.R. Hunter, P.K. Pike and B. Stavric. 1977, Anatoxin-a, a toxic alkaloid from AnabarnnJlos-aquae

NRC-44h, Can. J. Chem. 55, 1367. Edwards, C.. K.A. Beattie, CM. Scrimgeour and G.A. Codd, 1992.

Identification of anatoxin-a in benthic cyanobacteria (blue-green al-

232 P. Atrs et al. / En~~ironmentul Toxicology and Pharmacology 2 (1996) 223-232

gae) and in associated dog poisoning at Loch Irish, Scotland, Toxicon 30, 1165.

Eikrem, W. and J. Throndsen, 1993, Toxic prymnesiophytes identified from Norwegian coastal waters, in: Toxic Phytoplankton Blooms in the Sea, eds. T.J. Smayda and Y. Shimizu (Elsevier Science Publish- ers, Amsterdam, Netherlands) p. 687.

Falconer, LR., 1993, Algal Toxins in Seafood and Drinking Water (Academic Press, London, England) p. 224.

Fott, B., 1971, Algenkunde. Die Taxonomie der einzelnen Algenstamme (G. Fischer Verlag, Stuttgart, Germany) p. 23.

Gorham, P.R., J. McLachlan, U.T. Hammer and W.K. Kim, 1964, Isolation and culture of toxic strains of Anabuenn,Pos-aquae (Lyngb.) de Breb., Int. Assoc. Theor. Appl. Limnol. 15, 796.

Hamil, O.P., A. Marty, E. Neher, B. Sakman, F.J. Sigworth, 1981, Improved P- clamp technique for high-resolution current recording from cell-free membrane patches, Pfliigers Arch. 391, 85.

Haugen, J.E., O.M. Skulberg, R.A. Andersen, J. Alexander, Cl. Lilleheil, T. Gallagher and P.A. Brough, 1994, Rapid analysis of cyanophyte neurotoxins: An improved method for quantitative analysis of ana- toxin-a and homoanatoxin-a in the sub-ppb to ppb range, Algol. Stud. 75, 111.

Hillyard. D.R., V.D. Monje, I.M. Mintz, B.P. Bean, L. Nadasdi, J. Ramachandran, G. Miljanich, A. Azimi-Zoonooz, J.M. McIntosh, L.J. Cruz, J.S. Imperial and B.M. Olivera. 1992, A new conus peptide ligand for mamalian presynaptic Cal- channels. Neuron 9. 69.

Huber, C.S., 1972, The crystal structure and absolute configuration of 2,9-diacetyl-9-azobicyclo(4,2,l)non-2,3-ene, Acta Crystallogr. 328, 2577.

Lundy, P.M. and R. Few, 1993, Pharmacological characterization of voltage-sensitive Ca’+ channels in autonomic nerves, Eur. J. Pharma- col. 231, 197.

Lundy, P.M.. K. Stauderman, J. Comet and R. Frew. 1989, Effect of w-conotoxin GVIA on Ca’+ influx and endogenous acetylcholine release from chicken brain preparations. Neurochem. Int. 14, 49.

Meldahl, A.S.. S. Eriksen, V.A.T. Thorsen, 0. Sand and F. Fonnum. 1994, The toxin of the marine alga P~ninesi~m patelliferum in- creases cytosolic Cal+-currents in synaptosomes and voltage sensitive Cazf-currents in cultured pituitary cells, in: Biological Membranes: Structure. Biogenesis and Dynamics, NATO-AS1 Series. Vol. H82, ed. J.A.F. Op den Kamp (Springer Verlag. Berlin, Germany) p. 331.

Miller, R.J. and S.B. Freedman, 1984, Are dihydropyridine binding sites voltage sensitive calcium channels‘?, Life Sci. 34, 1205.

Nachshen, D.A. and M.P. Blaustein. 1980, Some properties of potassium-stimulated calcium influx in presynaptic nerve endings, J. Gen. Physiol. 76, 709.

Neher, E. and B. Sakmann, 1976, Single channel currents recorded from membrane of denervated frog muscle fibren, Nature 260. 799.

Richardson, I.W. and J.C. Szerb. 1974. The release of labelled acetyl- choline and choline from cerebral cortical slices stimulated electri- cally. Br. J. Pharmacol. 52, 499.

Sand. 0.. B. Chen, Q. Li. H.E. Karlsen, T. Bjaro and E. Haug, 1989, Vasoactive intestinal peptide (VIP) may reduce the removal rate of cytosolic Ca’+ after transient elevations in clonal rat lactotrophs. Acta Physiol. Stand. 137. 113.

Skulberg, O.M., 1996, Toxins produced by cyanophytes in Norwegian inland waters - health and environment, in: Chemical Data of Plant, Animal and Human Tissues as a Basis of Geomedical Investigations, ed. J. Lag (The Norwegian Academy of Science and Letters, Oslo, Norway) p. 197.

Skulberg. R. and O.M. Skulberg, 1990, Research with Algal Cultures - NIVA’s Culture Collection of Algae (Norwegian Institute for Water Research, Oslo, Norway).

Skulberg, O., W.W. Carmichael, R.A. Andersen, S. Matsunaga. R.E. Moore and R. Skulberg. 1992, Investigations of a neurotoxic oscilla- torialean strain (Cyanophyceael and its toxin. Isolation and characteri- zation of homoanatoxin-a, Environ. Toxicol. Chem. Il. 321.

Skulberg, O.M., W.W. Carmichael, G.A. Codd and R. Skulberg, 1993, Taxonomy of toxic Cyanophyceae (Cyanobacteria), in: Algal Toxins in Seafood and Drinking Water. ed. I.R. Falconer (Academic Press, London. UK) p. 145.

Smith, R.A. and D. Lewis, 1987, A rapid analysis of water for anatoxin-a. the unstable toxic alkaloid from Anaboena flos-aqua, the stable non-toxic alkaloids left after bioreduction and a related amine which may be nature’s precursor to anatoxin-a, Vet. Hum. Toxicol. 29, 2,153.

Spedding, M. and R. Paoletti. 1992, Classification of calcium channels and the sites of action of drugs modifying channel function, Pharma- col. Rev. 44, 3. 363.

Spivak, C.E., B. Witkop and E.X. Albuquerque, 1980, Anatoxin-a: a novel, potent agonist at the nicotinic receptor, Mol. Pharmacol. 18, 384-394.

Thomas, P.G. and A.J. Verkleij. 1990, The dissimilar interactions of the calcium antagonists flunarizine with different phospholipid classes and molecular species: a differential scanning calotimetry study, Biochem. Biophys. Acta 1030, 2 Il.

Wheeler, D.B., A. Randall and R.W. Tsien, 1994. Roles of N-type and Q-type Ca’+-channels in supporting hippocampal synaptic transmis- sion, Science 264. 107.

Wonnacott, S., K.L. Swanson, E.X. Albuquerque, N.J.S. Huby, P. Thompson and T. Gallagher, 1992, Homoanatoxin: A potent analogue of anatoxin-a, Biochem Pharmacol. 43. 419.

Yamamura, HI. and S.H. Snyder, 1973, High affinity transport of choline into synaptosomes of rat brain, J. Neurochem. 21, 1355.