1. Introduction

1.1 Cyanobacteria

Cyanobacterial evolution begun at least 2000 million years ago according to paleontological data (Hofmann, 1976 and Knoll & Golubic, 1991: cited by Golubic et al. , 1995). Cyanobacteria are a morphologically diverse group of oxygenic photosynthetic prokaryotes (Rippka et al. , 1979), which are phylogenetically closely related to each other and to chloroplasts (Giovannoni et al. , 1988). Chloroplasts have originated from cyanobacteria by one or more endosymbiotic events between non-photosynthetic eukaryotic organisms and cyanobacteria (Douglas, 1994).

Cyanobacteria and oxygenic Eukaryotes, algae and plants, have generated oxygen in the earth’s atmosphere. Atmospheric oxygen concentrations were high enough to promote the selection of heterocystous cyanobacteria approximately 2100 million years ago based on chemical analysis of ancient weathering profiles (Holland & Beukes, 1990: cited by Golubic et al. , 1995). The molecular phylogenetic study of Giovannoni et al. (1988) has shown that the last diverging cyanobacterial groups were the heterocystous groups in which the nitrogen (N) fixing enzyme, nitrogenase, was protected from oxygen in heterocysts.

1.1.1 Geographic and ecological distribution of cyanobacteria

Due to their early evolutionary history, cyanobacteria occur abundantly in a wide range of habitats (Schopf, 1994). Cyanobacteria are known as freshwater (Hoffmann, 1996) and marine (Hoffmann, 1994) organisms, and in addition, especially the filamentous forms, also occupy a variety of terrestrial habitats (Hoffmann, 1989). Cyanobacteria form symbiotic associations with a wide range of organisms: algae, fungi, pteridophytes, gymnosperms, angiosperms, some animals, bacteria, and non-photosynthetic protists. In these associations, the cyanobacterial symbionts (cyanobionts) are usually heterocystous and filamentous forms. Nostoc is commonly found, but also Anabaena , Calothrix , Fischerella , and Scytonema spp. have been observed in symbiotic associations (Rowell & Kerby, 1991).

The filamentous heterocystous Nodularia , the main object of this thesis, occurs in saline and brackish waters and in soil all over the world (Table 1). Aphanizomenon , which was also studied in this thesis, lives in both fresh (Sivonen et al. , 1990; Paerl, 1996) and brackish waters (Hällfors, 1979; Edler et al. , 1984).

1.1.2 Cyanobacterial blooms in the Baltic Sea

In the Baltic Sea, late-summer blooms of nitrogen fixing filamentous cyanobacteria are an annual phenomenon (Kononen, 1992). The geographical extent and the magnitude of these population explosions of cyanobacteria has been suspected to increase (Kahru et al. , 1994) due to the increased nutrient concentrations (Wulff et al. , 1990) and changed nutrient ratios in the Baltic Sea (Smayda, 1990).

In most open sea areas of the Baltic Sea, the blooms in July-August are dominated by Nodularia and Aphanizomenon . In addition, Anabaena is present in the blooms (Kononen & Niemi, 1984; Sivonen et al. , 1989a,b; Kononen et al. , 1993b). Several other cyanobacterial genera, e.g. Aphanothece , Chroococcus , Snowella, Merismopedia , Microcystis , Lyngbya , and Planktothrix have been recorded especially from coastal areas of the Baltic Sea (Hällfors, 1979; Edler et al. , 1984). In addition, several studies (Jochem, 1988; Kuosa, 1991; Heinänen et al. , 1995) have described the importance of cyanobacterial picoplankton, identified as Synechococcus , in the phytoplankton community of the Baltic Sea.

In the Baltic Sea, Nodularia has been confirmed as the only toxin-producing cyanobacterial genus so far (Sivonen et al. 1989a,b; Kononen et al. , 1993b), and animal poisonings have been associated with blooms of this genus (Nehring, 1993).

1.1.3 Toxicosis and exposure routes of cyanobacterial hepatotoxins

Hepatotoxins have most frequently been associated with incidents of animal poisonings. The first scientific report on a cyanobacterial poisoning is that by Francis (1878). The author reported that stock deaths in Australia had occurred as a result of drinking from a lake infested by a bloom of Nodularia spumigena . Since the finding of Francis, there have been numerous reports of animal poisonings that have been associated with Nodularia blooms (Table 2) and with other cyanobacterial genera (Sivonen, 1990a; Ressom et al. , 1994; Kuiper-Goodman et al. , 1999). There have also been several reports of human illness - even deaths - attributed to toxic cyanobacteria (Ressom et al. , 1994; Hunter, 1998; Codd et al. , 1999b; Kuiper-Goodman et al. , 1999). The poisoning episodes and monitoring surveys of toxic blooms (Sivonen, 1998; Codd et al. , 1999b; Sivonen & Jones, 1999) have revealed that cyanobacterial toxins are common and pose serious health hazards for animals and human beings throughout the world.

Cyanobacterial peptide toxins are linked to incidences of different human illnesses, including skin and eye irritation, allergy-like symptoms, gastro-enteritis, and hepatoenteritis caused by acute exposure to toxins (Carmichael & Falconer, 1993; Ressom et al. , 1994; Codd et al. , 1999b; Kuiper-Goodman et al. , 1999). Chronic exposure to microcystins through drinking water may increase incidences of human liver cancer (Codd, 1998; Codd et al., 1999b; Kuiper-Goodman et al. , 1999). A severe hepatoenteritis of the population in Palm Island, Queensland, Australia was also associated with drinking water contaminated with Cylindrospermopsis raciborskii (Byth, 1980; Hawkins et al. , 1985). In Brazil, a major human poisoning episode was attributed to water used for dialysis containing microcystins in 1996 (Jochimsen et al. , 1998; Pouria et al. , 1998). This poisoning caused the death of at least 55 people due to severe liver damage.

These episodes showed that hemodialysis water is an important exposure route for cyanobacterial toxins together with the consumption of drinking water. Furthermore, the recreational use of waters has caused illness in individuals who have been in skin contact with blooms (Carmichael & Falconer, 1993; Hunter, 1998; Codd, 1998; Codd et al. , 1999b). For example, in Australia, people have had skin, eye and respiratory symptoms after contact with a bloom of Nodularia (Soong et al. , 1992). Consumption of contaminated cyanobacterial cell dietary supplements, plant products, shellfish and fish may lead to minor exposure to cyanobacterial toxins (Codd et al. , 1999b). For example, cyanobacterial hepatotoxins have been found from salad lettuce (Codd et al. , 1999a) and from mussels (Chen et al. , 1993; Falconer et al. , 1992). In addition, showering and bathing are possible routes of exposure (Codd, 1998; Codd et al. , 1999b).

1.1.4 Ecological effects of cyanobacterial toxins

The ecological role of cyanobacterial toxins is still not understood despite the numerous studies on physiological and ecological bases for toxin production. Toxin production is most likely related to the physical, chemical, and biotic environment, in which cyanobacteria compete with other organisms to maximise their growth and reproduction (Ressom et al. , 1994; Paerl & Millie, 1996). For example, cyanobacteria are known to have toxic effects on phytoplankton and aquatic macrophytes (reviewed by Ressom et al. , 1994) and Nodularia blooms have been suspected to reduce fish abundance (Potter et al. , 1983; Lenanton et al. , 1985). In addition, field observations have revealed that zooplankton (Watanabe et al. , 1992; Kotak et al. , 1996), macroinvertebrates (Zurawell et al. , 1999), and mussels (Lindholm et al ., 1989 ; Falconer et al. , 1992; Chen et al. , 1993) accumulate microcystins and nodularins. These observations indicated that cyanobacterial toxins can have effects on higher trophic levels, thus they may be transfered in aquatic food chains.

Toxin production can also be a defence mechanism against grazers (see Ressom et al. , 1994; Hanazato, 1996). Toxic Microcystis and Nodularia (Reinikainen et al. , 1994; Hietala et al. , 1995; Walls et al. , 1997; Koski et al. , 1999) and toxins produced by these cyanobacteria (DeMott et al. , 1991) have been demonstrated to have harmful effects on zooplankton grazers in the laboratory. In field studies of the Baltic Sea, cyanobacterial blooms were not consumed by copepods and exposure of copepods to a bloom depressed their egg production (Sellner et al. , 1994, 1996). In some of these studies, also non-toxic cyanobacteria had negative effects on zooplankton, thus, there might be other factors than toxicity which affect food quality. For example, cell size and morphology affect ingestibility and assimilability of food (Koski, 1999).

Cyanobacteria are known to produce a wide range of compounds with dissimilar bioactivities (see section 1.3). For example, in Microcystis , different compounds have been suggested to be toxic to Daphnia and inhibit its filtering (Jungmann et al. , 1991). Recently, two variants of Microcystis PCC7806 strains, a wild-type strain and a mutant not able to produce toxin, were used to study the role of microcystins in the defence of Microcystis strains against Daphnia galeata (Rohrlack et al. , 1999). The wild-type strain was toxic to D. galeata , whereas the mutant was not lethal. Yet, both types of PCC7806 were able to reduce the Daphnia ingestion rate. The study gave strong evidence that microcystins were responsible for the poisoning of Daphnia , but they were not responsible for inhibition of the ingestion process.

1.2 Factors leading to cyanobacterial success

Cyanobacterial blooms have been thought to be a consequence of increased nutrient concentrations of fresh and marine waters where they occur seasonally during warm water temperatures. Although nutrients and temperature are regarded as the most important factors for the success of cyanobacteria (Paerl, 1988, 1996) several different factors have been presented to explain cyanobacterial success in aquatic environments (Hyenstrand et al. , 1998).

According to the review of Hyenstrand (1998), cyanobacteria are favoured by a low ratio of total N to total phosphorus (P) and by a high water temperature. Non-nitrogen fixing cyanobacteria require ammonium (NH ₄ ⁺ ) whereas a scarcity of N benefits the development of population of nitrogen fixing cyanobacteria, which differ from non-nitrogen fixing cyanobacteria for high requirements of trace elements (e.g. iron). Furthermore, cyanobacteria gain competitive advantage by storing P and by regulating their buoyancy. With the help of these properties, they can survive during periods of P-deficiency and regulate their vertical position in the water column. Moreover, cyanobacteria can replace eukaryotic phytoplankton in high pH or low CO ₂ situations, and under low light conditions. Cyanobacteria can avoid grazing e.g. by producing toxins.

1.2.1 Factors favouring bloom formation and growth of Nodularia and Aphanizomenon

In Australia and the Baltic Sea, low N and high P concentrations have been related to Nodularia blooms (Table 3). In the Baltic Sea, a low N:P-ratio has also been observed to favour the dominance of Aphanizomenon . In addition, P and in some cases N stimulated the growth of heterocystous cyanobacteria in most of the enrichment experiments in situ (Table 3).

Blooms of Nodularia are common in the N-deficient parts of the Baltic Sea, and they only occur exceptionally in the Kattegat which is connected to the Baltic Proper (the Baltic Sea). The absence of Nodularia in Kattegat, which has equally low N:P-ratios, is probably related to its high salinity (Granéli et al. , 1990). High salinities have been observed to depress the growth of Nodularia in Peel-Harvey Estuary, Orielton Lagoon and Great Salt Lake (Table 3). In addition, laboratory studies with cultures have shown that salinities higher than 30 ‰ repressed the growth (Horstmann, 1975; Nordin & Stein, 1980; Warr et al. , 1984; Blackburn et al. , 1996) and promoted the sporulation of Nodularia (Nordin & Stein, 1980; Jones et al. , 1994b).

In addition to the chemical controlling factors, cyanobacterial blooms are also controlled by physical factors such as temperature, light, and water stability (Paerl, 1996; Hyenstrand, 1998). High water temperatures have been reported to support Nodularia blooms in saline and in brackish waters (Table 3). Cold weather depresses initiation of cyanobacterial blooms, and cloudy and windy weather prevents the cyanobacteria from accumulating at the water surface (Table 3). The ecology and the physiology of heterocystous cyanobacterial blooms in the Baltic Sea have thoroughly been reviewed by Kononen (1992), Kononen and Leppänen (1997), and Sellner (1997).

In addition to enrichment tests done with natural communities in situ and in the laboratory (Table 3), unialgal cultures have been used to study the effects of several factors on cyanobacterial growth in the laboratory. The study of Nordin and Stein (1980) demonstrated that maximum growth of Nodularia strains occurred at salinities of 5-10‰ (test range 1-60‰), temperatures of 25-30°C (test range 15-35°C), pHs of 10.0-10.5 (test range 6.0-11.0), and light intensities of 6000 lx (test range 240-6000 lx). No preference was shown for dominant anions (Na ⁺ , Mg ₂ ⁺ ) or cations (Cl ^- , CO ₃ ^2- , SO ₄ ^2- ). All strains grew well with various levels of nitrate (0.25-1.0 g NO ₃ ^- -N l ^-1 ) but poor growth resulted when NH ₄ ⁺ or urea was the N source (Nordin & Stein, 1980). Later, Huber (1986b) found no stimulation of growth of Nodularia by NH ₄ ⁺ or NO ₃ ^- at concentrations from 0 to 20 mg N l ^-1 . Horstman’s study (1975) showed that Nodularia strains grew best at salinities of 5-15‰ (test range 0-30‰). Warr et al. (1984) studied the effect of different salinities and N sources on the growth of Nodularia harveyana . They found that growth was highest at 0-35‰ (test range 0-70‰). Moreover, the growth on NH ₄ ⁺ (20 mg N l ^-1 ) was higher than on NO ₃ ^- (20 mg N l ^-1 ) and on gaseous N. The study of Blackburn et al. (1996) showed that growth of Nodularia strains at 0 and 35‰ was lower than at 12 and 24‰. According to Melin and Lindahl (1973), the growth of the Baltic Sea Aphanizomenon culture was stimulated by the addition of PO ₄ ^3- , chelated trace elements or a combination of these. The highest biomass accumulations of N. spumigena and Aphanizomenon sp. occurred at the highest studied light intensity (test range 1-100 µmol m ^-2 s ^-1 ) (Holswilder, 1999). For both genera, the growth at the two lowest irradiances (1 and 6 µmol m ^-2 s ^-1 ) differed significantly from the two highest irradiances (60 and 100 µmol m ^-2 s ^-1 ). Furthermore, the biomass of N. spumigena cultures grown in moderately P-limited conditions (N:P = 32:1, molar ratio) was significantly higher than those of extremely P-limited (N:P = 64:1) and non-P-limited (N:P = 16:1) (Holswilder, 1999).

Gopal et al. (1975) found that an alkaline pH (test range 4-9) was most suitable for growth and that nutrients stimulated growth of N. spumigena strain. Growth increased when phosphate (PO ₄ ^3- ) and NO ₃ ^- were added. This soil strain was found to tolerate high concentrations of copper (1.5 µg l ^-1 ) and zinc (2.0 µg l ^-1 ). Dixit et al. (1985) studied the growth of Nodularia with different N sources. Ammonium chloride (0.5 mM) was highly toxic to this soil strain, while sodium nitrate (2 mM) and nitrite (2 mM) supported optimal growth. Furthermore, his studies indicated maximum growth at 20-25°C, pH 7.5-9.0 and light intensity 1000-5000 lx. Camm and Stein (1974) reported that urea was toxic at high concentrations of 8.8 mM and 4.4 mM to soil strains of N. spumigena whereas concentrations lower than those inhibited growth on N-free medium and on NO ₃ ^- -medium.

1.2.2. Factors controlling nitrogen fixation in Nodularia and Aphanizomenon

Many cyanobacteria have been experimentally shown to perform nitrogen fixation under anaerobic conditions, but fewer are able to grow at the expense of atmospheric N ₂ under aerobic conditions (see Flores & Herrero, 1994). Certain cyanobacteria, which are able to perform aerobic nitrogen fixation, have developed heterocysts as a protective structure for nitrogen fixation for the oxygen-sensitive enzyme, nitrogenase. Heterocysts are present within order Nostocales in three families: Nostocaceae , Scytonemataceae , and Rivulariaceae (Castenholz, 1989b).

The progress of cyanobacterial nitrogen fixation research until 1988 has been reviewed by Gallon and Chaplin (1988). Most estimates of nitrogen fixation are derived using the acetylene reduction method. This method is based on the property of nitrogenase to reduce compounds with a triplet bond such as dinitrogen (N ₂ ) and acetylene. The end product of the acetylene reduction, ethylene, can be measured using gas chromatography. The acetylene reduction method has been widely adopted as a method for quantifying nitrogen fixation, since it was first introduced by Stewart and co-authors (Stewart et al. , 1967). Most of the estimates of nitrogen fixation in field populations in the Baltic Sea are based on this method (see references in Table 3), but the ¹⁵ N method (Moisander et al. , 1996) and laser photoacoustic detection of ethylene (Zuckermann et al. , 1997) have been also used. The nitrogen fixation system in cyanobacteria has been intensively studied by molecular techniques. These techniques have made it possible to study the environmental regulation of cyanobacterial nitrogen fixation at a genetic level (see Flores & Herrero, 1994).

Nitrogen fixing cyanobacteria have a selective advantage in N-limited environments such as the open Baltic Sea. In the Baltic Sea, cyanobacterial nitrogen fixation has been estimated to be about 9% of the total N input (Larsson et al. , 1985). The estimated N inputs through nitrogen fixation vary in different areas of the Baltic Sea (Larsson et al. , 1985; Howart et al. , 1988a; Sellner, 1997). Estimations for nitrogen fixation rates and the importance for lakes, estuaries, and oceans have been reviewed by Howarth et al. (1988a). Nitrogen fixation rates are controlled by many chemical and physical factors (Bothe, 1982; Howarth et al. , 1988b). Light, temperature, nutrients and salinity are known to control nitrogen fixation in Nodularia and Aphanizomenon blooms (Table 3).

In laboratory experiments done with Nodularia cultures, nitrogen fixation was highest at salinities of 0- 35‰ and salinities higher than 35‰ decreased the nitrogen fixation (Warr et al ., 1984). In another study, highest nitrogen fixation was detected at salinity of 5-10‰ and lower salinities than 5‰ decreased the nitrogen fixation (Huber, 1986b). Phosphorus addition stimulated nitrogen fixation in P-starved Nodularia cultures (Huber, 1986b), while NH ₄ ⁺ addition inhibited nitrogen fixation (Sanz-Alférez & del Campo, 1994). A very high NH ₄ ⁺ concentration (42 mg l ^-1 ) inhibited nitrogen fixation by both Aphanizomenon and Nodularia (Moisander et al. , 1996). According to Huber (1986b), NH ₄ ⁺ completely inhibited nitrogenase activity whereas NO ₃ ^- did not. Light has also been demonstrated to control nitrogen fixation of Nodularia cultures (Huber, 1986b; Sanz-Alférez & del Campo, 1994; Zuckermann et al. , 1997). In outdoor cultures of N. harveyana nitrogenase activity was influenced by the growth rate and not by high light intensity (Pushparaj et al. , 1994).

1.2.3. Factors influencing cyanobacterial hepatotoxin production

There is no information available about environmental conditions that influence hepatotoxin production in Nodularia blooms in the field whereas factors affecting toxin concentrations in Microcystis -blooms in eutrophic lakes have been examined, and the results are contradictionary. In the Hartbeespoort Dam (South Africa), microcystin-YR, -LR, -YA and -LA concentrations correlated positively with solar radiation and water temperature, and microcystin-YR, and –YA correlated negatively with PO ₄ ^3- concentration (Wicks & Thiel, 1990). In the Coal, Driedmeat, and Little Beaver lakes (Canada), no correlation between microcystin-LR and water temperature was discovered, whereas the toxin concentration correlated positively with total and dissolved P and negatively with NO ₃ ^- (Kotak et al. , 1995). Earlier, in Coal Lake microcystin-LR was positively correlated with water temperature and P concentration (Kotak et al. , 1993). In Babtiste, Narrow, Skeleton, Steele, the Coal, Little Beaver and Driedmeat lakes (Canada), microcystin-LR correlated positively with total P and total N (Zurawell et al. , 1999). In Lake Grand-Lieu (France), microcystin-LR and one of microcystin–RR variants correlated negatively with solar radiation, and both microcystin-RR variants correlated positively with dissolved P and NO ₃ ^- (Vezie et al. , 1998). In Lake Tuusulanjärvi (Finland), microcystin-LR correlated positively with total N and total P and negatively with dissolved NO ₃ ^- (Lahti et al. , 1997). A corresponding analysis of field data on environmental factors during 78 water blooms from 72 Finnish lakes revealed that hepatotoxic Microcystis blooms correlated with high PO ₄ ^3- concentrations (Rapala & Sivonen, 1998).

Growth and nodularin levels in response to salinity changes have been studied in laboratory experiments (Blackburn et al. , 1996). Nodularin concentrations on both gravimetric and cellular basis decreased at the highest studied salinity (35 ‰). In addition, nodularin production has been demonstrated to be controlled by light and by P (Holswilder, 1999). Nodularin production was reduced under light limitation and increased under P limitation. In addition, N addition was shown to promote growth and to increase the nodularin content of a Nodularia -culture, which was grown on N-containing medium (Carmichael et al. , 1988).

Variation of cellular toxin levels under different growth conditions have been studied in the laboratory mainly with batch-cultures of hepatotoxic Microcystis , Oscillatoria , Anabaena , and neurotoxic Anabaena , Aphanizomenon , and Planktothrix (reviewed by Rapala, 1998 and by Sivonen & Jones, 1999). In these laboratory experiments, temperature, light, and nutrients were most frequently examined. In addition, the effects of pH, carbon dioxide, salinity, and micronutrients were examined. The majority of studies indicated that cellular toxin levels were highest under conditions most favourable for growth. Differences in responses of hepatotoxic and neurotoxic strains have been observed, e.g. P had a pronounced effect on hepatotoxin levels, but not on neurotoxin levels. Nitrogen had no effect on toxin production by nitrogen fixing species, while non-heterocystous species, such as Microcystis and Oscillatoria , produced more toxins under N-rich media. All these laboratory studies have provided evidence of environmental regulation of gravimetric toxin concentration. Orr and Jones (1998) suggested that toxin concentration in Microcystis is controlled by the effects of environmental factors on the rate of cell division, not through any direct effect on the metabolic pathways on toxin production. This suggestion was based on laboratory experiments with Microcystis aeruginosa cultures and on re-evaluation of data made on microcystin-producing Anabaena and Oscillatoria cultures presented by others (Sivonen, 1990b; Rapala et al. , 1997).

Cyanobacterial hepatotoxins are so-called intracellular toxins, and they are released from cells when they are damaged or lysing (see Sivonen & Jones, 1999). Chemical, physical, and biological degradation of cyanobacterial toxins have been demonstrated to occur (Hrudey et al. , 1999; Sivonen & Jones, 1999). Twist and Codd (1997) studied degradation of nodularin under light and dark conditions and found photochemical breakdown of nodularin. In addition, nodularin was degraded when components of Nodularia cells were present. These components most likely contained nodularin degrading and metabolising bacteria. In water of Lake Alexandrina, where N. spumigena blooms have occurred for months, nodularin was degraded by bacterial communities within 48-50 h (half-life 24 h) (Heresztyn & Nicholson, 1997).

Furthermore, three bacterial strains, which have been isolated from sediments of the Lake Tuusulanjärvi (Finland), were able to degrade nodularin (Lahti et al. , 1998). On the contrary, the bacterial strain isolated from water sample treated with Microcystis aeruginosa extract, was not able to degrade nodularin, although it has degradative activity against microcystins (Jones et al ., 1994a).

1.3. Bioactive peptides from cyanobacteria

Cyanobacteria contain a large number of secondary metabolites, compounds not essential for growth and reproduction. These include peptides, macrolides, and glycosides (Patterson et al. , 1994; Namikoshi & Rinehart, 1996). These compounds have been reported to possess a number of bioactivities: antiviral (Pattersson et al. , 1993, 1994), antifungal (Patterson et al. , 1994), cytotoxic (Patterson et al., 1991), protein phosphatase inhibitory (Honkanen et al. , 1995) and antineoplastic activities (Moore et al. , 1996). For example, Nodularia harveyana has antifungal, antibacterial, and alleopathic activities (Pushparajat et al. , 1999). Many of these secondary metabolites are toxic to animals and humans (Carmichael & Falconer, 1993; Falconer, 1996; Hunter, 1995, 1998; Codd et al. , 1999b).

1.3.1. Structure of nodularin

Cyanobacteria produce two main types of toxins: neurototoxins and hepatotoxins (Carmichael, 1992, 1994; Codd, 1998; Sivonen, 1998). Hepatotoxins are produced by several genera: Anabaena , Aphanizomenon , Cylindrospermopsis , Microcystis, Nodularia , Nostoc , and Oscillatoria (Carmichael, 1994; Sivonen, 1998; Sivonen & Jones, 1999). Hepatotoxins are divided into three groups: alkaloids (cylindrospermopsin), heptapeptides (microcystins) and pentapeptides (nodularins). Cylindrospermopsin is produced by Cylindrospermopsis raciborskii (Hawkins et al. , 1985, 1997), Aphanizomenon ovalisporun (Banker et al. , 1997), and by Umezakia natans (Harada et al. , 1994). Structurally related cyclic peptides, microcystins and nodularins, differ from each other in their number and type of certain amino acids (Carmichael, 1992; Sivonen, 1998). Microcystins, which are named after the genus Microcystis from which they were first identified, are produced by several cyanobacterial genera and consist of seven amino acids. The main differences between the different variants of microcystins are in the two variable L-amino acids, and methylation or non-methylation of certain amino acids (Sivonen & Jones, 1999). Nodularin is produced only by Nodularia and is composed of five amino acids (Rinehart et al. , 1988). Three of those amino acids are the same as in microcystins: D-MeAsp ¹ (D- erythro -b -methylaspartic acid), Adda ³ , and D-Glu ⁴ (D-glutamic acid)(Fig. 1). In addition to these, nodularin consists of L-Arg ² (L-arginine) and Mdhb ⁵ [2-(methylamino)-2-dehydrobutyric acid]. Both toxins contain a unique C ₂₀ amino acid, 3-amino-9-methoxy-2,6,8-trimethyl-10-phenyldeca-4,6-dienoic acid, which is abbreviated as Adda. This amino acid is essential for the hepatotoxicity (Harada et al. , 1990; Rinehart et al. , 1994). The same peptide structure, named nodularin by Rinehart et al. (1988), has been found in Nodularia blooms and strains sampled from Australia (Runnegar et al. , 1988), New Zealand (Carmichael et al. , 1988; Rinehart et al. , 1988) and the Baltic Sea (Eriksson et al. , 1988; Sivonen et al. , 1989a; Sandström et al. , 1990).

Nodularin show considerably less structural variation than microcystins (Rinehart et al. , 1994). To date, five different natural analogues of nodularin have been characterised (Table 4). Furthermore, the compound [L- Val ² ] , named motuporin, has been found from a marine sponge (de Silva et al. , 1992), in which the compound is probably produced by a symbiotic cyanobacterium. This analogue differs from nodularin by substitution of a valine (Val) residue for an Arg residue. In addition, two dihydronodularins have been synthesised chemically (Namikoshi et al. , 1993).

Most nodularins have an intraperitoneal LD ₅₀ of 50-150 µg kg ^-1 in mice (Table 4), which is calculated to be at least 200 times lower than the oral LD ₅₀ (Kiviranta et al. , 1990). In a similar manner as microcystins (Harada et al., 1990), nodularins with an 6 Z -Adda isomer and with modified D-Glu do not display toxicity. Furthermore, the cyclic structure is necessary for the toxicity, since linear peptides are biologically non-toxic (Choi et al. , 1993).

The linear peptide was first isolated when studying the biosynthesis of nodularin. This non-toxic peptide was thought to be a precursor of nodularin (Choi et al. , 1993; Rinehart et al. , 1994). Later, linear and cyclic peptides, spumigins and nodulapeptins, have been isolated from the toxic Nodularia strain AV1, but these compounds were not found from the non-toxic Nodularia HKVV strain (Fujii et al. , 1997a, b). In addition, glycosidic compounds have been isolated from Nodularia strains (Soriente et al. , 1992; Fujii et al. , 1997b). The function of these compounds is not known.

The immuno-gold labelling of hepatotoxins has revealed that the toxins are primarily localised in the thylakoid area and nucleoid in Microcystis and Nodularia cells, with smaller amounts in the cell wall and sheath. In the same study, nodularin was found both in vegetative cells and heterocysts (Shi et al. , 1995).

1.3.2. Mechanism of hepatotoxicity in mammals

After the hepatotoxins are released from the cyanobacterial cells in the digestive system and taken up in the ileum, they pass through the hepatic portal vein into the primary target organ, the liver (Carmichael & Falconer, 1993; Nishiwaki et al. , 1994). The liver specificity of the hepatotoxins is dependent on the active uptake of these compounds by the bile acid transporter mechanism (Carmichael, 1992; Carmichael & Falconer, 1993). In hepatocytes, they cause hyperphosphorylation of various proteins such as cytokeratin peptides 8 and 18 (Ohta et al. , 1992), and this leads to disturbances in the cell’s cytoskeleton and in morphological changes (e.g. Eriksson et al. , 1990, 1992a,b; Ohta et al. , 1992, 1994; Toivola et al. , 1997). Changes in cell shape and losses of the cell’s contacts with other hepatocytes and with sinusoidal capillaries lead to a lethal intrahepatic haemorrhage or hepatic insufficiency within a few hours to a few days (Carmichael, 1992, 1994; Carmichael & Falconer, 1993).

In mammals, signs of poisoning due to hepatotoxins include weakness, reluctance to move, pallor of the extremities and mucous membranes, anorexia, vomiting, cold extremities and hypovolaemic shock (Carmichael, 1992; Carmichael & Falconer, 1993). Before the animals die of intrahepatic haemorrhage and hypovolaemic shock, they frequently suffer muscle tremors and fall into a coma (Carmichael, 1992). The conclusion that death has resulted from intrahepatic haemorrhage can be confirmed by an autopsy that reveals an enlarged liver. In addition, a blood chemistry, which reveals low hepatic haemoglobin and iron concentrations, expose blood loss responsible for shock (Carmichael, 1992). Besides, necrosis and lysis of hepatocytes, accompanied by marked haemorrhages, are discovered in histological examinations (e.g. Runnegar et al. , 1988).

In all eukaryotic cells, intracellular signal transduction is principally linked to extracellular signals via reversible protein phosphorylation by protein kinases and phosphatases. Eukaryotic protein phosphatases (PPs) are structurally and functionally dissimilar enzymes, which belong to the PPP, PPM, and PTP gene families. PPs in the PTP family dephosphorylate phosphotyrosine whereas phosphoserine and phosphothreonine are dephosphorylated by PPMs and PPPs. The PPP family includes PP1, PP2A, and PP2B, which are the most abundant eukaryotic protein Ser/Thr phosphatases (Barford, 1996). PPs play a crucial role in a variety of cellular processes such as cell proliferation and differentiation (Shenolikar, 1994; Wera & Hemmings, 1995).

PPs have been shown to be inhibited by a variety of natural toxins. These include okadaic acid (Holmes & Boland, 1993; Honkanen et al. , 1994), a fatty acid polyether (Yasumoto et al. , 1985). This seafood toxin, which cause diarrhoeic shellfish poisoning (DSP), is produced by marine dinoflagellates and accumulates in filter-feeding organisms (Aune & Yndestad 1993; Steidinger, 1993; Scoging 1998). Okadaic acid was the first toxin reported to inhibit PP1 and PP2A (Takai et al. , 1987; Bialojan & Takai, 1988; Adamson et al. , 1989; Haystead et al. , 1989, and others). Since then, microcystins and nodularin have also been found to inhibit PP1 and PP2A (MacKintosh et. al. , 1990; Matsushima et al. , 1990; Yoshizawa et al. , 1990; Siegl et al. , 1990; Honkanen et al. , 1990, 1991, 1994; Ohta et al. , 1994; Suganuma et al. , 1992; Runnegar et al. , 1993).

The toxins inhibit distinct PPs differentially. For instance, the inhibitory effects of nodularin and microcystin-LR and –LA are higher for PP2A than for PP1 (Honkanen et al. , 1994). The different sensitivity of PP1 and PP2A to the toxins can be due to the presence of a cysteine residue at position 269 in PP2A and a phenylalanine residue at position 276 in PP1 (Lee et al. , 1999) .

Through inhibition of PPs activities these structurally diverse toxins, which are produced by different organisms, prevent dephosphorylation of phosphoserine and phosphothreonine and therefore increase protein phosphorylation (see Quinn et al. , 1993). Protein phosphorylation plays a central role in many adaptive responses to environmental signals also in cyanobacteria (Tandeau de Marsac & Houmard, 1993; Mann 1994). However, cyanobacterial PPs are resistant to their own toxins, such as microcystin-LR (Shi et al. , 1999).

It is known that the different toxins compete for binding to PP1, since prior binding of microcystins (Matshushima et al. , 1990; Yoshizawa et al. , 1990), nodularin (Yoshizawa et al. , 1990), and calyculin to PP1 (Suganuma et al. , 1990) prevent binding of okadaic acid. Furthermore, the prior binding of okadaic acid and inhibitor-1 and inhibitor-2 prevented PP1 from interacting with microcystin (MacKintosh et al. , 1990).

1.3.3. Tumour promotion caused by protein phosphatase inhibitors

Masami Suganuma was the first to report that okadaic acid is a tumour promoter (Adamson et al. , 1989). Later, structurally various compounds including microcystins and nodularin that inhibit PP1 and PP2A were found and named as "Tumor Promoters of the Okadaic Acid Activity Class" (Fujiki & Suganuma, 1993). Further studies of these compounds revealed that inhibition of PP1 and PP2A is a general tumour promotion pathway in various organs (Fujiki & Suganuma, 1999). The next step in tumour promotion, after inhibition of PPs, is the expression of the TNTµ gene and early-response genes (Sueoka et al. , 1997; Fujiki & Suganuma, 1999). These expressions are assumed to be associated with mRNA stabilisation.

In rat liver, repeated intraperitoneal injections of microcystin-LR and nodularin induced tumours (Nishiwaki-Matsushima et al. , 1992a; Ohta et al. , 1994). Skin tumour growth was promoted by oral consumption of Microcystis in drinking water (Falconer, 1991, 1996). Furthermore, several epidemiological studies in China have indicated that people taking their drinking water from ponds and ditches contaminated with Microcystis blooms have a much higher incidence of primary liver cancer than those using river or well water (Fujiki et al. , 1996; Hunter, 1998).

Nodularin and microcystin-LR have the same specific activity in inhibition of PP1 and PP2A (Yoshizawa et al. , 1990). Nodularin promotes liver tumour growth with higher activity than microcystin-LR (Ohta et al. , 1994; Fujiki et al. , 1996). In addition, nodularin is a liver carcinogen with both initiating and tumour-promoting activities, whereas microcystin-LR is a liver tumour promoter without an initiating activity (Ohta et al. , 1994; Fujiki et al. , 1996).

1.3.4. Structure-activity relationship of peptide toxins

The microcystins and nodularin inhibit PP1 and PP2A and show hepatotoxicity with similar potency. The cyclic stucture of the microcystins and nodularins share many common features. Furthermore, there are variants of microcystins and nodularin that also contain common residues in the variable regions. For instance, microcystin-LR and nodularin have Arg in common. The study of Nishiwaki-Matsushima and co-authors (1992b) with naturally occurring geometrical isomers of microcystin-LR and -LA has shown that the Arg residue did not significantly interact with the enzymes and can be substituted by other amino acids without loss of hepatotoxicity. This was further verified by the isolation of a nodularin variant, motuporin (de Silva et al , 1992). Motuporin is similar in structure to nodularin, but has Arg instead of Val. Despite this structural difference, both of these compounds inhibit PPs with similar potency. Furthermore, there are many naturally occurring toxic microcystins in which a number of residues replace the leucine (Leu) and/or the Arg in the peptide ring (Carmichael, 1992; Rinehart et al. , 1994), indicating that these residues are not critical for the inhibition of PPs. On the contrary, it has been reported that Adda, which is also common for microcystins and nodularins, is essential for the activity and that the peptides with the 6Z-Adda isomer are biologically inactive (Nishiwaki-Matsushima et al. , 1991; Rinehart et al. , 1994). Also, D-Glu has an important role in hepatotoxicity, since esterification of its free carboxyl group leads to a total loss of activity (Rinehart et al. , 1994). These findings suggest that only the Adda and Glu residues play important roles in the hepatotoxicity, thus they have functions in interactions with microcystins and nodularins to PP1 and PP2A.

Three-dimensional (3-D) structures of microcystins (Rudolp-Böhner et al. , 1994; Bagu et al. , 1995, 1997; Trogen et al. , 1996, 1998) and motuporin (Bagu et al. , 1995, 1997) have been determined by nuclear magnetic resonance spectroscopy (NMR). Conformational studies have shown that microcystin-LR (Bagu et al. , 1995; Trogen et al. , 1996), microcystin-RR (Trogen et al. , 1998), and motuporin (Bagu et al. , 1995) possess a saddle-shaped peptide ring. Furthermore, the Adda side-chain in both toxins and Arg in microcystin were shown to be flexible, whereas Val in motuporin was less flexible than Arg. The conformational similarities of microcystin and nodularin imply that these peptide toxins inhibit PPs in a similar way.

The NMR results differ from the results obtained using a molecular modelling analysis of the 3-D structures of microcystin-LR and nodularin, which predicted that these toxins have planar peptide rings (Lanaras et al. , 1991; Taylor et al. , 1992). The comparison of Lanaras and co-workers was later optimised by Taylor et al. (1992), who used molecular modelling to calculate the energetically lowest conformations for microcystin-LR. Later, molecular modelling was expanded to include several naturally occurring toxins: cantharidin, calyculin-A, okadaic acid, tautomycin (Quinn et al. , 1993; Gauss et al. , 1997; Gupta et al. , 1997). The conformational studies of natural toxins have been reviewed by Quinn et al. (1996), who has also developed a pharmacophore model for okadaic acid, calyculin-A, and microcystin-LR (Quinn et al. , 1993). This model contains a conserved acidic group, two potential hydrogen-bonding sites, and a non-polar side-chain.

The 3-D structures of the inhibitors alone do not explain how PPs are inhibited, therefore, the bound conformations of inhibitors to protein phosphatases are needed. Crystal structures have been determined for rabbit PP1 complexed with microcystin-LR (Goldberg et al. , 1995) and for human PP1 complexed with tungsten (Egloff et al. , 1995). These structures clarify how PP1 dephosphorylate substrates and how it is regulated.

The study of Goldberg et al. (1995) showed how PP1 is inhibited by microcystin. Microcystin interacts with the PP1 in a Y-shaped groove on the surface of the enzyme at three sites: the metal-binding site, the hydrophobic groove and the edge of the C-terminal groove near the active site (Goldberg et al. , 1995).

Lee and co-workers (Zhang & Lee, 1997; Lee et al ., 1999) have mutated residues in these grooves to assess the importance of these structures in substrate recognition. Strong effects were only observed for mutation of residues involved in phosphate binding and orientation at the active site (Lee et al ., 1999).

In the interaction between microcystin and PP1, the Glu residue of the microcystin binds indirectly to the metals via two of the metal-liganded water molecules. The long hydrophobic Adda group is inserted into the hydrophobic groove. The Leu residue of microcystin interacts with the side-chain of tyrosine 272 (Tyr 272) within the ß12/ß13-loop of the C-terminal groove (Goldberg et al. , 1995). The role of the ß12/ß13-loop in toxin binding has been confirmed by studies with different mutants (Shima et al., 1994; Zhang et al. , 1994, 1996; Lee et al. , 1999) .

Functional differences exist between the microcystins and nodularins with respect to their interaction with PPs. These enzymes are initially bound non-covalently and inhibited by these toxins. In contrast to the microcystins (Goldberg et al. , 1995; MacKintosh et al. , 1995; Runnegar et al. , 1995), motuporin does not form a secondary covalent bond after inhibition of PP1 (Bagu et al. , 1995, 1997). The explanation why nodularins are incapable of forming a covalent linkage with PP1 is the difference in the position of N-methyldehydrobutyrine (Mdhb) residue in motuporin relative to the counterpart N-methyldehydroalanine (Mdha) in microcystin-LR (Bagu et al. , 1995, 1997). The different position of Mdhb at the surface of the PP1-toxin complex compared to microcystins may facilitate a chemical interaction with further macromolecules, which may explain the carcinogenic properties of nodularins (Ohta et al. , 1994; Bagu et al. , 1997). Furthermore, the higher tumour promoting activity and the carcinogenicity of nodularin are thought to be caused by smaller size of nodularin compared to microcystin. For that reason, nodularin is more easily taken into the hepatocytes than microcystin-LR (see Fujiki et al. , 1996).

1.3.5. Biosynthesis of peptide toxins

Microbes synthesise non-ribosomally cyclic and linear peptides by large modular multienzyme complexes, peptide synthetases. The peptide sythetases use the so-called thiotemplate mechanism for the synthesis of peptides (von Döhren et al. , 1997; Konz & Marahiel, 1999).

Peptide synthetase genes have been identified from several Microcystis aeruginosa strains (Meißner et al. , 1996; Dittmann et al. , 1996, 1997). One of these strains (PCC7806) has been transformed to non-toxic form by the mutation of a microcystin synthetase gene, which demonstrates that this gene, called mcyB , encodes a microcystin synthetase (Dittmann et al. , 1997). Recently, mcyB was shown to hybridise, with variable signal intensity, to DNAs from several hepatotoxic strains of Oscillatoria , Microcystis , and Anabaena (Neilan et al. 1999). On the contrary, two peptide synthetase genes encoding anabaenopeptilides (cyclic depsipeptides) in the Anabaena sp. strain 90 (Rouhiainen et al. , 2000) hybridised insignificantly to non-heterocystous microcystin-producing strains. However, this probe gave strong signals with all hepatotoxic Anabaena strains, and it also cross-hybridised to most of the hepatotoxic Nostoc and Nodularia strains (Neilan et al. , 1999). Recently, three open reading frames encoding peptide synthetases involved in microcystin synthesis were identified (Nishizawa et al. , 1999).

The biosynthesis of nodularin has been studied by NMR analysis of nodularin obtained from experiments with several ¹³ C-labelled precursors (Choi et al. , 1993; Rinehart et al. , 1994). Carbons in Adda are derived from acetate, methionine, phenylalanine, and propionate (Moore et al., 1991; Choi et al. , 1993; Rinehart et al. , 1994). Acetate and pyruvate were also incorporated into Arg and Glu. Furthermore, D-MeAsp and Mdhb were labelled with propionate and methionine, respectively. The complete synthesis of the natural nodularin variant, motuporin, has been published (Valentekovich & Schreiber, 1995; Bauer & Armstrong, 1999). In addition, synthetic analogues of nodularin have been obtained (Mehrotra & Gani, 1996). A general route for the preparation of microcystins and nodularins has been provided by Kim et al. (1996).

1.3.6. Analytical methods of cyanobacterial toxins

Falconer (1993) and Harada et al. (1999) have written reviews of biological, biochemical, and chemical methods for assays of cyanobacterial toxins. Cyanobacterial toxins in water and in cells have been detected and identified by a range of methods that differ in selectivity and sensitivity (Harada, 1996; Harada et al. , 1999). Early methods for the assay of cyanobacterial toxins were based on the mouse bioassay. This method was widely used for determination of outright toxicity and toxin concentration. Furthermore, a number of invertebrates have been investigated for use as bioassays for toxins (Harada et al. , 1999). Later, biochemical tests such as the protein phosphatase inhibition assay (An & Carmichael, 1994; Ward et al. , 1998; Harada et al. , 1999) and the enzyme-linked immuno-sorbent assay (ELISA) (Chu et al. , 1989, 1990; Harada, 1996; Harada et al. , 1999) were developed for detection of peptide hepatotoxins, microcystins and nodularins.

Due to functional groups in the molecules cyanobacterial peptide toxins have common physico-chemical properties such as molecular weight and activities. For that reason they can be analysed by the same analytical methods (Harada et al. , 1999). The most commonly-used analytical system for cyanobacterial peptide toxins is high performance liquid chromatography (HPLC) combined with ultraviolet (UV) detection which relies on their retention times and UV spectra for identification (Harada, 1996; Meriluoto, 1997; Harada et al., 1999; Pelander, 2000). HPLC has been the most intensively used method when studying cellular toxin concentrations in cyanobacteria under different growth conditions (Sivonen & Jones, 1999).

1.4. Cyanobacterial taxonomy

Cyanobacteria were traditionally classified on the basis of their morphology. Despite the fact that their morphology is complex when compared to most other microbes, the taxonomy based on morphological characteristics does not necessarily result in a phylogenetically reliable taxonomy (Giovannoni et al. , 1988; Wilmotte, 1994). Nowadays, bacterial taxonomy relies on different kinds of information derived from phenotypic and genotypic data (Vandamme et al. , 1996).

The use of phenotypic and genotypic characteristics in cyanobacterial taxonomy was pioneered by Stanier and collaborators (e.g. Kenyon et al. , 1972; Herdman et al. , 1979a,b; Rippka et al. , 1979). The characteristics used include morphology, pigment and fatty acid composition, photoheterotrophic growth, nitrogenase activity, DNA base composition, and genome size. The taxonomic system of Rippka et al. (1979), which was based on identification of 178 cyanobacterial cultures in the Pasteur Culture Collection, recognised 22 genera placed in five taxonomic sections. Section I contains unicellular cyanobacteria that reproduce either by binary fission ( Gloeobacter , Gloeocapsa , Gloeothece , Synechococcus , Synechocystis ) or by budding ( Chamaesiphon ) whereas the unicellular members in section II divide only by multiple fission, which leads to formation of motile ( Dermocarpa ) or inmotile ( Xenococcus ) baeocytes, or by both binary fission and multiple fission ( Dermocarpella , Myxosarcina , Chroococcidiopsis , Pleurocapsa group). Members in sections from III to V are filamentous. The filamentous cyanobacteria in section III ( Spirulina , Oscillatoria , LPP group A, Pseudanabaena , LPP group B) compose only of vegetative cells while the filamentous cyanobacteria of sections IV and V contain additionally heterocysts and sometimes akinetes. In section IV, filamentous cyanobacteria are distinguishable by hormogonium formation ( Nostoc , Scytonema , Calothrix ) or its absence ( Anabaena , Nodularia , Cylindrospermum ). Divisions of cells of the filamentous, heterocystous cyanobacteria belonging to sections III and IV occur in one plane whereas members in section V ( Chlorogloeopsis , Fischerella ) divide on more than one plane.

The bacteriological taxonomic system created for cyanobacteria by Rippka et al . (1979) has been modified by Castenholz (1989a, b, c), Waterbury (1989), and Waterbury and Rippka (1989) in the volume three of "Bergey’s Manual of Systematic Bacteriology". This system also includes descriptions of cyanobacteria which are observed, but which have not been succesfully maintained in cultures. In addition, ecological features of cyanobacteria have been included. In their separate studies, Anagnostidis and Komárek (1988, 1990) and Komárek and Anagnostidis (1986, 1989) have revised the taxonomy of cyanobacteria using both bacteriological and botanical approaches. The authors made an extensive review of the literature and applied phenotypic and genotypic data concerning cyanobacterial taxonomy.

Molecular phylogenetic analysis have revealed at least ten distinct eubacterial phyla (Woese, 1987). One of these contains oxygenic photosynthetic prokaryotes, cyanobacteria and prochlorophytes, which have different pigment compositions and are genetically related on the basis of 16S rRNA sequences (Woese, 1987).

Cyanobacteria contain chlorophyll- a and phycobiliproteins and prochlorophytes possess chlorophyll- a as well as chlorophyll- b , but lack phycobiliproteins (Castenholz & Rippka, 1989; Lewin, 1989; Matthijs et al. , 1994). The prochlorophytes contain three genera: Prochloron , Prochlorothrix (Lewin, 1989), and Prochlorococcus (Partensky et al. , 1999). 16S rDNA sequence analysis has revealed that prochlorophytes emerged within the cyanobacteria, but on separate branches (Urbach et al. , 1992; Wilmotte, 1994). Prochlorophytes are phylogenetically nearest to Synechococcus (Urbach et al. , 1998).

The evolutionary relationships among cyanobacteria have been published by Giovannoni et al. (1988), Wilmotte (1994), and Turner (1997). The 16S rRNA sequence analysis (Fig. 2) were congruent to the taxonomic sections of II, III and IV defined by Rippka et al. , (1979). On the contrary, sections I and III were scattered in different lineages and sometimes mixed. The heterocystous filamentous strains were all in the same cluster. In this cluster, the Nodularia strain PCC73104 grouped with the Anabaena cylindrica PCC7122 strain. Furthermore, it was closely related to Nostoc (PCC73102 and PCC7120) and to the Cylindrospermum (PCC7417) strains (Wilmotte, 1994).

1.4.1. Molecular methods used in cyanobacterial taxonomy

Most bacterial genomes contain genes from multiple sources, even from genetically distant ones (Doolittle, 1999). Nowadays, it is suspected that the even most trusted chronometers, rRNA genes, can be transferred (Doolittle, 1999). These genes, which are functionally constant and are composed of highly conserved as well as more variable domains, have been utilised for phylogenetic analysis in bacteria (Vandamme et al. , 1996). Outlines of bacterial phylogenetic relationships emerged from the work of Woese (1987), who pioneered the comparison of rDNA sequences.

The first complete cyanobacterial 16S rDNA sequence was released by Tomioka and Sugiura (1983). In 1988, Giovannoni and co-authors published the first study of evolutionary relationships among cyanobacteria using partial 16S rDNA sequences (Giovannoni et al. , 1988). The sequencing of the 16S rRNA gene have resulted in a rRNA sequence database. At the end of 1998, 171 16S rRNA genes were sequenced from cyanobacteria.

These 171 cyanobacterial rDNA sequences (81 complete sequences comprising nearly 1500 nucleotides and 90 partial sequences) serve as the backbone for modern cyanobacterial taxonomy (Wilmotte, 1994). One partial (ARB_40A3CBCC, Giovannoni et al. 1988) and one complete (aj22447, Hayes & Barker, 1997) 16S rDNA sequence exist from Nodularia .

The molecular methods for studying genotypic relationships among cyanobacteria have been reviewed by Wilmotte (1994). In the present work, only the DNA- and rRNA sequence-based studies of free-living cultured cyanobacteria have been reviewed (Table 5).

Over the recent years, techniques based on sequences of rRNA genes have become the most widely used methods for identification, classification, and phylogeny of cyanobacteria. In addition to direct comparison of the DNA sequences, other methods based on amplification of rRNA genes have been used. The amplified ribosomal genes can be digested by restriction enzymes and the resulting patterns analysed, this technique has been termed restriction fragment length polymorphism (RFLP). Before the introduction of the polymerase chain reaction (PCR) method (Lane et al. , 1985), ribotyping based on labelled-rRNA probe have revealed the hybridised fragments generated after restriction enzyme digestion.

During the last few years, a battery of DNA-directed typing methods has been used for PCR-based genomic fingerprinting of cyanobacteria. Genomic fingerprints have been produced by PCR with random (RAPD) or by arbitrary primers (AP). In addition, fingerprints have been generated with primers corresponding to the repetitive extragenic palindromic (REP) and enterobacterial repetitive intergenic consensus (ERIC) sequences, and to the long and short tandemly repeated repetitive (LTRR and STRR) sequences. The function of repetitive and random amplified polymorphic DNA elements is not known. In addition, cyanobacterium-specific primers have been developed to identify cultured and natural samples. This method has been named "Specific PCR identification" or "Diagnostic PCR".

DNA base composition has been determined for several cyanobacterial strains in the Pasteur Culture Collection. Generally, large differences in DNA base composition reflect the fact that strains are not closely related, whereas similar guanine and cytosine (G+C) percentages give no information about genotypic relationships, therefore the taxonomic value of this method at the species level is low (Wilmotte, 1994). The species should be delineated according to DNA-DNA hybridisation studies, which established the base composition homology between different DNA strains. For strains of the same species at least 70% hybridisation is needed (Wayne et al ., 1987). Furthermore, protein-coding genes have been sequenced and probed. These genes can be used for identification and for classification of strains, but not for phylogeny, since they are not universally distributed and essential among bacteria (Woese, 1987), thus different molecular methods gave different taxonomic information (Vandamme et al. , 1996).

1.4.2. Morphological and genetic variation of Nodularia cultures and populations

Komárek and co-authors (1993) have studied the taxonomy of the genus Nodularia . Within this genus, two distinct groups of species, based on the ability to produce gas vesicles, were differentiated. The studies of the ultrastructure of genus Nodularia (Gumpert et al. , 1987; Šmarda et al. , 1988; Albertano et al. , 1996; Šmarda & Šmajs, 1996) have revealed large amounts of densely packed gas vesicles. Their densities in cells separated them into three distinct types (Šmarda & Šmajs, 1996). For example, the Baltic Sea Nodularia had three types of gas vesicles in species of N. baltica , N. spumigena , and N. litorea (Šmarda & Šmajs, 1996).

In addition to the size and the density of gas vesicles, Komárek et al. (1993) have used several other markers in the intrageneric taxonomy of Nodularia : the size of cells of all types, the morphology of akinetes, and the ecological properties of species types. Based on these features, Komárek et al. (1993) suggested distinction of several types of Nodularia species including three benthic species ( N. harveyana , N. sphaerocarpa , N. willei ) and four planktic species ( N. baltica , N. litorea , N. spumigena , N. crassa ). Four of these species (Table 6: N. harveyana , N. baltica , N. litorea , N. spumigena ) have been isolated from the Baltic Sea (Komárek et al. , 1993).

In the study of Hayes and Barker (1997), the genetic diversity of Baltic Sea Nodularia population seemed to be restricted to few genotypes. The diagnostic PCR study (also termed specific PCR identification) of Nodularia population in the Southern Baltic Sea in 1996 showed two PC-IGS (the intergenic spacer region of the phycocyanin operon) genotypes from 156 filaments. Later in 1994, the study of Barker et al. (1999) revealed three groups based on PC-IGS, two groups based on gvp A-IGS (the intergenic spacer region between two copies of gvpA gene), and three groups based on rDNA-ITS (the 16S-23S rRNA internal transcribed region) in thirteen clonal Nodularia strains. The authors could not find any correlation between the genotypic and phenotypic characters examined (trichome width, degree of coiling, and properties of gas vesicles).

The morphological and genetic variation using RFLP and DNA sequencing of the cpcBA -IGS region and RAPD-PCR of Nodularia strains from Australia and from other geographical sites have been studied (Bolch et al. , 1996, 1999). Geographically diverse Nodularia strains had a near cpcBA -IGS (the phycocyanin intergenic spacer region) sequence identity (Bolch et al. , 1999) suggesting that Nodularia is globally distributed as suggested also by Hayes and Barker (1997). However, genetically distinct geographical strains were showed by RAPD-PCR (Bolch et al , 1999) and by RFLP of the cpcBA -IGS (Bolch et al. 1996). Genetic groupings based on the sequence identity were supported by morphological features (size and morphology of vegetative cells, heterocysts and akinetes, and diameter and morphology of trichomes) (Bolch et al., 1999).

It has been suggested that the Baltic Sea Nodularia populations have exchanged genetic material, since the PC-IGS, gvpA -IGS, and rDNA-ITS genotypic groupings of Nodularia strains were not congruent (Barker et al. , 1999). The exchange of genetic material between genetically closely related cyanobacteria has been suggested also by Rudi et al. (1998). The authors proposed that the genetic exchange has led to the observed sequence homogeneity between these organisms and that it may also explain the similarity between the fossil and the recent species found by Schopf and others (1994). However, several studies have revealed that, for example, Synechococcus strains take up the DNA of any source (see Lorenz & Wackernagel, 1994) meaning that exchange may even occur between distant genera.