Cylindrospermopsin: occurrence, methods of detection and toxicology

Authors

  • C. Moreira,

    1. CIMAR/CIIMAR/Laboratory of Ecotoxicology, Genomics and Evolution, Porto University, Porto, Portugal
    2. Departamento de Biologia, Faculdade de Ciências, Universidade do Porto, Porto, Portugal
    Search for more papers by this author
  • J. Azevedo,

    1. CIMAR/CIIMAR/Laboratory of Ecotoxicology, Genomics and Evolution, Porto University, Porto, Portugal
    2. Escola Superior de Tecnologia da Saúde do Porto, Vila Nova de Gaia, Portugal
    Search for more papers by this author
  • A. Antunes,

    1. CIMAR/CIIMAR/Laboratory of Ecotoxicology, Genomics and Evolution, Porto University, Porto, Portugal
    Search for more papers by this author
  • V. Vasconcelos

    Corresponding author
    1. Departamento de Biologia, Faculdade de Ciências, Universidade do Porto, Porto, Portugal
    • CIMAR/CIIMAR/Laboratory of Ecotoxicology, Genomics and Evolution, Porto University, Porto, Portugal
    Search for more papers by this author

Correspondence

Vitor Vasconcelos, Rua dos Bragas, 289, Porto 4050-123, Portugal. E-mail: vmvascon@fc.up.pt

Abstract

Cyanobacteria are aquatic micro-organisms that pose a great threat to aquatic ecosystems by the production of dense blooms, but most importantly by the production of secondary metabolites, namely the cyanotoxins. One of these is cylindrospermopsin (CYN), a hepatotoxic polyketide-derived alkaloid with well-known associated cases of animal mortalities and human morbidity. First described as being associated with liver damage, this toxin is now considered a cytotoxic and a genotoxic toxin, due to its effects in other organs and in DNA. Its occurrence has been reported so far in eight different cyanobacteria species and in several water samples from four of the five continents. With a guideline value of 1 μg l−1, CYN is now considered the second most studied cyanotoxin worldwide. It is important to review the information regarding the findings made until now about this cyanotoxin 30 years since its first report.

Introduction

Cyanobacteria are supposed to be some of the oldest living beings on earth with fossil records, dating from 3 billion years ago (Sze 1993). These micro-organisms are very successful making a major contribution to global photosynthesis and nitrogen fixation (Skulberg 2005). With a worldwide distribution in both aquatic and terrestrial environments, cyanobacteria can be found in a wide range of ecosystems from the cold Antarctica to the hot springs of Japan, New Zealand and Italy (Papke et al. 2003; Jungblut et al. 2005). Although simple in morphology and nature, these prokaryotic organisms may pose a serious risk to aquatic ecosystem and its users. They have the ability to form dense blooms in eutrophic systems decreasing the water quality by the release of off flavours, water discoloration and accumulation of surface scums (Ibelings and Chorus 2007). Equally important is the ability of some bloom-forming species of cyanobacteria to produce harmful secondary metabolites named cyanotoxins. These toxins are secondary metabolites that have a harmful effect on tissues, cells or organisms (Carmichael 1992). They have been reported in both freshwater and marine environments with implications in human and animal poisonings (Dittmann and Wiegand 2006). Therefore, they need to be monitored to minimize their risks to both water quality and public health. In this regard, several countries have developed regulations or guidelines for cyanotoxins and cyanobacteria in drinking waters and, in some cases, in water used for recreation and agriculture (Burch 2006).

Toxins produced by cyanobacteria are classified as hepatotoxins (microcystins and nodularins) neurotoxins (anatoxins and saxitoxins), cytotoxins (cylindrospermopsin) and dermotoxins (Carmichael and Li 2006). Their production can be found in a diverse range of species. One of the most recently studied cyanobacterial toxins is cylindrospermopsin (CYN) (Falconer and Humpage 2005). This toxin has only been a subject of study since 1979 where it was first reported. With the increased number of publications on the detection of this toxin in several countries of nearly all continents, it is important to review the knowledge acquired so far. Therefore, we aim to compile most of the information on this toxin, by describing reports of its occurrence and toxicity levels, review its chemical structure and biosynthetic pathways, enumerate the existent methodologies applied in its detection and finally discuss its impact on human and animal health.

CYN-producing organisms

Cylindrospermopsis raciborskii was the first species reported as a producer of CYN. Other cyanobacteria species have been identified as CYN producers, namely Umezakia natans (Harada et al. 1994), Aphanizomenon ovalisporum (Banker et al. 1997), Anabaena bergii (Schembri et al. 2001), Raphidiopsis curvata (Li et al. 2001a), Aphanizomenon flos-aquae (Preußel et al. 2006), Anabaena lapponica (Spoof et al. 2006) and more recently Lyngbya wollei (Seifert et al. 2007), clearly indicating that its production is not species specific. A summary of each report is presented in Table 1.

Table 1. Summary of cylindrospermopsin-producing organisms
ContinentSpeciesCountryLD50ConcentrationReference
Europe Aphanizomenon ovalisporum Spain  Wörmer et al. (2008)
Anabaena lapponica Finland 242 μg g−1Spoof et al. (2006)
A. flos-aquae Germany 2·3–6·6 mg g−1Preußel et al. (2006)
Asia Cylindrospermopsis raciborskii Thailand250 mg kg−11·02 mg g−1Li et al. (2001b)
Umezakia natans Japan  Harada et al. (1994)
A. ovalisporum Israel465 mg kg−1 Banker et al. (1997)
Raphidiopsis curvata China 0·56 μg g−1Li et al. (2001a)
America A. ovalisporum USA 7·39–9·33 μg mg−1Yilmaz et al. (2008)
Oceania C. raciborskii Australia59–69 mg kg−1 Hawkins et al. (1985)
C. raciborskii Australia52 mg kg−1 Hawkins et al. (1997)
A. ovalisporum Australia0·1–1·3 mg kg−1 Shaw et al. (1999)
Anabaena bergii Australia  Schembri et al. (2001)
Lyngbya wollei Australia 0–33 μg g−1Seifert et al. (2007)

The first description of CYN occurred after an outbreak of hepatoenteritis in Palm Island, northern Queensland, Australia, in 1979 (Bourke et al. 1983). This was the first documented case of this cyanotoxin affecting 148 people, mostly children, which required hospitalization and presented symptoms of gastroenteritis (Byth 1980). Cylindrospermopsis raciborskii was the dominant species in the Solomon dam after this incident. This species was reported as nontoxic until this incident and known to form blooms in tropical environments (Hawkins et al. 1985). Later, Ohtani et al. (1992) characterized for the first time the chemical structure of this cyanotoxin from a C. raciborskii strain isolated from the Palm Island incident and named it after its given genus. Since then, it has been described in other cyanobacteria species all belonging to the Nostocales Order with two exceptions Umezakia natans (Order Stigonematales) as previously described by Harada et al. (1994) and more recently Lyngbya wollei (Order Oscillatoriales) by Seifert et al. (2007).

Chemical characterization of CYN and analogues

CYN is an alkaloid, ([C15H21N5O7S]; 415 43 g mol−1; glassy solid), a sulfate ester of a tricyclic guanidine moiety (rings A, B & C), with a uracil ring (D) (1) and its zwitterionic nature makes it a highly water-soluble molecule (Fig. 1). The gross structure of CYN was solved by mass spectroscopy (MS) and nuclear magnetic resonance (NMR), first by Ohtani et al. (1992). However, it was found later an incorrect feature in the first identification, the orientation of the hydroxyl group whose epimer was later identified by Banker et al. (2000) as 7-epicylindrospermopsin (3), a toxic minor metabolite of A. ovalisporum. Norris et al. in 1999 were able to isolate CYN and the analogue 7-deoxy-cylindrospermopsin (2) by solid-phase extraction (SPE) followed by semi-preparative chromatography. The CYN extinction coefficient ε-value and specific rotation [α]D are 9800 and +17·0°, respectively (Sano et al. 2008).

Figure 1.

The molecular structures of cylindrospermopsin (1) and its analogs 7-deoxy-cylindrospermopsin (2) and 7-epicylindrospermopsin (3).

Molecular characterization of the CYN cluster

Cyanobacteria produce a large number of bioactive compounds most of them with a nonribosomal peptide and/or polyketide structure; although widely widespread, they are not a general feature in this group of photosynthetic prokaryotes (Dittmann et al. 2001). These are collectively referred to as secondary metabolites and are compounds that are not used by the organism for its primary metabolism (cell division or metabolism) and include toxins (Carmichael 1992). The majority of these bioactive compounds isolated from cyanobacteria are extremely large classes of natural products that are polyketides, nonribosomal peptides or a mixture of both (Welker and von Döhren 2006; Barrios-Llerena et al. 2007). Biosynthesis of these compounds is performed by a family of multi-enzymatic complexes called nonribosomal peptide synthetases (NRPS) and polyketide synthases (PKS) organized into repeated functional units known as modules (Cane and Walsh 1999; Dittmann et al. 2001). Therefore, its transcription and translation are independent of the messenger RNA. Each PKS or NRPS module is made up of a set of three domains, two of which are catalytic and one which acts as a carrier, that together are responsible for the central chain-building reactions of polyketide or polypeptide biosynthesis (Cane and Walsh 1999). NRPS have three domains: the adenylation-(A)-domain responsible for amino acid recognition and activation; the peptidyl carrier protein for transport to the respective catalytic centres; and finally, the condensation-(C)-domain for the formation of the peptide bond.

Cyanobacterial NRPS and PKS are organized into gene clusters in the genome often within a single open reading frame (Welker and von Döhren 2006). Previously, Shalev-Alon et al. (2002) reported in A. ovalisporum CYN producer three putative genes involved in CYN biosynthesis designated as AoaA, AoaB and AoaC that encode an amidinotransferase, a hybrid NRPS/PKS and a PKS, respectively. With this study, an amidinotransferase-encoding gene in cyanobacteria was identified for the first time and its location in the genome is up-stream of the PKS on the reverse strand. AoaA is thought to catalyse the synthesis of guanidinoacetate, which is then recruited by AoaB for successive polyketide extension by AoaC and further polyketide synthase modules (Kellmann et al. 2006). AoaA three-dimensional protein was determined, and the amino acid sequence obtained showed a 40% identity/56% similarity to the human arginine/glycine amidinotransferase GATM, suggesting that AoaA may catalyse the transamination of glycine from arginine; however, such evidence needs to be further verified (Kellmann et al. 2006). Recently, Mihali et al. (2008) sequenced and characterized CYN biosynthesis gene cluster and proposed for the first time the complete biosynthetic pathway for this cyanotoxin. Using an adaptor-mediated ‘gene-walking’ technology, from the known partial sequence of the amidinotransferase gene from C. raciborskii AWT205, Mihali et al. (2008) sequenced the entire cluster. The cyr gene cluster spans 43 kb comprised of 15 open reading frames (ORFs) containing genes required for the biosynthesis, regulation and export of the toxin (Fig. 2). At both ends of the gene cluster are a further 3·5 kb that contain putative hyp accessory genes, which include molecular chaperones involved in the maturation of hydrogenases (Mihali et al. 2008).

Figure 2.

Cylindrospermopsin Cyr gene cluster structural organization (43 kb) showing genes encoding for amidinotransferase (CyrA), PKS/NRPS (CyrD, CyrF, CyrB, CyrE and CyrC), uracil ring (CyrG and CyrH), tailoring enzymes (CyrI, CyrJ and CyrN), transport (CyrK), regulation (CyrO) and transposase (CyrL and CyrM).

The first step in CYN biosynthesis is the formation of the carbon skeleton involving the synthesis of guanidinoacetate and is comprised of products transcribed from CyrA to CyrG genes. CyrA (AoaA analogous) contains an amidinotransferase that transfers a guanidino group forming the guanidinoacetate. Following is CyrB (AoaB analogous) consisted of a mixed NRPS/PKS containing eight domains: adenylation, peptidyl carrier protein, β-ketosynthase, acyltransferase, dehydratase, methyltransferase, ketoreductase and acyl carrier protein. The adenylation domain is thought to activate guanidinoacetate, which is then passed through the peptidyl carrier protein to the β-ketosynthase domain. The next step involves CyrC (AoaC analogous), a PKS of four domains (β-ketosynthase, acyltransferase, ketoreductase and acyl carrier protein), that elongates the chain with an acetate molecule. Afterwards is CyrD, a PKS of five domains (β-ketosynthase, acyltransferase, dehydratase, ketoreductase and acyl carrier protein). Its action is on the intermediate produced by CyrC, which in turn produces another intermediate product that will be the substrate for CyrE. This is a PKS containing the same five domains of CyrD, and that is responsible for the addition of one acetate. CyrF is also a PKS with three domains: β-ketosynthase, acyltransferase and acyl carrier protein. CyrF acts on the final product of CyrE increasing the carbon chain with an acetate group. Next occurs the formation of the uracil ring involving CyrG and CyrH where it takes place the transfer of a second guanidino group to the CYN molecule.

For CYN biosynthesis, it is necessary for the action of tailoring enzymes to complete its biosynthesis. Three enzymes are involved in the tailoring reactions, namely CyrI, CyrJ and CyrN. CyrN encodes adenylylsulfate kinases that are enzymes that catalyse the formation of PAPS, the sulfate donor for sulfotransferases. An interesting remark is that sulfotransferase genes were present only in CYN-producing strains; therefore, CyrJ may be a good marker for CYN toxicity as already tested in the work of Ballot et al. (2011). The final tailoring reaction is carried out by CyrI, which catalyses the hydroxylation of C-7, a residue that along with the uracil ring seems to confer much of CYN toxicity (Mihali et al. 2008).

In CYN transportation, the gene cluster contains an ORF designated CyrK. Finally, CYN gene cluster includes an ORF designated CyrO at its 3′ end with a likely role in transcriptional regulation and DNA binding.

Cylindrospermopsin Detection and Quantification

Chemical assays

This chapter gives an overview of the different chemical methods that have been used to analyse CYN in cyanobacterial cultured species and in environmental samples of fresh and brackish water, including drinking water and recreational water bodies, and also in contaminated animal tissues.

The knowledge about the chemistry of CYN and on the problems associated with complex matrices of samples containing this toxin is crucial to select the most reliable identification, extraction, separation and analyses techniques to monitor this cyanotoxin in the environment. Concerning the stability of the molecule, Chiswell et al. (1999) found that CYN is a very stable compound showing no degradation at boiling temperature (100°C for 5 min); when exposed to pH changes between 4 and 10 for a period of 8 weeks, it only degrades 25%, and in the solid state or in pure aqueous solutions, CYN does not show any degradation. However, while in culture exposed to sunlight, it decomposes rapidly (half-life 1·5 h). CYN degradation via chlorination occurs immediately at pH 6, but the molecule is more stable to this water treatment process at lower values of pH (Senogles et al. 2000).

Although nowadays there are some commercial companies that sell analytical standards of CYN and/or reference material (National Research Council of Canada), much of the work with this toxin rely on its purification using cultures or bloom material. For the purification of CYN from laboratory cyanobacterial cultures, researchers use mainly the species C. raciborskii (Norris et al. 2001a,b; Kubo et al. 2005; Lankoff et al. 2007) and A. ovalisporum (Forti) (Banker et al. 1997; Vasas et al. 2002) but also from U. natans (Watanabe) (Terao et al. 1994). There is also the alternative synthetic product, being the first total synthesis of (±)-CYN reported by Xie et al. (2000), however, with the yield of 3·5%. The fact that the synthetic product is not commercially available, it is probably due to a more costly process than the extraction/purification process.

Ohtani et al. (1992) reported bioassay-guided method to isolate CYN from cyanobacteria biomass by size-exclusion chromatography (Toyopearl HW40F) and identification/purification by HPLC-UV on a reverse-phase column at a distinct absorbance maximum of 262 nm, due to the uracil nucleus. Then, more detailed analytical and semi-preparative HPLC methods for the determination and purification of the toxin in the same sample type were reported (Harada et al. 1994; Hawkins et al. 1997). A much more sensitive and specific LC-MS/MS method was developed by Eaglesham et al. (1999) being able to determine trace amounts of CYN in water samples. LC-MS/MS is the ideal method for small amounts of toxin and for complicated sample types, having, however, the disadvantage of being a costly approach. HPLC-PDA is one of the alternative methods to quantify the toxin and analogues present in cyanobacterial cultures and samples that do not produce any background like the case of the environmental samples (Welker et al. 2002). Another approach to obtain background free cell and animal tissue extracts is the addition of a cleanup step by SPE. Carmichael et al. (2001), Norris et al. (2001a,b) did an extensive study to find the best SPE sorbent to retain CYN and deoxy-cylindrospermopsin to concentrate the toxin present in the media of C. raciborskii cultures. The graphitized carbon cartridges were most effective, being able to retain CYN very strongly and with the media concentration of 800 μg l−1. Later, Kubo et al. (2005) successfully used a double system of cartridges, which consisted of a styrene polymer cartridge and an anion exchange cartridge. The cartridge double system was able to isolate both molecules from the same culture, CYN and deoxy-cylindrospermopsin. Gallo et al. (2009) were able to isolate CYN from water and fish muscle, the later extracted by a hexane liquid–liquid extraction followed by HLB-SPE, a hydrophilic–lipophilic balanced reversed-phase sorbent. The extracts of both types of samples had excellent detection limits by LC/ESI-MS/MS, 0·04 ng ml−1 and 0·6 ng g−1, respectively. Recently, Wörmer et al. (2009) demonstrated that for CYN dissolved fraction, the samples should be treated with a SPE method that consists of a sample preparation with 1% formic acid and 0·1% sodium chloride as well as the use of a combination of dichloromethane/methanol (1 : 4) acidified with 5% formic acid as the solvent.

Molecular assays

Molecular methods are nowadays widely applied in cyanotoxin research because they are simple, rapid, cost effective, extremely sensitive and specific, allowing the simultaneous analyses of several target gene products (Pearson and Neilan 2008). Increased knowledge of the cyanotoxins gene clusters and DNA sequences allowed the development of primers that target these sequences and, consequently, detect them directly from both environmental and culture samples. Although the complete CYN gene cluster has only been recently sequenced, Schembri et al. (2001) were the first to identify genes implicated in CYN production. The sequenced genes were AoaB (597 bp) and AoaC (650 bp), and the respective primers that amplify conserved regions within these genes were published. These authors also found that the existence of these two genes was directly linked with the ability to produce CYN, as they were both either present or absent in each of the tested strains. Afterwards, Fergusson and Saint (2003) developed a multiplex PCR assay that simultaneously detected the presence of AoaB and AoaC genes and a region of the rpoC1 gene unique to C. raciborskii, a well-known CYN-producing strain. This assay contains three primer sets and two of them already described: rpoC1 by Wilson et al. (2000) and AoaB by Schembri et al. (2001). AoaC forward primer was modified to amplify a smaller DNA fragment (422 bp) using the reverse primer previously described by Schembri et al. (2001). This assay proved to be reliable and robust in CYN potential production, as the toxin profiles of the tested strains obtained by chemical assays agreed with the PCR results. Other primers that amplify CYN biosynthesis genes were published by Kellmann et al. (2006), and these amplify specifically the regions of the A-domain in AoaC (478 bp), β-ketosynthase domain in AoaB (514 bp) and almost the entire AoaA gene (1105 bp).

Recently, a real-time PCR assay was developed to quantify genes specific to C. raciborskii and CYN-producing species using primers already developed for conventional PCR, in the light of their previously proved specificity. This assay developed by Rasmussen et al. (2008) aimed to compare microscopy cell counts with rpoC1 copy numbers in both water samples and cultured strains. It also wanted to establish the presence of the three genes involved in the toxin production. The results obtained revealed that in the field samples, the number of copies of rpoC1 was close to the cell counts and AoaC detection matched the results of toxin testing. With this method, detection and enumeration of C. raciborskii and CYN in environmental and laboratory samples proved to be faster and more sensitive, with a low detection limit of 100 copies/reaction or 1000 cell ml−1 for both rpoC1 and AoaC genes. This assay appears to be promising in future approaches to the monitoring of CYN-producing cyanobacteria. Q-PCR may now be used to quantify CYN-producing species and the toxicity potential of the present cyanobacteria (Moreira et al. 2011).

Immunological assays

The enzyme-linked immunosorbent assay (ELISA) is an immunoassay applied to the quantification of toxins in water samples with a sensitive level of detection. ELISA kits for CYN are now available having a low detection limit range around 0·05 and 2·0 ppb, and they allow its detection from several types of samples other than water (fish tissue, fish plasma, etc.) (Abraxis LLC, Warminster, PA, USA). However, this immunological assay much like the molecular methods currently developed has the disadvantage of being nonselective for CYN analogues, like deoxy-cylindrospermopsin and 7-epicylindrospermopsin. Also, reports on the detection of congeners or some cross-reactivity due to environmental interferences in comparison with LC/MS data have been documented (Yilmaz et al. 2008; Bláhová et al. 2009; Berry and Lind 2010).

In vivo assays

In vivo assays using CYN involve essentially mouse assays as will be described in section 3 of this review. However, other animals have been used in CYN toxicology assays; one of these is the insect Locusta migratoria migratorioides with the intent to develop a cost-effective test in comparison with the mouse test (Hiripi et al. 1998). With this work, they showed that the insect test was more sensitive and cheaper allowing the use of more animals with the toxicity being in the same range as the mouse assay. Other type of animals used in CYN toxicity assays were snails (Helix pomatia and Lymnaea stagnalis) (Kiss et al. 2002). In their study, CYN extracts obtained from a C. raciborskii cultured strain showed similar membrane responses on identified neurons with the inhibition of the acetylcholine responses. This study brings to light the effect that CYN has on snails (neurotoxic) in contrast with the well-established effects (hepatotoxic) that it has on vertebrates. Other animals such as guinea pigs have been used in studying the potential allergenic effects of CYN to humans (Torokne et al. 2001). Later, Metcalf et al. (2002) used Artemia salina nauplii to assess CYN toxicity with C. raciborskii extracts as an alternative bioassay toxicity method. In their work, they state that A. salina bioassay is more sensitive than the mouse bioassay and the unit costs more reduced compared to the in vitro protein synthesis assays. Recently, Nogueira et al. (2004) showed that CYN-producing C. raciborskii cells were toxic to Daphnia magna and that this can be used as a test organism for the presence of CYN.

Cell line assays

Cell line assays have been proposed as a replacement for the traditional mouse assays. The use of cell lines is of interest because it avoids technical isolation procedures and batch-to-batch variations inherent to primary hepatocyte cultures (Froscio et al. 2009). However, Runnegar et al. (1994) were the first to use cells in CYN toxicity assays, namely rat hepatocytes. Since then, cell lines have been used in some CYN toxicity assays and in the evaluation of its genotoxicity, with the first work with cell lines using ovary cells (CHO K1) (Fessard and Bernard 2003). Other cell lines used so far in CYN toxicity studies are HepG2 (liver), BE-2 (bone marrow), Caco-2 (colon), MNA (brain), HDF (dermis), C3A (liver), NCI-N87 (stomach), HCT-8 (ileum), HuTu-80 (duodenum), Vero (kidney) and CHO-K1 (ovary) (Fessard and Bernard 2003; Bain et al. 2007; Lankoff et al. 2007; Neumann et al. 2007; Froscio et al. 2009). In the study of Neumann et al. (2007), they established that Caco-2 cells were the most sensitive to CYN, eliciting a response at 0·25 μg ml−1 while the others cell lines only responded up to concentrations of 1 μg ml−1. In a more recent work, Froscio et al. (2009) determined that CYN cell line sensitivity decreased in cell lines from gradually more distal regions of the gastrointestinal tract, rather than the hepatic-derived cell lines that appear as more susceptible, while the Vero cells had more variable results.

Toxicological data

Human exposure

The only report described so far of human toxicity associated with CYN has been the Palm Island hepatoenteritis incident in 1979, where 148 people were hospitalized. During October of that year, a cyanobacterial bloom occurred in the Solomon Dam, the only source of water supply on Palm Island, about 28 km off the north-east Australian coast (Bourke et al. 1983; Hawkins et al. 1985). The bloom was treated with copper sulfate, and 5 days after this treatment, the first case of hepatoenteritis occurred. An epidemiological investigation conducted by Bourke et al. (1983) revealed that the people affected were Aboriginal descents, mainly children, which had drunk the water from wells that are supplied by the reticulated system. The epidemic lasted for 21 days and consisted of three well-defined stages: a hepatitis phase that lasted 2 days, a lethargic phase that lasted 1––2 days and a diarrhoeal phase of 5 days of duration. Later, Carmichael et al. (2001) reported the presence of CYN in carbon filters and ion-exchange resins in a dialysis centre in Caruaru (Brazil) the same were an outbreak of microcystins occurred. Until now, no other cases of human intoxication by CYN were ever reported.

In terms of human exposure to cyanotoxins, besides ingesting contaminated water and water used in dialysis treatment, other activities such as recreation may result in the dermal contact of the toxins through swimming or bathing. In fact, several studies have been conducted to evaluate the potential for skin irritation to the exposure to CYN in bathing waters. In Australia, it was found a correlation between the time spent in the water with the number of cyanobacteria and symptoms such as diarrhoea, vomiting and eye irritation (Pilotto et al. 1997). However, no allergenic effects were found to be correlated with CYN (Torokne et al. 2001). In other cases, some positive and negative data have been obtained as to the effects of this toxin in the cutaneous toxicity (Stewart et al. 2006a) with one patient in another study with cutaneous hypersensitivity to CYN patch exposure (Stewart et al. 2006b).

Acute toxicity

The first description on the severe toxicity of CYN was by Hawkins et al. (1985), where they used the C. raciborskii strain isolated from the Palm Island incident and showed that the strain was toxic to mice. The tested mice appeared huddled, anorexic and often with slight diarrhoea with the autopsy showing pale livers with white foci. Histopathology of the livers revealed hepatocellular coagulative necrosis that ranged from centrilobular at lower doses to involving all hepatocytes at 168 mg kg−1 of dosage (Hawkins et al. 1985). More affected livers showed sinusoidal congestion and often haemorrhages, and surviving hepatocytes showed lipidosis. Fibrin thrombi were observed within the liver portal veins leading to liver infarcts. Other organs were also affected such as kidneys, lungs and small intestines. Lungs contained embolic material in the small arteries and capillaries similar to those found in the liver, kidneys showed epithelial cell necrosis and small intestines congestion and oedema. Remaining organs appear normal. This study was the first that demonstrated that this strain is primarily hepatotoxic although other organs can be involved (Hawkins et al. 1985).

Later, Ohtani et al. (1992) after identifying the CYN chemical structure reported that pure CYN after i.p. administration in mice had a LD50 of 2·1 mg kg−1 over 24 h and 0·2 mg kg−1 over 5–6 days.

After Harada et al. (1994) discovered CYN in U.natans, Terao et al. (1994) used the highly purified CYN from this cyanobacterium to clarify its mode of action in mice and to analyse the morphogenesis of the poisoning induced by comparing it with cycloheximide, a potent inhibitor of protein synthesis in the mouse liver. This became the first study conducted on the ultra-structural morphology in the organ cells after CYN intoxication. After injecting a dose of 0·2 mg kg−1 in mice, electron microscopy analyses showed that it was possible to establish four consecutive phases of the pathomorphological changes induced by CYN intoxication: an initial phase (up to 16 h) with the detachment of ribosome from the membranes of the rough surfaced endoplasmic reticulum and, accumulation of free ribosome in the cytoplasm of hepatocytes, with condensation and reduction in the size of nucleoli. A second phase, at 24 h, was characterized by a membrane proliferation, followed by a third phase of fat droplet accumulation. The last phase consisted in cell death. Other organs like kidney had no special changes until 40 h after intraperitoneal injection. Massive necrosis of lymphocytes in the cortical layer of the thymus was also observed. Occasionally, single cell necrosis was seen in the heart of mice. Similar results were obtained with this strain by light microscopy in the liver and kidney (Harada et al. 1994). Biochemical analyses showed that protein synthesis was completely inhibited by CYN and that total amount of P450 was greatly diminished. With this study, it became clear that CYN induces various severe injuries mainly in the liver, with one of its direct effects as a potent inhibition of protein synthesis in various cell types. Later on, Seawright et al. (1999) showed that the stomach is also one of the affected organs when an oral dose of CYN is administrated. This study provided more appropriate information in regard to human exposure through drinking water. The extracts of a CYN-producing culture dosed orally caused severe intoxication in mice with the stomach filled with a mixture of diet and culture extracts that could also be found in the proximal third of the small intestine. Stomach dissection revealed multiple and small ulcerations that affected the oesophageal part of the mucosa. In this study, other organs were evaluated such as liver, kidneys as well as thymus and spleen. Macroscopically, both liver and kidneys were swollen and pale, while the spleen was shrunken and thymus atrophic. Microscopically, liver lesions were different than the ones observed by Hawkins et al. (1985). In a following study, Falconer et al. (1999) reported also changes in liver and kidney after oral and intraperitoneal administration of CYN in mice. Gross pathology showed that after 24–48 h with a dosage of a half of the LD50, the liver appeared mottled while the kidneys appeared pale. Histologically, there was a decrease in the number of red blood cells within the kidney glomerulus. With further exposure, tubule lumina of that glomerulus were enhanced and the amounts of pyknotic nuclei and of cellular necrosis had increased. Ultrastructural examination revealed the same results. In terms of liver damage, there was an increase in vacuolation and granulation of the cytoplasm. Cellular necrosis increased with time and dosage, and similar pathology was observed by both oral and i.p. administration.

In the light of the inter-individual variability observed between assays in previous toxicity works, Carmichael et al. (2001), Norris et al. (2001a,b) conducted a following study where they demonstrated that most CYN is excreted in the first 12 h, primarily in the urine, and in some of the tested mice, it took place in a significant manner by the faecal route. The animals in the assay were allowed to feed freely. The different excretion patterns obtained within exposed mice may explain the substantial inter-individual variability that has been previously reported in CYN toxicity assays. Nevertheless, the authors claim that liver appears as the main organ for accumulation and toxicity. A further study by Falconer and Humpage (2001) characterized CYN as a potential carcinogen. This was based on a study where 53 mice were administered by gavage with CYN cell extracts, and after a treatment period of 30 weeks, six treated animals showed histological evidence of neoplastic processes with three animals showing tumours. In a following study conducted by Chong et al. (2002), they tried to explain CYN uptake mechanism in the previous established affected organs. Their study showed that while in other toxins such as microcystin and lophyrotomin the uptake was through the bile acid transport system in hepatocytes, in CYN, the mechanism may involve more than one transport system. They claim that it is possible that the bile acid transport system is responsible for the uptake of CYN into the hepatocytes, however; another uptake route is also probable due to this protection offered by the bile acids on rat hepatocytes, only effective at 48 h. They propose that this second transport system might be simple passive diffusion due to CYN low molecular weight.

Humpage and Falconer (2003) after exposing mice to chronic dosages of purified CYN and also crude aqueous extract of a cultured C. raciborskii strain for 10–11 weeks, respectively, observed an increase in body weight at low dosages (30–60 μg kg−1 day−1) and a decrease at high dosages (432 and 657 μg kg−1 day−1), which resulted in an adverse effect. Also, the kidney weights were higher at lower dosages than in the liver experiments. Also, Reisner et al. (2004) reported that CYN had an adverse effect on red blood cells morphology by producing an acanthocyte form associated with a significant increase in the haematocrite. Furthermore, they observed variation in cholesterol levels in the plasma and liver of CYN-exposed mice to 0·6 mg l−1 of the toxin in the water for a 3-week period. These results are also in agreement with those obtained by Sukenik et al. (2006); however, the trial period was bigger (42 weeks) and the dosages were from 100 to 550 μg l−1 of mice exposed to the toxin in the water. Finally, in a different and more recent study shows that CYN may inhibit progesterone production through the inhibition of the cellular response of hCG (human chorionic gonadotrophin) in human granulosa cells. This study first documents the potential of CYN as an endocrine disrupter (Young et al. 2008).

Cytotoxicity of CYN

Runnegar et al. (1994) published the first work that established CYN as cytotoxic. Using CYN from C. raciborskii of the Solomon dam incident in cultured rat hepatocytes, they attempted to determine CYN mode of action by monitoring reduced glutathione (GSH) and lactate dehydrogenase (LDH). For LDH, an increase in its release was observed associated with increasing levels of CYN, with a 67% of LDH release at a toxic concentration of 5 μmol l−1 and, consequently, cell deaths. For GSH, a 50% cell reduction at a nontoxic dose of 1·6 μmol l−1 and a profound depletion at higher doses of CYN (5 μmol l−1) were observed. Cell GSH depletion preceded any increase in LDH release with GSH significantly depleted 6 h before toxicity even became significant. This depletion in cell GSH was later attributed to the inhibition of the final common pathway of GSH synthesis and not to the increase in GSH efflux or GSH utilization (Runnegar et al. 1995). This inhibition of cell GSH synthesis at nontoxic CYN concentrations is an important contributor of CYN cytotoxicity, as its depletion makes cells and tissues more susceptible to other potential contaminants. Later, Norris et al. (2002) investigated the role of GSH and P450 in vivo in mice and obtained the same results as Runnegar et al. (1994). They demonstrated that P450 is essential in the mechanism of action in CYN toxicity.

Runnegar et al. (2002) investigated the role of the sulfate group in CYN toxicity. This group is located in the C-12 position of the CYN molecular structure and was demonstrated that has no part in the uptake into cells and that has the same effects in the in vivo and in vitro assays. An interesting feature from this work is that CYN analogues lacking an intact C-ring and the methyl and hydroxyl groups of ring A could inhibit protein synthesis but at higher concentrations than the pure CYN.

Froscio et al. (2003) showed that the inhibition of protein synthesis was irreversible and tried to answer if the protein synthesis inhibition occurs independently of the CYP450-derived CYN metabolites generated by the mediated toxicity in hepatocytes. In their study, they showed that when applying a P450 inhibitor activity it diminished the CYN toxicity but not the effects on protein synthesis, implying that the two events had no association. Furthermore, it was demonstrated that protein synthesis inhibition occurred well before the onset of toxicity (0·5 μmol l−1 at 4 h). Recently, CYN was shown to inhibit the eukaryotic protein synthesis apparatus. Together with these findings, it was also suggested that CYN may target another protein of the translation system and not the ribosome (Froscio et al. 2008).

Genotoxicity of CYN

CYN besides considered as a cytotoxic toxin has also been described as genotoxic. Shaw et al. (2000) first showed that a covalent binding of CYN occurs after observing adduct spots in mouse liver DNA. Also, Humpage et al. (2000) found that CYN induced an increase in micronucleus frequency and an increase in centromere-positive micronuclei resulting in whole chromosome loss. They explained this by proposing two distinct mechanisms of action: first, acentric fragments are produced causing strand breaks; then, whole chromosomes are lost due to mal-segregation of the chromosomes during anaphase (Humpage et al. 2000). Later, Shen et al. (2002) suggested that induction of strand breakage at DNA level is probably one of the key mechanisms for causing the cytogenetic damage previously proposed by Humpage et al. (2000). Following, Humpage et al. (2005) investigated the metabolism of genotoxicity and also whether if CYN is able to induce an increase in the levels of reactive oxygen species (ROS). With this work, it was demonstrated that genotoxicity is the mechanism of action under CYN exposure and that this leads to the cytotoxicity of CYN contributing to its increase. P450-derived metabolites are responsible to both cytotoxicity and genotoxicity induced by CYN and that ROS are not mediators of CYN cytotoxicity and genotoxicity. Other works such as those from Bain et al. (2007) established that CYN activates the p53 transcription factor, which is a protein that is involved in the activation of the gene expression of proteins important in DNA repair or apoptosis. This protein increases in expression after exposing cells to 1 μg ml−1 of CYN at 6 and 24 h of exposure. In contrast, Lankoff et al. (2007) reported that CYN increases the frequency of necrotic cells and necrotic cell death in a dose- and time-dependent manner and has a very small impact on apoptosis. In their work, they also determined that CYN had no clastogenic activity in CHO-K1 cells; however, it significantly reduces the frequencies of mitotic indices. Similar results were obtained previously by Fessard and Bernard (2003).

Cylindrospermopsin in the Environment

Occurrence and persistence

CYN occurrence has been reported either in field samples (water) or in isolated species so far in four of the five continents: Oceania, Asia, America and Europe. In field samples, reports so far include Oceania (New Zealand and Australia) and Europe (Germany, Spain, France, Czech Republic and Poland). The first description of the presence of CYN in a field sample in the Oceania continent was by Saker et al. (1999a) where they isolated two distinct morphotypes of CYN-producing C. raciborskii from the Solomon dam water system in February of 1996. Another example is also from Saker et al. (1999b) where in this study they isolated a CYN-producing C. raciborskii from the water of a farm dam to which it was attributed the death of the cattle that drank from that water. Further description was made by Fabbro et al. (2001) where they isolated CYN-producing C. raciborskii strains from a river in Queensland, Australia. In New Zealand, the first report was from a scum that formed on the surface of the recreational Lake Waitawa near Wellington. The sample was tested for the presence of CYN by LC-MS analysis but not quantified due to the lack of standard for the assay (Stirling and Quilliam 2001). In Australia, documented reports on the presence of CYN in field samples are mainly attributed to the occurrence of C. raciborskii blooms, the first species where this toxin was described (see chapter one of this review), and that constitutes in this country a major concern (Eaglesham et al. 1999).

In Europe, Fastner et al. (2003) reported the presence of CYN for the first time. The toxin was found in water samples from two German lakes (Lake Melangsee and Lake Langer See); however, CYN production was not attributed to any of the C. raciborskii isolates. Quesada et al. (2006) detected CYN in a Spanish reservoir associated with a bloom of A. ovalisporum with values as high as 9·4 μg l−1 in the sestonic fraction. Later, Rücker et al. (2007) found its presence in 21 lakes and Fastner et al. (2007) found its presence in 63 lakes from Germany showing its high prevalence in this country with concentrations reaching up to 12 μg l−1. Czech Republic, France and Poland are countries where CYN has been recently detected and quantified from water samples without assigning its production to a particular species. In the Czech Republic, CYN was quantified in low amounts in water blooms of Aphanizomenon at concentrations up to 200 μg g−1 of dry weight of biomass (Bláhová et al. 2008). In France, CYN was detected in 6 water bodies in concentrations varying from traces to 1·95 μg l−1 (Brient et al. 2009). Finally, CYN was reported in two lakes from Western Poland at concentrations ranging from 0·16 to 1·8 μg l−1 (Kokociński et al. 2008). The occurrence of CYN in isolated species has been previously enumerated and described in chapter one of this review.

There are a few studies on CYN persistence in the environment. The first study to be conducted on CYN stability under environmental conditions was by Chiswell et al. (1999). In their study, under artificial environmental parameters such as temperature, pH, light and UV intensity, CYN was examined. Natural sunlight conditions were also tested. The data obtained showed that there is a slight decrease in CYN concentrations. Under direct sunlight, the biggest decrease occurred in the summer months.

When studying the effect of temperature in C. raciborskii batch cultures, Saker and Griffiths (2000) demonstrated that there was a decrease in the CYN concentrations when the temperature reached growth values (35°C) for the tested cultures. However, when the temperature dropped, the ability to produce CYN was fully restored, indicating that the capacity of CYN production was blocked at 35°C but not destroyed. However, in the light of these findings, no consequences for CYN persistence under such environmental conditions were extrapolated. In opposition, CYN degradation has been studied. Several chemical methods, such as oxidation by chlorine, ozone and ultra-violet light with the addition of titanium dioxide, have already been studied. The results from this work showed that each method is effective in CYN removal under various conditions and that chlorination efficiency is pH dependent and that UV degradation is more efficient with the addition of titanium dioxide (Senogles et al. 2001). Another alternative method has been applied in CYN degradation, which is the microbial method. In this, bacteria are inoculated in the water with CYN, and after a certain period of time, it was found a reduction in the toxin concentration (Smith et al. 2008). Other studies refer to grazing effects of the ciliate Paramecium cf. caudatum on the CYN-producing C. raciborskii cultures by reducing its density in water samples and have been proposed to be used as a management strategy for this toxin (Fabbro et al. 2001). In terms of drinking water, Hoeger et al. (2004) evaluated the CYN removal in a water treatment plant. The data obtained showed that the efficiency to remove CYN from raw water was 46% after a treatment by flocculation and 100% using filtration and chlorination. They also observed that 38·1% of CYN was released from the cells after flocculation. Later, Wörmer et al. (2010) observed a 27% CYN degradation when exposed to direct sunlight during a 22-day trial period and that UV-A was the main contributor to CYN degradation. The effect of the water depth was also examined with samples from 4 m depth showing no variation in CYN concentration when compared with 1-m-depth samples.

In an experiment using the CYN-producing A. ovalisporum, Bácsi et al. (2006) showed an alteration in the CYN content in the S-deprived cultures. They saw that sulfate or phosphate starvation of cells induced a decrease in CYN content, and after a 48-h starvation period when sulfate was added, CYN content started to increase. Under phosphate starvation, a more moderate decrease was observed. Rücker et al. (2007) observed that CYN dissolved fraction in water containing the CYN producer Aphanizomenon is higher than the CYN particulate fraction with the proportion of dissolved CYN being more than 80% of total CYN in 31% of the tested samples. They also found no significant correlations between morphometric parameters like the lakes area, volume or mean depth and CYN occurrence. Particulate CYN showed positive correlations with total phosphorous, chlorophyll a, phytoplankton and cyanobacterial biovolume, and negative correlation with Secchi depth. They also concluded that the probability of CYN occurrence is higher in eutrophic lakes than in other lakes.

Hawkins et al. (2001) aiming at testing the phenotypical variation of a CYN batch culture discovered that CYN intracellular concentration was higher at the stationary phase and that the CYN extracellular concentration in the batch culture was higher by the end of log phase. These findings suggest that a single genotype in a slow-growing persistent bloom of C. raciborskii in natural waters will held the highest levels of CYN and a highest risk for consumers since a large amount of CYN may be released. Recently, Bar-Yosef et al. (2010) established the CYN biological role in the sense that has the unique function to serve as an inducer of the excretion of alkaline phosphatase (APase) by other organisms to the water systems.

Water regulations

Cyanotoxins are a major concern to aquatic ecosystems as well as for public health due to the use for drinking, agricultural or recreational purposes. The occurrence of cyanotoxins and namely CYN has lead national governments to establish guidelines and/or recommendation values for this toxin in the water. The guideline value of 1 μg l−1 in drinking water for CYN was proposed by Humpage and Falconer (2003) where they estimated the no observed adverse effect level (NOAEL) after an in vivo mouse oral dosing assay. In the estimation of this value, two variables were assumed, the average body weight of an adult human (60 kg) and the average value of water consumption (2·0 l day−1), which resulted in a 0·9 of total toxin intake leading in practical purposes to the guideline value of 1 μg l−1. After this guideline was proposed, several nongovernmental agencies, such as research institutions, assumed this value for toxicity assessment in water quality management and control. Later, few governments adopted this guideline value for CYN in drinking water, namely Brazil and New Zealand (Burch 2006). No guideline values for CYN in recreation and bathing waters have been adopted yet. Although these two countries have recommended values for CYN in drinking water, the established values differ greatly. While in Brazil the federal legislation includes a guideline value for CYN of 15 μg l−1, in New Zealand, the maximum acceptable value is of 1 μg l−1 (Burch 2006).

Exposure to domestic and wild animals

So far, only one incident of CYN intoxication on livestock animals has been reported (Thomas et al. 1998; Saker et al. 1999a). In the first report, one cow and three calves were found dead around the dam from a herd of 300 cattle (150 cows and 150 calves) that drank water from the dam containing a bloom of cyanobacteria. Necropsy showed severe abdominal and thoracic haemorrhagic effusion, hyperaemic mesentery, pale and swollen liver, extremely distended gall bladder containing dark yellow bile and the heart with extensive epicardial haemorrhages. During the next 3 weeks, a further eight animals died. Water examination showed the presence of toxic C. raciborskii similar to the Palm Island incident with a toxicity of 153 mg kg−1 mouse (LD50 at 7–8 h). Later, Saker et al. (1999b) showed through HPLC/MS that the toxin present was in fact CYN but at lower concentrations than those observed by Hawkins et al. (1985, 1997).

The only reports on animals were on the amphibian Bufo marinus. In this, two studies were conducted one by Kinnear et al. (2007b) and the other by White et al. (2007). In the first, they analysed the possibility of CYN uptake and toxicity by exposing the young animals to whole cell extracts, which represented CYN extracellular, and to live cells, which represented both CYN extracellular and intracellular that were administered as food source. Exposure to whole cell extracts resulted in a haemorrhagic liver, with hepatocytes appearing atrophied, degenerated and disorganized. The cells also exhibited increased lipid droplets, and some blood vessels contained debris (Kinnear et al. 2007b). Exposure to live C. raciborskii resulted in identical effects but with concomitant necrotic tissue. Kidneys were also affected in both assays. In this sense, they showed that the effects associated with intracellular toxin exceeded those recorded for extracellular exposure and also that it appears that oral and transdermal uptake of CYN is likely to occur. In a second study by White et al. (2007), they determined the adverse effects of CYN in Bufo marinus under the same conditions as described in the study of Kinnear et al. (2007b). They observed as previously that live C. raciborskii toxic cells were more toxic than whole cell extracts resulting in a 66% of mortality, changes in the normal tadpole behaviour and lower relative growth rate. They also observed the bioaccumulation of the toxin in the live cells trial being up to 19 times superior to the extracellular toxin trial. These data indicate the ability to bioaccumulate CYN via grazing in spite the fact that it is unknown if C. raciborskii is part of its natural diet.

Exposure to aquatic animals

Some studies on aquatic animals report the occurrence of CYN and are based in its effects, bioaccumulation and transfer into higher tropic levels of the food chain. The first work to be published on the exposure to CYN in aquatic animals was by Saker and Eaglesham (1999) who report the effects of a severe bloom of C. raciborskii on Redclaw crayfish (Cherax quadricarinatus) in a aquaculture pond in Townsville, Australia. They observed the presence of CYN in the hepatopancreas and muscle of Redclaw crayfish in a proportion of 5 : 1 and also in the viscera in concentrations similar to those in the muscle tissue of the crayfish. No histological abnormalities were observed in the exposed crayfish in spite of the high concentrations observed (1 g of freeze-dried hepatopancreas tissue contained enough toxin to be lethal to a 20-g mouse if administered by i.p. injection).

Another study by Saker et al. (2004) showed that the freshwater mussel Anodonta cygnea accumulated CYN after exposing the animals to a CYN-producing culture of C. raciborskii for a 16-day period at cell density ranging from 265 000 to 1900 000 cells ml−1. At the end of the experiment, the amounts of toxin expressed as a percentage of the total toxin content of the Anodonta were as follows: haemolymph (68·1%), viscera (23·3%), foot/gonad (7·7%) and mantle (0·9%). Haemolymph and viscera bioaccumulated more than 90% of CYN. They also observed that following a 2-week depuration period, approximately 50% of the toxin still remained in the tissues.

Bioaccumulation and toxicity have also been reported in the freshwater gastropod Melanoides tuberculata. White et al. (2006) reported that this gastropod is able to bioaccumulate CYN when exposed to live cells of C. raciborskii in much higher concentrations than Anodonta cygnea. Nevertheless, they could not bioconcentrate the toxin suggesting that CYN uptake occurs through their feeding grazing mechanism rather by dissolved toxin uptake, as small quantities were detected in the mantle and in the shell wall. The study also gives rise to the problem of CYN biomagnification in higher levels of the trophic food chain like in fish and ultimately humans. Finally, another study conducted by Kinnear et al. (2007a) with the same species reports sublethal responses after exposure to live C. raciborskii cells containing CYN. The exposure to whole cell extracts or to live C. raciborskii cells did not cause changes in mortality, relative growth rates or behaviour in the adult animals. However, it had an effect in the release of hatchlings with the treated animals with high toxin concentration recording the least of the hatchlings.

In fish, the only work published so far was by Berry et al. (2009) using zebrafish (Danio rerio) embryos. They found that with pure CYN, no developmental effects in the fish embryos were observed but with the cell extracts embryo malformations were seen.

Conclusions

The knowledge discovered so far has allowed classification of CYN as a dangerous toxin due to its proved cytotoxicity, carcinogenic potential and genotoxicity. CYN was found so far in eight different cyanobacterial species in four of the five continents. With the increasing reports of its occurrence worldwide without the assignment of the producing species, it suggests that besides the already known species, there are others that are associated with the production of CYN. There is a need to develop methods, besides isolation and culturing of cyanobacterial isolates, to determine the presence and occurrence of this toxin in the aquatic systems worldwide. In this sense, the HPLC and more recently the LC-MS methods have allowed us to determine the presence of CYN but also of its analogues namely deoxy-cylindrospermopsin and 7-epicylindrospermopsin in environmental samples. The molecular methods allow us to determine the presence of the genes involved in CYN synthesis but also help us to identify the species responsible. However, in the light of the current knowledge, new primers based on the recently sequenced CYN gene cluster may be used to enhance new phylogenetic as well as biogeographic studies. This can help us to understand how this toxin has evolved as well as possible biogeographic patterns that may explain how this toxin has dispersed through the world. The legislation such as that of Brazil and New Zealand should be implemented in assigned countries and the continuous screening for the presence of CYN in places such as the African continent where its presence still remains undetected is a need.

Acknowledgements

This research was funded by the PesT-C/MAR/LA0015/2011 project from Fundação para a Ciência e Tecnologia (FCT).

Ancillary