Cyanotoxins: New Generation of Water Contaminants

Cyanobacteria, more commonly known as blue-green algae, are found worldwide in various aquatic environments as well as in water distribution systems (Atikovic 2003; Carmichael 1994; Madigan et al. 2003). Blooms of cyanobacteria have recently become spatially and temporally more prevalent in the United States and worldwide as a consequence of increasing nutrient levels such as nitrates and phosphates from fertilizers and detergents. Cyanobacterial blooms impart color, odor, and taste problems in water. More importantly, such blooms produce and release toxic compounds that dramatically impair the quality of water bodies. Up to 50% of the recorded blooms can be expected to contain toxins (Carmichael 1992). These compounds have severe and sometimes irreversible effects on mammalian health. Episodes of human and animal poisoning by consumption of water contaminated with cyanobacterial toxins have been reported since the late 1800s (Carmichael 1994). Exposure to cyanobacterial toxins can affect the number and diversity of wild animal populations, cause bioaccumulation of toxins in the tissues of fish and shellfish, and indirectly affect other organisms through the food chain. Moreover, the presence of cyanobacteria and cyanobacterial toxins in sources of drinking water supply has raised major concerns. Another major issue is the lack of guidelines or regulations of cyanobacteria and cyanotoxins in terms of maximum contaminant level (MCL) and analytical detection methods. In the past few years, major research effort has been targeted toward the treatment of these toxins, especially the hepatotoxin microcystin-LR (MC-LR).

Cyanobacteria are present in various aquatic environments, including water distribution systems and can produce scum (usually greenish) observed at the bottom and surface of lakes. In ecosystems, the presence of cyanobacteria is crucial as they provide oxygen to specific aerobic microorganisms (symbiotic living), convert inorganic nitrogen to organic forms that plants and phytoplankton can utilize and act as a source of energy to zooplankton (primary producers). Recent studies have shown that zooplankton consumes cyanobacterial algal blooms selectively depending on the availability of food. The nonconsumable blooms can be lethal to or can reduce the number and size of progeny of zooplankton (Carmichael 1994). Cyanobacteria contain and release toxins as a metabolic byproduct and also during “cell death” (lysis), except for cylindrospermopsin, which is secreted. Cyanotoxins can cause irritations of the skin and other organs (dermatotoxins and irritant toxins), cell damage (cytotoxins), and liver damage (hepatotoxins), or affect the nervous system (neurotoxins) (Wiegand and Pflugmacher 2005). About 50 species of cyanobacteria are known to produce toxins, but not all compounds produced during cyanobacterial blooms are toxic to humans and animals. Cyanobacterial toxins can also be grouped based on structure: cyclic peptides (hepatotoxins), alkaloids (neurotoxins), and lipopolysaccharides (LPSs). The toxins that are of most concern in the United States are the derivatives of microcystin, which are cyclic peptides (hepatotoxins), and cylindrospermopsin, anatoxin-a, saxitoxin, and anatoxin-a(s), which are all alkaloids (neurotoxins). The cyanotoxin that has been the focus of most researchers dealing with such problems around the world is the specific derivative of microcystins, MC-LR.

MCs are a group of toxins known to promote liver cancer. These cyclic heptapeptides consist of five invariant amino acids and two variant amino acids. The latter are characteristic of the $>$ 60 derivatives of microcystin that have been found so far. Their high stability under extreme conditions and high solubility in water are attributed to their cyclic chemical structure and their functional groups, respectively. As naturally occurring organic compounds, MCs can undergo biodegradation. Because of their stable structure, they can persist in aquatic systems for days before any significant degradation occurs. MC-LR (L and R stand for leucine and arginine, respectively) is the most toxic and most frequently found derivative of microcystins in water resources. The chemical structure of MC-LR is shown in Fig. 1. MC-LR was found to have a half-life of three to four days in aquatic systems under laboratory conditions (Kenefick et al. 1993; Cousins et al. 1996). The presence of pigments and humic acids has been found to enhance the degradation of the toxins. Humic acids have the ability to absorb light in the UV-Vis range, and act as photosensitizers for the formation of very highly oxidizing species (free radicals) including hydroxyl radicals (Welker and Steinberg 1999; Welker and Steinberg 2000). Even so, some studies have reported persistence of MC-LR for up to nine days in concentrations as high as $1300–1800\mu \mathrm{g}∕\mathrm{L}$ before any significant degradation occurred (Jones and Orr 1994).

Health-related episodes in humans and animals caused by MC-LR contamination have been reported in several countries, including the United States, Australia, China, Great Britain, and Brazil (Carmichael 1994; Robertson et al. 1997; Pouria et al. 1998). After the first human fatal incident occurred in Brazil in 1996, the World Health Organization (WHO) set the provisional concentration limit of MC-LR in potable water to $1\mu \mathrm{g}∕\mathrm{L}$ . The WHO has also established the tolerable daily intake (TDI) to $0.04\mu \mathrm{g}{\mathrm{kg}}^{-1}{\text{day}}^{-1}$ since cyanobacterial toxins bioaccumulate in aquatic microorganisms that humans consume and because of their use as dietary supplements. The state of Oregon has adopted $1\mu \mathrm{g}{\mathrm{g}}^{-1}$ $\left(1\mathrm{ppm}\right)$ of microcystins as a standard for cyanobacterial products. The intake for a person that weights $60\mathrm{kg}$ and consumes $2\mathrm{g}$ of product containing $1\mu \mathrm{g}{\mathrm{g}}^{-1}$ MC-LR is $0.033\mu \mathrm{g}\text{per}\mathrm{kg}$ of body weight, which is lower than the provisional limit (WHO 1999). However, for a given consumer of these products, considerations must also be made for microcystin intake from other sources (such as drinking water, recreational water, and shower).

Cyanobacteria and their toxins are currently in the U.S. Environmental Protection Agency (USEPA) “Drinking Water Contaminant Candidate List” (CCL) (USEPA 2005). The CCL was first formed in 1998 and included 60 contaminants that require regulatory determination. Occurrence of cyanobacteria and cyanotoxins were expected in the public water systems and they were therefore included in the list. In May 2001, a workshop was held at the USEPA in Cincinnati for “Creating a Cyanotoxin Target List for the Unregulated Contaminant Monitoring Rule.” The final report was published in August 2003 (USEPA 2003). Last February, the USEPA released the second CCL with 51 contaminants with respect to the Safe Drinking Water Act (SDWA) requirements and again, cyanobacteria and their toxins were included. While action is being taken, MC-LR has still not been regulated in terms of MCL and best available technology (BAT).

The lack of guidelines or regulations for cyanobacteria or cyonotoxins is one aspect of the problem; the other is the absence of toxicity data, which is manifested in the diverse MCL values found worldwide. The Australian, Canadian, New Zealand, and French regulatory standards set the contamination limit of MC-LR to 1.3, 1.5, 1.0, and $1.0\mu \mathrm{g}∕\mathrm{L}$ , respectively, whereas the United States has no standard (Hoeger et al. 2005). Microcystins, besides promoting liver cancer, also disrupt the proper functioning of an important group of cell enzymes, protein phosphatases (PP). Inhibition occurs by binding to the PP through a carboxyl group, two hydrogen bonds, and an aliphatic chain (ADDA) (Goldberg et al. 1995). If any of the three components is missing or altered, then MCs fail to bind to the receptors. Specifically, two derivatives of MCs (LR and RR) where the bond, at ${\mathrm{C}}_6$ -methyl and ${\mathrm{C}}_7$ -hydrogen (circled in Fig. 1) was in cis configuration, rather than the usual configuration, trans, have been found to be nontoxic (Harada et al. 1990). Toxicity data are crucial not only for each type of toxin, but also for the combined effects of different toxins since most harmful algal blooms (HABs) produce a mixture of toxins in natural environments. In the human fatal incident that occurred in the dialysis clinic in Brazil in 1996, MC-LR was found in the tissues of the patients at high concentration, but the patients also exhibited symptoms that where attributed to neurotoxins. Hence, the incident could not be conclusively attributed to MC-LR alone (Pouria et al. 1998).

The low standard of the WHO for MC-LR and the high toxicity of the toxin ( ${\mathrm{LD}}_{50}$ around $50\mu \mathrm{g}∕\mathrm{kg}$ ) makes the development of highly sensitive analytical methods a necessity. Reversed-phase high-pressure liquid chromatography (HPLC) with photodiode-array (DAD), electrochemical, fluorescence, and mass spectrometry (LC-MS) detectors, as well as capillary electrophoresis (CE) with UV or MS detectors, have been developed and used to determine the presence and concentration of microcystins in the samples. Methods based on gas chromatography (GC) with mass spectrometer (MS) or flame ionization detectors (FID) have also been developed. Chromatographic analysis using HPLC with UV detectors is the most commonly used method to identify and quantify microcystins. Biochemical assays, which include enzyme-linked immunosorbent assay (ELISA) and protein phosphatase inhibition assays, are sensitive, rapid, and suitable for large-scale screening. Both tests, however, are predisposed to false positives and unable to differentiate between microcystin and nodularin (another hepatotoxin) variants. Several immunoassays have been developed to detect microcystins. Commercially available ELISA kits using groundwater and surface water are limited by the matrix effect, detection of both inactive and active microcystins and nodularins, and detection of toxic and nontoxic microcystin variants (Rivasseau and Hennion 1999). An enrichment step by solid-phase extraction (SPE) using ${\mathrm{C}}_{18}$ column can enhance the sensitivity and avoid false positives. ELISA and SPE followed by HPLC were highly correlated using spiked drinking and surface water samples (Rivasseau et al. 1998; Metcalf and Codd 2000). Despite its limitations, ELISAs allow rapid on-site detection of toxins without pretreatment in the field. Most research groups are utilizing reverse phase HPLC with a DAD detector. Even so, the USEPA has still not assigned a best available technology for MC-LR due to the existence of various variants and analogues for each variant.

The absence of guidelines from the USEPA for the best technologies to remove cyanobacterial toxins from drinking water has propelled research for exploring both conventional and emerging technologies for the treatment of cyanobacterial toxins. Various treatment technologies have been tested for the inactivation, degradation, and removal of MC-LR in water resources. These are coagulation, sedimentation, chlorination, and activated carbon, as well as emerging hydroxyl radical-based processes known as advanced oxidation technologies (AOTs). AOTs usually utilize radiation (UV, solar), strong oxidants (i.e., peroxide, ozone) and/or catalysts for the formation of highly oxidizing species (i.e., hydroxyl radicals). Fenton reagent, ozonation, and titanium dioxide photocatalysis are examples of AOTs utilized for the degradation of MC-LR (Lawton and Robertson 1999).

Frequently used methods in drinking water treatment like coagulation, flocculation, and sedimentation are very efficient for the removal of algal cells and some of the toxins. In contrast to the microcystins, cyanobacteria cells are very fragile and can be easily damaged. It has been reported from pilot plant studies that the application of mechanical force during water treatment can increase the number of cells that undergo lyses and consequently the overall toxin concentration (Schmidt et al. 2002). To control the phytoplankton blooming rate in water resources, a number of chemicals have been tested including potassium permanganate, copper sulfate, and chlorine. These chemicals inhibit phytoplankton production by disrupting enzymatic reactions such as photosynthesis and cell wall synthesis (Lam et al. 1995). The addition of copper sulfate removes the algal blooms, but increases the concentration of soluble MC-LR. Lam et al. (1995) tested the impact of frequently used chemical substances during water treatment including potassium permanganate and chlorine. They also tested common coagulants (alum, ${\mathrm{Al}}_2{\left(\mathrm{S}{\mathrm{O}}_4\right)}_3.14{\mathrm{H}}_2\mathrm{O}$ ) and lime softeners $\left(\mathrm{Ca}{\left(\mathrm{O}\mathrm{H}\right)}_2\right)$ that influence the algal bloom by precipitating cyanobacterial cells and their growth-limiting elements (phosphorus). They concluded that chlorine was more effective than permanganate for the removal of phytoplankton but induces a greater release of soluble MC-LR. Therefore, a waiting period is essential before the water is used for consumption or these oxidants should only be used after the complete removal of the algal cells from the treated water. On the other hand, alum and lime addition resulted in a reduced or zero MC-LR release, respectively. The cell removal was achieved through cell coagulation and sedimentation. Consequently these methods can be safer for removing cyanobacterial cells and some of the soluble toxin. A major concern is the treatment of the sediment containing cyanobacterial cells and toxins and the formation of chlorinated byproducts (CBPs) that can be more harmful for mammalian health than the toxin itself. This issue has not been investigated thoroughly yet, so no concrete conclusions can be made.

Activated carbon, in both granulated and powdered form, was used for the removal of taste and odor from water, and then tested for its efficiency for removing microcystins. Its performance has been found to depend on parameters such as the derivatives of microcystins and their respective solubilities, the concentration of the toxin and the dose and origin of the activated carbon (Lawton and Robertson 1999; Mohamed et al. 1999). MC-LR is estimated to have a diameter of approximately $3\mathrm{nm}$ , and therefore the micropores of activated carbon are left unutilized. Usually, higher concentration (more than 8 times the normal) of activated carbon (AC) is necessary for the treatment of the toxins and an early breakthrough of the MC-LR is observed. A survey on AC filters for domestic application showed than none of them was able to totally remove microcystins with the first pass (Lawton et al. 1998).

A very common oxidant that is used in environmental technologies because of its efficiency and its environmental friendly (“green”) characteristics is hydrogen peroxide $\left({\mathrm{H}}_2{\mathrm{O}}_2\right)$ . Additional components that can be used for its activation (i.e., generation of hydroxyl radicals) include UV radiation or catalysts (i.e., transition metals). Decomposition of hydrogen peroxide with iron (ferrous or ferric), known as the Fenton reagent, results in the formation of hydroxyl radicals (primary oxidizing species) (Gajdek et al. 2001). The Fenton reagent has been tested for the degradation of a number of organic contaminants and demonstrated high removal efficiencies. In many cases, the Fenton reagent is used in combination with UV radiation, known as photo-Fenton reagent, as an approach to overcome certain limitations of the former such as the requirements of large amounts of iron and iron precipitation (Anipsitakis and Dionysiou 2003). Bandala et al. (2004) tested Fenton and photo-Fenton for the degradation of MC-LR. The authors reported 84% removal efficiency of MC-LR with photo-Fenton in $25\min$ and 100% within the next $10–15\min$ , with an initial concentration of toxin and iron at $8\mu \mathrm{M}$ and $0.25\mathrm{mM}$ , respectively. An increase in peroxide concentration from $0.1\text{to}0.5\mathrm{mM}$ did not have a significant effect on the degradation. Gajdek et al. (2001) using using Fenton reagent reported similar results to previous studies. No direct comparison can be made between the two studies because in the latter, the investigators used a significantly higher initial concentration of toxin (75 times).

Ozonation is another AOT used for water purification, especially for the removal of aromatic contaminants, because it reacts very fast with conjugated double bonds. The reaction with the ozone and double bonds first goes through ozonides (C-O-O-O-C bridge) followed by the cleavage of the aliphatic chain to carbonyl compounds, ketones, and aldehydes, depending on the position of the double bond. MC-LR's unique ${\mathrm{C}}_{20}$ amino acid, known as ADDA, is very susceptible to ozonolysis because of the two-olefin groups at positions 4-5 and 6-7. Similar like Fenton reagent, treatment ozonation is highly pH-dependent process. The standard hydrogen electrode (SHE) potential of ozone is reduced to half when the pH is alkaline ( ${\mathrm{SHE}}_{\mathrm{pH}<7}=2.07\mathrm{V}$ versus ${\mathrm{SHE}}_{\mathrm{pH}>7}=1.24\mathrm{V}$ ) (Rositano et al. 1998). Other parameters that can affect the performance of this process are the presence of natural organic matter (NOM) and alkalinity. The potential problem of ozonolysis is the formation of brominated DBPs when there are high bromide concentrations in the treated water. Nevertheless, ozonation can be a very efficient method for the complete removal of cyanotoxins from water in addition to the odor-causing compounds such as geosmin and 2-methylisoborneol (MIB) (AWWA 1999). However, the characterization of the intermediates of oxidation has not yet been performed, which is a major disadvantage for the establishment of the method as possible BAT.

Rositano et al. (1998) investigated the effectiveness of chlorine, hydrogen peroxide, potassium permanganate, and ozone for the degradation of MC-LR, and found ozone to be the most effective as it removed 99% of the toxin in the first $15\mathrm{s}$ . In addition, the combination of ozone and hydrogen peroxide was found to be more effective than ozone alone. Similar results were obtained by the Water Services Association of Australia (WSAA). The results of their study indicated that monochloramine and hydrogen peroxide were completely ineffective for treating MC-LR. The oxidation efficiency for each method as determined in the study is summarized below from the greatest to the least (WSAA 1996):

Titanium dioxide photocatalysis is another popular emerging technology for the destruction of cyanobacterial toxins in water, especially MC-LR. Illumination with UV radiation of the $\mathrm{Ti}{\mathrm{O}}_2$ particles causes photoexcitation of electrons and the formation of electron-hole pairs. The holes, which are strong oxidizing species ( $E_o=+2.53\mathrm{V}$ at $\mathrm{pH}=7.0$ ), react with water and hydroxyl anions adsorbed on the surface of $\mathrm{Ti}{\mathrm{O}}_2$ and form hydroxyl radicals (Fujisjima et al. 2000). The hydroxyl radicals ( $1.8-2.7\mathrm{V}$ depending on the pH) are highly active and nonselective oxidizing species. These radicals have very high second-order reaction rate constants with most organic contaminants and react rapidly with organic contaminants that are adsorbed on the catalyst surface or are within the diffusion layer around the catalyst (Lawton and Robertson 1999).

Robertson et al. (1997) used 1% (m/v) of suspended $\mathrm{Ti}{\mathrm{O}}_2$ for the removal of MC-LR from water and they found that the toxin was completely removed in short treatment times (40 min at pH=4) even at high concentrations like 200 mM (Robertson et al. 1997). The degradation of MC-LR is highly dependent on the pH of the treated solution. The maximum degradation efficiency was observed at initial pH of 3.0. Feitz et al. (1999) explained how the acid-base chemistry of MC-LR is controlled by three amino acids. D-methylaspartic acid and the D-glutamic acid have free carboxylic groups. The pKa for both is around 3.0. L-arginine has two quanidino groups (basic group, $\mathrm{pKa}=12.48$ ). With increasing pH, MC-LR loses two protons from the carboxylic groups, making the overall charge $-1$ . This applies for most of the pH range $(3<\mathrm{pH}<12)$ . At extremely basic pH, MC-LR loses the proton from the protonated basic group and the overall charge is $-2$ . Thus, the overall charge transition (dissociation) of MC-LR in an aqueous medium can be summarized as:

$\mathrm{Ti}{\mathrm{O}}_2$ photocatalysis is a technology with great dependency on the pH of the treated water. With increasing pH, titanium dioxide’s overall surface charge changes from positive to neutral to negative with the point of zero charge being around $\mathrm{pH}\approx 6.4$ . Studies with $\mathrm{Ti}{\mathrm{O}}_2$ slurries for the destruction of biological toxins and several other organic contaminants reported increased degradation and mineralization efficiencies at acidic pH because of the decrease of the repulsion between catalyst, target compound, and intermediates (Feitz et al. 1999; Schmelling et al. 1997). Feitz et al. (1999) studied extensively the effect of pH on the degradation of MC-LR with titanium dioxide photocatalysis and set the optimum pH at 3. At this pH the toxin has an overall charge of $(-1)$ while the $\mathrm{Ti}{\mathrm{O}}_2$ is positively charged, increasing the attraction forces between catalyst and contaminant.

Lui, Lawton, and Robertson (2003) conducted studies on the photocatalytic oxidation pathway of MC-LR with $\mathrm{Ti}{\mathrm{O}}_2$ photocatalysis. They concluded that the major mechanism for the photocatalytic degradation of MC-LR was isomerization, substitution, and cleavage of the conjugated dienes of the ADDA amino acid. The study indicated elimination of the toxicity of MC-LR but also further decomposition of the primary intermediates. The oxidation pathways were identified. The main limitation of this treatment is the low mineralization efficiency, which is about 10% (Lawton et al. 1999).

The contamination of drinking water with cyanobacterial toxins is a seasonal phenomenon that lacks a standardized detection method, a technology for removing the toxins from drinking water, and formal guidelines and regulation from the USEPA. USEPA appears to be determined to take necessary measures against these groups of contaminants as reflected in the organization of the “International Symposium on Cyanobacteria Harmful Algal Blooms” (September 6–10, 2005, in North Carolina). Among the sponsors of the conference are the USEPA, the U.S. National Oceanic and Atmospheric Administration (NOAA), the U.S. Army Corps of Engineers (USACE), and the U.S. Centers for Disease Control (USCDC). To bridge the gap between the problems posed by the abundance of microcystins in the environment and the lack of proper guidelines or regulations and controlling technologies, extensive research is being done on the use of both conventional and emerging treatment technologies for the elimination of cyanobacterial toxins from water. Of all the technologies that have been explored so far for microcystin removal, $\mathrm{Ti}{\mathrm{O}}_2$ photocatalysis has been studied more extensively with respect to degradation efficiency, toxicity studies, and identification of intermediates. All the technologies explored have their own advantages and limitations, and choosing the most efficient, safest, and cost-effective technology should be done on a case-per-case basis.

This editorial does not necessarily reflect USEPA policy. Mention of trade names or commercial products does not constitute endorsement.