Environ. Sci. Technol. 2007, 41, 4397-4404
Selective Oxidation of Key Functional Groups in Cyanotoxins during Drinking Water Ozonation G R E T C H E N D . O N S T A D , ‡,† SABINE STRAUCH,‡ JUSSI MERILUOTO,§ GEOFFREY A. CODD,| AND U R S V O N G U N T E N * ,‡,⊥ Swiss Federal Institute of Aquatic Science and Technology, Eawag, Ueberlandstrasse 133, P.O. Box 611, CH-8600 Duebendorf, Switzerland, Department of Biochemistry and Pharmacy, Åbo Akademi University, BioCity, Tykisto¨katu 6A, 20520 Turku, Finland, Division of Environmental and Applied Biology, College of Life Sciences, University of Dundee, Dundee DD1 4HN, Scotland, U.K., and Institute of Biogeochemistry and Pollutant Dynamics, ETH Zurich, 8092 Zurich, Switzerland
Chemical kinetics were determined for the reactions of ozone and hydroxyl radicals with the three cyanotoxins microcystin-LR (MC-LR), cylindrospermopsin (CYN) and anatoxin-a (ANTX). The second-order rate constants (kO3) at pH 8 were 4.1 ( 0.1 × 105 M-1 s-1 for MC-LR, ∼3.4 × 105 M-1 s-1 for CYN, and ∼6.4 × 104 M-1 s-1 for ANTX. The reaction of ozone with MC-LR exhibits a kO3 similar to that of the conjugated diene in sorbic acid (9.6 ( 0.3 × 105 M-1 s-1) at pH 8. The pH dependence and value of kO3 for CYN at pH > 8 (∼2.5 ( 0.1 × 106 M-1 s-1) are similar to deprotonated amines of 6-methyluracil. The kO3 of ANTX at pH > 9 (∼8.7 ( 2.2 × 105 M-1 s-1) agrees with that of neutral diethylamine, and the value at pH < 8 (2.8 ( 0.2 × 104 M-1 s-1) corresponds to an olefin. Secondorder rate constants for reaction with OH radicals (•OH), kOH for cyanotoxins were measured at pH 7 to be 1.1 ( 0.01 × 1010 M-1 s-1 for MC-LR, 5.5 ( 0.01 × 109 M-1 s-1 for CYN, and 3.0 ( 0.02 × 109 M-1 s-1 for ANTX. Natural waters from Switzerland and Finland were examined for the influence of variations of dissolved organic matter, SUVA254, and alkalinity on cyanotoxin oxidation. For a Swiss water (1.6 mg/L DOC), 0.2, 0.4, and 0.8 mg/L ozone doses were required for 95% oxidation of MC-LR, CYN, and ANTX, respectively. For the Finnish water (13.1 mg/L DOC), >2 mg/L ozone dose was required for each toxin. The contribution of hydroxyl radicals to toxin oxidation during ozonation of natural water was greatest for ANTX > CYN > MC-LR. Overall, the order of reactivity of cyanotoxins during ozonation of natural waters corresponds to the relative magnitudes of the second-order rate constants for their reaction with ozone and •OH. Ozone primarily attacks the structural moieties responsible for the toxic effects of MC-LR, CYN, * Corresponding author phone: +41 44 823 5270; fax: +41 44 823 5210; e-mail:
[email protected]. ‡ Swiss Federal Institute of Aquatic Science and Technology. § Åbo Akademi University. | University of Dundee. ⊥ Institute for Biogeochemistry and Pollutant Dynamics. † Current address: University of Washington, Department of Environmental & Occupational Health Sciences, Box 357234, Seattle, WA 98195. 10.1021/es0625327 CCC: $37.00 Published on Web 05/09/2007
2007 American Chemical Society
and ANTX, suggesting that ozone selectively detoxifies these cyanotoxins.
Introduction Cyanotoxins are currently discussed as “emerging contaminants” in the water industry, even though their producers, namely, cyanobacteria, have been present in the natural environment for more than 3.5 billion years (1). Cyanobacteria were pioneers as primary producers of organic matter and likely responsible for the first production of oxygen in the Earth’s early atmosphere. Early recognition of the occurrence and adverse health effects of cyanotoxins in Denmark, Australia, and Poland is apparent in several nineteenth century publications (2). More recently, insufficient oxidation of microcystin-LR (MC-LR) in the chlorinated drinking water supply of a haemodialysis clinic in Brazil led to the death of 52 patients in 1996 (3). The Palm Island Mystery disease of 1979, a severe outbreak of hepatoenteritis in an Australian Aboriginal community, has been attributed to the release of the cell-bound cyanotoxin cylindrospermopsin (CYN) in a water supply treated with copper sulfate (4). Furthermore, occurrence of anatoxin-a (ANTX) was found to be responsible for canine neurotoxicosis in Scotland in 1990 (5). Occurrence at elevated concentrations (>1 µg/L) is more common for MC-LR (and >80 MC variants) and for CYN than for ANTX. Of these cyanotoxins, only MC-LR has a provisional guideline value of 1 µg/L (free plus cell-bound) in drinking water as set by the World Health Organization (6). Additional toxicological studies are necessary to set such guidelines for CYN and ANTX. The conjugated double bonds in the side chain ADDA (3-amino-9-methoxy-2,6,8-trimethyl-10-phenyldeca-4,6-dienoic acid) of MC-LR (see Figure 1) contribute to the toxicity of MCs. MCs are readily transported into liver cells (hepatocytes) where they inhibit several protein phosphatases and, in acute poisonings, cause pooling of blood in the liver and death in mammals due to hemorrhagic shock (7-9). Primarily regarded as hepatotoxins, MCs can also cause developmental abnormalities, tumor promotion, and neurological effects (2, 10). Cylindrospermopsin, a planar-shaped alkaloid, has been characterized as a hepatotoxin, cytotoxin, and genotoxin (2, 10). The side chain uracil on CYN binds to RNA to suppress protein synthesis and act as a hepatotoxin, or it binds to DNA to promote DNA strand breakage as a genotoxin. The neurotoxic alkaloid ANTX is a neuromuscular blocking agent which simulates acetylcholine to overstimulate respiratory muscles and causes death by respiratory paralysis (2, 10). A range of physicochemical, antibody- and enzyme-based methods are available to detect and quantify these cyanotoxins relative to their toxic functions (11). To protect drinking water consumers, control of cyanotoxins can be approached at the source or during drinking water treatment. Since many cyanobacteria form buoyant blooms on the surface of the reservoir, it is sometimes sufficient to choose an intake point far below the surface; however, there are exceptions where cyanobacterial communities can survive at depth near stratification of a lake or reservoir. The majority of the total pool of cyanotoxins (at least for MC or ANTX) is retained within the cyanobacterial cells, and once these are broken (by chemical/physical stress or age), the cyanotoxins are released in the water and are much more difficult to remove (12). The first step in treatment of cyanotoxins should be the removal of intact cells. Conventional treatment by coagulation and flocculation VOL. 41, NO. 12, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4397
FIGURE 1. Structures of target cyanotoxins, model compounds, and proposed sites of ozone attack. sometimes results in release of cell-bound toxins by physical stress. Preoxidation of raw water with chlorine, chlorine dioxide, permanganate, or ozone has also been shown to lyse the cell walls of cyanobacteria by chemical/physical stress, releasing intracellular toxins in postfiltration and insufficiently oxidizing extracellular toxins (13, 14). Nonetheless, intact cyanobacterial cells should be removed in the early stages of treatment, followed by removal of extracellular dissolved toxins by polishing treatment such as ozonation and granular-activated carbon (GAC) filtration (12). Relative to other water treatment oxidants such as chlorine and chlorine dioxide (15, 16), ozone has not been quantitatively characterized in its ability to oxidize dissolved cyanotoxins. Below pH 8, chlorine is a very feasible option for oxidative treatment of dissolved MC-LR. Acero et al. (15) measured half-lives of MC-LR of 6-25 min (from pH 6 to 8, 20 °C) for a chlorine concentration of 1 mg/L. When chlorine is applied to ammonia-containing waters, the resulting monochloramine is much less potent, with a measured halflife of MC-LR exceeding 14 h at pH 8 (15). Similarly, Kull et al. (16) measured the half-life of dissolved MC-LR in the presence of 1 mg/L chlorine dioxide, which ranged from 11 to 13 h (pH 6-8), proving chlorine dioxide to be an impractical solution for MC-LR because of the long contact time of chlorine dioxide required for oxidation. Previous studies of ozone applications made valuable qualitative comparisons but were often site-specific and did not adequately quantify rate constants (13, 17-20). In comparison of different oxidants for the elimination of MCLR, Rositano et al. (21) found ozone plus hydrogen peroxide > ozone . chlorine > hydrogen peroxide > permanganate. The reaction between ozone and MC-LR was so rapid that the rate constant could not be determined dynamically by Rositano et al. (21). Other studies uncovered that natural organic matter (NOM) had the greatest influence on MC-LR oxidation by ozone, being tied to ozone demand. Complete toxin oxidation was achievable when the ozone demand was satisfied, and ozone residual was measurable following ozonation (18, 19, 22). Hall et al. (20) found that ozone > permanganate > chlorine dioxide for oxidative removal of 4398
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 12, 2007
MC-LR and that permanganate > ozonation for ANTX. Ozone was identified as a good oxidant for elimination of ANTX (20-22), but it has not been evaluated for elimination of CYN. Ozone has great potential for oxidation of the double bonds and amine moieties found in MC-LR, CYN, and ANTX (23-25). We have chosen to highlight the compounds in Figure 1 for their reactivity with ozone and similarity to target functional groups on the cyanotoxins. Sorbic acid emulates the ADDA side chain of MC-LR, and 2-acetoamidoacrylic acid mirrors the rare amino acid Mdha in the peptide ring of MC-LR. Acetylcholine and ANTX both contain substituted amine and carbonyl moieties. The ring-locked guanidine structure of creatinine represents the same moiety in CYN, and 6-methyluracil is the best model for the uracil side chain of CYN. In this study, we determined second-order rate constants for the reaction of cyanotoxins with ozone and hydroxyl radicals (•OH are always present during ozonation). This is essential for understanding and optimizing ozonation for oxidative removal of cyanotoxins in natural waters varying in water quality.
Materials and Methods Standards and Reagents. MC-LR was isolated and purified from Microcystis and Anabaena cultures. CYN and ANTX were isolated from Cylindrospermopsis raciborskii and Oscillatoria cultures, respectively. Isolation, purification, standardization, and handling of cyanotoxins are given in Supporting Information (SI) (11). All other standard compounds were obtained at g99% purity. All aqueous solutions were prepared with preozonated (3-4 mg/L ozone dose) deionized water (resistivity 18.2 MΩ cm), which no longer contained residual ozone. Stock solutions of aqueous ozone (1-1.6 mM) were prepared by bubbling O3-containing oxygen through icecooled deionized water (26). Analyses. High-performance liquid chromatographic analysis was performed on an Agilent 1100 series HPLC with UV/DAD (Agilent, Palo Alto, CA), 100 µL injection volume, 4 °C injection temperature, and a 125 mm C18 column (CC125/4 Nucleosil 100-5 C18, Macherey-Nagel, Du ¨ ren,
Germany). Detection wavelengths were absorption maxima, given in SI. The HPLC conditions and solvent gradient were adapted from previous methods (11) to simultaneously measure competitor compounds and avoid coelution with oxidation products. Natural water samples were analyzed for dissolved organic carbon (DOC) using a High TOC II analyzer (Elementar Analysensysteme, Hanau, Germany). Analysis of microcystin variants (MC-LR, -RR, and -YR) at the Lengg drinking water treatment plant, Zu¨rich, Switzerland, was conducted by online SPE/HPLC/DAD (see SI and ref 27), and represents the sum of free (extracellular) and cell-bound (intracellular) concentrations of MC variants. Competition Kinetics. A competition kinetic study was a valuable tool for determination of the second-order rate constants of dissolved cyanotoxins with ozone or •OH because it could be conducted at micromolar concentrations that allowed conservation of the toxins and yet still prevented the need for preconcentration prior to HPLC analysis. Measurement of ozone rate constants above 1000 M-1 s-1 and •OH rate constants in general was conducted using competition kinetics (28, 29), according to
( ) ( )
ln
[T(n)] [C(n)] kT ) ln [T(0)] [C(0)] kC
(1)
where T is the toxin, C is the competitor, n is the specific oxidant dose, kT is the second-order rate constant for the toxin-oxidant reaction, and kC is the second-order rate constant for the competitor-oxidant reaction, assuming 1:1 stoichiometry with the oxidant. In most cases, the toxin and competitor were present at equimolar concentrations (1 µM with toxins and 1-10 µM for intercomparison of competitors) in phosphate-buffered solutions (10 mM). Determination of kO3,T required the addition of an •OH scavenger (50 mM tertbutanol, t-BuOH). Dilute ozone stock solutions (4-20 µM) were used for dosing competition kinetics experiments (see SI). The ozone concentration in the dilute stock solution was monitored at 258 nm ( ) 3000 M-1 cm-1) in a UV spectrophotometer or by dosing of parallel solutions of transcinnamic acid and detection of benzaldehyde production by HPLC/DAD, 254 nm (1 mol of ozone reacts with 1 mol of cinnamic acid to produce 1 mol of benzaldehyde (30)). Concentrations of toxins and competitors were determined by HPLC/DAD. The relationship for the pH dependency of the apparent rate constants was adapted from Hoigne´ and Bader (28)
kapp,O3 ) kdb + (1 - R) kHT+ + R kT
R)
1 (2) (1 + 10pKa-pH)
where R ) [T]/[T]tot or the degree of HT+ dissociation, [T]tot ) [T] + [HT+], kdb ) second-order rate constant of a double bond (non-pH-dependent species), kHT+ ) second-order rate constant for the protonated form of toxin, and kT ) secondorder rate constant for the deprotonated form of toxin. Second-order rate constants (k) were determined with 95% confidence intervals (see SI). •OH Oxidation. For determination of second-order rate constants with •OH, kOH,T, γ radiolysis was employed to generate the •OH; para-chlorobenzoic acid (pCBA) was used as a competitor, and kOH,T values were calculated using eq 1 (see SI and ref 29). Toxins and pCBA were present at equimolar concentrations (5 µM) in pH 7 phosphate-buffered solutions (5 mM). The kOH values for the toxins were not expected to vary appreciably with pH. Natural Water Samples. Samples from Lake Zu ¨ rich were taken 30 m below the surface at the inlet to the Lengg drinking water treatment plant, Zu ¨ rich, Switzerland. Grab samples were taken from the outlet of Lake Greifensee, at the headwaters of the Glatt River, Switzerland. The lake waters
were filtered through 0.45 µm filters (cellulose nitrate, Sartorius AG, Go¨ttingen, Germany) and analyzed for DOC, alkalinity, and ozone consumption. Samples from Lake A¨ yho¨nja¨rvi, Rauma, Finland, were collected from the input to the drinking water treatment plant, filtered through 1.2 µm glass fiber filters (GF/A, Whatman, Maidstone, England), and stored in a plastic bottle for shipping on ice overnight to EAWAG. Upon arrival, the samples were filtered through an additional 0.45 µm filter. All samples were stored at 4 °C and were used within 3 months of collection. The alkalinity of diluted water from Lake A¨ yho¨nja¨rvi was adjusted with the addition of bicarbonate (804 mg of NaHCO3 added to 2.3 L of LA water, added slowly to keep solution below pH 11). Batch Ozonation of Natural Waters. Natural waters were buffered to pH 8 (10 mM borate) with sodium tetraborate decahydrate (instead of phosphate, to avoid precipitation of calcium phosphate) and were fortified with pCBA (0.5 µM) and one toxin (0.15 µM MC-LR, 0.4 µM CYN, or 0.9 µM ANTX). The 2 L sample was split, and one liter was ozonated in the presence of an •OH scavenger (50 mM t-BuOH). Aliquots (30 mL) were dosed with an ozone stock solution to yield 2-60 µM O3 with mixing in amber vials at 20 °C. Parallel solutions of cinnamic acid were used to confirm ozone dosing by monitoring benzaldehyde concentration produced in solution. Samples were transferred to amber HPLC vials for analysis after 30 min. The fraction of the toxin oxidized by •OH was calculated as (23) ln( k ∫[OH]dt [pCBA] ) ) [T] ∫[OH]dt + k ∫[O ]dt k ln( [T] ) [pCBA]t
kOH,T
fOH ) kOH,T
OH,T
0
O3,T
3
(3)
t
OH,pCBA
0
fO3 ) 1 - fOH
(4)
Ozone consumption kinetics (without toxins) were determined by dosing buffered 250 mL natural water samples with ozone stock solution and, after brief mixing, collecting samples via a dispenser at time points from 15 s to 60 min. Ozone residual was measured by the Indigo method (26) and confirmed using cinnamic acid.
Results and Discussion Chemical Kinetics. As shown in Figure 1, the reactivity of the target cyanotoxins with ozone is dominated by two functional groups: double bonds and secondary amines. Table 1 shows the magnitude of the rate constants for these compounds, which required application of competition kinetics methods. Second-order rate constants for reaction of ozone with MC-LR and the olefin model compounds sorbic acid and 2-acetamidoacrylic acid (Figure 1) were determined relative to cinnamic acid. Cinnamic acid was used as the primary competitor for these compounds because it has an apparent rate constant of the same magnitude, ∼105 M-1 s-1, at pH 7 as expected for the ADDA group (Table S1) (30). Salicylic acid was chosen as the primary competitor for CYN and ANTX because of its smaller reactivity with ozone and the strong pH dependency of the rate constants (28). Rate constants for CYN and ANTX were confirmed relative to cinnamic acid at high pH (pH 8, Figure 2a and Figures S1S2). Competition kinetics results plotted in Figure 2 exhibited excellent linearity, and the apparent kO3,T (Figure 2a) and kOH,T (Figure 2b) values determined from eq 1 from the same competitor were very reproducible ( 7 were determined using the apparent rate constant of salicylate at pH 7, thus providing a conservative estimate of kT. Measurements relative to cinnamic acid were not used to calculate the rate constants for ANTX in Table 1 because the salicylate data set was larger (N ) 12 vs 4) and exhibited greater confidence, and the values of kT determined with salicylate and cinnamic acid differ considerably at pH 7-8 (Figure S2a). While the ozone-ANTX reactivity is better defined at low pH, kdb is still 2 orders of magnitude lower than that of cyclohexene (36). This might be explained by the electron-withdrawing effect of the R-ketone substituent on the double bond. A similar reduction in the rate constant was observed for carboxylic acid groups
in fumaric acid, kO3 ≈ 6 × 103 M-1 s-1 at pH 2 (28). Although the ozone reactivity of the model structure acetylcholine (Figure 1) was not measured in this study, it is not expected to react because of the carboxyl group and the positively charged quaternary ammonium moiety. Figure 4 superimposes the pH profiles for the kO3,T values of dissolved MC-LR, CYN, and ANTX. The corresponding half-life times are given for an ozone concentration of 1 mg/ L. Dotted lines represent extrapolated curves beyond experimental data (solid lines). Within the pH range of natural water resources (double arrow, Figure 4), MC-LR has the shortest half-life upon reaction with ozone, while the order of CYN and ANTX oxidation by ozone varies dramatically with pH. Double-bond reactivity with ozone dominates for MC-LR at all pH values and for ANTX at pH < 8. Clearly, oxidation and detoxification of amine moieties in CYN and ANTX are favored under the high pH conditions of ozonation. Although oxidation byproducts of MC-LR, CYN, and ANTX were not investigated in this study, the decrease of the UV signal as measured by HPLC/DAD was indicative of oxidized double bonds and conjugated systems present in these compounds and, therefore, adds support to our hypothesis that ozone attacks these functional groups. Hoeger (37) identified ozonation products of MC-LR which confirmed oxidation of the double bonds in the ADDA side chain, but more product identification is needed to elucidate mechanisms of CYN and ANTX oxidation by ozone. Table 1 also shows kOH,T values for dissolved cyanotoxins. The order MC-LR > CYN > ANTX corresponds roughly to the size of the molecule and the number of hydrogens that can be abstracted by •OH (see Figure 1). Relative to other environmental contaminants, the magnitude of kOH for MCLR is equivalent to chloroethene, CYN to sulfamethoxazole, and ANTX to atrazine (23). During ozonation of natural waters, a typical Rct value on the order of 10-8 can be observed, where Rct ≈ [•OH ]/[O3] (23). An initial ozone concentration of 1 mg/L results in a transitory [•OH ]ss of 2 × 10-13 M. For these conditions, the half-life times of •OH-toxin reactions can be estimated to be 5 min for MC-LR, 10 min for CYN, and 20 min for ANTX. Generally, the attack of •OH on each toxin is nonselective and therefore not as efficient as ozone in removing toxicity. Toxin Oxidation in Natural Waters. To test the kinetics determined in clean water systems under realistic conditions, experiments were conducted in lake waters from Switzerland and Finland. Table 2 shows the relevant water quality parameters of the Swiss and Finnish lake waters. As shown VOL. 41, NO. 12, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4401
TABLE 2. Water Quality Parameters of Natural Waters source
pH
Lake Zu¨ rich, 7.9 Switzerland Lake Greifensee, 8.4 Switzerland Lake A ¨ yho¨ nja¨ rvi, 7.2 Finland
DOC UV254 SUVA alkalinity as (mg/L) (m-1) (m-1 L/mg) HCO3- (mM) 1.6
2.6
1.6
2.6
3.6
6.0
1.7
3.8
13.1
35.2
2.7
0.43
FIGURE 5. Ozone consumption by the natural water matrix in three surface waters dosed with 3 mg/L of O3 (62.5 µM) at 20 °C and pH 8. Diluted A2 yho1 nja1 rvi water was adjusted to the DOC and alkalinity of Lake Greifensee. Inset: Comparison of measured ozone exposures (left axis) to CT requirements for E. coli, G. lamblia, and virus inactivation (right axis). Key: O, Lake A2 yho1 nja1 rvi water; 4, Lake Greifensee water; - -, diluted A2 yho1 nja1 rvi water; (, Lake Zu1 rich water. in Figure 5, ozone stability was highest in water from Lake Zu ¨ rich, followed by Lake Greifensee and Lake A¨ yho¨nja¨rvi (3 mg/L ozone dose). When the DOC and alkalinity (bicarbonate addition) in Lake A¨ yho¨nja¨rvi were adjusted to the conditions of Lake Greifensee, the two waters exhibited similar ozone consumption kinetics (dotted line in Figure 5). Therefore, for these two fairly eutrophic lake waters, the trend of ozone consumption appears to be primarily a function of NOM concentration more than NOM composition, which is probably similar because of its autochthonous origin. Consumption of ozone by the natural matrix can impede disinfection. The inset in Figure 5 shows the resulting ozone exposures (integration of ozone residual vs time). The ozone exposures required to obtain at least 99.9% inactivation of Escherichia coli, Giardia lamblia, and viruses (38, 39) are also indicated. Inactivation of E. coli is achievable within the first 15 s in all waters. Inactivation of viruses occurs immediately in the Swiss waters, but only after 3 min in the Finnish water. While Giardia inactivation can be obtained within the first 25 s for the Swiss waters, it is not possible over 60 min in the Finnish water. Consequently, the natural matrix can consume ozone to the point of limiting the disinfection and oxidation capabilities of ozone. Because of the magnitude (>500 M-1 s-1) of the apparent kO3,T at pH 8, which does not allow dynamically monitored experiments, dose experiments were conducted in each natural water to determine the ozone doses necessary for toxin oxidation. As evident from Figure 5, the natural matrix of Lake A¨ yho¨nja¨rvi consumes ozone much more quickly than that of Lake Zu ¨ rich, and therefore, more ozone was required to completely oxidize the toxins in Lake A¨ yho¨nja¨rvi. Figures 6a-c show the oxidation of toxins as a function of the ozone dose in waters from Lake Zu ¨ rich (LZ) and Lake A¨ yho¨nja¨rvi (LA), revealing the difference between toxin oxidation by 4402
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 12, 2007
FIGURE 6. Oxidation of toxins by ozone and •OH in two different natural waters: LA, Lake A2 yho1 nja1 rvi; LZ, Lake Zu1 rich (n ) 2). Bars show toxin remaining after oxidation by ozone only (presence of t-BuOH). Lines show toxin remaining after oxidation by ozone and hydroxyl radicals (O3 + •OH, without t-BuOH, natural conditions). Error bars indicate the standard deviation: MC-LR, microcystin-LR; CYN, cylindrospermopsin; ANTX, anatoxin-a. ozone alone (using an •OH scavenger, t-BuOH, bars) and toxin oxidation by ozone and •OH (generated by ozone decomposition (23), lines). The difference is noticeable in Lake Zu ¨ rich water for a small ozone dose of 0.1 mg/L. Under these conditions, ozone is rapidly transformed into OH radicals through direct electron-transfer reactions (40). Therefore, the addition of an OH radical scavenger leads to a significant loss of the overall oxidant exposure (O3 and •OH). In Lake Zu ¨ rich water, an ozone dose of 0.2 mg/L was needed to oxidize >95% MC-LR, 0.4 mg/L ozone for >95% CYN, and 0.8 mg/L ozone for >95% ANTX. At lower ozone doses, the difference was evident between toxin oxidation by ozone and •OH or by ozone alone, although the required dose for toxin oxidation was the same for both cases. The ozone dose necessary for toxin oxidation in water from Lake A¨ yho¨nja¨rvi was g2 mg/L, and •OH had very little influence on toxin oxidation because of the high •OH radical scavenging rate caused by the high NOM concentration. However, when the DOC and alkalinity of Lake A¨ yho¨nja¨rvi were adjusted to the conditions of Lake Greifensee, the toxin oxidation charts were similar, reinforcing the dependence of ozone consumption and toxin oxidation on NOM concentration in these eutrophic waters, but different enough to reveal the influence of NOM composition (Figure S3). Rositano et al. (22) also
Oxidative Treatment of Cyanotoxins. Higher doses of ozone are required for removal of cell-bound toxins than are needed for extracellular dissolved toxins because ozone is also consumed during cell lysis (20). The cells should be physically removed prior to ozonation in the treatment train via dissolved-air-flotation or coagulation and filtration, so that ozone can selectively oxidize the extracellular toxins. However, if the toxin load is low, as shown in Figure 8, preozonation may be sufficient to oxidize the toxins completely. Preozonation (1 mg/L of O3) plus rapid sand filtration, employed by the Lengg drinking water treatment plant in Zu ¨ rich, Switzerland, can satisfy disinfection requirements and effectively oxidize MC-LR and microcystin variants to below 20 ng/L which is far lower than the provisional drinking water guideline of 1µg/L recommended by the World Health Organization.
Acknowledgments FIGURE 7. Toxin oxidation in Lake Greifensee (pH 8) as a function of ozone dose, where fOH is the fraction of the toxin oxidized by •OH according to eqs 3 and 4 (n ) 2): MC-LR, microcystin-LR; CYN, cylindrospermopsin; ANTX, anatoxin-a.
MC-LR was purified by Lisa Spoof, Åbo Akademi University (AAU). CYN and ANTX were purified by James Metcalf and Marianne Reilly, University of Dundee. Samples from Lake A¨ yho¨nja¨rvi were collected by Olli Sjo¨vall, AAU. This study was performed under the framework of TOXIC: Barriers Against Cyanotoxins, European Union Project EVI1-CT-200200107. Financial support was provided by the Swiss Federal Department for Education and Science (Bundesamt fu ¨r Bildung und Wissenschaft). Microcystin occurrence and treatment data was provided by Marcel Leemann, Zu¨rich Water Works (Wasserversorgung). We appreciate the expert technical advice and laboratory assistance provided by Marc Huber, Michael Dodd, Marc-Olivier Buffle, Manuel SanchezPolo, and Elisabeth Salhi.
Supporting Information Available FIGURE 8. Treatment of microcystin-containing water from Lake Zu1 rich in a full-scale treatment plant (Lengg, Water Supply, Zurich). Microcystin occurrence in January 2004. MC-equivalents ) sum of MC-variants (LR, YR, RR, RR-des). observed that NOM concentration (4.6-15.5 mg/L DOC) had a greater influence on ozone consumption and toxin oxidation than NOM composition (SUVA 1.4-2.1 m-1 L/mg) in selected Australian lake waters and showed similar trends of removal of MC-LR > ANTX by ozonation in these waters. In this study, Lake Greifensee served as a matrix in which the oxidation of the three toxins was compared in the absence of an •OH scavenger. The relative reactivity of the toxins with ozone (and •OH), MC-LR > CYN > ANTX at pH 8 and 20 °C, is clearly reflected in Figure 7. The fraction of the toxins reacting with •OH (fOH) is also shown (calculated by eq 3). For an ozone dose of 0.12 mg/L, fOH can be as high as 38% (for ANTX). However, it is typically less than 20% for higher ozone doses. The role of oxidation by •OH decreases in the order ANTX > CYN > MC-LR. Overall, the degree of toxin oxidation depends on the relative rate constants for reaction with ozone and the ozone consumption by the natural water matrix. While the experiments were allowed to run to completion (∼30 min) with ozone completely depleted, the required ozone doses for toxin oxidation cannot be used to calculate ozone CT values for elimination of each toxin (in contrast to other studies (15, 41), where oxidant exposure CT values could be determined dynamically for elimination of MC-LR by chlorine and chlorine dioxide). However, using the apparent second-order rate constants for the reaction of ozone with the toxins and the ozone CT requirements for inactivation of E.coli, viruses and G. lamblia, we can predict that >99.9% of the toxins will be oxidized in all cases satisfied for disinfection (e.g., 3 log inactivation of G. lamblia).
Further description of toxin preparation and analysis, as well as competition kinetics methods and results, are presented together with toxin oxidation data. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Schopf, J. W. In Contributions in Phycology; Prasad, A. K. S. K., Nienow, J. A., Rao, V. N. R., Eds.; J. Cramer: Berlin, 1996; pp 13-32. (2) Codd, G. A.; Lindsay, J.; Young, F. M.; Morrison, L. F.; Metcalf, J. S. In Harmful Cyanobacteria; Huisman, J., Matthijs, H. C. P., Visser, P. M., Eds.; Springer: Dordrecht, The Netherlands, 2005; pp 1-23. (3) Jochimsen, E. M.; Carmichael, W. W.; An, J.; Cardo, D. M.; Cookson, S. T.; Holmes, C. E. M.; Antunes, M. B. C.; Melo Filho, D. A.; Lyra, T. M.; Barreto, V. S. T.; Azevedo, S. M. F. O.; Jarvis, W. R. Liver failure and death after exposure to microcystins at a hemodialysis center in Brazil. New Engl. J. Med. 1998, 338, 873-878. (4) Griffiths, D. J.; Saker, M. L. The Palm Island mystery disease 20 years on: A review of research on the cyanotoxin cylindrospermopsin. Environ. Toxicol. 2003, 18, 78-93. (5) Edwards, C.; Beattie, K. A.; Scrimgeour, C. M.; Codd, G. A. Identification of anatoxin-a in benthic cyanobacteria (bluegreen algae) and in associated dog poisonings at Loch Insh, Scotland. Toxicon 1992, 30, 1165-1175. (6) Guidelines for drinking water quality, 3rd ed.; World Health Organization: Geneva, 2004; Vol. 1. (7) Runnegar, M.; Berndt, N.; Kaplowitz, N. Microcystin uptake and inhibition of protein phosphatases: effects of chemoprotectants and self-inhibition in relation to known hepatic transporters. Toxicol. Appl. Pharmacol. 1995, 134, 264272. (8) Mackintosh, C.; Beattie, K. A.; Klumpp, S.; Cohen, P.; Codd, G. A. Cyanobacteria microcystin-LR is a potent and specific inhibitor of protein phosphatases 1 and 2A from both mammals and higher plants. FEBS Lett. 1990, 264, 187-192. (9) Hastie, C. J.; Borthwick, E. B.; Morrison, L. F.; Codd, G. A.; Cohen, P. T. W. Inhibition of several protein phosphatases by a nonVOL. 41, NO. 12, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4403
(10) (11) (12) (13)
(14) (15) (16)
(17) (18)
(19)
(20) (21) (22) (23) (24) (25)
(26) (27)
covalently interacting microcystin and a novel cyanobacterial peptid, nostocyclin. Biochim. Biophys. Acta 2005, 1726, 187193. Falconer, I. R.; Humpage, A. R. Health risk assessment of cyanobacterial (blue-green algal) toxins in drinking water. Int. J. Environ. Res. Public Health 2005, 2, 43-50. TOXIC. Cyanobacterial Monitoring and Cyanotoxin Analysis, series B; Meriluoto, J., Codd, G. A., Eds.; Abo Akademi University Press: Abo, Finland, 2005; Vol. 65. Falconer, I. R. Cyanobacterial Toxins of Drinking Water Supplies: Cylindrospermopsins and Microcystins; CRC Press: Boca Raton, FL, 2005. Schmidt, W.; Willmitzer, H.; Bornmann, K.; Pietsch, J. Production of drinking water from raw water containing cyanobacterias Pilot plant studies for assessing the risk of microcystin breakthrough. Environ. Toxicol. 2002, 17, 375-385. Lam, A. K. Y.; Prepas, E. E.; Spink, D.; Hrudey, S. E. Chemical control of hepatotoxic phytoplankton bloomssImplications for human health. Water Res. 1995, 29, 1845-1854. Acero, J. L.; Rodrı´guez, E.; Meriluoto, J. Kinetics of reactions between chlorine and the cyanobacterial toxins microcystins. Water Res. 2005, 39, 1628-1638. Kull, T. P. J.; Backlund, P. H.; Karlsson, K. M.; Meriluoto, J. Oxidation of the cyanobacterial hepatotoxin microcystin-LR by chlorine dioxide: Reaction kinetics, characterization, and toxicity of reaction products. Environ. Sci. Technol. 2004, 38, 6025-6031. Shawwa, A. R.; Smith, D. W. Kinetics of microcystin-LR oxidation by ozone. Ozone: Sci. Eng. 2001, 23, 161-170. Newcombe, G.; Cook, D.; Brooke, S.; Ho, L.; Slyman, N. Treatment options for microcystin toxins: Similarities and differences between variants. Environ. Technol. 2003, 24, 299308. Hoeger, S. J.; Dietrich, D. R.; Hitzfeld, B. C. Effect of ozonation on the removal of cyanobacterial toxins during drinking water treatment. Environ. Health Perspect. 2002, 110, 11271132. Hall, T.; Hart, J.; Croll, B.; Gregory, R. Laboratory-scale investigations of algal toxin removal by water treatment. Water Environ. J. 2000, 14, 143-149. Rositano, J.; Nicholson, B.; Pieronne, P. Destruction of cyanobacterial toxins by ozone. Ozone: Sci. Eng. 1998, 20, 223238. Rositano, J.; Newcombe, G.; Nicholson, B.; Sztajnbok, P. Ozonation of NOM and algal toxins in four treated waters. Water Res. 2001, 35, 23-32. von Gunten, U. Ozonation of drinking water: Part I. Oxidation kinetics and product formation. Water Res. 2003, 37, 14431467. Hoigne´, J.; Bader, H. Rate constants of reactions of ozone with organic and inorganic compounds in water. I. Non-dissociating organic compounds. Water Res. 1983, 17, 173-183. Lawton, L. A.; Robertson, P. K. J. Physico-chemical treatment methods for the removal of microcystins (cyanobacterial hepatotoxins) from potable waters. Chem. Soc. Rev. 1999, 28, 217-224. Bader, H.; Hoigne´, J. Determination of ozone in water by the indigo method. Water Res. 1981, 15, 449-456. Water quality-Determination of microcystins-Method using solid phase extraction (SPE) and high performance liquid
4404
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 12, 2007
(28) (29) (30)
(31)
(32)
(33) (34) (35)
(36) (37)
(38) (39) (40) (41)
(42)
chromatography (HPLC) with ultraviolet (UV) detection; ISO/ DIS 20179; International Organization for Standardization: Geneva, 2004. Hoigne´, J.; Bader, H. Rate constants of reactions of ozone with organic and inorganic compounds in water. II. Dissociating organic compounds. Water Res. 1983, 17, 184-194. Huber, M. M.; Canonica, S.; Park, G.-Y.; von Gunten, U. Oxidation of pharmaceuticals during ozonation and advanced oxidation processes. Environ. Sci. Technol. 2003, 37, 1016-1024. Leitzke, A.; Reisz, E.; Flyunt, R.; von Sonntag, C. The reactions of ozone with cinnamic acids: formation and decay of 2-hydroperoxy-2-hydroxyacetic acid. J. Chem. Soc., Perkin Trans. 2 2001, 5, 793-797. Theruvathu, J. A.; Flyunt, R.; Aravindakumar, C. T.; von Sonntag, C. Rate constants of ozone reactions with DNA, its constituents and related compounds. J. Chem. Soc., Perkin Trans. 2001, 2, 269-274. Hilal, S.; Karickhoff, S. W.; Carreira, L. A. A rigorous test for SPARC’s chemical reactivity models: Estimation of more than 4300 ionization pKa’s. Quant. Struc.-Act. Relat. 1995, 14, 348355. Carreira, L. A. SPARC Online Calculator; Department of Chemistry, University of Georgia: Athens, GA, 2004; http:// ibmlc2.chem.uga.edu/sparc. Koskinen, A. M. P.; Rapoport, H. Synthetic and conformational studies on anatoxin-a: A potent acetylcholine agonist. J. Med. Chem. 1985, 28, 1301-1309. Mun ˜ oz, F.; von Sonntag, C. The reaction of ozone with tertiary amines including the complexing agents nitrilotriacetic acid (NTA) and ethylenediaminetetraacetic acid (EDTA) in aqueous solution. J. Chem. Soc., Perkin Trans. 2000, 2, 2029-2033. Keady, H. D.; Kuo, C. H. A data acquisition system for rapid kinetic experiments. Chem. Eng. Commun. 1983, 23, 291304. Hoeger, S. J. Problems during drinking water treatment of cyanobacterial-loaded surface waters: Consequences for human health. PhD. Dissertation, Biologie, Naturwissenschaften, Universita¨t Konstanz: Konstanz, Germany, 2003. Disinfection profiling and benchmarking. Technical guidance manual; U.S. EPA 816-R-03-004; U.S. Environmental Protection Agency: Washington, DC, 2003; Appendix B, CT tables. von Gunten, U. Ozonation of drinking water: Part II. Disinfection and by-product formation in presence of bromide, iodide or chlorine. Water Res. 2003, 37, 1443-1467. Buffle, M.-O.; von Gunten, U. Phenols and amine induced HO radical generation during the initial phase of natural water ozonation. Environ. Sci. Technol. 2006, 40, 3057-3063. Kull, T. P. J.; Sjo¨vall, O. T.; Tammenkoski, M. K.; Backlund, P. H.; Meriluoto, J. A. O. Oxidation of the cyanobacterial hepatotoxin microcystin-LR by chlorine dioxide: Influence of natural organic matter. Environ. Sci. Technol. 2006, 40, 1504-1510. Lu ¨ ck, E.; Jager, M.; Raczek, N. In Ullmann’s Encyclopedia of Industrial Chemistry, 6th ed.; Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim, Germany, 2005.
Received for review October 21, 2006. Revised manuscript received February 9, 2007. Accepted March 23, 2007. ES0625327
