Environ. Sci. Technol. 2010, 44, 7361–7368
Cyanotoxin Mixtures and Taste-and-Odor Compounds in Cyanobacterial Blooms from the Midwestern United States JENNIFER L. GRAHAM,* KEITH A. LOFTIN, MICHAEL T. MEYER, AND ANDREW C. ZIEGLER United States Geological Survey, Kansas Water Science Center, 4821 Quail Crest Place, Lawrence, Kansas 66049
Received March 19, 2010. Revised manuscript received August 10, 2010. Accepted August 11, 2010.
The mixtures of toxins and taste-and-odor compounds present during cyanobacterial blooms are not well characterized and of particular concern when evaluating potential human health risks. Cyanobacterial blooms were sampled in twentythree Midwestern United States lakes and analyzed for community composition, thirteen cyanotoxins by liquid chromatography/ mass spectrometry and immunoassay, and two taste-and-odor compounds by gas chromatography/mass spectrometry. Aphanizomenon, Cylindrospermopsis and/or Microcystis were dominant in most (96%) blooms, but community composition was not strongly correlated with toxin and taste-and-odor occurrence. Microcystins occurred in all blooms. Total microcystin concentrations measured by liquid chromatography/mass spectrometry and immunoassay were linearly related (rs ) 0.76, p < 0.01) and LC/MS/MS concentrations were lower than or similar to ELISA in most (85%) samples. Geosmin (87%), 2-methylisoborneol (39%), anatoxin-a (30%), saxitoxins (17%), cylindrospermopsins (9%), and nodularin-R (9%) also were present in these blooms. Multiple classes of cyanotoxins occurred in 48% of blooms and 95% had multiple microcystin variants. Toxins and taste-and-odor compounds frequently co-occurred (91% ofblooms),indicatingodormayserveasawarningthatcyanotoxins likely are present. However, toxins occurred more frequently than taste-and-odor compounds, so odor alone does not provide sufficient warning to ensure human-health protection.
Introduction Cyanobacteria cause many water-quality concerns, including potential production of toxins and taste-and-odor compounds. With multiple potential producers for most compounds, co-occurrence in mixed assemblage cyanobacterial blooms can be expected (1-4); however, few studies have characterized the mixtures of toxins and taste-and-odor compounds present during blooms. Toxin mixtures are of particular concern when evaluating potential human-health risks. Cyanotoxins are diverse and may be classified into three main groups based on toxic mechanism: hepatotoxins, such as cylindrospermopsins and microcystins; neurotoxins, such as anatoxins and saxitoxins; and dermatoxins, such as lyngbyatoxins (1). Many cyanotoxins have multiple variants * Corresponding author phone: 785-832-3511; fax: 785-832-3500; e-mail:
[email protected]. 10.1021/es1008938
Not subject to U.S. Copyright. Publ. 2010 Am. Chem. Soc.
Published on Web 09/10/2010
with a range of toxicities. For example, there are over 80 known microcystin variants (5) and toxicity differences among them vary by an order of magnitude (1). Such diversity complicates risk assessment and development of regulations for human-health protection. Regardless, several countries have set national standards or guidelines for cyanotoxins in drinking and recreational waters (2). Cyanotoxins currently are on the U.S. Environmental Protection Agency drinkingwater contaminant candidate list (6) and several states include cyanotoxins, typically microcystins measured by enzymelinked immunosorbent assay (ELISA), in their freshwater beach-monitoring programs (7). Most taste-and-odor events associated with cyanobacteria are caused by either geosmin or 2-methylisoborneol (MIB) (8). Unlike toxins, geosmin and MIB have no known effects on human health and there are no regulations for these compounds. Aesthetic issues associated with geosmin and MIB occur at low (∼0.01 µg/L) concentrations and remedial actions often are taken as soon as taste or odor is detected in a drinking-water supply (4). The purpose of this study was to determine co-occurrence of toxins and taste-and-odor compounds in cyanobacterial blooms. During August 2006, a cyanobacterial bloom from each of 23 lakes in the Midwestern U.S. was sampled. Measured compounds included 13 toxins in 6 classes (anatoxins, cylindrospermopsins, lyngbyatoxins, microcystins, nodularins, and saxitoxins) and 2 taste-and-odor compounds (geosmin and MIB).
Materials and Methods Sample Collection and Analysis. Twenty-three eutrophic lakes and reservoirs (herein referred to as lakes) in Iowa, northeastern Kansas, southern Minnesota, and northwestern Missouri with a known history of late-summer cyanobacterial blooms (9) were sampled once during August 7-11, 2006 (Figure 1; Table SI-S1 in Supporting Information). At each lake, a single grab sample was collected in a 1-L amber-glass bottle from a near-shore area where cyanobacteria were visible. Samples were collected right at the water surface to include any cyanobacterial accumulations or scums (10). All samples (n ) 23) were analyzed for chlorophyll, cyanobacterial community composition, total and dissolved toxins, and dissolved taste-and-odor compounds. Total taste-andodor was measured in a subset of 16 samples. A subsample for phytoplankton analysis was immediately collected in a 250-mL amber-glass bottle and preserved with glutaraldehyde. The remaining sample was shipped on ice overnight to the U.S. Geological Survey (USGS) Organic Geochemistry Research Laboratory (Lawrence, KS). Samples were processed immediately upon receipt at the laboratory (10). After gently inverting bottles 3-4 times to ensure homogenization, subsamples for total toxin and taste-andodor analysis were collected into 125-mL amber-glass bottles. Chlorophyll samples were collected on 0.45-µm glass-fiber filters. The 0.45-µm filtrate was used for dissolved-phase analyses; samples for toxin analysis were placed in LC/MS/ MS vials and samples for taste-and-odor analysis were placed into 60-mL amber-glass vials. Chlorophyll, toxin, and total taste-and-odor samples were stored frozen and dissolved taste-and-odor samples were stored at 4 °C until analysis. Phytoplankton samples were sent to BSA Environmental Services, Inc. (Beachwood, OH) for determination of relative abundance with identification to genus. Details on chlorophyll analysis are provided in Table SI-S1. Cyanobacterial cells were lysed by three sequential freeze-thaw cycles to allow determination of total toxin and taste-and-odor VOL. 44, NO. 19, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 7361
FIGURE 1. Location of lakes sampled and occurrence and relative abundance of microcystin variants as detected by LC/MS/MS (n ) 23). concentrations (11). After freeze-thaw, samples were filtered as described for dissolved-phase analyses. Geosmin and MIB were analyzed by solid-phase microextraction gas chromatography/mass spectrometry (SPME GC/MS; detection limit 0.005 µg/L) as described in Zimmerman et al. (12). Abraxis enzyme-linked immunosorbent assays (ELISA) were used to measure microcystins and nodularins (detection limit 0.1 µg/L; -adda specific), cylindrospermopsins (0.04 µg/L), and saxitoxins (0.02 µg/L). ELISA is nonspecific and reported concentrations potentially include multiple variants, deg7362
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 19, 2010
radates, and precursors; thus, the entire toxin class is referred to when discussing ELISA results. Specific toxins (anatoxin-a, cylindrospermopsin, deoxycylindrospermopsin, lyngbyatoxin-a, microcystin-LA, -LF, -LR, -LW, -LY, -RR, -YR, and nodularin-R) were measured by a direct-inject multianalyte liquid chromatography/tandem mass spectrometry (LC/MS/MS) method (11), developed based on previous research (13-15), with a minimum reporting limit of 0.01 µg/L. LC/MS/MS analysis is compound specific and concentrations of individual toxins are reported.
Total microcystin concentrations are reported as the sum of the individual microcystin variants measured by LC/MS/MS for which standards were available. For LC/MS/MS analysis, cyanotoxins were separated on a Shimadzu Prominence liquid-chromatograph system (Kyoto, Japan) and detected with an Applied Biosystems API 5000 tandem mass spectrometer (Foster City, CA) in multiple reaction-monitoring mode by electrospray positive and negative ionization. Source parameters were optimized for all measured cyanotoxins (Table SI-S2). Cyanotoxins were separated on a Waters Atlantis dC18 column (3.0 mm ×150 mm, 3-µm particle size, Milford, MA) preceded by a Waters Atlantis dC18 guard cartridge (3.9 mm ×20 mm, 3-µm particle size, Milford, MA) using a reverse-phase gradient at a total flow rate of 0.4 mL per minute; mobile phase A was aqueous 0.1% formic acid and mobile phase B was 100% methanol with 0.1% formic acid. An active and passive needle rinse (0.1% tetrahydrofuran in methanol) was used to prevent carryover from needle contamination. Samples were quantitated by standard addition. Unspiked samples were amended with an aqueous solution of simetone as an internal standard. Standard addition samples were amended with a solution containing all measured cyanotoxins. Additional LC/ MS/MS details, including standard sources, transitions, compound-dependent mass-spectrometry values, and separation gradients, are provided in Supporting Information Tables SI-S2 and SI-S3 and Loftin et al. (11). Precision for LC/MS/MS analysis of cyanotoxins was evaluated on a subset of samples (n ) 7). The mean percent relative standard deviation for individual cyanotoxins ranged from 0.37-20 (grand mean ) 11). There was no carry over in blank samples. Details on precision for LC/MS/MS analysis of cyanotoxins are given in Table SI-S4. Statistical Analyses. Cyanobacterial communities and associated cyanotoxins were defined using cluster analysis and multidimensional scaling (MDS) (16). All data were square-root transformed prior to the calculation of BrayCurtis similarity matrices. The square-root transformation down-weights the influence of the most common taxa on Bray-Curtis similarities, but is not so severe that rare taxa have the same influence as common taxa (16). The number of MDS restarts was set at 50, and minimum stress was set at 0.01. Differences in microcystin variant composition with respect to cyanobacterial community composition were determined using analysis of similarities (ANOISM) (16). Cluster, MDS, and ANOISM analyses were conducted using PRIMER 6.1.9. Relations between toxins and individual measures of the cyanobacterial community, toxins and tasteand-odor compounds, and microcystin concentrations measured by LC/MS/MS and ELISA were developed using Spearman-Rank correlation (17) in SAS 9.1. To directly compare microcystin concentrations with those measured by ELISA, all microcystin variant and nodularin-R concentrations from LC/MS/MS analysis were cross reactivity corrected and summed (11). All statistical analyses were nonparametric, with significance set at p < 0.05.
Results and Discussion Cyanobacterial Community Composition. A wide range of bloom conditions occurred in these eutrophic midcontinent lakes,withchlorophyllconcentrationsrangingfrom28-190 000 µg/L (median ) 190 µg/L, n ) 23). Cluster analysis and ordination by MDS indicate five general cyanobacterial community types, characterized by dominant genera: mixed assemblages of Anabaena, Aphanizomenon, and/or Microcystis (57% of blooms); Microcystis (17%); Cylindrospermopsis (17%); Anabaena (4%); and a mixed assemblage of Anabaena, Planktothrix, and Aphanocapsa (4%; Figure 2). All of these dominant genera have known toxin and/or taste-and-odor producing strains (Table 1), and with the exception of
FIGURE 2. Multidimensional scaling (MDS) of cyanobacterial bloom communities (n ) 23) in the Midwestern U.S. and the associated cyanotoxins. Communities are described based on the dominant genera in each community (>60% of overall cyanobacterial community composition). Solid lines indicate similarity of 50% based on cluster analysis. Microcystins were detected in all communities and presence of this toxin is not indicated. All other toxins and taste-and-odor compounds are indicated as follows: A ) anatoxin-a, C ) cylindrospermopsins, G ) geosmin, M ) 2-methylisoborneol, N ) nodularin-R, S ) saxitoxins. Aphanocapsa, have been linked to cyanotoxin poisonings worldwide (1-3). In addition to the six dominant genera, six other potential toxin and/or taste-and-odor producers were identified, with 3-9 potential producers (median ) 5) present in individual blooms (Tables 1 and SI-S5). Potential producers of anatoxins, microcystins, saxitoxins, and geosmin occurred in every bloom, and potential producers of cylindrospermopsins were present in all but one. MIB (56% of blooms) and lyngbyatoxin (43%) producers were less common. No known nodularin producers occurred. Although potential producers of measured compounds were common, conclusive determination of the producer(s) of specific compounds in mixed field populations without strain isolation and culture is difficult (4). Toxins. Microcystins are generally maintained intracellularly until natural death or lysis by management practices, such as application of algicides (1). Whether other toxins also are similarly maintained is not as well understood. In our study, the cyanotoxins were largely cell-bound. Dissolved concentrations never represented more than 30% of total concentrations (median ) 5%), and there were no patterns in dissolved-phase occurrence among toxin classes. Dissolved toxin data are provided in Table SI-S6, but are not discussed further herein. Total cyanotoxin data are shown in Table 2. Of the six measured toxin classes, five were detected (lyngbyatoxins were not detected). Microcystins occurred in all blooms, anatoxin-a in 30%, saxitoxins in 17%, and cylindrospermopsins and nodularin-R in 9%. Microcystins. Most blooms (91%) had detectable microcystins by both LC/MS/MS and ELISA (Table 2). Two blooms with low-level detections (60% of the cyanobacterial community alone or in combination with other dominant genera in one or more lakes. CYL, cylindrospermopsins; MC, microcystins; ATX, anatoxins; SAX, saxitoxins; LYN, lyngbyatoxins; GEOS, geosmin; MIB, 2-methylisoborneol. - indicates no known strains produce the compound. b Based on Table 1 in Graham et al., 2008 (10).
TABLE 2. Total Toxins Measured by ELISA and LC/MS/MS and Dissolved Taste-and-Odor Compounds As Measured by GC/MS in Cyanobacterial Blooms (n = 23) from the Midwestern U.S. (All Data Are Reported in µg/L; All Measurements were e20%)a ELISA lake
LC/MS/MS
CYL
MC
SAX
ATX
LA
LF
LR
Beeds Binder Carter Clear Crystal East Okoboji Prairie Rose Rock Creek Spirit Upper Gar
0.12 -
2.3 40 2.6 2.8 39 12 1.5 13,000 2.6 19
0.19 0.04 -
0.29 9.5 0.02 -
0.02 0.20 0.02 0.03 54 0.05 1.4
0.03 0.04 51 -
Clinton Miami Perry Prairie Sabetha
0.14 -
0.15 36 0.14 3.1
0.02 -
-
1.3 -
0.05 -
Albert Lea Budd Elysian Loon Lura Okamanpeedan
-
2.5 17 5.3 4.9 7.7 0.98
0.02 -
1.1 0.16 0.14
0.28 0.38 -
Minnesota 0.02 1.7 0.34 12 0.02 4.7 0.02 1.6 0.07 7.7 0.27
Bilby Ranch Mozingo
-
1.7 0.55
-
0.02 -
0.52 0.11
-
number of detects mean of detects median of detects minimum maximum
2 0.13 0.13 0.12 0.14
22 600 3.0 0.14 13,000
4 0.07 0.03 0.02 0.19
7 1.6 0.16 0.02 9.5
12 4.9 0.24 0.02 54
9 5.7 0.04 0.02 51
Iowa 0.39 6.0 1.1 13 0.24 0.21 2,100 0.73 6.2 Kansas 0.20 18 0.07 1.6
Missouri 0.06 0.39 21 100 1.6 0.06 2,100
GC/MS
LW
LY
RR
YR
NOD
SMC
GEOS
MIB
0.07 56 -
0.10 0.03 0.32 200 0.06
2.0 9.4 0.59 110 0.03 0.75 16,000 1.4
0.04 0.81 0.43 4.0 0.02 240 0.16
0.19 -
2.4 17 0.46 1.7 127 0.30 0.98 19,000 0.78 9.2
0.01 0.02 0.01 0.01 0.11 0.01 0.69 0.02 0.86
0.06 0.01 0.01 0.01 0.03
0.07 -
0.20 0.28 -
0.04 3.0 0.25
0.13 -
-
0.44 23 0.07 1.9
0.02 0.05 0.01
0.03 0.04
0.02 0.38 0.03 0.05 0.09 -
0.10 0.35 0.09 0.04 0.69 0.02
0.64 13 0.40 2.4 9.8 0.22
0.03 0.26 0.03 0.14 0.38 0.04
0.01 -
2.5 27 5.7 4.2 19 0.54
0.03 0.05 0.01 0.28 0.02
-
0.03
0.02
0.13 -
0.01 -
-
0.72 0.54
0.01 0.01
0.01 -
9 6.3 0.07 0.02 56
14 14 0.10 0.02 200
18 900 1.1 0.03 16,000
15 16 0.14 0.01 240
2 0.10 0.10 0.01 0.19
22 880 2.2 0.07 19,000
19 0.12 0.02 0.01 0.86
8 0.03 0.02 0.01 0.06
a ATX, anatoxin-a; CYL, cylindrospermopsins; GEOS, geosmin; MC, microcystins; LA, LF, LR, LW, LY, RR, YR, microcystin variants; MIB, 2-methylisoborneol; NOD, nodularin-R; SAX, saxitoxins; SMC, sum of microcystin variants as measured by LC/MS/MS; Cylindrospermopsin, deoxycylindrospermopsin, and lyngbyatoxin-a were not detected by LC/MS/MS. indicates concentration was below the analytical detection limits (ELISA: MC
