supplements
Fischer A1, Hoeger SJ1, Fastner J2, Robertson A3, Dietrich DR1
1 Human and Environmental Toxicology, University of Konstanz, Konstanz, Germany
2 Federal Environmental Agency, Section II 3.3 - Drinking-water resources and treatment, Berlin, Germany
3 NOAA, Northwest Fisheries Science Center, Environmental Conservation Division, Harmful Algal Blooms Program, Seattle, Wisconsin, USA
in preparation
6.1 Abstract
Blue-green algae dietary supplements (BGAS) are still consumed at a large scale for their putative beneficial health effects, despite the publicized fact that they might be contaminated with cyanobacterial hepato- and neurotoxins.
Microcysin-assigned intoxications of consumers of BGAS products, which are based on Aphanizomenon flos-aquae harvested in Upper Klamath Lake in Oregon, have been reported and, hence, warnings against this hepatotoxic cyanotoxin have been proclaimed. Recent studies detected the neurotoxic excitatory amino acid β-N-methylamino-L-alanine (BMAA) in the vast majority (95%) of the cyanobacterial genera tested and also unexpectedly in the brain tissue of Alzheimer’s patients from Canada. This may indicate an unknown source of exposure to this cyanobacterial excitotoxin, since it has been associated with the increased incidents of an extensively endemic neurodegenerative disease among the indigenous population of Guam. In this study we therefore examined the presence of microcystins and BMAA in
assay, ELISA and LC-MS/MS, respectively. While we detected none of these products positively for BMAA, 10 out of 11 contained microcystins with 5 exceeding the maximum acceptable concentration of 1 µg/g dw defined by the Oregon Health Division.
Keywords: blue-green algae supplements, microcystin, BMAA
6.2 Introduction
Cyanobacteria are cosmopolitans, existing in nearly every environment imaginable. Their worldwide distribution led to versatile human exposure. While some of them, primarily Spirulina, Arthrospira, and Nostoc species, served men as aliments for centuries (Gao, 1998; Abdulqader et al., 2000; Carmichael et al., 2000) others are infamous for producing a variety of toxic secondary metabolites, usually classified into hepatotoxins and neurotoxins.
Especially the hepatotoxic microcystins (MCs) have been responsible for severe human (Teixera et al., 1993; Jochimsen et al., 1998) and animal poisonings (Odriozola et al., 1984; Carbis et al., 1995; Ernst et al., 2001). These cyclic heptapeptides comprise a family of more than 80 congeners (Spoof, 2005;
Zurawell et al., 2005; Humpage, 2008) with microcystin-LR (MCLR) being the most intensively investigated. The main variations in the general structure cyclo(-D-Ala1-L-X2-D-methylAsp3-L-Z4-Adda5-D-Glu6-N-methyldehydro-Ala7) predominantly occur in two of the seven amino acids (X and Z at position 2 and 4, respectively) and hence are used for the nomenclature of MCs (Carmichael et al., 1988a) according to the one-letter code for amino acids. Their primary molecular mode of action is the inhibition of several serine/threonine-specific protein phosphatases (PPs), especially PP1 and PP2A (Honkanen et al., 1990;
MacKintosh, 1993; Runnegar et al., 1993).
Besides hepatotoxins, some cyanobacteria are capable of producing potent neurotoxic alkaloids like anatoxins and saxitoxins (PSPs) that also have been associated with severe human (Falconer, 1993; Garcia et al., 2004) and animal poisonings (Mahmood et al., 1988; Negri et al., 1995b). Anatoxins mimic the effect of acetyl choline (Aronstam and Witkop, 1981; Swanson et al., 1986),
Chapter VI MC and BMAA in BGAS
whereas saxitoxins block sodium channels of nerve cells (Catterall, 1980;
Kuiper-Goodman et al., 1999), hence, both are leading to paralysis, respiratory depression and respiratory failure at sufficiently high doses (Carmichael et al., 1975; Kao, 1993; Carmichael, 1997).
Due to eutrophication of surface waters, mass development of potentially toxic cyanobacteria is widespread along with the hazard of intoxications, especially in poorer countries with limited or no drinking water treatment at all (Herath, 1995;
de Figueiredo et al., 2004; Hoeger et al., 2004). However, in industrialized countries that are less threatened by contaminated drinking water or different toxin-accumulated food items, an exceptional source of potential exposure to cyanotoxins is on the rise since the early 1980’s. This relatively new threat is posed by voluntarily ingested dietary supplements, mostly based on Spirulina (Arthrospira) spp. or Aphanizomenon flos-aquae. These blue-green algae supplements (BGAS), as they are commonly referred to, are consumed for their advertised putative beneficial health effects, such as increased alertness and energy, “detoxification”, weight loss, efficacy against various viral infections, including herpes and influenza, and even against cancer and mental disorders like depression or attention-deficit disorders (Gilroy et al., 2000; Lukassowitz, 2002; Saker et al., 2005). In 1999 the number of consumers was estimated to be over one million in North America alone (Falconer et al., 1999).
While Aphanizomenon flos-aquae has been reported to be a potential producer of neurotoxins, like anatoxin-a (Rapala et al., 1993; Carmichael, 1997) and various PSP toxins (Jackim and Gentile, 1968; Ikawa et al., 1982; Mahmood and Carmichael, 1986a; Carmichael, 1997; Pereira et al., 2000; Ferreira et al., 2001; Liu et al., 2006), the strain Aphanizomenon flos-aquae Ralfs ex Born. &
Flah. var. flos-aquae from Upper Klamath Lake in Oregon, USA, which is harvested for processing into the supplements usually marketed as “AFA”, turned out to be non-toxic with regard to neuro- and hepatotoxicity (Carmichael et al., 2000). However, potentially toxin producing cyanobacteria coexist in Lake Klamath, including Microcystis aeruginosa, Anabaena flos-aquae and Oscillatoria spp. (Carmichael et al., 2000). Indeed, contamination of A. flos-aquae products with the hepatotoxic microcystins (MCs) have been reported (Gilroy et al., 2000; Kuiper-Goodman et al., 2000; Lawrence et al., 2001; Saker
(MAC) of 1 µg MC/g dw was enforced by the Oregon Health Division and the Oregon Department of Agriculture (Gilroy et al., 2000).
In general, Spirulina is considered a non-toxic genus (Salazar et al., 1996;
Salazar et al., 1998), although anatoxins have been detected in supplements labeled Spirulina (Draisci et al., 2001; Osswald et al., 2008). Moreover, contaminations with microcystins occur less frequently in these products due to controlled cultivation conditions in ponds (Gilroy et al., 2000; Kuiper-Goodman et al., 2000). Contrarily, Arthrospira fusiformis is a potential producer of microcystins itself (Ballot et al., 2004; Ballot et al., 2005) and a case of hepatotoxicty that has been assigned to consumption of Arthrospira has been reported (Iwasa et al., 2002). Moreover, this species may also produce the neurotoxic alkaloid anatoxin-a (Ballot et al., 2004; Ballot et al., 2005). However, due to morphological similarities Spirulina and Arthrospira are easily confused, hence, neuro- and hepatotoxin detection in Spirulina might possibly be attributed either to mislabelling of the products or to a contamination with another toxic cyanobacterial species.
Recently, it was found that 20 out of 21 cyanobacterial genera and 29 out of 30 cyanobacterial strains tested, including a marine strain of Aphanizomenon flos-aquae, produce the excitatoric amino acid β-N-methylamino-L-alanine (BMAA) (Cox et al., 2005). BMAA is an excitotoxic amino acid that acts as an agonist of animal glutamate receptors (Weiss et al., 1989a; Weiss et al., 1989b; Myers and Nelson, 1990) and elevates intracellular calcium levels (Brownson et al., 2002). It is considered as the causative agent in one hypothesis for the increased incidents of amyotrophic lateral sclerosis/parkinsonism-dementia complex (ALS/PDC) among the Chamorro, the indigenous population of Guam, as a result of bioaccumulation in their traditional diet, i.e. cycad seeds and flying foxes with endosymbiotic cyanobacteria of the genus Nostoc being the primary source (Spencer et al., 1987a; Cox and Sacks, 2002; Bannack and Cox, 2003;
Cox et al., 2003). This controversial hypothesis and BMAA toxicity are reviewed by various authors (Ince and Codd, 2005; Papapetropoulos, 2007; Steele and McGeer, 2008). The neurodegenerative disease elicits symptoms similar to amyotrophic lateral sclerosis, Parkinson’s disease and Alzheimer’s. Besides Guam, there are two further high-incidence foci of ALS/PDC in Irian Jaya, West
Chapter VI MC and BMAA in BGAS
New Guinea and on Kii peninsula of Japan where cycad seeds are part of the traditional diet as well (Spencer et al., 1987b; Spencer et al., 1987c).
BMAA appeared to occur in a free and protein-bound form. The latter is hypothesized to build an endogenous neurotoxic reservoir from which free BMAA might slowly be released causing chronic damage to the brain over years or even decades (Murch et al., 2004b). Interestingly, BMAA was not only detected in the brain tissue of ALS/PDC patients from Guam, but also in those of Alzheimer’s patients from Canada (Cox et al., 2003; Murch et al., 2004a) suggesting a further source for BMAA intoxication that is not restricted to Guam.
Indeed, Johnson et al. (Johnson et al., 2008) positively analyzed BMAA in all 21 samples of Nostoc commune that is part of the traditional diet of the indigenous people in the mountains of Peru, by HPLC-FD, UPLC-UV, UPLC/MS, and LC/MS/MS. The averaged concentrations ranged from 6.18 to 18.90 µg/g. In addition, Metcalf et al. (Metcalf et al., 2008) detected free and protein-bound BMAA in all 12 environmental samples tested, including blooms, scums, and mats from waterbodies throughout the UK confirming the widespread distribution of BMAA in cyanobacteria. The predominant genera were identified as Microcystis, Planktothrix, Anabaena, Aphanizomenon, Gomphosphaeria, Nodularia, Pseudanabaena and Oscillatoria. Besides BMAA, they also reported the presence of other cyanotoxins in all but 2 of the samples: predominantly MCs, but also nodularin, anatoxin-a and saxitoxin, as well as an unidentified substance that elicited acute neurotoxicity in a bioassay.
Therefore, the scope of this study was to examine different BGAS in order to assess 1) the known health risk of a potential contamination with MCs by colorimetric protein phosphatase inhibition assay, Adda-ELISA and LC-MS/MS and 2) the possible additional health hazard caused by potentially contained BMAA using LC-MS/MS.
6.3 Material & Methods
Chemicals and reagents
All chemicals were of the highest analytical grade commercially available.
MCLF were obtained from Alexis (Switzerland). They were dissolved in 75%
MeOH and the concentrations of the stock and the working solutions were proven photometrically by using the molar absorption coefficient of MCLR and MCRR (39800 mol l-1 cm-1) published by Harada et al. (Harada et al., 1990b).
This coefficient was applied for all microcystins, since no molar absorption coefficients are published for the other congeners. Although this is the molar absorption coefficient for MCLR/-RR dissolved in 100% MeOH it turned out to be applicable for 75% MeOH as well (Meriluoto and Spoof, 2005). Additionally, concentrations were confirmed by HPLC-DAD analysis according to Lawton et al. (Lawton et al., 1994). L-BMAA hydrochloride was purchased from Sigma (Germany) and dissolved in 100% H2O, diluted to the stock and working concentrations according to the manufacturer’s specifications.
Methanolic extraction of microcystins from blue green algae supplements 11 products of different brands containing Aphanizomenon flos-aquae from Lake Klamath (Code: Aph # 1 - 11) were examined. Three (Aph # 2 - 11) or four (Aph # 1) subsamples of each product were examined and extracted separately. 375 mg of each subsample were dissolved in 75% MeOH by vigorous shaking and ultrasonicated for 30 min (15 min in the following steps) in an ice-cold ultrasonic bath. After centrifugation at 4,000 x g for another 30 min the resulting supernatants were collected and stored on ice. The pellets were resuspended in 75% MeOH as described above. This methanolic extraction was repeated 3 times and the respective supernatants were pooled. The total volume of ~45 ml/sample was evaporated passively to a small residue on ice, which was in turn filled up to 15 ml with purified water, followed again by vigorous shaking and ultrasonication for 15 min in an ice-cold ultrasonic bath.
Purification of microcystins from aqueous extracts
A solid phase extraction using C18 silica cartridges (Waters, Sep-Pak® Vac 6cc (1g)) was performed to purify and concentrate microcystins potentially present in the aqueous extracts. The cartridges were preconditioned with 2x 5 ml MeOH (100%) and equilibrated with 2x 3 ml H2O prior to adding the respective extracts. After elution of the extracts the cartridges were washed three times with 4 ml H2O. MCs potentially bound to the cartridges were eluted with 3 x 4.5
Chapter VI MC and BMAA in BGAS
ml MeOH (100%). These eluates were collected and evaporated passively over night to a small residue on ice prior to being completely dried in by vacuum centrifugation. Subsequently, these dried extracts were redissolved in 600 µl MeOH (100%) by vigorous shaking and untrasonification for 7.5 min. Then 2,400 µl H2O were added leading to a final concentration of 20% methanol, followed again by vigorous shaking and ultrasonification for 7.5 min. A last centrifugation step at 13,000 x g for 20 min was applied to remove particles.
The resulting supernatants were used for MC analysis.
Colorimetric protein phosphatase inhibition (cPPIA) assay with recombinant PP1
All three (four) extracts of the 11 A. flos-aquae samples (Aph # 1 - 11) were diluted serially (1:3) and analyzed three times in duplicates. In every assay MCLR served as internal standard for quantification and was applied diluted serially (1:3) at concentrations from 333.3 nM to 0.02 nM on every plate. The concentration of microcystin in the samples was calculated from the obtained standard curve and expressed in MCLR equivalents/g dw. The assay was carried out as described by Herersztyn and Nicholson with one modification (Heresztyn and Nicholson, 2001): instead of PP2A we used PP1 (rabbit skeletal muscle, recombinant (E. coli); New England Biolabs (USA)) in a stock concentrations of 2,500 units/ml, which resulted in a concentration of 3 units/ml in the final assay.
P-nitrophenylphosphate from Acros Organics (USA) was used as substrate.
Adda-Enzyme Linked Immunosorbent Assay (Adda-ELISA)
All A. flos-aque samples, but sample Aph # 1 were analyzed by a competitive indirect ELISA in a previous work (Hoeger et al., 2003), using a commercially available kit (Microcystins (Adda specific) ELISA Kit, Abraxis™ (USA)) according to the manufacturer’s specifications. Therefore, only three separately generated extracts of BGAS sample Aph # 1 and as a positive control one extract of each BGAS sample Aph # 8 and # 9 were analyzed in duplicates in different dilutions on two separate plates of the ELISA kit according to the manufacturer’s specifications. Thus, the data of sample Aph # 1 is composed of
calculated from the obtained standard curve using the MCLR standards provided in the kit, hence, are given as MCLR equivalents. The range of this ELISA is between 0.15 nM and 5 nM (approximately 0.15 µg/l and 5 µg/l).
Liquid Chromatography Tandem Mass Spectrometry (LC-MS/MS) analysis of microcystins
The LC-MS/MS analyses of the BGAS extracts (three extracts of AFA # 1 - 11) were carried out on an Agilent 1100 series HPLC system (Agilent Technologies, Waldbronn, Germany) coupled to a API 4000 triple quadrupole mass spectrometer (Applied Biosystems/MDS Sciex, Framingham, MA) equipped with a turbo-ionspray interface.
The extract was separated using a Purospher STAR RP-18 endcapped column (30 x 4 mm, 3 µm particle size, Merck, Germany) at 30°C. The mobile phase consisted of 0.5% formic acid (A) and acetonitrile with 0.5% formic acid (B) at a flow rate of 0.5 ml min-1 with the following gradient programme: 0 min 25% B, 10 min 70% B, 11 min 70% B (Spoof et al., 2003). The injection volume was 10 µl.
The mass spectrometer was operated in the multiple reaction monitoring mode (MRM) for the confirmation of microcystins listed in Table 6.1. The ion dwell time was 200 ms. Quantitation of microcystins was achieved using the MRM mode. Standard curves were established for MCRR, YR, LR, LW, LF and -LA (Alexis chemicals, Switzerland) and analyzed in a line with the unknowns (one calibration curve after 10 unknowns). In the MRM mode the limit of detection (LOD) for the single MC congeners were in the range of 0.64 - 2.5 pg o.c.
The samples were also analyzed for the presence of other MC congeners for which no standards are commercially available using the precursor ion mode as described previously by Hiller et al. (Hiller et al., 2007). Though applied for qualitative purposes only, the analyses of samples with known microcystin concentrations by this method gave higher LODs (~1 µg/g dw) than with the MRM mode.
Chapter VI MC and BMAA in BGAS
Tab. 6.1: Transitions and parameters for the detection of microcystins by LC-MS/MS in the MRM mode; DP: declustering potential, CE: collision energy, CXP: cell exit potential;
*, [M+2H]2+
Extraction of BMAA from blue green algae supplements
Three of the A. flos-aquae products analyzed for MC contamination (Code: Aph
# 1, # 8 and # 9) and two additional Spirulina products (Code: SP # 1 and # 2) were examined. Of each product two samples composed of 3 subsamples were extracted separately. The samples were extracted in 0.1 M trichloroacetic acid (TCA) and sonicated with a probe sonicator (600 watts; 30 kHz) for 2 min on ice. Subsequently, they were centrifuged at 15,800 x g for 5 min at 4°C. The supernatants, consituting the free fraction of potentially contained BMAA, were dried by vacuum centrifugation and stored at -20°C until further processing. The pellets were washed three times with 1 ml acetone (-20°C) prior to acidic hydrolysis with 6 M HCl for 24 h at 110°C. Afterwards, the hydrolysates, constituting the protein-bound fraction of potentially contained BMAA, were dried and stored as described for the supernatants. All fractions were reconstituted in 10% ACN/H2O, followed by centrifugal filtration at 10,000 x g using centrifugal filter units (Millipore; ultrafree-MC GV 0.22 µm). The filtrates were transferred into UPLC-vials (Waters; glass maximum recovery vials) and stored at -20°C until analysis.
Two subsamples per replicate were spiked with 10 ng BMAA: both during extraction, one prior to hydrolysis and again both prior to final analysis.
Toxin Transitions
m/z [M+H]+ → m/z DP CE CXP LOD (pg o.c.) Microcystin-RR 519.7* → 135 90 42 12 0.64 Microcystin-LA 910.5 → 135 96 77 6 1.3 Microcystin-LF 986.5 → 135 106 77 8 1,5 Microcystin-LR 995.5 → 135 140 90 28 2.5 Microcystin-LW 1025.5 → 135 101 89 14 1.8 Microcystin-YR 1045.5 → 135 145 90 13 3
Liquid Chromatography Tandem Mass Spectrometry (LC-MS/MS) analysis of BMAA
Extracts of the blue-green algal supplements were sent to Dr. Alison Robertson at the National Oceanic and Atmospheric Administration, Northwest Fisheries Science Center for confirmatory mass spectrometric analysis of BMAA. Sample extracts were shipped on dry ice and stored at -80°C upon arrival. All solvents and water used for our analyses were of HPLC grade and purchased from JT Baker Laboratory Chemicals (Phillipsburg, NJ).
Samples were analyzed by liquid chromatography tandem mass spectrometry (LC-MS/MS) in positive ion mode using multiple reaction monitoring (MRM) for the confirmation of BMAA. All MS experiments were performed using a Waters UPLC® system coupled to a Quattro micro triple quadrupole tandem mass spectrometer (MicroMass, Waters, US) fitted with an electrospray ionization interface (capillary: +2.6 kV, cone: +45 V). Normal phase analyte separation was achieved using hydrophilic interaction with a TosoTSKgel Amide-80 analytical column (5 µm, 250 mm x 2.1 mm, Waters, MA) with a guard column (2 mm x 1 mm), both maintained at 40°C. A flow rate of 0.25 ml/min resulted in a stable backpressure of approx. 1600 psi and was used for all analyses.
Samples (10 µl injection volume) were separated using a linear gradient from 90-30% B for 10 min, followed by a 1 min hold at 30% B, and then re-equilibration to 90% B for 5 min. Mobile phase consisted of water (A) and acetonitrile (B) both containing 50 mM formic acid.
Method optimization, calibration, retention time verification, matrix effects and characteristic MS/MS fragmentation was determined using a standard solution of L-BMAA hydrochloride from Sigma (≥97% purity; B107; Lot 087K47032) at the same concentration as used for spiking. Identification of suspect toxin peaks was undertaken using multiple reaction monitoring (MRM) for three characteristic transition ions of the protonated BMAA including; 119>102 (collision: 10 eV), 119>88 (collision: 15 eV), and 119>74 (collision: 10 eV). The relative ion ratios for these three transitions was also recorded and used for additional verification of BMAA. All MS/MS parameters were optimized prior to analysis using full MS and MS/MS scans and quantification was determined using an 8-point calibration curve (50 pg/ml - 5 µg/ml DA). Matrix effects were assessed by spiking and standard addition experiments and samples were
Chapter VI MC and BMAA in BGAS
subsequently diluted as required to eliminate any observed effects such as ionization suppression and BMAA retention shifts. Data analysis including peak integration and quantitation was performed using MassLynx software (version 4.1; Waters Laboratory Informatics).
Mass spectrometer source and analyzer parameters were as follows: Extractor 2 V; RF Lens 0.2 V; source temperature 130°C; desolvation temperature 385°C;
cone gas flow 200 L/Hr; desolvation gas flow 500L/Hr; collision cell entrance 50 and exit -8; and multiplier at 650 V.
Statistics
In order to compare the different methods of MC analysis the corresponding mean values (MCLR equivalents and total MC concentration) of BGAS samples Aph # 1, Aph # 8 and Aph # 9 were statistically analyzed by One-way ANOVA followed by Tukey's Multiple Comparison Test using GrAphPad Prism 4.03 software.
6.4 Results
Colorimetric protein phosphatase inhibition (cPPIA) assay with recombinant PP1
The inhibitory capacities of 11 A. flos-aquae extracts (Aph # 1 - 11) on recombinant PP1 were tested in three independent colorimetric protein phosphatase inhibition assays. The concentrations determined by comparing the inhibition of the BGAS samples with the inhibiton of MCLR standard dilutions ranged from 0.02 - 7.95 µg MCLR equivalents/g dw (Tab. 6.2). Every sample elicited inhibitiory effects on PP1, whereas the MAC of 1 µg/g dw defined by the Oregon Health Division and the Oregon Department of Agriculture was exceeded by 7 samples.
Adda-Enzyme Linked Immunosorbent Assay (Adda-ELISA)
Three of the 11 BGAS extracts were investigated by the Adda-specific ELISA.
The assay revealed MC contaminations in all three BGAS samples (Aph # 1, #
BGAS samples Aph # 8 and # 9 (Tab. 6.2). Moreover, the MC concentrations in these two samples were in the same range as deteced by Hoeger et al. (Hoeger et al., 2003).
Tab. 6.2: MC concentrations in BGAS as determined by cPPIA, Adda-ELISA (expressed as MCLR equivalents) and LC-MS/MS; values of cPPIA and ELISA represent mean ± standard deviation of at least three independent replicate analyses of three (Aph # 2 - 11) or four (Aph # 1) separately extracted samples per BGAS; values of LC-MS/MS represent mean, maximum and minimum (in brackets) of one analysis of three (Aph # 2 - 11) or four (Aph # 1) separately extracted samples per BGAS.
*, (Hoeger et al., 2003)
-, not determined in this study
n.d., none of the commercially available standard MCs detectable.
cPPIA Adda-ELISA LC-MS/MS
cPPIA Adda-ELISA LC-MS/MS