Cyanobacterial neurotoxin BMAA and brain pathology in stranded dolphins

Dolphin stranding events occur frequently in Florida and Massachusetts. Dolphins are an excellent sentinel species for toxin exposures in the marine environment. In this report we examine whether cyanobacterial neurotoxin, β-methylamino-L-alanine (BMAA), is present in stranded dolphins. BMAA has been shown to bioaccumulate in the marine food web, including in the muscles and fins of sharks. Dietary exposure to BMAA is associated with the occurrence of neurofibrillary tangles and β-amyloid plaques in nonhuman primates. The findings of protein-bound BMAA in brain tissues from patients with Alzheimer’s disease has advanced the hypothesis that BMAA may be linked to dementia. Since dolphins are apex predators and consume prey containing high amounts of BMAA, we examined necropsy specimens to determine if dietary and environmental exposures may result in the accumulation of BMAA in the brains of dolphins. To test this hypothesis, we measured BMAA in a series of brains collected from dolphins stranded in Florida and Massachusetts using two orthogonal analytical methods: 1) high performance liquid chromatography, and 2) ultra-performance liquid chromatography with tandem mass spectrometry. We detected high levels of BMAA (20–748 μg/g) in the brains of 13 of 14 dolphins. To correlate neuropathological changes with toxin exposure, gross and microscopic examinations were performed on cortical brain regions responsible for acoustico-motor navigation. We observed increased numbers of β-amyloid+ plaques and dystrophic neurites in the auditory cortex compared to the visual cortex and brainstem. The presence of BMAA and neuropathological changes in the stranded dolphin brain may help to further our understanding of cyanotoxin exposure and its potential impact on human health.


Introduction
Harmful algal blooms (HABs) are becoming increasingly frequent in fresh water lakes, estuaries, and the sea [1][2][3] likely due to climate warming and increased phosphorous and nitrogen effluents [2]. HABs are sometimes associated with widespread dying-off of sea grass and fish ascribed to the large biomass of algae producing hypoxic conditions in the water [4]. HABs composed of cyanobacteria contain multiple cyanotoxins including microcystins, nodularin, cylindrospermopsin, and anatoxins, while those composed of dinoflagellates can contain brevitoxins [5][6][7]. Acute and chronic exposures to such toxins can be harmful to both humans and animals resulting in respiratory illnesses, severe dermatitis, mucosal damage, cancer, organ failure and death [7].
Dolphins that inhabit Florida coastal waters are often exposed to recurring HABs [29][30][31]. Coastal subpopulations of bottlenose dolphins in the southeastern United States have an increased risked of cyanobacteria bloom exposure, compared to common dolphins, a pelagic species, which spend relatively little time in coastal or near-shore environments [27,32]. Bottlenose dolphins have a median lifespan of 30 to 32 years with an maximum of 50 years in captivity [33]. Irrespective of age or sex, bottlenose dolphins have been observed stranded all along the coastal United States [34][35][36]. Stranding events have been attributed to trauma [37], infections [35], neuropathology [38], navy sonar [39,40] or toxin exposures [29]. Regardless of cause of stranding, dolphins are valuable bio-indicators of the health of the marine environment and provide clues of potential environmental risks to humans. We hypothesize that dolphins are likely to accumulate BMAA both by exposure to HABs and by the ingestion of prey previously exposed to the cyanotoxin [27,28].
In this study, we examined an opportunistic sample of brain tissues from fourteen stranded dolphins obtained from the Rosenstiel School of Marine and Atmospheric Science and the International Fund for Animal Welfare (IFAW). BMAA was measured in select brain regions of stranded dolphins. In addition, we examined brain tissues of seven dolphins stranded in Massachusetts to seek a correlation between neuropathological changes and the presence of BMAA.

Dolphins
Adult (n = 9) and sub-adult (n = 5) stranded dolphins were obtained from ten different sites in Florida and Massachusetts between 2005 and 2012 (S1 Table). The seven Florida dolphins were found stranded in sites known to have recurring harmful algal blooms: Atlantic Ocean [41], Banana River [42], Indian River Lagoon [42,43] and Gulf of Mexico [30]. The seven Massachusetts dolphins were found stranded in various sites in Cape Cod Bay [44]. Cyanobacteria species capable of producing BMAA were detected in all these regions as determined by surveillance data from the Phytoplankton Monitoring Network (S2 Table) [45]. Dolphins were classified as adult or sub-adult based on body length from the Geraci & Lounsbury 1993 stranding guide [46]. Dolphins varied in manner of stranding, species and sex (S1 Table). Six of fourteen dolphins (43%) were observed alive while stranded. Physical assessments were performed on the dolphins when first sighted [46]; those assessed to be in poor health and unable to be released were euthanized on site, followed by a full necropsy within 24 to 48 hours by IFAW. Dolphin specimens in Massachusetts were obtained under a Stranding Agreement between National Ocean and Atmospheric Administration (NOAA) National Marine Fisheries Services Greater Atlantic Region Fisheries Office and IFAW. That agreement enables access to and management of stranded pinnipeds and cetaceans, and where applicable, landowners were consulted when animals were located on private property. Clinical assessments and euthanasia were conducted according to IFAW's protocols approved by the Woods Hole Oceanographic Institution Institutional Animal Care and Use Committee. Euthanasia was done by intravenous barbiturate, following guidelines from Barco et al 2016 [47]. Euthanasia decisions were made on the basis of prognosis. The Florida specimens were obtained from deceased dolphins found off the Florida coast, either beached or floating. Necropsies on these dolphins were performed at the Hubbs Seaworld Research Institute. These specimens were obtained with permission from the National Marine Fisheries Services (S1 Table).
At necropsy, external and internal examinations were performed and tissue samples were collected for histopathology and infectious disease testing (Morbillivirus, Brucella spp., etc. [48]) where available (S1 Table). Brains from the Massachusetts' cohort (n = 7 dolphins) were harvested, and a mid-sagittal cut was made to separate each brain into right and left hemispheres. One hemisphere was frozen and stored in an ultra-low freezer at -80˚C for toxicological analysis and the other hemisphere was fixed in 10% neutral buffered formalin for one month for histopathological assessments. Cryopreserved brain tissues were sampled for analytical measures of BMAA (see methods below). Formalin-fixed brain hemispheres were used to assess neuropathological changes (see methods below). Toxicological and histological assessments of dolphin brain specimens were performed with the approval from the NOAA Southeast Region Stranding Program and National Marine Fisheries Service.

Brain extraction and analytical methods
Frozen punch biospecimens (200 mg) from the cerebral cortex of each dolphin were made using a 6.0mm Biopunch (Ted Pella, Inc., Redding CA, USA). Punch biospecimens were obtained from the auditory cortex (ACtx) and the visual cortex (VCtx). Following sampling, tissue biospecimens were subjected to total acid hydrolysis and assayed for total concentrations of BMAA and BMAA structural isomers N-(2-aminoethyl) glycine (AEG) and 2,4-diaminobutyric acid (DAB) as described by Mondo et al 2012 [26]. No attempts were made to differentiate free from protein bound BMAA. Briefly, a 1:6 W/V of 6 N HCl (1000 μL) was added to each cortical biopsy, followed by sonication. Samples were then hydrolyzed on a heat block for 18 hours at 110˚C, filtered at 13,800 x g for 3 minutes at 4˚C (0.22 lm Ultrafree-MC, Millipore, Bedford, MA, USA), and concentrated in a speed-vacuum (Thermo-Savant SC250DDA Speed Vac Plus with a Savant refrigerator trap RVT 4104). The dried extract was then resuspended to 1000 μL of 20 mM HCl. An aliquot of the sample extract was derivatized with AQC using the AccQ-Fluor reagent (Waters Crop, Milford, MA). The derivatized samples (20 μL resuspended HCl extract, 60 μL of borate buffer, and 20 μL AccQ-Tag) were run in parallel with buffer and AQC blanks, and L-BMAA, AEG, DAB and reference amino acid standards. Since control dolphin specimens could not be obtained, fresh frozen human brain tissues were used in our assay. For a negative control, frozen punch biopsies were obtained from the cortex of a 68 year-old human female with no known neurological changes and previously found to be free of BMAA [18]. For a positive control, punch biopsies were obtained from a 77 year-old human male with ALS whose tissues were previously shown to contain a high concentration of BMAA [18]. In addition, control sample matrixes were also spiked with known amounts of BMAA to confirm the efficiency of extraction.
Detection was achieved using a Waters 2475 Multi k-Fluorescence Detector with excitation at 250 nm and emission at 395 nm. Measurements of BMAA in dolphin cortical tissues were compared with those of the human positive and negative controls (described above), and those spiked with known amounts of a reference BMAA standard (Sigma B-107; >95% purity, St. Louis, MO, USA). The limits of detection (LOD) and the limit of quantification (LOQ) were based on the standard deviation (SD) of response and slope (S), calculated from linearity of the response of BMAA. The LOD (2.7 ng) and LOQ (7.0 ng) were obtained by using formula (3.3 x SD)/S and (10 x SD)/S, respectively. The efficiency of the analyte recovery was estimated adding known amounts of a BMAA standard spiked into a reference sample that was below the LOD.

UPLC-MS/MS method
BMAA and its structural isomers AEG and DAB were separated, detected and quantified by ultra-performance liquid chromatography/ tandem mass spectrometry (UPLC-MS/MS) using a fully validated method as previously described [52,53]. In brief, punch biospecimens of cerebral cortex (200 mg) were weighed and suspended in 1.0 mL of 6 N HCl sealed with N 2 gas blown into the tubes for 30 s to displace oxygen. Brain samples were hydrolyzed for 18 hours at 110˚C. A subsample of 400 μL was filtered (0.22 μm PVDF Ultrafree MC centrifuge filters; EMD Millipore; Billerica, MA, USA) and a 100 μL aliquot was dried overnight (Labconco Centrivap; Kansas City, MO, USA). The sample was reconstituted in 1.0 mL 20 mM HCl and derivatized as described above. BMAA, AEG and DAB were separated by reverse phase C18 chromatography (BEH column 150 × 2.1 mm 1.7 μm; Waters) and eluted with a gradient of 20 mM ammonium formate with 0.2% formic acid (A) and 0.1% formic acid in acetonitrile; (B). Gradient was delivered by a Waters Acquity I-Class UPLC (Milford, MA, USA) (0 min, 95% A; 1.0 min, 95% A; 7 min, 85% A; 7.5 min, 78% A; 8 min, 15% A; 8.5 min, 15% A; 8.6 min, 95% A; 10 min, 95% A) with a flow rate of 0.7 mL/min at 52˚C. Compounds were clearly separated with BMAA elution at 6.

Immunohistochemistry
To investigate neuropathological changes, formalin-fixed brain tissue blocks (2 x 2 x 0.2 cm 3 ) were collected from the ACtx, VCtx and medulla (Md) from seven dolphins. Tissue blocks were processed through xylenes and alcohols and embedded in paraffin wax. Microtome sections were cut at 5 μm and mounted on 1 x 3 inch glass slides. Tissue sections were deparaffinized in three changes of xylene for 10 minutes each, followed by two changes of absolute ethanol for 5 minutes each, then 95% ethanol for 5 minutes. Tissues sections were stained with hematoxylin and eosin (H&E) (for cytoarchitecture) and thioflavine-S (for plaques, tangles, and dystrophic neurites). Modified Bielschowsky's (MB) silver staining was performed at AML laboratories using an American MasterTech stain kit (Lodi, CA, USA) [54].
To probe for the specific protein antigens noted below, after rehydration sections were incubated in 3% H 2 O 2 in methanol for 10 minutes, followed by rinsing in 3 changes of distilled water (DiH 2 O) each for 5 minutes. Slides were then incubated in 98% formic acid for 45 seconds, followed by a wash in 3 changes of DiH 2 O on a Thermolyne Roto Mix shaker, and incubation in phosphate-buffered saline pH 7.4 (PBS) for 5 minutes. To block non-specific antibody bindings, 10% normal donkey serum in PBS pH 7.4 (PBS) was applied to tissue sections in a humidity chamber and incubated at room temperature for 30 minutes. Sections were then probed with mouse anti-β-amyloid (1:800, Covance, Princeton, NJ, USA); anti-neuronal nuclei protein (Anti-NeuN Antibody, clone A60; 1:200, Millipore, MA, USA); or antiglial fibrillary acidic protein (GFAP; 1:200, DAKO/Agilent, CA, USA). Antibodies were allowed to incubate at 4˚C overnight, and then the slides were rinsed with PBS for 3 x 10 minutes, followed by additional application of 2% normal donkey serum for 10 minutes then PBS rinse. A donkey anti-mouse-biotin (1:200; Jackson Immunoresearch, West Grove, PA, USA) conjugated goat anti-mouse secondary antibody was incubated on tissue sections for 2 hours at room temperature, rinsed with PBS wash for 10 minutes followed by application of ExtrAvidin peroxidase (1:5,000, Sigma-Aldrich, St. Louis, MO, USA) in PBS for incubation for 1 hour. ExtrAvidin peroxidase was detected using DAB solution (100 mL DAB = 98 mL PBS + 2 mL 25mg/mL DAB + 16.6 μL 3% H 2 O 2 ) for 10 minutes. Slides were then washed in two changes of PBS, rinsed with distilled water, counterstained with Gill No. 1 Hematoxylin for 20 seconds, and rinsed under running tap water for 5 minutes. Frontal cortex samples from a patient with advanced AD were used as a positive control for all neuropathology markers listed above.

Digital and fluorescent microscopic pathology
Histology slides were scanned at 40x resolution using a TissueScope LE (Huron Digital Pathology, Waterloo, ON, CAN). Digital scanning allowed for complete micropathological review of the entire dolphin brain tissue sections and margins at an optimal resolution of (0.2 μm/pixel at 40x). High quality tif file images (975 x 477 pixels) were exported from TissueScope LE and imported into NIH ImageJ 64 VER1.44o (NIH, Bethesda, Maryland, USA) for analysis of Aβ + density and Aβ + plaque area. To determine number and area size of Aβ + plaques, a total of 6 images were analyzed per dolphin (ACtx n = 3 scans and VCtx n = 3 scans). Aβ + plaques were annotated and quantified using ImageJ cell counter plug-in software and presented as the mean and standard error of Aβ + deposits per 5x scan. A threshold was applied to regions of interest to determine the area of each Aβ + plaques. Slides stained with MB silver, thioflavine-S, 4',6-diamidino-2-phenylindole (DAPI), GFAP and Neu-N were visualized at 20x magnification using a Zeiss Apotome fluorescent microscope (USA).

Statistics
Statistical analyses were performed using Prism Version 7 software (Graph Pad, La Jolla, CA, USA). Single comparisons tests were analyzed using Mann-Whitney U and Wilcoxon matched-pairs test. Multiple comparisons were analyzed using Friedman test with Dunn's multiple comparison test. For correlation analyses, the Pearson's Correlation Coefficient and Spearman correlation tests were used to determine significance. All data and results were expressed as the Mean ± Standard Error; significance level of alpha = 0.05. Due to the limited size of our sample cohort, Cohen's d was used to determine effect size and D'Agostino & Pearson test was used to determine normality of data. A table of descriptive statistics for each comparison can be found in supplementary material (S3 Table).

Stranded dolphins
A total of fourteen dolphins were investigated from the following dolphin species: bottlenose dolphin (Turiops truncates) and common dolphin (Delphinus delphis) (S1 Table). Nine dolphins (64%) in our sample cohort were observed dead. Five dolphins (36%) were part of a mass stranding. Four dolphins (29%) had conditions that have been shown to cause strandings: one with mechanical trauma (Hubbs 0805 Tt) and three cases of Brucella infections (IFAW 12-223Dd, 12-228Dd and 12-229Dd) [55]. Seven dolphins (50%) were female; seven dolphins were male (50%). One dolphin (7%) was pregnant at the time of death. Nine dolphins (64%) in the study were categorized as adult and five as sub-adult (36%). The bottlenose dolphins had 1.3-fold greater mean length than the common dolphins (P<0.0001, t-Test) (S1 Table). Bottlenose dolphins were also 4-fold greater in weight than the common dolphin (P<0.0001, t-Test) (S1 Table). All dolphins except for two (Hubbs 0805-Tt and Hubbs 0630-Tt) had no detectable injury resulting from human interaction. Dolphin Hubbs 0805-Tt had an old scar on her dorsal fin from a boat injury. Dolphin Hubbs 0630-Tt had fishing gear entangled around the flukes and hooks in the oral cavity. External examinations of several dolphins (29%) showed signs of stranding stress (IFAW12-201Dd), cauliflower-like lesions on flukes, peduncle and dorsal fin (IFAW12-228Dd), a large circular wound (Hubbs 0805 Tt) and several broken ribs (Hubbs 0541-Tt). One dolphin (7%) (IFAW12-198Dd) had incidental post-stranding injuries to the eyes as a result of gull pecking. Gross examinations of internal structures were unremarkable, except for two dolphins (14%) with non-specific lung (Hubbs 0541 Tt) and spleen (IFAW12-198Dd) pathology. One dolphin had signs of hepatitis (IFAW12-201Dd) [48].

Detection of BMAA
BMAA was detected in 13 of the 14 dolphins (93%) beached in both Florida and Massachusetts (Table 1, S1 Table, Fig 1). One dolphin injured by boat collision was negative for BMAA (Hubbs 0805 Tt). BMAA levels in the other 13 dolphin brains ranged from (20-748 μg/g). The overall mean concentration of BMAA was 287.9 ± 64.3 μg/g and the median concentration was 170 μg/g. The bottlenose dolphins that stranded in Florida had 3-fold (P = 0.01; t-Test) higher concentrations of BMAA than the common dolphins from Massachusetts (Table 1, S1  Table). Sex had no effect on the mean concentration of BMAA. The average BMAA/length of dolphin was 1.2 ± 0.2 μg/g/cm. BMAA concentrations were positively correlated with the length of dolphin (n = 13; Pearson r = 0.6002; P = 0.01) (S3 Table). The mean concentration of BMAA in the stranded dolphin brains was 1.4-fold higher than in reference brains of patients with AD and ALS (S1 Table) [18]. In some dolphins, BMAA levels detected were up to 3.6-fold greater than found in those of AD and ALS reference brains (S1 Table). The levels of BMAA and the BMAA isomers (AEG and DAB) were not region-specific and the mean concentration of BMAA, AEG, and DAB did not differ in auditory or visual cortex (Table 2). DAB was detected in 7 of 7 (100%) of dolphins from Massachusetts. The mean concentration of DAB detected is 2.4-fold higher than BMAA averaged across all dolphins. AEG was detected in 6 of 7 (86%) of dolphins stranded in Massachusetts. The mean concentration of AEG was 4-fold lower than BMAA and 9.4-fold lower than DAB (P<0.0001; ANOVA). The detection of BMAA and BMAA structural isomers in the dolphin brain was confirmed using LC-MS/MS in representative animals sampled from the cohort (S4 Table).

Neuropathology in stranded dolphins
Gross examinations of external brain structures were unremarkable for all stranded dolphins examined. Brain structures showed normal anatomic landmarks with normal gyral and sulcal patterns (Fig 2A). No atrophy, white matter lesions, infarcts or hemorrhages were observed on gross inspection of coronal sections of the brain. Internal gray and white matter structures were also unremarkable, with clear delineation. The lateral, third and fourth ventricles were normal in size (Fig 2B). Microscopic examination of H&E sections from the ACtx, VCtx and Md tissue sections revealed evidence of cellular injury and age-related changes (see below). Microscopic examination showed hypoxic, shrunken, and eosinophilic neurons in upper and lower cortical layers and within brainstem nuclei (Fig 2C). Gliosis, mild neuronal satellitosis and neuronal atrophy were present in cortical layers and brainstem (Fig 2D). Age-related histological changes included accumulation of lipofuscin, chromatolysis, and corpora amylacea ( Fig 2E-2G). In addition, moderate numbers of eosinophilic plaques were seen ( Fig 2H) and edema was observed throughout parts of the cerebral cortex and brain stem of all dolphins (Fig 2C-2H). The pathological changes described above were present in all dolphins in varying degrees regardless of age, sex or BMAA concentrations.

Aβ + Pathology
Anti-Aβ staining showed Aβ + plaques, ranging in occurrence from rare to very frequent, and intracellular localization of Aβ in the cerebral cortex of all stranded dolphins (Fig 3A-3C). Aβ + plaques were observed throughout the ACtx and VCtx. Rare to sparse Aβ + plaques were localized adjacent to neurons in the Md region of the brainstem (Fig 3D). The Aβ + plaques varied from focal clusters of well-defined plaques to diffused and poorly organized plaque-like structures. Aβ + plaques were observed on both routine H&E and MB silver stained sections ( Fig  3E-3P). Most Aβ + plaques were senile (diffused) plaques, but rare argophyllic compact neuritic plaques and neuropil threads were observed in the cerebral cortex (Fig 4A-4C). Intracellular silver staining neurofibrillary tangles were observed in cortical neurons with dystrophic neurites, indicating the presence of intracellular inclusions of insoluble proteins (Fig 4D-4F). Thioflavin-S staining was used to confirm the Aβ + staining of plaques and neuronal inclusions (Fig 5A-5H). GFAP + fluorescent immunohistochemistry (IHC) provided evidence of reactive astrocytes around thioflavin-S + plaques (Fig 5A-5D). Semi-quantitative analysis showed the densities of Aβ + plaques were positively correlated in the ACtx and VCtx (n = 7; r = 0.89; P<0.05) but, was 1.3-fold denser in the ACtx than in the VCtx (P = 0.01; t-Test) ( Table 3, S3  Table). Dolphins with Brucella spp. infections had fewer Aβ + plaques when compared to dolphins without Brucella spp. infections (P = 0.02; t-Test) (S3 Table). The size of the Aβ + plaques did not differ between the two cortical regions. Neither age group nor sex of dolphins affected the Aβ + plaque deposition (S3 Table). Nor was Aβ + plaque deposition correlated with brain BMAA concentration (S3 and S5 Tables).

Discussion
As the world's climate warms, HABs are becoming more frequent, including in eastern China, which has seen some of the largest cyanobacterial blooms on Earth, and North America. Cyanobacteria produce powerful cyanotoxins that impact aquatic and terrestrial life. Exposures to cyanotoxins are a public health concern as they are linked to organ system damage and disease.
Examining the levels of the cyanobacterial toxin BMAA in apex predators, such as dolphins and sharks, provides a powerful bio-indicator of the potential for human exposures [26,27,49]. BMAA is produced by cyanobacteria that are sometimes seen as surface blooms, but can occur in the water column and in benthic mats in lakes, shallow estuaries and bays. In marine systems, BMAA enters the food chain via crustaceans and bottom-feeding fish, and then accumulates into long-lived apex predators like sharks [26,27]. We now report that another apex predator, dolphins, when exposed to cyanobacteria or diets of crustaceans and fish containing BMAA, bioaccumulate the toxin in their neuroproteins. In our study, BMAA was present in the brains of stranded dolphins at 1.4-fold greater amounts than in the brains of patients with AD and ALS [18]. Dolphins beached in Florida were observed to have nearly 3-fold higher concentrations of BMAA in their brains than dolphins beached in Massachusetts. The differences in BMAA concentration may be due to differences in diet, occurrences of algal blooms, varying phytoplankton species or modes of stranding [56].
The recognition that the very high incidence of neurodegenerative disease Guam ALS/PDC was likely due to the production of BMAA by endophytic cyanobacteria resident in specialized coralloid roots of cycad trees led to the hypothesis that BMAA is a cause of ALS/PDC and non-endemic ALS, and provided a link to neurodegenerative disorders [19,20]. In Guam, cycad seeds contain BMAA and are used to make flour by the indigenous Chamorro. BMAA is also biomagnified up the food chain to animals like pigs, deer and flying foxes that are part of the Chamorro diet [20,50,57]. Once ingested, BMAA can cross the blood-brain barrier and become incorporated into proteins where it is associated with neuropathological changes as seen in Guam ALS/PDC [12,20]. In this paper, we have now shown that BMAA is detectable in brains of dolphins that also show neuropathological changes characteristic of human neurodegenerative disease. Here, BMAA is also shown to accumulate with increasing length and age class. Thus, our data suggest dolphins may provide a naturalistic model of BMAA toxin exposures in marine environments. The effects of this exposure however are not yet clear.
Dolphins have a highly evolved cerebral cortex that underlies a sophisticated sonar navigation system [58]. Previous studies have shown that dolphins have age-related Aβ + deposits in their brains [58,59]. Aβ + plaques, a diagnostic hallmark of AD, were observed to be widespread in the auditory and visual cortex and to a lesser extent the brainstem of the stranded dolphins examined in our study. Numerous Aβ + intracellular inclusions and dystrophic neurites were also observed. The increased deposition of Aβ + plaques with accompanying histological changes (e.g. lipofuscin, corpora amylacea) suggest cellular brain aging. Accumulation of BMAA in the brains of dolphins may exacerbate these age related changes. This observation also raises concern about the potential risk to human health of chronic exposure to BMAA associated with increasingly frequent algal blooms in surrounding lakes and coastal waters.  Lastly, Marine mammals and fish have long been known to concentrate methylmercury (MeHg). Many water bodies have both high concentrations of mercury and frequent algal blooms. High consumption of fish has been linked both to increased blood levels of mercury and to higher incidence of neurodegenerative disease [60]. The co-occurrence of MeHg and BMAA has synergistic neurotoxicity compared to single exposures [61] and has been suggested as a link to dementia [62]. We detected BMAA in all but one dolphin beached in Florida and Massachusetts. Further studies are needed to investigate the synergistic effect of MeHg on   the dolphin brain. The potential combined neurotoxicity of BMAA and MeHg underscores the impending impact of climate change on the marine food web.

Conclusion
We have detected BMAA in the cerebral cortex of stranded dolphins. We also report Alzheimer-like neurodegenerative changes in the brains of dolphins containing BMAA. The presence of BMAA suggests that dolphins provide an excellent sentinel species for toxin exposures in the marine environment.
Supporting information S1 NeuroBioBank, for providing donated postmortem human brain tissues used in these studies. We gratefully acknowledge Margaret Basile, MS and John Pablo, PhD formerly of University of Miami Department of Neurology for their technical assistance.