The analysis of underivatized β-Methylamino-L-alanine (BMAA), BAMA, AEG & 2,4-DAB in Pteropus mariannus mariannus specimens using HILIC-LCMS/MS

Abstract

β-Methylamino-L-alanine (BMAA) has been identified as the potential cause of the amyotrophic lateral sclerosis/ parkinsonism–dementia complex (ALS/PDC) observed in the Chamorro people of Guam. The principal hypothesis for BMAA exposure and intoxication relies on the biomagnification of BMAA in flying fox specimens ingested by the Chamorro people. Although high levels of BMAA were quantitated in flying fox specimens utilizing liquid chromatography-fluorescence (LC-FL), there have not been any confirmatory analyses conducted to date. Therefore, a method for the tissue homogenization, extraction and direct analysis of BMAA (including BAMA, 2,4-DAB and AEG) was utilized. The approach was applied to mammalian dried skin and hair from various rodent species (negative controls) and archived flying fox (Pteropus mariannus mariannus) specimens. A positive control sample of homogenized mussel (Mytelius edulis) with native BMAA was used to verify the method. It was determined that the direct analysis using HILIC MS/MS required additional quality control in order to allow for the confident identification of BMAA due to the near co-elution of BAMA. BMAA was not present above 0.2 μg g−1 (free fraction) or 2.8 μg g−1 (total fraction) in the flying fox specimens. While analysis did not result in BMAA detection in flying fox or negative control samples, the positive control sample and spiked samples were successfully detected.

1. Introduction

β-Methylamino-L-alanine (BMAA) is a non-essential amino acid that has received significant attention due to research postulating its causal role in four neurodegenerative diseases, including amyotrophic lateral sclerosis/parkinsonism–dementia complex (ALS/PDC), ALS, Parkinsonism, and Alzheimer’s disease (AD) (Bradley et al., 2013; Cox and Sacks, 2002; Papapetropoulos, 2007). Original assertions of this hypothesis are based on the presence of BMAA in food sources (e.g. cycads) of the Chamorro people of Guam, an epidemiological cluster for the disease (Kurland and Mulder, 1954; Mulder et al., 1954; Stanhope et al., 1972) and the observed Parkinson’s-like neurotoxicity symptoms exhibited by Macaque monkeys exposed to high levels of BMAA (Spencer et al., 1987). Derivatized BMAA detected in unwashed cycad (Cycas) has been reported as high as 2657 μg g−1 and confirmed present using direct analysis techniques (Banack and Cox, 2003; Duncan et al., 1989; Rosén and Hellenäs, 2008). However, it was thought the levels found in washed cycad flour and other preparations ingested by the Chamorro people were too low in concentration to cause ALS/PDC (Duncan et al., 1990). Therefore, the focus was shifted to the flying fox (Pteropus mariannus mariannus), another dietary item that was observed to consume cycads (Cox and Sacks, 2002; Monson et al., 2003). The augmented hypothesis included the premise that the flying fox bioaccumulated BMAA through ingestion of cycads, which were then consumed by the Chamorro people, providing a large dose of the compound (Banack and Cox, 2003). All determinations of BMAA in flying fox samples, to date, have been made through high performance liquid chromatography coupled with fluorescence (LC-FL) detection of 6- aminoquinolyl-N-hydroxysuccinimidyl-carbamate (AQC)-derivatized samples, TLC and single quadrupole LC-MS methods (Banack et al., 2006; Banack and Cox, 2003; Murch et al., 2004b).

An important diversion from the original hypothesis, with widespread implications, is the proposition that BMAA in cycads originated from symbiotic cyanobacteria (Nostoc) living in the coralloid roots (Cox et al., 2003). The same non-specific methods used to measure BMAA in flying fox specimens were used for determinations of Nostoc cultures, where 0.3 μg g−1 BMAA was reported present (Cox et al., 2003). Following the positive detections in Nostoc, a variety of free living cyanobacteria cultures tested positive for BMAA (up to 6478 μg g−1 free) using the same approach (Cox et al., 2005). The authors therefore assert that BMAA was a global problem. However, research conducted by other groups has resulted in significant disparities in BMAA findings, with clear differences in reported BMAA presence and concentrations. It has been interpreted that much of the past research on BMAA detection was poorly implemented, not clearly presented and was lacking solid evidence to support author conclusions (Faassen, 2014). This has also been acknowledged in the toxicological aspects of BMAA research (Chernoff et al., 2017; Karamyan and Speth, 2008).

A significant issue with any BMAA analysis, with or without derivatization, is the potential presence of low molecular weight compounds resulting in interference with the target compound. Jiang et al. (2012) reported that of 206 isomers for BMAA, 7 are potentially present in nature and may interfere with AQC-derivatization analysis. This included β-amino-N-methyl-alanine (BAMA); diaminobutyric acid (2,4- DAB; 2,3-DAB; 3,4-DAB), N-2-aminoethylglycine (AEG), 3-amino-2- (aminomethyl)propanoic acid and 2,3-diamino-2-methylpropanoic acid. To date, only BAMA, 2,4-DAB, AEG and BMAA have been analyzed using AQC-derivatization.

AQC-derivatization adds a fluorescence group to target amines, thereby allowing for detection via fluorescence (Cohen and Michaud, 1993). This procedure also increases the molecular weight of amines and allows for better retention in reversed-phase chromatography (Cohen and Michaud, 1993). However, this method is prone to the misidentification of target compounds due to the low specificity of fluorescence detection and the co-elution of other derivatized compounds, resulting in false positive and overestimations in BMAA quantification (Faassen et al., 2012). The selectivity of AQC-derivatized BMAA analysis has been enhanced through the application of mass spectrometry/mass spectrometry (MS/MS) coupled with isomer specific diagnostic ions and high mass resolution. However, not all groups have adopted these practices (Faassen, 2014). The widespread use of derivatization procedures with MS/MS detection has resulted in the detection of BMAA in many different matrices, including field specimens of algae/cyanobacteria, invertebrates (e.g. mollusks, crustaceans) and vertebrates (e.g. fish) (Christensen et al., 2012; Faassen et al., 2009; Jiang et al., 2014b, 2012; Spácil et al., 2010). Evidence pointing to eukaryotic algae (e.g. diatoms) as potential sources of BMAA is convincing, with both sufficiently specific direct and indirect analysis techniques showing BMAA presence (Jiang et al., 2014a; Réveillon et al., 2016, 2015). In contrast, to date, Leptolyngbya PCC 73110 (laboratory culture) has been the only cyanobacterial strain found to contain BMAA using highly selective LC-MS/MS techniques (Jiang et al., 2013; Spácil et al., 2010). This has yet to be confirmed using analysis techniques without derivatization.

Hydrophilic interaction liquid chromatography (HILIC) allows for the separation of small polar compounds that normally do not retain well using reversed-phase chromatographic conditions, removing the need for derivatization to achieve BMAA retention (Buszewski and Noga, 2012). Matrices with reported BMAA using direct analysis are cycads, aquatic invertebrates (e.g. mussels), field-collected cyanobacteria samples and non-axenic diatom cultures (Beach et al., 2015; Faassen et al., 2012, 2009; Kerrin et al., 2017; Krüger et al., 2010; Monteiro et al., 2017; Réveillon et al., 2016, 2015; Rosén et al., 2016). One study reported BMAA presence in field collected cyanobacteria samples using direct analysis, but the only BMAA isomer included in the analysis was 2,4-DAB; with other isomers BAMA (similar retention) and AEG excluded (Faassen et al., 2009). This is also true of the Monteiro et al. (2017) work, where only BMAA was directly analyzed. Since BAMA is so closely retained, analysis without this standard to verify chromatographic separation could result in the misidentification of BMAA for BAMA, which exhibit similar fragmentation (Beach et al., 2015). Beach et al. (2015) implemented differential mobility spectrometry in conjunction with HILIC LC-MSMS (HILIC–DMS–MS/MS), allowing for additional confidence when differentiating the near coeluting isomers of BAMA with BMAA. Capillary electrophoresis has also been employed for direct analysis, including BAMA (Kerrin et al., 2017). The separation of these and other isomers is an important aspect of BMAA analysis, with or without derivatization (Beach et al., 2015; Jiang et al., 2012). Currently, isomers detected in shellfish by direct analysis include, BAMA, BMAA, 2,4-DAB, AEG and 3,4-DAB (Beach et al., 2015; Kerrin et al., 2017).

BMAA analysis can be conducted on three different fractions (free, bound soluble, and bound precipitated forms) or as total (Faassen et al., 2016; Rosén et al., 2016). Free BMAA extraction is rather straightforward, where an extractant solvent (typically TCA) is used followed with sample clean-up and analysis. The soluble bound fraction requires additional treatment (e.g. hydrolysis) to release the bound BMAA. Acid hydrolysis can be used on the pellet as well to account for ‘protein bound’ fractions. Conversely, hydrolysis can be conducted on a sample without extraction accounting for total BMAA (bound & free fractions), followed by cleanup and analysis. It has been shown that BMAA can be formed via hydrolysis from sources that are not proteins (bound soluble fractions) and may even be an artefact of the hydrolysis procedure, which requires additional attention to methods implemented to achieve BMAA values (Beach et al., 2018; Rosén et al., 2016). Significant interference from hydrolyzed samples due to an increase in small molecular compounds freed from protein bound fractions requires additional quality control (i.e. spikes) to verify analytical observations.

Sample preparation, including homogenization, extraction, and clean-up techniques, can add variability to results. In general, a reduction in steps can help improve reproducibility, but can sometimes compromise the analysis due to matrix effects. Therefore, samples require the addition of an internal standard and/or standard addition for accurate calibration and quantitation. In this work, total and free fractions were analyzed in a positive control sample (homogenized mussel), flying fox dried skin/hair specimens and negative control skin/ hair specimens from rodents. The purpose of this research was to further support the direct analysis of BMAA and its isomers and, through the analysis of flying fox specimens, to allow researchers to make headway in understanding the complex proposed etiology of ALS/PDC.

2. Materials and methods

2.1. Samples

The flying fox specimens used in this work were acquired from the Museum of Vertebrate Zoology, University of California (Berkeley, CA, USA). The ventral skins and fur were obtained from the same area as the identical desiccated, preserved specimens analyzed by Banack and Cox (2003); accession numbers 114606, 114607 and 114609. The negative controls were also collected from desiccated, preserved samples of skin from the ventral area.

The reference material CRM-Asp-Mus-d (homogenized mussel tissue from Mytelius edulis; Lot #20112) was purchased from the National Research Council of Canada Certified reference materials program as a positive control. The homogenates were lyophilized to dryness prior to extractions. All the samples used in the current research are listed in Table 1.

Table 1.

The samples used in method development and analyzed in this study.

Sample ID	Species	Common Name	Sample Weight
Negative Control Samples (dried skin with fur)
312	Neotoma fuscipes	Dusky-footed wood rat	25 mg
AT12	Sciurus niger	Fox squirrel	44 mg
Flying fox specimens (dried skin with fur)
114606	Pteropus mariannus mariannus	Flying Fox	66 mg
114607	Pteropus mariannus mariannus	Flying Fox	58 mg
114609	Pteropus mariannus mariannus	Flying Fox	32 mg
Positive Control Sample (mussel homogenate)
CRM-Asp-Mus-d	Mytilus edulis	Blue Mussel	8 g (w.w.)

2.2. Sample preparation

2.2.1. Homogenization

2.2.1.1. Negative control samples & flying fox samples.

The samples were maintained frozen at −20 °C until processing. Each sample was cut into approximately 2 mm sections and placed in a 7-mL tube with 2.8 mm ceramic beads (6.3 g). An Omni Bead Ruptor 24 (Omni International Inc., Kennesaw Georgia) was used to homogenize samples. In order to fully homogenize both skin and hair, a two-step process was used. The samples were homogenized initially dry (6.8 m/s, 30 s) to pulverize the hair. Three additional homogenization steps (6.8 m/s, 30 s) were used after the addition of 0.1 M trichloroacetic acid (TCA; > 99.5% purity, Fisher Scientific, Waltham, MA) at sample concentrations of 5–20 mg mL−1, depending on the initial weight. In between cycles, samples were allowed to cool (placed in refrigerator, 10 min) to mitigate heat generation. After homogenization, aliquots were removed for free and total BMAA procedures. Aliquots were stored at −20 °C until extractions were conducted.

2.2.1.2. CRM-Asp-Mus-d.

The positive control sample was used to verify destruction of native BMAA and its isomers did not occur during homogenization and was assessed as hydrolyzed material. The protocol used for skin/hair samples was applied to the lyophilized mussel tissue. A 50 mg subset of the CRM-Asp-Mus-d sample was transferred to a 7 mL homogenization vial and spiked with d3-BMAA (50 μg g−1 ; Abraxis Kits, Warminster, PA, USA). Dry homogenization was conducted followed with the addition of TCA (2.5 mL of 0.1 M) and 3× wet homogenizations, as with flying fox samples. The slurry was transferred to 6 × 50 mm glass hydrolysis tubes in 1 mg (50 μL) subsets (n = 8).

For comparison, vortex mixed material was prepared without using the bead ruptor. An aliquot of 0.1 M TCA (2.5 mL) was added to 50 mg of sample, one with d3-BMAA spiked at 50 μg g−1 and another spiked at 10 μg g−1 so that final d3-BMAA concentration in solution matched that of the external standard curve d3-BMAA concentration (100 ng mL−1 ). The samples were vortex mixed, and subsets removed as 1 mg (n = 4; 50 μL) from the 50 μg g−1 d3-BMAA spiked material and as 5 mg (n = 4; 250 μL) from the 10 μg g−1 d3-BMAA spiked material.

2.2.2. Free BMAA extraction

Table 2 shows the replicate extractions and pre-extraction spikes conducted on the samples. β-N-Methylamino-L-alanine hydrochloride was purchased from Tocris Bioscience (Bio-Techne, Minneapolis, MN, USA), β-amino-N-methylalanine from Enamine (Kyiv, Ukraine), N-(2- aminoethyl)glycine from TCI America chemicals (Portland, OR, USA), and 2,4-diaminobutyric acid dihydrochloride from Sigma–Aldrich (St. Louis, MO, USA). The 114609 sample was limited and only two free extractions were conducted (one spiked), in contrast to the remaining flying foxes which had 3 extractions each (one spiked per sample).

Table 2.

Sample replicates and spikes conducted for the free and total (hydrolyzed) fractions. All the samples were analyzed with HILIC LC-MS/MS.

	Sample and total number aliquots prepared (n)	#Sample Replicates	#Spiked Replicates	BMAA, BAMA, 2,4-DAB & AEG Fortified Levels pg g−1	d3-HMAA
FREE	Neotoma fuscipes (n = 3)	1	2	10 & 50	nonea
	Sciurus niger (n = 3)	1	2	10 & 50	nonea
	114606 (n = 3)	2	1	0.5	sample & spike
	114607 (n = 3)	2	1	1	sample & spike
	114609 (n = 2)	1	1	2	spike only
	CRM-Asp-Mus-d (n = 7)	3	4	range = 0.01–2.00	all
TOTAL	Neotoma fuscipes (n = 4)	1	10, 10 & 20	10 & 20 spikes
	Sciurus niger (n = 4)	1	10, 10 & 25	10 & 25 spikes
	114606 (n = 4)	1	100 & 500	sample & 100 spike
	114607 (n = 4)	2	50 & 300	sample & 50 spike
	114609 (n = 3)	1	250 & 1000	250 spike
	CRM-Asp-Mus-d (n = 12) as 1 mg	6	range = 5 – 100	all
	CRM-Asp-Mus-d (n = 4) as 5 mg	2	5 & 20	all

^aInternal Standard not available at the time of sample preparation.

The samples were extracted in the following manner: 10 mg (original dry weight) of each homogenized hair/skin sample was dispensed into 15 mL glass centrifuge vials. The CRM-Asp-Mus-d samples were prepared as 50 mg subsets and vortex mixed followed with bath sonication (10 min) in 2.5 mL 0.1 M TCA. All samples were cooled at −4 °C for 10 min prior to centrifugation at 1500 g for 10 min. The supernatants were retained and the pellets were rinsed with 0.1 M TCA (1–1.5 mL), cooled again, and centrifuged. The pooled supernatants were subjected to solid phase extraction (SPE) using 150 mg Oasis MCX (Waters Corporation, Milford, MA). Columns were conditioned with 100% MeOH (5 mL), equilibrated with 0.1 M TCA (5 mL), loaded with sample, rinsed with 0.1 M HCl (2 mL), rinsed with 100% MeOH (2 mL) and eluted with 5.5% of NH4OH (certified ACS plus - 28–30 w/w %) in MeOH (4 mL). Elutions were blown to dryness (N2; 60 °C) and the samples were reconstituted with 50% acetonitrile in 1% acetic acid. The samples were filtered through 0.45 μm PVDF prior to analysis at sample concentrations of 20 mg mL−1 (skin/hair) and 50 mg mL−1 (CRM-AspMus-d).

2.2.3. Total BMAA hydrolysis and extraction

The samples were prepared in the following manner: 1 mg of each homogenized sample was dispensed into 6 × 50 mm glass vials and lyophilized to dryness; with exception to the 4 positive control samples assessed at 5 mg. The 5 mg samples were prepared in order to evaluate the impact higher matrix had on the hydrolysis and analysis methods as implemented. All replicates and spikes were prepared in the same manner (Table 2). Each reaction vessel, holding up to 12 vials, contained a standard control (standards ranging from 10 to 100 ng each of BMAA, BAMA DAB and AEB) and a blank vial (10 μL deionized water). The dried vials were placed in a reaction vessel after 30 μL of 6 M HCl (constant boiling) was added to each vial. Hydrolysis was conducted for 20–24 h under vacuo at 105 °C using an Eldex Hydrolysis/Derivatization Work Station (Fig. S1) (Eldex Laboratories Inc., Napa, CA, USA). Once cooled to room temperature, the vacuum was released; vials were wiped clean and allowed to dry under vacuum. The controls (blanks and standards) were reconstituted in 500 μL 50% ACN in 1% acetic acid and analyzed using LC-MS/MS to verify losses did not occur during the lyophilization and hydrolysis procedures and that cross-contamination did not occur.

The samples were reconstituted in 0.1 M TCA and quantitatively transferred to 15 mL centrifuge tubes (total volume = 2 mL). SPE (MCX-cation exchange) was conducted in the same manner as the free BMAA samples. Elutions were blown to dryness and the samples were reconstituted with 50% acetonitrile in 1% acetic acid. The samples were filtered through 0.45 μm PVDF prior to analysis at a sample concentration of 2 mg mL−1, except the 5 mg samples, which were reconstituted to 10 mg mL−1.

2.3. Analysis techniques

2.3.1. HILIC MS/MS – direct analysis

Liquid chromatography coupled with mass spectrometry/mass spectrometry (LC-MS/MS) was utilized for the determination of BMAA, BAMA, 2,4-DAB, AEG and d3-BMAA. Separation was achieved using a SeQuant ZIC-HILIC column (EMD Millipore, Billerica, MA). Injection volumes were 5 μL with a flow rate of 350 μL min−1. Mobile phase A consisted of 0.1% formic acid in water and mobile phase B was 0.1% formic acid in acetonitrile. This study began with the analysis of negative control and flying fox samples (Table 1). After the initial analysis, a year passed prior to analysis of the CRM-Asp-Mus-d control sample. In that time, the purchase of a new ZIC-HILIC column required modification of HPLC conditions to maintain isomer separation. The two gradient conditions are found in Table 3. A Thermo Scientific Surveyor HPLC system coupled with a LTQ XL^™ Linear Ion Trap Mass Spectrometer (Thermo Fisher Scientific, Waltham MA, USA) was used in analysis. The [M+H]+ ion for BMAA (m/z 119) was fragmented using a normalized collision energy of 35% and the product ions (m/z 73, 76, 91, 101 and 102) were monitored. The [M +H]+ ion for the internal standard d3-BMAA (m/z 122) was fragmented using a normalized collision energy of 40% and the product ion (m/z 105) was monitored. Spectra were obtained with an isolation window of 1.0 Da, capillary temperature was set to 275 °C and source voltage was 5 kv. Thermo Xcalibur 2.2 (Thermo Fisher Scientific, Waltham MA, USA) software was utilized for instrument control, data acquisition and processing.

Table 3.

The chromatographic gradients used for ZIC-HILIC separation at the beginning of the study (Gradient#1), and the second set of conditions used with a new column months later (Gradient#2). Mobile phase A = 0.1% formic in water and mobile phase B = 0.1% formic in acetonitrile.

Gradient#1			Gradient#2
Time (min)	% A	% B	Time (min)	% A	% B
0	10	90	0	5	95
5	10	90	19	40	60
20	50	50	25	60	40
23	10	90	27	5	95
25	10	90	30	5	95

2.3.1.1. Linearity, method detection limits and interpretation.

Linearity (BAMA, BMAA, 2,4-DAB & AEG) was assessed from 1 to 400 ng mL−1 using external standard curves, from 5 to 1000 μg g−1 for hydrolyzed matrix curves and from 0.01 to 50 μg g−1 for free matrix curves. Method detection limits (MDLs) were determined by using a S:N ratio of 3 coupled with matrix spike observation. The method quantification limits (MQL) were set at a S:N ratio of 10. Differentiation of isomers was made by retention time, confirmed with matrix spikes (including d3- BMAA) and relative abundance of fragment ions. The quantification ions used in extracted ion chromatograms for interpretations were 102 m/z (BMAA & BAMA), 101 m/z (2,4-DAB) and 101 &102 m/z (AEG). Quantitation was achieved using standard addition curves, in addition to the internal standard method for BMAA using d3-BMAA.

3. Results and discussion

3.1. Homogenization

The flying fox and rodent samples required additional processing before they could be used in extractions. The samples were comprised of dry skin with hair attached, making them difficult to homogenize. Neither traditional glass tissue grinders nor mortar and pestle approaches provided sufficient homogenization. Therefore, a bead ruptor was employed and optimized prior to extractions. Hair was pulverized dry prior to adding solvent to homogenize the skin. This approach resulted in a homogenized mixture that could be reproducibly aliquoted for experiments (Fig. S2).

The positive control sample (mussel) did not require homogenization but was used to verify that the procedure did not compromise the analytes of interest. When comparing the use of a bead ruptor to vortex mixing the lyophilized mussel tissue in the same volume of TCA, there was not a statistical difference when using a student t-test in calculated BMAA (p = 0.9165), BAMA (p = 0.9889), 2,4-DAB (p = 0.4576) or AEG (p = 0.3977) concentrations (df = 4). Additionally, the prehomogenization d3-BMAA spike was returned 66 ± 6% for the bead ruptor processed subset and 69 ± 4% for vortexed subset. A comparison of student t-test achieved data presented as box plots for total BMAA, BAMA, 2,4-DAB and AEG measurements can be viewed in supplementary data Fig. S3.

Although rarely elaborated upon, homogenization and extraction procedures for complicated matrix analysis is a crucial part of any analytical work. For instance, in Banack and Cox (2003), it was stated that the samples were rehydrated in TCA and macerated, but no other details are provided on how this was achieved. Additionally, little is known on how homogenization for similar samples was achieved in the Banack et al. (2006) or Murch et al. (2004a, b) work. This provides additional challenges when relating analytical data from one study to another, even when similar sample types are processed.

3.2.1. Standards

The chromatographic separation of isomers was similar to other work (Beach et al., 2015; Réveillon et al., 2014). The baseline separation of BAMA and BMAA was not fully achieved in this work and changes in retention time were observed, both resulting from matrix and inter-day variability. Table 4 lists the retention times and standard deviations (minutes) observed in both legs of this work, with Fig. 1 showing representative chromatograms of standards used to calibrate direct analysis (Table 3). Matrix spikes (including d3-BMAA) and spectra (product ion ratios) added confidence to BMAA and isomer identification. The MS/MS spectra of standards, matrix spike (flying fox) and positive control (native BMAA) can be viewed in supplementary material (Figs. S4 – S5). External and a matrix (total-hydrolyzed) uncorrected standard curves can be viewed in supplementary data Fig. S6.

Table 4.

The retention times (min) with standard deviations observed for both gradients used in the two sections of this work, separated by a year and described in the methods section (Table 3). Variability (matrix derived and interday) was observed but was mitigated with the use of the internal standard d3-BMAA and matrix spikes.

	BAMA (min)	BMAA (min)	2,4-DAB (min)	AEG (min)	d3-BMAA (min)
Gradient#1	18.54 ± 0.29	18.75 ± 0.30	19.93 ± 0.26	21.13 ± 0.28	18.78 ± 0.31
Gradient#2	19.93 ± 0.36	20.16 ± 0.44	21.76 ± 0.51	23.09 ± 0.45	20.18 ± 0.40

Fig. 1.

LC-MS/MS chromatograms (direct analysis) of BAMA, BMAA, 2,4-DAB and AEG (m/z 119 → 101, 102) and d3-BMAA (m/z 122 → 105) standards at 100 ng mL−1 using Gradient #1 and Gradient #2 (described in Methods Table 3).

3.3. Free BMAA

3.3.1. Positive control – CRM-Asp-Mus-d

The positive control mussel sample contained free BMAA, BAMA, 2,4-DAB, AEG and two unknowns (Unknown A & Unknown B) (Fig. 2). This material is reported to also contain 3,4-DAB in addition to other unidentified compounds (Beach et al., 2015; Kerrin et al., 2017). Data derived from this material helped support the feasibility of this method to be used in direct analysis for BMAA and highlights the additional need for standard addition and/or the use of an internal standard. The Fig. 2 chromatogram was achieved using Gradient #2 (Table 3) and demonstrates the observed isomers detected in all 3 replicates of extracted material, which was analyzed at a sample concentration of 50 mg mL−1. Standard addition curves (ranging from 0.01 to 2.00 μg g−1 ) were used in quantitation and can be viewed in Figs. S8–S9.

Fig. 2.

Direct LC-MS/MS analysis of the CRM-Asp-Mus-d positive control sample showing FREE extracted BMAA and isomers. The BAMA and AEG levels were determined to be below the limits of quantification (Table 5).

3.3.2. Negative controls

The negative control samples (Scurius & Neotoma) were below detection for free BAMA, BMAA, AEG and 2,4-DAB. An example chromatogram of the non-spiked sample and 10 μg g−1 spiked sample can be viewed in Fig. 3.

Fig. 3.

Chromatogram of a negative control sample (Sciurus) extracted for FREE BMAA and the same sample spiked pre-extraction at 10 μg g−1 with BAMA (18.15 min), BMAA (18.33 min), 2,4-DAB (19.49 min), and AEG (20.69 min). The TIC chromatograms in the above windows represent the m/z 119 → 73, 76, 91, 101, 102.

3.3.3. Flying fox samples

Levels of BAMA, BMAA, 2,4-DAB and AEG were below detection in the flying fox samples (114606, 114607 & 114609). An example chromatogram of the 114607-sample fortified (pre-extraction) with the internal standard d3-BMAA can be viewed in Fig. 4. Although all three specimens were previously reported to have high free BMAA levels (1287–7502 μg g−1 ) when analyzed post AQC-derivatization with LCFL (Banack and Cox, 2003), BMAA was not detected using direct analysis. The MDLs for this matrix using direct analysis was 0.20 μg g−1, less than four orders of magnitude than the levels reported to be present (Table 7).

Fig. 4.

Free BMAA analysis of the flying fox sample 114607 (analyzed using direct LC-MS/MS, with the extracted ion chromatogram (EIC) for BMAA (m/z 119 → 102; top) and d3-BMAA (bottom). If present, free BMAA would be observable in the top chromatogram at the same retention time as d3-BMAA. This specimen was below detection for BMAA (< 0.20 μg g−1 ) even though it was previously reported to contain 7502 μg g−1 of free BMAA when AQC derivatized and analyzed using LC-FL.

Table 7.

The BMAA results from flying fox specimens using data achieved in this study as compared to previous work.

	Banack and Cox (2003)	Current Study
Flying Fox	AQC LC-FL	H0.1C LC-MS/MS
Specimen ID	Free (µg g−1)	Free (µg g−1)	Total (µg g−1)
114606	1879	< 0.20	< 2.8
114607	7502	< 0.20	< 2.8
114609	1287	< 0.20	< 2.8

3.4. Total BMAA

3.4.1. Positive control - CRM-Asp-Mus-d

Total BMAA and isomers BAMA, 2,4-DAB and AEG were confirmed present in the mussel material and quantitated using both standard addition and the internal standard technique (Table 5). A significant difference between hydrolyzing 1 mg vs. 5 mg of material was observed in this work. The 1 mg hydrolyzed sample, regardless of original sample preparation (vortex mixing vs. bead ruptor homogenization) resulted in the detection of all 4 isomers tested, while the use of 5 mg resulted in 2,4-DAB loss (including 5 & 20 μg g−1 spikes) and lower total BMAA, BAMA and AEG. The internal standard (d3-BMAA; recover y= 2%) was either lost during the hydrolysis procedure, extraction or ion suppression lowered the response due to increased matrix released from protein fractions. Beach et al. (2018) found ion suppression post hydrolysis to be significant (65%) due to interferences from proteinogenic amino acids, which likely impacted the higher concentration samples in this work. Since standard addition was used in quantification, ion suppression was not the only factor impacting the results. Additional contributions from the higher sample weights also likely resulted in inefficient hydrolysis and/or losses to extraction. An example chromatogram of the BMAA and isomers detected in the CRM material can be viewed in Fig. S9, with both 1 mg (C) and 5 mg (D) hydrolyzed extracts for comparison. Chromatograms of the four replicate samples can be viewed in the supplementary data (Figs. S10–S13), also showing another unknown that eluted near 15 min in retention. Standard addition curves for the 1 mg hydrolyzed extracts used to calculate total BMAA, BAMA, 2,4-DAB & AEG in the sample can be viewed in Fig. 5.

Table 5.

Summary of BMAA, BAMA, 2,4-DAB and AEG concentrations for all samples analyzed using HILIC LC-MS/MS. The standard addition (SA) method for quantification is reported as the x-intercept and 95% confidence interval. The (IS) internal standard method is reported as the average and standard deviation of replicate extractions. All data is reported as dry weight.

	Sample Prep	Sample	ID	Quant Method	BMAA	BAMA	2,4-DAB	AEG
					μg g−1	μg g−1	μg g−1	μg g−1
FREE	10 mg TCA Extraction, SPE	Flying Fox	114606	IS	ND	ND	ND	ND
		Flying Fox	114607		ND	ND	ND	ND
		Flying Fox	114609		ND	ND	ND	ND
		Neotoma fuscipes	302		ND	ND	ND	ND
		Sciurus niger	AT12		ND	ND	ND	ND
	MDL (μg/g)				0.20	0.15	1.65	0.75
	50 mg TCA Extraction; SPE	lyophilized mussel	CRM-ASP-MUS-D	IS	0.033 ± 0.006
				SA	0.040 ± 0.010	0.011a ± 0.011	2.347 ± 0.415	0.042a ± 0.030
	MDL (μg/g)				0.006	0.010	0.25	0.040
TOTAL	Bead ruptor, 1 mg Hydrolysis, SPE	Flying Fox	114606		ND	ND	ND	ND
		Flying Fox	114607		ND	ND	ND	ND
		Flying Fox	114609		ND	ND	ND	ND
		Neotoma fuscipes	302		ND	ND	ND	ND
		Sciurus niger	AT12		ND	ND	ND	ND
	MDL (μg/g)				2.80	2.00	21.80	2.30
	Bead ruptor, 1 mg Hydrolysis, SPE	lyophilized mussel	CRM-ASP-Mus-d	IS	0.59 ± 0.02
				SA	0.58 ± 0.18	1.94 ± 0.38	12.13 ± 2.54	0.57 ± 0.21
	MDL (μg/g)				0.05	0.03	1.50	0.15
	Vortex, 5 mg Hydrolysis, SPE	lyophilized mussel	CRM-ASP-Mus-d	IS	0.47a ± 0.03
				SA	0.46a ± 0.22	1.55 ± 1.28	ND	1.16a ± 0.19
	MDL (μg/g)				0.18	0.10	>20	0.15

IS = Internal Standard (d3BMAA) method was used in quantification.

SA = Standard Addition was used for quantification (Figs. S7 and S5).

ND = not detected above the method detection limit (MDL).

Method quantification limit (MQL) = 3.3^a MDL.

^aAbove the MDL below the QL, shown for comparison purposes only.

Fig. 5.

Standard addition curves of the hydrolyzed mussel sample for the 1 mg hydrolyzed CRM-Asp-Mus-d positive control sample. Concentrations relating to the xintercept are in Table 5. BMAA, BAMA and AEG curves shown above include the non-spiked replicates (n = 4) & 5 μg g−1 replicate spikes, with higher spiked concentrations not shown.

Total BMAA was quantitated at 0.59 ± 0.02 μg g−1 using standard addition and 0.58 ± 0.18 μg g−1 using the internal standard technique (corrected concentration using matrix fortified d3-BMAA and an external BMAA/d3-BMAA curve). Previously, BMAA was reported in this material at 1.2 ± 0.2 μg g−1 using direct LC-MS/MS and at 1.7 ± 0.2 μg g−1 using capillary electrophoresis – MS/MS (Beach et al., 2015; Kerrin et al., 2017). Differences in quantitation may be due to the source of the standard used to quantify. Since there currently is not a certified reference standard of BMAA, standards provided by manufacturers may not be accurate in their reported quantities. This could be resolved in future work when a certified reference standard becomes available. The same quantification issues are likely prominent with the isomers as well. Although certified standards are of great importance to accurate assessments, this method is more likely to be impacted by the employed hydrolysis procedure. It has been demonstrated that BMAA in shellfish is released from two compartments, one that was rapidly released during initial hydrolysis conditions (after 0.5 h) and a second source that results in the release and/or artefactual formation of BMAA over time (Beach et al., 2018). The authors did not observe a maximum BMAA concentration, even over a 120-h period, which may help explain quantitative variability of the same sample between studies. The confirmed presence of BMAA and other isomers in the positive control sample supports the current implemented method for BMAA detection in biological matrices.

3.4.2. Negative controls

The negative control samples (Scurius & Neotoma) were below detection for total BAMA, BMAA, 2,4-DAB and AEG. Matrix spikes were recovered (Table 6) and curves illustrated good linearity to 1000 μg g−1. Matrix curves of all skin/hair samples (negative controls and flying fox) can be viewed in Fig. S14.

Table 6.

Spike recoveries as determined from external curves using HILIC LC-MS/MS.

	Sample Prep	Sample Type	n=	d3-HMAA	n=	BMAA	BAMA	2,4-DAB	AEG
FREE	TCA Extraction, SPE	Mussel (50 mg)	6	58 ± 6%	3	73 ± 29%	51 ± 5%	115 ± 7%	43 ± 7%
		Flying Fox (0 mg)	2	95 ± 0%	3	114 ± 14%	122 ± 13%	120 ± 16%	54 ± 28%
		Sciurus & Neotoma (10 mg)	0	NS	2	93 ± 3%a	93 ± 17%	78 ± 3%	53 ± 23%
TOTAL	Bead Ruptor, Hydrolysis, SPE	Mussel (1 mg)	4	66 ± 6%	2	73 ± 10%	64 ± 13%	85 ± 23%	60 ± 6%
		Sciurus & Neotoma (1 mg)	2	71 ± 2%	2	65 ± 9%	71 ± 4%	72 ± 3%	68 ± 8%
		Flying Fox (1 mg)	6	68 ± 21%	2	89 ± 20%	67 ± 19%	58 ± 9%	102 ± 25%
	Vortex, Hydrolysis, SPE	Mussel (1 mg)	4	69 ± 4%	2	77 ± 5%	68 ± 1%	58 ± 32%	65 ± 3%
		Mussel (5 mg)	4	2 ± 3%	2	2 ± 7%	44 ± 13%	ND	67 ± 6%
	Hydrolysis	Standard Controls	6	NS	6	82 ± 23%	95 ± 29%a	98 ± 27%	98 ± 22%
		Negative Controls	6	ND	6	ND	ND	ND	NDb

ND = Not detected above the MDL.

NS = Not Spiked.

^an = 3.

^b= trace amount detected below MDL.

3.4.3. Flying fox samples

Total BMAA and its isomers were not detected above the method detection limits in the flying fox samples (MDLBMAA = 2.80 μg g−1 ). Chromatograms of the sample 114606 fortified with d3-BMAA and spiked with BMAA, BAMA, 2,4-DAB, AEG at 100 μg g−1 can be viewed in Fig. 6. Method detection limits for the skin/hair matrix were an order of magnitude higher than those determined for the mussel samples. All chromatograms relating to hydrolyzed fractions of flying fox samples, including spikes, can be viewed in supplementary data (Figs. S15–S17). Banack and Cox (2003) reported on BMAA measured in three 50- year-old dried preserved specimens of Pteropus mariannus mariannus from the Museum of Vertebrate Zoology (University of California, Berkeley). Levels of free BMAA were reported at 1879, 7502 & 1287 μg g−1 for flying fox samples identified as 114606, 114607, 114609, respectively (Table 7). The original BMAA analysis was conducted using LC-FL of AQC-derivatized material, other isomers (e.g. BAMA, DAB) were not analyzed, and standard addition (if used) was not described. A limit of detection was reported at 0.00013 μmol per injection, without specifics on injection volume. Additional analyses conducted in the Banack and Cox (2003) study (thin layer chromatography and single quadrupole mass spectrometry) were not utilized in quantitation and not adequately described to interpret or reproduce. The only other work conducted to date that supports BMAA presence in flying fox specimens employed the same non-specific methodologies (Banack et al., 2006; Murch et al., 2004b). The original AQC-derivatization and LC-FL analysis method was briefly explored in this work, but it was determined that LC-FL of AQC derivatized material (as implemented) was inadequate for the determination of BMAA. The lack of sensitivity in the matrix (higher detection limits than direct analysis), lengthy chromatographic runs (65 min), poor spike recoveries (32 ± 12% for free BMAA), low specificity and difficulty with interpretation (retention time shifts, interfering peaks) resulted in low confidence in the data. The LC-FL method and data can be viewed in the supplementary data; Figs. S18–22 & Table S1. The analysis of BMAA in flying fox specimens using HILIC-LC-MS/MS did not support the original findings. Since the AQC-derivatization and analysis of BMAA has been shown to produce false positive data in other work (Faassen et al., 2012), the possibility of false positive data from the original flying fox specimen analyses cannot be ignored.

Fig. 6.

The flying fox sample 114606 (E) and matrix spike (F) hydrolyzed and analyzed using direct LC-MS/MS, both fortified with d3-BMAA. BMAA, BAMA, 2,4- DAB, and AEG were below detection (E). The second set of chromatograms (F) show the matrix spike fortified with BAMA (18.29 min), BMAA (18.49), 2,4-DAB (19.70 min), and AEG (20.92 min) at 100 μg g−1. All fortifications were made prior to hydrolysis.

4. Conclusion

BMAA has been reported to be present in cyanobacteria, diatoms, cycads, flying fox specimens, shellfish and in the brains of people who suffered from neurodegenerative disease (Banack and Cox, 2003; Beach et al., 2015; Cox et al., 2005; Faassen et al., 2009; Jiang et al., 2014a; S. Murch et al., 2004a). However, due to the varying levels of specificity in analytical approaches used to identify BMAA, there is no definitive data on BMAA levels in the environment or in animal tissues (Faassen, 2014). The analytical approach for BMAA detection requires the implementation of appropriate quality controls (e.g. matrix spikes) to validate observations made by researchers. Furthermore, the methods employed need to be adequately described in a transparent manner in order to provide other researchers the information needed to interpret and reproduce experiments. Some of the studies supporting BMAA’s role in ALS/PDC were conducted using non-specific methods, and have not been reproduced independently with sufficiently specific methodology (Banack et al., 2011, 2006; Banack and Cox, 2003; Cox et al., 2005; S. Murch et al., 2004a). These non-specific methods have also been applied to brain tissue samples from Chamorro ALS/PDC patients from Guam, with high ppm levels of BMAA reported (Cox et al., 2003; S. Murch et al., 2004a). The report of BMAA in the brains of ALS/PDC patients, coupled with high levels measured in flying foxes has shaped the foundation of the current BMAA based etiology for neurodegenerative disease. Because the data utilized to assert these hypotheses are based on non-specific methods prone to false positive data, it is necessary to independently confirm the observations of the original work using more selective methodologies, such as those that do not employ derivatization.

However, there are caveats to analyzing BMAA without derivatization. The near co-elution of BAMA and BMAA using HILIC separation requires additional quality control and transparency in reporting BMAA and BAMA positive samples, such as shellfish (Jiang et al., 2012; Réveillon et al., 2015). The analysis of a positive control sample containing native BMAA and BAMA supported the need to employ the use of internal standards and standard addition techniques. It also reinforced the presence of BMAA in shellfish, in addition to at least 3 isomers, reiterating the importance of expanding upon existing approaches to include other potential isomers present in samples currently being tested for BMAA. The use of techniques involving high-field asymmetric waveform ion mobility spectrometry (FAIMS) and differential mobility spectrometry (DMS) should be considered in future experiments, as implemented by Beach et al. (2018, 2015). This ion filter was not available for the present study, so additional quality controls were required to verify interpretations were conducted accurately.