MS

Accepted Manuscript Title: A validated method for quantifying hypoglycin A in whole blood by UHPLC-HRMS/MS Author: J´er´emy Carlier J´erˆome Guitton C´ecile Moreau Baptiste Boyer Fabien B´evalot Laurent Fanton Jean Habyarimana Gilbert Gault Yvan Gaillard PII: DOI: Reference:

S1570-0232(14)00734-X http://dx.doi.org/doi:10.1016/j.jchromb.2014.11.029 CHROMB 19220

To appear in:

Journal of Chromatography B

Received date: Revised date: Accepted date:

16-7-2014 27-10-2014 29-11-2014

Please cite this article as: J. Carlier, J. Guitton, C. Moreau, B. Boyer, F. B´evalot, L. Fanton, J. Habyarimana, G. Gault, Y. Gaillard, A validated method for quantifying hypoglycin A in whole blood by UHPLC-HRMS/MS, Journal of Chromatography B (2014), http://dx.doi.org/10.1016/j.jchromb.2014.11.029 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

A validated method for quantifying hypoglycin A in whole blood by UHPLC-HRMS/MS Jérémy Carlier a,b,*, Jérôme Guitton c,d, Cécile Moreau a, Baptiste Boyer e, Fabien Bévalot f, Laurent Fanton g, Jean Habyarimana h, Gilbert Gault i, Yvan Gaillard a Laboratoire LAT LUMTOX, 800 av. Marie Curie Z.I. Jean Jaurès 07800 La Voulte-sur-Rhône, France

ip t

a

b

Ecole Doctorale Interdisciplinaire Sciences-Santé, Université Claude Bernard, Hôpital Louis Pradel, 28 av. du Doyen Lépine 69677 Bron, France

e

f

g

cr

Laboratoire de pharmacologie, Centre Hospitalier Lyon Sud, chemin du Grand Revoyet 69495 Pierre-Bénite, France

us

d

Laboratoire de toxicologie, Faculté de pharmacie de Lyon, 8 av. Rockefeller 69373 Lyon, France

Institut médico-légal, Centre Hospitalier de Clermont-Ferrand, rue Montalembert 63033 Clermont-Ferrand, France Laboratoire LAT LUMTOX, 71 av. Rockefeller 69003 Lyon, France

an

c

Institut médico-légal, Faculté de médecine, 12 av. Rockefeller 69008 Lyon, France

FARAH, Faculté de médecine vétérinaire, Université de Liège, 20 bvd. de Colonster 4000 Liège 1, Belgique

i

VETAGROSUP, Ecole nationale vétérinaire de Lyon, 1 av. Bourgelat 69280 Marcy l’Etoile, France

M

h

Ac ce p

te

E-mail address: [email protected] (J. Carlier).

d

* Corresponding author. Tel.: +33 475 62 05 24; fax: +33 475 85 54 58.

1 Page 1 of 26

ABSTRACT Hypoglycin A (HGA) is the toxic principle in ackee (Blighia sapida Koenig), a nutritious and readily

ip t

available fruit which is a staple of the Jamaican working-class and rural population. The aril of the unripe fruit has high concentrations of HGA, the cause of Jamaican vomiting sickness, which is very often fatal.

cr

HGA is also present in the samara of several species of maple (Acer spp.) which are suspected to cause seasonal pasture myopathy in North America and equine atypical myopathy in Europe, often fatal for horses.

us

The aim of this study was to develop a method for quantifying HGA in blood that would be sensitive enough

an

to provide toxicological evidence of ackee or maple poisoning. Analysis was carried out using solid-phase extraction (HILIC cartridges), dansyl derivatization and UHPLC-HRMS/MS detection. The method was

M

validated in whole blood with a detection limit of 0.35 µg/L (range: 0.8 to 500 µg/L). This is the first method applicable in forensic toxicology for quantifying HGA in whole blood. HGA was quantified in two serum

d

samples from horses suffering from atypical myopathy. The concentrations were 446.9 and 87.8 µg/L. HGA

te

was also quantified in dried arils of unripe ackee fruit (Suriname) and seeds of sycamore maple (Acer

Ac ce p

pseudoplatanus L.) (France). The concentrations were 7.2 and 0.74 mg/g respectively.

Keywords:

Forensic science Hypoglycin A

Ultra-high performance liquid chromatography-high resolution tandem mass spectrometry (UHPLCHRMS/MS) Blighia sapida Koenig

2 Page 2 of 26

Jamaican vomiting sickness Acer pseudoplatanus L.

Ac ce p

te

d

M

an

us

cr

ip t

Equine atypical myopathy

3 Page 3 of 26

1. Introduction

Ackee (Blighia sapida Koenig) is a tree in the Sapindaceae family that grows to height of around 15 metres. It is of West African origin and was imported to tropical regions of the Caribbean and Central America, where it now grows naturally. The

ip t

pear-shaped yellowish-orange dehiscent fruit turns red when it ripens and splits open to reveal two-to-four large shiny black seeds (Fig.1) [1]. The aril of the ripe fruit is a nutritious food, which is a staple of Jamaican and West African working-class and

cr

rural populations. The fruit is exported, whole or canned, to the United States, Canada and Great Britain. The aril of the unripe fruit is poisonous. Ackee poisoning is not uncommon [2], particularly after natural disasters (floods) that affect other food crop

us

harvests [3]. Its toxicity is mainly due to the presence of a phytotoxin, hypoglycin A (HGA), which is 10 to 20 times more concentrated in the aril of unripe fruit than ripe fruit, which is not poisonous [4-7]. The structure of HGA (L-α-amino-β-

an

methylene cyclopropane propionic acid) is close to alanine, with a cyclopropane cycle and an exomethylene group. After ingestion, HGA undergoes rapid metabolization, which leads, after transamination and decarboxylation, to the formation of a

M

toxic metabolite (methylene-cyclopropane-acetyl-coenzyme A). The metabolite reacts with some essential cofactors for βoxidation (coenzyme A, carnitine) and inhibits Acyl-CoA dehydrogenase, which also plays a role in β-oxidation [8, 9]. This

d

results in reduced fatty acid metabolism, bringing about increased glucose use and blocking hepatic gluconeogenesis, leading to

te

severe hypoglycaemia preceded by depletion of hepatic glycogen stores [10-12]. HGA poisoning causes ‘Jamaican vomiting sickness’ syndrome, which is characterized by gastrointestinal distress, with symptoms such as nausea, abdominal pain and

Ac ce p

uncontrollable vomiting [13]. The central nervous system is also impaired, leading to such clinical manifestations as paresthesia, lethargy, drowsiness, hypotonia, seizures, and coma. Ackee poisoning, which affects children in particular, can lead to the death of the victim. At a biological level, severe hypoglycaemia occurs.

HGA is also present in several species of the Acer genus, in variable concentrations [14, 15]. Valberg, Votion et al. recently established a link between the consumption of sycamore maple (Acer pseudoplatanus L.) seeds, box elder (Acer negundo L.) seeds, and equine atypical myopathy (AM) (seasonal pasture myopathy), which affects horses in Europe and North America [15, 16]. AM causes necrosis in postural and respiratory muscles, usually resulting in the death of the horse within 72 hours. AcylCoA dehydrogenase deficiency was identified in the skeletal muscle of a poisoned animal [17]. The quantity of HGA has been reported in box elder seeds [15] but not in sycamore maple.

4 Page 4 of 26

Several methods have been reported for measuring the quantity of HGA in B. sapida fruit and A. negundo seeds. These methods mainly comprise aqueous or alcoholic extraction of the fruit, followed by derivatization of HGA with ophthaldialdehyde (OPA) or phenyl isothiocyanate (PITC) [6, 7, 15, 18-20]. The compound is analysed in high-performance liquid chromatography coupled with UV spectrometry (at 254 nm). We recently proposed a method using gas phase coupled

with

mass

spectrometry

detection

(GC-MS)

after

derivatization

with

ip t

chromatography

N,O-

Bis(trimethylsilyl)trifluoroacetamide (BSTFA). This method enables the measurement of HGA in fruit samples. We were also

cr

able to establish the presence of HGA in the gastric fluid of a child who had died as a result of ackee poisoning [2].

us

To our knowledge, only one method of measuring HGA in blood and plasma has been published [21]. It is not possible to provide toxicological evidence of ackee or maple poisoning due to the high limit of quantification (1.4 mg/L), since HGA is

an

potentially poisonous in lower concentrations.

With the aim of lowering this limit of quantification, we have developed a method based on ultra-high performance liquid

M

chromatography coupled with high-resolution tandem mass spectrometry (UHPLC-HRMS/MS). The preparation of the sample

d

comprised solid-phase extraction (SPE) followed by derivatization with dansyl chloride.

Ac ce p

2.1. Chemical and reagents

te

2. Experimental

Working standards of HGA and internal standard (IS) mescaline-d9 were bought from LGC Standards (Molsheim, France). HGA was solubilized in a mixture of methanol : deionized water (50 : 50, v/v) and mescaline-d9 was diluted in methanol. Both were stored at a temperature of – 20°C.

Acetonitrile for LC-MS was obtained from Carlo Erba (Val de Reuil, France), isopropyl alcohol for LC/MS and deionized water (Optima LC/MS) were obtained from Fisher Chemical (Illkirch, France) and acetone and methanol were obtained from Sigma-Aldrich® (Saint-Quentin Fallavier, France). Dansyl chloride was bought from Sigma-Aldrich® (Saint-Quentin Fallavier, France).

5 Page 5 of 26

Saline phosphate buffer (PBS) was prepared with 0.2 g of potassium chloride KCl (≥ 99.5 % ; Carlo Erba, Val de Reuil, France), 8 g of sodium chloride NaCl (≥ 99.5 % ; Merck, Darmstadt, Germany), 0.2 g of monopotassium phosphate KH2PO4 (≥ 99.5 % ; Merck, Darmstadt, Germany) and 1.15 g of monosodium phosphate NaH2PO4 (≥ 99 % ; Sigma-Aldrich®, Saint-Quentin Fallavier, France) diluted in 1 L of deionized water and the pH adjusted to 11.0 with sodium hydroxide 1 M (Sigma-Aldrich®;

ip t

Saint-Quentin Fallavier, France). Ammonium formate buffer (2 mM) was prepared with ammonium formate (Optima LC/MS; Fisher Chemical, Illkirch, France) diluted in deionized water. The pH was subsequently adjusted to 3.0 with formic acid (Optima

cr

LC/MS; Fisher Chemical, Illkirch, France).

us

2.2. Extraction procedure

an

2.2.1. From body fluids

Extraction of HGA was a solid phase extraction (SPE) performed on CHROMABOND® HILIC cartridge (3 mL, 500 mg)

M

from Macherey-Nagel® (Hoerd, France):

The conditioning step was carried out with 3 mL of deionized water followed by 3 mL of isopropyl alcohol. The sample (250

d

µL of whole blood) was mixed with 5 µL of IS (mescaline-d9) at 1 µg/L and 4.75 mL of isopropyl alcohol, vortex mixed for 30

te

s and centrifuged at 3000 g for 10 min. The supernatant was then applied to the cartridge and allowed to drain under vacuum at 1

Ac ce p

mL/min. SPE rinse was carried out with 3 mL of isopropyl alcohol. The cartridge was allowed to dry under vacuum for 10 min and the elution was performed by 3 volumes of 1 mL of methanol. The eluate was evaporated to dryness under a stream of air at + 80°C.

2.2.2. From Ackee fruit and sycamore maple seeds

Quantification of HGA in arils from ackee fruit harvested in Suriname was also performed. The fruit was desiccated at + 25°C for one month and pulverized using a ball mill from Retsch® (Haan, Germany) until a fine powder was obtained. Exactly 50.0 mg of the powder was dissolved in 5 mL of methanol, vortex mixed, sonicated for 1 h and centrifuged for 10 min at 3000 g. The supernatant was diluted to a concentration of 5 mg/L. A methanolic extract of powdered sycamore maple seeds harvested in France was prepared the same way (100 mg/L).

2.3. Dansyl derivatization

6 Page 6 of 26

The residue was reconstituted with 200 µL of a PBS buffer at pH = 11.0 and vortex mixed with 200 µL of dansyl chloride (10 g/L in acetone). After 10 minutes at + 60°C, the solution was evaporated to dryness under a stream of air at + 80°C. The residue was then reconstituted with 50 µL of a 2 mM ammonium formate buffer at pH = 3.0 : acetonitrile (90 : 10, v/v) and centrifuged at 14000 g for 4 minutes, the supernatant being transferred into a vial for injection into the chromatographic system

ip t

(Fig. 2).

cr

2.4. Validation parameters in whole blood

us

The method was developed using whole blood provided by the regional blood bank (EFS Rhône-Alpes, France). The method was validated in whole blood according to the international standards [22, 23]. For the validation process, blank whole blood

an

samples were spiked with the appropriately diluted standard solutions.

The linearity of the method was assessed on four calibration ranges (eight calibration points: 0.8, 2, 5, 10, 20, 50, 100 and

M

500 µg/L) on four different days. The equation of the calibration curve was calculated by the least squares method and the validity of the regression model was confirmed by an ANOVA test. The slope was compared to zero by a Student test.

te

validated.

d

Quantification ranges were also carried out in urine (blank matrices spiked with the standards) but the methods were not fully

Ac ce p

The limit of detection (LOD) and the lower limit of quantification (LLOQ) were calculated by performing the analysis of ten different sources of blank matrices (post-mortem blood) and according to the formulas: LOD = mbl + 3SDbl and LLOQ = mbl + 10SDbl (mbl: mean of the measured concentration of the blanks; SDbl: standard deviation of the measured concentration of the blanks; n = 10). The upper limit of quantification (ULOQ) was chosen as the concentration of the upper calibration standard.

The extraction recovery was determined by comparing the mean peak areas from the samples spiked with compounds before the complete SPE extraction with those obtained from the samples spiked after the extraction (50 µg/L, n = 10).

Matrix effect was determined by comparing the mean peak areas from whole blood samples spiked after the extraction and direct injection of the same amount dissolved in the mobile phase (50 µg/L, n = 10).

The within- and between-day precision and the accuracy of the method were calculated on ten calibration points at 2, 50 and 350 µg/L (blank matrices spiked) on three different days (n = 90). The within-day precision and accuracy of the method were

7 Page 7 of 26

calculated on ten calibration points at the LLOQ (n = 10) and at 1000 µg/L with a dilution factor of 1/20 in another source of post-mortem blood (n = 10). Outliers were detected by a Dixon test and removed from the validation data.

ip t

2.5. Applications

The method has been applied to two serum samples from horses grazing in the Belgian provinces of Liege (horse 1) and

cr

Hainaut (horse 2). The two horses were suffering from AM and did not survive the disease.

A quantification range was carried out in horse serum kindly provided by the Laboratoire des Courses Hippiques

us

(Verrières-le-Buisson) (eight calibration points: 0.8, 2, 5, 10, 20, 50, 100 and 500 µg/L). The equation of the calibration curve

an

was calculated by the method of least squares and the validity of the regression model was confirmed by an ANOVA test.

2.6. Quantification procedure in fruit and seeds

M

HGA was quantified by the standard addition method: 0, 20, 40, 60, 80 or 100 µL of HGA at 0.1 mg/L (0, 10, 15, 20, 25 or 30 µL of HGA at 1 mg/L for the maple) and 5 µL of IS at 1 µg/L were added to 100 µL of the extract and evaporated to dryness

d

under a stream of air at + 80°C. The samples were then treated following the same derivatization procedure applied to the body

te

fluids as described above. The equation of the calibration curves was calculated by the method of least squares and the validity

Ac ce p

of the regression models was confirmed by an ANOVA test.

2.7. UHPLC-HRMS analysis

2.7.1. UHPLC conditions

Separation was performed by UHPLC (Ultimate® 3000; Thermo Scientific, Courtabœuf, France) using an ACQUITY UPLC® HSS C18 column (length: 150 mm; internal diameter: 2.1 mm; particle size: 1.8 µm) (Waters, Guyancourt, France) at + 25°C, for a gradient mobile phase composed of acetonitrile containing 0.1% of formic acid (Solvent A) and a 2 mM ammonium formate buffer at pH = 3.0 (Solvent B). The chromatographic run time was 10 min. The separation started with a mobile phase consisting of 5 : 95 (v/v) A : B for 1 min; followed by a linear gradient to 95 : 5 (v/v) A : B within 2 min; the conditions were isocratic for 2 min; then the mobile phase returned to initial conditions within 0.5 min and was equilibrated for 4.5 min. The

8 Page 8 of 26

mobile phase flow rate was 400 µL/min for the whole time of the separation. The temperature of the sampler was + 10°C and the injection volume was 10 µL.

ip t

2.7.2. Mass spectrometric conditions

Mass spectrometric detection was performed using a quadrupole-Orbitrap high-resolution detector (Q Exactive; Thermo Scientific, Courtabœuf, France) after ionization by heated electrospray in positive-ion mode. Nitrogen was both the collision and

cr

nebulizing gas (4.5; Linde gas, Molsheim, France). The mass spectrometer operated in full-scan mode and targeted-MS² mode

us

alternately (Table I). The full-scan product ion spectrum of HGA was used to confirm the identity of the toxin (Fig. 2). The mass spectrometric conditions and the collision energy were optimized by post-column infusion of dansyl-derivatized HGA (dns-

an

HGA) in 50 : 50 A : B (v/v) with a syringe pump (10 µL/min).

M

3. Results and discussion

3.1. Liquid chromatography

d

Initially, the polar nature of HGA (low-mass zwitterion (Fig.2) [24]) inclined us towards hydrophilic interaction

te

chromatography (HILIC) (bare silica) separation. The chromatographic peak, obtained from a pure solution, had good resolution. However, analysis of the blood showed the presence of isobaric compounds (at ± 5 ppm) at the same retention time as the HGA.

Ac ce p

The interference due to the presence of the endogenous isobaric compounds could not be eliminated, despite modifications to the extraction protocol and the chromatographic conditions. A derivatization step with dansyl chloride was tested. Dansyl chloride derivatization is a technique that has already been used for the analysis of amino acids with LC-MS/MS [25]. It proved perfectly suited for analysing HGA, a leucine-related compound with a primary amine group. Increasing the hydrophobic nature of HGA through dansyl chloride (dns-HGA) led us to select a type C18 analytical column.

Optimizing the separation parameters enabled us to obtain perfectly resolved peaks without interference from endogenous compounds. The retention times of dns-HGA and dansyl-derivatized mescaline-d9 (dns-mescaline-d9) were 4.0 and 4.2 min respectively (Fig. 3). The analysis time is short, with a gradient of 10 min.

3.2. Mass spectrometry

9 Page 9 of 26

The fragmentation of HGA produces two major transitions, corresponding to the breaking of the C-C bond on either side of the Cα amine function transporter. Dns-HGA produces a richer fragmentation spectrum (Fig.2). The fragmentation and ionization conditions were optimized by infusing dns-HGA, obtained from a pure HGA solution, in the system. An additional advantage of the derivatization is that it improves the ionisation of the molecule, resulting in a far more intense signal. The mass spectrometry

ip t

parameters are set out in Table 1. The two ionisation modes – positive and negative – were tested during infusion. Ionisation in positive mode proved more suitable, providing a better signal/noise ratio. The [M + H]+ ion (m/z 375.1370) was selected for

cr

identifying and quantifying HGA. The [M + Na]+ adduct was also present but its signal was weak (< 5%) compared with the [M

us

+ H]+ ion.

Using HRMS enabled us to achieve much higher sensitivity and better determination, given the short half-life of HGA due to

an

its rapid metabolism [13]. The resolution selected was not maximum, in order to obtain enough scans per peak to ensure repeatability of measurements (more than 13 scans). The mass tolerance was thus fixed at 10 ppm to mitigate any variations in

M

the m/z ratio, but without increasing the noise value.

te

3.3.1. Extraction recovery and matrix effect

d

3.3. Validation parameters

Ac ce p

Extraction comprised a protein precipitation stage with isopropanol, followed by SPE with CHROMABOND® HILIC cartridges attached by zwitterionic groups (ammonium-sulfonic acid modified silica). Liquid/liquid extraction was tested using several organic solvents combining different pH values. Non-modified silica-based cartridges (Strata Si-1 Silica, Phenomenex®) were also tested initially, but the extraction yields in both cases were disappointing (less than 10%). With SPE in HILIC mode in the conditions described, it was possible to obtain an extraction yield of 58%. With a ratio of 250 µL of total blood to 4.75 mL of isopropanol it is possible to limit the proportion of water and consequently the direct elution of the compound of interest when the sample is deposited on the cartridge. The quantification limit of the method can be virtually lowered by increasing the sample above 250 µL, but the volume of isopropanol added has to stay proportional to maintain the1/20 ratio (blood/isopropanol), extending the required deposit time.

With the combination of SPE and derivatization it was possible to analyse the HGA by removing the interfering isobaric compounds from the whole blood. However, some endogenous compounds remain throughout the analysis of dns-HGA, as

10 Page 10 of 26

shown by the high matrix effect (-45%). It is worth noting that the matrix effect does not compromise the sensitivity, accuracy or precision of the method (Tables II and III).

ip t

3.3.2. Precision, accuracy, LOD, LLOQ and choice of the internal standard

The validation parameters in whole blood are reported in tables II and III. The method meets with the usual requirements for precision (≤ 15% or ≤ 20% at the LLOQ) and accuracy (between 85 and 115% or 80 and 120% at the LLOQ). The LOD was

cr

fixed at 0.35 µg/L. The calibration range was tested from 0.80 to 500 µg/L, enabling quantification over a wide concentration

us

range. Calibration ranges were also set in urine and plasma at these concentrations. Although no validation took place in these two matrices, with the method described here it is possible to consider HGA analysis in various biological matrices.

an

Initially, we considered leucine-d3 as internal standard, since it has very similar physicochemical properties to HGA. However, the signal from the endogenous leucine derivative (C18H23N2O4S) at mass M + 3 (isotopic contribution of 13C) was a

M

source of bias for an internal standard added to a concentration of a few hundredths of µg/L. Indeed, the isotopic contribution at mass M + 3 is 0.8%, but the blood concentration of leucine is 10 to 20 mg/L. Leucine-d10 may be an alternative, but the

d

drawback is that it is very expensive. We tested and selected mescaline-d9 as internal standard, as its behaviour around

te

extraction and derivation, its retention time and its response are close to those of HGA. The choice of a deuterated analogue

drug.

Ac ce p

avoids any potential interference due to the presence of mescaline in a sample resulting from consumption of the psychotropic

3.4. Applications

Two applications are given as illustrations of the present assay: the quantification of HGA in horse serum and the quantification of HGA from vegetable matrices. The results from both applications show the interest and feasibility of our method to of quantifying HGA in various matrices.

3.4.1. Analysis in biological matrix

The concentrations of HGA measured in samples of horse serum are shown in table IV (Fig. 4). To the best of our knowledge, this is the first quantitative determination of HGA in blood to be reported in the literature. This constitutes proof of a direct link between the consumption of maple seeds and relatively high concentrations of HGA. However, it is not possible to

11 Page 11 of 26

extrapolate what may be found in humans from the concentrations found in horses. Indeed, the quantities of HGA consumed (depending on the quantity of plant compounds ingested and the concentration of HGA in the plants) and the pharmacokinetics of HGA are probably very different between the two species. It is, however, possible to consider that the blood concentrations in cases of human poisoning may be significantly lower than those found in horses. This would explain why the method proposed

ip t

by Fincham et al. in 1977 (LOQ of 1.4 mg/L) was unable to identify HGA in the blood of people suspected of poisoning after consuming ackee [21], but with the performance of the technique described here it may be possible to envisage detecting HGA

cr

in humans.

us

3.4.2. Analysis of ackee fruit and sycamore maple seeds

an

Quantification of the fruit was performed from a methanolic extract of the samples. Several extraction solvents had been tested to optimize the extraction conditions: water, water with 5% formic acid, water with 1% ammonia, acetonitrile, methanol, water/methanol mixture (50/50, v/v). Methanol proved to be the most effective extraction solvent. Extraction was performed

M

several times on the same powder, so an average value could be obtained. The HGA concentration in the methanolic extracts of unripe ackee arils and sycamore maple seeds is reported in table IV. A concentration of 7.2 mg/g in the ackee aril matches the

d

concentrations already published for fruit in stage 1 or 2 of the maturity scale (unripe) [7]. This is the first HGA concentration in

te

A. pseudoplatanus seeds to be reported. A concentration of 0.74 mg/g is consistent with those measured by Valberg et al. in box

4. Conclusion

Ac ce p

elder maple seeds (9 to 77 µg/seed) [15], the two species have a similar HGA concentration from electrophoresis [14].

This article describes the development and validation of an original method for quantifying HGA in blood samples by UHPLC-HRMS/MS. The evolution of techniques over 40 years has made it possible to increase the sensitivity of the assay considerably, and may enable the analytical documentation of ackee poisonings which are currently diagnosed biochemically. This article presents the first concentrations of HGA measured in the serum of horses suffering from EAM, showing a link between the disease and the consumption of sycamore maple seeds.

References

[1] J. Bruneton, Plantes toxiques, végétaux dangereux pour l'Homme et les animaux, third ed., Tec & Doc, Paris (France), 2006. [2] Y. Gaillard, J. Carlier, M. Berscht, C. Mazoyer, F. Bevalot, J. Guitton, L. Fanton, Forensic Sci. Int. 206 (2011) e103.

12 Page 12 of 26

[3] R. Joskow, M. Belson, H. Vesper, L. Backer, C. Rubin, Clin. Toxicol. 44 (2006) 267. [4] C.H. Hassall, K. Reyle, Biochem. J. 60 (1955) 334. [5] G.W. Chase, Jr., W.O. Landen, Jr., A.G. Soliman, J. Assoc. Off. Anal. Chem. 73 (1990) 318. [6] K.D. Golden, O.J. Williams, Y. Bailey-Shaw, J. Chromatogr. Sci. 40 (2002) 441.

ip t

[7] C.S. Bowen-Forbes, D.A. Minott, J. Agric. Food Chem. 59 (2011) 3869. [8] C. Von Holt, Biochim. Biophys. Acta 125 (1966) 1.

cr

[9] J. Salaün, Top. Curr. Chem. 207 (2000) 2.

[11] P.C. Feng, S.J. Patrick, Br. J. Pharmacol. Chemother. 13 (1958) 125.

us

[10] C.H. Hassall, K. Reyle, Nature 173 (1954) 356.

[12] C. Von Holt, M. Von Holt, H. Bohm, Biochim. Biophys. Acta, 125 (1966) 11.

[14] L. Fowden, H.M. Prat, Phytochemistry 12 (1973) 1677.

an

[13] K. Tanaka, E.A. Kean, B. Johnson, N. Engl. J. Med. 295 (1976) 461.

M

[15] S.J. Valberg, B.T. Sponseller, A.D. Hegeman, J. Earing, J.B. Bender, K.L. Martinson, S.E. Patterson, L. Sweetman, Equine Vet. J. 45 (2013) 419.

d

[16] D.M. Votion, G. van Galen, L. Sweetman, F. Boemer, P. de Tullio, C. Dopagne, L. Lefere, A. Mouithys-Mickalad, F.

te

Patarin, S. Rouxhet, G. van Loon, D. Serteyn, B.T. Sponseller, S.J. Valberg, Equine Vet. J. 46 (2014) 146. [17] D.M. Votion, ISRN Vet. Sci. 2012 (2012) 1.

Ac ce p

[18] M. Brown, R.P. Bates, C. McGowan, J.A. Cornell, J. Food Saf. 12 (1991) 167. [19] G.M. Ware, J. AOAC Int. 85 (2002) 933.

[20] T.B. Whitaker, J.J. Saltsman, G.M. Ware, A.B. Slate, J. AOAC Int. 90 (2007) 1060. [21] A.G. Fincham, West Indian Med. J. 26 (1977) 62. [22] V.P. Shah, K.K. Midha, S. Dighe, I.J. McGilveray, J.P. Skelly, A. Yacobi, T. Layloff, C.T. Viswanathan, C.E. Cook, R.D. McDowall, et al., Eur. J. Drug Metab. Pharmacokinet. 16 (1991) 249. [23] A. Polettini, Applications of LC-MS in toxicology, first ed., Pharmaceutical Press, London (UK), 2006. [24] E.V. Ellington, C.H. Hassall, J.R. Plimmer, C.E. Seaforth, J. Chem. Soc. (1959) 80. [25] R. Rebane, M.L. Oldekop, K. Herodes, J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 904 (2012) 99.

13 Page 13 of 26

Ac ce p

te

d

M

an

us

cr

ip t

FIGURES & TABLES CAPTIONS Fig. 1. Blighia sapida Koenig: photograph of a ripe fruit (stage 5-6 of maturity) [7]. Fig. 2. Full-scan product ion spectrum of HGA and dns-HGA (HRMS/MS). HGA, hypoglycin A; dns-HGA, dansylhypoglycin A; HRMS/MS, high resolution tandem mass spectrometry; MM, molecular mass. Fig. 3. Chromatogram obtained from the analysis of a sample of blood spiked with HGA (5 µg/L) (0.25 ng on column) or free of the compound. dns-HGA, dansyl-hypoglycin A. Fig. 4. Chromatograms obtained from the analysis of serum of horses suffering from atypical myopathy. dns-HGA, dansyl-hypoglycin A.

14 Page 14 of 26

Table I 60

Auxiliary gas flow rate (a.u.)

10

Sweep gas flow rate (a.u.)

5

Spray voltage (kV)

+ 5.0

Capillary temperature (°C)

+ 400

Heater temperature (°C)

+ 250

cr

Sheath gas flow rate (psi)

ip t

High resolution mass spectrometry parameters.

positive

Scan time (min)

3.5 - 4.7

Scan range (amu)

373 ̶ 456

Resolution

70000

Mass tolerance (ppm)

10.0

Quantifier ion (amu) 375.1374

dns-mescaline-d9

454.2366

M

dns-HGA Targeted-MS/MS mode: confirmation

positive

Scan time (min)

3.5 ̶ 4.7

d

Ionization mode Collision energy (%)

an

Ionization mode

us

Full-scan mode: quantification

40

Ac ce p

te

Resolution 17500 dns-HGA, dansyl-hypoglycin A; dns-mescaline-d9, dansyl-mescaline-d9.

15 Page 15 of 26

Table II Concentration ± S.D. (µg/L)

Accuracy (%)

Between-day precision (%)

0.8

0.81 ± 0.02

101.0

2.7

2

1.96 ± 0.18

98.0

9.0

5

4.98 ± 0.28

99.6

5.7

10

10.0 ± 0.8

100.3

8.4

20

19.2 ± 1.4

95.8

7.3

50

50.9 ± 2.1

101.8

4.1

100

94.7 ± 3.0

94.7

500

544 ± 42

108.7

us 3.2

an

7.8

Ac ce p

te

d

M

HGA, hypoglycin A; S.D., standard deviation.

cr

Spiked (µg/L)

ip t

Validation data for quantification of HGA in whole blood: calibration ranges.

16 Page 16 of 26

Table IV Serum HGA concentration from horses suffering from atypical myopathy, and content in ackee arils and sycamore maple seeds. Precision (%)

horse 1 (serum) (µg/L)

450.5

N.C.

horse 2 (serum) (µg/L)

32.7

N.C.

7.2 ± 0.4

A. pseudoplatanus L. (seeds) (mg/g), n =5

0.74 ± 0.09

0.9990

cr

B. sapida Koenig (arils) (mg/g), n = 4

r²

ip t

Concentration ± S.D.

5.4

0.9850

12.0

0.9989

Ac ce p

te

d

M

an

us

HGA, hypoglycin A; S.D., standard deviation; r², coefficient of determination; N.C., not calculated.

17 Page 17 of 26

Table III Validation data for quantification of HGA in whole blood: Q.C. Spiked (µg/L)

Concentration ± S.D. (µg/L)

Accuracy (%)

LLOQ

0.78 ± 0.11

2

Precision (%) Between-day

97.8

13.8

N.C.

1.79 ± 0.22

89.4

11.9

50

45.3 ± 6.4

90.5

7.5

350

353 ± 26.7

100.8

6.5

1000 (diluted 1/20)

1140 ± 155

114.0

13.6

ip t

Within-day

3.3

cr

14.4 4.7

us

N.C.

Ac ce p

te

d

M

an

HGA, hypoglycin A; Q.C., quality control; S.D., standard deviation; LLOQ, lower limit of quantification; N.C., not calculated.

18 Page 18 of 26

Ac

ce

pt

ed

M

an

us

cr

i

Fig. 1

Page 19 of 26

Ac ce p

te

d

M

an

us

cr

ip t

Fig. 2

Page 20 of 26

Ac

ce

pt

ed

M

an

us

cr

i

Fig. 3

Page 21 of 26

Ac ce p

te

d

M

an

us

cr

ip t

Fig. 4

Page 22 of 26

Table I

cr us an

Ac ce p

te

d

M

Sheath gas flow rate (psi) 60 Auxiliary gas flow rate (a.u.) 10 Sweep gas flow rate (a.u.) 5 Spray voltage (kV) + 5.0 Capillary temperature (°C) + 400 Heater temperature (°C) + 250 Full-scan mode: quantification Ionization mode positive ̶ Scan time (min) 3.5 - 4.7 Scan range (amu) 373 456 Resolution 70000 Mass tolerance (ppm) 10.0 Quantifier ion (amu) dns-HGA 375.1374 dns-mescaline-d9 454.2366 Targeted-MS/MS mode: confirmation ̶ Ionization mode positive Scan time (min) 3.5 4.7 Collision energy (%) 40 Resolution 17500 dns-HGA, dansyl-hypoglycin A; dns-mescaline-d9, dansylmescaline-d9.

ip t

Table I High resolution mass spectrometry parameters.

Page 23 of 26

Table II

Table II Validation data for quantification of HGA in whole blood: calibration ranges. Spiked (µg/L)

Concentration ± S.D. (µg/L)

Accuracy (%) 101.0 98.0 99.6 100.3 95.8 101.8 94.7 108.7

2.7 9.0 5.7 8.4 7.3 4.1 3.2 7.8

Ac ce pt e

d

M

an

us

cr

ip

t

0.8 0.81 ± 0.02 2 1.96 ± 0.18 5 4.98 ± 0.28 10 10.0 ± 0.8 20 19.2 ± 1.4 50 50.9 ± 2.1 100 94.7 ± 3.0 500 544 ± 42 HGA, hypoglycin A; S.D., standard deviation.

Between-day precision (%)

Page 24 of 26

Table III

Table III Validation data for quantification of HGA in whole blood: Q.C. Concentration ± Spiked (µg/L) Accuracy (%) S.D. (µg/L)

Ac ce pt e

d

M

an

us

cr

ip

t

Precision (%) Within-day Between-day LLOQ 0.78 ± 0.11 97.8 13.8 N.C. 2 1.79 ± 0.22 89.4 11.9 3.3 50 45.3 ± 6.4 90.5 7.5 14.4 350 353 ± 26.7 100.8 6.5 4.7 1000 (diluted 1/20) 1140 ± 155 114.0 13.6 N.C. HGA, hypoglycin A; Q.C., quality control; S.D., standard deviation; LLOQ, lower limit of quantification; N.C., not calculated.

Page 25 of 26

Table IV

Table IV Serum HGA concentration from horses suffering from atypical myopathy, and content in ackee arils and sycamore maple seeds. Concentration ± S.D.

Precision (%)

r²

Ac ce pt e

d

M

an

us

cr

ip

t

horse 1 (serum) (µg/L) 450.5 N.C. 0.9990 horse 2 (serum) (µg/L) 32.7 N.C. B. sapida Koenig (arils) (mg/g), n = 4 7.2 ± 0.4 5.4 0.9850 A. pseudoplatanus L. (seeds) (mg/g), n =5 0.74 ± 0.09 12.0 0.9989 HGA, hypoglycin A; S.D., standard deviation; r², coefficient of determination; N.C., not calculated.

Page 26 of 26