1. Introduction
Grass pea (
Lathyrus sativus) is the most cultivated
Lathyrus species [
1,
2], particularly in South Asia and Sub-Saharan Africa [
1], and has been used as human food and animal feed since ancient times as an important source of carbohydrates, proteins, and also antioxidant compounds [
3]. It is resistant to both flood and drought, and thrives on poor soils that do not support many other crop species growth [
2]. However,
Lathyrus species have been implicated as the cause of a neurological disorder, called neurolathyrism, in both men and animals [
2]. Neurolathyrism is characterized by an irreversible neurodegeneration of motor neurons, resulting in a spastic paraparesis of the legs that can vary from mild walking difficulties to bedridden state [
4,
5,
6]. Neurolathyrism is caused by an excessive consumption of grass pea seeds as staple food for several months [
5,
6] and it is intricately linked to poverty, malnutrition, and periods of famine [
2,
4,
5,
7]. Importantly, grass pea is harmless to humans and animals when consumed as part of a balanced diet [
1].
The discovery of the neuroexcitatory potential of β-
N-Oxalyl-
l-α,β-diaminopropionic acid (β-ODAP), a non-protein amino acid present in grass pea [
7], led to proposing a role as the neurotoxin responsible for the disease [
5,
6]. The mechanism of its action has not been conclusively elucidated [
6]. Nevertheless, potential pharmacological benefits of β-ODAP should not be neglected [
1]. The activation of protein kinase C (PKC) by β-ODAP adds a new dimension to explore its possible therapeutic potentials in areas such as Alzheimer’s disease, hypoxia, and long-term potentiation [
7]. Interestingly, β-ODAP was also found to be present in the non-leguminous plants
Panax species, and it is considered a bioactive therapeutic amino acid component in
Panax notoginseng, commonly referred as dencichine. Studies have reported β-ODAP as the component responsible for the medicinal herb’s main haemostatic and platelet-increasing properties in vivo [
8,
9]. It has also been described that the haemostatic effect is present at a low dose of β-ODAP, while neurotoxicity occurs at higher doses [
8,
9].
In
Lathyrus seeds, β-ODAP is accompanied by lower concentrations of its isomer α-ODAP [
2,
5]. Unlike β-ODAP, α-ODAP is considered non-toxic, and technological processing of
Lathyrus seeds or flour, such as cooking, fermentation, or pre-soaking in alkaline solutions, can decrease β-ODAP concentration and increase the concentration of its α isomer, reducing
Lathyrus toxicity [
1,
2,
10]. Diaminopropionic acid (DAP) can also be formed by steaming, presumably resulting from hydrolysis of the oxalyl compound, decreasing β-ODAP content [
8,
11].
Given the increasing need for resilient food crops, grass pea breeding has gained importance, and major efforts have been aimed at reducing β-ODAP content, with the development of low β-ODAP varieties through selective breeding [
1,
2]. High-throughput accurate quantification methods for this specific compound are extremely important for these breeding programs, where a large number of samples need to be quantified in a simple and fast way. Several analytical methodologies have been suggested for β-ODAP quantification, but some were not able to distinguish the toxic β-ODAP from its non-toxic α isomer, such as colorimetric methods [
12], GC-MS (gas chromatography-mass spectrometry) after sample derivatization [
9] and TLC (thin layer chromatography) [
13]. Some other methodologies have been successful in analyzing both α- and β-ODAP, namely by NMR (nuclear magnetic resonance) [
14] and CZE (capillary zone electrophoresis). However, the high operation pH (9.2) in CZE may result in β-ODAP hydrolysis to diaminopropionic acid (DAP) [
13,
15,
16].
Although these methods have been employed, HPLC (high performance liquid chromatography) has been the most used for β-ODAP quantification [
8,
9,
12,
13,
15,
17,
18,
19,
20,
21]. However, the analysis of ODAP by HPLC is a challenge due to its high polarity and weak UV absorption [
8]. The majority of the currently available HPLC based methods for β-ODAP quantification involve the use of a reversed phase C18 column and UV detection, after a multi-step derivatization of the sample [
17,
18,
19,
22]. However, poor derivative stability, side reactions, and reagent interferences may occur, which may lead to interference signals [
8,
13,
20]. Furthermore, these methods are usually laborious, inconvenient, and time-consuming [
8,
13]. More recently, a UHPLC-MS/MS method using a C18 column has been developed for the quantification of β-ODAP,
L-homoarginine, and asparagine in
Lathyrus sativus [
21]. However, since α-ODAP has not been analyzed, it is possible that both isomers had been quantified simultaneously. Similarly, Gresta et al. (2014) have developed a HPLC-ELS (evaporative light scattering) method for the quantification of both isomers, using a reversed-phase column [
14].
Despite the frequent use of reversed phase chromatography for ODAP analysis, hydrophilic interaction chromatography (HILIC) is suitable for analyzing compounds in complex systems that elute near the void (highly polar) in reversed-phase chromatography [
23]. HILIC is a variant of normal phase chromatography [
8] and typical stationary phases consist of classical bare silica or silica gels modified with many polar functional groups [
23]. The predominant retention mechanism in HILIC separation is not always easily predictable [
24] and it is commonly accepted that the partitioning of the analyte occurs between two layers [
23,
24]: the water-deficient mobile phase and water-enriched layer on the surface of the polar stationary phase which is partially immobilized onto the surface of the stationary phase. Nonetheless, other interactions, such as hydrogen-bonding, hydrophobic, and ion-exchange, may also be involved [
24]. Precisely defining which mechanism prevails is currently a complex and difficult task [
23]. The number of HILIC-compatible columns has increased in the last few years. Generally, stationary phases are silica-based and can often be charged in some pH region [
25]. Considering the various types of stationary phases and the poorly understood retention mechanism in HILIC, the choice of an appropriate column for simultaneous analysis of α- and β-ODAP is a challenge [
25,
26]. Previous HILIC studies have also shown that different selectivities and retention times were obtained using the same mobile phase on different stationary phases [
25,
26]. Therefore, the possible secondary interactions between the stationary phase and the analytes have to be considered for the column choice [
25,
26].
HILIC can be conveniently coupled to mass spectrometry (MS), especially in the electrospray ionization (ESI) mode [
23]. HPLC-MS/MS is a powerful tandem method for quantification of analytes in complex matrices, providing retention time, mass/charge ratios, and relative abundance (intensity) data [
27]. MS/MS detectors are preferred over other chromatographic detectors, due to the better sensitivity, specificity, and wide range applicability [
21]. Some authors have employed HILIC columns using MS/MS (tandem mass spectrometry) detectors [
8,
20] for ODAP analysis. However, these methodologies were not able to distinguish both ODAP isomers.
In this work, a procedure proposed by McKeown (2015) for the selection of the best HILIC-type column was followed [
28]. A simple and fast HPLC-ESI-MS/MS method for quantification of both α- and β-ODAP, without sample derivatization, was developed and validated according to the most recent guidelines (European Communities, Food and Drug Administration, European Medicines Agency and Harmonized Tripartite Guideline), and applied to the analysis of 107 grass pea and two red pea (
Lathyrus cicera) accessions representative of the main grass pea-growing geographical regions in the world, using a neutral HILIC column (chemically bonded diol phase).
3. Materials and Methods
3.1. Chemicals and Materials
For extract preparation and HPLC-MS/MS analysis, Milli-Q water (18.2 MΩ.cm resistivity) was obtained from a Millipore-Direct Q3 UV system (Millipore®, Burlington, MA, USA); formic acid ≥ 95% was from Sigma-Aldrich®, St Louis, MO, USA; and acetonitrile HPLC Plus Gradient was from Carlo Erba®, Val de Reuil, France. β-ODAP standard was obtained from Lathyrus Tecnhologies, Hyderabad, India.
Since the isomerization of β- to α-ODAP increases with increase in temperature [
38], six solutions of β-ODAP in water, at 40 mg L
−1, were heated at 120 °C for 120 min. These solutions were used for HPLC-MS/MS method development.
3.2. Samples
A collection of 107 grass pea accessions representative of the main grass pea-growing geographical regions of the world plus two
L. cicera accessions, kindly provided by Prof. Fernand Lambein (IPBO, Ghent, Belgium), USDA-ARS (Pullman, WA, USA), INIA-CRF (Madrid, Spain), IAS-CSIC (Córdoba, Spain), ICARDA, HAO-DEMETER (Thermi, Greece), and IFVC (Novi Sad, Serbia) germplasm collections, were analyzed for their α-ODAP and β-ODAP contents under the scope of the European FP7 project LEGATO-LEGumes for the Agriculture of TOmorrow. Accessions were grouped in big (>20 g/100 seeds weight) or small seed size, light or dark seed color, and as a combination of both traits into varietal groups [
39]: Mediterranean type (big light seeds), Indian type (small dark seeds), and all the other size and color combinations in an Intermediate group (
Table S2).
All the accessions were multiplied under the same edapho-climatic conditions at the IAS-CSIC experimental farm (Córdoba, Spain) and kindly provided by Prof. Diego Rubiales. Dry mature seeds of grass pea were milled (Falling n° 3100–Perten, Sweeden) to a particle size of 0.8 mm and flour stored at −20 °C until analysis.
3.3. Optimization of the Extraction Procedure
In order to achieve a higher ODAP extraction efficiency, water and different ethanol:water solvent ratios (60:40, 30:70, 70:30) described in the literature for ODAP extraction [
8,
14,
16,
19,
29] were tested, using commercial grass pea samples, based on methods previously described [
8,
14,
16,
22,
38]. Water was selected as the best extraction solvent. Briefly, samples were accurately weighed (48 mg) and ODAP (α and β) was extracted with 2 mL of solvent, vortexed for 90 s, and extracted for 2 h, in ice, using a Selecta
® Rotabit orbital shaker and centrifuged (20 min, 10,000 rpm). Extracts were kept at −20 °C until analysis, for a maximum of two months. Before HPLC-MS/MS analysis, extracts were diluted (1:50) in acetonitrile:water 40:60, in order to match the mobile phase, and filtered through a 0.20 μm PTFE syringe filter (Chromafil
® Macherey-Nagel, Germany). Extracts were prepared in triplicate.
3.4. HPLC-MS/MS Method Development
The HPLC-MS/MS analyses were performed on a Waters Alliance HPLC system (Waters
®, 2695 separation module, Dublin, Ireland) comprising a quaternary pump, an on-line solvent degasser, autosampler, and column oven. The HPLC method was optimized following the procedure proposed by McKeown (2015) for the selection of the best HILIC column: HILIC-A, HILIC-B, and HILIC-N (100 Å, 5 µm 3.0 × 150 mm), ACE
®, Scotland, using the mobile phases A: 10 mM ammonium formate, pH 3.0, 4.7 or 6.0 in acetonitrile:water (94:6
v/
v) and B: 10 mM ammonium formate, pH 3.0, 4.7, or 6.0 in acetonitrile:water (50:50
v/
v) [
28].
The MS/MS conditions were optimized using a pure β-ODAP standard. Tandem mass spectrometry (MS/MS) detection was performed on a Micromass® Quattro Micro triple quadrupole (Waters®, Ireland) using an ESI source operating at 140 °C, applying a capillary voltage of 2.7 kV and a cone voltage of 10 V, in positive ion mode. MassLynx software (version 4.1) was used to control the system, for data acquisition and processing. High purity nitrogen (N2) was used both as drying gas and as a nebulizing gas. Ultra-high purity argon (Ar) was used as collision gas. Collision energies were optimized using a β-ODAP standard solution and was set at 10 eV. The analysis was performed in MRM mode in order to achieve a higher selectivity and sensitivity.
After method optimization, validation and application to real samples were performed in a HILIC-N (bonded neutral character phase), column at 25 °C, using an injection volume of 10 µL. The mobile phase consisted of an isocratic method of 40% of 2% HCOOH in acetonitrile (eluent A) and 60% of 2% HCOOH in Milli-Q H2O (eluent B) at a flow rate of 0.40 mL min−1, for 18 min. Autosampler temperature was set at 10 °C.
3.5. Method Validation
Validation was carried out by determination of specificity, limit of detection (LOD), and quantification (LLOQ), linearity and linear range, precision, accuracy, matrix effect, dilution integrity, recovery, and stability assays.
3.5.1. Specificity
Since ODAP is an endogenous compound of grass pea, it was not possible to determine the specificity in grass pea extracts. Extracts of other legumes were prepared and analyzed, namely faba bean (Vicia faba) (n = 1), common bean (Phaseolus vulgaris) (n = 1), chickpea (Cicer arietinum) (n = 1), pea (Pisum sativum) (n = 2), and lentils (Lens culinaris) (n = 2), and analyzed as blank assays.
A standard solution of β-ODAP at 100 mg L−1 was infused into the mass spectrometer in order to determine the two product ions with the highest signals. These transitions were used as the quantification transition (MRM1) and the confirmation transition (MRM2), in order to evaluate the method specificity. The same transitions were used for the quantification of α-ODAP. MRM1/MRM2 signal transition ratio was determined for α- and β-ODAP and compared with the values obtained for the sample extracts. The retention time of both isomers in solution and samples was also compared. Carry-over was addressed by injecting blank samples after six injections of a high concentration calibration standard of β-ODAP at 3100 ng mL−1, corresponding to the upper limit of quantification (ULOQ).
3.5.2. Limit of Detection (LOD) and Lower Limit of Quantification (LLOQ)
A β-ODAP solution of 40 mg L−1 was prepared in water and diluted in acetonitrile:water (40:60), until a signal-to-noise ratio of 3:1 (LOD) and 10:1 (LLOQ). The LOD and LLOQ values were confirmed by the analysis of six solutions at these concentrations, prepared from six independent stock solutions (around 40 mg L−1).
3.5.3. Linearity and Linear Range
The linearity study was performed using nine calibration standards, in order to cover the quantification of α- and β-ODAP simultaneously, with the concentrations of 25 ng mL−1 (LLOD), 75, 100, 300, 700, 1300, 1900, 2500, and 3100 ng mL−1 (ULOQ). All the calibration standards in the different batches (n = 8) were prepared from fresh β-ODAP standard solutions of around 40 mg L−1 in water, and diluted in acetonitrile:water (40:60), in order to match the mobile phase constitution. The determination coefficient (r2) was calculated. Since no commercial standard is available, α-ODAP was quantified using the β-ODAP calibration curve and expressed as β-ODAP equivalents.
3.5.4. Precision and Accuracy
Accuracy (intra-day and inter-day accuracy) and precision (experimental precision, injection repeatability, and inter-day precision) were determined at the LLOQ (25 ng mL−1), low (75 ng mL−1), mid (1300 ng mL−1), and high (2500 ng mL−1) β-ODAP concentration levels (CLs).
For the determination of intra-day accuracy and experimental repeatability, six β-ODAP solutions at approximately 40 mg L−1 were prepared in water, and diluted to the LLOQ, low CL, mid CL and high CL in acetonitrile:water (40:60), and their concentrations were calculated against a calibration curve. In order to determine the injection repeatability, one solution at the LLOQ, low CL, mid CL, and high CL was injected 10 times in the equipment.
To evaluate the inter-day accuracy and precision, two β-ODAP solutions were prepared for each sample batch (sequence of analysis) and diluted to the same CLs described above (LLOQ, low, mid, and high). All the quality control (QC) standards analyzed in all sample batches (
n = 8) were used for the determination of inter-day accuracy and precision. Their concentrations were determined against the calibration curve of the respective sample batch, and compared with their nominal values. Accuracy was reported as percent of the nominal value [
32,
33], and precision was expressed as relative standard deviation (RSD) [
30,
32,
33].
Injection repeatability was also evaluated analyzing three real samples (LS 104, LS 035, and LS 025), corresponding to different CLs of α- and β-ODAP, and results were expressed as relative standard deviation (%).
3.5.5. Matrix Effect
To address matrix effects, ion suppression, or enhancement, six calibration curves were prepared in six grass pea’s extracts. Since β-ODAP is an endogenous compound of grass pea, slopes of the curves in pure solvent (acetonitrile:water (40:60)) and in sample matrix were compared, and the matrix factor (MF) was quantified as the ratio between the slope of the matrix calibration curve and pure solvent [
40].
Extracts of six samples (LS 043, LS 067, LS 076, LS 104, LS 114, and LS 118) were selected to study the matrix effect, since they have different endogenous CLs of β-ODAP. Water extracts were diluted (1:50) in acetonitrile:water (40:60), and 180 μL were spiked with 20 μL of different CLs (0.25 to 31 μg mL
−1) of β-ODAP standards in solvent (acetonitrile:water (40:60)), to obtain spiked concentrations of 25, 75, 100, 300, 700, 1300, 1900, 2500, and 3100 ng mL
−1 of β-ODAP in the samples. The background peak areas (endogenous peak areas) of β-ODAP in the matrices were determined adding 20 μL of solvent to 180 μL of each extract. In order for this method to be reproducible, only the CLs which gave an increase of >20% in the matrix peak areas (endogenous β-ODAP content of the samples mentioned above) after spiking with β-ODAP were considered for the calibration curve in the matrices [
41].
3.5.6. Dilution Integrity
Six grass pea extracts randomly chosen (LS 043, LS 067, LS 076, LS 103, LS 104, and LS 114) were spiked with 10 mg L−1 of β-ODAP in water, before dilution. Samples were diluted 1:50 in acetonitrile:water (40:60), analyzed in the HPLC-MS/MS, and compared to non-spiked extracts. The accuracy was determined for each sample. Results were expressed as %.
3.5.7. Method Recovery
To address method recoveries, 48 mg of four different samples (LS 059, LS 075, LS 104, and LS 114) were extracted with 2 mL of β-ODAP aqueous solutions of 3.75, 65, and 125 mg L
−1. Thus, after extracts dilution (1:50), β-ODAP concentrations corresponded to 75 ng mL
−1 (low CL), 1300 ng mL
−1 (mid CL), and 2500 ng mL
−1 (high CL), respectively. Peak areas of the extracted samples were compared to the peak areas of samples spiked with the analyte post-extraction. Recovery was reported as a percentage of the known amount of the analyte carried through the sample extraction and processing steps of the method [
30,
33].
3.5.8. Stability Assays
The stability of β-ODAP in solution (stock solution and working solutions) was evaluated. Two fresh β-ODAP stock solutions were prepared in water (40 mg L−1), diluted to a high, mid, and low CL (2500, 1300, and 75 ng mL−1) in acetonitrile:water (40:60) (working solutions) and analyzed by HPLC-MS/MS. The stability of the β-ODAP stock solutions was evaluated at −20 °C, up to three months. Autosampler stability of β-ODAP working solutions in acetonitrile:water (40:60) was also evaluated, up to one week. The stability of stock solutions (40 mg L−1) and working solutions of β-ODAP was also evaluated at room temperature (23 °C), up to two months. Freeze-thaw stability of the two β-ODAP stock solutions was also verified after three cycles: stock solutions were stored and frozen in the freezer at −20 °C for at least 24 h, and thereafter thawed at room temperature. After complete thawing, β-ODAP stock solutions were refrozen again, in the same conditions. Control standards were analyzed against fresh calibrations curves prepared in solvent, and the obtained concentrations compared to the nominal concentrations.
The stability of α- and β-ODAP in the sample matrix (grass pea extracts in water and after 1:50 dilution in acetonitrile:water (40:60)) was evaluated in the freezer (−20 °C), up to three months; fridge (4 °C), up to one week; and room temperature (23 °C), up to two months. Freeze-thaw stability (three cycles) was also evaluated. The autosampler stability was verified in the diluted extracts, up to one week. Three samples were chosen for the stability assay, corresponding to a low (LS 104), mid (LS 035), and high (LS 025) CLs of ODAP, and were analyzed as triplicates. Samples were analyzed against fresh calibrations curves prepared in solvent, and the obtained concentrations compared to the fresh extracts’ initial concentrations.
3.6. Application of the Method
ODAP was extracted from 107 grass pea and two red pea accessions, and its content was analyzed by HPLC-MS/MS following the developed methodology, using freshly prepared extracts. The HPLC-MS/MS analysis was performed for all the extracted samples (triplicates of extraction), and a calibration curve was analyzed for each batch of analysis (n = 8), to ensure accuracy. Blanks (acetonitrile:water (40:60)) and duplicates of QCs standards at three CLs (75, 1300, and 2500 ng mL−1 of β-ODAP), were also run between every 20 injections to verify the instrument response. Summary statistics and correlations were calculated (GenStat, 19th Ed.).