1. Introduction
The faba bean (
Vicia faba L.) is a nitrogen-fixing legume that has both economic and environmental benefits. In Australia, 2.3 million tons of different pulses, including faba bean, field pea (
Pisum satvium L.), chickpea (
Cicer arietinum L.), lentils (
Lens culinaris Medik.), Australian sweet lupin (
Lupinus angustifolius L.) and mungbean (
Vigna radiata L.), were produced annually from 2009 to 2015 with primary production in South Australia, Victoria and New South Wales [
1]. Globally, in 2016, 4.5 million tons of faba bean crop was cultivated, with China, Ethiopia, and Australia being the main producers [
2].
One of the earliest domesticated crops, faba beans have a long history of being used for human consumption [
2,
3], but they are also commonly used as a crop for animal feed, foraging, and medicine, and are one of the most versatile globally produced crops [
2]. The consumption of faba beans, like most pulses, contributes to a balanced diet, due to the high protein content of 26–38% of the seed [
1,
4,
5,
6], which is also high in lysine, carbohydrates [
1,
5], fibre, and phytochemicals [
5].
Despite the potential benefits, faba beans also produce anti-nutritional compounds that can have adverse effects by reducing nutrient digestibility [
3], limiting their use in food and feed formulations [
1]. Vicine (2,6-diamino-4,5-dihydroxypyrimidine-5-β-D-glucopyranoside) and convicine (2,4,5,trihydroxy-6-aminopyrimidine-5-β-D-glucopyranoside) are the two major anti-nutritional compounds found in faba beans, with levels varying depending on cultivar, maturation, cultivation climate, and soil properties [
3,
7]. Vicine and convicine (v-c) are reported to be synthesised in the testa during the seed filling stage [
2,
8,
9], before being transported to the cotyledon, where they accumulate into a concentration of 6–14 g/kg [
1,
2,
5,
10].
Once consumed, v-c are enzymatically degraded by the β-glucosidase enzyme in the small intestine to the reactive aglycone divicine (2,6-diamino-4,5-dihydroxypyrimidine) and isouramil (6-Amino-2,4,5-trihydroxypyrimidine) [
1,
2,
4,
5,
6,
11,
12,
13]. Accumulation of the aglycones can potentially be toxic to individuals with a genetic deficiency of glucose-6-phophate dehydrogenase, leading to haemolytic anaemia or favism [
1,
3,
4,
5,
10,
11,
14]. Favism affects approximately 400 million people globally [
1,
4,
11], with the highest prevalence in Asia, the Mediterranean and Africa [
6,
11].
Concentrations of v-c in faba bean seeds and flour can be reduced or eliminated by roasting, boiling, or microwaving. Furthermore, soaking in water, a weak acid, alkaline solution, or fermentation [
1,
2,
5,
6,
10] prior to consumption also reduces the risk of accumulating aglycones. These processes are important, as air classification during industrial scale processing has been shown to increase the concentration of v-c [
1,
2] up to 4-fold in the protein fraction [
2].
The accurate determination of v-c in faba beans is especially important due to the increased interest in developing low-v-c cultivars. This research is underway, though not yet commercially available [
1,
2,
4,
5,
6]. Current methods of extraction commonly involve long complex acid-extraction protocols using perchloric acid [
1,
4,
5,
6,
10,
13,
14] and hydrochloric acid [
3,
13]; or large sample weights >1 g with an organic solvent such as methanol [
10,
12] or ethanol [
6] and up to 50% water. Purves et al. performed a detailed extraction optimisation study comparing the responses of samples extracted with water; water with 1% formic acid; 70:30 acetone:water; 70:30 methanol:water; and 70:29:1 methanol:water:formic acid, finding that extraction with an organic solvent provided consistent results without the risk of continued biological activity that may occur in the water extract [
4].
Analytical techniques used for the quantitation of v-c initially used spectrophotometry and colorimetry methods that were suitable at high concentrations [
2,
4]. Now, techniques predominantly employ liquid chromatography (LC) with UV detection [
1,
2,
3,
4,
5,
6,
10,
12,
13,
14], all of which lack the selectivity and sensitivity to accurately quantitate low levels of v-c [
2]. Generally, methods use silica-based C
18 columns for compound separation, and HILIC [
4] and silica-based pentafluorophenylpropyl [
1] columns have also been reported, although the long-term stability and reproducibility of the methods remain unclear. Utilising mass spectrometry (MS) provides greater selectivity and sensitivity, allowing for the accurate determination of cultivars with standard v-c levels, while also allowing for the accurate determination of low-v-c cultivars.
Despite the potential selectivity and sensitivity improvements, MS has predominantly been used as a confirmatory tool, complimenting UV quantitation for v-c [
1,
5,
6,
13]. Purves et al. used LC-MS to quantify the v-c concentration in 13 faba bean seeds, reporting significant improvements in selectivity and sensitivity compared to the UV analysis of the same samples, particularly for convicine [
4]. The accurate measurement of v-c is essential to determine whether a crop is suitable for consumption; however, the current reported methods require improvement, especially for convicine [
2].
Here, we report on a simple high-throughput method for the extraction and analysis of vicine and convicine in faba bean flour, improving on the previously reported method in terms of extraction simplicity, and the accuracy and precision of quantitation.
3. Materials and Methods
3.1. Reagents and Standards
All extraction and mobile-phase solvents were of HPLC grade. Methanol (≥99.9% pure), acetonitrile with 0.1% formic acid (≥98.5% pure), and water with 0.1% formic acid were purchased from Fisher Chemical (Fair Lawn, NJ, USA).
Vicine and convicine standards were purchased from Novachem Pty Ltd. (Heidelberg West, VIC, Australia) as the distributor for Toronto Research Chemicals (Toronto, ON, Canada). A stock solution of 25,000 ng/mL vicine and 10,000 ng/mL convicine was prepared using 10% methanol in water. Serial dilution was performed to prepare working standard solutions of 2500, 1250, 250, 125, 25 and 12.5 ng/mL vicine and 1000, 500, 100, 50, 10, and 5 ng/mL convicine in 80% methanol.
3.2. Sample Preparation
Seven commercially available Australian faba bean cultivars (PBA Amberley, PBA Bendoc, Farah, PBAMarne, PBA Nasma, PBA Samira and PBA Zahra) were sources from trials grown at Curyo, Victoria, Australia. The samples were milled to a homogenous powder of less than 0.5mm for analysis.
3.3. Extraction Optimisation
Each sample (10.0 + 0.2 mg) was weighed into an Axygen 2.0 mL microcentrifuge tube with analytical balance (Sartorius. MSU225S, Göttingen, Germany). Samples were extracted with 1 mL of 80% methanol (methanol and milli-Q water, 80:20, v:v), vortexed for 1 min (Ratek multitube vortex mixer, MTV1, Boronia, Victoria, Australia), sonicated for 5 min (SoniClean, 250TD, Thebarton, South Australia, Australia), and centrifuged at 13,000 rpm for 5 min (Eppendorf, 5415D, Hamburg, Germany). The supernatant was transferred to a pre-labelled 2.0 mL LC-MS vial for analysis. The pellet was re-extracted a further three times, with the supernatant transferred to an empty vial each time for analysis to determine extraction efficiency.
3.4. Method Validation
Samples were weighed and extracted as previously described. The supernatant was transferred to a second microcentrifuge tube, the pellet re-extracted a second time, and the supernatants combined. Samples were diluted 1:20 to fit within the linear range of the instrument. The method was validated for linearity; limit of detection (LOD); limit of quantitation (LOQ); accuracy; precision; repeatability; and matrix effect.
LOD and LOQ were determined by multiplying the standard error of the intercept—obtained using the regression data analysis tool in MS excel—by 3.3 for LOD and 10 for LOQ, then dividing the result by the slope of the curve.
Accuracy, precision, and repeatability of the standards were determined using the calculated concentration of the standards across five replicate injections compared to the expected concentrations. Extract repeatability was determined from comparing the results of seven replicate extractions of each cultivar, obtained as described above.
Matrix effect was determined by adding standards at two different levels—high (50,000 ng/mL vicine and 20,000 ng/mL convicine (HS)) and low (50 µL of 2500 ng/mL vicine 1000 ng/mL convicine (LS))—to different extracts of each of the seven cultivars. Each undiluted cultivar extract (50 µL) was combined with 50 µL of the high or low standard, and then made up to a final volume of 1 mL with 900 µL of 80% methanol.
3.5. Instrumentation and Data Analysis
Samples were analysed using a Thermo Scientific Vanquish ultra-high-performance liquid chromatography (UHPLC) system (Thermo Fisher Scientific, Bremen, Germany) coupled to a Thermo Fisher Q Exactive Plus mass spectrometer (QE MS) (Waltham, MA, USA; Thermo, Bremen, Germany). All MS data were acquired in positive electrospray ionization (ESI) mode over a mass range of 80–1200 m/z. The resolution was 35,000, the normalized collision energy was 30 V, and the maximum ion time was 200 milliseconds. The source heater temperature was maintained at 310 °C, and the heated capillary maintained at 320 °C. The sheath, auxiliary and sweep gases (N2) were 28, 15 and 4 units, respectively. Spray voltage was set at 3.6 kV. Prior to data acquisition, the system was calibrated with Pierce® LTQ Velos ESI Positive Ion Calibration Solution (Thermo Scientific, product no. 88323).
Analytes were separated on a Phenomenex Synergi Polar-RP (150 mm × 2 mm, 4 μm) HPLC column with an isocratic mobile phase of 20% A (0.1% formic acid in water) and 80% B (0.1% formic acid in acetonitrile) over 4 min. The column was cleaned for 3 min at 100% B, before returning to the initial conditions for 3 min for re-equilibration and a total analysis time of 10 min. The flow rate of the method is 0.15 mL/min, with the column maintained at 40 °C.
All acquired data were quantitatively processed using Tracefinder 5.1 Build 110 (Thermo Fisher Scientific, San Jose, CA, USA).