Rapid and Simultaneous Quantification of Five Quinolizidine Alkaloids in Lupinus angustifolius L. and Its Processed Foods by UPLC–MS/MS

Abstract

Quinolizidine alkaloids (QAs) are toxic secondary metabolites of Lupinus plants. This study reports the simultaneous quantification of five alkaloids from Lupinus angustifolius L. and its processed foods by ultraperformance liquid chromatography with electrospray ionization mass spectrometric detection. After optimizing the extraction conditions, the analytical method was validated for the QAs in lupin beans and in three major processed foods through detection and quantification limits, linearity, precision, and accuracy. The detection and quantification limits of the QAs were 0.5–1.7 and 1.5–5.7 mg kg^–1, respectively. The linearity was greater than 0.9992 and the precision was less than 3.1%. The recoveries of three different concentrations of each QA were 89.5–106.2%. For both raw lupin beans and processed foods, lupanine was the major QA, and 13-hydroxylupanine and angustifoline were detected in lower amounts. This method could be widely used for accurate risk assessment of QAs.

1. Introduction

In recent years, legumes, especially blue lupins such as Lupinus angustifolius L., have been a major grain crop in southern Australia, Europe, America, and South Africa.¹ Lupins are consumed as food and feed instead of animal protein because of their high protein content (34–43%, dry matter) and essential amino acid content.² The lupin protein has been reported to reduce cholesterol and blood pressure, as well as acutely reduce glycemia in type 2 diabetes.³

However, Lupinus species are known to contain quinolizidine alkaloids (QAs), which are toxic secondary metabolites to animals and humans,⁴ causing arrhythmia, mydriasis, vomiting, depression, labored breathing, trembling, or convulsions.⁵ Lupanine, 13-hydroxylupanine, and angustifoline are tetracyclic compounds having a pyridone nucleus, and are representative toxic alkaloid compounds detected in Lupinus.^6,7 The acute oral lethal dose (LD₅₀) of QAs to rats is 2279 mg kg^– 1 for L. angustifolius seeds and 1664 mg kg^–1 for lupanine.⁸ In addition, the mouse LD₅₀ of lupanine and sparteine for intraperitoneal administration are 175 and 36 mg kg^–1, respectively.⁹ Accordingly, the Australia and New Zealand Food Authority (now Food Standards Australia New Zealand) has set the maximum level of 200 mg kg^–1 for QAs in lupin seeds or processed products.¹⁰

Lupanine is the major QA in the Lupinus species. 13-Hydroxylupanine and angustifoline have also been detected in large amounts in Lupinus, while sparteine and lupinine have been reported in smaller amounts in the literature.¹¹ For accurate quantitative analysis of a compound, a standard material with guaranteed purity is required. There are five QAs that can be obtained commercially with good purity including lupanine, 13-hydroxylupanine, angustifoline, sparteine, and lupinine. Therefore, an accurate analytical method for QAs in the Lupinus species and lupin processed products would include these five substances.

There are several reported experimental methods for analyzing QAs in Lupinus species. Analysis of the alkaloids was initially done by the thin-layer chromatography method.¹² Later, QAs were analyzed by the capillary electrophoresis method,¹³ infrared and Raman spectroscopy,¹⁴ and near-infrared spectroscopy.¹⁵ Recently, an analytical method was developed using gas chromatography–mass spectrometry (GC–MS) for accurate analysis of QAs with similar structures.^16,17 However, GC–MS analysis has a complicated preconditioning process, long analytical time, and poor reproducibility. Unfortunately, most of the above methods were not properly validated, such as by extraction efficiency, instrumental (blank) validation, and standard spiked sample validation. Therefore, in this study, a liquid chromatography–mass spectrometry (LC–MS) analytical technique was developed and validated with a simple pretreatment process and short analysis time. The QA extraction method needs to establish an effective procedure such as with minimal time and effort from various food samples. Optimal extraction conditions for QAs were established by comparing different extraction solvents [methanol and acetonitrile (ACN) base], methods (shaking and sonication), extraction times (5, 20, 40, 60, 120, and 180 min), and efficiencies. In addition, the optimal MS analysis conditions were established by confirming the precursor and fragment ions of each QA using ultraperformance liquid chromatography–tandem mass spectrometry (UPLC–MS/MS).

This study aimed to (i) develop and validate an analytical method for the rapid and simultaneous quantification of five QAs—lupinine, 13-hydroxylupanine, lupanine, angustifoline, and sparteine—by UPLC–MS/MS, (ii) apply the established method to the lupin (L. angustifolius) raw materials and lupin processed products, and (iii) determine the content of QAs in the raw lupin and lupin processed products.

2. Results and Discussion

2.1. Validation of Analytical Methods and Quality Assurance

2.1.1. Optimization of the Extraction Procedure

In order to determine the optimal extraction method, 80% methanol and 80% ACN were selected as the extraction solvents, and shaking and ultrasonic extractions were performed with different extraction times (5, 20, 40, 60, 120, and 180 min) on homogenized raw lupin beans. The extraction efficiency results of the five QAs are shown in Table S1. First of all, the efficiencies of the extracting solvents were compared, and it was determined that 80% methanol was more efficient than 80% ACN using the same extraction method. Second, it was determined that the ultrasonic extraction method had higher efficiency than shaking for the same extraction time. Finally, using ultrasonic extraction with 80% methanol, it was determined that the extraction was completed within 60 min or less for all QAs. Therefore, the optimal extraction procedure for the QAs was determined to be ultrasonic extraction with 80% methanol for 60 min.

2.1.2. Performance of the Analysis Method

The instrument validation results such as the limit of detection (LOD), limit of quantification (LOQ), linearity, and the standard solution equation for the QAs are given in Table 1. The working solution concentration range was 10–400 μg L^–1. Angustifoline had the lowest LOD at 0.503 μg mL^–1, and lupanine had the highest LOD at 1.300 μg mL^–1. Similarly, the lowest LOQ was for angustifoline (1.509 μg mL^–1) and the highest was for lupanine (3.901 μg mL^–1). In the case of linearity, sparteine was the lowest at 0.9977 and angustifoline was the highest at 0.9997. As seen by the calibration curve equation of each QA standard solution, lupanine showed the highest sensitivity, which was approximately 3 times higher than that of lupinine, which had the lowest sensitivity.

Table 1. UPLC–MS/MS Method Validation Parameters for QA Analysis.

lupin alkaloids	retention time (min)	range (μg/mL)	LOD (μg/mL)	LOQ (μg/mL)	linearity (R2)	equation
lupinine	2.46	10–400	1.300	3.901	0.9996	y = 1883.4x – 3820.8
13-hydroxylupanine	2.49	10–400	1.245	3.735	0.9992	y = 3073.1x + 7254.7
lupanine	2.64	10–400	1.915	5.746	0.9993	y = 5160.0x + 15,708
angustifoline	2.81	10–400	0.503	1.509	0.9997	y = 4590.1x + 6991.3
sparteine	3.07	10–400	0.785	2.356	0.9977	y = 2090.1x + 8526.4

Method validation was carried out for raw lupin beans and three processed foods with lupin—noodles, biscuits, and bean milk. Specificity was verified to distinguish the QA from interfering substances. Recovery and precision were verified by the standard addition method. The recovery test was performed by adding standard solutions at three concentrations (low level: 25 mg kg^–1, medium level: 500 mg kg^–1, and high level: 2000 mg kg^–1). The results are shown in Table 2. The range of recovery was 89.2–108.4% and the precision range was 0.3–5.4%. The obtained validation results confirm that there was no interference of matrix compounds for the quantitative analysis of the QAs. In addition, the methods described in this study satisfied the required criteria for the application of analytical methods.¹⁹ Analysis of QAs by GC requires a pretreatment process of reconstituting after concentrating nitrogen. However, the pretreatment method developed for UPLC–MS/MS omits this step and thus saves time and effort. In addition, the instrumental analysis time of this study was 5 min, which was significantly shorter than that of alkaloid’s 40 min determined using a general LC technique.^20,21

Table 2. Method Validation Results of the QA Analysis^a.

		spiked recovery
		low levelb		medium levelc		high leveld
sample	compound	intradaye	interdayf	intraday	interday	intraday	interday	LOD	LOQ
L. angustifolius	lupinine	89.5 ± 1.8	89.2 ± 2.6	90.5 ± 1.3	91 ± 2.3	94.9 ± 1.9	94.9 ± 2.4	2.865	8.594
	13-hydroxylupanine	103.4 ± 0.5	104.5 ± 1.4	104.3 ± 1.3	105.4 ± 1.7	106.2 ± 1.3	106.0 ± 1.0	2.239	6.716
	lupanine	97.3 ± 4.1	98.9 ± 3.1	97.9 ± 1.5	99.7 ± 2.0	99.0 ± 2.8	100.3 ± 1.9	4.549	13.646
	angustifoline	96.8 ± 1.0	97.7 ± 1.9	99.0 ± 1.6	99.2 ± 1.5	99.1 ± 1.0	100 ± 1.4	2.060	6.181
	sparteine	105.7 ± 0.5	107.0 ± 1.2	105.6 ± 0.4	106.7 ± 1.0	100.0 ± 3.2	102 ± 2.2	3.663	10.988
noodles	lupinine	91.8 ± 1.2	91.7 ± 1.9	91.4 ± 0.1	91.5 ± 1.5	94.4 ± 0.9	94.7 ± 2.0	1.188	3.563
	13-hydroxylupanine	104.1 ± 1.7	104.8 ± 1.8	102.6 ± 0.6	104.4 ± 1.6	99.6 ± 0.5	101.0 ± 1.3	1.318	3.954
	lupanine	100.9 ± 5.4	102.9 ± 3.3	97.4 ± 3.1	99.0 ± 2.7	97.7 ± 1.4	98.5 ± 1.5	4.323	12.970
	angustifoline	98.7 ± 1.1	100.0 ± 1.6	97.7 ± 0.2	97.9 ± 1.3	98.2 ± 0.6	98.6 ± 1.1	1.017	3.052
	sparteine	105.5 ± 0.9	107.6 ± 1.7	103.3 ± 2.1	105.0 ± 1.8	98.8 ± 1.0	100.3 ± 1.6	2.406	7.217
biscuits	lupinine	97.8 ± 0.4	95.7 ± 2.3	93.2 ± 2.6	92.1 ± 2.6	94.5 ± 2.4	93.8 ± 2.5	2.416	7.249
	13-hydroxylupanine	105.4 ± 2.1	104.3 ± 1.7	99.6 ± 1.5	99.9 ± 1.0	99.6 ± 2.2	99.6 ± 1.4	2.339	7.017
	lupanine	100.9 ± 3.6	102.1 ± 2.6	96.6 ± 4.0	97.4 ± 2.8	96.9 ± 1.3	97.4 ± 1.4	1.343	4.030
	angustifoline	98.5 ± 0.8	98.9 ± 1.2	95.5 ± 1.2	95.7 ± 1.5	96.9 ± 1.2	97.7 ± 2.3	1.132	3.396
	sparteine	105.0 ± 0.6	106.7 ± 1.4	103.7 ± 1.0	104.3 ± 1.5	98.8 ± 0.6	99.3 ± 0.7	2.677	8.031
bean milk	lupinine	93.4 ± 1.1	92.0 ± 1.7	91.7 ± 1.7	91.1 ± 1.8	94.1 ± 1.8	94.2 ± 2.1	2.807	8.422
	13-hydroxylupanine	99.0 ± 0.9	100.3 ± 1.6	96.8 ± 1.8	97.8 ± 1.6	96.8 ± 1.8	97.1 ± 1.5	2.869	8.607
	lupanine	99.6 ± 3.2	101.6 ± 2.9	98.6 ± 1.2	99.0 ± 1.9	97.9 ± 1.2	98.6 ± 1.4	2.641	7.924
	angustifoline	98.8 ± 1.3	99.7 ± 1.8	97.5 ± 0.8	97.6 ± 1.6	97.1 ± 0.1	97.6 ± 1.1	0.821	2.464
	sparteine	106.4 ± 0.3	108.4 ± 1.8	105.7 ± 0.3	106.4 ± 1.1	97.8 ± 2.6	99.9 ± 2.2	2.881	8.644

^aValues are mean ± standard deviations of three (n = 3) measurements.

^b25 μg/L standard solution added to sample.

^c500 μg/L standard solution added to sample.

^d2000 μg/L standard solution added to sample.

^eThree repeated analyses on the same day.

^fRepeated analysis on 3 different days.

2.2. Quantitative Analysis of Samples

QA quantitative analysis was performed using the established optimal analysis method on raw L. angustifolius (9) and seven types of processed foods containing lupin—pickles (6), noodles (19), biscuits (20), bean milk (2), bean patties (3), baking powder (2), and sauces (2). The results are given in Table 3, and the analytical chromatogram of L. angustifolius as a representative sample is shown in Figure 1.

Table 3. Concentration (mg kg^–1) of QAs in Raw L. angustifolius and Processed Foods^a.

sample		nob	lupinine	13-hydroxylupanine	lupanine	angustifoline	sparteine	total lupin alkaloids
raw material	L. angustifolius	9	N.Dc	2975 ± 1848	35,707 ± 22,771	825 ± 515	34.3 ± 22.3	39,541 ± 25,115
				126–4608	278–51,810	35.6–1347	21.6–63.2	440–57,157
lupin processed food	pickles	6	N.D.	19 ± 33.9	258 ± 487.6	9.0 ± 15.5	N.D.	286 ± 536.7
				0.3–87	12–1248	0.4–40		13–1375
	noodles	19	N.D.	15 ± 8.3	43 ± 25	5.3 ± 3.3	2.5 ± 11.3	62 ± 35
				2.9–27	12–91	0.9–11	N.D.–51.1	20–1
	biscuits	20	N.D.	1.2 ± 3.1	6.6 ± 7.9	0.4 ± 0.6	N.D.	8.2 ± 10.8
				N.D.–14	N.D.–33	N.D.–2.0		N.D.–47
	bean milk	2	N.D.	0.8 ± 0.4	3.4 ± 4.4	N.D.	N.D.	4.2 ± 4.8
				0.6–1.0	0.3–6.5			0.9–7.6
	bean patties	3	N.D.	36 ± 45.1	185 ± 274	15 ± 22.6	N.D.	236 ± 341.7
				9.0–88	19–502	0.4–41		33–630
	baking powder	2	N.D.	84 ± 11.7	551 ± 107.6	40 ± 4.9	N.D.	676 ± 124.3
				76–93	475–627	37–44		588–764
	sauces	2	N.D.	20 ± 9.4	66 ± 31.6	6.7 ± 3.4	N.D.	92 ± 44.3
				13–27	43–88	4.4–9.1		61–124

^aValues are mean ± standard deviations of three measurements.

^bNumber of samples analyzed.

^cN.D.: not detected.

Figure 1.

Representative analyzed L. angustifolius chromatograms of lupin alkaloids. (A) Lupinine, (B) 13-hydroxylupanine, (C) lupanine, (D) angustifoline, and (E) sparteine.

The QA content can be affected by the species and the cultivation environment during the growth period.²² However, lupanine and 13α-hydroxylupanine have been reported to be the major QAs in L. angustifolius.¹¹ In this study, the concentration of lupanine was 35,707 ± 22,771 mg kg^–1, which was about 90.3% of the sum of the five analyzed QAs (39,541 ± 25,115 mg kg^–1). The concentrations of 13-hydroxylupanine, angustifoline, and sparteine were 2975 ± 1848, 825 ± 515, and 34.3 ± 22.3 mg kg^–1, respectively. In the processed foods, lupanine (258 ± 487.6 mg kg^–1) was the major compound in pickles and accounted for 90.2% of the total pickle QAs, which is a similar percentage to that found in raw L. angustifolius. 13-Hydroxylupanine and angustifoline were detected in pickles at 19 ± 33.9 and 9.0 ± 15.5 mg kg^–1, respectively, and sparteine was not detected. These low amounts indicate that the QA contents were reduced by the debittering process involving soaking or washing with water.²³ The total QAs in noodles, biscuits, bean milk, and sauces were lower than the average of 92 mg kg^–1. The total QAs in bean patties and baking powder were 236 ± 641.7 and 676 ± 124.3 mg kg^–1, respectively, which are higher values than those of the other processed products. However, the proportions of each of the QAs in all the processed foods were similar to that of raw L. angustifolius (Table S2).

In this study, a rapid and simultaneous quantification method for five QAs from L. angustifolius L. and its processed foods using UPLC–MS/MS was developed, optimized, and validated. Analytical conditions such as the extraction method, separation, and MS analysis conditions for the QAs were optimized. This drastically shortened the analysis time than other conventional methods. Besides, the validation results satisfied the Association of Official Analytical Chemists (AOAC) standard. The established analytical method is expected to help risk assessment through accurate analysis of QAs in lupin samples. From the analysis of raw L. angustifolius, lupanine was found to constitute approximately 90.3% of the total QA content. Lupanine was also identified as the main compound in lupin processed foods; however, much of the alkaloid was removed during processing, so the lupanine content was relatively lower than that in the raw material. Despite the lower content, monitoring is still considered necessary because some samples exceeded the normal health limit of 200 mg kg^–1. The rapid analysis method of lupine alkaloids established in this study is considered to be useful for safety management of L. angustifolius and processed foods.

3. Materials and Methods

3.1. Reagents and Instrumentation

Lupanine (≥98%), angustifoline (≥98%), and 13-hydroxylupanine (≥98%) were purchased from ChemFaces (China); sparteine (>98%) was purchased from Santa Cruz Biotechnology, Inc. (USA). Lupinine and heptafluorobutyric acid (HFBA, ≥98%) were purchased from Sigma-Aldrich (USA). Methanol (HPLC grade, ≥99.8%) was purchased from J.T. Baker (USA). Deionized water (≥18.2 MΩ·cm) was prepared using an ultrapure water system (arium pro, Sartorius, Germany).

For analysis, an Acquity 1 UPLC system (Waters, Millford, MA, USA) was employed with a Xevo TQ-S triple quadrupole mass spectrometer (Waters, Millford, MA, USA). These instruments were controlled using the MassLynx software program (Waters).

3.2. Sample Collection and Preparation

A total of 84 commercially available samples including raw L. angustifolius (9) and seven types of processed foods that contain lupin, such as pickles (6), noodles (19), biscuits (20), bean milk (2), bean patties (3), baking powder (2), and sauces (2), were purchased from online markets from May to August 2018. All samples were homogenized using a laboratory mill (MR 350CA, Braun, Spain), transferred to properly labeled polypropylene tubes, and stored at −20 °C until analysis.

3.3. Sample Extraction

To optimize the extraction method, two extraction solvents (80% methanol and 80% ACN) and two different extraction processes (shaking and sonication) with various times (5, 20, 40, 60, 120, and 180 min) were examined. All optimization experiments used ground lupin bean powder as the sample. The sample (1 g) was dissolved in 50 mL of the extraction solvent and the desired extraction process was performed. The extract was centrifuged at 3200g for 15 min, and the resulting supernatant was filtered through a 0.22 μm syringe filter. The filtrate was diluted 5-fold with the extraction solvent and then directly injected into the UPLC instrument for analysis.

3.4. Analytical Conditions

Separation of the five QAs was performed using an Atlantis T3 C18 column (i.d. = 100 mm × 2.1 mm and particle diameter = 3 μm; Waters) and 0.1% HFBA in water and 0.1% HFBA in MeOH as mobile phases A and B, respectively (flow rate: 0.5 mL min^–1). The sample injection volume was 5 μL, the column oven temperature was 35 °C, and the autosampler temperature was set to 10 °C. The following gradient elution program was used: 0–0.5 min, 95% A; 0.5–4.0 min, 0% A; 4.0–4.5 min, 0% A; and 4.5–5.0 min, 95% A.

The mass spectra were acquired in the multiple reaction monitoring (MRM) mode using the following optimized parameters: dwell time = 0.3 ms per transition, capillary voltage = 2.5 kV, cone voltage = 20 V, source temperature = 150 °C, and desolvation temperature = 500 °C. The nitrogen flow rates for the cone and desolvation were 150 and 800 L h^–1, respectively. Argon was used as the collision gas at a flow rate of 0.15 mL min^–1.

The QAs were identified and quantified based on m/z values determined in the MRM mode. The conditions and fragmentation patterns are shown in Table 4.

Table 4. Mass Spectrometry Data for QAs.

compound name	chemical formula	molecular weight (g/mol)	adduct	precursor ion, m/z	collision energy (CE)	fragments, m/za
lupinine	C10H19NO	169.268	[M + H]+	170.1539	24	152.1436, 136.1124, 124.1125, 96.0815
13-hydroxylupanine	C15H24N2O2	264.369	[M + H]+	265.1911	26	247.1809, 221.1636, 152.1073, 114.0919, 69.9728
lupanine	C15H24N2O	248.370	[M + H]+	249.1961	25	150.1280, 136.1124, 114.0918, 69.0707
angustifoline	C14H22N2O	234.343	[M + H]+	235.1805	18	193.1339, 150.0916, 136.1122, 112.0762, 84.0816
sparteine	C15H26N2	234.387	[M + H]+	235.2169	27	152.1436, 134.0969, 97.0971, 84.0816, 69.0706

^aThe product ion with the highest intensity is underlined.

3.5. Validation of Analytical Methods and Quality Assurance

The analytical method was validated by determining the selectivity, standard solution concentration linearity, limits of detection and quantification, precision, and recovery. In addition, intraday (three repeated analyses on the same day) and interday (repeated analysis on 3 different days) validation methods were used. The validation and extraction studies were performed using raw lupin beans and three types of lupin-processed samples (noodles, biscuits, and bean milk) according to a previously reported procedure.¹⁸

3.5.1. Selectivity

The QA selectivity was evaluated by the peak retention time of each compound measured in the MRM mode. The results were compared with the sample peaks of standard QA solutions.

3.5.2. Linearity

The linearity was evaluated from the calibration curves of each QA using a nonweighted least-squares linear regression analysis method. Standard working solutions were prepared at a concentration range of 10–400 μg L^–1 (with calibration levels of approximately 10, 20, 50, 100, 200, 300, and 400 μg L^–1).

3.5.3. Limits of Detection and Quantification

The limit of detection (LOD) and limit of quantification (LOQ) were assessed experimentally with 3 and 10 times the standard deviation of the blank divided by the slope of the analytical standard curve.

3.5.4. Precision

Precision was obtained as percent coefficient of variation (CV %) from the relative standard deviation of 10 repeated determinations using each standard solution of QA in a spiked sample.

3.5.5. Accuracy

Accuracy was verified by adding three concentrations (25, 500, and 2000 mg L^–1) of the standard material solution from each sample type and then determining the recovery (%).

Glossary

Abbreviations

HFBA

heptafluorobutyric acid

QA

quinolizidine alkaloid

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.0c01929.

• Concentration (mg kg^–1) of lupin alkaloids in lupin beans according to various extraction methods and relative amounts (%) of QAs in raw L. angustifolius and processed foods (PDF)

This work was supported by a grant (18162MFD037) from the Ministry of Food and Drug Safety in 2018 and the World Institute of Kimchi (grant number KE2002-2-2), funded by the Ministry of Science and ICT, Republic of Korea.

The authors declare no competing financial interest.