Identification and Quantification of Major Faba Bean Seed Proteins.

Faba bean (Vicia faba L.) holds great importance for human and animal nutrition for its high protein content. However, better understanding of its seed protein composition is required in order to develop cultivars that meet market demands for plant proteins with specific quality attributes. In this study, we screened 35 diverse Vicia faba genotypes by employing the one-dimensional sodium dodecyl sulfate-polyacrylamide gel electrophoresis (1D SDS-PAGE) method, and 35 major protein bands obtained from three genotypes with contrasting seed protein profiles were further analyzed by mass spectrometry (MS). Twenty-five of these protein bands (MW range: ∼ 9-107 kDa) had significant (p ≤ 0.05) matches to polypeptides in protein databases. MS analysis showed that most of the analyzed protein bands contained more than one protein type and, in total, over 100 proteins were identified. These included major seed storage proteins such as legumin, vicilin, and convicilin, as well as other protein classes like lipoxygenase, heat shock proteins, sucrose-binding proteins, albumin, and defensin. Furthermore, seed protein extracts were separated by size-exclusion high-performance liquid chromatography (SE-HPLC), and percentages of the major protein classes were determined. On average, legumin and vicilin/convicilin accounted for 50 and 27% of the total protein extract, respectively. However, the proportions of these proteins varied considerably among genotypes, with the ratio of legumin:vicilin/convicilin ranging from 1:1 to 1:3. In addition, there was a significant (p < 0.01) negative correlation between the contents of these major fractions (r = -0.83). This study significantly extends the number of identified Vicia faba seed proteins and reveals new qualitative and quantitative variation in seed protein composition, filling a significant gap in the literature. Moreover, the germplasm and screening methods presented here are expected to contribute in selecting varieties with improved protein content and quality.

Introduction

Faba bean (Vicia faba, hereafter Vf) seeds contain about 29% protein,[1] and the crop is well adapted to various climates and is grown for both human and animal nutrition.[2,3] Given its high yield potential[4] and unparalleled nitrogen-fixation capacity,[5]Vf is among the few crops with great potential to meet the dietary needs of the growing human population while maintaining sustainability of agricultural production systems.[6] Much of the research on seed quality to date has focused on the reduction or removal of antinutrients, namely, vicine and convicine[7−9] and seed coat tannins,[10−12] with surprisingly little effort dedicated to improving the protein composition.

Utilization of plant proteins for human or animal nutrition is largely determined by the nutritional and functional properties of their constituent protein classes. It is estimated that Vf seed proteins contain ∼ 80% globulin which in turn is comprised of legumin and vicilin/convicilin, also referred to as 11S and 7S, respectively, based on their ultracentrifugation sedimentation coefficients,[13] respectively. Globulins belong to the cupin superfamily,[14] and the legumin and vicilin types have a high degree of structural homology.[15,16]

Legumin is a major Vf seed protein, representing about 50% of the storage proteins.[17,18] It is encoded by multiple genes belonging to type-A (Methionine-containing) and type-B (Methionine-lacking) subunits.[18] Only few genes encoding type-A (A1 and A2), type-B (LeB2, LeB4, LeB6, and LeB7), and one high-molecular mass legumin polypeptide (LeB3) have been described[18−20] in the literature. However, Tucci et al.[21] reported 29 biochemically distinct disulfide-linked αβ legumin subunit pairs with molecular weights between 39–81 kDa, suggesting that the number of legumin-encoding genes could be much more than is currently known. Vicilin is also a heterogeneous protein in its native trimer form.[21] Regarding convicilin subunits, at least two structural genes have been described,[22] though the question of whether convicilin can be considered a vicilin subunit or a distinct class of globulin is yet to be resolved in Vf.

The relationship between subunit composition of major storage proteins and the overall seed protein quality has been studied in other legumes like soybean,[23] where molecular markers for specific legumin and vicilin-like subunit variants with desirable qualities have been developed.[24] In Vf, it is generally accepted that selection for a higher legumin:vicilin ratio could enhance its nutritional quality[1,25,26] since some major legumin subunits contain relatively higher proportions of sulfur-containing amino acids (S-AA) compared to vicilin. However, given the underlying genetic complexity of these broadly defined classes of storage proteins, concrete exploitation of genetic variation in seed protein composition for the development of cultivars with improved protein profiles would require identification of the genes encoding the major seed storage proteins, as well as understanding their synthesis, transport, and storage mechanisms. To date, studies have referred to just a few major protein subunits of legumin and vicilin[17,21,25−27] and although Liu et al.[28] identified several additional nonglobulin seed storage proteins from Vf by mass spectrometry, the identification of the full set of proteins that contribute to the nutritional and functional properties of the Vf seed is far from complete.

The one-dimensional sodium dodecyl sulfate–polyacrylamide gel electrophoresis (1D SDS-PAGE) method has been exploited in the qualitative and quantitative analysis of protein composition in various legume species.[24,29−31] However, the main problem inherent in this method is that depending on the particular electrophoresis conditions used, unrelated proteins of similar mobility can partially or completely overlap, which can lead to over- or underestimation of certain protein subunits. An alternative method of protein separation based on size-exclusion high-performance liquid chromatography (SE-HPLC) has been widely used in studying seed proteins of wheat, notably in determining the proportions of gliadin and glutenin fractions associated with certain quality attributes, including pasta-cooking and bread-making qualities.[32−35] The advantage of this method is that proteins can be quantified in their native condition and the sample analysis is amenable for automation.

In this study, our aim is to (1) assess the diversity in subunit composition of major Vf seed proteins in genetically diverse germplasm; (2) accurately identify the most abundant seed proteins; and (3) quantify the proportions of legumin and vicilin/convicilin proteins using a panel of diverse Vf genotypes.

Materials and Methods

Reagents

Sodium phosphate, calcium chloride, trichloroacetic acid, dithiothreitol, iodoacetamide, triethylammonium bicarbonate, and a Bradford assay reagent were obtained from Sigma-Aldrich (UK). PageBlue, NuPAGE LDS sample buffer, and NuPAGE MES SDS running buffer were sourced from Thermo Fisher Scientific (UK). Acetonitrile, sulfuric acid, and HPLC grade water were obtained from Fisher Scientific (UK). A sequence grade porcine trypsin enzyme was obtained from Promega (UK).

Plant Materials

Thirty-five Vf genotypes, including inbred lines derived from breeding materials, landraces, and cultivars from different locations around the world, were used in this study for 1D SDS-PAGE and SE-HPLC protein subunit profiling (Table ). This genetically diverse population contained genotypes collected by the University of Reading (UK), the Agricultural Research Center (Egypt), Nordic Seeds (Denmark) and the University of Saskatchewan (Canada). The majority of these genotypes are parents of Vf mapping populations, which are already existing[36] or currently under development.

Table 1

List of Vf Genotypes Used for Qualitative and Quantitative Analysis of Seed Protein Composition

genotype	original source	germplasm category	countrya	mapping populations foundb
LG Cartouche		cultivar	UK
Lynx		cultivar	UK
Vertigo		cultivar	UK	RSBP
Wizard		cultivar	UK
Fanfare		cultivar		RSBP
Icarus	Icarus	inbred line from cultivar	Ecuador	7-way MAGIC; Icarus × Ascot
NV640	Maris Bead	inbred line from cultivar	UK	RSBP
NV643	Albus	inbred line from cultivar	Poland	Albus × BPL10; RSBP; 7-way MAGIC
NV672	Betty	inbred line from cultivar		RSBP
NV866	Disco/2	inbred line from cultivar	France	Hedin/2 × Disco/2; 4WP; RSBP
NV639–2	Hedin	inbred line from cultivar	Germany	RSBP
RV501	Robin Hood	inbred line from cultivar	UK
RV502	Sutton	inbred line from cultivar	UK
RV503	Casata Midwinter	inbred line from informal cultivar	UK	RSBP
RV504	Crimson Flowered-3	inbred line from heirloom cultivar	UK	RSBP
RV505	Diana	inbred line from cultivar	Canada	7-way MAGIC
RV506	Cuscan Super Yellow-1	partial Inbred line from landrace	Peru	RSBP
RV507	Iantos-3	partial inbred line from landrace	Peru	RSBP
RV508	Mustard Yellow	partial inbred line from landrace	Peru	RSBP
RV509	Sakha4	inbred line from cultivar	Egypt	RSBP
RV510	Nubaria3	inbred line from cultivar	Egypt	RSBP
RV511	Misr3	inbred line from cultivar	Egypt	RSBP
RV512	Giza716	inbred line from cultivar	Egypt	RSBP
NV735	Mélodie	inbred line from cultivar	France	Melodie×ILB938–2; RSBP
RV319–2		inbred line	UK
NV153	ig12658	inbred line from Landrace	Ethiopia
NV648–1	BPL10	inbred line	Unknown	Albus × BPL10; RSBP
NV734	ILB938–2	inbred line	Colombia	Melodie×ILB938–2; 4WP; 7-way MAGIC
NV657	INRA 29H	inbred line	France	RSBP
L170	ig132238	inbred line	China	4WP
NV651–3	BPL21	inbred line	Unknown	RSBP
NV658–2	CGN07715 cf-3	inbred line	Unknown
L43	ig114476	inbred line	Bangladesh	4WP
RV322	HEL170	inbred line	China	RSBP
NV873–13	F5 from NV644xNV153	recombinant inbred line	Unknown	RSBP

Country of release (for cultivars) or collection (landrace materials).

RSBP: Reading Spring Bean Population (currently under development); 7-way MAGIC: Multiparent advanced generation intercross (under development); 4WP: 4-Way cross population.[36]

Total Protein Extraction

Five to 10 seeds per genotype were dried in an oven at 80 °C for 48 h and ground using a Laboratory Mill 3303 (Perten Instruments, Warrington, UK). The flour was then sieved through a 1 mm diameter sieve to obtain a homogenous sample. Total seed proteins were extracted according to the procedure reported by Mertens et al.[37] with some modifications. Briefly, we used 0.1 M phosphate buffer (pH 7.2) containing 5 g L–1 of potassium sulfate with a sample/buffer ratio of 1:10 (w/v). Samples were vortexed briefly and stirred for 30 min at 300 rpm followed by centrifugation at 20,000 × g for 30 min at room temperature. The supernatant was then transferred to a new tube and stored at −20 °C until further analysis. The protein concentration in protein extracts was measured using the Bradford method[38] with a SpectraMax i3x microplate reader (Molecular Devices, UK).

Protein Fractionation

The total seed protein extracts were fractionated by sequential extraction in aqueous and salt solutions to obtain fractions enriched for water-soluble and salt-soluble proteins (for details, see Figure S1). The globulin precipitation step was conducted according to the procedure reported by Krishnan et al.[39] A total of five protein fractions (hereafter F1–5) were obtained: water soluble (F1), globulin-depleted water soluble (F2), salt-soluble (F3), globulin-depleted salt-soluble (F4), and globulin-enriched fraction (F5). These fractions were then analyzed by SE-HPLC and SDS-PAGE.

Protein and Sulfur Content Analysis

Nitrogen and sulfur contents (%) were determined using an isotope ratio mass spectrometer (DELTA V IRMS, Thermo Fisher, UK). The analysis was carried out in duplicate using oven-dried flours of ∼ 1 mg. Nitrogen content data were then converted to protein content as: protein (%) = %N × 5.4.[40]

1D SDS-PAGE Analysis

One-dimensional SDS-PAGE analysis of the total protein extract (∼ 15 μg per well) was performed using NuPAGE 10% Bis–Tris precast gels. Before gel loading, the samples were mixed with the NuPAGE LDS sample buffer and a sample reducing agent following manufacturer’s instructions. Gels were run in the NuPAGE MES SDS buffer in an XCell SureLock Mini-Cell at a constant current of 70 mA and a maximum voltage of 200 V for 1 h. Before staining, the gels were fixed with 12% trichloroacetic acid for 15 min and washed twice with 250 mL of deionized water for another 15 min on a rocker. Gels were then stained with 50 mL of a PageBlue protein-staining solution for 2 h followed by destaining overnight with deionized water.

Identification of Major Seed Protein Subunits

In-Gel Protein Digestion

Individual protein bands were carefully excised from gel lanes of the selected genotypes (LG Cartouche, NV657, and NV734) and were destained in 0.6 mL tubes with 400 μL of 50% acetonitrile (MeCN) and 50% 10 mM triethylammonium bicarbonate (TEAB) overnight. Gel pieces were then reduced with 10 mM dithiothreitol (DTT) in 10 mM TEAB for 30 min at 50 °C followed by alkylation with 50 mM iodoacetamide in 10 mM TEAB for 30 min in the dark. After washing three times with 400 μL of 10 mM TEAB and once with MeCN, the dehydrated gel samples were resuspended in 10 μL of 10 mM TEAB containing 200 ng of porcine trypsin and incubated at 25 °C overnight. The gel digests were placed on dry ice for 5 min, then allowed to thaw, and 30 μL of 10% MeCN/5% formic acid was added. After 15 min of sonication, peptide extracts were transferred to 250 μL PCR tubes. This step was repeated twice, and the resultant extract was pooled and dried in a centrifugal vacuum concentrator.

Mass Spectrometry Analysis

The dried peptides were resuspended in 20 μL of LC–MS buffer A (0.1% formic acid in water), and 10 μL of the sample was injected into an Ace C18 column (150 × 2.1 mm, 5 μM particle size with 300 Å pore size) and analyzed by LC–MS using a Thermo Scientific LTQ-Orbitrap XL interfaced with an Accela HPLC instrument. Buffer B was 0.1% formic acid in MeCN. The gradient was as follows: 0–2 min; 5% B, 20 min; 60% B, 20.1–23 min; 80% B, 23.1–30 min; 5% B. The column oven was maintained at 30 °C, and at 15 °C for the autosampler. The first 2 min and the last 6 min of each run were excluded from the analysis. A data-dependent acquisition (DDA) strategy was employed. In brief, ions were measured using the Orbitrap at 30,000 resolution, scanning from 400–2000 m/z. Three ions from each MS1 scan that were most abundant and multiply charged were chosen for MS2. MS2 was performed using collision-induced disassociation (CID) in the ion trap and scanned out at a unit resolution. The acquired data were analyzed using an in-house version of MASCOT search engine (Matrix Science, UK) via Mascot Daemon with file conversion performed using ProteoWizard. The acquired MS spectra were searched against the NCBI nonredundant protein (https://www.ncbi.nlm.nih.gov), common Repository of Adventitious Proteins (ftp://ftp.thegpm.org/fasta/cRAP), and other contaminant databases. The search parameters were set as follows: type of search = MS/MS ion search, enzyme = trypsin, variable modifications = acetyl (protein N-term), carbamidomethyl (C), Gln- > pyro-Glu (N-term Q), oxidation (M), mass values = monoisotopic, protein mass = unrestricted, peptide mass tolerance = ±10 ppm, fragment mass tolerance: ±1 Da, max missed cleavages = 2, and instrument type: ESI-TRAP.

Protein Composition Analysis by SE-HPLC

Size-exclusion HPLC analysis was conducted with the Waters Alliance 2695 Separations Module using a Phenomenex BioSep-SEC-S2000 column with silica resin (300 × 7.8 mm, 5 μm particle size and 145 Å pore size). The same extraction buffer (0.1 M phosphate buffer containing 5 g L–1 of potassium sulfate, pH = 7.2) was used as a mobile phase with a flow rate of 0.5 mL min –1. The injection volume of protein sample was 20 μL, and detection was at 210 nm using the Waters 2996 photodiode array (PDA) detector. Two biological replicates were analyzed for each genotype, and the raw chromatogram data were exported for peak integration in Origin software (OriginLab Corporation, Northampton, MA, USA).

Results and Discussion

A Comprehensive Survey of Vf Seed Proteins

In order to capture the most common seed protein variants, we first conducted a preliminary 1D SDS-PAGE screening of 35 diverse genotypes for their seed protein profiles (Figure S2). From this analysis, we identified three genotypes—LG Cartouche, NV657, and NV734—with distinct protein profiles (Figure ) and used them for protein band identification. Forty-six bands, with apparent molecular weights (MW) ranging from less than 10 to ∼ 145 kDa on reducing 1D SDS-PAGE gels, were detected collectively from these three genotypes. Thirty-five of these bands were excised from the gel and subjected to mass spectrometry analysis, with 25 of them reporting significant (p ≤ 0.05) matches with proteins in the database, mainly from Vf and related legumes (Table ). Failure to identify the remaining 10 bands can be explained in term of their relatively lower abundance, which made it technically challenging to elute enough protein.

Figure 1

SDS-PAGE profile of three Vf genotypes with distinct seed protein profiles which were used for seed protein identification by mass spectrometry analysis.

Table 2

Major Proteins Identified by Mass Spectrometry Analysis of Protein Bands Excised from Reducing SDS-PAGE Gels of Vf Seed Proteins and their Significant (p ≤ 0.05) Matches (from Vf and Other Legume Species) in the Database

SDS-PAGE banda	band apparent MW (kDa)	accession	score	num. of significant sequences	emPAI	description	species
1	106.9	gi|126405	565	15	1.08	seed linoleate 9S-lipoxygenase-3	Pisum sativum
		gi|164512572	128	2	0.18	convicilin	Vf
2	96.3	gi|126405	508	15	1.18	seed linoleate 9S-lipoxygenase-3	Pisum sativum
		gi|164512572	178	4	0.39	convicilin	Vf
3	88.8	gi|164512572	120	2	0.18	convicilin	Vf
		gi|187766747	99	1	0.26	Gly m Bd 28 K allergen	Glycine max
4	83	gi|164512572	165	6	0.68	convicilin	Vf
		gi|22053	154	9	1.34	vicilin: precursor	Vf
5	75.2	gi|357480003	391	8	0.81	heat shock 70 kDa protein	Medicago truncatula
		gi|126162	94	4	0.74	legumin type B	Vf
6	73.1	gi|562006	364	12	1.26	PsHSP71.2	Pisum sativum
		gi|164512572	123	4	0.40	convicilin	Vf
7	64.7	gi|164512572	1145	25	6.89	convicilin	Vf
		gi|126164	101	3	0.30	legumin type B: precursor	Vf
8	54.1	gi|164512572	1074	21	4.67	convicilin	Vf
		gi|403336	312	7	0.68	legumin-related high-molecular weight polypeptide	Vf
		gi|3122060	123	6	0.78	elongation factor 1-alpha	Vf
9	50.0	gi|137584	1344	22	6.28	vicilin: precursor	Vf
		gi|403336	589	11	1.25	legumin-related high-molecular weight polypeptide	Vf
10	48.2	gi|137584	1374	22	6.28	vicilin: precursor	Vf
		gi|403336	342	7	0.68	legumin-related high-molecular weight polypeptide	Vf
		gi|12580894	176	6	0.69	putative sucrose-binding protein	Vf
11	45.4	gi|12580894	1018	18	4.40	putative sucrose-binding protein	Vf
		gi|22008	226	9	1.16	legumin A2 primary translation product	Vf
		gi|126166	178	8	1.84	legumin type B	Vf
12	43.5	gi|2578438	98	3	0.26	legumin (minor small)	Pisum sativum
		gi|403336	90	3	0.26	legumin-related high-molecular weight polypeptide	Vf
13	40.2	gi|22008	662	14	2.51	legumin A2 primary translation product	Vf
		gi|164512572	208	7	0.78	convicilin	Vf
		gi|259474	312	6	1.42	legumin propolypeptide alpha chain	Vf
14	38.4	gi|22008	875	14	2.61	legumin A2 primary translation product	Vf
		gi|126166	628	12	3.78	legumin type B	Vf
		gi|22053	392	11	1.75	vicilin: Precursor	Vf
15	37.6	gi|542002	823	9	2.67	legumin type B alpha chain; precursor	Vf
		gi|137584	506	16	3.24	vicilin: precursor	Vf
		gi|22008	312	10	1.31	legumin A2 primary translation product	Vf
16	36.2	gi|542002	926	8	2.28	legumin type B alpha chain: precursor	Vf
		gi|137584	747	19	4.83	vicilin: precursor	Vf
		gi|22008	253	7	0.83	legumin A2 primary translation product	Vf
17	31.5	gi|137584	277	11	1.71	vicilin: precursor	Vf
		gi|137582	203	4	0.44	vicilin: precursor	Vf
18	30.4	gi|137584	300	11	1.73	vicilin: precursor	Vf
		gi|137582	157	4	0.45	vicilin: precursor	Vf
19	26.0	gi|22008	76	2	0.18	legumin A2 primary translation product	Vf
		gi|29539109	54	3	0.35	allergen len c	Lens culinaris
20	24.0	gi|12580894	53	1	0.09	putative sucrose-binding protein	Vf
21	22.3	gi|259475	399	5		legumin propolypeptide beta chain	Vf
		gi|403336	369	5		legumin-related high-molecular weight polypeptide	Vf
22	13.7	gi|51704211	97	2	0.98	albumin-1 E	Pisum sativum
23	12.4	gi|51704211	72	1	0.27	albumin-1 E	Pisum sativum
		gi|27466894	70	2	0.68	thioredoxin h	Pisum sativum
		gi|763805274	50	1	0.25	hypothetical protein	Gossypium raimondii
24	10.2	gi|51704209	60	1	0.29	albumin-1 C	Pisum sativum
25	9.5	gi|205277584	56	2	1.15	defensin-like protein	Vf
		gi|205277582	55	2	1.19	defensin-like protein	Vf

Band numbers in the first column refer to the band numbers shown in Figure .

Figure 2

SE-HPLC chromatogram of Vf seed protein extract from NV639–2 which is overlaid with the SDS-PAGE profile of protein fractions collected at a 1 min interval across the analysis time. Observable peaks are numbered from 1–21 and labels on the left refer to some of the major protein subunits identified in this study.

Nearly all analyzed bands contained more than one type of protein, and a total of 106 proteins were identified (for detailed list, see Table S1). As expected, the most abundant proteins were globulins, with polypeptides belonging to legumin, vicilin, and convicilin identified in 13, 8, and 4 of the 25 bands, respectively (Table ). This wide molecular mass distribution of legumin and vicilin subunits was previously reported[21] using antibodies specific to these proteins. However, in the case of convicilin, for which a single discrete band near 68 kDa has been so far reported in the literature,[1,17,21,28] we have identified multiple bands, including a major band at ∼ 54 kDa (Figure , Table ). Although this is a new observation for Vf, it is not surprising considering that multiple convicilin subunits with MW ranging from 52–99 kDa have been reported in the related species Medicago truncatula.[31,41] To further investigate whether the two major convicilin bands identified (7, 8 in Figure ) represent the convicilin A and B genes reported in the past,[22] we compared the protein sequences derived from these convicilin genes with the MS peptide sequences from band 7 and 8 of LG Cartouche and ILB 938–2. While convicilin B-specific peptides were found in both bands 7 and 8 in both genotypes, three peptides unique to convicilin A were found only in band 8 of LG Cartouche (Table S2). Furthermore, peptide sequences from band 7 of LG Cartouche contained a 37 AA long peptide which aligned to a region with significant polymorphism between convicilin A and B. Interestingly, this unique peptide had nine and five mismatches with A and B genes, respectively, but had 100% similarity with a convicilin accession (CAP06324.1) from Lathyrus ochrus. Taken together, these results indicate that convicilin structural diversity in Vf is greater than previously thought, comprising of at least two B-type isoforms, as well as A and other unnamed convicilin polypeptides which appear to be expressed in a genotype-dependent manner.

Mass spectrometry analysis also identified several less abundant but nonetheless distinct and well-conserved protein bands. These include two distinct lipoxygenase bands (∼ 96 and 106 kDa), a heat shock protein (∼ 73–75 kDa, depending on the genotype), a sucrose-binding protein (∼ 45 kDa), albumins (∼ 10.2, 12.4 and 13.7 kDa), and defensins (Table , Figure ). From a nutritional quality point of view, lipoxygenase is considered antinutritional due to its role in lipid oxidation, which also leads to undesirable flavors during food processing.[42] The studied genotypes show noticeable variation in the intensity of lipoxygenase bands (Figure S2), but establishing the significance of this variation requires further scrutiny. In other legumes such as soybean[43] and pea,[44] efforts to develop genotypes lacking the major seed protein lipoxygenase have been successful.

Protein Subunit Diversity Among Vf Genotypes

In total, we identified 15 protein bands polymorphic among the Vf genotypes, with variation being concentrated in less abundant proteins with MW of more than 70 kDa or less than 20 kDa (Figure S2). The most interesting protein variants were found in the α subunits of legumin, represented in the majority of Vf genotypes by a single legumin band of about 38 kDa and by rare legumin α subunits of about 36 and 40 kDa in LG Cartouche and NV657, respectively (Figure ). MS analysis showed that the higher MW legumin α subunit in NV657 is an A-type legumin while that of a lower mass in LG Cartouche is a B-type legumin α subunit (Table ). Further evidence that these genotypes contain novel legumin subunits comes from the observation that unreduced proteins of these genotypes have two distinct major bands of αβ polypeptides (data not shown). These natural variants in subunit composition can be exploited to address questions on the genetic architecture of seed protein composition and the impact of discrete protein subunit variants on the nutritional and processing quality of the overall seed protein.

Although the majority of the analyzed bands contained one predominant protein type, the existence of some bands where there is an overlap between major bands of different protein classes underpins the need for an alternative method to the conventional SDS-PAGE-based densitometric approach for quantifying protein composition. Nonetheless, this expanded and refined list of identified seed proteins can be utilized as a reference for qualitative SDS-PAGE-based screening for protein subunit variants of interest in breeding and research materials like mutant or mapping populations.

SE-HPLC Analysis of Seed Proteins

Total Seed Protein Extract

The total seed protein extract from the NV639–2 inbred line was separated using a Phenomenex BioSep-SEC-S2000 column, producing chromatographic peaks between 10 and 28 min of the analysis time (Figure ), and four major peaks (1, 2, 4, and 18) accounted for more than 70% of the total chromatogram peak area. To confirm the identity of proteins associated with these peaks, SE-HPLC peak fractions were collected at 1 min intervals and separated by 1D SDS-PAGE. By comparing these gels with the annotated SDS-PAGE (on the basis of MS analysis), it was determined that peaks 2 and 4 were legumin and vicilin/convicilin aggregates with retention times of 12.4 and 14.0 min, respectively (Figure ). Proteins with smaller molecular weights were eluted in the expected order, suggesting that the selected column was suitable for the separation of Vf proteins. However, despite having strong signals at 214, 254, and 280 nm, no detectable proteins were found in peak 1 and all other peaks eluted after ∼ 21 min (Figure ). We therefore hypothesized that peak 1 corresponds to protein–phenol complexes that could not be detected by SDS-PAGE. Sȩczyk et al.[45] found that some phenolic compounds preferentially interact with globulins, leading to changes in their SE-HPLC and SDS-PAGE profiles. Regarding peak 18, Defaix et al.,[46] who used the same type of column used in our study, suggested that the strong signal near the end of the analysis was due to phenolic compounds. To further investigate this hypothesis, SE-HPLC profiles of protein extracts of dehulled and whole seeds were compared; since Vf seed coats contain higher phenols, we would expect the proportion of peak 1 and 18 to be substantially reduced in the dehulled sample. Indeed, dehulled protein samples showed a nearly 50% decrease in both peaks (Figure S3). On this basis, only the peak area between 11.5 and 21.5 min was considered for SE-HPLC protein composition analysis.

Fractionated Seed Proteins

To further confirm that the peaks resolved by SE-HPLC belong to the major seed proteins, we separated the protein fractions prepared by sequential extraction (denoted as F1–F5 in Figure ) both by SE-HPLC and 1D SDS-PAGE. Comparison of the separation profiles obtained for F1–F5 protein fractions in the two systems (Figure A,B) showed that the functional proteins like lipoxygenase, heat shock protein, and albumin have relatively higher solubility in water and they were enriched in F1 and F2, with an elution time between 15 and 20 min under the SE-HPLC conditions used in this work. However, since these peaks, unlike globulin peaks, were poorly resolved by SE-HPLC, they are referred collectively as “other proteins” as shown in Figure A. On the other hand, different globulin subclasses were enriched in specific extraction buffers, reflecting their different physicochemical properties. For instance, legumin is soluble in water and to a higher degree in a phosphate buffer (F3), which could be further enriched by the addition of Ca+2 (F5), producing a major HPLC peak with a retention time of 12.4 min. However, the vicilin/convicilin subclass of globulin appears to contain a mixture of subunits with varying properties. As shown in Figure B, protein bands corresponding to subunits of convicilin (∼ 54 kDa) and vicilin (∼ 37 kDa) were extractable in water (F1) and did not precipitate in the presence of Ca+2 (F2). Conversely, other subunits of convicilin (∼ 65 kDa) and vicilin (∼ 50 kDa) were soluble in a phosphate buffer (F3) and precipitated, to a certain degree, with the addition of Ca+2 (F4 and F5). Even though the fractions F1, 4, and 5 have a vicilin/convicilin peak of nearly a similar magnitude, the SDS-PAGE profile of these different fractions showed distinct subsets of vicilin/convicilin (Figure B). It was therefore concluded that convicilin and vicilin polypeptides form heterogeneous subclasses of the globulin type protein with distinct physicochemical properties but eluted as a single peak with a retention time of 14 min under the SE-HPLC conditions used in our study. This observation would explain why O’Kane et al.,[47] who conducted various fractionation and physicochemical characterization of vicilin and convicilin proteins in pea, concluded that convicilin is a α subunit of vicilin.

Figure 3

(A) SE-HPLC chromatogram and (B) SDS-PAGE profile of fractionated proteins of NV639–2 line. Fractions (F1–5) are water-extractable proteins (F1), globulin-removed water-soluble fraction by addition of 10 mM CaCl2 (F2), pellet from F1 extracted with 0.1 mM phosphate buffer at pH = 7.2 (F3), globulin-depleted salt-soluble proteins by addition of 10 mM CaCl2 (F4), and pellet from F4 suspended in 0.1 mM phosphate buffer at pH = 7.2 (F5).

Quantification of Legumin and Vicilin/Convicilin Contents by SE-HPLC

Since one of the major indicators of protein quality is the content of S-AA which in turn is determined by the relative proportions of the major protein classes, the SE-HPLC method was used to quantify legumin and vicilin/convicilin contents in a panel of 35 genetically diverse Vf genotypes. Overall, legumin and vicilin/convicilin accounted for 50 and 27% of the protein extract, respectively. Among the genotypes, legumin accounted for 35 to 62% of the quantified peak area while vicilin/convicilin for 20–34% (Figure ). These results are comparable with the findings of Utsumi, Yokoyama, and Mori[27] who reported ranges of 42 to 47% and 28 to 31% for 11 and 7S globulins in crude protein extracts of six Vf cultivars analyzed by the sucrose density gradient fractionation technique. In another study, the Vf legumin and vicilin content reportedly varied between 40 to 45% and 20 to 25%, respectively.[3] Moreover, according to our study, globulin peaks represent 77% of the total protein peak area, which is very close to the estimated 70 to 80% globulin content in Vf seed proteins reported by other studies.[3,17,48] The present study appears to capture a wider variation in Vf protein composition than previously reported, likely reflecting the fact that the plant materials we used spanned a deliberately broad genetic base.

Figure 4

Bar graph showing the proportions of legumin and vicilin/convicilin in the total seed protein extracts of 35 Vf genotypes. Protein percentages are determined from the relative area of SE-HPLC peaks belonging to each protein class in two biological replicates.

The legumin to vicilin/convicilin ratio (L/V) varied from 1 to 3, which is comparable to the 2.1 to 3.6 range reported previously.[26] Among the genotypes of special interest for their high L/V ratio are two inbred lines, NV153 and NV658–2, which have been previously used as parents in mapping populations (Table ). However, it is important to mention that this ratio is highly sensitive and can be affected by many factors, including genotype, environment, protein extraction method, and quantification techniques. In this work, the reproducibility of the results was measured by comparing five replicates of a single genotype (NV639–2) that were independently extracted and analyzed in different batches. The coefficient of variation between the five replicates was higher in the legumin fraction (6%) compared to vicilin/convicilin (3%). However, biological replicates of each genotype analyzed in the same run were highly correlated (r2 > 0.98) (Figure S4). This indicates the importance of including the batch as a cofactor for statistical analysis.

Finally, we exploited this quantitative data from a wide spectrum of germplasm to examine the possible limits and trade-offs between the two main classes of storage proteins and overall sulfur and protein contents. The Pearson’s correlation analysis showed that the legumin content significantly but negatively correlated with vicilin/convicilin (r = −0.83, p < 0.001) and with “other proteins” (r = −0.87, p < 0.001) (Figure ). On the other hand, the seed sulfur content correlated negatively with the content of vicilin/convicilin, the S-AA-deficient fraction[1] but correlated positively with the total protein content. Interestingly, the total protein content was independent of the proportion of major protein fractions, suggesting that protein composition can be improved without compromising the protein content. Similar independence of the total protein and globulin fractions has been observed in pea,[30] while, in contrast, a highly significant negative correlation between the content of certain 7S fractions and the total seed protein content has been reported in soybeans.[35]

Figure 5

Correlation matrix between proportion of globulin fractions and other seed composition parameters at a significance level of p ≤ 0.05.

In conclusion, this study provides a contemporary survey of the major seed proteins and their subunit composition among genetically diverse Vf germplasm. The MS-based identification of many major protein bands is a timely update linking a greater diversity of seed storage protein sequences to specific protein subunits which can be readily resolved with 1D SDS-PAGE gels; this new information can be used in the screening of germplasm with unique protein profiles, such as naturally occurring or induced mutations related to the reduced content of undesirable or increased content of desirable proteins. Also, we have demonstrated the potential of SE-HPLC as a method to efficiently determine the contents of legumin and vicilin/convicilin in Vf. This work paves the way for further understanding the Vf seed protein composition and the development of cultivars with desired protein quality.