Reaction process and characteristic changes in soybean oil bodies during the formation of volatile flavour compounds in soymilk

Abstract

Oil bodies (OBs) are lipid-storing organelles of soybean seeds and the source of substrates for the enzymatic oxidation reaction of soymilk flavour formation. However, the triacylglycerols (TAGs) in the OBs are surrounded by a membrane composed of phospholipids embedded with integral oleosins to prevent them from being hydrolyzed by lipase. Therefore, the specific reaction process of the OBs in the soymilk flavour formation process becomes complicated and is not well understood. In this study, the optimal substrate, OBs, were extracted by sucrose and washed with water at pH 11.0. The oxidation products (i.e., lipid hydroperoxides, and volatile flavour compounds) of the reaction systems, with OBs as substrate, showed that phospholipase A2 (PLA2) can directly hydrolyze external phospholipids (PLs) of OBs, and pancreatic lipase can hydrolyze the internal TAGs of OBs only after oleosins or PLs are hydrolyzed. Phospholipase D (PLD) cannot hydrolyze PLs of OBs. These findings are of great significance for understanding the mechanism of flavour formation in soymilk and further improving the flavour of soymilk.

Introduction

The lipids in plant seeds are stored mainly in small, discrete, intracellular organelles called the oil bodies (OBs). OBs have triacylglycerols (TAGs) matrix cores, and the cores are surrounded by phospholipid (PL) monolayers embedded with OBs intrinsic proteins (mainly oleosins) (Huang, 1992). The oleosin/PLs OBs coat can shield TAGs from exterior environmental stresses (such as temperature, desiccation, and pH) and maintain the integrity of OBs during dehydration and seed dormancy (Napier et al., 1996; Chen and Ono, 2010; Shimada and Hara-Nishimura, 2010). However, once imbibition occurs and germination is triggered, TAGs in the OBs are mobilized and degraded to supply energy for seed germination and seedling growth. It is understood that many enzymes (i.e., protease, phospholipase, and lipase) bound to the OBs surface initiate lipid mobilization (Matsui et al., 1999; Rudolph et al., 2011). Although lipid OBs and lipid droplets are rather stable, there are different views on the necessity of destroying the integrity of the protein coat and/or the PLs monolayer prior to TAGs matrix hydrolysis during OBs mobilization. Maize OBs were hydrolyzed by PLA2 only after treatment of OBs by trypsin (Tzen and Huang, 1992). Partial trypsin digestion of OB oleosins of cucumber seedlings led to oxidative degradation of TAGs (Matsui et al., 1999; Rudolph et al., 2011). A lipid body-associated thiol-protease in sunflower cotyledons was found to gradually degrade oleosins during seedling growth (Reza and Chander, 2002). In rapeseed seedlings, the partial degradation of oleosin isoforms was also observed prior to the complete degradation of lipid OBs (Murphy, 2001). Furthermore, data on the degradation of the PLs monolayer by a patatin-like phospholipase (PLase) have been reported only for cucumber and sunflower (Noll et al., 2000; Gupta and Bhatla, 2007). In contrast to the OBs mentioned above, almond OB mobilization in vivo was achieved by lipase without prior action of a protease (Beisson et al., 2001). Early studies indicated that prior limited degradation of oleosins is necessary for subsequent lipolytic action in most oilseeds. To date, most of the research regarding the mobilization of soybean OBs is speculated from studies on related plant seed OBs.

Soymilk is a traditional processed food derived from soybean, and it has a number of nutritional and health benefits. However, its characteristic beany flavour is off-putting to Western consumers, limiting its application and promotion (Morr and Ha, 1991; Keast and Lau, 2006). Studies on the offflavours of soybean products have shown that lipoxygenase (LOX), an enzyme naturally present in soybeans, mediates the conversion of polyunsaturated fatty acids to lipid hydroperoxides (LOOHs), resulting in degradation products that are responsible for the volatile flavours generated (Kobayashi et al., 1995). The reaction process of volatile flavour formation in soymilk is shown in Fig. 1. However, the typical off-flavours are not present in the dried intact soybean seeds. Only when the integrity of the soybean is destroyed, such as in processing soyflour or soymilk, large numbers of volatile flavour compounds are generated. The reason may be that in the dried intact soybean seeds, the soybean endogenous enzymes are in a dormant state and cannot hydrolyze lipids to produce free fatty acids (FFAs) (Mellor et al., 2010). In addition, the steric barrier formed by OBs oleosins in soybean cells prevents access of the LOX to its lipid substrates (Leprince et al., 1998). During soymilk grinding, a series of enzymatic oxidation reactions related to flavour occur in the OBs released from soybean seeds cells. The oleosin isoforms isolated from soybean seeds showed significant PLA2 activity with the release of a small amount of FFAs, which indicated they may be involved in the formation of off-flavours in soymeal (Kumari et al., 2016). Phospholipase D (PLD), an endogenous phospholipase in soybeans, converts phosphatidylcholine (PC) to phosphatidic acid (PA). The PLD pathway is of great interest, in connection with the expression of LOX activity (List et al., 1992). However, whether PLD facilitates the off-flavours of soybean products has not been confirmed. Previous studies on the formation mechanism of soymilk volatile flavour compounds were mainly based on free polyunsaturated fatty acids or esterified polyunsaturated fatty acids as the substrates and LOX as the catalyst (Zhuang et al., 1991; Iassonova et al., 2009). However, the study of lipid storage organelles, i.e., soybean OBs, as enzymatic substrates has not yet been reported.

Fig. 1.

Reaction process diagram of volatile flavour formation in soymilk. FFAs: free fatty acids; LOOHs:lipid hydroperoxides.

The studies mentioned above showed that the shell of OBs oleosins not only maintains the stability of the OBs but also plays an important role in protecting the components of the OBs from enzymatic hydrolysis. Enzymatic hydrolysis of OBs in vitro would produce OBs with different properties, which would affect the flavour characteristics of soymilk and its related products. However, no studies have systematically examined the behaviors of OBs components and oxidation products of the extracted soybean OBs upon treatment with lipase (EC.3.1.1.3), or phospholipase, or trypsin (EC.3.4.21.4) and LOX. Moreover, it is not clear whether oleosins hydrolysis is a necessary prerequisite for OBs oxidation to occur. In this study, a model reaction system with soybean OBs emulsion as a substrate was established. The enzymatic hydrolysates (proteins and lipid components) and oxidation products (LOOHs, volatile flavour compounds) of the OBs after enzymatic treatment were measured. Furthermore, OBs extracted under alkaline conditions from raw soymilk were dispersed into deionized water to prepare an OBs suspension and to repeat the experiment above. The results from soymilk were compared.

Materials and Methods

Materials    Soybeans (Heinong 64) were harvested in 2017 and obtained from the Nenjiang fumin Agricultural and Sideline Products Co., Ltd (Heihe, China). Sucrose, with analytical purity, was purchased from Sinopharm (Beijing, China). Phospholipase A2 (600–2400 units/mg), phospholipase D (≥ 60 units/mg), pancreatic lipase (≥ 20 000 units/g), trypsin (∼1 500 units/mg), analytically pure acrylamide, N,N'-methylene diacrylamide, bromophenol blue, mercaptoethanol, tricarboxylic aminomethane (Tris), glycine, tetramethylenediamine (TEMED), and Coomassie Blue G250 were purchased from Sigma-Aldrich (Shanghai, China). The protein standards (MW range of 10–250 kDa) were purchased from Bio-Rad (Shanghai, China).

Preparation of raw soymilk    An aliquot of 100 g of soybeans was soaked in 500 mL deionized water and refrigerated at 4 °C for 18 h. The soaked soybeans were ground in deionized water (precooled in 4 °C, seed/deionized water, 1/9, w/w) using a MJ-60BE01B Waring blender (Midea, Feshan, China) for 2 min. The raw soymilk was harvested by filtering through four layers of gauze to remove all soybean residues. Some of the soymilk was immediately taken for the measurement of volatile flavour compounds according to the method described in the Detection of volatile flavour compounds section, and the remaining soymilk was stored in an ice water bath for OBs extraction within 30 min.

Extraction of soybean OBs    Soybean OBs were extracted according to the method described by Zhao et al. (2016). The prepared soymilk was divided equally into six parts, and the pH of each part was adjusted to 6.0, 7.0, 8.0, 9.0, 10.0, or 11.0 with 0.1 and 1 M NaOH or HCl solution. Sucrose, 25% of the mass of soymilk, was added and mixed well at 4 °C. The mixtures were centrifuged (40 000 g, for 30 min at 4 °C) using a CR21N high-speed refrigerated centrifuge (Hitachi, Tokyo, Japan). The upper floating fractions were collected and evenly dispersed into deionized water (OBs/deionized water, 1/10, w/w) at 4 °C; the procedure was repeated twice. The floating fractions were collected and named pH 6.0-, 7.0-, 8.0-, 9.0-, 10.0-, and 11.0-OBs, respectively. These OBs were diluted 10 times with deionized water and their protein concentrations were determined by the micro-Kjeldahl method. After proper adjustment, equal volumes (0.5 mL) of the OBs emulsion and SDS-PAGE sample buffer were mixed to achieve a protein concentration of 2 mg/mL.

Extraction of LOX    Soybean LOX was extracted according to the methods described by Axelrod et al. (1981) and Fukushige et al. (2005), with some modifications. Dry soybean seeds (200 g) were milled to a fine flour using an XA-1 high-speed grinder (Yinhe, Hebi, China), sieved (60-mesh), and mixed with deionized water (soybean flour/deionized water, 1:6, w/w). The resultant slurry was adjusted to pH 4.5 with HCl (0.5 M), stirred for 1 h at 4 °C, and filtered through four layers of gauze. The resultant filtrate was centrifuged at 10 000 g at 4 °C for 30 min. The supernatant was adjusted to pH 6.8 with 2 M NaOH. Ammonium sulphate powder was added slowly to a concentration of 40%, and the mixture was stirred for 1 h at 4 °C before centrifugation (10 000 g, for 30 min, at 4 °C). The supernatant was again adjusted to pH 6.8 with 2 M NaOH. Ammonium sulphate was added to a final concentration of 60% and the mixture was stirred for 1 h at 4 °C before centrifugation. The supernatant was removed, and the precipitate was dissolved in sodium phosphate buffer (100 mL, 0.02 M, pH 6.8). The enzyme solution was subjected to dialysis using a MD44 dialysis bag (Solarbio, Shanghai, China) prior to vacuum freeze-drying. The activity of LOX was determined via the method described by Axelrod et al. (1981).

Enzymatic hydrolysis of OBs

Trypsin hydrolysis    The pH 11.0-OBs were dispersed into deionized water (OBs/deionized water, 1/10, w/w), and the pH was adjusted to pH 7.0. To seven 15 mL glass vials, 4.50 mL of the OBs emulsion and 0.20 mL of trypsin solution were added. The mixture was incubated in a 40 °C water bath with magnetic stirring. At 30 s, 1 min, 2 min, 5 min, 10 min, 15 min, and 30 min, a vial was removed from the 40 °C water bath and immediately heated in a boiling water bath for 5 min to inactivate the trypsin. After proper adjustment according to the above measured protein concentration, equal volumes (0.5 mL) of the OBs suspension and tricine-SDS-PAGE sample buffer were mixed to a concentration of 2 mg/mL.

Enzymatic hydrolysis and thin-layer chromatography (TLC) analysis    The pH 11.0-OBs were operated as in the Trypsin hydrolysis section. To five 15 mL glass vials, 4.50 mL OBs emulsion were added. Subsequently, 0.10 mL lipase solution, 0.20 mL PLA2 (EC.3.1.1.4) solution, 0.2 mL PLD (EC.3.1.4.4) solution, 0.20 mL trypsin solution and 0.20 mL PLA2 solution, and 0.20 mL trypsin solution and 0.20 mL PLD solution were added to the vials. An appropriate amount of deionized water was added, bringing the final volume of the system to 6 mL. The vials were incubated in a 40 °C water bath with magnetic stirring for 30 min. Then, 20 mL chloroform/methanol (2/1, v/v) were added to extract lipids and the vials were placed in a 50 °C water bath for 2 h. The lipids were obtained by rotary evaporation of organic solvent and dissolved in chloroform/methanol (2/1, v/v) to a sample concentration of 2 mg/mL. To the TLC plate, 10 µL samples were spotted. The TLC plate (silica gel G, Qingdao Haiyang Chemical Co., Ltd) was developed in chloroform/methanol/acetic acid/water 85:15:10:3 (v/v/v/v) to separate the lipids. Spots were visualized by staining with iodine. The PL components of the OBs reaction system were identified by comparison with PLs standards. The concentration of the enzyme preparation solution used in the OBs enzymatic hydrolysis is shown in Table 1.

Table 1. Components of the reaction system with oil bodies as substrate

Model number	Oil bodies (mL)	Trypsin (mL)	Phospholipase D(mL)	Phospholipase A2(mL)	Lipoxygenase (mL)	Lipase (mL)	Deionised water(mL)
1	4.50	–	–	–	1.00	–	0.50
2	4.50	0.20	–	–	1.00	–	0.30
3	4.50	-	–	–	1.00	0.10	0.40
4	4.50	0.20	–	–	1.00	0.10	0.20
5	4.50	–	0.20	–	1.00	–	0.30
6	4.50	0.20	0.20	–	1.00	–	0.10
7	4.50	–	0.20	–	1.00	0.10	0.20
8	4.50	–	–	0.20	1.00	–	0.30
9	4.50	0.20	–	0.20	1.00	–	0.10
10	4.50	–	–	0.20	1.00	0.10	0.20

“–”:Not added

Enzymes chosen above were all dissolved in deionized water, the concentrations of trypsin, phospholipase D, phospholipase A2, lipoxygenase, and lipase solutions were 2.0 mg/mL, 10.0 mg/mL, 0.5 mg/mL, 6.0 mg/mL, and 7.0 mg/mL. The volatile flavours contained in all enzyme solutions were negligible.

Particle size observation of OBs    After trypsin and/or PLA2 hydrolysis, the OBs emulsions were diluted 10 times and 100 times with deionized water, respectively, for microscopic observation and particle size scanning. One drop of the diluted OBs emulsion was transferred onto the glass slide and covered with a coverslip. The glass slide was observed at a magnification of 400× using a CX31 optical microscope (Olympus Corporation, Tokyo, Japan). The hydrodynamic diameter (expressed by average diameter), size distribution by volume, and ζ-potential were analyzed via dynamic light scattering measurements using a Zetasizer Nano ZS (Malvern Instruments, Malvern, UK). A milk module was employed. The scanning time was 60 s, and the temperature was 25 °C.

OBs reaction system    The components of the reaction system are given in Table 1. To a 15 mL HS vial, 4.5 mL of OBs emulsion (pH 11.0-OBs/deionized water, 1/20, w/w, pH 7.0) were added as the substrate. The solutions of a certain enzyme or multiple enzymes were added (as shown in Table 1) and mixed well. The total volume of each system was brought to 6.0 mL with deionized water as appropriate. The vial was then fitted with a septum cap. Samples were immediately stirred (magnetic stir bar, 120 rpm) in a 40 °C water bath for 30 min.

Separate OBs reaction system runs were made for the determination of LOOHs and volatile flavour compounds. Prior to headspace solid-phase microextraction (HS-SPME) analysis, the 2-methyl-3-heptanone (1 µL of 0.5025 mg/mL in anhydrous methanol) internal standard was added via syringe. For the analysis of flavour chemicals, a SPME fiber was inserted into the headspace of the vial at the commencement of the main incubation period (t = 0). At the end of the incubation period (t = 30 min), the fiber was immediately desorbed into a GC-MS system, according to the procedure described below. For the analysis of LOOHs, 0.3 mL of the OBs reaction emulsion was removed via syringe at t = 30 min for analysis, according to the procedure described below.

Detection of lipid hydroperoxides    LOOHs were measured using the method developed by DeLong et al., (2002). Briefly, OBs reaction emulsion (0.3 mL) and deionized water solvent (0.2 mL) were added to freshly prepared FOX reagent (4.0 mL) and mixed well. The reaction was incubated in the dark for 45 min and centrifuged at 3 500 rpm for 10 min. The absorbance of each sample at 560 nm was measured on a UV-2450 ultraviolet spectrophotometer (Shimadzu, Kyoto, Japan). The concentration of LOOHs was calculated from the H₂O₂ standard curve (R² = 0.9993). The concentration of LOOHs is reported as H₂O₂ µM equivalents.

Detection of volatile flavour compounds    The volatile flavour compounds were measured according to the method described by Jung and Ebeler (2013). The CAR–PDMS SPME fiber (85 µm) was exposed to the HS above samples (6 mL) containing 2-methyl-3-heptanone internal standard (1 µL of 0.5025 mg/mL) in a 15 mL HS vial for 30 min at 40 °C with stirring (magnetic stir bar, 120 rpm). SPME fibers were desorbed into a splitless injector at 260 °C for 7 min on a combined 3800/1200L GC-MS system (Varian Inc., Palo Alto, USA), equipped with a DB-WAX column (30 m × 0.25 mm i.d., 0.25 µm df (df: film thickness dimension)). The GC oven temperature program was as follows: 40 °C (3 min hold), increase to 100 °C at 6 °C/min, increase to 230 °C at 10 °C/min (7 min hold). Scanned data acquisitions were made in positive ionization (EI) mode over the mass range of 33–350 m/z. The relative quantitation of the volatile flavour compounds was conducted via the method reported by Yuan and Chang, (2007). The standard curve was established by plotting response factors against the amount of standard volatile compounds, with 2% soybean protein as a matrix.

SDS-PAGE and tricine-SDS-PAGE    Samples used for SDS-PAGE/tricine-SDS-PAGE were prepared as described in the Extraction of soybean OBs and Trypsin hydrolysis sections. To the samples, 10 µL of β-mercaptoethanol were added, and the mixture was heated in a boiling water bath for 3 min and centrifuged at 25 000 g for 10 min. Ten microliters of each sample supernatant were loaded into a sample well by microsyringe.

SDS-PAGE was performed according to the method described by Chen and Ono (2010) with 5% stacking gel and 12.5% separating gel. SDS–PAGE was performed at 15 mA until completion. After electrophoresis, the gel was stained with Coomassie Brilliant Blue G-250.

Tricine-SDS-PAGE was carried out according to the method by Schägger (2006), using 4% stacking gel and 16% separating gel. The samples were separated at a constant voltage of 30 V until all samples entered the stacking gel, followed by a constant voltage of 100 V until completion. After electrophoresis, the gel was stained with Coomassie Brilliant Blue G-250.

Statistical analysis    All experiments were performed with two replications. The results are presented as the average values. The correlation analysis was performed between LOOHs and volatile flavour compounds in the OBs reaction system and soymilk by SPSS version 19.0 (SPSS Inc., 2014).

Results and Discussion

Effect of extraction pH on proteins of OBs    Previous studies have shown that there are two types of integral proteins in soybean OBs: oleosin and caleosin. Oleosin is the main protein of OBs integral proteins and has three isoforms with molecular masses of 16, 18, and 24 kDa. Caleosin is the minor integral protein and has two isomers with molecular masses of 27 and 30 kDa (Lin et al., 2002). These proteins are strongly bound to OBs through embedment of their hydrophobic central domains into the TAGs matrix, and they are called intrinsic proteins. The proteins associated with the soybean OBs have been detected by SDS-PAGE. At least ten bands were resolved by SDS-PAGE (Fig. 2), which showed that in addition to intrinsic proteins, there were other soybean proteins present in pH 7.0-OBs. The proteins (soybean storage proteins and bioactive proteins) adsorbed on the surface of OBs by noncovalent and covalent interactions are called extrinsic proteins. They mainly include β-conglycinin (α, α′, and β), glycinin (A3 and acidic peptides), γ-conglycinin, and LOX.

Fig. 2.

SDS–PAGE profiles of OBs by extracted thrice at different pHs. Lane 1, marker; Lanes 2-7, pH 6.0-, 7.0-, 8.0-, 9.0-, 10.0-, and 11.0-OBs. OBs: oil bodies.

Fig. 2 shows that the protein bands resolved by SDS-PAGE gradually become lighter with increasing pH. Compared with intrinsic proteins, the reduction of extrinsic proteins was more severe. When the extraction pH was <8, some extrinsic proteins remained on the OBs surface, in addition to OBs integral proteins. When the extraction pH was >8, LOX, β-conglycinin, γ-conglycinin, and glycinin were almost removed from the OBs surface, and most of the 16 kDa oleosin was released from the OBs. The reason might be that the positively charged amino acid residues of the N-terminal and C-terminal domains were deprotonated by alkaline pH, which decreased the number of salt bridges between oleosins and PLs. As a result, a fraction of oleosins could no longer anchor onto the OBs surface. For the pH 11.0 extraction, most of the 30 kDa caleosin was released from the OBs, and only the OBs integral oleosins remained on the surface. It was reported that half of ά and α subunits of β-conglycinin were disulfide (SS) linked with 30 kDa caleosin, which might be the reason that some 30 kDa caleosin was released in integral OBs (Wadahama et al., 2012). These results reveal that high alkaline pH extraction removed all extrinsic proteins, as well as some intrinsic proteins. LOX is the key enzyme that catalyzes the oxidation of lipids and generates a large amount of volatile flavour compounds. Therefore, the elimination of LOX is necessary to improve flavour quality of soymilk. We anticipated LOX would be removed after three replicate extractions.

TLC analysis OBs phospholipids    Early studies showed that neutral lipids (mainly TAGs) and PLs in OBs contribute 94–98% and 0.6–2% (w/w) to the total weight of dry plant seed OBs, respectively (Tzen et al., 1993). TLC analysis of lipid components of the soybean OBs confirmed the above results (Fig. 3). Fig. 3 shows that the components of PLs observed in soybean OBs were phosphatidylethanolamine (PE), PC, and phosphatidylinositol (PI), and the components of PLs in OBs were hydrolyzed by phospholipase. PLA2 was found to be active on the PLs of OBs, as inferred from the disappearance of PE and PI, as well as the production of lysophospholipids (LP) and phosphatidyl serines (PS). The addition of trypsin can further facilitate the hydrolysis of PLA2. It is likely that PLA2 readily hydrolyzed the hydrophilic phosphoric acid groups on the surface of OBs but could not access the hydrophobic phosphate acyl groups that extend into the center of the OBs. By contrast, the OBs were not susceptible to PLD, since there was no change in PLs components, even in the presence of trypsin. The reason may be that PLD cannot specifically act on the ester bonds of PLs embedded in OBs and evenly dispersed in an aqueous solution. The selected lipase does not have the activity of phospholipase. PLA2 hydrolyzed stereoselective PLs in the sn-2 position to produce stable LP, which were more effective emulsifiers than other PLs (Vacklin et al., 2005). From these results, the reaction products of PLA2 and PLD were different, due to the different sites and states at which they acted on PLs.

Fig. 3.

TLC diagram of the phospholipids of OBs after enzymatic hydrolysis. Lane 1, standard; Lanes 2–7, Original OBs, lipase treatment, PLA2 treatment, PLD treatment, trypsin and PLA2 treatment, trypsin and PLD treatment. TAGs: triacylglycerols; PA: phosphatidic acid; PE: phosphatidylethanolamine; PC: phosphatidylcholine; PS: phosphatidyl serines; PI: phosphatidylinositol; LP: lysophospholipids; PLA2: phospholipase A2; PLD: phospholipase D.

Particle size and morphology of OBs    The particle size distribution of OBs is shown in Figs. 4a–4f. All OBs samples showed monomodal distribution, and OBs with trypsin and PLA2 shifted toward larger sizes. The results of Zetasizer Nano ZS show that the average particle sizes of OBs, OBs with trypsin, OBs with PLA2, and OBs with trypsin and PLA2 were 334, 352, 399, and 505 nm, respectively. The stability of isolated OBs may be damaged after treatment with trypsin. The reason was that the N-terminal and C-terminal domains of the oleosins were removed by trypsin, leaving behind the central hydrophobic domain. In terms of the particle size of the OBs after proteolytic digestion, the OBs aggregated, but no larger aggregates were formed. This was due to the steric hindrance caused by the shielding oleosins. The OBs PLs, from which the hydrophilic domains of the oleosins had been removed, were hydrolyzed with PLA2, and the particle size of the OBs became significantly larger. This may be because the hydrolyzed PLs in the OBs coat diminished the stability of the OBs, causing the OBs to aggregate and coalesce, thus increasing particle size. Therefore, oleosin and PL are indispensable for maintaining OBs stability.

Fig. 4.

Particle size distributions and optical microscopy images of OBs after enzymatic treatment.(a: original OBs, b: trypsin treatment, c: PLA2 treatment, d: trypsin and PLA2 treatment, f: PLD treatment, g: trypsin and PLD treatment)

As revealed by optical microscopy, OBs extracted from soybean seeds maintained their discrete morphology and rule integrity via surface charges and steric hindrance of oleosins and PL on the OBs surface (Huang, 1992); however, the morphology and particle size of the OBs changed significantly after PLA2 hydrolysis. PLA2 caused a small number of 1 µm oil droplets to exist in the OBs emulsion, and the size and dispersion of the OBs particles were generally uniform. After trypsin and PLA2 enzymolysis, the particle size of the OBs was larger than that of PLA2, and a few large oil droplets existed. However, there was no small peak at the larger size in the particle size analysis (Fig. 4d), which may be due to the uneven distribution of large droplets, leading to the failure to absorb this part of the droplets samples. The average particle sizes of OBs with PLD and OBs with trypsin and PLD were 335 and 349 nm, respectively. After treatment with PLD, the shapes of OBs were regular and spherical, without other adverse changes (Figs. 4e–4f). Thus, PLD has little effect on the morphology and particle size of OBs. In general, the microscopic observations were consistent with the OBs particle size determination results.

Enzymatic hydrolysis of oleosins    Trypsin treatment of oleosins in the pH 11.0-OBs resulted in the production of smaller polypeptides, as revealed by SDS-PAGE (Fig. 5). Lane 2 is the control group, which consisted of OBs without trypsin. The protein bands did not change, indicating that the oleosins remained stable for a long time, without the action of other exogenous proteases. After 30 s of trypsin hydrolysis, the oleosins were rapidly enzymized. Moreover, 24 kDa oleosin was completely enzymatically hydrolyzed within 2 min, resulting in some peptide segments of low molecular mass. With increasing hydrolysis time, the early-formed peptides were cleaved by trypsin into smaller peptides. However, the caleosin remained unchanged, even after 30 min of hydrolysis. The same phenomenon was observed in the trypsin hydrolysis of maize OBs (Tzen and Huang, 1992). Peptides with low molecular mass dissociated from the oleosins did not dissolve in the aqueous phase because the C-terminal and N-terminal domains of the oleosins were cleared by the trypsin. The PLs embedded in the OB surface were also exposed as a result of the enzymatic hydrolysis and dissolution of oleosins. In addition, the protective barrier of the TAGs core in the OBs deteriorated, exposing TAGs and making them accessible substrates for certain enzymes, such as LOX and lipase.

Fig. 5.

Tricine-SDS-PAGE profiles of OBs at different hydrolysis time. Lanes 1–2, marker, original OBs; Lanes 3–9, OBs at different hydrolysis time: 30s, 1min, 2min, 5min, 10min, 15min, 30 min.

Volatile compounds of the reaction system    LOOHs are the primary products of LOX-catalyzed oxidation of polyunsaturated FFAs, which can be converted into volatile flavour compounds (Kobayashi et al., 1995; Andreou and Feussner, 2009). To assess oxidation during the enzymolysis treatment of OBs, the formation of LOOHs was monitored, along with volatile flavour compounds in the reaction system (Table 1) and soymilk. These results are shown in Tables 2 and 3. Off-flavours are the most prominent characteristics of soymilk products, and the study of off-flavours has attracted extensive attention of food researchers. The products of LOX-oxidized OBs (Model 1) indicate that OBs emulsions appear to be less effcient substrates than FFAs micelles (Beisson et al., 2001). It can also be concluded that the soybean endogenous enzymes have little effect on the OBs during grinding. Compared with oil emulsions containing FFAs micelles, the main reason for the oxidative inertness of OBs is that the PL/protein shell of the OBs reduces its activity as a catalytic substrate, so that the reaction of LOX on the OBs is not significant.

Table 2. LOOHs of oil bodies reaction system and soymilk

	Model										Soymilk
	1	2	3	4	5	6	7	8	9	10
Concentration (µM)*	–	1.72	–	9.80	–	1.86	–	5.88	9.02	13.83		22.86

“–”: Not detected

The data shown are the average value.

*  The concentration of lipid hydroperoxides is indicated with H₂O₂ µM equivalent.

Table 3. Volatile substances of oil bodies reaction system and soymilk

Volatile substances name	Model										Soymilk
	1	2	3	4	5	6	7	8	9	10
Ethyl acetate	0	0	0	1.25	0	0	0	0.89	1.25	2.15	2.89
Pentanal	1.00	3.05	0.99	15.25	1.01	3.49	1.38	12.89	21.71	25.61	13.70
1-Penten-3-one	0.95	1.08	0.96	18.54	0.99	1.83	0.85	25.60	37.50	40.55	4.55
2,3-Pentanedione	0.27	0.28	0.25	0.68	0.30	0.34	0.26	0.62	1.44	1.99	3.22
Hexanal	35.5	66.8	34.60	356.90	36.80	68.60	35.90	164.50	215.2	425.7	640.79
(E)-2-Pentenal	1.62	1.67	1.60	15.44	1.56	2.10	1.64	9.20	16.40	21.03	20.74
1-Penten-3-ol	0	0	0	2.53	0	0	0	2.27	2.67	4.15	5.68
Heptanal	1.79	2.36	1.75	10.72	1.88	2.48	1.74	6.00	6.98	10.74	12.08
(E)-2-Hexenal	2.71	4.85	2.70	6.74	2.66	4.61	2.71	19.84	25.54	42.55	85.47
2-Pentyl-furan	0.17	0.34	0.15	12.79	0.17	0.38	0.20	1.66	3.46	28.33	2.49
3-Octanone	0	0	0	1.55	0	0	0	1.17	1.33	2.62	38.85
1-Pentanol	0.66	1.11	0.64	5.65	0.65	1.31	0.68	5.35	6.06	11.13	16.12
Octanal	0.52	1.22	0.53	2.15	0.55	1.42	0.50	0.55	1.03	2.89	5.85
1-Octen-3-one	1.14	1.38	1.15	5.90	1.25	1.33	1.28	4.37	7.57	15.89	17.04
(E)-2-Heptenal	2.46	4.34	2.44	51.85	2.47	4.10	2.50	94.37	119.37	130.50	34.56
2,3-Octanedione	2.85	3.61	2.82	1.99	2.76	3.54	2.99	5.00	5.62	6.55	1.85
1-Hexanol	0.50	0.99	0.51	9.18	0.54	0.85	0.57	1.89	1.41	12.45	310.01
Nonanal	3.19	4.51	3.15	3.87	3.04	4.28	3.45	4.07	6.12	8.68	12.98
3-Octen-2-one	0.09	0.20	0.10	2.05	0.12	0.15	0.17	0.88	1.24	2.45	2.39
(E)-2-Octenal	1.77	3.87	1.75	15.32	1.85	1.88	1.95	9.27	13.67	20.33	12.45
Acetic acid	0.36	0.37	0.35	1.05	0.32	0.41	0.37	0.21	0.26	1.12	1.58
1-Octen-3-ol	3.48	4.50	3.50	23.59	3.81	4.51	3.97	25.21	35.48	55.8	170.55
Heptanol	0.17	0.18	0.18	3.51	0.19	0.20	0.18	0.46	0.66	3.78	5.5
(E,E)-2,4-Heptadienal	0.36	0.39	0.35	5.38	0.39	0.57	0.49	18.62	23.02	27.52	2.62
Decanal	0.65	0.67	0.66	0.89	0.57	0.77	0.60	0.90	0.96	1.22	1.11
Benzaldehyde	1.55	1.88	1.50	3.55	1.45	1.59	1.41	2.83	3.20	4.21	3.21
(E)-2-Nonenal	0.85	1.04	0.82	2.63	0.90	0.86	0.93	1.12	2.58	4.02	5.78
1-Octanol	0.54	0.65	0.55	3.01	0.57	0.61	0.57	1.36	2.03	3.24	3.83
(E)-2-Decenal	0	0	0	2.10	0	0	0	0	0	2.21	4.42
(E,E)-2,4-Nonadienal	0.33	0.46	0.32	1.48	0.35	0.45	0.36	1.51	2.23	2.85	0.85
2-dodecenal	0	0	0	0.67	0	0	0	0.91	1.39	1.55	1.11
(E,E)-2,4-Decadienal	0.75	1.04	0.73	10.16	0.74	1.14	0.98	3.93	6.67	12.66	0.33
Hexanoic acid	0.63	0.70	0.65	4.18	0.71	0.94	0.90	2.12	4.95	5.12	1.32
Total amount (µg/L)	66.86	113.50	65.70	602.60	68.60	114.73	69.53	429.60	579.00	941.60	1445.90

The data shown are the average value.

When trypsin was added to the OBs reaction system, the concentration of LOOHs and volatile flavour compounds in the OBs system increased, compared with the control groups. This indicated that the hydrolysis of oleosins (Fig. 5) provided more access for phospholipase, lipase, or LOX to act on the innernal lipids of OBs. The concentration of volatile flavour compounds in Model 1 was only 66.86 µg/L, which was equivalent to that in Model 3. This indicates that OBs are not the optimal substrates for LOX. The concentration of volatile flavour compounds in Model 4 was 9.17 times that of Model 3. This is because lipase hydrolyzes the TAG matrix of OBs to generate large numbers of FFAs that are easily oxidized by LOX. The concentration of volatile flavour compounds in Models 8 and 9 increased, and those in Models 5 and 7 did not change. This is because PLD, unlike PLA2, cannot hydrolyze PLs in OBs, even in the presence of trypsin. The increase in the concentration of volatile compounds in Model 10 indicates that lipase can hydrolyze internal TAGs through the OBs surface, when the PLs in the OBs surface are hydrolyzed by PLA2. As shown in Table 3, the types of volatile flavour compounds generated by the OBs reaction system were all within the range of the volatile flavour compounds of soymilk, but there were certain differences in their concentrations. This showed that the reaction process in the OBs reaction system was basically the same as that of soymilk, but the degree of reaction of the two was different, which was related to the difference in their composition and content.

The absence of LOOHs detection in Models 1, 3, 5 and 7, as shown in Table 2 is consistent with the presence of volatile flavour compounds in the corresponding Models shown in Table 3. This was because the formation of volatile flavour compounds in the OBs reaction system is linked to the OBs substrate itself, rather than the OBs oxidation reaction. Pearson's correlation coefficient was calculated to determine the relationship between the concentration of LOOHs (Table 2) and the concentration of volatile compounds generated in the corresponding OBs reaction systems and soymilk (Table 3). The results show that there is a strong positive correlation between the concentration of volatile flavour compounds and the concentration of LOOHs generated in the OBs reaction systems and soymilk (r = 0.997). These results are in agreement with the results of a previous study (Mizutani and Hashimoto, 2004).

Conclusions

The OBs extracted from soymilk have a barrier composed of integral proteins and adsorbed extrinsic proteins on the surface, which prevents LOX from accessing the internal lipid substrate. With the increase in extraction pH, the extrinsic proteins were gradually removed, and the OBs containing only integral proteins were obtained, i.e., pH 11.0-OBs (Fig. 2). Oleosins could be rapidly hydrolyzed by trypsin to produce protein peptides with low molecular weight (Fig. 5). As the hydrolysis of oleosins destroyed the integrity of the surface layer of the OBs, LOX was accessible to its internal lipid substrate, producing a certain amount of oxidation products (Tables 2 and 4). Among lipase, PLA2, and PLD, only PLA2 could hydrolyze the PLs layer of OBs (Fig. 3), with an increase in particle size (Fig. 4) and the promotion of the oxidation of LOX to produce more oxidation products (Tables 2 and 4). Based on Model 4 and Model 10, the destruction of the OBs surface layer, as caused by the hydrolysis of oleosins or PLs in the OBs surface layer, can provide a channel for the lipase to hydrolyze the TAGs inside the OBs. It can therefore be concluded that the destruction of the OBs surface layer (not only the hydrolysis of oleosins) is a prerequisite for the occurrence of LOX catalytic oxidation. These results suggest that in addition to inhibiting or inactivating the activity of soybean LOX, inhibiting or inactivating the activity of its phospholipase and trypsin is also important for improving the flavour of soymilk.

Acknowledgements    Gas chromatography-mass spectrometry was performed in the State Key Laboratory of Food Science and Technology, Jiangnan University. We thank LetPub (www.letpub.com) for its linguistic assistance and scientific consultation during the preparation of this manuscript.

Conflict of Interest    The authors declare no conflict of interest.

References

•  Axelrod,  B.,  Cheesebrough,  T. M., and  Laasko,  S. (1981). Lipoxygenase from soybeans. Method Enzymol., 71, 441-451.
•  Andreou,  A. and  Feussner,  I. (2009). Lipoxygenases-structure and reaction mechanism. Phytochemistry, 13-14, 1504-1510.
•  Beisson,  F.,  Ferté,  N.,  Bruley,  S.,  Voultoury,  R.,  Verger,  R., and  Arondel,  V. (2001). Oil-bodies as substrates for lipolytic enzymes. Bba-mol. Cell Biol. L., 1531, 47-58.
•  Chen,  Y. and  Ono,  T. (2010). Simple extraction method of non-allergenic intact soybean oil bodies that are thermally stable in an aqueous medium. J. Agric. Food Chem., 58, 7402-7407.
•  DeLong,  J. M.,  Prange,  R. K.,  Hodges,  D. M.,  Forney,  C. F.,  Bishop,  M. C., and  Quilliam,  M. (2002). Using a modified ferrous oxidation–xylenol orange (FOX) assay for detection of lipid hydroperoxides in plant tissue. J. Agric. Food Chem., 50, 248-254.
•  Fukushige,  H.,  Wang,  C.,  Simpson,  T. D.,  Gardner,  H. W., and  Hildebrand,  D. F. (2005). Purification and identification of linoleic acid hydroperoxides generated by soybean seed lipoxygenases 2 and 3. J. Agric. Food Chem., 53, 5691-5694.
•  Gupta,  A. and  Bhatla,  S. C. (2007). Preferential phospholipase A2 activity on the oil bodies in cotyledons during seed germination in Helianthus annuus L. cv. Morden. Plant Sci., 172, 535-543.
•  Huang,  A. H. C. (1992). Oil bodies and oleosins in seeds. Ann. Rev. Plant. Physiol. Mol. Bio., 43, 177-200.
•  Iassonova,  D. R.,  Johnson,  L. A.,  Hammond,  E. G., and  Beattie,  S. E. (2009). Evidence of an enzymatic source of off flavors in “lipoxygenase-null” soybeans. J. Am. Oil Chem. Soc., 86, 59-64.
•  Jung,  D. M. and  Ebeler,  S. E. (2003). Headspace solid-phase microextraction method for the study of the volatility of selected flavor compounds. J. Agric. Food Chem., 51, 200-205.
•  Keast,  R. S. J. and  Lau,  J. J. (2006). Culture-specific variation in the flavor profile of soymilks. J. Food Sci., 71, 567-572.
•  Kobayashi,  A.,  Tsuda,  Y.,  Hirata,  N.,  Kubota,  K., and  Kitamura,  K. (1995). Aroma constituents of soybean [Glycine max (L.) Merrill.] milk lacking lipoxygenase isozymes. J. Agric. Food Chem., 43, 2449-2452.
•  Kumari,  S.,  Memba,  L. J.,  Dahuja,  A.,  Vinutha,  T.,  Saha,  S., and  Sachdev,  A. (2016). Elucidation of the role of oleosin in off-flavour generation in soymeal through supercritical CO₂ and biotic elicitor treatments. Food Chem., 205, 264-271.
•  Lin,  L.,  Tai,  S. S.,  Peng,  C., and  Tzen,  J. T. (2002). Steroleosin, a sterol-binding dehydrogenase in seed oil bodies. Plant Physiol., 128, 1200-1211.
•  Leprince,  O.,  Vanaelst,  A. C.,  Pritchard,  H.W., and  Murphy,  D. J. (1998). Oleosins prevent oil-body coalescence during seed imbibition as suggested by a low-temperature scanning electron microscope study of desiccation-tolerant and - sensitive oilseeds. Planta, 204, 109-119.
•  List,  G. R.,  Mounts,  T. L., and  Lanser,  A. C. (1992). Factors promoting the formation of the nonhydratable soybean phosphatide. J. Am. Oil Chem. Soc., 69, 443-446.
•  Matsui,  K.,  Hijiya,  K.,  Tabuchi,  Y., and  Kajiwara,  T. (1999). Cucumber cotyledon lipoxygenase during postgerminative growth. Its expression and action on lipid bodies. Plant Physiol., 119, 1279-1287.
•  Murphy,  D. J. (2001). The biogenesis and functions of lipid bodies in animals, plants and microorganisms. Prog. Lipid Res., 40, 325-438.
•  Morr,  C. V. and  Ha,  E.Y.W. (1991). Processing to prevent formation of off-flavors in soy products. Comments Agric. Food Chem., 2, 247-260.
•  Mellor,  N.,  Bligh,  F.,  Chandler,  I., and  Hodgman,  C. (2010). Reduction of off-flavor generation in soybean homogenates: a mathematical model. J. Food Sci., 75, 131-138.
•  Mizutani,  T. and  Hashimoto,  H. (2004). Effect of grinding temperature on hydroperoxide and off-flavor contents during soymilk manufacturing process. J. Food Sci., 69, 112-116.
•  Napier,  J. A.,  Stobart,  A. K., and  Shewry,  P. R. (1996). The structure and biogenesis of plant oil bodies: the role of the ER membrane and the oleosin class of proteins. Plant Mol. Biol., 31, 945-956.
•  Noll,  F.,  May,  C., and  Kindl,  H. (2000). Phospholipid monolayer of plant lipid bodies attacked by phospholipase A₂ shows 80 nm holes analyzed by atomic force microscopy. Biophys. Chem., 86, 29-35.
•  Rudolph,  M.,  Schlereth,  A.,  Körner,  M.,  Feussner,  K.,  Berndt,  E.,  Melzer,  M.,  Hornung,  E., and  Feussner,  I. (2011). The lipoxygenase-dependent oxygenation of lipid body membranes is promoted by a patatin-type phospholipase in cucumber cotyledons. J. Exp. Bot., 62, 749-760.
•  Reza,  S. H. and  Chander,  B. S. H. R. (2002). Differential sensitivity of oleosins to proteolysis during oil body mobilization in germinating sunflower seedlings. Plant Cell Physiol., 43, 1117-1126.
•  Schägger,  H. (2006). Tricine-SDS-PAGE. Nature Protocols, 1, 16-22.
•  Shimada,  T. L. and  Hara-Nishimura,  I. (2010). Oil-body-membrane proteins and their physiological functions in plants. Biol. Pharm Bull., 33, 360-363.
•  Tzen,  J. T. C., and  Huang,  A. H. C. (1992). Surface structure and properties of plant seed oil bodies. J. Cell Biol., 117, 327-335.
•  Tzen,  J. T. C.,  Cao,  Y.,  Laurent,  P.,  Ratnayake,  C., and  Huang,  A. H. C. (1993). Lipids, proteins, and structure of seed oil bodies from diverse species. Plant Physiol., 101, 267-276.
•  Vacklin,  H. P.,  Tiberg,  F.,  Fragneto,  G., and  Thomas,  R. K. (2005). Phospholipase A₂ hydrolysis of supported phospholipid bilayers: a neutron reflectivity and ellipsometry study. Biochemistry, 44, 2811-2821.
•  Wadahama,  H.,  Iwasaki,  K.,  Matsusaki,  M.,  Nishizawa,  K.,  Ishimoto,  M.,  Arisaka,  F.,  Takagi,  K., and  Urade,  R. (2012). Accumulation of β-conglycinin in soybean cotyledon through the formation of disulfide bonds between α′-and α-subunits. Plant Physiol., 158, 1395-1405.
•  Yuan,  S. H. and  Chang,  S. K. (2007). Selected odor compounds in cooked soymilk as affected by soybean materials and direct steam injection. J. Food Sci., 72, 481-486.
•  Zhuang,  H.,  Hildebrand,  D. F.,  Andersen,  R. A., and  Hamiltonkemp,  T. R. (1991). Effects of polyunsaturated free fatty acids and esterified linoleoyl derivatives on oxygen consumption and C6 aldehyde formation with soybean seed homogenates. J. Agric. Food Chem., 39, 1357-1364.
•  Zhao,  L.,  Chen,  Y.,  Yan,  Z.,  Kong,  X., and  Hua,  Y. (2016). Physicochemical and rheological properties and oxidative stability of oil bodies recovered from soybean aqueous extract at different pHs. Food Hydrocoll., 61, 685-694.