Reducing Washout of Proteins from Defatted Soybean Flakes by Alkaline Extraction: Fractioning and Characterization

Abstract

Human health, sustainable development, numerous environmental issues, and animal welfare are increasingly driving research and development of plant-based protein products that can serve as meat substitutes. This trend is expected to continue in the coming years due to growing consumer awareness, with people gradually shifting from animal-based foods to more sustainable plant-based options. Soy proteins are a valuable source of plant proteins and are widely used in human and animal diets due to their nutritional value and health benefits. In this study, soybean protein extraction by two methods was compared: water extraction (lower salt content) and Tris-HCl extraction (higher salt content), aiming to characterize the resulting protein fractions. These fractions were studied using differential precipitation based on the isoelectric point. Protein identification by SDS-PAGE, scanning electron microscopy (SEM) for cellular structure assessment, and Fourier-transform infrared spectroscopy (FTIR) were used to determine residual protein left in the solid fraction after extraction using the two methods. Electrophoresis assays revealed the presence of the four main protein fractions (2S, 7S, 11S, and soy whey proteins) in the defatted soybean flakes, establishing the protein profile of Brazilian soybeans and for the two main waste streams of the production process—spent flakes and whey. The separation of fractions was carried out by differential precipitation. FTIR analysis indicated higher residual protein levels in solid residues after the water extraction method compared to the Tris-HCl extraction method. SEM analysis revealed the removal of protein bodies in both extraction methods and the presence of residual oil-containing bodies. Both methodologies are viable alternatives for the industrial separation of soybean protein fractions. Differential precipitation could be implemented to produce isolated products and improve the nutritional profile, increase process yield thus generating less industrial waste and driving the process towards environmental sustainability.

1. Introduction

Soybean (Glycine max) is widely recognized for its significant contribution to the global production of plant-based food protein, feed, and oil. Although global consumption of plant-based proteins is lower than that of animal origin, soy is uniquely positioned to help meet the caloric and proteic needs of a growing global population [1].

Soybean contains approximately 40% protein, 20% oil, 25% carbohydrates, and 5% crude fiber (dry weight) [2]. Approximately 17% of the carbohydrates are soluble, and 21% are insoluble, distributed among three main structures in the grain: the seed coats (8%), the cotyledon (90%), and the hypocotyl, or embryonic axis (2%).

Soybean storage proteins are divided into four main groups, namely 2S, 7S, 11S, and 15S, with albumins present in the 2S fraction, whereas globulins are mainly found in the 7S, 11S, and 15S fractions [2]. The 7S and the 11S fractions, mainly composed of β-conglycinin and glycinin, respectively, constitute 80% of the total protein content [3].

The molecular weight of β-conglycinin (7S) typically ranges from 150 kDa to 180 kDa and is mainly composed of the subunits α′ (~76 kDa), α (~72 kDa), and β (~53 kDa) [2]. Two additional components are also identified in this same fraction, the γ-conglycinin protein (210 kDa) [4] and the allergenic protein Gly m Bd 30K (34 kDa), described by Ogawa et al. [5].

Glycinin (11S) is a hexamer with a molecular weight of 360 kDa, in which each of the six identified subunits consists of a basic polypeptide and one or two acidic polypeptides linked by sulfur bridges [6]. The molecular mass of the basic polypeptides is approximately 20 kDa, while the acidic polypeptides have molecular weights of 44, 37, and 10 kDa, respectively [7].

The 2S fraction is found in smaller quantities compared to the 7S and 11S fractions and consists mainly of anti-nutritional and allergenic factors [8,9]. Its main components are two types of trypsin inhibitors, Bowman–Birk (8 kDa) and Kunitz (21.5 kDa) [7,10,11,12].

Soy whey proteins are defined as the fraction that makes up the supernatant of the acid precipitation of soy proteins. In addition to including proteins from the 2S and 7S fractions that remain soluble [13], it also contains lipoxygenase (102 kDa), β-amylase (61.7 kDa), and lectin (119 kDa) [3,14,15,16]. The latter is a tetramer composed of identical units of approximately 30 kDa [17].

The alkaline aqueous extraction of proteins from soybean defatted flakes depends on pH, temperature, and ionic strength, relying on the differential solubility of the various components in the defatted flakes [7,18] . The precipitation of the extracted protein is commonly carried out at the isoelectric point (pI), which falls in the range of 4.2 to 4.5 in the case of soybean proteins [19]. However, the two main soy fractions, 7S and 11S, have different pIs, as well as water and fat absorption capacities, emulsification, foam formation ability, gel formation, and adhesive performance [20]. These different properties lead to a varied range of characteristics in the final protein isolate product.

Understanding the soybean protein fractions in different steps of the isolated protein process enables the improvement of yield and product output. This, in turn, allows for the reduction in protein loss to spent flakes and whey streams, generating less industrial waste. Thus, the process is rendered more environmentally friendly and sustainable, while at the same time making it more competitive as a raw material for feed protein sources.

There is little background on the impact of different extraction methods on the protein profile extracted from soybeans grown in Brazil, which was responsible for 40% of the global soybean production in 2023 [21].

In light of these considerations, the aim of this study was to analyze protein fractions in soybean defatted flakes and waste streams, that is, spent flakes and whey, as well as to evaluate the impact of extraction methods with higher and lower salt amounts. The differential precipitation process was also investigated as an alternative for achieving full protein recovery in the production process, as well as an alternative to incorporate specific features of soybean protein fractions into food and feed products.

2. Materials and Methods

2.1. Materials

Defatted soybean flakes, spent flakes, and soy whey samples (the latter being referred to as industrial whey samples in this paper) were provided by a grain processing company (Brazil). Samples were frozen at −20 °C until use. All chemicals used in this research were purchased from Sigma-Aldrich (St. Louis, MO, USA) and were of analytical grade.

2.2. Alkaline Extraction Methods

The general process for obtaining isolated soybean protein from defatted soybean flakes is schematically illustrated in Figure 1.

In this work, the alkaline extraction protocols were adapted from Din et al. [22] and Liu et al. [23], hereinafter referred to as water extraction (WE) and Tris-HCl extraction (THE), respectively. The WE protocol uses lower salt concentration compared to the THE method. For the WE protocol, soluble compounds were extracted from 10 g of dry soybean defatted flakes or from 20 g of humid spent flakes using water in ratios of 1:10 (w/v) and 1:5 (w/v), respectively, to correct for the sample moisture. The pH was adjusted to 7.5 with a 2 M NaOH solution at room temperature, and the mixtures were agitated at 160 rpm on a benchtop shaker for 1 h. For the THE protocol, soluble compounds were only extracted from 10 g of dry soybean defatted flakes, using a Tris-HCl buffer solution (0.03 M, pH 8.5) in a 1:10 (w/v) ratio. The samples were agitated at 160 rpm on a benchtop shaker at 45 °C for 1 h, according to the methodology published by Liu et al. [23]. Aqueous extracts from both procedures were centrifuged using an ultracentrifuge (Hitachi CR21G III) at 9000× g and a temperature of 4 °C for 30 min. The supernatants were collected by pouring, and the solids were stored in a freezer at −20 °C for SEM image analysis.

2.3. Differential Protein Precipitation

The process of differential protein precipitation was conducted for all the samples, in duplicate, according to the method adapted from Din et al. [22] as illustrated in Figure 2.

To the supernatants obtained from alkaline extractions, 0.98 g·L⁻¹ of dry sodium bisulfite was added. A 2M HCl solution was added to adjust the pH to 6.4. The mixtures were refrigerated at 4 °C overnight. Insoluble fractions were removed from the solution by centrifugation at 6500× g for 20 min at 4 °C. Solid NaCl was added to the supernatants to adjust the salt concentration to 0.25 mol·L⁻¹, and the pH was adjusted to 5.0 using a 2 M HCl solution. The solutions were left in an ice bath for 1 h, then the insoluble fractions were removed by centrifugation at 9000× g for 30 min at 4 °C. The obtained supernatants were diluted two times in volume with chilled distilled water, and the pH was adjusted to 4.8 with a 2 M HCl solution. Insoluble fractions were removed from the solution by centrifugation at 6500× g for 20 min at 4 °C. The supernatants were also collected, hereinafter referred to as laboratory whey samples.

2.4. Protein Resuspension and Drying

All the collected insoluble fractions were resuspended in distilled water with the aid of vortex agitation, and the pH was adjusted to 7.5 with NaOH. The samples were stored in a freezer at −20 °C for SDS-PAGE analysis. Aliquots were also lyophilized for SEM analysis.

2.5. Separation of Fraction 2S in the Industrial Whey

An aliquot of 12 mL of the industrial whey sample was filtered through a 50 kDa Amicon Ultra-15 50K centrifugal filter (Merck, Rahway, NJ, USA) at 5000× g centrifugation for 30 min, as per manufacturer instructions. Both the permeate and retentate samples were collected and stored at −20 °C for SDS-PAGE analysis.

2.6. Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis (SDS-PAGE)

SDS-PAGE was conducted according to the method described by Laemmli [24], using the BioRad Mini-PROTEAN TGX system (Biorad, Hercules, CA, USA). Gels with 12% for resolution and 4.0% for stacking were employed, along with the BenchMark Protein Ladder (Invitrogen, Waltham, MA, USA) as a molecular weight marker. Electrophoresis was carried out at a constant current of 60 V for 20 min followed by 150 V for 1 h, and then stained with Coomassie Brilliant Blue R250. For samples with protein concentrations below the detection limit of Coomassie Brilliant Blue R250, silver nitrate staining was employed according to Kavran et al. [25].

2.7. Scanning Electron Microscopy (SEM)

The defatted soybean flakes and the solid residues from the alkaline extractions with water and Tris-HCl were lyophilized and subjected to SEM image analysis according to the procedure adapted from Rahman et al. [26]. Images were captured using a Phenom ProX benchtop scanning electron microscope (ThermoFischer, Waltham, MA, USA) at 10 kV.

2.8. Fourier Transform Infrared Spectroscopy (FTIR)

The solid residues from the alkaline extractions with water and Tris-HCl underwent FTIR analysis. The spectrum was obtained using a MIR-FTIR Frontier molecular absorption spectrophotometer (Perkin Elmer, Norwalk, CT, USA). The UATR (universal attenuated total reflectance) analysis mode was used, and the spectra were collected in transmittance, with 16 scans and a resolution of 4 cm⁻¹ in the range of 4000–650 cm⁻¹.

3. Results and Discussion

3.1. Composition of Samples

The composition of the samples of defatted soybean flakes and spent flakes, as determined by the sample-supplying company, is presented in

Table 1. Quality control analyses of industrial samples.

Sample	Moisture (%)	Solids (%)	Protein (%)	Oil (%)	Fiber (%)
Defatted flakes	5.72 ± 0.09	-	52.80 ± 0.16	0.52 ± 0.03	3.28 ± 0.18
Spent Flakes	-	13.10 ± 0.85	3.77 ± 0.97	-	-

The results are the average of triplicate experiments.

.

Protein content in Table 1 is presented as is for both samples. It is, however, more often referred to as dry base. For defatted flakes and spent flakes, the dry base protein was on average 56.0% (standard deviation = 0.15%) and 28.6% (standard deviation = 5.79%), respectively.

The defatted soybean flakes sample presents higher fiber and lower oil content when compared to data published by Perkins et al. [27], who reported protein content of defatted soybean flakes of 55%, 0.8% oil, 2.4% fiber, and 7% moisture. This variability can be attributed to the different oil extraction processes, soybean genetics, or geographical factors [28].

3.2. Profile of Protein Fractions

Figure 3 displays the SDS-PAGE profile of protein fractions of the industrial whey samples. Samples IW and +50_IW in Figure 3a represent the industrial whey before filtration and its retentate after filtration, respectively. It is possible to identify, on both, characteristic bands of the α, α’, and β subunits of β-conglycinin (72 kDa, 76 kDa, and 53 kDa), as well as a band of approximately 34 kDa, which may correspond to the Gly m Bd 30K compound accompanying the 7S fraction. Furthermore, the presence of bands at 100 kDa, 60 kDa, and 33 kDa may indicate the presence of the characteristic whey proteins lipoxygenase, β-amylase, and lectin subunits. Bands of approximately 20 kDa, in turn, may indicate the presence of the 2S fraction represented by the Kunitz trypsin inhibitor. Similar labeling was made by Zarkadas et al. [29] while working on the identification of storage proteins in samples of different soybean genotypes.

Seeking to increase the sharpness of the bands present in sample −50_IW (the permeate) in Figure 3a, silver nitrate staining was performed, the result of which is presented in Figure 3b. Silver staining resulted in bands of lower resolution but higher intensity, allowing for the identification of the 34 kDa and 20 kDa bands (Gly m Bd 30K and Kunitz trypsin inhibitor) of the 7S and 2S fractions, respectively.

The acidic (44, 37, and 10 kDa) and basic (20 kDa) glycinin polypeptides are not identified in any of the samples from these gels (Figure 3a, b). This may indicate that the industrial processing was efficient in precipitating the 11S fraction while still in the plant facility—assuming this fraction was indeed extracted from the soybean defatted flakes. It is also possible that this result indicates that the industrial extraction or material preparation conditions were not sufficient to remove part of the glycinin from the soybean-defatted flakes in the first place. This possibility is compatible with the residual protein content of 28.6% in the spent flakes, as reported by the sample-supplying company. Pierce et al. [30] in a study on the polysaccharides composing dietary fiber, reported levels as low as 10.05% for residual protein in commercial fiber samples.

Figure 4 presents the SDS-PAGE profile of fractions precipitated at pH 6.4 and pH 4.5. All the samples produced during differential precipitation at pH 6.4 (DF_WE_11S, DF_THE_11S, and SF_WE_11S) exhibit bands corresponding to the acidic (44, 37, and 10 kDa) and basic (20 kDa) polypeptides of glycinin [7]. The sample generated by extracting this fraction with water (DF_WE_11S) lacks the bands characteristic of the 7S and 2S fractions, supporting the efficiency of the differential precipitation process proposed by [22].

The samples from the fraction precipitated at pH 4.5 (DF_WE_7S and DF_THE_7S) show well-defined bands for the α, α’, and β subunits of β-conglycinin (72 kDa, 76 kDa, and 53 kDa) [2] along with bands of lower intensity corresponding to glycinin polypeptides. This indicates that the differential precipitation of the 7S fraction, according to the adapted protocol by Din et al. [22], is efficient, although not entirely selective. The authors of the protocol, optimizing parameters such as enzyme addition and fraction recycling to improve yields, reported lower purity for the 7S fraction compared to the 11S fraction for all variations of experiments conducted. In their experiments, purity was at its highest for the water extraction of 11S fraction at 98.64%. The procedure to obtain the 7S fraction from the spent flake samples did not generate a solid precipitate; therefore, these samples were not included in the gel.

Industrially, simultaneous precipitation of both fractions is economically advantageous as it increases the product generation capacity. On the other hand, efficient selective precipitation can be used for manipulating the properties and functionality of the final product since the 7S fraction has distinctive characteristics, such as better emulsification and foam production properties, when compared to the 11S fraction [31].

Figure 5 presents the SDS-PAGE profile of fractions precipitated at pH 5.0 and samples of the laboratory-prepared soy whey. The fractions precipitated at pH 5.0 (DF_WE_Int, DF_THE_Int, and SF_WE_Int) from all the samples show characteristic bands of β-conglycinin (72 kDa, 76 kDa, and 53 kDa) and glycinin (40 kDa, 36 kDa, and 20 kDa), in addition to the absence of bands from the 2S fraction. According to Liu et al. [23], the removal of the fraction precipitated at this point (referred to as intermediate) is performed with the goal of improving the purity of the following fraction to be precipitated (7S fraction), at the expense of its yield.

Laboratory whey samples (DF_WE_LW and DF_THE_LW) exhibit bands that may correspond to the presence of lipoxygenase, β-amylase, and lectin subunits [3,14,15,16]. They also show characteristic bands of the 7S fraction (72 kDa, 76 kDa, and 53 kDa), suggesting that it was not completely precipitated in the previous steps. These results are in accordance with those published by Din et al. [22], who showed a maximum yield of 30.35% in the recovery of the 7S fraction from soybean defatted flakes, as well as a recovery of 1.6 to 2.0 times higher for the 11S fraction compared to the 7S fraction. In addition, the laboratory soy whey sample, whose starting material was spent flakes (SF_WE_LW), does not show the presence of whey-characteristic proteins. That may result from the fact that the original spent flake sample had already undergone industrial processing, having its whey proteins extracted and directed to the plant’s wastewater stream, which represents a loss of valuable protein, at the same time negatively impacting the environment.

3.3. Residual Protein in Spent Flakes

The FTIR spectra analysis is presented in Figure 6. The sample of solid residue from the extraction of soybean defatted flakes with water (in red) shows seven peaks, while five peaks were identified for the sample extracted with Tris-HCl (in blue).

Table 2. Data obtained from FTIR analysis and possible labeling by the authors.

Peak #	X (cm−1)	Transmittance (%)	Extracting Solution	Band Name
1a	3279.67	91.16	Tris-HCl	Amide A
1b	3280.25	79.74	Water
2b	2929.06	85.46	Water	Amide B
3a	1634.84	82.70	Tris-HCl	Amide I
3b	1635.29	63.85	Water
4a	1532.15	84.44	Tris-HCl	Amide II
4b	1537.94	68.72	Water
5a	1387.87	86.70	Tris-HCl	Amide III
5b	1394.81	75.76	Water
6b	1236.71	77.88	Water	Amide III
7a	1048.03	81.84	Tris-HCl
7b	1047.48	63.48	Water

shows labeled bands in the spectra, based on Kong et al. [32], along with the transmittance values of each peak that were used to qualitatively assess the extraction efficiency of each method.

Of the identified peaks, six of the identified peaks have wavelengths corresponding to the characteristic bands of the main chains of peptides, as shown in Table 2. The Amide I and Amide II bands indicate the presence of residual protein in the solid samples, suggesting that the extraction may have been incomplete. This hypothesis is supported by the results published by Din et al. [22], which also indicate partial extraction using both solvents. Additionally, it is supported by the composition analysis of the spent flakes generated by the industrial process, which reports a residual protein content in the fiber of 28.6% on a dry basis, as reported in Section 3.1.

The peaks in the spectrum resulting from the Tris-HCl extraction experiment consistently showed higher transmittance values than those from the water extraction method. This indicates greater absorption and, consequently, a higher amount of residual protein in samples extracted with water, suggesting less efficient protein recovery. These data reinforce the assumptions previously described regarding the comparison of the efficiency of the two extraction methodologies.

3.4. Microscopic Structure

The SEM images of the samples at three different magnifications are presented in Figure 7. In the SEM images of soybean defatted flakes (Figure 7a–c), it is possible to observe bodies with an approximately cylindrical shape deposited on flat fragments. Bodies with a similar shape in soy flour samples were identified as protein bodies by Xing et al. [33].

In the images of both solid residue samples (Figure 7d–i), the absence of protein bodies is noticeable. However, the presence of residual protein after extraction is indicated by the FTIR analysis in these same samples. These protein bodies may be in areas not visualized by SEM analysis or within cells whose structure was not disrupted during processing. Cartabia et al. [34], in their study of the characterization of fungal strains for bio-based materials, suggested that the SEM technique applied by them helped as a complementary, non-exhaustive method for characterization.

In the solid residue samples, it is also possible to observe the presence of clear and cylindrical bodies of approximately 0.5 µm, as pointed out in Figure 8a,b. These bodies correspond to the dimensions and shapes established by other authors for oil bodies, components of cotyledons [35,36]. The presence of oil bodies in insoluble fiber samples derived from defatted flakes suggests that the solvent-based oil extraction process performed before the sample collection was incomplete. These findings align with the chemical composition analysis in Section 3.1, which reported residual soybean oil in defatted soybean flakes, comprising 0.52% of the content.

The insoluble fiber, which accounts for 3.28% of the starting sample, is thought to be a part of the channel-like structures shown in Figure 8c, which may represent the cell walls of the cotyledons.

4. Conclusions

This work presents the protein profile for samples of Brazilian soybean grains and for the two main waste streams of the isolate protein production process (spent flakes and whey). Two methodologies of protein alkaline extraction were applied on defatted soybean flakes to study their effectiveness and resulting protein profiles. Both procedures were able to extract proteins of the 2S, 7S, and 11S fractions from the white flakes into the aqueous phase. The THE method yielded better results than the WE methodology, in terms of protein concentration in the extract under the tested conditions. It was also possible to replicate the procedure of differential precipitation in the alkaline extract obtained from defatted soybean flakes with both methods, resulting in isolated fractions of 7S and 11S, in addition to a mixture of whey proteins and fraction 2S in different proportions. This methodology can be considered a viable alternative for the industrial separation of soybean protein fractions since it introduces a modification of the commonly used soy protein isolate production process. It is possible that differential precipitation could be implemented to obtain products with specific physicochemical properties or with an improved nutritional profile. Today, the industrial process to obtain ISP (isolate soy protein), which is based on successive washings and protein precipitation, presents some limitations. A portion of the protein originally found in the seed is lost to effluent. Thus, these methodologies can also be applied to increase production yield aiming for full protein recovery. Future studies on this topic include the quantification of protein recovery yields, as well as the purity of recovered fractions through the identification of individual proteins. Other recommended studies are evaluating different extraction methods, using water as a solvent due to its ease of use in industrial settings, and exploring the effects of higher pH ranges.