Development of Soy-Based Meat Analogues via Wet Twin-Screw Extrusion: Enhancing Textural and Structural Properties Through Whole Yeast Powder Supplementation

For this study, SPI was sourced from Linyi Shansong Biological Products Co., Ltd. (Linyi, China); the protein sample contained 91.4 g/100 g of protein and 6.56 g/100 g of moisture on a dry basis. The whole yeast powder (dry-based), provided by Angel Yeast Co., Ltd. (Yichang, China), comprised 48.3 g/100 g of protein, 13.1 g/100 g of carbohydrates, 4.7 g/100 g of crude fat, and 23.53 g/100 g of crude dietary fiber. The yeast protein powder content was measured to be 84.2 g/100 g on a dry basis, with a moisture content of 3.3 g/100 g. Sodium chloride, urea, phosphate buffer solution, and β-mercaptoethanol were all purchased from Kunming Meibo Biotechnology Co., Ltd. (Kunming, China). The reagents required for the experiments were of analytical grade. To prepare the mixtures, different proportions of whole yeast powder and SPI were combined, with the following yeast powder-to-SPI ratios: 0.5:9.5, 1:9, 2:8, 3:7, and 4:6. Table 1 provides the precise compositions and sample quantities.

All high-moisture extrusion experiments were conducted using a laboratory-scale twin-screw extruder (FMHE22; Fumai Food Co., Ltd., Linyi, China). The feed rate of the SPI–WYP dry blend was maintained at 9.5 kg/h, with deionized water introduced at 5.9 kg/h. The extruder was equipped with seven sequential heating zones, with temperatures progressively set at 60, 80, 90, 120, 130, 140, and 140 °C. A schematic diagram of the twin-screw extruder is presented in the. For each formulation, raw materials were divided into three portions and processed under identical operating conditions, with each sample prepared in triplicate. The resulting extrudates were subsequently divided into two portions: one was vacuum freeze-dried, ground, and sieved through a 100-mesh screen for further analysis; the other was stored at 4 °C for subsequent use. Supplementary Material

The samples were observed using a scanning electron microscope (Apreo 2C, Thermo Fisher Scientific, Bremen, Germany). The observation process and parameters followed the method described in [13]; the operating voltage of the SEM was 10 kV. Magnifications of ×500, ×1500, were used.

The samples were cut into dimensions of 40 mm × 20 mm × 10 mm using a knife and were fixed onto an iron rod for macrostructure observation through photography.

To further evaluate the alignment and density of fibrous structures, fiber orientation analysis was performed using ImageJ v.1.54 and its OrientationJ plugin. The fiber alignment of the extrudate microstructure was quantitatively assessed using the OrientationJ plugin in ImageJ (National Institutes of Health, Bethesda, MD, USA). SEM images (×1500 magnification) were first converted to 8-bit grayscale and then analyzed using OrientationJ Distribution to extract the angular distribution of fiber orientation. The analysis parameters included a smoothing factor of 2 and a region of interest (ROI) covering the central textured area. The output included an orientation histogram ranging from −90° to +90°, with the peak angle representing the dominant fiber alignment direction. A sharper and higher peak near 0° indicated better unidirectional alignment. The coherency index and standard deviation were also extracted to evaluate the degree of orientation quantitatively. SEM images of samples with varying substitution levels of whole yeast powder or yeast protein powder were also processed using ImageJ binary thresholding. Additional details are provided in the. Supplementary Material (Figure S4)

The extrudates were torn along the direction of mold extrusion, fixed onto an iron rod, and photographed in a studio using a smartphone (iPhone 15, Apple Co., Cupertino, CA, USA) [14].

The texture characteristics of the samples were tested using a texture analyzer (TA-XT Plus, Stable Micro Systems, Godalming, UK). The samples were cut into strips of 20 mm × 20 mm × 10 mm using a knife and fixed onto the testing platform. The testing parameters followed the method described in [15]. The vertical force (FV) and parallel force (FL) were recorded, and their ratio (FV/FL) was used to evaluate the degree of fiber alignment (fiberization). In addition, chewiness and hardness were measured to characterize the overall texture properties of the samples.

Low-field nuclear magnetic resonance (LF-NMR) measurements were performed using a benchtop NMR analyzer (PQ001-20-020V, Niumag Analytical Instrument Co., Ltd., Suzhou, China) operating at 20 MHz. The Carr–Purcell–Meiboom–Gill (CPMG) pulse sequence was used to determine the spin–spin relaxation time (T₂) and the corresponding relative peak area (P₂). The acquisition parameters were set as follows: echo time (TE) of 0.3 ms, 1024 echoes per scan, repetition time (TR) of 5000 ms, and four accumulated scans. Signal acquisition and T₂ distribution analysis were performed using the MultiExp Inv 2023 software (Niumag).

The intermolecular interactions of proteins were tested with slight modifications based on the method described in [16]. The samples were dissolved separately in the following solutions: 0.05 mol/L NaCl (S1), 0.6 mol/L NaCl (S2), 0.6 mol/L NaCl + 1.5 mol/L urea (S3), 0.6 mol/L NaCl + 8 mol/L urea (S4), and 0.6 mol/L NaCl + 8 mol/L urea + 0.5 mol/L β-mercaptoethanol (S5). The supernatant was collected, and the protein concentration was determined using a BCA protein assay kit. The differences in protein concentration between S5 and S4, S4 and S3, S3 and S2, and S2 and S1 were used to evaluate the contributions of ionic bonds, hydrogen bonds, hydrophobic interactions, and disulfide bonds, respectively.

The surface hydrophobicity of the sample was determined using the bromophenol blue (BPB) binding method, as referenced by [17]. The measurement parameters were set as follows: excitation wavelength 390 nm, emission wavelength 470 nm, slit width 5 nm, and scan speed 600 nm/min. 2 g of the sample was placed in 10 times the volume of PBS buffer (20 mmol/L, pH 6) and homogenized using a high-speed disperser at 18,000 rpm. The protein concentration in the suspension was then measured and adjusted to 5 mg/mL. Additionally, 1 mL of the sample suspension and PBS (control) were mixed with 200 μL of BPB (1 mg/mL) by vortexing. After stirring at room temperature for a period of time, all samples and the control were centrifuged at 4 °C, 2000× g for 15 min. The supernatant was collected, diluted 10 times with PBS, and the absorbance was measured at 595 nm. The amount of BPB bound was calculated as follows (1):

In the formula, Ac and As represent the absorbance values of the control and the sample at 595 nm, respectively.

The color changes of the extrudates were measured using a colorimeter (Hunter Associate Lab, Reston, VA, USA). Samples were cut into blocks with dimensions of 20 × 20 × 10 mm, and the measurement parameters followed the procedure described in [18]. In the CIELAB color space, L* represents lightness, ranging from 0 (black) to 100 (white); a* and b* represent chromaticity, ranging from −128 to 127, where positive a* indicates red and negative a* indicates green, while positive b* indicates yellow and negative b* indicates blue. The total color difference (ΔE*) was calculated using the following Equation (2):

Changes in the protein secondary structure of the extruded samples were analyzed using Fourier transform infrared (FTIR) spectroscopy (Nicolet iS50; Thermo Fisher Scientific, Germany). The samples were mixed with potassium bromide (KBr) at a ratio of 1:100, and the spectra were recorded over the range of 500–4000 cm⁻¹. The amide I region (1600–1700 cm⁻¹) was used for secondary structure analysis, and peak fitting was conducted using PeakFit v4.12 software. Fourier self-deconvolution was applied to the spectra of extrudates containing different substitution levels of whole yeast powder (see Supplementary Figure S3).

All results were obtained from three independent experiments and expressed as mean ± standard deviation (SD). Statistical analysis was performed using analysis of variance (ANOVA) in SPSS Statistics 27.0 software (IBM Corp., Armonk, NY, USA), followed by Tukey’s honestly significant difference (HSD) test for multiple comparisons. A significance level of p < 0.05 was considered statistically significant. Data visualization was performed using Origin 2021, and the types of graphs created include line charts and bar charts.

Color is one of the core attributes influencing product attractiveness and value, shaping the product’s visual identity while also implicitly reflecting its nutritional quality and market positioning [19].

As shown in Table 2, the brightness (L* value) increased progressively from 37.89 ± 0.23 at 0% WYP to 49.16 ± 0.70 at the highest substitution level (40% WYP). This suggests that the addition of whole yeast powder and yeast protein powder contributed to a brighter appearance, likely due to their inherent color properties. For the a* value, only minor variations were observed among samples, ranging from 3.17 ± 0.14 to 3.91 ± 0.04, indicating that substitution had no significant effect on the red–green color axis. However, the b* value increased from 14.86 ± 0.45 at 0% WYP to 18.22 ± 0.02 at 40% WYP, indicating enhanced yellowness with increasing substitution levels. This result is likely attributed to the intrinsic yellowness of whole yeast and yeast protein powders, which impart a more yellow hue to the product. Moreover, the total color difference (ΔE) increased from 0.00 at 0% substitution to 11.80 ± 0.24 at 40% WYP, suggesting that the overall color difference became more pronounced with increasing substitution, particularly at higher levels. Previous studies have shown that color changes in plant-based meat analogues are closely related to Maillard reactions occurring within the extruder barrel during processing [20,21].

The addition of whole yeast powder and yeast protein powder increased the brightness (L* value) and enhanced the yellowness (b* value) of the product, while having minimal impact on the red–green contrast (a* value). The total color difference (ΔE) also significantly increased with higher substitution levels, particularly at higher levels (e.g., 40% YP), reaching perceptible differences (ΔE > 3).

Scanning electron microscopy (SEM) enables the detailed visualization of the fibrous structure and internal microarchitecture of high-moisture mixed-protein extrudates [13,22].

As shown in Figure 1, SEM images revealed that high-moisture extrudates composed solely of soy protein isolate (SPI) exhibited limited fibrous and interwoven structures in the absence of whole yeast powder (WYP) substitution [15]. When WYP substitution ranged from 5% to 20%, the formation of torn fibrous and interwoven structures progressively increased with rising WYP content. However, further substitution beyond 20% resulted in a decline: extrudates with 30% WYP showed fewer fibrous structures than those with 20%, and this reduction was more pronounced at 40% WYP. As shown in the binarized images (Figure 1), red-highlighted regions represent aligned microstructures with high edge responses, indicating the degree of fiber orientation and structural continuity. The control group (0% WYP) displayed scattered and irregular red areas, suggesting the absence of organized directional structures. As the WYP substitution level increased, particularly at 10% and 20%, the red regions became more continuous and vertically aligned, indicating the progressive development of fibrous protein networks. Notably, the 20% WYP sample exhibited the most prominent and organized vertical structures, confirming the highest degree of fiber alignment and structural organization. In contrast, the 20%YPP, 30%WYP, and 40%WYP groups showed disrupted and randomly oriented structures, with fragmented or curved patterns, indicating reduced fiber formation ability and less structural coherence. These results were validated through the texture tests below.

These findings indicated that whole yeast powder substitution between 5% and 20% increased fibrous interwoven structures in high-moisture mixed-protein extrudates, with the highest amount at 20%WYP. Adding more whole yeast powder beyond 20% disrupted the existing fibrous interwoven structure. Although 20%YPP extrudates showed some fibrous structure in both SEM images and visual observation [10], the 20%WYP extrudates exhibited more fibrous interwoven structures upon tearing, suggesting superior extrudability of 20%WYP over 20%YPP. A possible reason for this phenomenon is that under high pressure and shear forces in the twin-screw extruder, pure yeast protein powder undergoes denaturation, destroying its structure [23]. In contrast, whole yeast powder contains dietary fiber that does not denature, allowing more fibrous structures to remain in the mixed-protein extrudate. Similarly, ref. [14] reported that the incorporation of insoluble dietary fiber into soybean isolate protein during high-moisture extrusion promoted the formation of filamentous structures. In addition, the arrangement of these fiber structures follows the extrusion direction, which is similar to the experimental results reported by [24,25].

Texture is a critical characteristic of meat analogs, as it greatly influences their perceived mouthfeel and overall palatability. In this study, a texture analyzer was used to evaluate the impact of high-moisture extrusion parameters on the hardness, chewiness, and fibrousness of co-precipitated proteins. The degree of fiber formation was considered a key indicator of the fiber structure [15,22]. Numerous studies have indicated that the development of fibrous structures is strongly associated with protein–protein interactions during the extrusion process, particularly due to the oriented rearrangement of protein macromolecules [26,27].

As shown in Table 3, different whole yeast powder substitution levels exerted a pronounced effect on the products’ hardness and chewiness. As the WYP substitution ratio increased, both hardness and chewiness (measured in parallel and perpendicular directions) increased significantly, indicating a positive correlation between substitution level and textural properties. These findings are consistent with those reported by [10], in which yeast protein powder and soy protein were used in high-moisture extrusion.

According to Table 3, all samples exhibited a fibrous degree greater than 1, indicating the presence of fibrous structures in each sample [7]. Under low-to-moderate substitution ratios (5–20%WYP), the fibrous degree of the high-moisture mixed-protein extrudates increased with higher substitution levels, reaching a maximum (1.84 ± 0.02, p < 0.05) at 20%WYP. As the substitution level increased further (30–40%), it led to a considerable reduction in the fibrous degree, with the lowest value (1.69 ± 0.06, p < 0.05) observed at 40%WYP. This finding showed that an excessive amount of whole yeast powder could hinder the formation of fibrous structures [7]. A possible explanation is that the superior viscoelastic properties of the SPI–WYP and SPI–YPP mixtures may prolong their residence time within the extruder, the extended exposure to high shear forces and elevated temperatures promotes more extensive protein unfolding, thereby facilitating the development of stronger and more defined fibrous structures [28]. In most studies, the fibrous level of soybean protein-based meat analogs typically ranges from 0.7 to 1.7 [29]. However, in this study, when the whole yeast powder substitution level reached 20%, the fibrous level exceeded 1.7, reaching 1.84 ± 0.02. Moreover, the fibrous degree for whole yeast powder was higher than that for YPP (1.81 ± 0.04). Although there was no significant difference in fibrous degree between whole yeast powder and yeast protein powder, the extrudates containing whole yeast powder showed greater hardness and chewiness than those containing yeast protein powder. The observed difference may result from the dietary fiber content in whole yeast powder, whereas yeast protein powder is composed predominantly of purified protein and lacks fiber. Additionally, SEM images indicated that the fibrous structures formed by whole yeast powder were more orderly and complete than those formed by yeast protein powder, displaying a smoother appearance in the parallel direction. This finding suggests that 20%WYP achieved better extrudability than 20%YPP, indicating that whole yeast powder substitution promoted the formation of fibrous structures in meat analogs.

Low-field nuclear magnetic resonance (LF-NMR) is widely employed to assess the distribution and physical state of water in food systems. The T₂ relaxation time reflects the mobility of water molecules, typically ranging from 1 to 10 ms for bound water, 10 to 100 ms for immobilized water, and 100 to 1000 ms for free water [15]. A longer T₂ relaxation time indicates greater water mobility, whereas a shorter T₂ suggests restricted molecular movement [30,31]. Similarly, longer T₂ values are associated with higher water activity, while shorter values indicate lower water activity levels [32].

As shown in Figure 2, the T₂ relaxation time of immobilized water significantly decreased at low-to-moderate WYP substitution levels (5–20%), suggesting enhanced water entrapment within the protein matrix and stronger interactions between the protein network and water molecules [33]. These results indicate that moderate WYP incorporation effectively enhances water-holding capacity by stabilizing water in a semi-bound state. However, when the substitution level exceeded 20%, the T₂ relaxation time of immobilized water increased, indicating a reduced ability of the protein matrix to retain water, possibly due to structural disruption or excessive phase separation at higher WYP levels [33]. In comparison, the T₂ relaxation time of bound water showed a slight decrease at lower substitution levels, followed by a moderate increase beyond 20%WYP, although the changes were less significant than those observed for immobilized water. In addition, the 0%WYP sample exhibited a very weak capacity to bind water molecules. T₂ decreased as the whole yeast powder content increased from 5% to 20%, which may suggest a partial transition from immobilized to bound water and more intimate interactions between water and protein molecules. The reason for this phenomenon may be that the incorporation of whole yeast powder leads to a more compact network structure. Ref. [34] reached similar conclusions in their study of lemon juice gels used as 3D printing food materials.

Moreover, the T₂ relaxation times of free water, immobilized water, and bound water in the 20%WYP extrudate were shorter than those in the 20%YPP extrudate, indicating a stronger water-binding capacity in the sample [33]. A higher degree of water binding typically reflects a denser microstructure and greater fibrousness, which aligns with the SEM observations and further confirms that the extrudability of 20%WYP was superior to that of 20%YPP. The cause may be that the addition of yeast protein promoted protein molecule aggregation and induced the formation of a protein network structure with enhanced water-binding ability [35,36]. This effect may be attributed to the reduction in pore size between SPI and yeast protein, resulting in a denser structure. By enhancing the interactions between water and protein molecules, the whole yeast powder rendered the extrudates’ structure more compact. The improved structural integrity influenced the water distribution in the meat analogs, as a more compact structure led to reduced water loss during extrusion. The 20%WYP samples exhibited the most pronounced fibrous characteristics, and with texture data showing the highest fibrousness (1.84 ± 0.02, p < 0.05) at 20%WYP substitution. Moreover, greater water-holding capacity resulted in lower water mobility [37].

Intermolecular interactions in the extrudates were assessed by evaluating changes in protein solubility across various selective extraction solutions [26,38].

As shown in Figure 3a, at low-to-moderate WYP substitution levels (5–20%), the ionic bond content gradually decreased with increasing WYP levels. At higher substitution levels (30–40%), further increases in WYP content led to a significant decline in ionic bond content. As shown in Figure 3b, the hydrogen bond content significantly decreased following the addition of WYP.

As shown in Figure 3c, after adding whole yeast powder, the hydrophobic interactions in all mixed-protein samples significantly increased. The cause may be that yeast protein itself contains a certain proportion of hydrophobic amino acids, which further contributes to enhancing hydrophobic interactions [39,40]. However, at 40%WYP substitution, hydrophobic interactions significantly decreased. This suggested an overall weakening of protein aggregation stability.

As shown in Figure 3d, after the addition of whole yeast powder, the disulfide bond content in the mixed-protein samples significantly increased, peaking at 20% WYP, consistent with the trend in Figure 3c. The increased disulfide bonds enhanced interactions between the two proteins, leading to tighter water binding at the microstructural level and resulting in fibrous aggregates [41]. The disulfide bond content of the 20%WYP sample (95.33 ± 0.92, p < 0.05) was higher than that of the 20%YPP sample (78.41 ± 0.78, p < 0.05), indicating that the extrudability of 20%WYP was superior to that of 20%YPP. The solubility of the extruded samples in different extraction solutions can be attributed to the disruption of specific intermolecular interactions. These results indicate that the overall structure of the protein extrudates is stabilized by a synergistic combination of multiple types of chemical interactions [42].

Hydrophobic interactions help maintain the protein structure and also affect the stability and functional properties of myofibrillar proteins. The surface hydrophobicity of the protein can be determined by the binding capacity of the hydrophobic chromophore bromophenol blue (BPB) [17]. In globular proteins, surface hydrophobicity reflects the extent of molecular unfolding, wherein previously buried nonpolar amino acid residues become exposed. An increase in surface hydrophobicity indicates the exposure of internal hydrophobic domains and potential denaturation of protein monomers [43].

Figure 4 shows that as the whole yeast powder substitution level increased, the surface hydrophobicity of high-moisture mixed-protein extrudates significantly increased. This phenomenon is attributed to interactions between yeast protein and soybean protein, triggering protein structural reorganization and exposing hydrophobic groups initially buried within the structure [6]. Thus, whole yeast powder content positively correlated with the surface hydrophobicity of the samples. Specifically, as the substitution rate of whole yeast powder (WYP) increases from 5% to 40%, the surface hydrophobicity gradually increases. The 5% WYP, 20% YPP, and 40% WYP groups exhibit similar levels of hydrophobicity, while the 20% and 30% WYP groups reach a similar hydrophobicity threshold. Notably, the surface hydrophobicity peaked at a 20% WYP substitution, with a value of (29.51 ± 0.28, p < 0.05), significantly higher than that of the 20% YPP sample (26.31 ± 0.30, p < 0.05). These findings suggest that introducing whole yeast powder enhances network formation driven by hydrophobic interactions among proteins, demonstrating dose-dependent structural optimization and maximizing hydrophobic group exposure within a specific substitution range [44].

Both surface hydrophobicity and hydrophobic interactions were significantly enhanced at a moderate substitution level (20%WYP), indicating that the introduction of yeast protein promoted the exposure and aggregation of hydrophobic groups to some extent, thus facilitating protein rearrangement and fibrous structure formation during extrusion. However, excessive substitution (e.g., 40%WYP) increased surface hydrophobicity but significantly reduced hydrophobic interactions, possibly due to protein conformational instability or structural incompatibility, thereby hindering hydrophobic-driven protein network formation [10].

The secondary structure of proteins in the extrudate samples was evaluated by analyzing the characteristic absorption peaks in the amide I region (1600–1700 cm⁻¹) of the FTIR spectra. This region primarily originates from C=O stretching vibrations, accompanied by C–N stretching, C–N deformation, and N–H bending within the protein backbone. These vibrational modes provide both qualitative and quantitative insights into secondary structure elements, including α-helices, β-sheets, β-turns, and random coils [45].

As illustrated in Figure 5, the primary absorption peak of SPI was observed at 1654 cm⁻¹, with additional peaks detected at 1530 cm⁻¹ and 1234 cm⁻¹, corresponding to C=O stretching, N–H bending, C–N stretching, and N–N bending vibrations, respectively. With the incorporation of whole yeast powder, increased damping of the C–H stretching vibrations in the 2800–3000 cm⁻¹ region was observed, along with an intensified absorption peak at 1530 cm⁻¹. This enhancement may be attributed to the introduction of additional C=O groups through the addition of whole yeast powder, resulting in stronger absorption in this region. A similar observation was reported by [46] when β-glucan was incorporated into SPI during high-moisture extrusion.

As presented in Table 4, at low-to-moderate substitution levels (0% WYP–20% WYP), the β-sheet content significantly increased (p < 0.05) with increasing whole yeast powder substitution, reaching its peak at 20% WYP (0.267 ± 0.003). Additionally, the β-sheet content of 20% WYP was significantly higher than that of 20% YPP (0.267 ± 0.003 > 0.260 ± 0.003, p < 0.05). At higher substitution levels (30% WYP–40% WYP), the β-sheet content tended to stabilize, remaining consistently higher than the control group (0% WYP). A similar increasing trend was also observed for antiparallel β-sheet structures, which reached their maximum at 20% WYP (0.123 ± 0.004, p < 0.05). The content of antiparallel β-sheets at 20% WYP was higher than that of 20% YPP (0.123 ± 0.004 > 0.119 ± 0.003, p < 0.05). This upward trend is consistent with the findings reported by [47], who suggested that the formation of antiparallel β-sheets may be associated with the development of fibrous structures in extruded protein systems. Overall, the β-sheet content exhibited an increasing trend (p < 0.05) with rising WYP substitution levels, consistent with previous reports on structural transitions in extruded protein systems [10]. In contrast, no notable changes (p > 0.05) were observed in α-helix content across the different substitution levels, which is also in agreement with earlier findings [10], indicating that the dominant conformational shift occurred primarily in the β-sheet domains. The increase in β-sheet structures at low-to-moderate substitution levels contributed to fibrous protein structure formation and strengthened protein–protein interactions. These findings align with [48], who noted that increased β-sheet content facilitated fiber formation. The growth in β-sheet structures is closely associated with protein aggregation. During high-moisture extrusion, proteins undergo deformation and aggregate, ultimately forming aligned fibrous structures in plant-based meat analogues [49]. This structural transition indicates that β-sheet formation may be the key driver of protein alignment and anisotropic texture development in high-moisture extruded systems.

The original contributions presented in the study are included in the article, further inquiries can be directed to the corresponding author.