Effects of Fermentation with Tetragenococcus halophilus and Zygosaccharomyces rouxii on the Volatile Profiles of Soybean Protein Hydrolysates

Abstract

The effects of fermentation with lactic acid bacteria (LAB) and yeast on the aroma of samples were analyzed in this work. The volatile features of different soybean hydrolysates were investigated using both GC-MS and GC-IMS. Only 47 volatile flavor compounds (VFCs) were detected when using GC-IMS, while a combination of GC-MS and GC-IMS resulted in the identification of 150 compounds. LAB-yeast fermentation could significantly increase the diversity and concentrations of VFCs (p < 0.05), including alcohols, acids, esters, and sulfurs, while reduce the contents of aldehydes and ketones. Hierarchical clustering and orthogonal partial least squares analyses confirmed the impact of fermentation on the VFCs of the hydrolysates. Seven compounds were identified as significant compounds distinguishing the aromas of different groups. The partial least squares regression analysis of the 25 key VFCs (ROAV > 1) and sensory results revealed that the treatment groups positively correlated with aromatic, caramel, sour, overall aroma, and most of the key VFCs. In summary, fermentation effectively reduced the fatty and bean-like flavors of soybean hydrolysates, enhancing the overall flavor quality, with sequential inoculation proving to be more effective than simultaneous inoculation. These findings provided a theoretical basis for improving and assessing the flavor of soybean protein hydrolysates.

Keywords: GC-IMS; GC-MS; lactic acid bacteria; soybean protein hydrolysates; volatile flavor compounds; yeast.

Figure 1

Volatile flavor compounds in the soybean protein hydrolysates obtained via different processes (A) Venn diagram; (B) cluster heatmap of different categories of volatile flavor compounds based on the GC-MS data after standardization. Darker red and darker blue, respectively, indicate higher and lower concentrations of volatile flavor compounds; (C) changes in the relative contents of volatile flavor compounds.

Figure 2

Heat map and hierarchical cluster analysis of the volatile flavor compounds in different soybean hydrolysate samples based on the GC-MS data after standardization. Darker red and darker blue respectively indicate higher and lower concentrations of volatile flavor compounds.

Figure 3

Topographic plots of soybean hydrolysate samples. Each colored dot on the spectrum represents a volatile compound, with red color indicating a high intensity and blue color indicating a low intensity. (A) 3D-topographic features of the volatile compounds (within the blue background, the red vertical lines on the left represent the reactive ion peaks); (B) 2D-topographic features of the volatile compounds; (C) 2D-topographic plot with a difference comparison model. Red and blue dots indicated that the concentrations of the compounds were higher or lower than the reference, respectively. The red vertical line at the abscissa 1.0 represents the RIP peak (Reaction Ion Peak, normalized); (D) Gallery plot fingerprints of the volatile compounds in soybean hydrolysate samples. Each dot represents a volatile flavor compound, with the intensity of its color indicating its concentration.

Figure 3

Topographic plots of soybean hydrolysate samples. Each colored dot on the spectrum represents a volatile compound, with red color indicating a high intensity and blue color indicating a low intensity. (A) 3D-topographic features of the volatile compounds (within the blue background, the red vertical lines on the left represent the reactive ion peaks); (B) 2D-topographic features of the volatile compounds; (C) 2D-topographic plot with a difference comparison model. Red and blue dots indicated that the concentrations of the compounds were higher or lower than the reference, respectively. The red vertical line at the abscissa 1.0 represents the RIP peak (Reaction Ion Peak, normalized); (D) Gallery plot fingerprints of the volatile compounds in soybean hydrolysate samples. Each dot represents a volatile flavor compound, with the intensity of its color indicating its concentration.

Figure 4

Orthogonal partial least squares analyses (OPLS-DAs) of the volatile compounds in soybean hydrolysate samples based on the GC-MS and GC-IMS data. The three OPLS-DA models are as follows: (A,B) SH-0 and SH-1; (C,D) SH-0 and SH-2; (E,F) SH-1 and SH-2. (A,C,E) represent the score plots, while (B,D,F) illustrate the variable importance in projection values for the respective models. A1, A2, and A3; B1, B2, and B3; and C1, C2, and C3 denote the three replicates of the SH-0, SH-1, and SH-2 groups, respectively. The Var ID numbers are consistent with the compound numbers in Tables S1 and S2.

Figure 4

Orthogonal partial least squares analyses (OPLS-DAs) of the volatile compounds in soybean hydrolysate samples based on the GC-MS and GC-IMS data. The three OPLS-DA models are as follows: (A,B) SH-0 and SH-1; (C,D) SH-0 and SH-2; (E,F) SH-1 and SH-2. (A,C,E) represent the score plots, while (B,D,F) illustrate the variable importance in projection values for the respective models. A1, A2, and A3; B1, B2, and B3; and C1, C2, and C3 denote the three replicates of the SH-0, SH-1, and SH-2 groups, respectively. The Var ID numbers are consistent with the compound numbers in Tables S1 and S2.

Figure 4

Orthogonal partial least squares analyses (OPLS-DAs) of the volatile compounds in soybean hydrolysate samples based on the GC-MS and GC-IMS data. The three OPLS-DA models are as follows: (A,B) SH-0 and SH-1; (C,D) SH-0 and SH-2; (E,F) SH-1 and SH-2. (A,C,E) represent the score plots, while (B,D,F) illustrate the variable importance in projection values for the respective models. A1, A2, and A3; B1, B2, and B3; and C1, C2, and C3 denote the three replicates of the SH-0, SH-1, and SH-2 groups, respectively. The Var ID numbers are consistent with the compound numbers in Tables S1 and S2.

Figure 5

Partial least squares regression (PLSR) analyses of key volatile compounds. (A) Cross validation diagram (n = 200); (B) Biplot loadings; (C) variable importance in projection values; (D) observation and prediction diagram. A1, A2, and A3; B1, B2, and B3; and C1, C2, and C3 denote the three replicates of the SH-0, SH-1, and SH-2 groups, respectively. The Var ID numbers are consistent with the compound numbers in Table 1.

Figure 5

Partial least squares regression (PLSR) analyses of key volatile compounds. (A) Cross validation diagram (n = 200); (B) Biplot loadings; (C) variable importance in projection values; (D) observation and prediction diagram. A1, A2, and A3; B1, B2, and B3; and C1, C2, and C3 denote the three replicates of the SH-0, SH-1, and SH-2 groups, respectively. The Var ID numbers are consistent with the compound numbers in Table 1.