Integrated Metabolomics and Volatolomics for Comparative Evaluation of Fermented Soy Products

Abstract

Though varying metabolomes are believed to influence distinctive characteristics of different soy foods, an in-depth, comprehensive analysis of both soluble and volatile metabolites is largely unreported. The metabolite profiles of different soy products, including cheonggukjang, meju, doenjang, and raw soybean, were characterized using LC-MS (liquid chromatography-mass spectrometry), GC-MS (gas chromatography-mass spectrometry), and headspace solid-phase microextraction (HS-SPME) GC-MS. Principal component analysis (PCA) showed that the datasets for the cheonggukjang, meju, and doenjang extracts were distinguished from the non-fermented soybean across PC1, while those for cheonggukjang and doenjang were separated across PC2. Volatile organic compound (VOC) profiles were clearly distinct among doenjang and soybean, cheonggukjang, and meju samples. Notably, the relative contents of the isoflavone glycosides and DDMP (2,3-dihydro-2,5-dihydroxy-6-methyl-4H-pyran-4-one) conjugated soyasaponins were higher in soybean and cheonggukjang, compared to doenjang, while the isoflavone aglycones, non-DDMP conjugated soyasaponins, and amino acids were significantly higher in doenjang. Most VOCs, including the sulfur containing compounds aldehydes, esters, and furans, were relatively abundant in doenjang. However, pyrazines, 3-methylbutanoic acid, maltol, and methoxyphenol were higher in cheonggukjang, which contributed to the characteristic aroma of soy foods. We believe that this study provides the fundamental insights on soy food metabolomes, which determine their nutritional, functional, organoleptic, and aroma characteristics.

Keywords: fermented soy product; metabolic pathway; metabolomics; volatolomics.

Figure 1

Manufacturing steps during the fermentative manufacturing of cheonggukjang, meju, and doenjang from raw soybean.

Figure 2

PCA and PLS-DA score plots, derived from the (A,B) GC-TOF-MS; (C,D) UHPLC-LTQ-Orbitrap-MS dataset; and (E,F) HS-SPME/GC-TOF-MS. The datasets are indicated - ; soy, - ; cheonggukjang, ; meju, ; doenjang.

Figure 3

Schematics showing the relative abundance of the significantly discriminant metabolites (primary and secondary metabolites, as well as VOCs), detected comprehensively from the fermented soy foods (cheonggukjang, meju, and doenjang) and unfermented soybean. The metabolites are shown in the form of biosynthetic pathways, adapted from the Kyoto Encyclopedia of Genes and Genomes (KEGG) database, relatively higher in cheonggukjang. Nucleotide compounds were relatively more abundant in cheonggukjang.

Figure 4

Total flavonoid, phenolic contents, and antioxidant activity tests. (A) 2,2-azino-bis-(3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt (ABTS) and (B) ferric reducing antioxidant power (FRAP), as well as (C) total flavonoid contents (TFC) and (D) total phenolic contents (TFC). S, soy; C, cheonggukjang; M, meju; D, doenjang. Values are expressed as the average of three biological replicates (n = 3). Bar graphs denoted by the same letter were not significantly different, according to Duncan’s multiple range test (p-value < 0.05).

Figure 5

Heat map representation of correlation analysis between relative abundance of significantly discriminant metabolites and antioxidant activity (ABTS and FRAP), total flavonoid content (TFC), and total phenolic content (TPC). Each square indicates Pearson’s correlation coefficient values (r). Red and blue represent positive (0 < r < 1) and negative (−1 < r < 0) correlation, respectively. *; p-value less than 0.05, according to Duncan’s multiple range test.