Swapping Peas for Mung Beans Reshapes the Microbes and Enzymes of Baijiu Starter

Replacing peas with mung beans in the grain mix used to make medium-temperature Daqu, the fermentation starter at the heart of much Chinese Baijiu production, substantially changed the starter’s microbial community, protein expression and enzymatic performance, according to a multi-omics study published in Food Chemistry: X. In the commercial Daqu made with mung bean substitution, saccharifying power rose by 25.38 percent, from 390.00 ± 21.63 to 489.00 ± 12.00 mg/g·h, and fermenting power jumped by 117.78 percent, from 0.45 ± 0.03 to 0.98 ± 0.04 g CO₂/g·72 h. At the same time, liquefying power fell from 1.55 ± 0.03 to 1.13 ± 0.01 g/g·h and esterifying power declined from 666.92 ± 18.18 to 595.55 ± 20.91 mg/5 g·7 d, showing that the substitution reshaped functionality in both directions rather than simply improving it.

Daqu is a solid-state fermentation starter that supplies microorganisms, hydrolases and metabolites to subsequent Baijiu fermentation, and the quality of the final spirit is closely tied to its enzymatic properties, particularly saccharifying and fermenting power. Medium-temperature Daqu, produced with maximum incubation temperatures of 50–60 °C, is the most widely used of the three Daqu classes, which also include high-temperature (60–70 °C) and low-temperature (40–50 °C) variants. It is manufactured in an open fermentation system from non-sterile raw materials: wheat and barley mainly supply starch for microbial growth, while peas contribute protein and nitrogen sources associated with microbial metabolism and functional enzyme formation. Mung bean Daqu is made by substituting mung bean for pea within the traditional process and is used for mung bean-flavor Baijiu, which gave the researchers a practical reason to investigate how the swap alters the starter’s function.

The samples came from a commercial Baijiu distillery in Anhui Province. Traditional medium-temperature Daqu without mung bean (NMBQ) was prepared from 85 percent wheat, 10 percent barley and 5 percent peas. In the mung bean Daqu (MBQ), peas were replaced with mung bean at the same proportion, with the rest of the formulation and production conditions unchanged to allow a direct comparison. For each type, three Daqu bricks were randomly collected from the same production batch, ground, pooled and homogenized into one composite biological sample, a procedure repeated three times to yield three composite samples per type, drawn from 18 bricks in total. The authors note that because all biological replicates came from a single production batch, potential batch effects cannot be fully separated from the observed differences, so the community differences should be interpreted as strongly associated with mung bean substitution rather than attributed exclusively to it.

Physicochemical measurements showed that MBQ had lower moisture content than NMBQ (p < 0.01) and significantly higher total acidity (p < 0.05), two environmental factors closely linked to microbial growth, metabolic activity and extracellular enzyme activity during Daqu fermentation. Enzymatic assays conducted according to the Chinese industry standard QB/T 4257–2011 then revealed the divergent functional profile: higher saccharifying and fermenting powers in MBQ, but lower liquefying and esterifying powers. Since most enzymes in Daqu derive from its microbial community, the researchers turned to metabolomics, metagenomics and metaproteomics to trace how the raw material change propagated through community composition, genetic potential and actual protein expression.

Untargeted LC–MS/MS metabolomics generated 3037 and 2474 ion features in positive and negative ionization modes respectively. Principal component analysis separated MBQ from NMBQ clearly, with PC1 and PC2 explaining 36.70 percent and 9.66 percent of total variance, and Adonis analysis attributed a substantial share of metabolic variation to the substitution (R² = 0.6568, P = 0.001). Screening with a variable importance in projection score above 1 and FDR-adjusted p < 0.05 identified 1591 differential metabolites, dominated by carboxylic acids and derivatives (22.97 percent), organooxygen compounds (9.88 percent), fatty acyls (7.78 percent) and prenol lipids (7.41 percent). Differential pathway enrichment showed that MBQ was enriched in flavone and flavonol biosynthesis — the strongest differential enrichment — along with biosynthesis of plant secondary metabolites, the one-carbon pool by folate, folate biosynthesis and cofactor biosynthesis, whereas NMBQ was relatively enriched in amino acid metabolism pathways including tryptophan, alanine-aspartate-glutamate and tyrosine metabolism.

Metagenomic sequencing predicted 5,972,468 genes with a cumulative length of 2,390,527,818 bp; after redundancy removal, 1,307,657 non-redundant genes remained. Microbial diversity differed significantly between the two starters: MBQ showed a higher Chao1 richness index but a lower Shannon index (both p < 0.01), indicating greater richness but lower evenness, possibly due to dominance of specific taxa. Principal coordinate analysis based on Bray–Curtis distances separated the two types at the genus level, with PC1 explaining 93.63 percent of variation, and ADONIS confirmed significant community-structure differences (R² = 0.936, p = 0.003). Ascomycota, Mucoromycota and Bacillota dominated both starters at the phylum level, and the dominant genera — Aspergillus, Rhizopus, Lichtheimia, Paecilomyces, Rasamsonia and Bacillus — collectively accounted for more than 53.28 percent in NMBQ and 66.98 percent in MBQ.

The clearest taxonomic shifts emerged among the secondary genera. Pichia (8.50 percent), Lactiplantibacillus (6.41 percent) and Weissella (5.17 percent) were relatively enriched in NMBQ, while Saccharopolyspora (7.33 percent) was specifically enriched in MBQ. LEfSe analysis with an LDA threshold above 3.0 identified Saccharopolyspora, Thermoactinomyces, Staphylococcus and Leuconostoc as bacterial markers of MBQ, whereas NMBQ carried more lactic acid bacteria, including Lactiplantibacillus, Weissella, Levilactobacillus, Companilactobacillus and Pediococcus, together with Acetobacter and Streptomyces. Among fungi, MBQ was enriched in filamentous genera such as Aspergillus, Paecilomyces, Rasamsonia, Lichtheimia and Talaromyces, commonly associated with extracellular enzyme production and degradation of starch, polysaccharides and proteins, while Pichia, Monascus and Wickerhamomyces marked NMBQ. Redundancy analysis associated the MBQ marker genera with its higher saccharifying and fermenting powers and the NMBQ markers with its higher liquefying and esterifying powers, though the authors stress that these direct contributions require further experimental validation.

Functional annotation revealed a shift in carbohydrate-active enzyme potential. Glycoside hydrolases were significantly more abundant in MBQ, while glycosyltransferases were higher in NMBQ (p < 0.01), and key starch-degrading subfamilies GH13 and GH15 were enriched in MBQ. At KEGG level 3, MBQ was enriched in amino acid degradation pathways, fatty acid degradation, glyoxylate and dicarboxylate metabolism and benzoate degradation, whereas NMBQ was enriched in transport and cellular process pathways, including ABC transporters, ribosome, two-component system and the phosphotransferase system. Interestingly, some enrichment patterns diverged between data layers: tryptophan metabolism was enriched in MBQ at the gene level but in NMBQ at the metabolite level, and cofactor biosynthesis showed the opposite pattern. The authors attribute such discrepancies to the difference between genetic potential and actual metabolic output, which depends on protein expression, enzyme activity, substrate availability and cross-feeding within the community.

Metaproteomics, which searches detected peptides against a reference database built from the metagenomic results, identified 7124 microbial proteins, with plant-taxa proteins excluded. Differential analysis using fold change thresholds of at least 1.5 or at most 0.67 with p < 0.05 found 1983 differentially expressed proteins — 598 upregulated and 1385 downregulated in MBQ. KEGG enrichment of these proteins highlighted glycolysis/gluconeogenesis, pyruvate metabolism, and starch and sucrose metabolism (p < 0.05). Within the starch degradation pathway, nine differential proteins were annotated as α-amylase, six of them significantly downregulated in MBQ, accounting for 67.39 percent of the total relative abundance of that differential protein subset — a pattern that may partly explain MBQ’s lower liquefying power. In contrast, proteins annotated as glucoamylase and β-fructofuranosidase were significantly upregulated, providing direct enzymatic evidence for enhanced release of fermentable sugars from dextrins and sucrose, consistent with MBQ’s higher saccharifying power.

Downstream carbon metabolism showed a coherent redirection. All differential proteins annotated as L-lactate dehydrogenase and pyruvate dehydrogenase were downregulated in MBQ, with fold changes of 0.11–0.24 and at most 0.36 respectively, while differential proteins for pyruvate decarboxylase and alcohol dehydrogenase included both upregulated and downregulated members, with the most abundant differential proteins showing fold changes of 4.40 and 2.46 and representing 57.28 percent and 53.02 percent of their respective subsets. This suggests pyruvate in MBQ was preferentially channeled toward acetaldehyde and ethanol formation rather than lactate and acetyl-CoA, plausibly contributing to the 117.78 percent increase in fermenting power. The enhanced fermenting power, however, was not matched by esterifying power: no differential proteins were directly annotated as esterification enzymes, and although several fatty acid metabolism proteins were downregulated, medium-chain fatty acids did not differ significantly and acyl-CoA intermediates fell outside the untargeted metabolomic method’s coverage, leaving the mechanism behind the reduced esterifying power unresolved. The authors conclude that mung bean substitution may suit production goals favoring saccharification and fermentation, but the reduced liquefying and esterifying powers mean it should be tailored to specific aims, and independent multi-batch trials across different seasons and Qu-rooms are needed to confirm reproducibility before broader industrial application.

Subject of Research: Multi-omics analysis of how substituting mung bean for pea alters the microbial community, protein expression and enzymatic properties of medium-temperature Daqu for Baijiu production.

Article Title: Multi-omics insights into microbial community and functional expression in medium-temperature Daqu following mung bean substitution

Article References: Wang, R., Yan, L., Ding, F., Zhou, H., Xu, B., Xu, S., Yuan, Y., Wu, X., & Li, X. (2026). Multi-omics insights into microbial community and functional expression in medium-temperature Daqu following mung bean substitution. Food Chemistry: X, 40, Article 104680. https://doi.org/10.1016/j.fochx.2026.104680

Image Credits: AI Generated

DOI: 10.1016/j.fochx.2026.104680

Keywords: Daqu, Baijiu, mung bean, metaproteomics, metagenomics, metabolomics, saccharifying power, fermenting power, microbial community, food fermentation, enzyme expression, Food Chemistry: X