Reveal the microbial communities and functional prediction during the fermentation of Fen-flavor Baijiu via metagenome combining amplicon sequencing

Abstract

Purpose

Microbial resources are abundant in fermented grains of the Chinese Fen-flavor Baijiu, which is closely related to the quality of Baijiu. The purpose of this study was to investigate the microbial community structure and function in Daqu and fermented grains.

Methods

We systematically compared two technical approaches, amplicon sequencing, and metagenomic sequencing, to analyze the microbial communities during Baijiu fermentation.

Result

The results showed that lactic acid bacteria (LAB) and yeasts were the main microorganisms in the fermentation process. Firmicutes (Lactobacillus, Pediococcus, and Weissella) were the dominant bacteria, and Ascomycota (Issatchenkia or Pichia) was the dominant fungus in fermented grains. Moreover, Pichia kudriavzevii, Lichtheimia ramosa, and Companilactobacillus paralimentarius were the dominant species at the initial stage of fermentation by metagenomic sequencing. Latilactobacillus curvatus, Loigolactobacillus coryniformis subsp. coryniformis, and Lentilactobacillus parabuchneri became dominant during the middle stage of fermentation. Lentilactobacillus parabuchneri and Lactobacillus acetotolerans were the dominant species in the final stage of fermentation. Spearman correlation analysis showed that LAB inhibited the growth of yeasts.

Conclusion

Combining the two sequencing methods provided valuable insights into the dynamic succession of microorganisms during the fermentation of Baijiu. It had had a particular significance for mining microbial species resources in fermented grains.

Introduction

Chinese traditional Baijiu is one of the oldest distilled spirits in the world (Xu et al. 2017). Baijiu can be divided into 12 types based on aroma characteristics (Xu et al. 2022), and each type of Baijiu is produced by its unique processing technology and complex fermentation system. The Fen flavor (light flavor) is one of the basic flavor types of Baijiu (Hong et al. 2020; Xu et al. 2017). Fen-flavor Baijiu is distilled after spontaneous fermentation of fermented grains (Jiupei) under a specific environment and unique process, and ethyl acetate is the hallmark aroma component (Wu et al. 2021). Fermented grain is formed by mixing Daqu and sorghum in a certain proportion. Daqu is a unique multi-strain starter for Chinese Baijiu, produced by solid-state fermentation from barley and pea (Huang et al. 2020a, b; Zheng et al. 2012). It not only provides functional microorganisms for the fermentation of Baijiu but also provides a variety of enzymes for the saccharification and fermentation of fermented grains and participates in the metabolism of flavor substances (Wang et al. 2020; Hou et al. 2022; Y. Huang et al. 2017a, b). Therefore, Daqu is being exploited as a new microbial resource (Huang et al. 2017a, b). In fermentation, the microbial communities in fermented grains adapt to the environment, continuously undergo dynamic succession, decompose raw materials during growth and metabolism and generate a variety of compounds in the fermentation system (Kong et al. 2014; Pang et al. 2020). At the same time, raw materials regulate flavor formation by driving microorganisms in Baijiu fermentation (Liu et al. 2019a, b). Therefore, the composition and succession of microbial communities in the fermentation system of Baijiu are directly related to the quality of Baijiu. Studying microbial community structure in Daqu and fermented grains has been a critical topic in Baijiu research.

The emergence of next-generation sequencing (NGS) technology has been used to explore microbial structure in a variety of sample systems. It can predict microbial function through gene annotation (Wensel et al. 2022), which makes up for the limitations of traditional culture methods in the study of microbial community structure (He et al. 2017). Amplicon sequencing and metagenomic sequencing are the most widely used next-generation sequencing technologies (Chakravorty et al. 2007; Knight et al. 2018), which have been used to study microorganisms that are closely related to the quality of Chinese Baijiu, including Daqu (Zhang et al. 2022); fermentation pit muds of Baijiu (Xu et al. 2020); fermented grains (Wang et al. 2022); and fermentation environment (Ma et al. 2022). At present, some studies have used amplicons sequencing and metagenomic sequencing to dissect microorganisms in the fermentation of Fen-flavor Baijiu (Huang et al. 2020a, b; Kang et al. 2022a). The two sequencing methods have similar results in the analysis of microbial diversity (Zuo et al. 2022). However, amplicon sequencing can only obtain structures for bacteria or fungi, respectively (Davis et al., 2016). Metagenome sequencing can understand the structure of microorganisms (bacteria and fungi) in a sample and identify microbial species (species or subspecies) with higher accuracy and predict their functions (Knight et al. 2018; Zhao & Eun 2020). The combination of amplicon and metagenomic sequencing can better interpret microbial community structure and function.

In this study, amplicon sequencing and metagenome sequencing were combined to reveal the microbial diversity succession, sample differential microorganisms, symbiosis, and potential metabolic functions during the fermentation of Fen-flavor Baijiu. This study provided information for exploring and utilizing microbial resources in fermented grains of traditional Chinese Baijiu. Then, the correlation between microorganisms and metabolic components in Baijiu could be studied, which was of great significance for the improve quality of Baijiu.

Materials and methods

Sampling

Daqu and fermented grains of Baijiu were collected from Shanxi Fenjiu Distillery (Fenyang, Shanxi, China). All samples, including Daqu and fermented grains, were from the same production batch. Samples were collected from Daqu, early, middle, and late stages of fermentation, i.e., at 0, 7, 15, and 28 days, the pH values were 6.65, 5.21, 4.67, 3.80 and 3.78, respectively. Sample collection and processing methods were as follows: firstly, samples obtained at different time points were from the same earthenware jars. Secondly, positions were fixed for each sampling, and samples were quantitatively collected at 50 cm below the surface of the fermented grains and mixed well. Next, the samples were transported back to the laboratory and refrigerated at − 80 °C. The Daqu sample and the fermented grains samples were divided into three samples for parallel testing, and they were labelled as JQ, JP0, JP7, JP15, and JP28, respectively.

DNA extraction and inspection

The genomic DNA of the samples was extracted by the CTAB method, as previously described (Zhang et al. 2022). The quality of genomic DNA was assessed with 1% agarose gel. DNA concentration was measured using Qubit® dsDNA Assay Kit in Qubit® 2.0 Fluorometer (Life Technologies, CA, USA). OD260/OD280 value was between 1.8 ~ 2.0; DNA contents above 1 μg are used to construct the library. Amplicon and metagenomic sequencing were completed at Beijing Novogene, Beijing, China.

Amplicon sequencing

PCR amplification and sequencing

Bacterial 16S rDNA genes were amplified by PCR using primers 341F/806R with barcode (Liu et al. 2019a, b), and fungal ITS genes were amplified using primers ITS1F/ITS2 (Li et al. 2020). The diluted genomic DNA was used as a template. PCR was performed with Phusion® High-Fidelity PCR Master Mix with GC buffer and high-efficiency high-fidelity enzyme (New England Biolabs, Beijing, China) to ensure amplification efficiency and accuracy. The PCR products were detected by electrophoresis in 2% agarose gel, and the samples were mixed in equal amounts according to the concentration of PCR products. After thoroughly mixing, the samples were detected again by electrophoresis in 2% agarose gel, and the target bands were recovered by gel recovery kit (Qiagen, Hilden, Germany). NEBNext® Ultra™ II DNA Library Prep Kit was used to construct the library, and the constructed library was subjected to Qubit and Q-PCR quantification. The qualified libraries were sequenced using NovaSeq 6000.

Sequencing data processing and analysis

Each sample data was split from the sequencing data, and the reads of the sample were spliced to obtain raw tags and quality control to obtain high-quality clean tags. Chimera filtering was performed on the clean tags to obtain effective tags that could be used for subsequent analysis. The effective tags were denoised by the DADA2 module in QIIME2 software, and sequences with an abundance of less than five were filtered out to obtain the final amplicon sequence variants (ASVs) and feature sheets (Callahan et al., 2016). For the obtained ASVs (Callahan et al., 2017), on the one hand, species annotation was performed on the representative sequence of each ASV to obtain the corresponding species information and species-based abundance distribution, as well as species richness and evenness information within the sample. On the other hand, multi-sequence alignment of ASVs was performed, and nonmetric multidimensional scaling (NMDS) was used to explore the differences in community structure among different samples. T-test statistical analysis methods were used to test the significance of differences in species composition and community structure. Spearman correlation analysis was used to analyze the interaction between different microorganisms.

Metagenomic sequencing

Library construction and sequencing

A total amount of 1 μg DNA per sample was used as input material for the DNA sample preparations. Sequencing libraries were generated using NEBNext® Ultra™ DNA Library Prep Kit for Illumina (NEB, USA) following the manufacturer’s recommendations, and index codes were added to attribute sequences to each sample. Briefly, the DNA sample was fragmented by sonication to a size of 350 bp. Then, DNA fragments were end-polished, A-tailed, and ligated with the full-length adaptor for Illumina sequencing with further PCR amplification. At last, PCR products were purified (AMPure XP system), and libraries were analyzed for size distribution by Agilent 2100 Bioanalyzer and quantified using real-time PCR.

The index-coded samples were clustering on a cBot Cluster Generation System according to the manufacturer’s instructions. After cluster generation, the library preparations were sequenced on an Illumina HiSeq PE150 platform, and paired-end reads were generated.

Sequencing data processing and analysis

We were preprocessing the Raw Data from the Illumina HiSeq sequencing platform using Readfq software to obtain a Clean Date. Metagenome assembly of clean data was performed by SOAPdenovo software. An abundance of information on each sample in the gene catalogue was obtained. Species annotation information of UniGene was obtained from the gene catalogue and microRNA database. Combined with the gene abundance table, species abundance tables at different taxonomic levels were obtained. We compared the gene sequences to obtain functional information with two available databases: Kyoto Encyclopedia of Genes and Genomes (KEGG) and carbohydrate-active enzymes (CAZy).

Results

Amplicon sequencing

Microbial community diversity

Figure 1 shows the relative abundance of microorganisms at the phylum and genus level for bacteria and fungi based on amplicon sequencing. Firmicutes was the dominant phylum in all samples at the phylum level of bacteria (Fig. 1A). In Daqu, the dominant phyla were mainly Firmicutes (69%), Proteobacteria (7%), Cyanobacteria (4%), and Bacteroidota (4%). At the beginning of fermentation (JP0), Firmicutes, Proteobacteria, and Cyanobacteria in fermented grains were provided by Daqu. Actinobacteria was probably provided by raw materials such as sorghum, as it only was found in sample JP0 and was detected in Daqu and fermented grains at the middle and late stages of fermentation. The relative abundance of Firmicutes increased to more than 80% as fermentation progressed. The relative abundance of Proteobacteria decreased in the early stage (0–7 days) and middle stage (7–15 days) of fermentation but increased in the late stage (15–28 days) of fermentation. The abundance of Bacteroidota continued to increase during fermentation.

Fig. 1

Microbial community diversity based on amplicon sequencing (top 10 in relative abundance). Bacterial diversity at phylum (A) and genus level (B). Fungal diversity at phylum (C) and genus level (D)

At the genus level of bacteria (Fig. 1B), Lactobacillus was the dominant genus in all samples, followed by Pediococcus and Weissella. The relative abundance of Lactobacillus continued to increase as fermentation progressed, with the most incredible abundance at 15 days of fermentation and a slight decrease at the late stage of fermentation, probably due to the higher ethanol content in the late stages, which was not suitable for the growth of Lactobacillus. Pediococcus increased in relative abundance in the early stages of fermentation and remained stable during the middle of fermentation but decreased during the late stages. Weissella was present throughout the whole fermentation stage, and the relative abundance was relatively stable in the early and middle stages, with a slight increase in the late fermentation stage. The abundance of Bacillus in JQ and JP28 was more significant than 1%, and the relative abundance was very low in the early and middle stages of fermentation. Staphylococcus was detected at the beginning of fermentation (JP0) but not in Daqu (JQ), indicating that Staphylococcus was mainly derived from sorghum. As fermentation progressed, the relative abundance of Staphylococcus decreased and was detected only in JP28. Enterococcus was only detected in the 28-day fermentation sample (JP28). The above Lactobacillus, Pediococcus, Weissella, and Enterococcus all belonged to lactic acid bacteria (LAB), indicating that LAB was the dominant bacteria in both the Daqu and fermented grain samples.

At the phylum level of fungi (Fig. 1C), Ascomycota was the dominant fungus in both Daqu and the whole fermentation process, with relative abundance above 90%. At the genus level (Fig. 1D), Issatchenkia (68%), Candida (11%), and Pichia (4%) were the primary dominant fungi in Daqu, while Pichia had a relative abundance of less than 1% during fermentation. Issatchenkia and Saccharomycopsis are the dominant fungi during fermentation. The relative abundance of Issatchenkia decreased from 46 to 21% at the early stage of the fermentation (0–7 days), after which it gradually increased. However, the relative abundance of Saccharomycopsis increased rapidly from 24% on day 0 to 64% on day 7 and then decreased continuously. In addition, Candida and Aspergillus were present in each sample, with a higher relative abundance in JP0 and JP28 than in JP7 and JP15. Hanseniaspora increased abundance during the early fermentation stage (0–7 days) and was not detected during the middle and late stages. In summary, yeasts (Issatchenkia, Candida, Pichia, Saccharomycopsis, Hanseniaspora, Cystofilobasidium, and Wickerhamomyces) were the predominant fungi in Daqu and fermented grains.

Previous studies have shown that the dynamic succession of microbial communities is related to many factors, among which biological factors (intraspecies genomic diversity and microbial structure) (Tan et al. 2022), fermentation parameters (starch, moisture, acidity, and reducing sugar) (Wang et al. 2021), and fermentation environment (container, factory, climate, ecology) (Kang et al. 2022b; Wang 2022) and other factors are considered to drive the succession of core and distinct microbial communities.

Alpha-diversity index analysis

We compared the α-diversity of microbiota between Daqu and fermented grains using the Chao1, pielou_e, Shannon, and Simpson index (Fig. 2) to evaluate changes in microbial diversity. The diversity of bacteria was higher than that of fungi throughout fermentation. The JP0 sample had a significantly higher than that of the other samples. The Chao1 and Shannon indices of bacteria decreased significantly first and then increased slightly, while the pielou_e and Simpson indices of bacteria did not change significantly. Shannon and Simpson of fungi first decreased significantly in the early stage of fermentation, increased significantly in the middle stage, and decreased in the late stage.

Fig. 2

Alpha-diversity index analysis of bacteria (A–D) and fungi (E–H). ^a–cMean values in the same index with different lowercase letters differ significantly (p < 0.05)

Beta-diversity analysis

Nonmetric multidimensional scaling (NMDS) (Kruskal 1964) was used to analyze the degree of difference between the samples, which was reflected by the distance between points. If the stress is less than 0.2, NMDS can accurately reflect the degree of difference between samples. The result is shown in Fig. 3. For both bacteria and fungi, there was a significant difference between one replicate sample of Daqu and three replicates of JP0, while the replicate samples of JP7, JP15, and JP28 were close. Moreover, the community structure of samples in the middle and late fermentation stages was similar, indicating that the microbial community structure in fermented grains gradually stabilized with the fermentation process.

Fig. 3

NMDS analysis of bacteria (A) and fungi (B) community. A closer distance between points indicates higher similarity; n = 3

Microbial difference analysis

T-test was used to analyze the microbial differences in different samples and the species with significant differences among the groups at each taxonomic level (p < 0.05). The analysis results between bacteria and fungi at the genus level are shown in Fig. 4. Weissella, Cladosporium, and Fusarium in Daqu were significantly more abundant than in JP0 samples. At the early stage of fermentation (JP0 vs. JP7), the abundance of Pediococcus, Muribaculaceae, and Saccharomycopsis increased significantly. At the middle stage of fermentation (JP7 vs. JP15), there was no significant change in bacteria; Issatchenkia increased significantly, while Saccharomycopsis decreased significantly. At the late fermentation stage (JP15 vs. JP28), the abundance of Pediococcus, Acetobacter, Leuconostoc, and other bacteria decreased significantly, while Staphylococcus increased significantly. However, Aspergillus, Thermoascus, Rhizopus, and other fungi increased significantly, and Hanseniaspora decreased significantly.

Fig. 4

T-test analysis of differences of bacteria (A–C) and fungi (D–G) at genus level in different fermentation stages. The bacteria in JP7 and JP15 samples were not significantly different

Correlation network of bacterial and fungal community

Respectively choose the relative abundance of bacteria and fungi in the level before the results of 40, adopt Spearman correlation analysis of testing, and select the relevant coefficient |R|> 0.6 and p < 0.05 microorganisms were plotted, and the results are shown in Fig. 5. In the correlation network diagram of bacteria (Fig. 5A), Lactobacillus was negatively correlated with most bacteria, indicating that the growth of Lactobacillus could inhibit the growth of other bacteria during the fermentation process. In addition, Pediococcus was positively correlated with Weissella and negatively correlated with Bacillus. In the correlation network diagram of fungi (Fig. 5B), Issatchenkia was found to be negatively correlated with Saccharomycopsis and Hanseniaspora, as well as Candida and Hanseniaspora.

Fig. 5

Correlation networks of bacteria (A) and fungi (B) based on amplicon sequencing. Different nodes represent different genera, node size represents the degree of connectivity of the genus, the same color represents the level of the same phylum, the thickness of the line represents the size of correlation, the red line indicates positive correlation, and the blue line indicates negative correlation

Metagenomics sequencing

Microbial community structure

The microbial community structure in the samples based on metagenomic sequencing is shown in Fig. 6. At the phylum level (Fig. 6A), Firmicutes, Ascomycota, and Mucoromycota were the dominant phylum in Daqu and fermented grains. Among them, the relative abundance of Ascomycota in Daqu was 39%, which gradually decreased during the fermentation process, reaching the lowest (4%) on the 15th day of fermentation (JP15), and increased slightly in the late stage of fermentation. The relative abundance of Firmicutes increases during the early and middle stages of fermentation and decreases somewhat during the late stages.

Fig. 6

Microbial community diversity at phylum (A) and genus level (B) based on metagenomic sequencing (top 10 in relative abundance). Abundance clustering heat of microbial (top 35) map genus level (C)

At the genus level (Fig. 6B), Lactobacillus (17.4%), Pichia (33.2%), Lichtheimia (6.5%), and Leuconostoc (2.1%) were the predominant microorganisms in Daqu. At the beginning of fermentation (JP0), the relative abundance of Lactobacillus, Pichia, and Lichtheimia was 20.7%, 24.0%, and 13.1%, respectively, and the microbial structure was similar to that of Daqu. During the fermentation process, the relative abundance of Lactobacillus and Pediococcus reached a maximum on the 15th day of fermentation (JP15) and decreased slightly at the late fermentation stage. Leuconostoc was highest on the 7th day of fermentation and then reduced gradually. The relative abundance of Pichia and Lichtheimia decreased during fermentation’s early and middle stages and increased slightly during the late stages. Saccharomyces showed an increasing trend during the whole fermentation process. Overall, the rapid growth of LAB (Lactobacillus, Leuconostoc, Pediococcus, Weissella) inhibited the growth of yeast (Pichia, Saccharomyces, Candida) and molds (Lichtheimia, Rhizopus) in the early and middle stages of fermentation, while in the late stage of fermentation, the fermentation tended to be complete with LAB was slightly reduced, and yeast increased somewhat.

Combining the clustering heat map of the top 35 microorganisms in terms of abundance at the genus level (Fig. 6C) revealed that the diversity of microbial populations gradually decreased as the fermentation proceeded. Still, the abundance of the dominant bacteria increased, with Lactobacillus, Pediococcus, and Saccharomyces mainly acting as fermenters in the late stages of fermentation.

In conclusion, microorganisms at the beginning of fermentation were mainly from Daqu, and yeasts were more abundant than LAB and were the dominant species in fermented grains. In the early and middle stages of fermentation, microbial succession in fermented grains was significant, the abundance of yeast decreased, and LAB continued to increase. At the later stage of fermentation, the microbial community structure tended to be stable, LAB became the main dominant species, and the relative abundance of yeast increased slightly.

Figure 7 shows the microbial community structure in each sample, with inner to outer circles showing species abundance at phylum, genus, and species levels, respectively. In JQ and JP0, Pichia kudriavzevii and Lichtheimia ramosa were the main dominant fungi, and Companilactobacillus paralimentarius was the main dominant bacteria. With the fermentation process, fungi’s abundance rapidly decreased, while bacteria’s abundance increased, and continuous succession occurred. At the genus level, a fermentation system with Lactobacillus as the main dominant bacteria was gradually formed, while at the species level, the species was still undergoing succession. In the late fermentation stage, the abundance of Lactobacillus acetotolerans increased from 0.2 to 9% and became the dominant species together with Lentilactobacillus parabuchneri. Huang et al. (2020a, b) found that Lactobacillus acetotolerans, Lentilactobacillus buchneri subsp. buchneri, and Lentilactobacillus hilgardii were the main dominant strains in fermented grains. Wang and Xu (2019) used a culture-dependent method to find that Pichia anomala and Saccharomyces cerevisiae were the dominant yeast species in Fen-flavored Baijiu. Different from the dominant species in this study, all strains were detected, probably due to different production batches, resulting in differences in species at the species level.

Fig. 7

Microbial community structure in each sample (abundance > 0.3%). The circles from inside to outside are the species abundance at phylum, genus, and species level, and the same color represents the same phylum

Correlation network of microbial community

Metagenomics sequencing provides information on all the microorganisms, based on the species abundance by the Spearman correlation analysis between total microbial relationship, select the relevant coefficient | R |> 0.6, p < 0.05 for drawing of microorganism; the result is shown in Fig. 8. The result is 34 nodes, 196 edges, 181 negative correlations (red lines), and 15 positive correlations (blue lines). Lactobacillus was negatively correlated with Pichia, Bacillus, and Pseudomonas, while Candida was positively correlated with Lactococcus and Weissella. In conclusion, during the fermentation of fermented grains, lactic acid and bacteriocin produced by LAB inhibited the growth and metabolism of some microorganisms (Porto et al. 2017), and antagonistic relationships existed among different yeasts, resulting in changes in microbial community structure.

Fig. 8

Correlation network of the microbial community. Different nodes represent different genera, node size represents the degree of connectivity of the genus, the same color represents the level of the same phylum, the thickness of the line represents the size of correlation, the red line indicates positive correlation, and the blue line indicates negative correlation

Functional gene annotation by blasting to KEGG databases

KEGG and CAZy databases are using annotated metagenomic data to explore the functional characteristics and differences of microorganisms in Daqu and fermented grains at different fermentation stages. Figure 9 shows the results of functional gene annotation based on the KEGG database. Metabolism is the functional gene with the most significant number of annotations (Fig. 9A), which was the most abundant gene in Daqu and fermented grains, and its abundance reached 15.5% at day 15 of fermentation (Fig. 9B). According to the cluster heat map of available abundance under KEGG level 2 classification (Fig. 9C), the function of Daqu (JQ) is similar to that at the beginning of fermentation (JP0), and energy metabolism had the highest relative abundance. During fermentation, carbohydrate and amino acid metabolism increased and became the central functional gene. In addition, membrane transport was the primary functional gene in fermentation’s middle and late stages.

Fig. 9

Functional gene annotation results based on the KEGG database. A Statistical chart of the number of KEGG unigenes. The numbers on the bar chart represent the number of unigenes on the annotation. B Relative abundance histogram at level 1 of KEGG. C, D Cluster heat map at levels 2 and 3 of KEGG; the cluster tree on the left is the functional cluster tree. The corresponding value of the middle heat map is the Z-value obtained after the standardization of the relative abundance of each line of functions

In the metabolic pathway analysis at level 3 of KEGG (Fig. 9D), 396 categories were annotated. At the beginning of fermentation (JP0), the metabolic pathway of microorganisms was mainly oxidative phosphorylation (ko00190). The primary metabolic function during fermentation gradually evolved into carbohydrate metabolism (purine metabolism (ko00230), pyrimidine metabolism (ko00240), pyruvate metabolism (ko00620), starch and sucrose metabolism (ko00500), glycolysis/gluconeogenesis (ko00010), amino sugar and nucleotide sugar metabolism (ko00520), pentose phosphate pathway (ko00030), galactose metabolism (ko00052)), amino acid metabolism (alanine, aspartate, and glutamate metabolism (ko00250), cysteine and methionine metabolism (ko00270), and nucleotide metabolism (purine metabolism (ko00230), and pyrimidine metabolism (ko00240).

KEGG analysis showed that the microbial communities in fermented grains were mainly growing and reproducing in the early fermentation stage. As the progress of fermentation, microorganisms metabolize carbohydrates and amino acids in raw materials to produce small molecular compounds (such as ethanol, lactic acid, and ethyl acetate).

Functional gene annotation by blasting to CAZy databases

Figure 10 shows the results of functional gene annotation based on the CAZy database. All genes were grouped into six categories at level 1 of CAZy. Glycoside hydrolases (GH) and glycosyl transferases (GT) were annotated in higher numbers (Fig. 10A) and were also the predominant carbohydrate enzymes in all samples (Fig. 10B). Figure 10C shows the clustering results of functional abundance at level 2 of CAZy. The main GHs in JQ are endo-polygalacturonase (GH28), mannosyl-oligosaccharide 1,2-α-mannosidase (GH38), and lactase (GH1), which is similar to that in JP0 samples. As fermentation progresses, lactase (GH1), murein polymerase (GT51), and levansucrase (GH32) increase and become the significant enzymes in JP7. However, in the middle and late stages of fermentation, α, α-trehalase (GH65), lysozyme (GH73, GH25), (1, 4)-α-D-glucan 1-α-D-glucosylmutase (GH13), and mannosylglycoprotein endo-β-mannosidase (GH2) increased and became the main enzymes.

Fig. 10

Functional gene annotation results based on CAZy database. A The statistical chart of CAZy unigenes. B Relative abundance bar graph at level 1 of CAZy. C, D Cluster heat map at level 2 and EC of CAZy

In summary, the types of carbohydrate enzymes metabolized by microorganisms constantly change at different fermentation stages. In the early fermentation stage, hydrolysis enzymes are mainly generated to decompress macromolecular carbohydrates. In addition, the abundance of lysozyme increased at the late stage of fermentation, which promoted the dissolution of dead colonies. In addition, the abundance of lysozyme increased at the late stage of fermentation, which promoted the dissolution of dead colonies.

Discussion

The analysis of microbial community structure in fermented grains of Baijiu helped understand the rule of microbial succession in the fermentation system, which was essential to control the quality of liquor and was beneficial to the mining of microbial resources. In this study, we use amplicon (16S rDNA and ITS) sequencing and metagenomic sequencing to analyze the diversity and functions of microbial communities in the fermented grains and Daqu of Baijiu.

By amplicon sequencing, we found that Lactobacillus and Issatchenkia were identified as the dominant bacteria and fungi during the fermentation process, respectively, and the community diversity of bacteria in fermented grains was higher than that of fungi. However, amplicon sequencing can only obtain the community information of bacteria or fungi by amplifying the 16S rDNA or ITS gene sequence, so we cannot obtain the specific distribution of bacteria or fungi in the entire fermentation system. Metagenomes can compensate for this defect and obtain complete microbial community information.

Metagenome sequencing results showed that Lactobacillus, Weissella, Issatchenkia, Pichia, Candida, and Lichtheimia were the dominant microorganisms in Daqu. It is similar to the microbial structure of Daqu studied by Yang et al. (2021). Moreover, we found that yeasts were more abundant than LAB at the beginning of fermentation, and the Lactobacillus, Issatchenkia, or Pichia in fermented grains are mainly provided by Daqu, while Saccharomycopsis may come from raw materials or the environment (Du et al. 2019). With the progress of fermentation,

Lactobacillus and Weissella coexist and play an important role in many fermented foods (An et al. 2021; Yongsawas et al. 2022).

In addition, some less abundant microorganisms in Daqu and fermented grains, such as Bacillus, Aspergillus, and Lichtheimia, also played essential roles during fermentation. Aspergillus and Lichtheimia were present in each sample and associated with amino acid metabolism (Liang et al. 2020). Some studies have found that Bacillus and yeast are the most effective microorganisms in the first 10 days of Baijiu fermentation (Wang & Xu 2019). In this study, Bacillus was only present in Daqu and the samples on the 28th day of fermentation, and the abundance was low. However, the relative abundance of Jiang-flavor Baijiu was 13.32–31.98% (Wang et al. 2021). When fermented grains’ acidity and ethanol content increased (Hu et al. 2021), the growth environment for microorganisms became harsh, increasing the abundance of microorganisms with good tolerance, such as Bacillus.

Amplicon sequencing can only identify microorganisms at the genus level but cannot obtain specific information about species at the species level. Metagenomic sequencing can sequence and analyze the entire genome of a sample, which can obtain more detailed information on microbial classification and genes. Annotation of species at the species level using metagenomic sequencing showed that

Pichia kudriavzevii, Lichtheimia ramosa, Companilactobacillus paralimentarius, Lentilactobacillus parabuchneri, and Lactobacillus acetotolerans were the main strains in the fermentation process.

In addition, metagenomic sequencing can also obtain the functional information of microbial communities, which can better obtain the information of microorganisms in a sample. In this study,

KEGG and CAZy annotations revealed the metabolic functions of microorganisms in fermented grains. The annotated functions of JQ and JP0 samples were similar, and energy metabolism was the primary function. Carbohydrate metabolism and amino acid metabolism become the main functions in fermentation’s middle and late stages. Glycoside hydrolases (GH) and glycosyl transferases (GT) are the major enzymes produced by microorganisms. The results were similar to those of Zhang et al. (Zhang et al. 2022). Studies have found that ethanol production in fermented grains is mainly due to yeasts participating in the glycolytic pathway to promote glycation, lipid production, and alcohol fermentation (Huang et al. 2017a, b; Li et al. 2018; Q et al. 2012; Xu et al. 2022). Aroma components are associated with amino acid metabolism and biosynthesis (An et al. 2020; Zhao et al. 2016), and the dynamic interaction between yeast and LAB promotes the metabolism of flavor chemicals (Tan et al. 2022).

It is worth noting that the difference between the two sequencing methods was mainly reflected in Issatchenkia and Pichia, which were closely related. Furthermore, the difference was mainly due to the different databases used. For amplicon species annotation, the SILVA database was used for bacteria and UNITE database for fungi. Metagenomic sequencing is an NR database, and the species information obtained by annotation is more comprehensive, including bacteria, fungi, archaea, and viruses. By comparing the two sequencing methods, we suggest that metagenomic sequencing should be used in the microbial studies of traditional fermented foods to understand the microbial fermentation mechanism better.

Conclusions

Amplicon sequencing can better analyze microbial diversity but can only detect species information at the genus level. In contrast, metagenome sequencing can explore the microbial structure and obtain functional information about microorganisms. This study combined amplicon sequencing and metagenomic sequencing technology to reveal the microbial community structure and gene function prediction during the fermentation of Fen-flavor Baijiu. The Spearman coefficient was used to analyze the correlation between microorganisms. The results showed that at the beginning of fermentation, the microorganisms were mainly from Daqu, and the abundance of yeast was higher than that of LAB. With the progress of fermentation, the abundance of yeast decreased, while lactic acid bacteria continued to increase. At the later fermentation stage, the microbial community structure tended to be stable, LAB became the main dominant species, and the relative abundance of yeast increased slightly. Functionally, fermented grains’ crucial functions were amino acid and carbohydrate metabolism. Finally, we concluded that fermented grains could be a good strain pool for development. In the future, the physicochemical properties and aroma components in fermented grains could be determined, and the correlation between physical and chemical composition and microorganisms could be analyzed further to clarify the specific functions of microorganisms and explore the resources of microbial strains, which is also one of the research hotspots in Baijiu.