Analyzing bacterial community in pit mud of Yibin Baijiu in China using high throughput sequencing

“Yibin Baijiu” (YB) is a special Chinese strong-aroma Baijiu (CSAB) that originated in Yibin, a city in western China. YB is fermented in cellars lined with pit mud (PM), the microbiota in which may affect YB quality. In this study, high throughput sequencing of the 16S rRNA gene was used to demonstrate the bacterial community structure and diversity in PM of YB. In addition, the physicochemical characteristics of PM were also analyzed, including moisture content, pH, and available phosphorous, ammonia nitrogen, and humic acid levels. Results showed that Firmicutes was the dominant phylum in all PM samples with abundance > 70.0%, followed by Euryarchaeota (11.3%), Bacteroidetes (6.5%), Synergistetes (3.0%), Actinobacteria (1.4%), and Proteobacteria (1.2%). Furthermore, 14 different genera with average relative abundance of > 1% were detected. The Chao1 and Shannon indexes did not vary significantly between the sub-layer and middle-layer PM (P > 0.05). However, Linear discriminant analysis Effect Size (LEfSe) analysis showed that the relative abundance of Lactobacillus in the sub-layer PM was significantly higher than in middle-layer PM. pH differed significantly (P < 0.05) between the two groups. Canonical correspondence analysis revealed that bacterial community in PM correlated significantly with available phosphorous content and pH. Our study provides basic data for further elucidating the diversity of microbiota in the PM of YB and the potential mechanism of Baijiu production.


INTRODUCTION
Baijiu is one of the six most famous distilled liquors worldwide (Fan & Qian, 2006;Jin, Zhu & Xu, 2017;McGovern et al., 2004). It includes four main aroma types: strong, light, sauce, and rice. Among these, the Chinese strong-aroma Baijiu (CSAB) accounts for 70% of the total Baijiu production (Liu & Sun, 2018;Liu et al., 2017b) . Yibin is one of the birth places of CSAB culture (Zou, Zhao & Luo, 2018;You et al., 2016), and it is located at the confluence of the Min River, Jinsha River, and Yangtze River, which has good water quality. In addition, it has unique weak acid clay soil with strong viscosity, good water retention Hence, this study aimed to investigate the bacterial community structure and diversity in PM of YB using high-throughput sequencing. Furthermore, the physicochemical characteristics of PM were determined to investigate the correlation between bacterial community structure and environmental factors, which will assist us in accumulating basic data for further elucidating the mechanism of Baijiu brewing, and will promote the long-term development of YB.

Sampling
PM samples were collected in December 2017 from 13 distilleries located in Yibin (28 • 76 N, 104 • 63 E). In total, 68 samples were collected, among which 37 were collected from the middle layer of cellar walls and 31 were collected from the substrate layer (100 g PM at each position). The PM sampling sites are shown in Fig. S1. PM samples were numbered according to the information regarding the distillery, cellar, and layer, and were divided into two groups, the middle-layer and the sub-layer (Table S1). The samples were aseptically collected in sterilized bags and maintained at low temperature during transport from the distilleries to the laboratory.

Physicochemical properties
The basic physicochemical properties of PM, including moisture content, pH, and available phosphorous, ammonia nitrogen, and humic acid content were also determined as described previously (Shen, 2014).

DNA extraction, polymerase chain reaction (PCR), and Illumina HiSeq sequencing
DNA was extracted from the PM using the FastDNA SPIN kit for soil (Mpbio, USA) according to the manufacturer's instructions. The crude DNA was quantified based on the absorbance at 260 nm using a NanoDrop TM one spectrophotometer (Thermo Fisher Scientific, USA), and the purity was assessed using agarose gel electrophoresis (Beijing Liuyi Biological Technology Co., Ltd. China).

Nucleotide sequence accession numbers
The HiSeq sequencing data were submitted to the Sequence Read Archive (SRA) of the NCBI database as BioProject ID PRJNA597727.

Data analysis
All the raw HiSeq-generated 16S rRNA gene sequencing data were processed using the QIIME pipeline (V1.9.1). After initial quality control (QC) processing, pairs of reads were merged using FLASH (V1.2.7, Magoč & Salzberg, 2011) to acquire raw tags. Then, they were screened using the QIIME function (Caporaso et al., 2010) to obtain clean tags. Chimeras were detected and removed using the Uchime algorithm to obtain effective tags (Edgar et al., 2011). The effective tags, at a similarity threshold of 97%, were grouped into the same operational taxonomic units (OTUs). A representative sequence for each OTU was extracted and annotated using the SILVA database V119 (Quast et al., 2012).
Community richness and diversity were estimated using the Chao1 (Chao & Bunge, 2002) and Shannon indices (Shannon, 1997), respectively. Bacterial community differences in two groups of PM were evaluated using principal coordinates analysis (PCoA) in Fast UniFrac (Hamady, Lozupone & Knight, 2010). Linear discriminant analysis effect size (LEfSe) was used to compare the significance of species-specific differences between the two groups (Segata et al., 2011). Canonical correspondence analysis (CCA) between bacterial communities in the PM and physicochemical properties was performed using the vegan package in R (Oksanen et al., 2018). All the statistical analyzes, including Spearman's test, were performed with the SPSS software (V22.0, IBM Corp., Armonk, NY, USA). All tests for significance were two-sided, and P-values <0.05 were considered statistically significant.

Physicochemical properties of the PM
The physicochemical characteristics of the PM of the two groups are shown in Table S2 . Among the physicochemical indicators analyzed using one-way analysis of variance (ANOVA) of SPSS Statistics 22, pH differed significantly (P < 0.05) between the two groups. However, no significant differences in moisture content, and ammonia nitrogen, available phosphorus, and humic acid levels were observed.

Diversity of bacterial community in PM
In total, 68 samples were analyzed for amplicon sequencing. On an average, 99,639 clean tags were generated for each sample with a minimum of 53,888 tags. In total, 53,493-208,410 effective tags for each sample with an average sequence length of 444 bp were generated. Rarefaction curves were used to assess the degree of completion of a taxonomic survey (Good, 1953). The rarefaction curve showed that all curves approached the saturation plateau, proving that the identified sequences of PM samples represented almost all the bacterial sequences (Fig. S1).
Eventually, 220,790 operational taxonomic units (OTUs) were detected based on 97% similarity in 16S rRNA gene sequences, of which the middle-layer group and sub-layer group had 186,289 and 148,904 OTUs, respectively. Furthermore, 114,403 OTUs were shared between the two groups. The mutual and unique OTUs for the two groups are shown in Venn diagrams (Fig. S2).
The bacterial richness (Chao1) indices for all samples ranged from 1,214.30 to 42,441.82, and were fairly similar in the two different groups, with 11,296.93 ± 4,816.73 in the middlelayer group and 10,549.07 ± 7,638.48 in the sub-layer group. Furthermore, the diversity (Shannon) indices were 7.82 ± 1.68 in the middle-layer group and 7.57 ± 1.21 in the sub-layer group. The Chao1 and Shannon indexes did not vary significantly between the two groups (P > 0.05), suggesting similar overall species diversity ( Fig. 1). Furthermore, the PCoA analysis showed that there was no explicit clustering of samples within the groups, demonstrating significant inter-individual variations in the bacterial communities ( Fig. 2).

Comparison of dominant bacterial communities in different Baijiu PM
Based on the results of previous studies, we compared the dominant microbiota of strong-aroma Baijiu in different regions to ascertain the common dominant bacteria, and the unique dominant bacteria of YB ( and Methanobacterium were also detected in our study with the relative abundance above 1%, and almost all of them belonged to phylum Euryarchaeota. Three unique bacteria were also detected in YB PM, namely Ruminiclostridium 5, Gelria, and Syntrophaceticus. In addition, the common dominant microorganisms of strong-aroma Baijiu and those of other aroma types (miscellaneous-aroma and sauce-aroma) were compared. The dominant microbiota composition in the Baijiu PM of different aroma types differed considerably. Only Lactobacillus was the dominant microorganism shared by three aroma types of Baijiu PM. In addition, Enterobacter was the other dominant microorganism shared by miscellaneous-aroma and sauce-aroma, which was not found in strong-aroma Baijiu PM.  The detailed information regarding the dominant microflora of the three aroma types of Baijiu PM is shown in Table 3.

DISCUSSION
This is the first study to use the Illumina Hiseq sequencing platform for investigating the population diversity of bacteria in the middle-layer and sub-layer PM of YB. In addition, the relationship between bacterial community structure and environmental factors was also assessed. The results indicated that the dominant bacteria in PM of YB were different from that of other regions and aroma types of Baijiu. The pH and available phosphorus content significantly affected the PM bacterial community structure.
According to previous studies, the total number of effective tags of the 16S rRNA gene of microorganisms in PM obtained from the Roche 454 or Illumina Miseq sequencing platforms were less than 10,000 (Liu et al., 2017a;Tao et al., 2014b). In our study, more than 200,000 effective tags were detected in a single sample with an average of 94,023 tags for all samples. Thus, the result of sequencing can adequately reflect most of the sample diversity noted. In total, 49 phyla, 332 families and 811 genera were detected, which may be helpful for further understanding the phylogenetic status of numerous uncultured microbes in the PM ecosystem. Although the relative abundance of Firmicutes, Euryarchaeota, Bacteroidetes, and Synergistetes differed slightly between middle-layer and sub-layer samples, they were still the key components of the microbiota in the PM of YB. Previous studies also showed that Firmicutes, Euryarchaeota and Bacteroidetes were the three most abundant phyla in PM (Liang et al., 2015;Liu et al., 2017a;Tao et al., 2014a). Firmicutes include syntropyhic bacteria that can degrade volatile fatty acids such as butyrate and its analogs, and is represented mainly by the Clostridia and Bacilli (Garcia-Peña et al., 2011). Euryarchaeota includes diverse groups of methanogens that are interspersed with non-methanogenic lineages. Methanogens are often symbiotic with hexanoic acid bacteria and can effectively increase the content of ethyl hexanoate in strong-aroma Baijiu via ''hydrogen transfer'' effect between species (Zhao et al., 2017). Bacteroidetes includes three major types of bacteria: Bacteroidia, Flavobacteriia, and Sphingobacteriia. Most of them are capable of degrading macromolecular carbohydrates such as cellulose to produce acid (Guo et al., 2014). All these phyla in PM played a vital role in Baijiu fermentation.
At the genus level, the dominant bacteria in PM from different regions producing strong-aroma Baijiu were compared. In total, 11 dominant genera were detected by Liu et al. (2017a) in CSAB's PM collected from Luzhou, Sichuan province (about 110 kilometers from Yibin city), among which 8 genera were also the dominant bacteria in YB PM. Tao et al. (2014a) reported 14 dominant genera in CSAB's PM collected from Mianzhu, Sichuan province (about 300 kilometers from Yibin city), 6 of which were different from those of YB PM. In comparison, we observed that the composition of the dominant genera in PM in the Mianzhu area differed from that in the Yibin area due to their geographical distance. In contrast 7 of the 17 dominant genera in Jiangsu area (about 1,700 kilometers away from Yibin city) Baijiu PM were shared by YB PM. The dominant genera in Jiangsu area Baijiu PM and YB PM differed significantly, which further suggested that geographical location considerably affects Baijiu PM bacterial community. Previous studies have demonstrated that the composition of PM bacterial community varied by geographical location (Huang et al., 2015;Liang et al., 2016). Interestingly, the dominant genera in the Baijiu PM of Hunan region, which is 470 km away from Yibin City, was also similar to the dominant genera in YB PM. It was speculated that the identical latitude of Hunan and Yibin, which resulted in both having the same climate, especially temperature, significantly affected the diversity and structure of the microbiota. However, the farther are two places, the more is the similarity in the dominant genera of the Baijiu PMs from these areas.
The dominant genera in the PM of strong aroma Baijiu differed considerably from those of Baijius of other aromas (Table 3). Obviously, the differences in these dominant bacteria contributed significantly to the formation of different aroma types of Baijiu. Lactobacillus was a dominant genus among Baijiu PMs of several aroma types, (Huang et al., 2017;Wang, Zhang & Liu, 2016;Hu et al., 2016), leading to the speculation that Lactobacillus was critical for Baijiu brewing. Gelria was found to be the dominant genus in our study, which has been rarely reported in previous studies. In syntrophic association with a hydrogenotrophic methanogen, Gelria can utilize various amino acids and sugars to produce acetate, propionates, H 2 , NH 4 + , and CO 2 (Plugge et al., 2002). Fan et al. (2019) reported that Gelria showed a significant positive correlation with the content of major flavor substances such as alcohols, esters, and aldehydes in Baijiu. Therefore, as a unique dominant genus of YB PM, we speculated that Gelria may contribute to the formation of unique flavor of YB. Furthermore, a positive correlation between humic acid content and Gelria abundance was evident in the heat map (Fig. 6). Humic acid is an important organic matter in PM, which is formed by the accumulation of microbial metabolites and can completely reflect the ability of PM to supply nutrients to microorganisms (Zhu et al., 2018). We speculated that the metabolic activity of Gelria promoted the formation of humic acid, which in turn provided nutrition for the growth and metabolism of Gelria.
The difference between the physicochemical indicators of the middle-layer and sub-layer PM of YB was analyzed using ANOVA. Results showed that the pH of the sub-layer was significantly lower than that of the middle-layer. Meanwhile, CCA analysis showed that the pH of PM correlated considerably with the bacterial community structure in PM.
Spearman's analysis revealed that the Clostridium sensu stricto 12, Lachnoclostridium, and Lactobacillus were correlated negatively with pH value. LEfSe also showed that Lactobacillus and Clostridiaceae_ 1 were significantly less abundant in the middle-layer of PM with higher pH. In particular, the above three genera that were significantly related to PM pH belonged to Firmicutes, the most abundant phylum. Among them, Clostridium sensu stricto 12 and Lachnoclostridium belonged to the order Clostridiales, while Lactobacillus belonged to the order Lactobacillales. Bacteria of order Clostridiales possess the metabolic ability to convert organic substances into organic acids (such as butyric, caproic acids, etc.), alcohols, CO 2 /H 2 , and minerals (Kenealy, Cao & Weimer, 1995;Zheng et al., 2013). Furthermore, Clostridiales contributed to the formation of ethyl butyrate and ethyl caproate when butyric and caproic acids reacted with alcohols via enzymatic and non-enzymatic catalysis (Ding et al., 2014b). Lactobacillales also plays an important role in Baijiu brewing, as the lactic acid produced can form ethyl lactate. Ethyl lactate, ethyl butyrate, and ethyl caproate are vital flavor compounds in CSAB (Fan & Qian, 2005).
Interestingly, a strong correlation of the bacterial community with the available phosphorous level was also observed using CCA analysis. The available phosphorus provides nutrients for the growth and reproduction of microorganisms. Phosphorus is present in the cell nuclear membrane and is a source of the energy substance ATP. Zheng et al. (2013) reported that the low pH and high content of available phosphorus in PM may have resulted in the orthogenesis of microorganisms and Clostridiales-, Lactobacillaceae-, and Bacillales-dominated microbial community structure. However, Spearman's analysis showed the absence of any significant correlation between bacterial presence and the available phosphorous level. This may be because Spearman's analysis only performed a pairwise correlation between environmental factors and the most abundant 35 genera and did not reflect the overall situation. In addition, the available phosphorous level may correlate significantly with some low-abundance genera in the PM, which was not evident in the Spearman's heat map.
In this study, we analyzed the community characteristics of bacteria in PM of YB, and observed that fungi also played an indispensable role in fermentation. We will study the community structure and diversity of fungi in the future. Furthermore, using a combination of metagenomic technology and metabolite detection technology, we will investigate the functional microorganisms that affect the quality and flavor of YB in PM. This will provide theoretical guidance and technical support for the application of functional bacteria in YB development.

CONCLUSIONS
This study demonstrated the population diversity of bacteria in the PM of YB from 13 distilleries using HiSeq sequencing. Our results suggested the presence of a complex community structure and abundant species in the PM of YB. No significant difference was observed in bacterial diversity between middle-layer and sub-layer PM. The LEfSe analysis showed that the relative abundance of Lactobacillus in the sub-layer PM was significantly higher than that in middle-layer PM. pH differed significantly (P < 0.05) between the two groups. Available phosphorous and pH strongly affected bacterial community structure in PM. Our study provided basic information for further elucidating the diversity of microbiota in the PM of YB and the potential mechanism of Baijiu production.

ADDITIONAL INFORMATION AND DECLARATIONS Funding
This study was supported by Yibin Science and Technology Bureau (NO. 2017CNY001-4). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Grant Disclosures
The following grant information was disclosed by the authors: Yibin Science and Technology Bureau: 2017CNY001-4.