|Home | About | Journals | Submit | Contact Us | Français|
Acetoin (3-hydroxy-2-butanone) formation in vinegar microbiota is crucial for the flavor quality of Zhenjiang aromatic vinegar, a traditional vinegar produced from cereals. However, the specific microorganisms responsible for acetoin formation in this centuries-long repeated batch fermentation have not yet been clearly identified. Here, the microbial distribution discrepancy in the diacetyl/acetoin metabolic pathway of vinegar microbiota was revealed at the species level by a combination of metagenomic sequencing and clone library analysis. The results showed that Acetobacter pasteurianus and 4 Lactobacillus species (Lactobacillus buchneri, Lactobacillus reuteri, Lactobacillus fermentum, and Lactobacillus brevis) might be functional producers of acetoin from 2-acetolactate in vinegar microbiota. Furthermore, A. pasteurianus G3-2, L. brevis 4-22, L. fermentum M10-3, and L. buchneri F2-5 were isolated from vinegar microbiota by a culture-dependent method. The acetoin concentrations in two cocultures (L. brevis 4-22 plus A. pasteurianus G3-2 and L. fermentum M10-3 plus A. pasteurianus G3-2) were obviously higher than those in monocultures of lactic acid bacteria (LAB), while L. buchneri F2-5 did not produce more acetoin when coinoculated with A. pasteurianus G3-2. Last, the acetoin-producing function of vinegar microbiota was regulated in situ via augmentation with functional species in vinegar Pei. After 72 h of fermentation, augmentation with A. pasteurianus G3-2 plus L. brevis 4-22, L. fermentum M10-3, or L. buchneri F2-5 significantly increased the acetoin content in vinegar Pei compared with the control group. This study provides a perspective on elucidating and manipulating different metabolic roles of microbes during flavor formation in vinegar microbiota.
IMPORTANCE Acetoin (3-hydroxy-2-butanone) formation in vinegar microbiota is crucial for the flavor quality of Zhenjiang aromatic vinegar, a traditional vinegar produced from cereals. Thus, it is of interest to understand which microbes are driving the formation of acetoin to elucidate the microbial distribution discrepancy in the acetoin metabolic pathway and to regulate the metabolic function of functional microbial groups in vinegar microbiota. Our study provides a perspective on elucidating and manipulating different metabolic roles of microbes during flavor formation in vinegar microbiota.
Solid-state fermentation (SSF) is one of the oldest and most economical ways of producing and preserving foods which may improve the nutritional values, sensory properties, and functional qualities of raw materials (1). The SSF processes of most known natural and industrial fermented foods, such as grape wine (2) and cheese (3), are driven by complex communities of microorganisms. As such, studies on the formation and function of food microbiota, especially as-yet-uncultivated microorganisms, during SSF processes are becoming increasingly important.
Traditional Chinese vinegars, referred to as cereal vinegars, are important seasoning and medicinal products in Chinese daily life (4). Solid-state acetic acid fermentation (AAF), an important step in producing the flavor compounds of cereal vinegar, is a repeated batch fermentation process that proceeds for many centuries without spoilage (4). In an open work environment, microbes that inhabit solid-state vinegar culture (termed Pei in Chinese) reproducibly metabolize nonautoclaved raw materials (e.g., sorghum, sticky rice, and wheat bran) and synthesize flavor compounds (5). Thus, the function of reproducible fermentation-based metabolism makes this acidic ecosystem (pH 3.0 to 3.5) amenable to adaptation for dissecting the formation and function of microbiota in food fermentation. Although many microbiological studies have been conducted to reveal the diversity and formation of microbial communities during the AAF of cereal vinegars (6,–9), the gap between community assemblage and the function of this microbial ecosystem still exists.
In vinegar microbiota, the utilization of ethanol and glucose originating from raw materials leads to diverse flavors and the formation of bioactive compounds. 2,3,5,6-Tetramethylpyrazine (TTMP), termed ligustrazine in traditional Chinese medicine, exists abundantly in the rhizome of Ligusticum wallichii. This alkaloid has been widely used in China to treat cardiovascular and cerebrovascular diseases (10). Aside from its therapeutic effect, TTMP is also a food flavor agent with a pleasant tonality of nutty, roasty, and toasty (11). It exists widely in fermented foods, including cheese, rum, Chinese liquor, and soy sauce. In our previous study, a high content of TTMP (>500 mg/liter) was detected in 5-year-aged Zhenjiang aromatic vinegar (12), a representative cereal vinegar that is certified with a protected geographical indication (PGI) (European Union no. 501/2012). Although TTMP is mainly accumulated during the aging process of vinegar, the precursors of TTMP, including acetoin (3-hydroxy-2-butanone) and diacetyl (2,3-butanedione), are biosynthesized in the AAF state (13). However, the mechanisms that underlie the formation of TTMP precursors by acid-tolerant vinegar microbiota remain poorly characterized.
The formation mechanism of TTMP is a contentious issue. Recently, a biochemical-chemical route for the synthesis of TTMP has gained convincing experimental support (14, 15). Acetoin and ammonium were demonstrated to be two key precursors of TTMP. In the AAF of cereal vinegar, microorganisms inhabiting vinegar Pei can utilize ethanol and glucose as substrates to synthesize acetoin through the diacetyl/acetoin metabolic pathway (see Fig. S1 in the supplemental material). 2-Acetolactate is an important precursor for the biosyntheses of the acetoin-diacetyl-2,3-butanediol group and the valine-leucine-isoleucine group (see Fig. S1). There are two possible routes for the origin of acetoin. The first is through the nonenzymatic decomposition of 2-acetolactate and the following biocatalysis by diacetyl reductase (DR); the other involves the direct transformation of 2-acetolactate by acetolactate decarboxylase (ALDC). The acetoin can be further transformed into 2,3-butanediol (2,3-BDO) by butanediol dehydrogenase (BDH). A previous study showed that acetoin, diacetyl, and 2,3-BDO were the dominant volatile compounds in Zhenjiang aromatic vinegar (16). Thus, it is of interest to understand which microbes are driving the formation of the acetoin-diacetyl-2,3-BDO group to elucidate the microbial distribution discrepancy in the diacetyl/acetoin metabolic pathway and to regulate metabolic function of the functional microbial group in vinegar microbiota.
In this study, the microbial functional group driving acetoin formation in vinegar microbiota was predicted by a combination of metagenomic sequencing and gene-targeted clone library analysis. The species distribution discrepancy in the diacetyl/acetoin metabolic pathway was revealed. Furthermore, functional species were isolated from vinegar microbiota by a culture-dependent method, and their acetoin-producing function was evaluated via mono- and coculture fermentation experiments. Last, the acetoin-producing function of vinegar microbiota was regulated in situ by augmenting functional species in vinegar Pei.
AAF was carried out in an open work environment with nonautoclaved raw materials. Glutinous rice is the raw material in starch saccharification and alcohol fermentation of Zhenjiang aromatic vinegar, which are similar to the techniques used for rice wine production (17). The rice wine (6.3 m3) from alcohol fermentation was mixed with wheat bran (2,550 kg) and rice hull (720 kg) in an AAF pool (0.8 m by 1.5 m by 11 m). Vinegar Pei on the 7th day from the last AAF batch was used as the starter (inoculum size, 10%) to initiate AAF. The AAF process lasted 18 days, and the temperature and humidity of Pei were maintained at 40 to 46°C and 60 to 70%, respectively. A sterilized cylinder-shaped sampler (Puluody, Xi'an, Shanxi, China) was used to collect Pei on the 7th day (about 500 g) from top to bottom at the center of three parallel AAF pools. The Pei was homogenized thoroughly in a sterile plastic bag and stored at −80°C immediately before further analysis.
Our previous study revealed that the pattern of bacterial community succession and flavor formation in vinegar Pei showed batch-to-batch uniformity, and the microbial structures of starters among different AAF batches were highly similar (similarity of 90%) (5). Here, we used mixed vinegar Pei on the 7th day from three parallel AAF pools as a representative sample for metagenomic sequencing. Vinegar Pei samples on the 7th day (about 500 g) were homogenized thoroughly in a sterile plastic bag. Then, the mixed Pei on the 7th day (10 g) was ground into powder with liquid nitrogen, and 0.5 g of Pei powder was used for the extraction of genomic DNA with a PowerSoil DNA isolation kit (Mo Bio Laboratories, Inc., Carlsbad, CA, USA). The concentration of total DNA was measured using a DyNA Quant 200 fluorometer (Hoefer, San Francisco, CA, USA). DNA purity was determined by A260/A280. DNA integrity was verified by 1% agarose gel electrophoresis under UV light. The DNA was stored at −20°C before further processing.
DNA library preparation followed the manufacturer's instructions (Illumina). The same workflow as described elsewhere was used to perform cluster generation, template hybridization, isothermal amplification, linearization, blocking and denaturization, and hybridization of the sequencing primers. The base-calling pipeline (Illumina Pipeline version 0.3) was used to process the raw fluorescent images and call sequences. We constructed one library (insert size, 330 bp) for the sample. The library was sequenced on an Illumina HiSeq 2000 platform and paired-end reads with approximate lengths of 100 bp were generated.
The raw reads in metagenomic sequencing files were trimmed using Trimmomatic to remove adaptors and the reads with low quality (18). Taxonomic and functional assignment of trimmed reads was carried out using MetaCV (19), which is a composition-based algorithm to classify short metagenomic reads (75 to 100 bp) into specific taxonomic and functional groups. For the taxonomic assignment, DNA reads were annotated with the GenBank reference database. For the functional assignment, DNA reads were annotated with the KEGG reference database. As a result, the specific organisms that participate in the diacetyl/acetoin metabolic pathway were identified.
Degenerate primers were designed to probe DR genes. A total of 32 protein sequences known to be responsible for DR were acquired from GenBank and aligned using ClustalW. Conserved regions in the gene were identified with Block Maker and then used to design consensus-degenerate hybrid oligonucleotide primers (CODEHOP) (20). Primer specificity was validated with genomic DNA prepared with a genomic DNA extraction kit (Generay, Shanghai, China) from pure cultures of 32 bacterial strains (see Table S1 in the supplemental material). The degenerate primers set were dr-f (AGCTATAGTTACAGGGGCCgsncrrggnat) and dr-r (GACGATCCCTGGGCAGwanscrttnac). The amplicon size was 513 bp.
PCR was performed on a T100 thermal cycler (Bio-Rad, Hercules, CA, USA) with primers and Ex Taq polymerase (TaKaRa, Tokyo, Japan) in a working volume of 25 μl. PCRs were performed at 95°C for 3 min, followed by 30 cycles at 95°C for 40 s, 54°C for 40 s, and 72°C for 40 s, and a final extension at 72°C for 10 min. The PCR product was purified with an agarose gel extraction kit (Generay, Shanghai, China), cloned into T-vector pMD19 (TaKaRa), and transformed into Escherichia coli JM109 cells. Randomly chosen clones were amplified and sequenced by a 3730XL DNA analyzer (ABI, Carlsbad, CA, USA).
Vinegar Pei (10 g) in the AAF was suspended in sterile physiological saline (90 ml) with glass beads in a 500-ml Erlenmeyer flask and incubated at 30°C for 30 min with shaking at 200 rpm. Then the solution in the flask was serially diluted, ranging from 10−2 to 10−5. For acetic acid bacteria (AAB) isolation, 200 μl of diluent was spread onto GYC agar, GYEC medium (9), or modified YPM medium (1% soft calcium carbonate, 1.2% agar, 0.5% ethanol) (21). Small colonies with transparent circles and different colony morphology were selected after incubation at 30°C for 2 to 3 days. Lactic acid bacteria (LAB) were isolated by the pour plate method with de Man-Rogosa-Sharpe (MRS) agar medium (22) or Elliker medium (23) with 0.04% bromocresol purple as director. After 2 to 3 days of incubation at 37°C, single colonies were picked up from the plates. The isolates which were Gram negative, catalase positive, and oxidase negative, had ellipsoidal-to-rod morphology, and grew individually, in pairs, or in short chains were supposed to be AAB; the isolates which were Gram positive, catalase negative, and oxidase-negative, had rod-to-coccoid morphology, and grew singly, in pairs, or in short chains were supposed to be LAB.
For the AAB phenotypic diversity, the following tests were performed: ethanol tolerance (5%, 10%, 15%, 20% [vol/vol]), acetic acid tolerance (3%, 5% [wt/vol]), bacterial cellulose production, temperature-tolerant property (45°C), and growth on a single carbon source (24). For the LAB phenotypic diversity, the following tests were performed: biogas, Voges-Proskauer (V-P), gelatin hydrolysis, high temperature resistance (40°C and 45°C), arginine hydrolysis, hydrogen sulfide production, 7% and 10% NaCl tolerance, litmus milk, and sole carbon source to produce acid (1% soft calcium carbonate, 1.5% agar, and 2% sole carbon source: glucose, acetate, lactate, fructose, maltose, mannitol, sucrose, and glycerol; base culture medium: 1% tryptone, 1% beef extract, 0.5% yeast extract, 0.2% K2HPO4, 0.058% MgSO4, 0.019% MnSO4, and 1% [vol/vol] Tween 80).
On the basis of the phenotypic characters, representative isolates of AAB and LAB were selected for 16S rRNA gene sequencing with primer set P0 (GAGAGTTTGATCCTGGCTCAG) and P6 (CTACGGCTACCTTGTTACGA), which were located in the E. coli rRNA gene positions 27f and 1495r, respectively (25). The sequencing analysis was performed by the Sangon Biotech Co., Ltd. (Shanghai, China), and the sequencing results were subjected to a BLAST search at NCBI.
Phylogenetic trees based on the 16S rRNA gene sequences of AAB and LAB were constructed by using the maximum likelihood method (1,000 times bootstrap) based on the general time reversible model. Evolutionary analyses were conducted in MEGA6.
Mono- and coculture fermentations of four strains (Acetobacter pasteurianus G3-2, Lactobacillus buchneri F2-5, Lactobacillus brevis 4-22, and Lactobacillus fermentum M10-3) were carried out in 250-ml Erlenmeyer flasks containing 50 ml of MRS medium (Oxoid) supplemented with 5% ethanol. Coculture flasks were inoculated with 0.2 ml of A. pasteurianus and 0.2 ml of Lactobacillus sp. Monoculture controls were inoculated with volumes identical to those for the coculture flasks. Each culture was incubated at 37°C and shaken for 6 h at 300 rpm every 12 h. Each experiment was carried out at least three times and in triplicate each time. Duncan's multiple-range test was used to determine the significant differences. Differences at a P level of <0.05 were considered statistically significant.
Quantitative real-time PCR (qPCR) was used to quantify the biomasses of LAB and AAB in mono- and coculture fermentations. qPCR amplification and detection were performed in a Chromo4 real-time 4-color 96-well PCR system (MJ Geneworks, USA) with a commercial mixture kit (SYBR Premix Ex Taq; TaKaRa, Dalian, China). Specific primers and PCR conditions for quantification of LAB and AAB were reported in previous studies (4).
A. pasteurianus G3-2 was inoculated aerobically in GYC medium (9) supplemented with 3% ethanol and incubated aerobically at 30°C for 24 h in a rotary shaker. Inocula of L. brevis 4-22, L. fermentum M10-3, and L. buchneri F2-5 were prepared in MRS medium (22) and incubated statically at 30°C for 24 h. All of the strains were propagated twice to obtain final cultures. During culture, the transferred volume was 1% (vol/vol). Cells in the final cultures were spun down (8,000 × g, 10 min) and resuspended in phosphate-buffered saline (PBS) solution. The cell concentrations of the microbial suspension liquids were adjusted to 1012 CFU/ml using optical density.
A total of 8 kg of vinegar Pei on the 7th day was collected from top to bottom at the center of the AAF pool (0.8 m × 1.5 m × 11 m) and mixed thoroughly. The mixed Pei (1 kg) was placed in a plastic bucket (14 cm inside diameter [i.d.], 18 cm height). Suspension liquids of A. pasteurianus G3-2 (100 ml), L. brevis 4-22 (100 ml), L. fermentum M10-3 (100 ml), L. buchneri F2-5 (100 ml), A. pasteurianus G3-2 (50 ml) plus L. brevis 4-22 (50 ml), A. pasteurianus G3-2 (50 ml) plus L. fermentum M10-3 (50 ml), and A. pasteurianus G3-2 (50 ml) plus L. buchneri F2-5 (50 ml) were added into the Pei in the plastic bucket, respectively. In the control group, 100 ml of PBS solution was added into the Pei. Vinegar Pei in the plastic bucket was mixed once a day, and three replicates of the mixed Pei were sampled from three different points near the center of the plastic bucket for the determination of acetoin content. All experiment groups were sampled three times using separate batches of Pei. Dunnett's t test (two-sided) was used to determine the significance.
Quantitative analyses of diacetyl, acetoin, 2,3-BDO, and TTMP in vinegar Pei were performed using a GC-2010 gas chromatograph (Shimadzu Co., Kyoto, Japan) with a headspace autosampler (TurboMatrix 16; PerkinElmer Inc., USA). The column was a DB-23 (60 m length, 0.32 mm i.d., 0.25 μm film thickness; Agilent Technologies, Santa Clara, CA, USA). Nitrogen was used as the carrier gas at a constant flow rate of 1.5 ml/min. The injector and detector temperatures were 200°C and 250°C, respectively. The gas chromatograph oven temperature was maintained at 40°C for 5 min, raised at 10°C/min to 180°C, and held for 5 min. Contents of diacetyl, acetoin, 2,3-BDO, and TTMP in the Pei were determined by the external standard method, with standard chemicals from Sigma-Aldrich Inc., Shanghai, China.
In the monoculture and coculture broths of functional microbes, contents of acetate and lactate were analyzed by reversed-phase high-performance liquid chromatography(HPLC) (Waters Atlantis T3 column, 4.6 by 250 mm; UVD, 210 nm). The mobile phase was sodium dihydrogen phosphate at 20 mM (pH 2.7).
Various microorganisms participated in the degradation of substrates and the formation of flavors in the microbial community of Zhenjiang aromatic vinegar. In this study, two separate 9.3-Gbp sequence files resulted from the Illumina paired-end sequencer. The numbers of raw read pairs and trimmed read pairs were 40,302,838 and 38,936,447, respectively. Short metagenomic sequence reads (75 to 100 bp) were classified into specific taxonomic and functional groups by MetaCV, and the number of annotated read pairs was 6,595,146. The diacetyl/acetoin metabolic pathway in the AAF ecosystem was reconstructed based on the results of MetaCV.
The relationship between microorganisms and enzymes in the diacetyl/acetoin metabolic pathway is illustrated in Fig. 1, in which DNA reads encoding enzymes are proportional to the diameters of bubbles. The read numbers of acetolactate synthase (ALS), ALDC, BDH, and ketol-acid reductoisomerase (KARI) genes were 47,792, 11,048, 11,068, and 32, respectively. Here, ALS is part of the biosynthesis of branched-chain amino acids (see Fig. S1 in the supplemental material). As shown in Fig. 1, various microorganisms harbored the ALS enzyme, including Euryarchaeota, Actinobacteria, Bacteroidetes, Cyanobacteria, Elusimicrobia, Bacillales, Lactobacillus, Leuconostocaceae, Clostridia, Rhizobiales, Rhodobacterales, Acetobacter, Burkholderiales, Betaproteobacteria, Enterobacteriales, Pseudomonadales, and Deltaproteobacteria, and likely participated in the formation of 2-acetolactate. Actinobacteria, Lactobacillus, Mollicutes, and Acetobacter likely participated in the formation of acetoin-diacetyl-2,3-BDO group (Fig. 1), while a few microorganisms, mainly Acetobacter, might function as a KARI producer.
Species-level diversities of ALS, ALDC, and BDH genes are shown in Fig. 2. L. buchneri (21.0%) and A. pasteurianus (5.8%) might be the major potential ALS producers. L. buchneri (26.5%), Lactobacillus reuteri (11.6%), A. pasteurianus (5.6%), L. fermentum (3.2%), and L. brevis (2.8%) might contribute to the production of ALDC. Diverse species such as Mycoplasma hominis (19.4%), Lactobacillus salivarius (5.3%), Corynebacterium efficiens (3.2%), Rhodococcus jostii (2.1%), and Clostridium lentocellum (1.9%) likely participated in the formation of 2,3-BDO. As for KARI, A. pasteurianus (62.5%) and Streptococcus infantarius subsp. infantarius (6.2%) might be the major potential producers.
The information about the DNA reads encoding enzyme DR was not obtained from the MetaCV analysis result, which indicated that DR-producing microorganisms might be scarce (not abundant enough) in vinegar Pei on the 7th day. To understand the phylogenetic diversity of DR producers, we analyzed vinegar microbiota by the clone library method using amplified fragments of the DR gene with degenerate primers. As shown in Fig. 2, the enzyme DR that metabolizes diacetyl into acetoin was mainly associated with Bifidobacterium indicum and L. fermentum. It is worth mentioning that the degenerate primers used in this study lack some specificity since some strains known to contain the DR gene, such as L. helveticus (see Table S1 in the supplemental material), did not amplify with their primers. Thus, the DR variants in vinegar Pei that do not map to known sequences might have resulted in the underestimate of DNA reads encoding enzyme DR.
According to the metagenomic sequencing results, A. pasteurianus, L. buchneri, L. fermentum, and L. brevis were determined to be potential functional producers for acetoin from 2-acetolactate in vinegar microbiota (Fig. 2). Thus, we used the culture-dependent method to isolate these species from vinegar Pei. A total of 112 isolates, including 53 presumed AAB and 59 presumed LAB were picked up from the plates and purified at least two times by streaking for isolation. Based on physiological and biochemical tests, 35 AAB strains and 30 LAB strains were selected to be identified at the species level by 16S rRNA gene sequencing. Phylogenetic trees based on the 16S rRNA gene and phenotypic characteristics of AAB and LAB in vinegar Pei are shown in Fig. 3 and and44.
All of the AAB isolates were clustered into 16 groups (Fig. 3). Six isolates were clustered together with A. pasteurianus SKU1108, and 3 isolates were identical to Acetobacter pomorum. Phenotypic tests showed that there is a high degree of variability for almost all the traits considered among A. pasteurianus strains. Gluconacetobacter intermedius 1-6 and Gluconacetobacter xylinus (1-15 and 1-17) could produce bacterial cellulose. LAB showed great diversity with 15 groups belonging to 8 Lactobacillus species, including L. fermentum, L. pontis, L. casei/paracasei, L. hilgardii, L. buchneri, L. brevis, L. plantarum, and L. helveticus (Fig. 4).
Acetoin-producing functions of the isolated A. pasteurianus G3-2, L. brevis 4-22, L. fermentum M10-3, and L. buchneri F2-5 were evaluated in mono- and coculture fermentation experiments (Table 1). Biomasses of A. pasteurianus G3-2, L. buchneri F2-5, L. brevis 4-22, and L. fermentum M10-3 in all of the mono- and coculture fermentation experiments were on the same order of magnitude (1012 copies/ml), indicating the growth of these cells in the flask (Table 1).
Compared to that in the control group (3.2 ± 0.7 mg/liter), the acetoin concentration in the monoculture of A. pasteurianus G3-2 (3.6 ± 1.7 mg/liter) did not significantly increase in the MRS-ethanol medium (P > 0.05). L. fermentum M10-3 in the monoculture significantly enhanced its acetoin production (103.3 ± 9.9 mg/liter) (P < 0.05), while acetoin concentrations in the monocultures of L. brevis 4-22, and L. buchneri F2-5 were not significantly different from that in the control group (P > 0.05). The acetoin concentrations in two cocultures (L. brevis 4-22 plus A. pasteurianus G3-2, and L. fermentum M10-3 plus A. pasteurianus G3-2) were significantly higher than those in the LAB monocultures (P < 0.05). However, L. buchneri F2-5 did not produce more acetoin when coinoculated with A. pasteurianus G3-2 (P > 0.05).
Compared with that in the monoculture of A. pasteurianus G3-2 (16.0 ± 0.2 g/liter), the acetate concentration was significantly enhanced in the cocultures of L. fermentum M10-3 plus A. pasteurianus G3-2 (24.1 ± 1.0 g/liter) (P < 0.05). However, acetate production in the coculture of L. brevis 4-22 plus A. pasteurianus G3-2 (6.3 ± 1.7 g/liter) was significantly lower than that in the monoculture of A. pasteurianus G3-2 (16.0 ± 0.2 g/liter) (P < 0.05).
All three LAB strains tested in this study produced lactate in the MRS-ethanol medium, while only 0.1 g/liter lactate was detected in the monoculture of A. pasteurianus G3-2 (Table 1). Lactate concentrations were increased in the cocultures of L. brevis 4-22 plus A. pasteurianus G3-2 (3.7 ± 0.2 g/liter) and L. buchneri F2-5 plus A. pasteurianus G3-2 (3.6 ± 0.3 g/liter); however, they were decreased in the cocultures of L. fermentum M10-3 plus A. pasteurianus G3-2 (0.5 ± 0.0 g/liter).
Effects of functional microbe augmentation on the acetoin content in vinegar Pei are shown in Fig. 5. In the bioaugmentation experiment, the temperature of vinegar Pei in all groups varied between 44 to 47°C (data not shown). After 24 h of fermentation, the acetoin contents in all bioaugmented vinegar Pei were lower than that in the control group (Table 1). The acetoin content in the vinegar Pei augmented with A. pasteurianus G3-2 plus L. fermentum M10-3 (3.3 ± 0.7 mg/g Pei) was significantly lower than that in the control group (6.2 ± 0.7 mg/g Pei) (P < 0.05). After 72 h of fermentation, the acetoin contents in vinegar Pei that was augmented with A. pasteurianus G3-2 plus L. brevis 4-22 (12.5 ± 1.2 mg/g Pei), L. fermentum M10-3 (9.3 ± 1.4 mg/g Pei), and L. buchneri F2-5 (15.4 ± 2.6 mg/g Pei) were significantly increased compared with that in the control group (6.8 ± 0.1 mg/g Pei) (Fig. 5).
The multispecies microbial community formed through centuries of repeated batch AAF accounts for the flavor quality of cereal vinegar. However, the specific vinegar microorganisms responsible for the flavor formation have not yet been identified. Elucidation of the molecular mechanisms underlying the effects of vinegar microbiota has been complicated by difficulties in linking the metabolic functions associated with the microbial community as a whole to individual microorganisms and activities. In this study, we linked the relative abundances of members of the microbial community, estimated on the basis of annotated read counts in the metagenome, to KEGG-based reconstructed diacetyl/acetoin metabolic pathway with the objective of identifying potential functional producers of acetoin. Various microorganisms that originated from the starter, the raw materials, and the environment of AAF were detected. Most taxa in Fig. 1, including archaeota (Crenarchaeota and Euryarchaeota), have not been reported as acetoin producers in vinegar microbiota in previous studies due to the inability to isolate them by a culture-dependent method. After the AAF phase is finished, the microorganisms in raw vinegar, including some conditional pathogenic bacteria (e.g., enterobacteria and Rothia) are killed by decoction (5). It is also worth noting that roughly half of the genes identified from whole-community metagenomic sequencing encode products of unknown function, and existing functional annotations are often incomplete or inaccurate (26). In this study, Mycoplasma hominis was presumed to be the major source of BDH activity (19% of sequence reads in the total BDH enzyme read) (Fig. 2), yet Mycoplasma species are known to be very fastidious commensal bacteria and pathogens at the mucosa. Since Mycoplasma strains belong to the group of low-GC, Gram-positive bacteria, it is probable that these sequences belong to microorganisms such as lactic acid bacteria and bacilli. Thus, as the information in the GenBank and KEGG reference databases is improved, the taxonomic and functional assignment of metagenomic reads in this study will be more accurate.
According to the results of MetaCV analysis, no DNA read encoding DR enzyme was detected, which indicated that DR-producing microorganisms might be scarce (not abundant enough) in vinegar Pei on the 7th day. However, the numbers of DNA reads encoding ALS and ALDC were large. This result demonstrated that a large proportion of acetoin in vinegar Pei on the 7th day might be biosynthesized directly from 2-acetolactate. Actinobacteria, Lactobacillus, Mollicutes, and Acetobacter likely participated in the formation of acetoin, diacetyl, and 2,3-BDO (Fig. 1). Further analysis of the species-level diversities of ALS and ALDC genes revealed that A. pasteurianus and 4 Lactobacillus species (L. buchneri, L. reuteri, L. fermentum, and L. brevis) might be the major potential producers of acetoin from 2-acetolactate in vinegar microbiota. Besides Lactobacillus, Mollicutes, and Acetobacter, many genera in Enterobacteriales (e.g., Klebsiella, Enterobacter, and Serratia), Cyanobacteria (e.g., Synechocystis), and Clostridia (e.g., Clostridium) have been reported for their ability to produce acetoin and 2,3-BDO (11, 27,–29).
According to the annotation results of the metagenomic reads, some major acetoin producers, including A. pasteurianus G3-2, L. buchneri F2-5, L. fermentum M10-3, and L. brevis 4-22, were isolated from vinegar Pei by a culture-dependent method. Via mono- and coculture fermentation experiments, we found little acetoin in the monocultures of A. pasteurianus G3-2 and L. brevis 4-22, but it existed in the coculture of A. pasteurianus G3-2 plus L. brevis 4-22 (Table 1). In the study of Moens et al. (30), A. pasteurianus 386B oxidized lactic acid into acetoin and acetic acid in cocoa pulp simulation medium. In this study, there was little lactic acid in the MRS-ethanol medium supplemented with 5% ethanol (Table 1), which might be the reason that acetoin was hardly formed in the monoculture of A. pasteurianus G3-2. When coinoculated with L. fermentum M10-3, A. pasteurianus G3-2 likely utilized the lactic acid produced by Lactobacillus strains to produce acetoin (Table 1). In the three isolated Lactobacillus strains, L. fermentum M10-3 produced more acetoin than the other two Lactobacillus strains (L. brevis 4-22 and L. buchneri F2-5) in coculture with A. pasteurianus G3-2 in the MRS-ethanol medium (Table 1). However, it could not be concluded that L. fermentum M10-3 was the only major acetoin producer in vinegar Pei, because of the inability to isolate many other microorganisms by a culture-dependent method and the metabolic difference of L. fermentum M10-3 in vitro and in situ. It was found that the acetoin productions of the same bacterial strains were inconsistent between MRS-ethanol medium and vinegar Pei because of the differences in nutritional composition and environmental conditions (Table 1 and Fig. 5).
In this study, a representative single sample, the Pei on the 7th day of AAF, was collected to determine metagenomic sequencing, acetoin metabolic pathway reconstruction, and a bioaugmentation experiment. It stands to reason that the metabolic role of any given species in the context of the community may vary along with fermentation time. Our previous study showed different patterns of community assembly in different AAF stages of Zhenjiang aromatic vinegar (5). For example, the relative abundance of Lactobacillus in the vinegar microbiota dramatically increased on the first day of AAF and then decreased gradually. However, the relative abundance of Acetobacter kept increasing during the whole fermentation process (5). Thus, the bubble size representing the relative abundance of an enzyme gene in Fig. 1 and the species assembly of the enzyme gene in Fig. 2 might change along with the fermentation time. It will be of interest in the future to explore the successions of taxonomic distribution and enzyme abundance in the diacetyl/acetoin metabolic pathway in the vinegar microbiota of Zhenjiang aromatic vinegar. Otherwise, as reported previously, endpoint metagenomes, which just tell us which organisms are present in the actual sample and which have been present at some point in the process but may have died and left their DNA, are traces in the sample (31). As such, unraveling the dynamic expression of genes in vinegar microbiota by metatranscriptomic and metaproteomic approaches is ongoing in further studies.
To conclude, we provided a collection of genes from the known diacetyl/acetoin pathway by using deep sequencing of vinegar microbiota, bridging an important gap, to accurately assess the flavor-producing potential of complex microbial communities from omics-derived data. Furthermore, according to the information on potential acetoin producers, we obtained A. pasteurianus, L. buchneri, L. fermentum, and L. brevis strains from vinegar Pei with selective culture media. The acetoin-producing functions of these strains were evaluated in vitro. Last, the acetoin-producing functions of vinegar microbiota were regulated in situ by augmentation of A. pasteurianus G3-2 plus L. brevis, L. fermentum M10-3, and L. buchneri F2-5 in the vinegar Pei.
This work was supported by two grants from the National Nature Science Foundation of China (no. 31271922 and 31530055), three grants from the High Tech Development Program of China (863 Project) (no. 2012AA021301, 2013AA102106, and 2014AA021501), and a grant from the Ministry of Education of the People's Republic of China (JUSRP51516).
Supplemental material for this article may be found at http://dx.doi.org/10.1128/AEM.01331-16.