Fermented grain samples from the first round of SFBJ pit fermentation (30 days) were collected from three distilleries in three regions: Songzi, Hubei (D1, D2); Shennongjia, Hubei (SNJ-1, SNJ-2); and Maotai town, Guizhou (GZ-1, GZ-2) (Supplementary Table S1). XQBJ fermented grains (sample X) were collected from Songzi, Hubei. For each fermentation pit, 2 kg of fermented grains were collected from the upper, middle, and lower layers and then thoroughly mixed to form the composite sample for analysis. Samples were stored at −20°C or 4°C for subsequent flavor and microbial analysis. DNA was extracted from fermented grains with high n-propanol content, and Lentilactobacillus diolivorans quantification was performed. Stacking-fermented (48 h) samples and Aged Zaopei (distilled fermented grains from XQBJ pit fermentation) were obtained from Hubei Baiyunbian Liquor Industry Co., Ltd. (Songzi, China).

Limosilactobacillus panis strain CGMCC 1.3925 was obtained from China General Microbiological Culture Collection Center. The LAB strains L. diolivorans ZX6, ZD9, ZD92, and ZD93 were isolated and identified from Baijiu fermented grains in this study. S. cerevisiae SCY62 was isolated from SFBJ fermented grains and stored in our laboratory. Rhizopus Qu (Angel Rhizopus), used as a saccharifying agent in XQBJ fermentation, was purchased from Angel Yeast Co., Ltd. (Yichang, Hubei, China).

MRS medium (Hopebio Company, Qingdao, China), containing 10 g/L peptone, 8 g/L beef extract, 4 g/L yeast extract, 20 g/L glucose, 5 g/L sodium acetate, 1 g/L Tween-80, 2 g/L dipotassium phosphate, 2 g/L ammonium citrate, 0.2 g/L magnesium sulfate, 0.04 g/L manganese sulfate, and adjusted to pH 6.5 ± 0.2, was used for culturing and enumerating the total number of LAB. Yeast extract-peptone-dextrose (YPD) medium, consisting of 20 g/L glucose, 20 g/L peptone, and 10 g/L yeast extract, was used for yeast culturing. Rose Bengal Agar (Hopebio Company, Qingdao, China) was employed for culturing and counting yeast populations.

Sorghum extract medium (SS) was used for fermentation and enrichment culture preparation. The sorghum extract preparation method was previously described in the literature (Deng et al., 2020). The obtained sorghum extract was diluted with water to achieve a final sugar concentration of 12 °Bx and adjusted to a pH of 5.8 using a 1:1 mixture of lactic acid to acetic acid. Fermentation media were sterilized at 121°C for 20 min. The modified MRS (SMRS) enrichment medium, a 1:1 mixture of SS and MRS medium with a pH of 6.1 ± 0.1, was used for n-propanol-producing microorganism enrichment and isolation. Nystatin and ampicillin trihydrate (Macklin Biochemical Technology Co., Ltd., China) were added at concentrations of 100 mg/L and 80 mg/L, respectively. Xiaoqu Baijiu solid fermentation medium (XBSM) consisted of 40% (w/w) fresh steamed sorghum (sorghum cooked and cooled with pH 6.95) and 60% (w/w) sterilized aged Zaopei (with pH 3.75). This medium was used for solid-state mixed fermentation of yeast and LAB.

To evaluate n-propanol production by L. diolivorans in liquid co-culture systems, S. cerevisiae SCY62 and L. diolivorans ZX6 were first cultured separately under the following conditions: SCY62 in YPD medium at 28°C with 180 rpm shaking for 24 h, and ZX6 in MRS medium under static conditions at 30°C for 72 h. Mixed-culture fermentations were conducted in 250 mL triangular flasks containing 150 mL of SMRS medium. The inoculation ratios (cells/cells) of S. cerevisiae SCY62 to L. diolivorans ZX6 were 1:0, 1:1, 1:0.1, and 1:0.01, respectively. The initial cell concentration of S. cerevisiae SCY62 in fermentation medium was 1 × 106 cells/mL. Static cultures were maintained at 30°C for 6 days.

Five hundred milliliters of plastic bottles were used as containers for solid-state fermentation. XQBJ solid-state fermentation was conducted using an XBSM medium. Twelve XBSM medium samples were inoculated with 5% (w/w) of Rhizopus Qu, 5 × 105 CFU/g of S. cerevisiae SCY62, and varying amounts of L. diolivorans ZX6 (0, 104, 105, and 106 CFU/g). The fermentation material was initially fermented by stacking in a plastic box at 28°C for 24 h, then transferred to 500 mL plastic bottles for anaerobic fermentation at 28°C for 10–20 days.

To simulate the pit fermentation of SFBJ, fermented grains that had completed stacking fermentation were collected from the Baijiu distillery, individually packed into 500 mL plastic bottles, and incubated. Five fermented grain samples from the first round of SFBJ stacking fermentation at Baiyunbian Liquor Industry Co., Ltd. were randomly selected and separated. Five groups labeled Pit No. 1 through No. 5 were established, each containing 15 parallel plastic bottles. Fifteen kilograms of mixed samples were taken from each stacking fermentation mash and distributed equally among the 500 mL plastic bottles in the five groups. The bottles were sealed and incubated at 28°C for 28 days. Gas chromatography and qPCR analyses were conducted on samples from each fermentation pit at 0, 7, 14, 21, and 28 days. Three sample bottles were taken at each time point for analysis. Additionally, these samples were used for LAB isolation (refer to Section 2.6).

The sample X characterized by its high n-propanol content in XQBJ was selected and inoculated into 150 mL of MRS, SS, and SMRS liquid triangular flask media at a 5% (w/w) inoculation rate. Static cultures were maintained at 30°C for 5 days. The sample with the highest n-propanol content was serially diluted and plated onto SMRS agar plates containing 100 mg/L of nystatin. Plates were incubated at 30°C for 3 days, and single colony morphology was observed. Forty single bacterial colonies with distinct morphological characteristics were selected for 16S rRNA gene sequence analysis and strain identification. Twenty bacterial strains representing different species were selected and inoculated into 250 mL Erlenmeyer flasks containing 150 mL of SMRS medium. Static cultures were maintained at 30°C for 7 days. The strain exhibiting the highest n-propanol production ability was selected, and its morphological characteristics were captured using scanning electron microscopy.

Total DNA was extracted from fermented grains following the method described in the literature (Liu et al., 2017). The genome sequence of L. diolivorans (GenBank GCA_007992335.1) was obtained from the National Center for Biotechnology Information (NCBI) database. Species-specific primers for L. diolivorans were designed based on the propanediol dehydratase large subunit pduC gene (GenBank Accession No. WP 004105132.1). We designed one primer to target a DNA sequence encoding a species-specific loop region absent in orthologous proteins, and positioned the second primer in a nearby low-similarity region to maintain the efficiency of the qPCR process. This dual strategy, exploiting both structural divergence and sequence dissimilarity, maximized specificity and amplification robustness. PCR was used to validate primer specificity using genomic DNA from the target organism and non-target organisms as templates (Du et al., 2020). A standard curve was generated as previously described in the literature (Du et al., 2020).

Two fermented grain samples exhibiting the highest n-propanol content after 7–14 days of bottle pit fermentation were selected from five stacked material samples for LAB separation. Following serial dilution of the fermented grains, LAB colonies were isolated by plating appropriate dilutions onto SMRS agar plates containing nystatin. Plates were incubated at 30°C for 3 days, and 120 colonies with distinct colony morphologies from different dilutions were selected and purified through repeated streaking. These purified colonies were numbered and used for rapid L. diolivorans identification and systematic bacterial classification.

For rapid L. diolivorans identification, the Ezup Column Bacteria Genomic DNA Purification Kit (Sangon Biotech, Shanghai, China) was used to extract DNA. Following PCR amplification with specific primers, detection was performed using 1% agarose gel electrophoresis and subsequent observation with an automated gel imaging analysis system.

For seed cultures, L. diolivorans ZX6 was grown in MRS medium under static conditions at 30°C for 72 h. Then, different fermentation media were inoculated at 2% (v/v) for the study. The growth and metabolic curves of L. diolivorans ZX6 were measured in MRS and SMRS media under static culture conditions for 10 days. MRS media with varying pH values (3.5, 4.0, 4.5, 5.0, 5.5, 6.0, and 6.5) were adjusted using a mixed acid solution (lactic acid: acetic acid = 1:1) and sodium hydroxide. To investigate the effects of pH and temperature on L. diolivorans ZX6 growth, experiments were conducted in an MRS medium at different pH or temperatures (20, 25, 28, 30, 32, 34, 37, 40, and 42°C) under static culture conditions at 30°C for 24 h. The impact of shaker speed (0, 50, 180 rpm) on n-propanol production by L. diolivorans ZX6 was evaluated in SMRS medium at 30°C for 7 days.

MRS media containing varying glucose concentrations (2%, 5%, 9%, 12%, 15%, 20%, 30%, and 35% w/v) were used to assess the impact of glucose concentration on n-propanol production by L. diolivorans ZX6. Fermentation experimental media with different carbon sources were prepared by replacing glucose in MRS medium with an equivalent amount (5% w/v) of sucrose, fructose, and xylose, respectively. All fermentations were conducted under static culture conditions at 30°C for 7 days.

L. diolivorans ZX6 was inoculated (2% v/v) into MRS medium containing individual sugars (5% w/v each of glucose, sucrose, fructose, and xylose) as the sole carbon sources. Fifty milliliters of each culture were collected during the n-propanol index production period (5 d) and centrifuged to obtain the sediment, frozen in liquid nitrogen for 30 min, and immediately transferred to −80°C for storage. Three independent repeat experiments were conducted for each treatment group. RNA extraction, quantification, and sequencing were performed by Shanghai Majorbio Bio-pharm Biotechnology Co., Ltd. (Shanghai, China).

The high-quality sequencing data were mapped to the L. diolivorans DSM 14421 genome (GenBank: GCA_001434255.1), and the original data were uploaded to NCBI (BioProject: PRJNA1225471). The expression level and quantify abundances of each gene were calculated based on previous literature (Chen et al., 2023). The pathways and enzymes of flavor metabolism related to n-propanol were obtained from the KEGG database and previous literature (Gao et al., 2024).

To analyze higher alcohol, ethanol, and acetic acid content in fermented broth or fermented grains, 50 g or 50 mL samples were subjected to pretreatment. Distillation was employed as a pretreatment method to quantify these compounds (Cao et al., 2010). Higher alcohols, ethanol, and acetic acid were analyzed using an Agilent 7820 gas chromatograph equipped with flame ionization detection (Agilent, USA) (López-Vázquez et al., 2010). A J&W CP-Wax 57 CB Acidic column (50 m × 0.25 mm × 0.2 μm, Agilent, USA) was used.

To analyze propionic acid and lactic acid concentrations, fermentation broths were centrifuged at 12,000 rpm for 10 min to remove cells. The supernatant was filtered through a 0.22 μm syringe filter and mixed with methanol in a 1:2 ratio. The mixture was shaken and stored at −20°C overnight to precipitate proteins and polysaccharides. Following this incubation period, the solution was centrifuged at 12,000 rpm for 10 min, filtered through a 0.22 μm syringe filter, and transferred to a liquid phase bottle for detection. High-performance liquid chromatography analysis was performed using an Agilent 1260 Infinity II with an Agilent C18 column, following the method described in a previous study (Sugiura et al., 2023). The peak areas of standard samples were used to determine propionic acid and lactic acid concentrations in the fermentation samples.

Reducing sugars were assayed using the 3,5-dinitrosalicylic acid method (Solarbio, Beijing, China). pH was measured using a pH meter.

Data were reported as mean ± standard deviation (SD) based on three independent determinations of each sample. Statistical analyses were performed using Microsoft Excel for Windows. Statistical significance was considered at P < 0.05, and results were expressed as mean ± SD.

The raw sequence data of the 16S rRNA amplicon sequencing have been deposited in NCBI with the accession numbers PQ136908, PQ136909, PQ136910, and PQ136911.

Considering the convenience of sampling, we initially selected XQBJ fermented grains with high n-propanol content (sample X) as the material for enrichment culture. Sample X was added to liquid culture media, including MRS, SS, and SMRS, for enrichment cultivation. Subsequently, samples were analyzed for n-propanol content. As shown in Table 1, n-propanol content was relatively low when fermented with MRS and SS media. However, when fermented with SMRS mixed media, the n-propanol content increased significantly to 326.60 mg/L, which was 5.9 and 12.6 times higher than that of MRS and SS media, respectively. This suggests that SMRS medium could effectively enriches microorganisms with high n-propanol production capabilities. The n-propanol content of sample X-N4, supplemented with antifungal nystatin, reached 431.70 mg/L. In contrast, n-propanol was not detected in sample X-N5, supplemented with antibacterial ampicillin. These results indicated that the n-propanol-producing microorganisms in the SMRS fermentation broth were bacteria.

To isolate n-propanol-producing bacteria from sample X-N4, 40 bacterial colonies were selected based on morphological differences and purified through repeated streaking on SMRS plates. These strains, named ZX1 to ZX40, were classified into 9 species of LAB and one species of Acetobacter pasteurianus based on 16S rRNA gene sequence homology analyses. Twenty bacterial strains representing distinct species were inoculated into SMRS liquid medium to evaluate their n-propanol production capacity through fermentation experiments. Among these, strain ZX6 demonstrated the highest n-propanol yield, reaching 3,499.40 mg/L after a 7-day fermentation (Supplementary Table S2). No n-propanol was detected in the fermentation broth of the other 19 bacterial strains. Strain ZX6 was rod-shaped and Gram-positive. Its cell morphology was examined using scanning electron microscopy (Figure 1A). The 16S rRNA gene of strain ZX6 was amplified, resulting in a 1,450 bp single fragment that was subsequently sequenced (GenBank Accession No. PQ136908). BLAST analysis revealed a high similarity (>99.7%) between the ZX6 sequence and the 16S rRNA gene sequence of L. diolivorans LMG 19667^T (AF264701) (Krooneman et al., 2002). A phylogenetic tree was constructed (Figure 1B). Based on colony morphology, morphological characteristics, and 16S rRNA gene sequence analysis, strain ZX6 was identified as L. diolivorans.

To evaluate the n-propanol production capability of L. diolivorans under brewing conditions, co-culture experiments with S. cerevisiae were conducted. Both liquid co-culture results (Supplementary Figure S1A) and solid-state co-culture results (Supplementary Figure S1B) of L. diolivorans ZX6 and S. cerevisiae SCY62 demonstrated a direct positive correlation between n-propanol content and the initial inoculation amount of L. diolivorans ZX6. When L. diolivorans ZX6 and S. cerevisiae SCY62 were mixed in a 1:1 (v/v) ratio, the n-propanol content in the SMRS fermentation broth was 246.01 mg/L. During solid-state fermentation, with an L. diolivorans ZX6 inoculation amount of 10⁶ CFU/g, the n-propanol content in the fermented grain reached 1,762.05 mg/kg. Under solid-state fermentation conditions, high inoculation of L. diolivorans resulted in n-propanol content in laboratory-scale fermented grains equivalent to that in the first round of fermented grains from SFBJ distillery.

Next, qPCR was conducted to quantify L. diolivorans in microbiome samples. Species-specific primers for L. diolivorans were designed: F: 5′-TTTCATTTGGTAAGAAGGATTCTTCTG-3′ and R: 5′-TAATATCACCATTTCCTGATGTGGAC-3′. These primers exhibited 100% exclusivity for the other 12 non-target microorganisms from Baijiu fermented grains (Supplementary Table S3), with no matches to other organisms in the NCBI database. We subsequently developed a qPCR method using these specific primers. The standard curve obtained was y = −3.4462x + 46.017 (R² = 0.99), demonstrating a linear correlation (R² = 0.99) over a range of 65.47 to 6.547 × 108 copies per reaction. This suggests that the qPCR method is suitable for quantifying L. diolivorans in the samples.

Significant differences were observed in the n-propanol content of the five fermented grains from the scaled-down pit fermentation experiment (Figure 2B). Sample No. 2, characterized by its high n-propanol content, was selected for the L. diolivorans bacterial count determination using qPCR. Initial L. diolivorans quantities were relatively low, at 5.51 × 10⁵ copies/g, but rapidly increased to 2.50 × 107 copies/g by the 7th day of fermentation. Subsequently, the bacterial count decreased significantly, reaching 7.08 × 105 copies/g at the end of fermentation (Figure 2A). The quantities of L. diolivorans in sample No. 5 (which had an n-propanol content in the middle range) showed a similar trend, reaching a maximum of 4.98 × 10⁴ copies/g on the 7th day and then gradually decreasing throughout the fermentation process. This decline might be attributed to the accumulation of acidity and alcohol content in the later stages of fermentation, which inhibits L. diolivorans growth. n-Propanol and ethanol concentrations in the fermented mash increased rapidly in the early stages of fermentation, reaching their peak by the 14th day and remaining relatively constant thereafter (Figures 2B, C). The pH of the fermented grains decreased significantly during the first 2 weeks of fermentation, from around pH 4.5 to pH 3.7, and then stabilized in the last 2 weeks (Figure 2D). Changes in the L. diolivorans bacterial count within the fermented grains generally aligned with the increasing trend of n-propanol content. However, the n-propanol content typically reached its peak around the 14th day of fermentation, exhibiting a slight lag compared to the peak L. diolivorans count.

One hundred and seven LAB strains, numbered ZD1 to ZD107, were isolated and purified from SFBJ fermented grains with high n-propanol content. Three strains with positive gel electrophoresis bands were identified: ZD9 (GenBank Accession No. PQ136909), ZD92 (GenBank Accession No. PQ136910), and ZD93 (GenBank Accession No. PQ136911). BLAST analysis revealed that these three strains shared high similarity (>99.7%) with the 16S rRNA gene sequence of L. diolivorans LMG 19667^T (AF264701), confirming their classification as L. diolivorans. This demonstrated the high reliability of the specific primer method for rapid L. diolivorans identification. 16S rRNA gene sequence analysis and homology alignment classified the 107 LAB strains into 7 genera and 12 species (Supplementary Figure S2A). The three most prevalent LAB species were L. buchneri, L. panis, and Lactobacillus pontis, accounting for 57.94%, 12.15%, and 10.28%, respectively. The other 9 species accounted for < 5%, and L. diolivorans accounted for only 2.80% (Supplementary Figure S2B). Fermentation tests of the three L. diolivorans strains from SFBJ fermented grain revealed n-propanol contents exceeding 2,400 mg/L, similar to L. diolivorans ZX6. However, no significant n-propanol production was detected in samples fermented by other LAB (Supplementary Table S4). This indicated that L. diolivorans was the primary microorganism isolated from SFBJ fermented grains with the capacity to synthesize high concentrations of n-propanol.