Microbial communities and volatile metabolites in different traditional fermentation starters used for Hong Qu glutinous rice wine

Accepted Manuscript Microbial communities and volatile metabolites in different traditional fermentation starters used for Hong Qu glutinous rice wine

Zi-Rui Huang, Wei-Ling Guo, Wen-Bin Zhou, Lu Li, Jia-Xin Xu, Jia-Li Hong, Hui-Peng Liu, Feng Zeng, Wei-Dong Bai, Bin Liu, Li Ni, Ping-Fan Rao, Xu-Cong Lv PII: DOI: Reference:

S0963-9969(18)30971-2 https://doi.org/10.1016/j.foodres.2018.12.024 FRIN 8149

To appear in:

Food Research International

Received date: Revised date: Accepted date:

30 May 2018 4 November 2018 20 December 2018

Please cite this article as: Zi-Rui Huang, Wei-Ling Guo, Wen-Bin Zhou, Lu Li, Jia-Xin Xu, Jia-Li Hong, Hui-Peng Liu, Feng Zeng, Wei-Dong Bai, Bin Liu, Li Ni, Ping-Fan Rao, Xu-Cong Lv , Microbial communities and volatile metabolites in different traditional fermentation starters used for Hong Qu glutinous rice wine. Frin (2018), https://doi.org/ 10.1016/j.foodres.2018.12.024

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Microbial communities and volatile metabolites in different traditional fermentation starters used for Hong Qu glutinous rice wine

Zi-Rui Huanga,d, 1 , Wei- Ling Guoa,d,1 , Wen-Bin Zhoua,d, Lu Lia, Jia-Xin Xua,d, Jia-Li Honga,d,

PT

Hui-Peng Liua,d, Feng Zenga,d, Wei-Dong Baic, Bin Liua,d,* [email protected], Li Nib,

RI

Ping-Fan Raob, Xu-Cong Lva,b,c,d,* [email protected]

a

SC

National Engineering Research Center of JUNCAO Technology, Fujian Agriculture and Forestry

NU

University, Fuzhou, Fujian 350002, P. R. China b

MA

Institute of Food Science and Technology, College of Biological Science and Technology, Fuzhou

University, Fuzhou, Fujian 350108, P. R. China

College of Light Industry and Food Science, Zhongkai University of Agricultural Engineering,

Guangzhou 510225, P. R. China d

ED

c

*

AC C

China

EP T

College of Food Science, Fujian Agriculture and Forestry University, Fuzhou, Fujian 350002, P. R.

Corresponding authors at: National Engineering Research Center of JUNCAO Technology, Fujian

Agriculture and Forestry University, Fuzhou, Fujian Province 350002, People’s Republic of China.

1

Co-first author: Zi-Rui Huang and Wei-Ling Guo contributed equally to this study.

ACCEPTED MANUSCRIPT ABSTRACT Hong Qu glutinous rice wine (HQGRW), as one of the most typical representatives of Chinese rice wine, is generally brewed from glutinous rice by adding two traditional wine fermentation starters—Hong Qu (HQ) and Bai Qu (BQ). The objective of this study was to determine the

PT

microbial communities and volatile metabolites of different traditional fermentation starters for HQGRW, and elucidate the potential correlation between microbiota and volatile metabolites. Both

RI

heatmap and principal component analysis (PCA) revealed the significant variances in volatile

SC

profiles among different wine starters. Microbiological analysis based on high-throughput

NU

sequencing (HTS) technology demonstrated that both of bacterial and fungal communities varied

MA

significantly in different starters. HQ was dominated mainly by bacteria of Bacillus ginsengihumi (20.17%), Pantoea sp. (10.39%), Elizabethkingia sp. (5.52%), Streptococcus sp. (5.03%)

ED

Brevundimonas sp. (3.03%), Rickettsia prowazekii (2.94%), Thermus thermophilus (2.54%), Bacillus amyloliquefaciens (1.48%), Bacillus aryabhattai (1.42%); fungi of Monascus purpureus

EP T

(39.7%), Aspergillus niger (27.35%), Xeromyces bisporus (8.39%), Aspergillus penicillioides (6.89%), Aspergillus flavus (2.33%) and Pichia farinose (0.79%). By contrast, BQ contained much

AC C

higher proportions of bacteria of Lactococcus lactis (10.45%), Lactobacillus brevis (9.99%), Pediococcus pentosaceus (8.29%), Weissella paramesenteroides (6.69%), Lactobacillus fermentum (4.83%), Gluconobacter thailandicus (3.93%), Lactobacillus alimentarius (3.59%), fungi of Rhizopus arrhizus (31.47%), Saccharomycopsis fibuligera (27.86%), Aspergillus niger (20.81%), Issatchenkia orientalis (3.79%), Saccharomycopsis malanga (3.15%), Clavispora lusitaniae (2.29%), Candida tropicalis (1.47%), Saccharomyces cerevisiae (1.11%) and Rhizopus microsporus (0.57%). Furthermore, core functional microbiota that might contribute to volatile flavour

ACCEPTED MANUSCRIPT development was explored through Spearman's correlation-based network analysis. Lactobacillus brevis, Lactobacillus alimentarius, Lactobacillus plantarum and Aspergillus niger were found to be strongly associated with acid compounds (FDR adjusted P < 0.01), while Pichia sp., Candida sp., Monascus purpureus, Lactobacillus brevis and Lactobacillus alimentarius were positively

PT

correlated with concentrations of aromatic esters associated with fruity and floral notes (FDR adjusted P < 0.01), implying that these microorganisms might make significant contributions to the

RI

flavour of rice wine. These findings demonstrated that the aromatic quality of HQGRW may be

SC

critically influenced by the microbiota in traditional fermentation starters. To conclude, this study

NU

would contribute to the development of novel defined starter cultures for improving the aromatic

MA

quality of HQGRW.

ED

Keywords: Hong Qu glutinous rice wine; traditional fermentation starters; microbial community;

AC C

EP T

volatile metabolites; high-throughput sequencing

ACCEPTED MANUSCRIPT 1. Introduction There are three most famous brewing wines in the world (including rice wine, grape wine and beer), of which the production of Chinese rice wine (huáng jiǔ) has existed for more than 4,000 years (Chen and Xu, 2010; Wang et al., 2014). As one of the most typical representatives of

PT

Chinese rice wine, Hong Qu glutinous rice wine (HQGRW) is generally brewed from glutinous rice with the addition of two traditional fermentation starters—Hong Qu (HQ) (also called red yeast rice)

RI

and Bai Qu (BQ) (also called Yao Qu) (Park et al., 2016). The consumption of HQGRW has

SC

increased in recent years, because it possesses a bright red color and multiple health-promoting

NU

functions due to the involvement of Monascus spp. during the traditional fermentation (Park et al.,

MA

2016; Luo et al., 2018). Starters of HQ and BQ are important for saccharification, fermentation and flavor-generation in the production of HQGRW. Nonetheless, until now, both of them are still made

ED

based on empirical knowledge under non-sterile environmental conditions, which often lead to the inconsistency of quality in terms of taste and flavour between different batches. Application of

EP T

defined starter cultures is one of the most effective approaches to standardize fermentation process and stabilize the aromatic quality of wine. Therefore, it is meaningful to explore the key

HQGRW.

AC C

microorganisms responsible for the production of volatile flavour in traditional fermentation of

Numerous studies have been conducted to characterize the bacterial and fungal communities of in traditional fermentation starters for Chinese rice wine (Guan et al., 2012; Lv et al., 2012a). Our previous studies based on traditional and molecular approaches, e.g., culture-dependent method (Lv et al., 2012b) and culture- independent PCR-denaturing gradient gel electrophoresis (DGGE) technique (Lv et al., 2017, 2015), have demonstrated that traditional fermentation starters for

ACCEPTED MANUSCRIPT HQGRW contained various types of microorganisms, such as molds, yeasts, and bacteria. However, both of culture-dependent approach and PCR-DGGE technique had difficulties in distinguishing the species those were present at population densities below 10 3 CFU/g or two orders of magnitude lower than the most abundant members of these communities (Cocolin et al., 2011;

PT

Prakitchaiwattana et al., 2004). Beside, DGGE bands with the same migration position might contain more than one amplicon, leading to the underestimate of microbial diversity (Sekiguchi et

RI

al., 2001). Advances in high-throughput sequencing (HTS) technology allow a deeper and more

SC

precise evaluation of complex microbiota with a reasonably low cost and in a relatively short period

NU

of time (Baltasar et al., 2014; Lee et al., 2017), and has been widely applied to acquire a

MA

comprehensive analysis of the microbial community in varied fermented foods (Liu et al., 2017; Nie et al., 2015; Portillo and Mas, 2016). Nevertheless, so far HTS technology has not been employed

ED

to investigate the microbial communities of different traditional fermentation starters for HQGRW. Previous studies on the volatile flavour of HQGRW have been conducted based on head space -

EP T

solid phase microextraction (HS-SPME) combined with gas chromatography - mass spectrometry (GC-MS), and results showed that the key aroma components were mainly volatile alcohols and

metabolites

AC C

esters (Zheng et al., 2015, 2014). But, the cryptic correlations between microbiota and volatile associated

with

traditional

fermentation

starters

for

HQGRW

remain

under-characterized.

In present study, we attempted to provide a detailed insight in the microflora of traditional fermentation starters for HQGRW using HTS technology, which decreases the limits imposed by culture-dependent approach and PCR-DGGE technique. Furthermore, the analysis of volatile metabolites in traditional fermentation starters was performed by the combination of HS-SPME and

ACCEPTED MANUSCRIPT GC-MS techniques. Finally, the potential correlations between the volatile metabolites and microbiota were uncovered through Spearman's correlation analysis. The results would enhance our understanding of the core microbiota in traditional fermentation starters and their contribution to the fermentation process, which may be helpful for the development of novel defined starter cultures to

PT

improve the wine quality.

RI

2. Materials and methods

SC

2.1. Sample collection

NU

Ten kinds of characteristic wine starters — HQ and BQ, were collected from well-known

MA

distilleries in different geographical areas of Fujian province, China. All the collected samples are exhibited in Fig. S1. The starter samples were ground into powder (100 mesh) and then stored at

ED

-20 °C prior to further analysis.

2.2. Bacterial and fungal total DNAs extraction

EP T

Bacterial and fungal total DNAs were extracted from wine starter samples using PowerSoil® Total DNA Isolation Kit (Mo Bio, Carlsbad, CA) according to the manufacturer's instructions

AC C

without modifications. The concentration of extracted DNA was measured by a NanoDrop 2000 UV–vis spectrophotometer (Thermo Fisher, Wilmington, MA, USA) and also checked by 0.8% agarose gel electrophoresis. The extracted DNA was stored at − 80 °C for further analysis. 2.3. Illumina sequencing of the bacterial and fungal communities Primers 341-F/806-R and ITS5-1737-F/ITS2-2043-R with specific barcodes were used to amplify the V3–V4 region of bacterial 16S rRNA genes and fungal ITS1 regions, respectively (Liu et al., 2018). The methods of the sequencing libraries of bacterial 16S rRNA genes and fungal ITS

ACCEPTED MANUSCRIPT rRNA genes were those we used previously (Liu et al., 2018). The sequencing libraries of bacterial 16S rRNA genes and fungal ITS rRNA genes were generated for high throughput sequencing using a TruSeq® DNA PCR-Free Sample Preparation Kit (Illumina, San Diego, CA, USA). Then both libraries were sequenced on an Illumina HiSeq2500 platform by Novogene Bioinformatics

PT

Technology Co., Ltd. (Beijing, China). 2.4. Bioinformatic analysis

RI

Raw sequencing reads obtained from the Illumina platform were quality- filtered with the

SC

QIIME software (Ver. 1.7) (Caporaso et al., 2010). The bacterial and fungal sequencing reads were

NU

annotated using the SILVA/16S rRNA database and the UNITE database by a QIIME-based

MA

wrapper of RDP-classifier (v.2.2), respectively. For both bacteria and fungi, a 97% identity threshold was set. The bacterial OTU sequences were double checked with the BLAS T search

ED

program to classify them at the species level. Sequences with identity scores greater than 97% were resolved at the species level. The abundance of OTUs was normalized, by using a standard

EP T

sequence number corresponding to the sample with the fewest sequences. The relative abundances of the representative taxa were further visualized with heatmap and hierarchical clustering (with

AC C

complete linkage), by using R software (ver. 3.3.3) with the “pheatmap” package. 2.5. Volatile profiles analysis

The volatile components in wine starter samples were extracted using a 50/30 μm divinyl-benzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS) fiber (Supelco, Bellefonte, PA, USA) and detected by a Trace GC-2010 Ultra gas chromatograph-DSQ II mass spectrometer (Thermo Electron Corporation, Waltham, MA, USA) equipped with an DB-wax capillary column (30.0 m × 0.25 mm × 0.25 μm, Agilent Technology, Santa Clara, CA, USA). Each sample (0.5g)

ACCEPTED MANUSCRIPT was placed in a 15 mL SPME glass vial together with 2 g of sodium chloride, 5 mL of ddH2 O and 10 μL of the internal standard 2-octanol (40.34 mg/L in absolute ethanol). The vial was tightly capped, and then incubated for 15 min at 50 °C to equilibrate and the sample was extracted (50 °C, 30 min). After extraction, the fiber was introduced into the injection port of the GC-MS system

PT

(250 °C, 3 min) and the analytes extracted from the fiber were thermally desorbed. The GC operation conditions were conducted according to previously reported protocols (Luo et al., 2008).

RI

The mass detector was operated in electron impact mode at an ionizing voltage of 70 eV using the

SC

full scan mode (45–400 amu). The ion source temperature was set at 230 °C. The compounds were

NU

tentatively identified by matching the mass spectra with the NIST11 mass spectral database and

MA

Wiley (NY, 320k compounds, Ver. 6.0). Semi-quantification of the volatile compounds was calculated according to the following formula (Mo, Fan, & Xu, 2012):

ED

C(μg/mL) = Ac×Cis/Ais (μg/mL)

where C is the relative concentration of analyzed sample, C is is the final concentration of

internal standard.

EP T

internal standard in sample, Ac is the peak area of analyzed sample, and Ais is the peak area of

AC C

The results were reported in the mean value of three replicates of fermentation starter samples. The volatile metabolite profiles were visualized using heatmap generated through R software (ver. 3.3.3).

2.6. Statistical analysis Principal component analysis (PCA) was performed to analyze the profiles of the microbiota and volatile compounds using SIMCA software (ver. 14.1) (UMETRICS, Sweden). The microbial taxa that differed significantly between HQ and BQ were analysed using STAMP (ver. 2.1.3)

ACCEPTED MANUSCRIPT (http://kiwi.cs.dal.ca/Software/STAMP) and shown by an extended error bar plot. Statistical differences between HQ and BQ were considered significant at FDR < 0.05 (P value corrected < 0.05) (Guo et al., 2018). Spearman's correlations coefficient (r) was calculated by using SPSS (ver. 19.0) (SPSS Inc., USA), |r| > 0.7 with FDR-value (FDR < 0.05) was considered as a robust

and

microbial

community

were

visualized

via

Cytoscape

software

(ver.

3.5.1)

SC

RI

(http://www.cytoscape.org/).

PT

correlation (Wang et al., 2017). The correlations networks between the selected volatile metabolites

NU

3. Results

MA

3.1. Microbial profiles of different traditional fermentation starters The microbial communities of the starter samples were analyzed using PCA to assess their

ED

variation and similarity (Fig. 1). The PCA biplot of bacterial community illustrated the differences between HQ and BQ (Fig. 1A). The first principal component (PC1), the second principal

EP T

component (PC2) and the third principal component (PC3) accounted for 26.6%, 25.4% and 13.0% of the total variation, respectively (Fig. 1A). Marked variability of fungal communities was also

AC C

observed across different starters. The PC1, the PC2 and PC3 accounted for 30.8%, 23.2% and 11.6% of the total variation, respectively (Fig. 1B). As can be observed from principal component plot shown in Fig. 1B (PC1 vs. PC2) (The PC1 vs. PC3 are shown in Fig. S2 and Table S2), traditional starters of BQ (BQ1~BQ5) lie on the left quadrants (the second and third quadrants), where were characterized mainly by S. fibuligera, R. microsporus, I. orientalis, R. arrhizus, M. circinelloides, M. indicus, C. glabrata, etc. For the samples of HQ (HQ1~HQ5), placed on the right quadrants (the first and fourth quadrants), were strongly characterized by M. purpureus, P. farinosa,

ACCEPTED MANUSCRIPT A. penicillioides, Candida sp., M. acetoabutans, X. bisporus, etc. (Fig. 1B). It was indicated that the relative abundance of some bacteria and fungi have significant differences between BQ and HQ (Fig. S3). The relative abundance analysis revealed the dominant bacterial and fungal species in the traditional fermentation starters (Fig. 2 and Table S1). The bacterial results illustrate that HQ was

PT

characterized mainly by the occurrence of uncultured bacterium (31.05%), followed by species ascribed to B. ginsengihumi (20.17%), Pantoea sp. (10.39%), Elizabethkingia sp. (5.52%),

RI

Streptococcus sp. (5.03%), Psychrobacter sp. (3.56%), Brevundimonas sp. (3.03%), R. prowazekii

SC

(2.94%), T. thermophilus (2.54%), B. amyloliquefaciens (1.48%) and B. aryabhattai (1.42%) (Fig.

NU

2A). BQ was mainly dominated by uncultured bacterium (14.23%), followed by species ascribed to

MA

Lactococcus lactis (10.45%), L. brevis (9.99%), P. pentosaceus (8.28%), W. paramesenteroides (6.69%), L. fermentum (4.83%), G. thailandicus (3.93%), Streptococcus sp. (3.78%) and L.

ED

alimentarius (3.59%) (Fig. 2A). In terms of fungal community, HQ was represented mainly by M. purpureus (39.70%), A. niger (27.35%), X. bisporus (8.39%), A. penicillioides (6.89%), A. flavus

EP T

(2.33%) and P. farinosa (1.28%) (Fig. 2B). BQ was characterized mainly by R. arrhizus (31.47%), S. fibuligera (27.86%), A. niger (20.81%), X. bisporus (4.79%), I. orientalis (3.79%) and S. malanga

AC C

(3.15%), while the proportions of M. purpureus, A. penicillioides, A. niger and A. flavus were evidently lower in BQ compared to HQ (Fig. 2B). 3.2. Co-occurrence/exclusion analysis revealed the relationships between different microbes Co-occurrence/exclusion analysis is a useful method to elucidate the correlations and core microorganisms in complex microbial communities. As shown in Fig. 3, one apparent pattern was that lactic acid bacteria (LAB), including L. lactis, W. paramesenteroides, L. fermentum, L. brevis, P. pentosaceus, appeared almost antagonist to B. amyloliquefaciens, Pantoea sp, B. ginsengihumi, B.

ACCEPTED MANUSCRIPT aryabhattai, Enterobacter sp. and other contaminants (Fig. 3A). As for fungal community, R. microsporus, S. fibuligera, I. orientalis and R. arrhizus showed strong exclusion with Candida sp., Moniliella acetoabutans, Trichomonascus ciferrii, A. penicillioides, P. farinosa and M. purpureus (Fig. 3B). Interestingly, Spearman's correlation analysis between bacteria and fungi indicated that B.

PT

ginsengihumi, B. amyloliquefaciens, B. aryabhattai and B. subtilis correlated positively with A. flavus, A. penicillioides, M. purpureus, Blastobotrys adeninivorans, A. niger, P. farinosa and C.

RI

blankii (Fig. 3C). In contrast, L. brevis, L. fermentum, L. lactis, P. pentosaceus and W.

SC

paramesenteroides were found to be negatively associated with A. penicillioides, C. smithsonii, M.

NU

purpureus and P. farinosa (Fig. 3C).

MA

3.3. Volatile metabolites in different traditional fermentation starters The analysis of volatile organic compounds (VOCs) in the different traditional fermentation

ED

starters was performed by SPME-GC-MS. The assessment allowed the identification of 118 volatile compounds, mainly including esters (32), alcohols (22), acids (23), aldehydes (9) and ketones (15)

EP T

(Fig. 4). The main differences in the volatile profiles among different traditional fermentation starters were also highlighted by PCA (Fig. 5). In detail, the PCA score plot showed 32.0%, 18.3%

AC C

and 14.0% variance by PC1, PC2 and PC3, respectively. As shown in Fig. 5A, traditional starters of BQ (BQ1, BQ2, BQ4 and BQ5) lie on the upper quadrants (the first and second quadrants), where were characterized mainly by methyl adipate (C18), pantolactone (E15), (Z)-3-nonen-1-ol (A15 ), benzyl alcohol (A19), globulol (A22), methyl palmitate (C25) and octyl octanoate (H10). The PCA ordination of the volatile profiles and sample variables demonstrated that isoamyl alcohol (A2), (E)-2-octen-1-ol (A11), 2-furanmethanol (A14), 2- methyl-pentanoic acid (B4), 2- methyl-propanoic acid (B8), 2-ethyl- hexanoic acid (B9), octanoic acid (B11), 2-heptenoic acid (B12), 2-octenoic acid

ACCEPTED MANUSCRIPT (B13), nonanoic acid (B14), benzoic acid (B16), isoamyl acetate (C4), phenethyl acetate (C17), 4-pentadecanyl butyrate (C20), isobutyl 2-ethylbutanoate (C22), methyl octylate (C23), 1-octen-3-one (E7), 5-pentyldihydro-2(3H)- furanone (E14), 2-methyl-phenol (F1) and eucalyptol (H3) were strongly correlated with wine starter BQ3, which was distinguished from other starter

PT

samples by the PCA biplot analysis (Fig. 5A and B). Interestingly, volatile profiles among samples of HQ (except for HQ1) were quite similar by principal component plots (PC1 vs. PC2 and PC1 vs.

RI

PC3), as they grouped together. (Z)-8-dodecen-1-ol (A18), ethyl butyrate (C2), ethyl 2-butenoate

SC

(C6), ethyl caprylate (C10), ethyl benzoate (C15), 2-nonen-4-one (E2), 2- heptanone (E3),

MA

furfural (H2) were highly associated with HQ1.

NU

2-octanone (E6), 3-octen-2-one (E10), 5-butyldihydro-2(3H)- furanone (E13), anisole (H1) and

3.4. Statistical correlations between microbiota and volatile metabolites

ED

Correlation analysis was conducted to analyze the interaction between volatile metabolites and microbiota composition (Fig. 6 and Fig. S4). Results showed that different microorganisms

EP T

contributed differently to the volatile flavours. Correlation based network analysis (|r| > 0.7 with P < 0.05) showed that L. brevis exhibited a positive correlation with 1-octanol (A10), 1-nonanol

AC C

(A13), .alpha.-terpineol (A16), .alpha.- methyl-benzenemethanol (A17), 2-methyl-pentanoic acid (B4), 3- methyl-pentanoic acid (B7), 2-octenoic acid (B13), benzeneacetic acid (B18), n-caproic acid vinyl ester, 4-pentadecanyl butyrate (C20) and 2-heptenal (D4) (Fig. 6A). P. pentosaceus showed significant positive correlations with 1-octanol (A10), levomenthol (A12), 1-nonanol (A13), .alpha.-terpineol (A16), .alpha.- methyl-benzenemethanol (A17), 2-octenoic acid (B13), benzeneacetic acid (B18), n-caproic acid vinyl ester (C14), 2-heptenal (D4), decanal (D6) and 3,4-dimethoxytoluene (H8) (Fig.

6A). L. alimentarius was positively correlated

with

ACCEPTED MANUSCRIPT 3-methyl-pentanoic acid

(B7),

2- methyl-propanoic acid

(B8),

2-heptenoic acid

(B12),

9,12-octadecadienoic acid (Z,Z)- (B20), phenethyl acetate (C17), 4-pentadecanyl butyrate (C20) and isobutyl 2-ethylbutanoate (C22) (Fig. 6A). On the other side, L. plantarum showed significant a negative correlation with (Z)-3-nonen-1-ol (A15), benzyl alcohol (A19) and globulol (A22) (Fig.

PT

6A). In terms of fungi, R. microsporus was positively correlated with isoamyl alcohol (A2), (A13),

.alpha.-terpineol

(A16),

.alpha.- methyl-benzenemethanol

RI

1-nonanol

(A17),

SC

2-methyl-pentenoic acid (B4), 3-methyl-pentenoic acid (B7), octanoic acid (B11), 2-octenoic acid

NU

(B13), benzeneacetic acid (B18), ethyl oenanthate (C9), n-caproic acid vinyl ester (C14) and

2-methyl-propanoic

acid

(B8),

MA

4-pentadecanyl butyrate (C20) (Fig. 6B). A. niger exhibited a positive correlation with 2-heptenoic

acid

(B12),

isoamyl

acetate

(C4)

and

ED

2,4-di-t-butylphenol (F3) (Fig. 6B). In particular, M. purpureus exhibited a high positive correlation with ethyl caprylate (C10), ethyl caproate (C8), 3-penten-2-one (E1) and 2-nonen-4-one (E2), but correlation

with

EP T

negative

isoamyl

alcohol

(A2),

1-nonanol

(A13)

and .alpha.- methyl-benzenemethanol (A17) (Fig. 6B). Furthermore, non-Saccharomyces yeasts,

AC C

such as Pichia sp. and Candida sp., showed significant positive correlations with 8-dodecen-1-ol (A18), (Z)-anisole (H1), furfural (H2), geosmin (H7), ethyl butyrate (C2), ethyl 2-butenoate (C6), ethyl stearate (C12), and 2-octanone (E6) (Fig. 6B).

4. Discussion In the present study, a comprehensive investigation of the microbial communities in different traditional fermentation starters for HQGRW was carried out through HTS technology. Compared

ACCEPTED MANUSCRIPT with our previous studies based on PCR-DGGE technique and MALDI-TOF/MS fingerprinting (Lv et al., 2017, 2016, 2013, 2012b, 2012a), 28 bacterial and 13 fungal genera, namely Acinetobacter, Anderseniella,

Bradyrhizobium,

Brevundimonas,

Brochothrix,

Clostridium,

Comamonas,

Elizabethkingia, Enterobacter, Escherichia, Geobacillus, Gluconacetobacter, Gluconobacte, Kluyvera,

Novosphingobium,

Paenibacillus,

Pantoea,

PT

Hyphomicrobium,

Pseudomonas,

Psychrobacter, Rahnella, Rickettsia, Ruminobacter, Solitalea, Sphingomonas, Streptococcus, Vagococcus,

Blastobotrys,

Clavispora,

Cunninghamella,

RI

Thermus,

Eurotiales,

Fusarium,

SC

Geotrichum, Issatchenkia, Lichtheimia, Moniliella, Ogataea, Phoma, Trichomonascus and

NU

Xeromyces, were successfully detected in the traditional fermentation starters for HQGRW by HTS

MA

technology for the first time. The results of comparison indicated that HTS technology have an aptitude for detecting low-abundance microbes, revealing higher microbial diversity than our

ED

previous studies. Our previous results obtained from DGGE bands sequencing of 18S rDNA fragments based on NS1/GCFung primers revealed that some DGGE bands cannot be acc urately

EP T

identified at the species level, which may be due to high genetic homology inside certain fungal genera (Lv et al., 2017). This may explain why the number of fungal species detected in this study

2012b).

AC C

is far larger than our previous study through PCR-DGGE based on 18S rDNA (Lv et al., 2013,

Different types of traditional fermentation starters potentially lead to the distinctive flavour of rice wine. In this study, the communities of bacteria and fungi have exhibited significant d ifferences between HQ and BQ, which may be put down to their raw material diversity as well as the differences of temperature, moisture and micro-habitats during production process. Traditionally, HQ is manufactured through the fermentation of polished round grain rice by inoculating Monascus

ACCEPTED MANUSCRIPT spp.. However, the microbial flora in HQ often showed complexity and diversity, which might be also influenced by the environmental microorganisms during the non-sterile fermentation process (Park et al., 2016). Compared with HQ, BQ is manufactured from raw starchy grains and several Chinese native medicine ingredients based on empirical knowledge. The traditional production of

PT

BQ also involves various microbes, mainly consisting of Rhizopus spp., Aspergillus spp. and yeasts, which can provide rich enzymes and synthesize specific flavour and functional components during

RI

fermentation (Li et al., 2014; Zhang, et al., 2009). Moreover, the different processing technologies

SC

of Hong Qu and Bai Qu are exhibited in Fig. S5.

NU

Following the HTS analysis, it was established that the most predominant bacterial species

MA

present in HQ was B. ginsengihumi, a gram-positive bacteria that can survive under low- moisture and high-temperature conditions (Ma, et al.,2014) and is frequently detected in various traditional

ED

fermented foods, such as Daqu (Wang et al., 2008), nuruk (Song et al., 2013), fermented soya bean (Kiers et al., 2000), etc. Our previous study on the bacterial community of HQ also pointed out that

EP T

several species within the genus of Bacillus were identified in most HQ starter samples (Lv et al., 2017, 2012a; Liu et al., 2018). Several members within the Bacillus genus may play important roles

AC C

in the brewing of some fermented foods, as they can secrete extensive hydrolytic enzymes (including amylases, acid protease, and fibrinolytic enzyme) and facilitate the formation of the flavour of rice wine (Simonen and Palva, 1993). Moreover, Bacillus can also produce a broad range of volatile compounds including pyrazines, aldehydes, ketones and alcohols (Azokpota et al., 2010). In addition to the Bacillus, Pantoea was the second predominant bacterial genus detected in starters of HQ. Currently, Pantoea is usually served as a biological control agent because of its excellent capability to produce antimicrobial substances (Coutinho and Venter, 2009; Walterson et al., 2014).

ACCEPTED MANUSCRIPT Apart from Bacillus and Pantoea, the genera Elizabethkingia, Streptococcus, Psychrobacter, Brevundimonas, Rickettsia, Thermus and some other low-abundant taxa also presented in HQ starters, although some of which were related to spoilage or pathogenic bacteria. BQ contain much higher proportions of lactic acid bacteria (LAB) group (P. pentosaceus, L.

PT

brevis, L. lactis, L. fermentum, L. alimentarius, L. plantarum and W. paramesenteroides) than HQ. LAB group, a kind of probiotics, is considered to be of great benefit to human health. It is also

RI

widely involved in many fermentation processes as part of a common starter culture. The richness

SC

of spoilage or pathogenic bacteria detected in BQ is lower than that of HQ. A possible reason is that

NU

BQ is made with a certain amount of Chinese medicinal herb (eg,. Angelica sinensis, Rhizoma

MA

cimicifugae, Rhizoma kaempferiae, Pericarpium zanthoxyli，Rhizoma nerdostachyos，Cinnamomum cassia, Eucommia ulmoides, Astragalus membranaceus，Fructus gardeniae，Herba menthae, and so

ED

on (Fu and Yang, 2010; Gu, 1996), which contains multifarious antimicrobial substances, creating conditions not only favorable for beneficial bacteria such as LAB but also unfavorable for the

EP T

growth of some potential pathogens (Hui, 2006). Besides, most genus of LAB, especially Lactobacillus, can produce lactic acid and a variety of antimicrobial substances such as bacteriocin

AC C

to inhibit the growth of pathogens and spoilage microorganisms during the brewing process (Cappello et al., 2017; Castellano et al., 2008; Grimaldi et al., 2005). Moreover, LAB plays critical roles in various fermented foods by offering pleasant sensory and nutritional benefits (Leroy and Vuyst, 2005). Our previous study also revealed that LAB experienced a rapid growth during the traditional brewing of HQGRW (Lv et al., 2013), suggesting that it may contribute to the formation of aroma characteristics. The production of esterases, lipases and alcohol acetyl transferases by LAB is highly significant for enhancing the flavour forming attributes of wine (Gammacurta et al.,

ACCEPTED MANUSCRIPT 2018). In recent years, interactions between LAB and yeasts were also described, since some yeast species and LAB may cooperate with each other and contribute to the production of some important aromatic compounds (Benincasa et al., 2015; Sieuwerts et al., 2018). A low ratio of LAB was always associated with good fermentation, whereas relatively high ratios of LAB inevitably resulted

PT

in failed fermentation (Hong et al., 2016). It has been reported that rice wine spoilage was usually associated with the rapid growth of Lactobacillus at the early stage of fermentation (Sami et al.,

RI

1997; Hong et al., 2016), because several species of Lactobacillus may convert carbohydrates to

SC

large amounts of acetic acid through heterofermentation, which acidifies the environment and

NU

finally inhibits the growth of other microbes, including yeasts and molds (Vol, 1999).

MA

Monascus spp., a unique pigment producing filamentous fungi, accounted for a considerable proportion of fungi in HQ. It can produce multiple functional secondary metabolites, mainly

ED

pigments, γ-aminobutyric acid and monacolin K. In addition to functional secondary metabolites, the correlation analysis between microbiota and volatile metabolites also revealed that M. purpureus

EP T

is positively correlated with ethyl caproate, ethyl caprylate, 3-penten-2-one and 2-nonen-4-one. In addition to M. purpureus, the genus Aspergillus, including A. penicillioides, A. niger, A. flavus, X.

AC C

bisporus and P. farinosa, etc., also accounted for a considerable proportion of fungi in HQ. As the predominant filamentous fungi in Daqu, the genus Aspergillus is good at secreting large quantities of a broad range of different enzymes into its environment, which may contribute to the saccharification of the starch in wine mash (Oda et al., 2006; Zheng et al., 2012). It was observed that Rhizopus and Saccharomycopsis genus were found ubiquitously as predominant in BQ. Rhizopus spp. (mainly R. oryzae, R. microsporus and R. arrhizus), with a strong production ability for amylase, is commonly found in various types of amylolytic fermentation starters for rice wine in

ACCEPTED MANUSCRIPT many countries or regions (Choi et al., 2012; Dung et al., 2007; Liu et al., 2018). Some species belonging to the genus of Rhizopus also make great contributions to the flavour formation of rice wine through synthesis of volatile compounds (Bramorski et al., 1998; Christen et al., 2000), although they decrease during the traditional brewing process due to the absence or a limitation of

PT

oxygen and the predominance of S. cerevisiae. The behavior of Rhizopus spp. was observed and its ability to produce volatile compounds during fermentation such as ethanol, 2-methylpropanol and

RI

3-methylbutanol was highlighted (Christen et al., 2000), which was consistent with the results of the

SC

present study.

NU

Surprising was the low- frequency detection of S. cerevisiae in traditional fermentation starters

MA

for HQGRW, which was in contrast with our previous findings using a culture-dependent procedure. S. cerevisiae has always been considered to be the most commonly used yeast species in rice wine

ED

fermentation due to its high efficiency of metabolizing sugars to alcohol (Thanh et al., 2008). During the brewing process of rice wine, the population of Saccharomyces increase dramatically,

EP T

along with the increase of ethanol, which will result in the decrease of microorganisms of alcohol intolerance. Through the application of HTS technology, non-Saccharomyces yeasts were detected

AC C

commonly present in starter samples with certain abundance, including the genera Candida, Saccharomycopsis and Pichia, which were also detected in our previous studies through PCR-DGGE fingerprints (Lv et al., 2012b). More and more researchers have paid attention to the non-Saccharomyces yeasts because their secondary metabolites are closely related to the taste and flavour of wine (Nuñez-Guerrero et al., 2016). Nonetheless, the mechanisms of function and work in non-Saccharomyces yeasts for Chinese rice wine are still poorly understood. Previous study found that non-Saccharomyces yeast of Candida could regulate alcohol content and enhance the

ACCEPTED MANUSCRIPT wine aromatic quality to some extent, by co- fermenting with Saccharomyces (Englezos et al., 2016). S. fibuligera can metabolize native starch to into maltose, dextrin, and glucose because it can produce various enzymes including α-amylase, glucoamylase, acid protease and β- glucosidase to efficiently saccharify starch (Chi et al. 2009; Horváthová et al. 2004). Therefore, S. fibuligera is a

PT

major amylolytic yeast in food fermentation. Previous studies indicated that non-Saccharomyces yeasts of Pichia were positively correlated with volatile acids, alcohols and esters, which was

RI

consistent with our results (Thera et al., 2016; Vestner et al., 2011). In further study, co-cultivation

SC

are needed to verify the interactions of non-Saccharomyces yeasts and S. cerevisiae, and their

NU

contribution to the distinctive flavor of HQGRW.

MA

The present study also indicated constant presence of opportunistic contaminants and pathogenic species (E. coli, Pseudomonas sp., Clostridium sp., Enterobacter sp., F. culmorum, P.

ED

georgiense and A. nomius) in traditional fermentation starters. These microbes detected in the starters with certain abundances may threaten food safety and human health potentially. Therefore,

EP T

the application of defined starter cultures is strongly recommended both for suppressing the growth of the undesirable microbes and for nurturing the functional microbes to improve the hygienic and

AC C

aromatic quality of HQGRW.

In conclusion, this study provided a comprehensive and thorough insight into the correlation between microbial communities and volatile metabolites in traditional fermentation starters used for HQGRW brewing. Results contribute to a better understanding of the traditional fer mentation mechanism, which could be useful for improving the industrial production of HQGRW and ensuring the high quality and safety. Further studies should be performed to understand the interactions between LAB, yeasts and molds to define the most important factors contributing to the

ACCEPTED MANUSCRIPT final flavour of rice wine. Microbial functions and fermentation mechanisms during traditional brewing needs to be performed

using multi-omics approach including metagenomics,

metaproteomics and metatranscriptomics.

PT

Acknowledgments This work was financially supported by National Natural Science Foundation of China (No.

RI

31601466), fund for outstanding young scientific talents of Fujian Agriculture and Forestry

SC

University (XJQ201607), Project Funded by China Postdoctoral Science Foundation (2016T90591),

MA

NU

and Natural Science Foundation of Fujian Province (2016J01095).

Conflict of Interest Statement: The authors declare that the research was conducted in the absence

ED

of any commercial or financial relationships that could be construed as a potential conflict of

AC C

EP T

interest.

ACCEPTED MANUSCRIPT References Azokpota, P., Hounhouigan, J.D., Annan, N.T., Odjo, T., Nago, M.C., & Jakobsen, M. (2010). Volatile compounds profile and sensory evaluation of beninese condiments produced by inocula of Bacillus subtilis. J. Sci. Food Agric, 90(3), 438–444.

PT

Baltasar, M., Rachid, C.T.C.C., Ángel, A., Leite, A.M.O., Peixoto, R.S., & Susana, D. (2014). Impact of next generation sequencing techniques in food microbiology. Curr. Genomics, 15(4),

RI

293–309.

SC

Benincasa, C., Muccilli, S., Amenta, M., Perri, E., & Romeo, F. V. (2015). Phenolic trend and

NU

hygienic quality of green table olives fermented with Lactobacillus plantarum starter culture.

MA

Food Chem, 186, 271–276.

Bramorski, A., Christen, P., Ramirez, M., Soccol, C.R., & Revah, S. (1998). Production of volatile

ED

compounds by the edible fungus Rhizopus oryzae during solid state cultivation on tropical agro-industrial substrates. Biotechnol Lett, 20, 359–362.

EP T

Caporaso J.G., Kuczynski J., Stombaugh J., Bittinger K., Bushman F.D., Costello E.K., … Knight R. (2010). QIIME allows analysis of high-throughput community sequencing data. Nature

AC C

Methods, 7(5), 335–336.

Cappello, M.S., Zapparoli, G., Logrieco, A., & Bartowsky, E.J. (2017). Linking wine lactic acid bacteria diversity with wine aroma and flavour. Int. J. Food Microbiol, 243, 16–27. Castellano, P., Belfiore, C., Fadda, S., & Vignolo, G. (2008). A review of bacteriocinogenic lactic acid bacteria used as bioprotective cultures in fresh meat produced in Argentina. Meat Sci, 79, 483. Chen, S., & Xu, Y. (2010). The influence of yeast strains on the volatile flavour compounds of

ACCEPTED MANUSCRIPT Chinese rice wine. J. I. Brewing, 116(2), 190–196. Chi, Z.M., Chi, Z., Liu, G.L., Wang, F., Ju, L.A., & Zhang, T. (2009). Saccharomycopsis fibuligera and its applications in biotechnology. Biotechnol. Adv, 27(4), 423-431. Choi, J.S., Jung, S.T., Choi, J.H., Choi, H.S., Baek, S.Y., Yeo, & S.H. (2012). Quality characteristics

PT

of Wheat Nuruks by storage conditions of liquid starters using Rhizopus oryzae N174. Korean J. Microbiol. Biotechnol, 40, 319–324.

RI

Christen, P., Bramorski, A., Revah, S., & Soccol, C.R. (2000). Characterization of volatile

SC

compounds produced by Rhizopus strains grown on agro- industrial solid wastes. Bioresource

NU

Technol, 71(3), 211–215.

MA

Cocolin, L., Campolongo, S., Alessandria, V., Dolci, & P., Rantsiou, K. (2011). Culture independent analyses and wine fermentation: an overview of achievements 10 years after first application.

ED

Ann. Microbiol, 61(1), 17–23.

Coutinho, T.A., & Venter, S.N. (2009). Pantoea ananatis: an unconventional plant pathogen. Mol.

EP T

Plant Pathol, 10(3), 325–335.

Dung, N.T.P., Rombouts, F.M., & Nout, M.J.R. (2007). Characteristics of some traditional

130–135.

AC C

Vietnamese starch-based rice wine fermentation starters (men). LWT - Food Sci. Technol, 40,

Englezos, V., Rantsiou, K., Cravero, F., Torchio, F., Ortizjulien, A., Gerbi, V., & Cocolin L. (2016). Starmerella bacillaris and Saccharomyces cerevisiae mixed fermentations to reduce ethanol content in wine. Appl. Microbiol. Biotechnol, 100(12), 5515. Fu, Z.K., & Yang, G.J. (2010). 200 questions of rice wine production. Chemical Industry Press, Beijing.

ACCEPTED MANUSCRIPT Gu, G.X. (1996). Brewing technology. China Light Industry Press, Beijing. Gammacurta, M., Lytra, G., Marchal, A., Marchand, S., Christophe, B.J., & Moine, V. (2018). Influence of lactic acid bacteria strains on ester concentrations in red wines: specific impact on branched hydroxylated compounds. Food Chem, 239, 252–259.

PT

Grimaldi, A., Bartowsky, E., Jiranek, V. (2005). Scree ning of Lactobacillus spp. and Pediococcus spp. for glycosidase activities that are important in oenology. J. Appl. Microbiol, 99,

RI

1061–1069.

SC

Guan, Z., Zhang, Z., Cao, Y., Chen, L., Xie, G., & Lu, J. (2012). Analysis and comparison of

NU

bacterial communities in two types of ‘Wheat Qu’, the starter culture of Shaoxing rice wine,

MA

using nested PCR-DGGE. J. I. Brewing, 118(1), 127–132. Guo, W.L., Pan, Y.Y., Li, L., Li, T.T., Liu, B., & Lv, X.C. (2018). Ethanol extrac t of Ganoderma

ED

lucidum ameliorates lipid metabolic disorders and modulates the gut microbiota composition in high-fat diet fed rats. Food & Function, 9(6), 3419–3431.

EP T

Hong, X., Jing, C., Lin, L., Wu, H., Tan, H., Xie, G., … Lan, W. (2016). Metagenomic sequencing reveals the relationship between microbiota composition and quality of Chinese Rice Wine. Sci.

AC C

Rep, 6, 26621.

Horváthová, V., Šlajsová, K., Šturdík, E., & Šturdík, K., 2004. Evaluation of the glucoamylase Glm from Saccharomycopsis fibuligera IFO 0111 in hydrolysing the corn starch. Biologia, 59(3), 361–365. Hui, Y. H. (2006). Handbook of food science, technology, and engineering. Crc Press, v. 1–4. Kiers, J.L., Van Laeken, A.E., Rombouts, F.M., & Nout, M.J. (2000). In vitro digestibility of Bacillus fermented soya bean. Int. J. Food Microbiol, 60(2), 163–169.

ACCEPTED MANUSCRIPT Lee, S., Lee, S., Singh, D., Ji, Y.O., Jeon, E.J., Ryu, H.S., … Lee, C.H. (2017). Comparative evaluation of microbial diversity and metabolite profiles in doenjang, a fermented soybean paste, during the two different industrial manufacturing processes. Food Chem, 221, 1578–1586.

PT

Leroy, F., & Vuyst, L.D. (2005). Modelling interactions between bacteriocin-producing sausage starter cultures or cocultures and listeria reveals how to improve the efficiency of listeria

RI

killing. Acta Horticulturae, 674, 239–243.

SC

Li, L., Gao, B., Zhang, W.X., Sun, A.M., Zhao, Y.Y., Wu, Z.Y., … Yang, J. (2014). Development of

NU

Panax quinquefolius Yaoqu and Panax quinquefolius Sake and determination of ginsenosides

MA

Rg1, Rb1, and Re in both samples by HPLC–MS/MS. Eur. Food Res. Technol, 239, 137–147. Liu, M., Tang, Y., Zhao, K., Liu, Y., Guo, X., Ren, D., … Zhang, X. (2017). Determination of the

ED

fungal community of pit mud in fermentation cellars for Chinese strong- flavor liquor, using DGGE and Illumina MiSeq sequencing. Food Res. Int, 91, 80–87.

EP T

Liu, Z., Wang, Z., Lv, X., Zhu, X., Chen, L., & Ni, L. (2018). Comparison study of the volatile profiles and microbial communities of Wuyi Qu and Gutian Qu, two major types of traditional

AC C

fermentation starters of Hong Qu glutinous rice wine. Food Microbiol, 69,105–115. Luo, F., Li, L., Wu, Z., Yang, J., Yi, X., & Zhang, W. (2018). Development of new red mold rice and determination of their properties. LWT - Food Sci. Technol, 87, 259–265. Luo, T., Fan, W., & Xu, Y. (2008). Characterization of volatile and semi‐ volatile compounds in chinese rice wines by

headspace solid

phase

microextraction

followed

by gas

chromatography‐ mass spectrometry. J. I. Brewing, 114(2), 172–179. Lv, X.C., Chen, Z.C., Jia, R.B., Liu, Z.B., Zhang, W., Chen, S., & Ni, L. (2015). Microbial

ACCEPTED MANUSCRIPT community structure and dynamics during the traditional brewing of Fuzhou Hong Qu glutinous rice wine as determined by culture-dependent and culture-independent techniques. Food Control, 57, 216–224. Lv, X.C., Jiang, Y.J., Liu, J., Guo, W.L., Liu, Z.B, Zhang, W., & Ni, L. (2017). Evaluation of

PT

different PCR primers for denaturing gradient gel electrophoresis (DGGE) analysis of fungal community structure in traditional fermentation starters used for Hong Qu glutinous rice wine.

RI

Int. J. Food Microbiol, 255, 58–65.

SC

Lv, X.C., Huang, X.L., Zhang, W., Rao, P.F., & Ni, L. (2013). Yeast diversity of traditional alcohol

NU

fermentation starters for Hong Qu glutinous rice wine brewing, revealed by culture-dependent

MA

and culture-independent methods. Food Control, 34(1), 183–190. Lv, X.C., Huang, Z.Q., Zhang, W., Rao, P.F., & Ni, L. (2012a). Identification and characterization of

ED

filamentous fungi isolated from fermentation starters for Hong Qu glutinous rice wine brewing. J. Gen. Appl. Microbiol, 58(1), 33–42.

EP T

Lv, X.C., Jia, R.B., Li, Y., Chen, F., Chen, Z.C., Liu, B., … Ni, L. (2016). Charac terization of the dominant bacterial communities of traditional fermentation starters for Hong Qu glutinous rice

AC C

wine by means of MALDI-TOF mass spectrometry fingerprinting, 16S rRNA gene sequencing and species-specific PCRs. Food Control, 67, 292–302. Lv, X.C., Weng, X., Zhang, W., Rao, P.F., & Ni, L. (2012b). Microbial diversity of traditional fermentation starters for Hong Qu glutinous rice wine as determined by PCR- mediated DGGE. Food Control, 28(2), 426–434. Ma, K., Maeda, T., You, H., & Shirai, Y., (2014). Open fermentative production of L- lactic acid with high optical purity by thermophilic Bacillus coagulans using excess sludge as nutrient.

ACCEPTED MANUSCRIPT Bioresource Technol, 151, 28–35. Mo, X. L., Fan, W. L., & Yan, X. (2012). Changes in volatile compounds of Chinese rice wine Wheat qu during fermentation and storage. Journal of the Institute of Brewing, 115(4), 300-307.

PT

Nie, Z., Zheng, Y., Du, H., Xie, S., & Wang, M. (2015). Dynamics and diversity of microbial community succession in traditional fermentation of Shanxi aged vinegar. Food Microbiol, 47,

RI

62–68.

SC

Nuñez-Guerrero, M.E., Páez-Lerma, J.B., Rutiaga-Quiñones, O.M., González-Herrera, S.M., &

NU

Soto-Cruz, N.O. (2016). Performance of mixtures of Saccharomyces and non- Saccharomyces

MA

native yeasts during alcoholic fermentation of agave duranguensis juice. Food Microbiol, 54(1), 91–97.

ED

Oda, K., Kakizono, D., Yamada, O., Iefuji, H., Akita, O., & Iwashita, K. (2006). Proteomic analysis of extracellular proteins from Aspergillus oryzae grown under submerged and solid-state

EP T

culture conditions. Appl. Environ. Microb, 72(5), 3448–3457. Park, K.H., Liu, Z., Park, C.S., & Ni, L. (2016). Microbiota associated with the starter cultures and

649–658.

AC C

brewing process of traditional Hong Qu glutinous rice wine. Food Sci. Biotechnol, 25(3),

Portillo, M.D.C., & Mas, A. (2016). Analysis of microbial diversity and dynamics during wine fermentation of Grenache grape variety by high-throughput barcoding sequencing. LWT Food Science and Technology, 72, 317–321. Prakitchaiwattana, C.J., Fleet, G.H., & Heard, G.M. (2004). Application and evaluation of denaturing gradient gel electrophoresis to analyse the yeast ecology of wine grapes. LWT -

ACCEPTED MANUSCRIPT Food Sci. Technol, 4(8), 855–877. Sami, M., Yamashita, H., Hirono, T., Kadokura, H., Kitamoto, K., Yoda, K., & Yamasaki, M. (1997). Hop-resistant Lactobacillus brevis, contains a novel plasmid harboring a multidrug resistance-like gene. J. Ferment. Bioeng, 84(1), 1–6.

PT

Sekiguchi, H., Tomioka, N., Nakahara, T., & Uchiyama, H. (2001). A single band does not always represent single bacterial strains in denaturing gradient gel electrophoresis analysis. Biotechnol.

RI

Lett, 23, 1205–1208.

SC

Sieuwerts, S., Bron, P.A., & Smid, E.J. (2018). Mutually stimulating interactions between lactic

NU

acid bacteria and Saccharomyces cerevisiae in sourdough fermentation. LWT - Food Sci.

MA

Technol, 90, 201–206.

Simonen, M., & Palva, I. (1993). Protein secretion in Bacillus species. Microbiol. Rev, 57, 109.

ED

Song, S.H., Lee, C., Lee, S., Park, J.M., Lee, H.J., Bai, D.H., … Park, Y. (2013). Analysis of microflora profile in Korean traditional nuruk. J. Microbiol. Biotechnol, 23(1), 40–46.

EP T

Thanh, V.N., Mai, L.T., & Tuan, D.A. (2008). Microbial diversity of traditional Vietnamese alcohol fermentation starters (banh men) as determined by PCR mediated DGGE. Int. J. Food

AC C

Microbiol, 128, 268–273.

Thera, M.A., Drissa, C., Kone, A.K., Guindo, A.B., Karim, T., Sall, A.H., … Youssouf, T. (2016). Phase 1 randomized controlled trial to evaluate the safety and immunogenicity of recombinant Pichia pastoris-expressed Plasmodium falciparum apical membrane antigen 1 (PfAMA1-FVO [25-545]) in healthy Malian adults in Bandiagara. Malar. J, 15, 442. Vestner, J., Malherbe, S., Du, T.M., Nieuwoudt, H.H., Mostafa, A., Górecki, T., & De, V.A. (2011). Investigation of the volatile composition of pinotage wines fermented with different malolactic

ACCEPTED MANUSCRIPT starter cultures using comprehensive two-dimensional gas chromatography coupled to time-of-flight mass spectrometry (GC×GC-TOF-MS). J. Agric. Food Chem, 59, 12732–12744. Vol., N. (1999). Implication of acetic acid in the induction of slow/stuck grape juice fermentations and inhibition of yeast by Lacobacillus sp. Am. J. Enol. Viticult, 50(2), 204–210.

PT

Walterson, A.M., Smith, D.D.N., & Stavrinides, J. (2014). Identification of a Pantoea biosynthetic cluster that directs the synthesis of an antimicrobial natural product. PLos One, 9(5), e96208.

RI

Wang, C.L., Shi, D.J., & Gong, G.L. (2008). Microorganisms in Daqu: a starter culture of Chinese

SC

Maotai-flavor liquor. World J. Microb. Biot, 24(10), 2183–2190.

NU

Wang, P., Mao, J., Meng, X., Li, X., Liu, Y., & Feng, H. (2014). Changes in flavour characteristics

region. Food Control, 44, 58–63.

MA

and bacterial diversity during traditional fermentation of Chinese rice wines from Shaoxing

ED

Wang, P., Wu, Q., Jiang, X., Wang, Z., Tang, J., & Xu, Y. (2017). Bacillus licheniformis affects the microbial community and metabolic profile in the spontaneous fermentation of Daqu starter for

EP T

Chinese liquor making. Int. J. Food Microbiol, 250, 59–67. Zhang, W.X., Wu, Z.Y., Zhang, Q.S., Wang, R., & Li, H. (2009). Combination of newly developed

AC C

high quality Fuqu with traditional Daqu for Luzhou- flavor liquor brewing. World J. Microbiol. Biotechnol, 25, 1721–1726. Zheng, C.Y., Gong, L.T., Huang, Z.Q., Liu, Z.B., Zhang, W., & Ni, L. (2014). Analysis of the key volatile aroma compounds in sweet Hongqu glutious rice wine. J. Chinese Inst. Food Sci. Technol, 14, 209–217. Zheng, C.Y., Huang, Z.Q., Liu, Z.B., Zhang, W. & Ni, L. (2015). Sensory evaluation of Hongqu glutinous rice wine using quantitative descriptive analysis method. J. Chinese Inst. Food Sci.

ACCEPTED MANUSCRIPT Technol, 15, 205–213. Zheng, X.W., Tabrizi, M.R., Nout, M.J.R., & Han, B.Z. (2012). Daqu – A traditional Chinese liquor

AC C

EP T

ED

MA

NU

SC

RI

PT

fermentation starter. J. I. Brewing, 117(1), 82–90.

ACCEPTED MANUSCRIPT Figure legends Fig. 1. Biplots of the two principal components (PC1 vs. PC2) after principal component analysis (PCA) of the relative abundance of (A) bacterial and (B) fungal flora found in different fermentation starters (5 HQ and 5 BQ) used for HQGRW brewing.

PT

Fig. 2. Heatmap and dendrogram of the abundant (A) bacterial and (B) fungal species or genus present in different fermentation starters. The color intensity is proportional to the relative

SC

classified tags by the total tags number of each sample.

RI

abundance of bacterial/fungal species, which was calculated by dividing the number of

NU

Fig. 3. Co-occurrence and co-exclusion relationships between the abundances of (A) bacteria and

MA

bacteria; (B) fungi and fungi; (C) bacteria and fungi. The figure presents a Spearman's rank correlation matrix of bacterial/fungal species with > 0.01% abundance. Strong correlations are

ED

indicated by large circles, whereas weak correlations are indicated by small circles. The color of the scale bar denotes the nature of the correlation, with 1 indicating a perfect positive

EP T

correlation (light blue) and -1 indicating a perfect negative correlation (light red). Fig. 4. Heatmap and dendrogram of the volatile organic compounds present in the ten traditional

AC C

fermentation starters (5 HQ and 5 BQ) used for HQGRW brewing. The heatmap plot indicates the relative abundance of volatile organic compounds in different samples (variables clustered on the vertical axis). The color intensity is proportional to the relative abundance of volatile organic compounds. Fig. 5. Score and loading Biplots of the three principal components after PCA of total volatile organic metabolites found in traditional fermentation starters used for HQGRW brewing. (A) PC1 vs PC2; (B) PC1 vs PC3. Samples of HQ are color coded as red, and samples of BQ are

ACCEPTED MANUSCRIPT color coded as blue. Fig. 6. Visualization of the correlation network according to significant correlations between volatile metabolite concentrations and the relative abundance of the dominant (A) bacteria, and (B) fungi in different traditional fermentation starters. Each node represents a volatile organic

PT

compound (VOC) or a microorganism. The red solid line and black dotted line represents positive and negative correlation, respectively. In addition, line width indicates the strength of

RI

correlation. Only the significant edges are drawn in the network using the Spearman

AC C

EP T

ED

MA

NU

SC

correlation test (FDR adjusted P < 0.01).

ACCEPTED MANUSCRIPT Highlights HQGRW is brewed with two traditional fermentation starters—Hong Qu (HQ) and Bai Qu (BQ).



BQ was mainly dominated by L. lactis, L. brevis, R. arrhizus, and S. fibuligera.



HQ was mainly dominated by B. ginsengihumi, Pantoea sp., M. purpureus, and A. niger.



Microbial communities varied significantly in different traditional fermentation starters.



L. brevis, P. pentosaceus, L. alimentarius, R. microspores, A. niger, and M. purpureus were related to some volatile alcohols and esters.

AC C

EP T

ED

MA

NU

SC

RI

PT



Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6