Characteristic flavor compounds and bacterial community of different gray sufu, a traditional Chinese fermented soybean curd

ABSTRACT

Gray sufu is a traditional Chinese fermented soybean curd, and its flavor is significantly influenced by bacterial community. In this study, volatile flavor compounds (VFCs) were examined by headspace solid‐phase microextraction (HS-SPME) in conjunction with gas chromatography and mass spectrometry (GC‐MS). Additionally, bacterial community was analyzed using high-throughput sequencing. The results showed that, a total of 156 VFCs were tentatively identified, in which 8 esters, 6 S/N-containing compounds, 3 alcohols, 3 aldehydes, 1 ketone and 3 other compounds were recognized as the main characteristic VFCs. The abundant S/N-containing compounds are the most important flavor characteristics of gray sufu. A total of 4 phyla and 12 genera were revealed as dominant bacteria. Firmicutes and Proteobacteria were the most abundant phylum, whereas Bacillus, Tetragenococcus and Lactococcus were the prevailing genera. According to the KEGG pathway annotation, amino acid metabolism (26.98%) was the highest metabolic pathway. Among them, aromatic amino acids metabolism accounted for the highest proportion, up to 34.94%. Correlation analysis showed that Lactobacillus, Tetragenococcus, Bacillus, Pseudomonas and Halanaerobium were the top five positive flavor-producing bacteria. The results will be helpful for understanding the fermentation mechanism of gray sufu and improving flavor quality by developing microbial regulation strategies.

KEYWORDS:

• Gray sufu
• volatile flavor compounds
• bacterial community
• functional predictive
• correlation analysis

Introduction

Gray sufu, also known as stinky tofu, is a traditional fermented soybean curd in China.^[1] The name of gray sufu comes from its gray surface and unique flavor of “smelling alwfully stinky, eating alwfully delicious.”^[2] Due to the abundant high-quality protein, free amino acids, unsaturated fatty acids, calcium and functional compounds, gray sufu can be used as a good source of high-quality protein and calcium.^[3] The production process of gray sufu can be approximately divided into three phases, including tofu preparation, pre-fermentation and post-ripening (Figure 1).^[4] Briefly, tofu is made from soybean bean by a series of procedures including grinding, boiling, coagulation and pressing. Then, the tofu cubes are generally inoculated with Actinomucor elegans, Mucor flavus or Mucor racemosus for 48–72 hours. Subsequently, the moldy tofu cubes (pheze) are pickled in salt for about 5 days. Finally, the dressing mixture is used to age the salt-cured moldy tofu cubes (salted pheze) for 2–6 months.^[2,5]

Figure 1. The fermentation process of gray sufu.

Flavor are very important indicators for sufu quality. Volatile flavor compounds (VFCs) contribute the most to the overall flavor. Currently, most studies focus on the VFCs of red and white sufu. The characteristic VFCs are dominated by esters, followed by alcohols and aldehydes, such as ethyl hexanoate, ethyl caprylate, ethyl caprate, ethyl dodecanoate, 1-octen-3-ol, phenylacetaldehyde.^[6–9] However, there are few studies on the VFCs of gray sufu.^[2] Especially, the unique flavor of “smelling alwfully stinky, eating alwfully delicious” needs to be further analyzed.

The VFCs formation mostly originates from the complex microbial metabolism during fermentation, including amino acid metabolism, esterification, elimination, oxidation-reduction reaction, alcohol fermentation and Maillard reaction. Previous works have provided a view of microbial community, and the correlation between microbes and flavor of red and white sufu.^[10–14] For example, nine bacterial and six fungal genera were identified as core functional microbiotas, which significantly affected the production of flavor compounds during white sufu natural fermentation.^[11] It has been shown that Lactobacillus can change the overall flavor composition of red sufu, which is conducive to the production of esters and alcohols, especially ethyl 2-methylbutyrate, phenylethanol and 3-methylbutanol. Leuconostoc, Weissella, Debaryomyces and Pichia may contribute to the production of phenylacetaldehyde, 2-pentylfuran, ethyl caprylate, ethyl caprate.^[13] Compared with other types of sufu, gray sufu has great differences in production technology and microbial composition. It is necessary to explore the correlation between the characteristic flavor and microbial community of gray sufu.

Therefore, the objectives of this study were 1) to determine VFCs of different gray sufu products, 2) to explore the microbial community, and 3) to analyze correlation between characteristic VFCs and bacterial community.

Materials and methods

Samples and standards

Ten types of gray sufu were purchased from various regions in China. The schematic diagram for the production of gray sufu is shown in Figure 1. Specific descriptions of the samples (abbreviation, trade name and region) were listed in Table 1. All samples were produced on similar dates and stored in the laboratory refrigerator (4°C) until analyzed. The standard chemicals including cyclohexanone (CAS 108-94-1), 2,4,6-trimethylpyridine (CAS 108-75-8), methyl nonanoate (CAS 1731-84-6) and n-alkanes (C7-C30) were purchased from Sigma Aldrich Trading Co., Ltd. (Shanghai, China).

Table 1. Details of the different commercial gray sufu.

NO.	Abbreviation	Trade name	Region
1	HLJ	Kedong	Heilongjiang province
2	JL	Qishen	Jilin province
3	TJ	Hongluan	Tianjin
4	BJ	Wangzhihe	Beijing
5	SD	Jimei	Shandong province
6	HB	Jiangbaochen	Hebei province
7	AH	Manaoquan	Anhui province
8	JS	Lican	Jiangshu province
9	JX	Xiangsijia	Jiangxi province
10	ZJ	Xianhengjiudian	Zhejiang province

Sample pretreatment

Approximately 50 g of sufu cubes were sampled from 10 cans of each kind of gray sufu. The surface of approximately 0.2 cm was collected and mixed with dressing mixture at a proportion of 2:1 (w/w). After mixing thoroughly in a sterile plastic bag, about 5 g of samples in triplicate were collected and stored at −80°C for the determination of microbial diversity. All the above operations were done under aseptic conditions. The remaining part without surface was collected to avoid the interference of dressing mixtures. Subsequently, about 10 g of well-mixed samples in triplicate were collected for the determination of VFCs. All the analyses were carried out in three replicates and expressed by the mean of three replicates.

VFCs

The VFCs of different gray sufu samples were detected by using headspace-solid phase microextraction (HS-SPME) coupled with gas chromatography-mass spectrometer (GC-MS) as described previously^[9] with some modifications. Briefly, 1 g of sufu samples were placed in a 10 mL SPME glass vial. Subsequently, 2 μL of internal standard solution was added, which was composed of cyclohexanone (2.053 mg/mL), 2,4,6-trimethylpyridine (0.843 mg/mL), and methyl nonanoate (0.788 mg/mL). The mixture was equilibrated in a constant stirring water bath at 55 ◦C. The 2 cm-50/30 μm DVB/CAR/PDMS Stable Flex manual holder (Supelco Inc., Bellefonte, PA, USA) was used to extract VFCs at 50 ◦C for 40 min. The fiber was then inserted into a GC-MS inlet and desorbed at 250°C for 6 min.

The VFCs analysis was carried out by GC-MS (HP6890/5975C; Agilent Ltd, Santa Clara, CA, USA) equipped with an HP-5 MS Ultra Inert column (60 m × 0.25 mm × 0.25 μm; Agilent Technologies, USA). The initial oven temperature was 40°C (4 min), then ramped to 160°C at a rate of 3°C/min and 280°C (50 min) at a rate of 10°C/min. The split ratio was 5:1. All other operation conditions were consistent as described before.^[9] According to the Kovats formula, the retention index (RI) was calculated by using external standard n-alkanes (C7-C30) under identical operating conditions. The VFCs were identified by matching the mass spectra and RI of the database (Wiley275/Nist2017). The VFCs with high matching degree were confirmed by combining the known compounds reported in the previous literature. The concentration of each VFCs was calculated based on the content and peak area of the internal standard. The odor activity value (OAV) was conducted to assess the contribution of each component to flavor.^[15] Principal component analysis (PCA) of VFCs was performed on the online tool of Majorbio Cloud Platform (https://cloud.majorbio.com/page/tools/).

Microbial diversity

The sufu samples were prepared according to the method described in our previous reports.^[9] The microbial diversity was detected by high-throughput sequencing. Briefly, the total genomic DNA was extracted using the E.Z.N.A Soil DNA kit (Omega Bio-tek, Norcross, GA, USA). Primers 338F (5′-ACTCCTACGGGAGGCAGCA-3′) and 860 R (5′-GGACTACHVGGGTWTCTAAT-3′) were used to amplify the V3-V4 hypervariable region of bacterial 16S rRNA gene. The Illumina MiSeq platform (Majorbio Biopharm Technology Co., Ltd, Shanghai, China) was used for purified amplicons. The sequences were subjected to merging by FLASH and quality filtering, followed by clustering into Operational Taxonomic Units (OTU) at a 97% similarity threshold using UPARSE.^[10]

Based on OTU cluster results, the multiple diversity index analysis was performed. Then, the statistics of community structure were carried out at each taxonomic level. On the basis of the above analysis, a series of in-depth statistical and visual analysis of community structure and phylogeny were conducted.

Analysis of correlation between VFCs and bacterial community

The correlation between characteristic VFCs and main bacteria was analyzed by R software (version 3.3.1).^[9] A total of 24 characteristic VFCs (OAVs >200) were selected for correlation analysis, including 8 esters, 6 S/N-containing compounds, 3 alcohols, 3 aldehydes, 1 ketone and 3 others compounds. The top 25 bacterial genera were further analyzed by spearman’s rank correlation coefficients (ρ). The correlation was visualized by heatmap via Heml Heatmap Illustrator (version 1.0.3.3). The color depth indicates the size of the values.

Results and discussion

Comparison of the types and contents of VFCs

The VFCs of different gray sufu samples were detected by HS-SPME/GC-MS. As shown in Figure 2 and Table S1, a total of 156 VFCs were tentatively identified, including 51 esters, 23 S/N-containing compounds, 19 alcohols, 14 acids, 9 aldehydes, 9 ketones, 8 alkenes and 10 other compounds. In terms of types, esters (51), S/N-containing compounds (23) and alcohols (19) are the main VFCs (Figure 1a). In HB, the number of VFCs was the highest, up to 91. In terms of contents, acids (615.07 mg/kg), alcohols (314.76 mg/kg), and S/N-containing compounds (174.09 mg/kg) were the major VFCs (Figure 1b). In JL (245.50 mg/kg), the total content of VFCs was the most abundant, followed by JX (239.67 mg/kg). AH had the lowest VFCs, only 65.36 mg/kg. There were only 14 VFCs shared by all gray sufu, in which included 6 esters, 2 S/N-containing compounds, 3 alcohols, 1 acid, 1 aldehyde and 1 other compound. The VFCs were distinctly different in both type and content, which may be an important reason for the unique flavor characteristics of different gray sufu samples. The VFCs exhibited substantial alterations during the fermentation process,^[12] which can be attributed to the diverse range of enzymes secreted by microorganisms in sufu, including protease, glutaminase, peptidase, lipase, cellulase, hemicellulose, α-amylase, and β-glucosidase. These enzymes facilitate the breakdown of amino acids, proteins, lipids, and carbohydrates present in the raw materials, resulting in the formation of distinctive flavor compounds or their precursors.^[16]

Figure 2. Comparison of the types (A) and contents (B) of VFCs.

Esters were the most abundant VFCs, which was in accordance with previous reports.^[9,12,17] Esters content in JL was the highest (29.69 mg/kg), followed by that in TJ (23.93 mg/kg). Six common esters were detected, including ethyl acetate, ethyl butyrate, isoamyl butyrate, 2-methylbutylbutyrate, methyl 2-methyloctanoate and octyl butyrate. Ethyl acetate has a fruity, sweet aroma that can contribute to olfactory complexity and enhance the bouquet of sufu.^[18] The acid group is a fatty acid with a medium-length chain and the alcohol group is ethanol. Esters are recognized as the most important flavor components in sufu.^[17] Esterification and alcoholysis are the principal catalytic processes involved in the biosynthesis of esters. Esterification refers to the creation of esters through the reaction between alcohols and carboxylic acids. Concurrently, alcoholysis entails the production of esters from the reaction between alcohols and acylglycerols, as well as alcohols and fatty acids, amino acids, and lipidyl CoA generated through carbohydrate metabolism.^[19]

A total of 23 S/N-containing compounds were identified. Among them, dimethyl disulfide and indole were detected in all gray sufu samples. In some samples, the S/N-containing compounds contents were extremely abundant, such as JL (31.41 mg/kg) and JX (30.48 mg/kg). It was speculated that these compounds were derived from the degradation of sulfur-containing amino acids or generated by microorganisms,^[20] and linked with alliaceous, sulfuric, sweaty, onion and cabbage aromas.^[21] Nitrogen-containing compounds mainly consisted of indole and pyrazines. Indole often shows an unpleasant smell of feces and rotting corpses at high concentrations, while fruity at low concentrations. From the perspective of amino acid structure, tryptophan is the only amino acid containing indole structure skeleton. Indole was confirmed to be derived from microbial degradation of tryptophan.^[22] Seven pyrazines were found in gray sufu, including 2,5-dimethylpyrazine, 2,3,5-trimethylpyrazine, 2,6-dimethyl-3-ethylpyrazine, 2,5-dimethyl-3-ethylpyrazine, tetramethylpyrazine, 2,5-dimethyl-3-propylpyrazine and 2,3,5-trimethyl-6-propylpyrazine. Pyrazines have special “roasting” or nutty flavors, which are generally formed by condensation of aminoketones produced by Maillard reaction and Strecker degradation. Some microorganisms can also metabolize and synthesize pyrazines.^[23]

Alcohols contained 19 compounds, in which ethanol, 3-methyl-1-butanol, and phenylethyl alcohol were found in all the gray sufu samples. In JL, the alcohols content was 88.64 mg/kg, which was about 8 times more than HB and SD. Most of the ethanol was purposely added, and other alcohols were derived from the catalytic reaction of different fungal enzymes during fermentation.^[24] Alcohols can impart alcoholic, sweet and fruity notes and were important flavor-contributing class for most types of sufu.^[9,12,17] As synthetic precursors, alcohols can form esters with acids, which is beneficial to the formation of the unique flavor of sufu. The acid class contained 14 compounds, and only acetic acid were shared by all the samples. Acids can directly influence the balance and taste of food, due to the strong odor. It is noteworthy that acids are also the precursors of other flavor compounds, such as ketones, alcohols, esters.^[25]

In addition, 9 aldehydes, 9 ketones, 8 alkenes and 10 other compounds were also detected. Among them, benzaldehyde and phenol were the common VFCs. Compared to VFCs mentioned above, the total contents of aldehydes, ketones, alkenes and other compounds were very low. For instance, the highest contents of aldehydes and alkenes were 3.36 mg/kg and 5.079 mg/kg, respectively. Benzaldehyde possesses a pleasant olfactory profile characterized by sweet, floral, and spice-like notes, rendering it a popular choice for the cosmetic and food sectors as an aliphatic fragrance and flavoring agent.^[26] Aldehydes exhibit low odor thresholds and highly pronounced olfactory characteristics, contributing to the delectable sensory attributes in sufu, encompassing sweet, fruity, nutty and caramel flavor.^[27]

PCA is a method that reduces multidimensional data to orthogonal coordinates based on maximum variance linear projection.^[28] In this study, PCA was used to determine the types and contents of VFCs in gray sufu (Figure 3). For VFCs types, PC1 and PC2 accounted for 72.846% and 10.856% of the total variance, respectively (Figure 3a). The first 2 principal components retained 83.702% of the original data information. For VFCs contents, the contribution rate of PC1 was 99.895%, while PC2 was only 0.046% (Figure 3b). The cumulative contribution rate of the two principal components was more than 99%, indicating that the two principal components can well represent the main information characteristics of VFCs contents. The distribution distances between the gray sufu samples represent the similarities and differences. Compared with contents, the differences of types were somewhat small, especially SD and JX. The results are in agreement with the conclusions discussed above (Figure 2a). The VFCs contents of JX were largely different from other gray sufu samples when projected on PC1. The main reason might be that the composition of dressing mixture in JX was edible oil, while in other gray sufu was brine. In addition, many factors such as soybeans type, tofu moisture content, pheze cubes size, salt content, fermentation conditions also contribute to VFCs.

Figure 3. PCA analysis on the types (a) and contents (b) of VFCs.

Analysis of characteristic VFCs

It is known that the flavor is determined by aroma and taste. VFCs were the important indicator of the aroma quality, while non-VFCs were commonly correlated with taste perception.^[29] The contributions of VFCs to the overall flavor depend on not only the amount of the components, but also the odor threshold values.^[9] To estimate the flavor contributions of the individual VFC in gray sufu, the OAVs were calculated. Generally, VFCs with OAVs ≥ 1 contribute to the overall flavor. The larger the OAV value, the greater the contribution.^[30] As shown in Table S2, a total of 69 VFCs (OAVs ≥1) contribute to the overall flavor of gray sufu, including 26 esters, 8 S/N-containing compounds, 8 alcohols, 6 acids, 7 aldehydes, 5 ketones, 3 alkenes and 6 other compounds.

In the study, 24 VFCs with OAVs > 200 were regarded as the main characteristic VFCs, containing of 8 esters, 6 S/N-containing compounds, 3 alcohols, 3 aldehydes, 1 ketone and 3 other compounds (Table 2). The esters (ethyl acetate, ethyl isobutyrate, ethyl butyrate, ethyl 2-methylbutyrate, ethyl isovalerate, ethyl valerate, isoamyl propionate, ethyl caproate) and S/N-containing compounds (methanthiol, dimethyl disulfide, dimethyl trisulfide, 2,6-dimethyl-3-ethylpyrazine, dimethyl tetrasulfide, indole) were dominant characteristic VFCs. Furthermore, all of the characteristic VFCs with OAV ≥ 1.00 × 10⁴ were belong to esters and S/N-containing compounds. The characteristic VFCs in sufu have been reported by gas chromatography-mass spectrometry (GC-MS) and gas chromatography-olfactometry (GC-O).^[8–10,17] Chen, et al.^[10] concluded that ethyl 3-methylbutanoate, ethyl isobutyrate, linalool, linalool, cis-4-heptenal and 2-methylpropanal were the characteristic VFCs of Mouding sufu. Li, et al.^[8] found that the key VFCs were trimethylpyrazine, linalool, 2-pentyl-furan, decanoic acid ethyl ester, methyl salicylate and octanoic acid ethyl ester in Chaling natural fermented red sufu. It is worth mentioning that, (E,E)-2,4-decadienal, 1-octen-3-ol, (E,E)-2,4-nonadienal, eugenol, ethyl hexanoate and phenylacetaldehyde were not only the aroma-active compounds in Guilin Huaqiao white sufu, but also played an important role in enhancing the umami aftertaste and palatability of umami solution.^[17] Compared with red sufu and white sufu, the abundant S/N-containing compounds are the most important characteristics of the unique flavor “smelling alwfully stinky, eating alwfully delicious.” Sulfur-containing compounds are also recognized as key flavor contributors to cheese,^[31] alliums,^[21] coffee^[32] and meat.^[33] The compounds such as methanthiol, dimethyl disulfide and dimethyl trisulfide can be used as food additives in China, and play an important role in food industry.^[34] In addition, six characteristic VFCs, including ethyl butyrate, ethyl butyrate, dimethyl disulfide, indole, 3-methyl-1-butanol and phenol, were shared by all the gray sufu. The OAV of ethyl isovalerate in HB was the highest, up to 6.83 × 10⁵, followed by dimethyl trisulfide in JX (1.04 × 10⁵). Both of ethyl isovalerate and dimethyl trisulfide are the important flavor compounds in cheese.^[35]

Table 2. Characteristic volatile flavor compounds of gray sufu samples (OAVs >200).

	ID	VFCs	Threshold (mg/kg)	OAV
				HLJ	JL	TJ	BJ	HB	SD	AH	JS	JX	ZJ
Esters (8)	ES-1	Ethyl acetate	0.005	20.41	2.22 × 10^2	11.24	1.26 × 10^2	78.97	4.42	36.42	15.22	2.40	54.61
	ES-4	Ethyl isobutyrate	0.000089	2.38 × 10^3	-	-	-	-	1.32 × 10^4	-	-	-	4.94 × 10^3
	ES-5	Ethyl butyrate	0.0009	3.17 × 10^3	1.00 × 10^4	1.08 × 10^4	1.17 × 10^4	2.27 × 10^3	6.79 × 10^3	4.28 × 10^2	4.51 × 10^3	1.24 × 10^4	3.83 × 10^3
	ES-8	Ethyl 2-methylbutyrate	0.000013	-	-	-	-	4.45 × 10^4	-	9.27 × 10^3	1.82 × 10^4	4.46 × 10^4	-
	ES-9	Ethyl isovalerate	0.00001	-	-	-	-	6.83 × 10^5	-	-	-	-	2.15 × 10^4
	ES-11	Ethyl valerate	0.0058	98.54	1.74 × 10^2	1.31 × 10^2	60.78	-	-	73.19	97.17	-	5.85 × 10^2
	ES-15	Isoamyl propionate	0.026	-	8.82	-	-	4.50	1.25	0.95	3.75	-	2.02 × 10^2
	ES-17	Ethyl caproate	0.0022	-	1.74 × 10^3	-	-	-	-	78.11	-	-	2.20 × 10^2
S/N-containing compounds (6)	SN-1	Methanthiol	0.0002	78.95	-	62.95	6.60 × 10^2	2.96 × 10^2	-	1.17 × 10^2	5.05 × 10^2	-	80.15
	SN-2	Dimethyl disulfide	0.0011	2.45 × 10^3	7.97 × 10^3	4.79 × 10^3	8.67 × 10^3	1.45 × 10^3	3.60 × 10^3	4.52 × 10^2	3.01 × 10^3	2.29 × 10^4	6.84 × 10^2
	SN-4	Dimethyl trisulfide	0.000016	-	9.04 × 10^4	-	2.09 × 10^4	3.46 × 10^3	1.96 × 10^4	5.7 × 10^2	4.17 × 10^4	1.04 × 10^5	1.62 × 10^4
	SN-5	2,6-Dimethyl-3-ethylpyrazine	0.00004	6.57 × 10^2	-	5.65 × 10^2	2.28 × 10^3	4.95 × 10^2	3.42 × 10^2	-	4.92 × 10^3	-	2.38 × 10^3
	SN-7	Dimethyl tetrasulfide	0.00002	3.30 × 10^3	1.16 × 10^3	-	-	-	4.64 × 10^3	1.26 × 10^3	-	1.77 × 10^3	-
	SN-8	Indole	0.011	1.34 × 10^3	8.22 × 10^2	4.17 × 10^2	1.03 × 10^3	1.83 × 10^2	31.30	6.10 × 10^2	1.32 × 10^3	48.20	46.95
Alcohols (3)	AL-2	3-Methyl-1-butanol	0.004	1.82 × 10^2	5.15 × 10^2	3.27 × 10^2	7.08 × 10^2	8.60 × 10^2	8.94 × 10^2	2.39 × 10^3	1.28 × 10^3	2.88 × 10^2	58.73
	AL-4	1-Hexanol	0.0056	2.15 × 10^2	1.56 × 10^3	3.86 × 10^2	-	1.80 × 10^2	1.87 × 10^2	4.49 × 10^2	8.35 × 10^2	-	81.86
	AL-7	Linalool	0.00022	1.06 × 10^2	5.28 × 10^2	1.19 × 10^2	3.23 × 10^2	2.62 × 10^2	-	60.24	1.24 × 10^2	-	38.49
Aldehydes (3)	AD-1	3-Methyl-butanal	0.00017	-	-	-	7.76 × 10^3	1.27 × 10^3	3.33 × 10^3	-	-	2.90 × 10^3	8.11 × 10^2
	AD-2	2-Methyl-butanal	0.001	-	-	-	-	2.16 × 10^2	9.07 × 10^2	-	-	-	-
	AD-4	Hexanal	0.005	-	2.41 × 10^2	3.36 × 10^2	3.60 × 10^2	1.09 × 10^2	1.18 × 10^2	2.56 × 10^2	-	3.23 × 10^2	2.31 × 10^2
Ketone (1)	KE-2	2-Methyl-3-hexanone	0.024	-	8.29	-	-	2.14 × 10^2	-	8.96	49.23	70.60	-
Others (3)	OT-1	Phenol	0.021	4.32 × 10^2	1.91 × 10^2	82.39	2.48 × 10^2	64.76	44.04	9.04	80.23	3.07 × 10^2	1.12 × 10^2
	OT-5	Anethole	0.015	0.86	4.90	1.06	-	2.51 × 10^2	-	2.42	-	2.59	1.09 × 10^2
	OT-6	Eugenol	0.00071	-	-	-	-	1.04 × 10^3	-	62.51	17.97	-	-

Note: “-” represented the VFCs were not detected. The thresholds for flavor compounds were detected in water media, and quoted from the book by Gemert (2011)

Overview of bacterial community composition

Currently, gray sufu production still adopts the traditional fermentation technology, but the open environment leads to many airborne microorganisms entering the fermentation substrate. This led to the development of a unique microbial community structure that imparts the unique flavor of gray sufu.^[36] As mentioned above, the production process of gray sufu can be approximately divided into three phases, including tofu preparation, pre-fermentation and post-ripening. Many different factors such as soybeans type, tofu moisture content, pheze cubes size, salt content, dressing mixture composition, fermentation conditions will have important impacts on microbial community. Therefore, the bacterial diversity of 10 gray sufu samples was determined by high-throughput sequencing. Alpha diversity indexes were calculated to assess richness and diversity of microbial community.^[12] As shown in Table 3, the Ace (755) and Chao1 (749) of SD were the highest, while Ace (149) and Chao1 (151) of HLJ were the lowest, suggesting that bacterial community in SD sample was the most richest. Shannon indexes ranged from 2.18 to 3.42, in which ZJ (3.42) was the highest, followed by JX (3.06) and TJ (3.03). Correspondingly, these three gray sufu samples had the lowest Simpson indexes, implying that the bacterial communities diversity in ZJ, JX and TJ were higher than other samples. The Good’s coverage indexes were more than 99.9% for all samples, indicating that the amplicon sequences selected randomly were appropriate to reveal the diversities of major bacteria.^[37]

Table 3. Alpha diversity of bacterial community in gray sufu.

Sample	Ace	Chao1	Shannon	Simpson	Coverage
HLJ	149	151	2.18	0.2232	0.9995
JL	426	436	2.84	0.1689	0.9991
TJ	528	443	3.03	0.1218	0.9991
BJ	234	206	2.37	0.2015	0.9998
SD	755	749	2.94	0.2171	0.9990
HB	502	418	2.90	0.1302	0.9997
AH	452	429	2.43	0.1986	0.9992
JS	411	322	3.06	0.1113	0.9993
JX	360	338	2.19	0.1952	0.9990
ZJ	468	470	3.42	0.0922	0.9994

The Venn diagram depicts the bacteria that are shared or unique for different gray sufu products.^[9] As shown in Figure 4a, there were 44 bacterial phyla in SD, followed by JL (40 bacterial phyla) and AH (38 bacterial phyla). A total of 8 bacterial phyla were shared by all the gray sufu. As can be seen in Figure 4b, the bacterial genera in SD were the highest, up to 643, followed by ZJ (515 bacterial genera) and JL (474 bacterial genera). Simultaneously, bacterial genera in HLJ were the lowest, which was in accordance with bacterial phyla. A total of 27 common bacterial genera were observed. Essentially, the changes of total bacterial phyla and genera were in agreement with the Ace and Chao1 indexes.

Figure 4. Venn diagrams of bacteria in gray sufu at the (a) phylum and (b) genus level.

Elucidating the microbial structure of gray sufu is the basis for understanding the mechanism involved.^[2] In the study, microorganisms with a relative abundance > 10% were defined as dominant bacteria, while microorganisms ≤ 10% were defined as others. A total of 4 dominant phyla and 12 genera were selected dominant bacteria (Figure 5).

Figure 5. The relative abundance percentage of dominant bacteria at the (a) phylum and (b) genus level.

As presented in Figure 5a, Firmicutes, Proteobacteria, Halanaerobiaeota and Actinobacteriota were the dominant bacterial phyla. Firmicutes and Proteobacteria were detected in all the gray sufu, which was consistent with the changes during gray sufu fermentation.^[2,36] Firmicutes and Proteobacteria were also dominant bacterial phyla in Baixi sufu,^[37] Mouding sufu^[10] and CS sufu.^[14] In gray sufu, Firmicutes was the most abundant in JX, up to 92.54%, followed by AH (75.63%). Proteobacteria in TJ, ZJ and JS were the highest, accounting for 64.98%, 64.71%, 59.79%, respectively. Different from other gray sufu samples, Halanaerobiaeota was the most abundant phylum in HLJ, occupying for 68.87%.

As presented in Figure 5b, there were 12 dominant bacterial genera, such as Bacillus, Tetragenococcus, Lactococcus, Anaerococcus, Halanaerobium, Halomonas. At genus level, the composition of dominant bacteria in different samples varied greatly. Bacillus were highest in AH (34.26%) and JS (22.43%), while Tetragenococcus were most abundant in BJ (38.53%) and JX (33.30%). In the preceding investigation, a substantial abundance of Bacillus was detected on the external surface of gray sufu.^[2] Bacillus has the capability to excrete proteolytic enzymes during fermentation, contributing to the development of flavors in fermented foods.^[38] Lactococcus the dominant bacterial genera in BJ (13.36%) and JS (10.51%). Anaerococcus was prevailing in HB (29.34%), JL (24.87%) and AH (22.10%). Halanaerobium and Halomonas were predominant in HLJ, accounting for 68.87% and 14.97%, which was in agreement with the dominant bacterial phyla. The dominant bacteria of sufu were very variable and could be modulated by many factors like climate changes, microbial starters, fermentation techniques and conditions, geographic factors and fermentation materials.^[39–41]

Functional predictive analysis of bacteria

In order to study the main metabolic functions of gray sufu, annotation analysis of KEGG pathways were performed. By annotating KEGG pathway at level 1, it was found that metabolism accounted for the highest proportion (46.35%), followed by environmental information processing (18.54%) and genetic information processing (16.22%) (Figure S1). Similar results were also obtained in cereal vinegar, in which the top three KEGG pathways were metabolism (42.3%), genetic information processing (28.3%), and environmental information processing (10.1%).^[42] The results indicated that, bacteria in gray sufu are involved in the metabolism of many compounds, which are needed for the survival and reproduction.^[43]

The KEGG pathway of metabolism at level 2 was further analyzed. As shown in Figure S2, the bacteria in gray sufu were mainly involved in nine pathways, including amino acid metabolism, carbohydrate metabolism, cofactors and vitamins metabolism, nucleotide metabolism, lipid metabolism, xenobiotics biodegradation and metabolism, glycan biosynthesis and metabolism, terpenoids and polyketides metabolism, and other secondary metabolites biosynthesis. It is worth noting that, amino acid metabolism accounted for the highest proportion, up to 26.98%, followed by carbohydrate metabolism (22.11%). Amino acid catabolism is the core function of bacteria, through which various bacteria metabolize protein into peptides and amino acids.^[2,44] Carbohydrate catabolism usually serves as the energy source for microbial growth and development during fermentation.^[45]

Studies have shown that amino acid metabolism is one of the most important biochemical pathways in the formation of cheese flavor.^[46–48] The flavor precursor amino acids are mainly aromatic amino acids (phenylalanine, tyrosine and tryptophan), branched chain amino acids (leucine, isoleucine and valine), and sulfur-containing amino acids (methionine and cysteine).^[46–48] Similar to cheese, gray sufu is also a fermented food with high protein as the main raw material. Therefore, the amino acid metabolism at level 3 was further annotated. As presented in Figure 6, aromatic amino acids metabolism accounted for the highest proportion, up to 34.94%. Sulfur-containing amino acids metabolism and branched-chain amino acids metabolism accounted for 18.89% and 14.49%, respectively. The results confirmed that aromatic amino acids, sulfur-containing amino acids and branched-chain amino acids are the main precursor amino acids for the formation of characteristic VFCs in gray sufu.

Figure 6. Annotation analysis of KEGG pathways at level 3.

Correlation analysis between VFCs and bacteria

It is putative that VFCs is one of the key factors that determine the final quality of fermented foods. The VFCs formation mostly originates from the complex microbial metabolism during gray sufu fermentation. It is important to explore the correlation between the VFCs and microbial community of gray sufu. Therefore, the correlation between main characteristic VFCs (OAVs >200) and bacteria (top 25 of relative abundance) was assessed by Spearman’s correlation coefficient (|ρ| > 0.7).

As shown in Figure 7, Lactobacillus, Tetragenococcus, Bacillus, Pseudomonas and Halanaerobium were the top five positive flavor-producing bacteria, whereas Idiomarina, Escherichia-Shigella, Pediococcus, Leuconostoc and Kurthia were the top five negative flavor-producing bacteria. Among them, Lactobacillus, possessed the largest number of significant positive correlations to flavors (8), followed by Tetragenococcus (7) and Bacillus (6). Lactobacillus was positively correlated with eight characteristic VFCs, namely ethyl acetate (ES1), ethyl isobutyrate (ES4), ethyl butyrate (ES5), ethyl valerate (ES11), ethyl caproate (ES17), dimethyl disulfide (SN2), 2-methyl-butanal (AD2) and phenol (OT1). Lactobacillus has also been reported to have positive correlation with the production of VFCs in Mouding sufu.^[10] Generally, Lactobacillus was regarded as the main producers of methyl and ethyl esters due to the lipase and esterase activities.^[49,50] Bacillus and Tetragenococcus were identified as the core functional microbiota in plain sufu.^[11,51] Bacillus has the capability to secrete a diverse array of hydrolases, such as lipases, amylases, and proteases, which facilitate the breakdown of lipids, starch, and proteins into a range of flavor compounds. Consequently, Bacillus is frequently employed to enhance the sensory profile during the fermentation process of white wine.^[52] Similarly, Tetragenococcus has been found to play a pivotal role in the production of flavor compounds during the initial stages of fermentation and recognized as a significant contributor to flavor development in fermented soy products.^[51,53]

Figure 7. Heatmap of correlation between characteristic VFCs (OAVs >200) and bacteria (top 25 of relative abundance). The color changes in heatmap demonstrated the positive (red) or negative (blue) correlation between characteristic VFCs and bacteria.

Previous studies have also confirmed the close correlation between VFCs and microorganisms in sufu. Seven bacterial genera, namely Pseudomonas, Tetragenococcus, Lysinibacillus, Bacillus, Pantoea, Staphylococcus, and Burkholderia-Caballeronia-Paraburkholderia, were strongly correlated with 32 characteristic flavor compounds in the early stage during sufu fermentation.^[15] Li, et al.^[8] discovered that Leuconostoc, Lactobacillus, Weissella, Debaryomyces, Tausonia and Pichia were the main microorganisms responsible for the flavor formation of Chaling red sufu. Based on Pearson correlation coefficients, Bacillus, Enterobacter, Lactobacillus, Sphingobacterium, Stenotrophomonas, Tetragenococcus, Trabulsiella, Unclassified, and Weissella were identified as core functional bacteria significantly affecting the flavor compounds formation during the fermentation of plain sufu.^[11] In summary, the findings from the correlation analysis provide confirmation that the presence of bacteria has a significant impact on the formation of VFCs. Achieving a thorough understanding of the assembly and manipulation of microbiota is essential for the production of gray sufu with consistent and desirable flavor quality.^[2]

Conclusion

The volatile flavor compounds (VFCs), bacterial community and correlation between characteristic VFCs and main bacterial community of different gray sufu were analyzed in the study. A total of 156 VFCs were tentatively identified, in which 8 esters, 6 S/N-containing compounds, 3 alcohols, 3 aldehydes, 1 ketone and 3 other compounds were recognized as the main characteristic VFCs. The abundant S/N-containing compounds are the most important characteristics of the unique flavor “smelling alwfully stinky, eating alwfully delicious” in gray sufu. Bacterial diversity analysis indicated that Firmicutes and Proteobacteria were detected in all the gray sufu. At genus level, the composition of dominant bacteria in different samples varied greatly. Bacillus were highest in AH (34.26%) and JS (22.43%), while Tetragenococcus were most abundant in BJ (38.53%) and JX (33.30%). According to the KEGG pathway annotation, sulfur-containing amino acids metabolism accounted for 18.89%. It is an essential reason for the rich S/N-containing compounds in gray sufu. Correlation analysis showed that Lactobacillus, Tetragenococcus, Bacillus, Pseudomonas and Halanaerobium were the top five positive flavor-producing bacteria. The results confirmed that the bacteria significantly affected the formation of VFCs. Achieving a thorough understanding of the assembly and manipulation of microbiota is essential for the production of gray sufu with consistent and desirable flavor quality.

Acknowledgement

The authors are grateful to Xuzhou University of Technology and State Key Laboratory of Functions and Applications of Medicinal Plants for supporting and providing necessary infrastructure. The authors also thank Miss Chunyan Li for the technical assistance of volatile flavor compounds determination.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Science and Technology projects of Xuzhou [award-id KC22057], N atural Science Foundation of the Jiangsu Higher Education Institutions [award-id 20KJB550013], North Jiangsu Science and Technology Project [XZ-SZ202051], National Natural Science Foundation of China [32102121] and the Research Projects of Xuzhou University of Technology [award-id XKY2019226; XKY2019228].