Untargeted metabolomics analysis of Mucor racemosus Douchi fermentation process by gas chromatography with time‐of‐flight mass spectrometry

Abstract Intensive study of the metabolome during the Douchi fermentation can provide new knowledge for optimizing the fermentation process. In this work, the metabolic characterization throughout the fermentation of Mucor racemosus Douchi was investigated using gas chromatography with time‐of‐flight mass spectrometry. A total of 511 peaks were found, and 114 metabolites were identified. The fermentation process was clearly distinguished into two main phases by principal components analysis and orthogonal partial least squares‐discriminant analysis. All the samples in the score plots were within the 95% Hotelling T 2 ellipse. Two separated clusters can be seen clearly in the score plot, which represents the two stages of fermentation: koji‐making (within 48 hr) and postfermentation (after 48 hr). Besides, clear separation and discrimination by both methods were found among different fermentation time within 15 days, while the discrimination cannot be found with more than 15 days of fermentation, indicating that the fermentation of Douchi was finished in 15 days. Due to the synergistic effect of protease and hydrolase accumulated in the early stage, proteins and other big molecular substances are rapidly hydrolyzed into a large number of small molecule components. However, the activity of enzymes decreased with the further fermentation, and some free amino acids were consumed in Maillard reaction. Therefore, there was no significant change in the content of small molecular substances after 15 days of fermentation. Furthermore, the levels of some metabolites such as alanine and lysine involved in the fermentation varied significantly throughout the processes. This study provides new insights for the metabolomics characteristics of Douchi fermentation.

Among the technologies in metabolomics, gas chromatography with time-of-flight mass spectrometry (GC-TOF/MS) has been widely used due to the advantages of high resolution and sensitivity (Adebo et al., 2018;Ding et al., 2009;Salvatore, Gyong, Tobias, Cataldi, & Oliver, 2010;Sun et al., 2015;Tobias et al., 2009). With the help of the effective method, comprehensive and quantitative analysis of metabolites can be achieved, which can be used to characterize metabolic mechanism at molecular level (Zhang et al., 2017). However, the comprehensive research and optimization of the Douchi fermentation with GC-TOF/MS technology are rarely reported so far.
The aim of the present study was to evaluate dynamics of the metabolome of the Douchi fermentation by using untargeted GC-TOF/MS metabolomics. The metabolites of the M. racemosus Douchi in different fermentation time were compared with the help of principal components analysis (PCA) and orthogonal partial least squares-discriminant analysis (OPLS-DA).

| Materials
Mucor racemosus was obtained from the strain CGMCC8700 kept in China General Microbiological Culture Collection Center. Soya beans, distilled wine (alcohol content 45%), and sodium chloride were purchased from the local markets.
(a) Cleaned soybean was soaked in water and then steamed for 20 min at 115°C. (b) The soybeans were cooled down to 35°C and then inoculated with M. racemosus rapidly, which incubated at 25°C for 48 hr. (c) The product was processed with 1% distilled wine and 8% sodium chloride, and then aged for several days (10, 15, 20, and 25 days) at 25°C. In this study, six batches of soybeans were used, while different fermentation time (0 hr, 24 hr, 48 hr, 5 days, 10 days, 15 days, 20 days, and 25 days, respectively) were investigated. So there were 48 samples in total.

| Metabolites extraction
Sample (60 mg) was added into 0.4 ml methanol/water (3:1, v/v), and then, 20 μl of adonitol solution (1 mg/ml stock in dH 2 O) was added as internal standard. The solution was mixed in the vortex for 30 s and ultrasound treated for 5 min. Then, the solution was centrifuged for 15 min at 9810 g, 4°C. The supernatant (0.3 ml) was transferred into a glass vial and dried by vacuum drying.
Methoxyammonium chloride solution (80 μl, 20 mg/ml in pyridine) was added into the sample and incubated for 30 min at 80°C.
About 100 μl N,O-bis(trimethylsilyl)trifluoroacetamide w/1% trimethylchlorosilane was added into the sample and incubated for 90 min at 70°C. In the experiment, the electron impact ionization was tuned at 70 eV and helium was used as carrier gas with an average linear velocity of 1.0 ml/min. The mass spectrometer was operated with a transfer line temperature of 270°C, ion source 220°C, and mass range from 50 to 500 amu at a rate of 20 spectra/s after a solvent delay of 370 s. The injection volume was 1 μl, and the temperature of injection was 280°C.
Simulation of the missing value and noise removal was achieved as previously described (Sun et al., 2015). Besides, the internal standard normalization method was used to standardize the data (Sun et al., 2015). Both of mass spectrum match and retention index match were considered in metabolites identification. LECO/Fiehn Metabolomics Library was used for the metabolite identification, and a similarity above 700 was adopted for giving the positive answer of the existence of the metabolite. Besides, the compound with the similarity <200 is defined as an "analyte," while the compound with a similarity between 200 and 700 is considered as a putative annotation.
Principal components analysis and OPLS-DA methods were used to display the similarity and difference with the help of the SIMCA14.1 software package (Umetrics). This software has the advantage of displaying Hotelling T 2 95% confidence limit ellipse in the score plot to show the presence of outliers. Furthermore, the first principal component of variable importance in the projection (VIP) value above 1.0 was adopted for giving the positive answer of the existence of the changed metabolite. The remaining variables were then assessed using Student's t test (t test) method. Variable was discarded when the value of p was above 0.05.

| Global analysis of metabolites with PCA and OPLS-DA
In order to discriminate the metabolites of Douchi samples with different fermentation time, PCA was performed. Figure 1  in the score plots. All the samples in the score plots were within the 95% Hotelling T 2 ellipse. As shown in Figure 1

| Optimization of the fermentation time
The

| Significantly different metabolites
For the established OPLS model, the first principal component is related to fermentation time, and thus, S-plot was used to find the important significantly different metabolites (Lee et al., 2016). As shown in Figure 6b, blue squares represent the very significantly increased metabolites, and the red triangles represent significantly increased metabolites during the whole fermentation course. The yellow triangles represent significantly decreased metabolites during the whole fermentation course. Table 1 is the identification of significantly F I G U R E 6 (a) Relationship of the TICs and the fermentation time with OPLS-DA Classification of different fermentation time with OPLS-DA; (b) S-plot for finding the important significantly different metabolites. The blue squares represent the very significantly increased metabolites, and the red triangles represent significantly increased metabolites during the whole fermentation course. The yellow triangles represent significantly decreased metabolites during the whole fermentation course. OPLS-DA, orthogonal partial least squaresdiscriminant analysis; TICs, total ion chromatographs different metabolites. The levels of some metabolites such as alanine and lysine involved in the fermentation varied significantly throughout the processes. In addition, the levels of putrescine and myo-inositol were dramatically increased, while the level of l-Malic acid was decreased. In the fermentation processing of Douchi, the active phytase produced by microorganism can hydrolyze the phytic acid to the inositol and phosphate (Quan, Fan, & Ohta, 2003). Besides, the trypsin inhibitor was destroyed in the fermentation and soy protein is easier to be hydrolyzed by protease, producing a series of intermediate products, such as soy peptides and amino acids. Application of untargeted metabolomics enables unbiased analysis of metabolites which may result in discoveries that were not anticipated by production engineers and thus may lead to provide new insights into the metabolomics characteristics during the Douchi fermentation process.

| CON CLUS ION
The metabolic characterization throughout the fermentation of M. racemosus Douchi was investigated using GC-TOF/MS. A total of 511 peaks were found, and 114 metabolites were identified. Clear separation and discrimination were found within 15 days of fermentation, while the discrimination cannot be found with more than 15 days of fermentation, indicating that the fermentation of Douchi was finished in 15 days. The levels of some metabolites such as alanine and lysine involved in the fermentation varied significantly throughout the processes. In addition, the levels of putrescine and myo-inositol were dramatically increased, while the level of l-Malic acid was decreased.
Application of untargeted metabolomics enables unbiased analysis of metabolites which may result in discoveries that were not anticipated by production engineers and provides new insights into the metabolomics characteristics during the Douchi fermentation process.

ACK N OWLED G M ENTS
This study was supported by National Natural Science Foundation of China (No. 31371828, 31571819, 31601551, and 31671931) and the "1515 Talent Project" of Hunan Agricultural University.

CO N FLI C T O F I NTE R E S T
The authors notify that there are no conflicts of interest.

E TH I C A L A PPROVA L
This study does not involve any human or animal testing.

S U PP O RTI N G I N FO R M ATI O N
Additional supporting information may be found online in the Supporting Information section at the end of the article.