Comprehensive Metabolomic Comparison of Five Cereal Vinegars Using Non-Targeted and Chemical Isotope Labeling LC-MS Analysis

Vinegar is used as an acidic condiment and preservative worldwide. In Asia, various black vinegars are made from different combinations of grains, such as Sichuan bran vinegar (SBV), Shanxi aged vinegar (SAV), Zhenjiang aromatic vinegar (ZAV), and Fujian Monascus vinegar (FMV) in China and Ehime black vinegar in Japan (JBV). Understanding the chemical compositions of different vinegars can provide information about nutritional values and the quality of the taste. This study investigated the vinegar metabolome using a combination of GC-MS, conventional LC-MS, and chemical isotope labeling LC-MS. Different types of vinegar contained different metabolites and concentrations. Amino acids and organic acids were found to be the main components. Tetrahydroharman-3-carboxylic acid and harmalan were identified first in vinegar. Various diketopiperazines and linear dipeptides contributing to different taste effects were also detected first in vinegar. Dipeptides, 3-phenyllactic acid, and tyrosine were found to be potential metabolic markers for differentiating vinegars. The differently expressed pathway between Chinese and Japanese vinegar was tryptophan metabolism, while the main difference within Chinese vinegars was aminoacyl-tRNA biosynthesis metabolism. These results not only give insights into the metabolites in famous types of cereal vinegar but also provide valuable knowledge for making vinegar with desirable health characteristics.


Introduction
As a fermented acidic condiment and preservative, vinegar has been used worldwide for more than 5000 years [1]. White vinegar is produced mainly from diluted alcohol [2], whereas black vinegar is produced from whole grains to elicit their distinct characteristics. In China, four famous black vinegar brands exist: Sichuan bran vinegar (SBV: 保寧醋), Zhenjiang aromatic vinegar (ZAV: 鎮江香醋), Shanxi aged vinegar (SAV: 山西老陳醋), and Fujian Monascus vinegar (FMV: 永春老醋). In the production of Chinese vinegar, prepared grain starters are first saccharified and then fermented to produce alcohol and acetic acid simultaneously [3]. Although the basic chemistry is the same, starters and ingredients differ substantially among the four brands.
The starter of SBV is bran koji (Fu Qu in Chinese) with wheat, rice, glutinous rice, and more than 60 medicinal herbs. After the starter fermentation is matured, uncooked wheat bran and corn are added and aerobically mixed for more than one month. After stirring to produce acetic acid, the fermented product is stored in a closed jar for maturation for at least six months before leaching. The starter of ZAV is koji (Da Qu in Chinese) with wheat, barley, and pea. The starter is mixed with cooked glutinous rice and is fermented in a fed-batch culturing style with wheat bran and rice hull in an open concrete basin with mechanical mixing. The maturation period is at least three months. The starter of SAV is similar to that of ZAV, with barley and pea at a ratio of 7:3. The starter, amounting to 60% of the raw materials, is mixed with sorghum for fermentation. Half of the material fermentation matured above is heated in a closed jar at 80~90 • C for 5~6 days (fuming) and then mixed with the remaining half for leaching. The filtrate is then matured for at least six months. Lastly, the starter of FMV is red koji (Hong Qu in Chinese) with red yeast japonica rice. The starter is mixed with steamed glutinous rice, and fermentation takes three years using fed-batch culturing in 1-year intervals.
In Japan, a different type of black vinegar (JBV) is produced [4][5][6]. It uses steamed rice with koji and yeast to produce alcohol first. Then, the broth is filtered before acetic acid fermentation takes place in a steel tank for 3 months. There is no fed-batch culturing or simultaneous fermentation of alcohol and acetic acid.
Non-targeted metabolic approaches, typically based on GC-MS and LC-MS, have been applied widely in various food analyses, such as tea [7] and milk [8], providing a view of metabolic composition. Chemical isotope labeling (CIL) LC-MS is a high-performance and quantitative profiling method. It is based on chemical derivatization using a "divideand-conquer" strategy in which different labeling methods are used to analyze different chemical groups, including amine/phenol, hydroxyl, carboxyl, and carbonyl metabolome groups [9].
In this study, we used a combination of GC-MS, conventional LC-MS, and CIL LC-MS for a comprehensive metabolome analysis of the five famous black vinegars in China and Japan. We aimed to identify and reveal broader chemical classes, compare the different metabolites, and reveal responsible differentiating metabolic pathways in cereal vinegar. These results not only clarify metabolites in cereal vinegar but also provide valuable knowledge for improving the quality of vinegar.

General Information of Identified Metabolites
A total of 1285 metabolites were identified with high confidence (Figure 1), in which 1190, 97, and 68 metabolites were identified by CIL LC-MS, label-free LC-MS, and GC-MS methods, respectively. Among them, 1130 (accounting for 87.9%), 67 (5.2%), and 28 (2.3%) metabolites were uniquely identified by CIL LC-MS, label-free LC-MS, and GC-MS, respectively. Among the common metabolites detected by two or more methods, 30 metabolites were identified by both CIL LC-MS and GC-MS, and 20 were identified by both CIL LC-MS and label-free LC-MS. Only 10 metabolites (<1% in total) were identified by all three methods. These results indicate that CIL LC-MS, label-free LC-MS, and GC-MS are complementary techniques for comprehensive metabolomic analysis.
In this study, we focused on determining the common metabolites found in all five types of vinegar and then examining their relative abundance differences inferred by ion intensities. We did not explore unique metabolites found in each type of vinegar. In a future study, it will be interesting to characterize unique metabolites that may play major roles in differentiating nutritional values and tastes.

Non-Targeted-Based Analysis of Metabolites
When the metabolites detected by GC-MS and label-free LC-MS were analyzed using a principal component analysis (PCA) biplot, the five vinegars were largely separated into three groups: JBV, FMV, and the rest (ZAV, SAV, and SBV) ( Figure 2). The PCA scree plots are shown in Supplemental Figure S1. The first component (PC1) explained ca. 40% of the variability and separated JBV and FMV from the rest. The second component (PC2) explained > 20% of the variability and separated JBV from FMV. In the GC-MS data, the top 10 contributing metabolites were plant-oriented saccharides and branched-chain amino acids: lactic acid (9), DL-valine (14), L-isoleucine (30), 2-hydroxyhexanoic acid (35), L-threonic acid (108), 3-phenyllactic acid (118), xylitol (145), D-xylopyranose (151), β-arabinopyranose (165), and L-iditol (209) ( Figure 2I). In the label-free LC-MS data, the top 10 metabolites were amino acids and dipeptides, especially leucine in positive mode and phenolics in negative mode ( Figure 2II,III). Furthermore, the PCA biplot showed that JBV, FMV, and SBV could be separated only by these metabolites in the GC-MS data ( Figure 2I). Meanwhile, JBV and FMV could be separated by label-free LC-MS in positive mode ( Figure 2II). 3-Phenyllactic acid (Phla) was identified by both GC-MS and label-free LC-MS in negative mode, where the ion abundances of ZAV, SAV, and SBV were much higher than those of FMV and JBV ( Figure 2I,III).

Pathway Annotation in Vinegar Metabolome
To reveal the main metabolic difference among FMV, JBV (Japan), and the rest (ZAV, SAV, and SBV, all from China), the metabolites identified by CIL LC-MS with high confidence (tier 1 and tier 2) were used as inputs for pathway analyses using MetaboAnalyst (https://www.metaboanalyst.ca/, accessed on 20 March 2021) [12]. As shown in Figure 5, the main difference between Chinese vinegars and Japanese vinegar ( Figure 5I, JBV vs. FMV and the rest) was tryptophan metabolism, followed by the citrate cycle (TCA cycle) and lysine degradation, while the main difference between FMV and the rest was aminoacyl-tRNA biosynthesis ( Figure 5II).

Discussion
Vinegar is rich in bioactive compounds and considered to exhibit physiological effects such as antioxidation, the prevention of obesity, and the regulation of blood pressure [6]. The integration of non-targeted and chemical isotope labeling (CIL) approaches allows for the detection of a large number of metabolites, shedding light on the metabolic composition of this complex food matrix.
Cyclo(pro-leu) and cyclo(phe-pro) are cyclic (pro)-containing dipeptides formed by a dehydration-condensation reaction, and they have been detected in various fermented and heated foods, in which they were recognized as not only flavor compounds but also bioactives; for example, cyclo(pro-leu) could deter influenza infection [15]. Moreover, other cyclic (pro)-containing dipeptides such as cyclo(pro-val) (cosine value = 0.933), cyclo(propro) (COSMIC = 0.953), and cyclo(his-pro) (COSMIC = 0.948) were also annotated with high confidence (Table S2). Linear dipeptides were also identified [16]. They greatly affect the taste of vinegar but have received relatively little attention in previous research [4,6].  (Table S3). Furthermore, FMV, JBV, and the rest (ZAV, SAV, and SBV) could be separated by glutamyl-valine (A-613), leucyl-aspartate (A-721), and valyl-lysine (A-2231) ( Figure 3I,II). PLS-DA cross-validation details for component 1 revealed Q2 = 0.852 and R2 = 0.988, which indicate that the model is stable and reliable. This indicates that dipeptides could be used as metabolic markers for vinegar discrimination and classification. In terms of amino acid-related pathways, tryptophan metabolism was the best differentiating pathway between Japanese (JBV) and Chinese (FMV, ZAV, SAV, and SBV) vinegars, and aminoacyl-tRNA biosynthesis was the most active within the Chinese vinegars, as inferred from the KEGG Ontology categories ( Figure 5).
From the metabolic analysis, we could also make an inference about the manufacturing process. The detection of an Amadori product N-(1-deoxy-1-fructosyl)phenylalanine, also identified in tea [21] and botanical extracts [22], implied the heating (Maillard) process of vinegar production. Its amount was the highest in JBV and the lowest in SAV.
Tetrahydroharman-3-carboxylic acid and harmalan, belonging to aromatic β-carboline alkaloids, can be found in traditional medicinal plants and kanjang (Korean soy source), exhibiting biological activities such as antimitotic activity [23] and anti-neuroinflammatory activities [24], and they might originate from the raw materials used in making the vinegar.

GC-MS Analysis
A 100 µL vinegar sample was extracted with 1 mL of mixed solvent (isopropanol: acetonitrile:water = 3:2:2, v/v/v) in a 2 mL Eppendorf tube using a Retsch MM400 Mixer Mill (Retsch GmbH&Co, Germany) at 30 Hz/s for 3 min and then centrifuged at 12,000× g for 5 min at 4 • C. After centrifugation, 200 µL of the supernatant was transferred to a new 1.5 mL Eppendorf tube and freeze-dried using a vacuum freeze drier (FDU-2110, EYELA, Japan). The dried sample was subjected to a two-step derivatization process with 100 µL of methoxyamine hydrochloride in pyridine (20 mg/mL) at 37 • C for 90 min, followed by 100 µL of MSTFA containing 1% trimethylchlorosilane at 37 • C for 90 min. The derivatized vinegar samples (1 µL) were injected into an Agilent GC 9000 coupled to 5977B mass-selective detector (MSD) (Agilent Technologies, St Louis, MA, USA). The separation was carried out on a DB-5MS fused silica capillary column (30 m × 0.25 mm × 0.25 µm film thickness, J&W Scientific, Folsom, CA, USA). The helium flow rate was 1.0 mL/min, and analysis was conducted in split mode (25:1). The column temperature was initially maintained at 80 • C for 2 min, raised from 80 • C to 200 • C at 4 • C/min, held at 200 • C for 1 min, then increased from 200 • C to 230 • C at 5 • C/min, and maintained at 230 • C for 5 min. The other MSD conditions were as follows: injector temperature was 250 • C, electron energy was 70 eV, and mass range was 20-600 amu. Compounds were identified using the NIST17 library as well as by comparing their retention indices (RIs) in the NIST Chemistry WebBook (http://webbook.nist.gov/, accessed on 15 January 2020). For relative intensity comparison, signal intensity of a metabolite was normalized by the total ion chromatogram (TIC) intensity. Each sample was analyzed in triplicate (n = 3).

Conventional or Label-Free LC-MS Analysis
The sample was centrifuged at 12,000× g for 10 min, and the supernatant was ultrafiltered through a 10 kDa cut-off filter device (Vivacon 500, Sartorius Stedim Lab Ltd, Stonehouse, UK). Samples were analyzed on a Shimazu L30A ultrahigh-pressure liquid chromatography (UHPLC) system coupled to a TripleTOF™ 5600 mass spectrometer (AB SCIEX, Foster City, CA, USA) equipped with a DuoSpray ion source. Chromatographic separation of the vinegar samples was performed on Zorbax Eclipse Plus C18 column (100 × 2.1 mm, 1.8 µm, Agilent Technology, Little Falls, DE, USA). The column was maintained at 40 • C, and the injection volume was 3 µL. Compounds were eluted using a binary mobile phase at a flow rate of 0.5 mL/min. The mobile phase contained solvent A (0.1% formic acid in water (v/v)) and B (0.1% formic acid in acetonitrile (v/v)). An elution linear gradient program was performed as follows: The raw DDA data (.wiff format) files were converted to ABF format using Reifycs Abf (Analysis Base File) Converter (Reifycs Inc., Tokyo, Japan) as input to MS-DIAL (v4.16) for peak detection and alignment [25]. The alignment results of signals normalized with total ion chromatogram (TIC) intensity were exported from MS-DIAL for relative quantification for principal component analysis. The peaks with representative MS/MS spectra were annotated by MS-FINDER (v3.30) [26], SIRIUS4 (v4.01) [27], and Global Natural Products Social Molecular Networking (GNPS) [28].

CIL LC-MS Analysis
Each sample was diluted to 2 mM after sample normalization using the total concentration of labeled metabolites measured by LC-UV. Sample labeling strictly followed the SOP in the provided kit from Nova Medical Testing Inc. (NovaMT) (Edmonton, AB, Canada) [10]. An Agilent eclipse plus reversed-phase C18 column (150 × 2.1 mm, 1.8 µm particle size) was used to separate the labeled sample, and Agilent 1290 LC linked to Bruker Impact II QTOF Mass Spectrometer was used to detect the labeled metabolites. Solvent A was water with 0.1% formic acid, whereas solvent B was acetonitrile with 0.1% formic acid. The LC gradient was 25% B from 0 to 10 min, 99% B from 10 to 13.1 min, and 25% B from 13.1 to 16 min, and it ended at 25% B. The flow rate was 400 µL/min. The column oven temperature was 40 • C, and mass range was 220-1000 m/z. The raw mass data were exported to csv file with Bruker DataAnalysis 4.4. Data analysis was performed using IsoMS Pro 1.2.5 (NovaMT Inc., Edmonton, AB, Canada) with the parameters: minimum m/z: 240; maximum m/z: 1000; saturation intensity: 20,000,000; retention time tolerance: 22 s; and mass tolerance: 10 ppm. Identified metabolites were searched against the in-house mass spectral library NovaMT Metabolite Database v2.0 based on the accurate mass and retention time with retention time tolerance of 30 s and mass tolerance of 10 ppm. For each sample, the analysis was performed in triplicate. For total metabolite concentration measurement, the amine/phenol labeled sample was centrifuged at 15,000 g for 10 min. An Agilent 1290 UPLC system with a photodiode array (PDA) detector was used for quantifying the total labeled or pooled sample concentration, as described in a previous study [29].

Pathway Analysis
To investigate and compare the metabolic pathways present in vinegar, the Kyoto Encyclopedia of Genes and Genomes (KEGG) was used as the backend knowledgebase. Non-targeted analysis of metabolites identified by a combination of GC-MS and UHPLC-QTOF-MS with high confidence (Tables S1 and S2) and the tier 1 and tier 2 metabolites identified by CIL LC-MS (Table S3) were mapped to KEGG pathways to reveal the main pathway distributions among the different types of vinegar using MetaboAnalyst (https: //www.metaboanalyst.ca/, accessed on 20 March 2021) [12].

Statistical Analysis
Statistical analyses were performed in Excel and other software. Principal component analysis (PCA) was carried out with correlation matrix by the FactoMineR packages in opensource software language R (version 3.6.3). Partial least-squares discriminant analysis (PLS-DA) was performed by MetaboAnalyst 5.0 (https://www.metaboanalyst.ca/, accessed on 20 March 2021) [12]. Venn diagrams were used to determine the commonly identified metabolites using chemical isotope labeling LC-MS, label-free LC-MS, and GC-MS and visualized by jvenn, an interactive Venn diagram viewer (http://www.bioinformatics.com. cn/static/others/jvenn/index.html, accessed on 20 March 2021 ) [30].

Conclusions
In this study, a combination of GC-MS, label-free LC-MS, and chemical isotope labeling LC-MS was used to profile the chemical compositions of five black vinegars. We aimed to identify and reveal broader chemical classes, compare the different metabolites, and reveal responsible differentiating metabolic pathways in cereal vinegar. Different types of vinegar showed different compositions of metabolites. Various diketopiperazines and linear dipeptides contributing to different taste effects were identified. Dipeptides, 3-phenyllactic acid, and tyrosine could be used as metabolic markers for differentiating cereal vinegars. As far as we know, cyclic peptides are reported here for the first time as vinegar metabolites and are considered important components. Hydroxyphenyllactic acid, 3-phenyllactic acid, lactic acid, and pyroglutamic acid were abundant organic acids. Amino acid-related pathways such as tryptophan metabolism and aminoacyl-tRNA biosynthesis were the top differentiating pathways. These results not only give insights into metabolic compositions in different cereal vinegars but also provide valuable knowledge for consumers and for making vinegar with desirable health characteristics.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/metabo12050427/s1, Table S1: List of metabolites identified from GC-MS with high confidence; Table S2: List of metabolites identified from UHPLC-QTOF-MS with high confidence; Table S3: List of peak pairs detected and metabolites identified from CIL LC-MS measurement of the samples with high confidence; Figure S1: PCA scree plots; Figure S2: Some details of main metabolites annotated by SIRIUS4, GNPs, and MS-FINDER.