Compositional differences among Chinese soy sauce types studied by 13C NMR spectroscopy coupled with multivariate statistical analysis

Author’s Accepted Manuscript Compositional Differences among Chinese Soy Sauce Types Studied by 13C NMR Spectroscopy Coupled with Multivariate Statistical Analysis Ghulam Mustafa Kamal, Xiaohua Wang, Bin Yuan, Jie Wang, Peng Sun, Xu Zhang, Maili Liu www.elsevier.com/locate/talanta

PII: DOI: Reference:

S0039-9140(16)30352-6 http://dx.doi.org/10.1016/j.talanta.2016.05.033 TAL16583

To appear in: Talanta Received date: 1 March 2016 Revised date: 8 May 2016 Accepted date: 13 May 2016 Cite this article as: Ghulam Mustafa Kamal, Xiaohua Wang, Bin Yuan, Jie Wang, Peng Sun, Xu Zhang and Maili Liu, Compositional Differences among Chinese Soy Sauce Types Studied by 13C NMR Spectroscopy Coupled with Multivariate Statistical Analysis, Talanta, http://dx.doi.org/10.1016/j.talanta.2016.05.033 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 galley proof before it is published in its final citable 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.

Compositional Differences among Chinese Soy Sauce Types Studied by 13

C NMR Spectroscopy Coupled with Multivariate Statistical Analysis

Ghulam Mustafa Kamala,b, Xiaohua Wanga,b, Bin Yuana,b, Jie Wanga, Peng Suna, Xu Zhanga*, Maili Liua* a

Key Laboratory of Magnetic Resonance in Biological Systems, State Key Laboratory of

Magnetic Resonance and Atomic and Molecular Physics, Centre for Magnetic Resonance, Wuhan Institute of Physics and Mathematics, Chinese Academy of Sciences, Wuhan, 430071, P. R. China b

University of Chinese Academy of Sciences, 100049, Beijing, P. R. China

*Corresponding Authors: 1.

Maili Liu, Professor/Research Director State Key Laboratory of Magnetic Resonance and Atomic and Molecular Physics, Centre for Magnetic Resonance, Wuhan Institute of Physics and Mathematics, Chinese Academy of Sciences, Wuhan, 430071, P. R. China E-mail: [email protected]

2.

Xu Zhang, Professor State Key Laboratory of Magnetic Resonance and Atomic and Molecular Physics, Centre for Magnetic Resonance, Wuhan Institute of Physics and Mathematics, Chinese Academy of Sciences, Wuhan, 430071, P. R. China E-mail: [email protected]

Abstract Soy sauce a well known seasoning all over the world, especially in Asia, is available in global market in a wide range of types based on its purpose and the processing methods. Its composition varies with respect to the fermentation processes and addition of additives, preservatives and flavor enhancers. A comprehensive 1H NMR based study regarding the metabonomic variations of soy sauce to differentiate among

1

different types of soy sauce available on the global market has been limited due to the complexity of the mixture. In present study,

13

C NMR spectroscopy coupled with

multivariate statistical data analysis like principle component analysis (PCA), and orthogonal partial least square-discriminant analysis (OPLS-DA) was applied to investigate metabonomic variations among different types of soy sauce, namely super light, super dark, red cooking and mushroom soy sauce. The main additives in soy sauce like glutamate, sucrose and glucose were easily distinguished and quantified using

13

C

NMR spectroscopy which were otherwise difficult to be assigned and quantified due to serious signal overlaps in 1H NMR spectra. The significantly higher concentration of sucrose in dark, red cooking and mushroom flavored soy sauce can directly be linked to the addition of caramel in soy sauce. Similarly, significantly higher level of glutamate in super light as compared to super dark and mushroom flavored soy sauce may come from the addition of monosodium glutamate. The study highlights the potentiality of

13

C NMR based metabonomics coupled

with multivariate statistical data analysis in differentiating between the types of soy sauce on the basis of level of additives, raw materials and fermentation procedures. Graphical Abstract

2

Key words: Soy sauce, 1H NMR, 13C NMR, multivariate analysis, metabonomics, PCA

1. Introduction Soy sauce or soya sauce is a condiment made from pastes of boiled soybeans, roasted grain, brine, and Aspergillus oryzae or Aspergillus sojae molds during the course of a well established fermentation process. After a two-step fermentation process, the paste is pressed, producing a liquid, which is the soy sauce, and a solid byproduct, which is often used as an animal feed [1-3]. Soy sauce is widely used in East and Southeast Asian cuisines as a traditional food ingredient and condiment. Initially soy sauce originated from China in 2nd century BC and then spread to all over the Asia. Presently it is widely used even in Western cuisines and prepared foods [4]. Soy sauce has a distinct yet basic taste of umami due to the acidic amino acids mainly aspartic acid and glutamic acid which come from the hydrolysis of soy proteins and wheat gluten during the course of fermentation. The other contributors to the distinct

3

taste of soy sauce are the low molecular weight compounds, such as acidic peptides and monosodium L-glutamate (MSG) which impart salty and umami taste to soy sauce [5-7]. A wide range of types of soy sauce are available with typical earthy, salty and brownish liquid used in seasoning of foods during cooking or serving at the table. Many different kinds of soy sauce are manufactured in China, Japan, Korea, Indonesia, Vietnam, Myanmar and other countries. The variation in composition and taste is mainly imparted by different ways and durations of fermentation, ratios of added water, salt and fermented soy, or through the addition of other ingredients like flavorings and preservatives. Super light and super dark are the two basic types of soy sauce most widely available. Light soy sauce is a thin (low viscosity), opaque; lighter brown soy sauce, brewed by first culturing steamed wheat and soybeans with Aspergillus, and then letting the mixture ferment in brine. It is the main soy sauce used for seasoning, since it is saltier, has less noticeable color and a distinct flavor. Dark soy sauce is a darker and slightly thicker soy sauce made from light soy sauce. This soy sauce is produced through prolonged aging. It usually contains added caramel, in some cases molasses to give it a distinctive appearance. It has a richer, slightly sweeter and less salty flavor than light soy sauce. This variety is mainly used during cooking, whereas partly used to add color and flavor to a dish after cooking, since its flavor develops during heating. Red cooking and mushroom flavored soy sauce are two extensions of dark soy sauce which are prepared by adding extra spices, additives and flavors. These two types are mostly used while cooking some specific dishes [8, 9]. The procedures of their fermentation are almost the same; however, their tastes are different mainly caused by different types and levels of additives, flavorings, taste developers and preservatives. Amino acids, sugars, organic acids and some other minor compounds are produced by enzymatic actions of different microorganisms during the fermentation process. The composition and level of these metabolites varies among soy sauce produced by different manufacturers and fermented from various raw materials, because of some compounds like isoflavones which are derived from specific raw materials such as soy beans [10]. While hundreds of chemical compounds have been identified in soy sauce, but the cause of its unique flavor still remains mysterious [8]. 4

Besides, a variety of preservatives (benzoic acid, p-hydroxy benzoate), flavorings (caramel, molasses, monosodium glutamate, lactic acid, citric acid, KCl) and sweeteners (acesulfame K and saccharine) have been reported to be added in the final soy sauce product. The allowance of these additives in foods should be based on the Codex General Standards for food additives [11, 12]. Many of these additives pose serious threats to the health of consumers if present in concentrations higher than the permissible level. So the consumers have a right to be aware of the compositional features of the foods they are eating. Presently metabonomics has proved to be a promising method to improve the understanding of quality parameters about dietary materials like differentiation based on composition, geographical variations and processing that may lead to food quality control and authentications [13, 14]. Metabonomics or metabolomics makes use of different equipments like mass spectroscopy (MS), nuclear magnetic resonance (NMR) or combination of NMR spectroscopy with LC-MS or MS for the comprehensive study of biomaterials [15]. The multivariate statistical modeling of the data obtained from these techniques can discriminate the metabonomic variations among samples very efficiently, so it got vast adoption in metabonomic studies. NMR is a more preferred and widely used technique suitable for metabonomics as it requires little or no sample preparation, and is non destructive in providing rich information about the metabolite structure in complex biological mixtures. The data obtained from NMR spectroscopy is multivariate. So, NMR spectroscopic measurements coupled with chemometrics or multivariate statistical analyses can provide an efficient way to extract useful information from the complex spectral data sets by providing dimensionality reduction in data set. In this way, pattern recognition can provide an outline to differentiate the samples and to address the possible metabonomic variations depicting the compositional differences. Two most preferable methods of pattern recognition analysis are principal component analysis (PCA) and partial least square-differential analysis (PLS-DA) which is widely used for the mapping of inherent patterns in large data sets [16-18]. Orthogonal partial least squarediscriminant analysis (OPLS-DA) is a further advancement of PLS-DA which shows an efficient discrimination among two or more groups (classes) using multivariate data sets [19, 20], it has an advantage over PLS-DA in a way that it uses single component as 5

predictor for the group/class whereas the other component can show variation orthogonal to the predicting component. In recent years, there has been increasing consumer interest to have a clear knowledge about the composition and the origin of the food products. A need therefore, exists to unveil how the types of soy sauce differ from each other. Studies related to the bioactive amines, isoflavones, amino acids, flavoring and taste compounds, phytochemicals and toxic compounds in soy sauce have been reported in literature [6, 7, 21, 22]. A few reports are also available considering the pattern recognition and geographical variations [8, 23] and metabolic changes during aging and fermentation of soy sauce [24, 25]. Herein, to evaluate the metabonomic differences in soy sauce types available on the local market, namely super light, super dark, red cooking and mushroom flavored soy sauce,

13

C NMR spectroscopy coupled with multivariate statistical data analysis was

applied. In most of the reported literature, techniques MS, GC-MS, LC-MS and 1H NMR have been used for this purpose. Although 1H NMR is commonly used for metabolic profiling, it suffers from signal overlap between chemically similar metabolites. 1H NMR based metabonomics analysis most often follows the water pre-saturated pulse sequences. The water pre-saturation usually causes the suppression of the resonances closer to the water signals, this in turn often causes false quantification of those metabolites. 13C NMR spectroscopy on the other hand being more informative does not bear this kind of problems. Therefore, to circumvent the problem,

13

C NMR spectroscopy with larger

chemical shift range has been used to sort out the metabonomic/compositional differences among soy sauce types on the market.

2. Material and Methods 2.1.

Sample Collection and Preparation A total of 24 samples from four different types of soy sauce (Super light, Super

dark, Red cooking and Mushroom flavored) were purchased from a local market. The pH of all the samples was monitored and adjusted to 5.0 ± 0.05 with phosphate buffer. Samples were prepared by mixing 60 µL soy sauce with 480 µL phosphate buffer (0.05 M sodium phosphate, pH 5.0 ± 0.05) and 60 µL 2.5 mM trimethylsilyl propanoic acid in 6

D2O (TSP), and then centrifuged at 13,000 rpm for 10 minutes. The supernatants were transferred to 5 mm NMR tubes. D2O provided field frequency lock while TSP served as internal reference standard. 2.2.

1

H and 13C NMR Analysis

1

H and

13

C NMR spectra were acquired on a Bruker Avance 600 NMR

spectrometer (Bruker Biospin GmbH, Germany) operating at 600.1699 and 150.912 MHz respectively at a temperature of 298 K using TXI CryoProbe. For the 1H NMR spectra, the H2O signal was suppressed by presaturation and the parameters for observation were as follows: number of data points, 32K; spectral width, 9000 Hz; acquisition time, 1.8 s; repetition time, 1s; number of scans, 128. All

13

C NMR spectra were acquired using

QDEPT135 pulse sequence; with which all carbons include quaternary carbons were obtained. For each sample 8K transitions were collected in 32K data points using a spectral width of 240 ppm with a repetition time of 2.0 s and an acquisition time of 0.45 s. Chemical shifts were referenced with that of TSP (1H & 13C, ơ 0.00 ppm). 1H and

13

C

spectra were apodised with 0.3 and 3.0 Hz exponential line broadening function before Fourier transformation (FT), respectively. Signal assignment for representative samples was facilitated by 2D 1H-1H total correlation spectroscopy (TOCSY), hetero-nuclear single quantum correlation spectroscopy (HSQC), and online spectral databases. 2.3.

NMR Data Analysis Chemical shifts migration present between different NMR spectra is one among

the most disturbing factor with respect to the multivariate data analysis [26]. The discovery of patterns in the spectra is often obscured due to this problem [27-29]. Instrumental factors, changes in pH, temperature, and concentrations of specific ions or salt and overall dilution are among the most important factors causing the chemical shift migration. However, the effect of these factors is not the same on all resonances. To address the alignment and phase problems, all spectra were first manually phased and baseline corrected by Topspin software 3.2 (Bruker Biospin, Germany). The raw data were then automatically aligned using NMRSpecs software which is free for academic research [30]. The regions without significant signals (106-230ppm) were removed prior to the spectral alignment. All spectral data were reduced into 0.2 ppm spectral buckets

7

and integrated. The aligned and bucketed spectra were normalized to the total spectral area in order to compensate for the total concentration difference. 2.4.

Multivariate Data Analysis The normalized NMR data sets were imported to SIMCA 14 (Umetrics, Umea,

Sweden) for multivariate statistical analysis. The par scaling was applied for all multivariate analyses. PCA, an un-supervised pattern recognition analysis was applied to reveal the intrinsic variations in the data set and to diagnose any possible outlier. OPLSDA models were then explored to find out the metabolites responsible for discrimination among them. The quality of the model was defined by total variance of the two components at a confidence level of 95%. The overall predictive ability of the model is assessed by cumulative Q2 representing the fraction of the variation of the Y that can be predicted by the model, which was extracted according to the internal cross-validation default method of the SIMCA software. All the models were validated using CV ANOVA test within SIMCA at p<0.05. This method applied to get these variations is simple and robust, having a general applicability to data mining from metabolomic and other similar kinds of data. The significantly varying metabolites extracted from OPLS-DA S plots were shown with standard error bar graphs (Fig. 7A & B). For standard error bar graphs, relative metabolite concentration was calculated from the integrals of selected metabolite NMR signals (least overlapping ones). 2.5.

Chemicals All chemical reagents were of analytical grade. Deuterium oxide (D2O) (99.9%),

trimethylsilyl propanoic acid (TSP) sodium salt (99.9 atom %), Na2HPO4, NaH2PO4 were purchased from Sigma (St. Louis, MO).

3. Results and Discussion 3.1.

1

H and 13C NMR Spectroscopy

Figure 1 & Fig. 2 show the representative 1D 1H and

13

C NMR spectra of super

light, super dark and red cooking soy sauce. The assignments of the metabolites have been carried out making use of 2D NMR (1H-1H TOCSY and 1H-13C HSQC), online NMR spectral database like SDBS (http://sdbs.db.aist.go.jp/sdbs/cgi-bin/cre_index.cgi), 8

MMCD

(http://mmcd.nmrfam.wisc.edu)

and

HMDB

(http://www.hmdb.ca)

published information [24]. A total of 22 signals were identified through

13

and

C NMR

spectra of all types of soy sauce (Fig. 1A, B & Fig. 2) with different ranges of concentration. The variation in metabolites concentration might be linked to the difference in manufacturing processes like fermentation period and the addition of additives (preservatives, sweeteners and salts) and flavors.

Figure 1A & 1B As shown in Figure 1A, the overall 1H NMR fingerprints of four categories of soy sauce are almost similar. However, a close inspection of spectra revealed that the spectral regions between 3.4-4.2ppm were highly overlapped due to the signals of oligosaccharides and glucose, the enlarged view for this region is shown in Fig. 1B. Sugars are the major compounds in complex food mixtures and they not only overlap with each other but also obscure the resonances of some minor compounds like organic acids [32, 33]. This effect may cause variations in intensities of the signals in different samples, and lead to errors in quantification of metabolites. It may not be much problematic in case of super light soy sauce samples but this problem aggravates in case of other three types of soy sauce, because they contain higher amounts of caramel as an additive. Moreover, the base line was also much distorted in case of the super dark, red cooking and mushroom flavored soy sauce spectra. This problem could not be managed by applying some base line and phase correction methods. we tried CPMG pulse sequence to avoid this problem but that was also not useful for solving these issues. Deconvolution of these overlapped signals is of much importance for quantification purposes. In addition, 1H NMR based analyses mostly makes use of water suppressed pulse sequences for the analysis of most biological samples. Resonances closer to the water peak (oligosaccharides and glucose mostly) are usually suppressed partially by 1H NMR analysis [31]. Furthermore, the carbohydrates are capable of adopting a range of dynamically interchanging conformations in solution [34]. This situation in complex biofluids causes remarkable chemical shift migration in signals in 1H NMR spectra. Although the phosphate buffer was used to remove the pH variation among samples, the 9

overlapping signal patterns, especially in the regions of 3.4−4.2 could not be removed by sample preparation. These overlaps led to some errors and uncertainties in signal recognition, alignment, and binning of data processing for PCA. The strong signal overlapping of 1H resonances increased the difficulty of and frequency of errors in the identification of variables in PCA and OPLS-DA models. Therefore,

13

C NMR spectra

were used in the classification of Chinese soy sauces types on the basis of raw materials, addition of additives and their traditionally different processes of manufacture. 13C NMR spectra of soy sauce yielded well separated signals which are more informative and useful for further information recovery.

Figure 2 Figure 2 shows the 13C NMR spectra of the corresponding soy sauce samples. The signals were narrow and less overlapped than those in 1H NMR spectrum. Although no clear differences were observed in overall patterns of the spectra, distinct differences in composition and concentrations of metabolites were observed among species. Besides, the overlaps observed in 1H NMR spectra due to oligosaccharide overlaps were not found in 13C NMR spectra. It has been shown previously that

13

C NMR spectroscopy provides

complementary component information and has potential in reducing the problems of overlap that occur in 1H NMR spectroscopy [35]. Therefore, the 13C NMR spectroscopy may be appropriate for metabolomics of soy sauce. In the manuscript, DEPT has been used for 13C detection, the sensitivity of 13C spectra is enhanced by polarization transfer via scalar coupling from 1H (I) to 13C (S). This polarization transfer is dependent on one bond coupling constant (1JIS), spin system (IS1, IS2, IS3) and transfer delay (Δ), therefore, the sensitivity enhancement is not uniform because of different attached protons and 1JIS values which in turn limit the use of polarization transfer to quantitative studies in metabolomics [36]. Similarly, the long T1 relaxation time reduces signal intensities which may also cause problems in quantification of metabolites. Nevertheless, for metabolomic studies, all samples are usually analyzed under the similar conditions, those parameters such as 1JIS, and T1 times generally remain stable in a certain range, therefore, it can be

10

used to differentiate different type of soy sauces by giving the overall pattern of metabolites that can be interpreted [37, 38]. 3.2.

Multivariate Statistical Analysis To see the differences in concentration of metabolites in different types of soy

sauce 24 samples representing different types of soy sauce were analyzed by 1H and 13C NMR spectroscopy. PCA is an unsupervised classification method requiring no prior knowledge of the data set and acts as a screening model to reduce the dimensionality of data while preserving most of the variance within it [39]. The PCA score scatter plot with R2X (62%) and Q2 (23%) is shown in Figure 3, which is derived from 13C NMR spectra of four different types of Chinese soy sauce, namely super light, super dark, mushroom flavored and red cooking soy sauce. R2X represents the goodness of fit of the PCA model [40]. The PCA model shows super light soy sauce samples were separated from the other soy sauce which shows unique metabolic attributes. Whereas, no clear separation among the super dark, red cooking and mushroom flavored soy sauce was observed indicating a non significant variation among them. It was expected because red cooking and mushroom flavored soy sauce are basically the extended varieties of dark soy sauce which are further developed by addition of flavorings and extra additives to dark soy sauce.

Figure 3

It was evident from PCA scatter plot that the main differences were present between the super light and the other three types of soy sauce. So, in order to exploit the different metabonomic characteristics of different types of soy sauce, pair wise OPLSDA models were calculated for super light soy sauce with the other 3 types of soy sauce. The important variables responsible for discrimination among different types of soy sauce were captured by OPLS-DA S plots. Cut-off values for the covariance of |p| ≥ 0.05 and for the correlation of |p(corr)| ≥ 0.5 were used.

Figure 4A, B & C

11

OPLS-DA model calculated for super light and super dark soy sauce is shown in Figure 4A & 4B and positively correlated right part enlarged and shown in Fig. 4C. Figure 4A shows the separation of super light soy sauce from super dark soy sauce with statistical significance of R2X (Cum.) 73%, R2Y (Cum.) 100% and Q2 (Cum.) 89%. Variables responsible for separation in OPLS-DA scatter plot were captured by S plot shown in Figure 4B. The S plot helps identify statistically and potentially biochemically significant variables, on the basis of both of their contribution and reliability. The significant variables were identified according to the

13

C NMR assignment information.

The names of the metabolites corresponding to the separation between super light and super dark soy sauce are leucine, valine, glutamate, glucose, sucrose, phenylalanine and lactate. Compared with super dark, super light soy sauce contained significantly higher levels of leucine, valine and glutamate, whereas, super dark soy sauce samples were characterized with increased concentrations of lactate, phenylalanine, glucose and sucrose.

Figure 5A, B & C

To clearly identify the underlying variables for separation between super light and red cooking soy sauce, another OPLS-DA model was further constructed (Figure 5A & 5B) and positively correlated right part enlarged and shown in Fig. 5C. OPLS-DA score scatter plot (Figure 5A) showed a clear separation among super light and red cooking soy sauce accounting for R2X (Cum) 58%, R2Y (Cum) 98% and Q2 (Cum) 85%. The metabolites contributing significantly to this separation in OPLS-DA score plot were captured by calculating an OPLS-DA S plot shown in Figure 5B. Lactate, acetate, phenylalanine, glucose and sucrose were the metabolite, more concentrated in red cooking soy sauce which contribute significantly towards the difference, whereas, super light soy sauce contained increased concentrations of leucine, isoleucine, valine and glutamate.

Figure 6A, B & C

12

Another OPLS-DA model was also performed representing super light and mushroom flavored soy sauce samples (Fig. 6A & B) and positively correlated right part enlarged and shown in Fig. 6C. The OPLS-DA score scatter plot showed clear separation among the two types of soy sauce with a statistical significance of R2X (Cum) 60%, R2Y (Cum) 98% and Q2 (Cum) 84%. The corresponding S plot (Fig. 6B) shows that the underlying metabolites responsible for the discrimination among the two types of soy sauce were Leucine, Isoleucine, valine and glutamate, which are more concentrated in super light soy sauce, whereas acetate, lactate, phenylalanine, glucose and sucrose were the main contributors towards the separation of mushroom flavored soy sauce. Metabolites with significantly different concentrations among soy sauce types were captured from the OPLS-DA models and plotted as standard error bar charts in terms of their concentrations relative to the concentration of TMSP (Figure 7A & B). The data were further analyzed using t-test, the resulting OPLS-DA models are explained below. 3.2.1. Carbohydrate: Two important sugars like sucrose and glucose were observed in all the four types of soy sauce. While comparing super light with super dark and red cooking and mushroom flavoured types, lower levels of sucrose and glucose were observed in super light soy sauce, whereas red cooking, mushroom flavoured and super dark soy sauce were characterized by increased levels of sucrose. Less concentration of sugar in soy sauce or any fermentation product indicates the elevated consumption of carbohydrates by micro organisms during the course of fermentation as studied earlier [41]. Earlier studies suggested that the oligosaccharide consumption by wine yeast in grape wines increased over time, a significant decrease in level of oligosaccharides were found during the 6 months of aging period after alcoholic fermentation. Moreover, carbohydrates in soy sauce are helpful in production of immunoglobulin A (Ig A) in vivo and in vitro [42]. Similarly, Tetragenococcus halophilus, a halophilic lactic acid bacterium (LAB) is also active in the fermentation processes of soy sauce and possesses immunomodulatory activity in vitro [43, 44]. Halophilic or osmophilic bacteria degrades soybean and wheat starches into polysaccharides and oligosaccharides over time, in this way soy sauce can be an important dietary source of improving host defences. However,

13

the fermentation process should be optimized to retain a necessary amount of sugars required to maintain peculiar characteristics of pure fermented soy sauce. Concentration of sucrose was significantly higher in all the other three types as compared to light soy sauce (Fig. 7A). This composition characterization was in accordance with their color classifications. Soy sauce is also classified as light or white soy sauce and dark or red soy sauce, based on its color. Light or white soy sauce is light in color, a natural result of fermentation. This product has a superior flavor, and is produced by following a lower fermentation temperature and longer fermentation time, to fully develop the flavor components. Dark soy sauce along with its variants like red cooking and mushroom flavored soy sauce are particularly desirable for cooking food where a dark color is preferred, they have an intense darker color and a higher viscosity compared with white or light soy sauce due to the addition of caramel [45]. Caramel is mainly produced by heating a mixture of sucrose (or saccharose), lactose and, maltose and used in traditional cooking, backing and food additive for coloring. While heating the mixture for caramel production, lactose breaks down to monosaccharide units whereas sucrose and maltose remain abundant in caramel being thermally more stable [46, 47]. Therefore, significantly higher levels of sucrose in red cooking, mushroom flavoured and dark soy sauce indicate the high caramel addition in the final product, and suggest that sucrose is a widely used source for caramel in China. This finding is confirmed by the presence of caramel as one of the ingredient mentioned in labels on the bottles of these types of soy sauce. The concentrations of glucose are much higher in red cooking soy sauce and mushroom flavored soy sauce, however, non significant variation was observed in concentration of glucose among super light and red cooking soy sauce. Besides, the concentration of other metabolites, such as acetate and lactate which are also involved in glucose metabolism were not found to be significantly different from the other two soy sauces. Apparently, concentration variations of all the other metabolites among all 4 types of soy sauce have similar trends, being in a narrow range and stable. This indicates that the raw materials and the fermentation procedures were almost similar for all the four types of soy sauce, whereas, extra high concentrations of glucose arise in red cooking and mushroom soy sauce may because of the manual additives. Red cooking and mushroom 14

flavored soy sauce are basically produced from dark soy sauce by addition of flavors. Therefore, the differences in concentration of glucose among these three types of soy sauce could be due to additives. 3.2.2. Lactate, glutamate and acetate: Higher levels of lactate were observed in super dark, mushroom flavoured and red cooking soy sauce, whereas decreased lactate concentration was found in super light soy sauce (Fig. 7A). Increased levels of lactate suggest the increased fermentation period indicating that halophilic or osmoprotolerant lactic acid bacteria were involved in the brine fermentation of soy sauce [48]. Lactate is the main organic acid and very important in determining the quality of soy sauce. It was produced naturally during fermentation or sometimes added manually. It imparts a mild and balanced acidic flavour to soy sauce with some lingering effects. Another important organic acid, acetate was also observed as a non-significantly discriminating higher metabolite in super dark, red cooking and mushroom flavored soy sauce as compared to the light soy sauce samples (Fig. 7A). Acetate along with formate is produced during fermentation by the activity of osmotolerant lactic acid bacteria. Both acetate and formate were reported to inhibit the growth of halophilic yeasts such as Saccharomyces rouxii and Torulopsis versatilis, during the formation of Japanese fermented soy sauce “shoyu” [49, 50]. Both acetate and lactate are produced during fermentation as TCA cycle intermediates. The similar trend in variation of these intermediates suggests the reliability of

13

C NMR spectroscopic quantification of

metabolites. Significantly higher amount of glutamate was observed in super light soy sauce as compared to super dark and mushroom flavoured soy sauce whereas a less significant variation in concentration of glutamate was observed among super light and red cooking soy sauce (Fig. 7A). Glutamate is an important amino acids present both in raw and fermented soy sauce [51] and serves as a key intermediate in LAB metabolism because of its utilization by most of the aminotransferases as the donor of amino groups. Glutamate is also produced by glutamate dehydrogenase during the course of fermentation, from 2oxoglutarate, a TCS metabolism intermediate. Glutamic acid and glutamates are flavour enhancers and umami (savoury) tasting condiments to foods. Glutamate is a natural constituent of many fermented foods like soy 15

sauce, fermented beans and cheese etc. Glutamate salt such as monosodium glutamate (MSG) is a widely used additive in final soy sauce products to impart characteristic umami taste in combination with bitter phenylalanine. However, MSG is not a healthy additive if present in concentrations higher than permissible level. A number of in vitro studies indicate that glutamate is a potent neurotoxin at higher concentrations and can destroy neurons by apoptosis [52]. Thus there is a need to show whether the MSG, glutamine rich raw material like wheat gluten instead of soybean was used in the manufacturing process or the glutamate was produced by fermentation.

Figure 7A & 7B

3.2.3. Leucine, isoleucine, valine and phenylalanine: Leucine, isoleucine and valine were observed in higher concentrations in super light soy sauce as compared to the other three types of soy sauce (Fig. 7B). Amino acids in soy sauce are produced by the microbial activity during the fermentation. The microorganisms break down the soy and wheat proteins into amino acids through proteolysis and peptidase activity and use them as a nitrogen source for their growth. The same was observed in significant consumption and synthesis of valine in grape wine fermentation [41]. However, no significant variations were observed in amino acids contents among four types of soy sauce. According to the taste characteristics amino acids are grouped as umami (aspartic acid and glutamic acid), sweet (alanine, glycine, serine, threonine, proline and lysine), bitter (arginine, histidine, isoleucine, leucine, methionine, phenylalanine, tryptophan, tyrosine and valine) and tasteless (cysteine)[53, 54]. So, phenylalanine along with MSG imparts an intense umami taste to the soy sauce which was obvious while tasting the super dark, mushroom flavoured and red cooking soy sauce samples under our study. 3.3.

Non Discriminant Metabolites: Tyrosine, aspartate, formate, malate, malonate,

alanine, choline, threonine, methionine and betaine were also observed in all types of soy sauce with very low signals causing no significant discrimination among types of soy sauce. Tyrosine is an important amino acids present in lesser amounts in soy sauce, peanut, chocolates, white wines. Tyrosine and its metabolites impact the human mental health, adapt to stress, blood pressure, skin colour, ability to stand pain and metabolic rate 16

[55, 56]. Very low signals of tyrosine were detected in soy sauce types under our investigation. Tyrosine is often converted to Tyramine by microbial decarboxylation in fruits and vegetables during fermentation process [57]. The low levels of tyrosine detected in all four types of soy sauce provide an evidence of the longer aging period of the soy sauce fermentation indicating the conversion of tyrosine to tyramine. Aspartate signals were also detected in spectra of all the three types of soy sauce without contributing towards discrimination. During fermentation process some strains of T. halophilus convert aspartate to alanine through decarboxylation and so there is decreases in aspartate level whereas increase in the level of alanine along with the increasing aging period [24, 58-60]. The low aspartate content and comparatively high levels of alanine can give insight into the fermentation period of the product. Weak signal of malonate was also detected in soy sauce types which is also produced as intermediates of TCA cycle during fermentation [61]. Choline was also observed showing a very low signal. Normally along with the increasing fermentation period, choline level decreases because it is converted to betaine and glycine by the activity of a lactic acid bacterium Tetragenococus halophila [62, 63]. Choline is an important essential nutrient available in a variety of food including dried raw soybeans (120mg/100g) and soy sauce (18mg/100g). It was firstly recommended by The US Institute of Medicine in 1998 [64, 65]. It is involved in neurotransmitter synthesis (acetylcholine), cell-membrane signalling (phospholipids), lipid transport (lipoproteins) and methyl group metabolism (homocystine reduction) etc. [66]. A choline deficient food is observed to cause many disorders like liver steatosis and damage in humans [67] growth retardation, renal dysfunction, haemorrhage or bone abnormalities in animals [68, 69]. Low signals of threonine, methionine and betaine were also detected in all soy sauce types. These are produced along with the other major discriminant amino acids by proteolysis and peptidase activity of microorganisms involved in soy sauce fermentation.

4. Conclusion We provide

13

C NMR spectroscopic method to evaluate the compositional

differences among different types of soy sauce. Super light soy sauce which is the primary soy sauce was efficiently distinguished from the super dark, red cooking and 17

mushroom flavoured soy sauce. Glutamate, sucrose and glucose were found to be the most significantly discriminating metabolites among all the four types of soy sauce. Less significant and almost similar trends of variation in concentrations of all the other metabolites indicates that almost similar fermentation procedures were followed. However, higher levels of glutamate, sucrose and glucose indicate the manual addition of monosodium glutamate and sucrose abundant caramel to the final soy sauce product. Our results can be helpful to further evaluate and authenticate these differences using an in house fermented soy sauce to compare with commercial soy sauces. The study also highlights the potentiality of

13

C NMR spectroscopy for the

metabonomic evaluation of the complex biological mixtures as most of the previously reported studies have made use of 1H NMR spectroscopy for this purpose.

13

C NMR

studies of complex systems can facilitate the assignments and deconvolution of otherwise overlapped signals in 1H NMR spectra, such as glucose, sucrose and glutamate which are the main components and additives of soy sauce and are difficult to quantify using 1H NMR spectroscopy.

Acknowledgement The authors would like to thank the financial support from National Major Basic Research Program of China (2013CB910200), National Natural Science Foundation of China (21120102038, 81227902, 21475146, 21221064, 20875098 and 21075132), and the Chinese Academy of Sciences (CAS) - The World Academy of Sciences (TWAS) president’s scholarship program.

References [1] P.B. Leboffe M, Microbiology Laboratory Theory and Application, 2nd ed.2012. [2] D. Fukushima, Soy Proteins for Foods Centering around Soy Sauce and Tofu, J Am Oil Chem Soc, 58 (1981) 346-354. [3] D. Fukushima, Industrialization of fermented soy sauce production centering around Japanese shoyu, in: S. KH (Ed.) Industrialization of indigenous fermented foods, Marcel Dekker, New York, 1989, pp. 1-88. [4] K. Farrel, Spices, Condiments and Seasonings, Van Nostrand Reinhold Company Inc., 1985. [5] T.K. Lioe H. N., Yasuda M, Evaluation of peptide contribution to the intense umami taste of Japanese soy sauce, Journal of Food Science, 71 (2006) 270-283.

18

[6] K.W. Lioe H. N., T. Aoki, & M. Yasuda, Chemical and sensory characteristics of low molecular weight fractions obtained from three types of Japanese soy sauce (Shoyu)Koikuchi, tamari and shiro shoyu, JFood Chemistry, 100 (2007) 1669-1677. [7] S. Kaneko, K. Kumazawa, O. Nishimura, Isolation and identification of the umami enhancing compounds in Japanese soy sauce, Bioscience, biotechnology, and biochemistry, 75 (2011) 1275-1282. [8] E. Kinoshita, Y. Ozawa, T. Aishima, Differentiation of soy sauce types by HPLC profile pattern recognition. Isolation of novel isoflavones, Advances in experimental medicine and biology, 439 (1998) 117-129. [9] R. Parkinson, 5 Types of Chinese Soy Sauce. [10] E. Kinoshita, T. Sugimoto, Y. Ozawa, T. Aishima, Differentiation of soy sauce produced from whole soybeans and defatted soybeans by pattern recognition analysis of HPLC profiles, Journal of agricultural and food chemistry, 46 (1998) 877-883. [11] T.Y. Chu, C.L. Chen, H.F. Wang, A rapid method for the simultaneous determination of preservatives in soy sauce, J Food Drug Anal, 11 (2003) 246-250. [12] C.A. Commission, Proposed Draft Codex Standard For Soy Sauce, Joint Fao/Who Food Standards Programme, 2004. [13] Y. Wang, H. Tang, J.K. Nicholson, P.J. Hylands, J. Sampson, E. Holmes, A metabonomic strategy for the detection of the metabolic effects of chamomile (Matricaria recutita L.) ingestion, Journal of agricultural and food chemistry, 53 (2005) 191-196. [14] K.S. Solanky, N.J. Bailey, B.M. Beckwith-Hall, S. Bingham, A. Davis, E. Holmes, J.K. Nicholson, A. Cassidy, Biofluid 1H NMR-based metabonomic techniques in nutrition research - metabolic effects of dietary isoflavones in humans, The Journal of nutritional biochemistry, 16 (2005) 236-244. [15] D.J. Crockford, E. Holmes, J.C. Lindon, R.S. Plumb, S. Zirah, S.J. Bruce, P. Rainville, C.L. Stumpf, J.K. Nicholson, Statistical heterospectroscopy, an approach to the integrated analysis of NMR and UPLC-MS data sets: application in metabonomic toxicology studies, Analytical chemistry, 78 (2006) 363-371. [16] S. Esslinger, J. Riedl, C. Fauhl-Hassek, Potential and limitations of non-targeted fingerprinting for authentication of food in official control, Food Research International, 60 (2014) 189-204. [17] E.T.M.D. Keun H. C., Antti H., Bollard M. E., Beckonert O., Holmes E., Lindon J. C., & Nicholson J. K., Improved analysis of multivariate data by variable stability scaling: Application to NMR-based metabolic profiling, Analytica chimica Acta, 490 (2003) 265-276. [18] L. Mannina, A.P. Sobolev, D. Capitani, Applications of NMR metabolomics to the study of foodstuffs: Truffle, kiwifruit, lettuce, and sea bass, Electrophoresis, 33 (2012) 2290-2313. [19] M. Bylesjo, M. Rantalainen, O. Cloarec, J.K. Nicholson, E. Holmes, J. Trygg, OPLS discriminant analysis: combining the strengths of PLS-DA and SIMCA classification, Journal of Chemometrics, 20 (2006) 341-351. [20] J. Trygg, S. Wold, Orthogonal projections to latent structures (O-PLS), Journal of Chemometrics, 16 (2002) 119-128. [21] L.R. Guidi, M.B. Abreu Gloria, Bioactive amines in soy sauce: Validation of method, occurrence and potential health effects, Food Chemistry, 133 (2012) 323-328.

19

[22] W.S. Fu, Y. Zhao, G. Zhang, L. Zhang, J.G. Li, C.D. Tang, H. Miao, J.B. Ma, Q. Zhang, Y.N. Wu, Occurrence of chloropropanols in soy sauce and other foods in China between 2002 and 2004, Food additives and contaminants, 24 (2007) 812-819. [23] L. Xu, Y. Li, N. Xu, Y. Hu, C. Wang, J. He, Y. Cao, S. Chen, D. Li, Soy sauce classification by geographic region and fermentation based on artificial neural network and genetic algorithm, Journal of agricultural and food chemistry, 62 (2014) 1229412298. [24] B.K. Ko, H.J. Ahn, F. van den Berg, C.H. Lee, Y.S. Hong, Metabolomic insight into soy sauce through (1)H NMR spectroscopy, Journal of agricultural and food chemistry, 57 (2009) 6862-6870. [25] H.G. Mao C, Du X, Cui M, Gao S, Biochemical Changes in the Fermentation of the Soy Sauce Prepared with Bittern, Advance Journal of Food Science and Technology, 5 (2013) 144-147. [26] T.M. Ebbels, J.C. Lindon, M. Coen, Processing and modeling of nuclear magnetic resonance (NMR) metabolic profiles, Methods Mol Biol, 708 (2011) 365-388. [27] M. Defernez, I.J. Colquhoun, Factors affecting the robustness of metabolite fingerprinting using 1H NMR spectra, Phytochemistry, 62 (2003) 1009-1017. [28] H. Witjes, W.J. Melssen, H.J. in 't Zandt, M. van der Graaf, A. Heerschap, L.M. Buydens, Automatic correction for phase shifts, frequency shifts, and lineshape distortions across a series of single resonance lines in large spectral data sets, J Magn Reson, 144 (2000) 35-44. [29] J.T.W.E. Vogels, A.C. Tas, J. Venekamp, J. VanderGreef, Partial linear fit: A new NMR spectroscopy preprocessing tool for pattern recognition applications, Journal of Chemometrics, 10 (1996) 425-438. [30] Y.G. Liu, Yunling.; Cheng, Ji.; Wang, Jie.; Xu, Fuqiang A Processing Method for Spectrum Alignment and Peak Extraction for NMR Spectra, Chinese Journal of Magnetic Resonance, 2 (2015) 382-392. [31] M. Liu, H. Tang, J.K. Nicholson, J.C. Lindon, Recovery of underwater resonances by magnetization transferred NMR spectroscopy (RECUR-NMR), J Magn Reson, 153 (2001) 133-137. [32] P. Maes, Y.B. Monakhova, T. Kuballa, H. Reusch, D.W. Lachenmeier, Qualitative and quantitative control of carbonated cola beverages using (1)H NMR spectroscopy, Journal of agricultural and food chemistry, 60 (2012) 2778-2784. [33] Y.B. Monakhova, A.M. Tsikin, T. Kuballa, D.W. Lachenmeier, S.P. Mushtakova, Independent component analysis (ICA) algorithms for improved spectral deconvolution of overlapped signals in 1H NMR analysis: application to foods and related products, Magn Reson Chem, 52 (2014) 231-240. [34] A. Almond, P.L. DeAngelis, C.D. Blundell, Dynamics of hyaluronan oligosaccharides revealed by 15N relaxation, J Am Chem Soc, 127 (2005) 1086-1087. [35] H.C. Keun, O. Beckonert, J.L. Griffin, C. Richter, D. Moskau, J.C. Lindon, J.K. Nicholson, Cryogenic probe 13C NMR spectroscopy of urine for metabonomic studies, Analytical chemistry, 74 (2002) 4588-4593. [36] B. Jiang, N. Xiao, H. Liu, Z. Zhou, X.A. Mao, M. Liu, Optimized quantitative DEPT and quantitative POMMIE experiments for 13C NMR, Analytical chemistry, 80 (2008) 8293-8298.

20

[37] F. Wei, K. Furihata, M. Koda, F. Hu, T. Miyakawa, M. Tanokura, Roasting process of coffee beans as studied by nuclear magnetic resonance: time course of changes in composition, Journal of agricultural and food chemistry, 60 (2012) 1005-1012. [38] F. Wei, K. Furihata, M. Koda, F. Hu, R. Kato, T. Miyakawa, M. Tanokura, (13)C NMR-based metabolomics for the classification of green coffee beans according to variety and origin, Journal of agricultural and food chemistry, 60 (2012) 10118-10125. [39] H.K. Choi, Y.H. Choi, M. Verberne, A.W. Lefeber, C. Erkelens, R. Verpoorte, Metabolic fingerprinting of wild type and transgenic tobacco plants by 1H NMR and multivariate analysis technique, Phytochemistry, 65 (2004) 857-864. [40] Y. Jung, J. Lee, J. Kwon, K.S. Lee, H. Ryu do, G.S. Hwang, Discrimination of the geographical origin of beef by (1)H NMR-based metabolomics, Journal of agricultural and food chemistry, 58 (2010) 10458-10466. [41] H.S. Son, G.S. Hwang, W.M. Park, Y.S. Hong, C.H. Lee, Metabolomic characterization of malolactic fermentation and fermentative behaviors of wine yeasts in grape wine, Journal of agricultural and food chemistry, 57 (2009) 4801-4809. [42] H. Matsushita, M. Kobayashi, R.-I. Tsukiyama, M. Fujimoto, M. Suzuki, K. Tsuji, K. Yamamoto, Stimulatory effect of Shoyu polysaccharides from soy sauce on the intestinal immune system, International Journal of Molecular Medicine, 22 (2008) 243247. [43] A.P.D.R.H. Marcello Villar, Jorge J. Sanchez, Raul E. Trucco, Guillermo Oliver Isolation and Characterization of Pediococcus halophilusfrom Salted Anchovies (Engraulis anchoita), Applied and Environmental Microbiology 49 (1985) 664-666. [44] S. Masuda, H. Yamaguchi, T. Kurokawa, T. Shirakami, R.F. Tsuji, I. Nishimura, Immunomodulatory effect of halophilic lactic acid bacterium Tetragenococcus halophilus Th221 from soy sauce moromi grown in high-salt medium, Int J Food Microbiol, 121 (2008) 245-252. [45] S.P. Y-H. Peggy Hsieh, and Jiangrong Li, in: E.R. Farnworth (Ed.) Handbook of Fermented Functional Foods, CRC Press, Taylor & Francis Group, Boca Raton, London, New York,, 2008, pp. 433-463. [46] N. Kuhnert, J.W. Drynan, J. Obuchowicz, M.N. Clifford, M. Witt, Mass spectrometric characterization of black tea thearubigins leading to an oxidative cascade hypothesis for thearubigin formation, Rapid Commun Mass Spectrom, 24 (2010) 33873404. [47] A. Golon, N. Kuhnert, Unraveling the Chemical Composition of Caramel, Journal of agricultural and food chemistry, 60 (2012) 3266-3274. [48] I.T. Nishimura I, Enomoto T, Dake Y, Okuno Y, Obata A, Clinical efficacy of halophilic lactic acid bacterium Tetragenococcus halophilusTh221 from soy sauce moromi for perennial allergic rhinitis, Allergology International, 58 (2009) 179-185. [49] D.J. Hentges, Influence of pH on the inhibitory activity of formic and acetic acids for Shigella, J Bacteriol, 93 (1967) 2029-2030. [50] F. Noda, K. Hayashi, T. Mizunuma, Antagonism between osmophilic lactic Acid bacteria and yeasts in brine fermentation of soy sauce, Appl Environ Microbiol, 40 (1980) 452-457. [51] B.S. Luh, Industrial-Production of Soy-Sauce, J Ind Microbiol, 14 (1995) 467-471. [52] Y.M. Zhang, B.R. Bhavnani, Glutamate-induced apoptosis in neuronal cells is mediated via caspase-dependent and independent mechanisms involving calpain and 21

caspase-3 proteases as well as apoptosis inducing factor (AIF) and this process is inhibited by equine estrogens, Bmc Neurosci, 7 (2006). [53] C. Schoenberger, Krottenthaler, M. & Back, W., Sensory and analytical characterization of nonvolatile taste-active compounds in bottom-fermented beers., Master Brewers Association of the Americas, 39 (2002) 210-217. [54] H. Kato, Rhue, M.R. & Nishimura, T., Role of free amino acids and peptides in food taste, American Chemical Society, Washington, DC, 1989. [55] J.B. Deijen, J.F. Orlebeke, Effect of Tyrosine on Cognitive Function and BloodPressure under Stress, Brain Res Bull, 33 (1994) 319-323. [56] J.B. Deijen, C.J.E. Wientjes, H.F.M. Vullinghs, P.A. Cloin, J.J. Langefeld, Tyrosine improves cognitive performance and reduces blood pressure in cadets after one week of a combat training course, Brain Res Bull, 48 (1999) 203-209. [57] K.R. Stratton, P.F. Worley, J.S. Litz, S.J. Parsons, R.L. Huganir, J.M. Baraban, Electroconvulsive Treatment Induces a Rapid and Transient Increase in Tyrosine Phosphorylation of a 40-Kilodalton Protein Associated with Microtubule-Associated Protein-2 Kinase-Activity, J Neurochem, 56 (1991) 147-152. [58] B.J. McMahon, C.J. Christensen, D.R. Gretch, J.L. Williams, D.G. Sullivan, D. Bruden, R.L. Carithers, T.W. Hennessy, C.E. Homan, H. Deubner, Alanine aminotransferase (ALT) levels over time in anti-HCV-positive persons who are HCV RNA positive compared with persons who are HCV RNA negative but riba positive., Hepatology, 30 (1999) 358a-358a. [59] M. Fernandez, A.B. Florez, D.M. Linares, B. Mayo, M.A. Alvarez, Early PCR detection of tyramine-producing bacteria during cheese production, J Dairy Res, 73 (2006) 318-321. [60] M. Fernandez, M. Zuniga, Amino acid catabolic pathways of lactic acid bacteria, Crit Rev Microbiol, 32 (2006) 155-183. [61] M. Aoshima, Y. Igarashi, Nondecarboxylating and decarboxylating isocitrate dehydrogenases: oxalosuccinate reductase as an ancestral form of isocitrate dehydrogenase, J Bacteriol, 190 (2008) 2050-2055. [62] J. Boch, B. Kempf, E. Bremer, Osmoregulation in Bacillus-Subtilis - Synthesis of the Osmoprotectant Glycine Betaine from Exogenously Provided Choline, J Bacteriol, 176 (1994) 5364-5371. [63] W.F.M. Roling, H.W. vanVerseveld, Characterization of Tetragenococcus halophila populations in Indonesian soy mash (Kecap) fermentation, Applied and Environmental Microbiology, 62 (1996) 1203-1207. [64] S.A.B.J.W.R. Patterson KY, H.J. Howe JC, USDA Database for the Choline Content of Common Foods, 2nd Release [65] I.o. Medicine, Dietary Reference Intakes for Folate, Thiamin, Riboflavin, Niacin, Vitamin B12, Panthothenic Acid, Biotin, and Choline Washington, DC, 1998. [66] J.T. Penry, M.M. Manore, Choline: An important micronutrient for maximal endurance-exercise performance?, Int J Sport Nutr Exe, 18 (2008) 191-203. [67] S.H. Zeisel, K.A. Dacosta, P.D. Franklin, E.A. Alexander, J.T. Lamont, N.F. Sheard, A. Beiser, Choline, an Essential Nutrient for Humans, Faseb J, 5 (1991) 2093-2098. [68] P.M. Newberne, A.E. Rogers, Labile Methyl-Groups and the Promotion of Cancer, Annu Rev Nutr, 6 (1986) 407-432. 22

[69] P. Handler, F. Bernheim, Choline Deficiency in the Hamster, P Soc Exp Biol Med, 72 (1949) 569-571.

Fig. 1 Assigned 1H NMR spectra of soy sauce. B is enlarged view of spectral regions from 3.1-5.1ppm. A1 and B1. Super light, A2 and B2. Super dark, A3 and B3. Red cooking and A4 and B4. Mushroom flavoured. Peaks: 1, TSP; 2, Isoleucine; 3, Leucine;, 4, Valine; 5, Threonine; 6, Alanine; 7, Lactate; 8, Acetate; 9, Glutamate; 10, Methionine; 11, Aspartate; 12, γ-amino butyric acid; 13, Choline; 14, αß-glucose; 15, ß-glucose; 16, Glucose & aliphatic region; 17, ß-glucose; 18, Solvent (H2O); 19, α-glucose; 20, Fumarate; 21, Tyrosine; 22, Phenylalanine; 23, Formate Fig. 2 13C NMR spectra of four types of Chinese soy sauces. A. Super light, B. Super dark, C. Red cooking and D. Mushroom flavoured. Peaks: 1, TSP; 2, Isoleucine; 3, Alanine; 4, Valine; 5, Threonine; 6, Acetate; 7, Leucine; 8, Glutamate; 9, Isoleucine; 10, Leucine; U1, Unknown; 11, Leucine; 12, Malonate; 13, Betaine; 14, Choline; 15, Aspartate; 16, Glutamate; 17, Valine; 18, Glucose & sucrose; 19, Tyrosine; 20, Phenylalanine; 21, Lactate; 22, Acetate Fig. 3 PCA score scatter plot derived from 13C NMR spectra of all types of Chinese soy sauces showing separation of Chinese super light soy sauce (green circle) from those of super dark (blue circle), red cooking (red circle) and mushroom flavored soy sauce (yellow circle). Fig. 4 (A) OPLS-DA score scatter plot derived from 13C NMR spectra of Super light (Green circle) and Super dark (Blue circle) soy sauce (B) S plot generated from OPLSDA model representing the metabolites responsible for discrimination among Super light and Super dark soy sauce. The range of the variables selected is highlighted with a rectangle. Cutoff values for the covariance of |p| ≥ 0.05 and for the correlation of |p(corr)| ≥ 0.5 were used. The variables in orange rectangles represent the metabolites responsible for differentiation in OPLS-DA score plot. (C) Right segment of S-plot enlarged to show the discriminating variables. Fig. 5 (A) OPLS-DA score scatter plot derived from 13C NMR spectra of Super light (Green circle) and Red cooking (Red circle) soy sauce (B) S plot generated from OPLSDA model representing the metabolites responsible for discrimination among Super light and Red cooking soy sauce. The range of the variables selected is highlighted with a rectangle. Cutoff values for the covariance of |p| ≥ 0.05 and for the correlation of |p(corr)| ≥ 0.5 were used. The variables in orange rectangles represent the metabolites responsible for differentiation in OPLS-DA score plot (C) Right segment of S-plot enlarged to show the discriminating variables.

23

Fig. 6 (A) OPLS-DA score scatter plot derived from 13C NMR spectra of Super light (Green circle) and Mushroom flavored (Yellow circle) soy sauce (B) S plot generated from OPLS-DA model representing the metabolites responsible for discrimination among Super light and Mushroom flavored soy sauce. The range of the variables selected is highlighted with a rectangle. Cutoff values for the covariance of |p| ≥ 0.05 and for the correlation of |p(corr)| ≥ 0.5 were used. The variables in orange rectangles represent the metabolites responsible for differentiation in OPLS-DA score plot (C) Right segment of S-plot enlarged to show the discriminating variables. Fig. 7A Relative concentrations of significantly different metabolites (Excluding amino acids) captured by OPLS-DA models derived from the 13C NMR spectra of super light (Blue), super dark (Red), red cooking (Green) and mushroom flavored (Purple) soy sauce. *Less significant **Highly significant Fig. 7B Relative concentrations of discriminating amino acids metabolites captured by OPLS-DA models derived from the 13C NMR spectra of super light (Blue), super dark (Red), red cooking (Green) and mushroom flavored (Purple) soy sauce. *Less significant **Highly significant

24

Figure 1

25

Figure 2

26

Figure 3

27

Figure 4

28

Figure 5

29

Figure 6

30

a

b 31

Figure 7

Highlights: 

13

C NMR spectroscopy offering long range of chemical shifts facilitates the

assignment and deconvolution of sucrose, glucose and glutamate which are otherwise difficult to be separated and deconvoluted due to serious overlaps in 1H NMR spectroscopy. 

Three important metabolites like sucrose, glucose and glutamate were found to be the main cause of discrimination among Chinese soy sauce types.



The significantly higher concentration of glucose, sucrose can be linked directly to the addition of caramel whereas, that of glutamate due to the addition of monosodium glutamate (MSG).

32