1. Introduction

Fresh or processed laver Pyropia yezoensis is traditionally and extensively consumed as popular food-stuff in China, Japan and Korea ^{[1, 2]}. The laver product is processed from fresh thallus of P. yezoensis by washing, cutting, roasting, and seasoning. These processing procedures are able to result in great nutrition changes. For instance, the mechanical rupture of seaweed tissues causes the leakage of nutrient substances. The high temperature roasting results in the degradation of proteins and polysaccharides. Importantly, a body of evidence has elucidated that the boiling changes the biochemical compositions and mineral contents of seaweeds ^[3]. The drying also causes the marked changes of the nutritional compositions of seaweed Sargassum hemiphyllum. Freeze-dried seaweed has the highest content of total amino acids, total polyunsaturated fatty acids and total vitamin C, compared with sun-dried and oven-dried seaweed ^[4].

Many studies have been conducted for the analyses of nutrient composition of fresh P. yezoensis. A series of primary metabolite or secondary metabolite contents have been elucidated such as proteins ^[5], polysaccharide ^{[6, 7]}, porphyran ^{[8, 9]}, vitamins ^[10] and minerals ^[11]. It has been indicated that fresh P. yezoensis is abundant in proteins, sulfated galactan, choline, taurine, inositol, eicosapentanoic acid and minerals such as Zn, Cu, Mn and Se ^{[5, 11]}. The contents of free amino acids are dependent on the genotype of P. yezoensis blades and the progression of harvest number ^{[12, 13]}. However, little information is available regarding the effects of processing on the nutrient composition of P. yezoensis.

The NMR-based metabolomic analysis has been extensively used for the characterization of complex foodstuffs. This is because the advantages of metabolomics in food analysis relate to it enabling direct sample analysis and holistic detection of large numbers of compounds in a single experiment. Such a strategy has been already successfully applied to green tea ^[14], pine-mushroom ^[15], Cheonggukjang ^[16], soy sauce ^[17], and crab paste ^[18]. It is particularly interesting to note that the metabolomics approach has been proved powerful for detecting the metabolome of Porphyra haitanensis and its alterations in response to high temperature stress ^[19].

In this study, the nutrient composition of raw materials, semi-finished products, and finished products of P. yezoensis was systematically analyzed using the NMR-based metabolomics strategy coupled with multivariate data analysis. The objective of this study was to understand the effects of food processing on the nutrient composition of P. yezoensis products.

2. Materials and Methods

Acetonitrile, K₂HPO₄·3H₂O, and NaH₂PO₄·2H₂O (all of analytical grade) were purchased from Guoyao Chemical Co., Ltd. (Shanghai, China). Sodium 3-trimethylsilyl [2,2,3,3-d₄] propionate (TSP) and deuterated water (D₂O, 99.9% D) were purchased from Sigma-Aldrich Inc. (St. Louis, MO, USA). Phosphate buffer (K₂HPO₄/NaH₂PO₄, , pH 7.4), containing 10% D₂O (v/v) and 0.005% TSP (w/v), was prepared in H₂O ^[20].

All samples of P. yezoensis were provided by ShenXian Laver Co., Ltd. (Lianyungang, Jiangsu Province, China). Raw material, semi-finished product, and finished product of P. yezoensis were obtained from the same thalli of P. yezoensis. The seaweed processing was described briefly as follows. (1) The raw materials were prepared by the following procedures: the fresh thalli of P. yezoensis were washed thoroughly with tap water, and cut coarsely, then dried in the air oven at 60°C for 90 min and preserved at 4°C. (2) The semi-finished products were obtained from the above raw materials after roasting at 160°C for 60 sec. (3) The finished products were obtained from the semi-finished products after seasoning procedure using a mixture of monosodium glutamate, soy sauce, and sugar as the seasoning paste, and subsequent roasting at 90°C for 60 sec. All these three products were manufactured based on standard procedure, and for sale to the customers.

The raw material, semi-finished product and finished product were sampled randomly and divided into ten portions, respectively. Each portion (about 0.05 g) was extracted with 600 µL of 50% aqueous acetonitrile by shaking with a TissueLyser (Qiagen/Retsch, Germany) at 20 Hz for 90 sec. The resultant extracts were denoted as A, B and C, sequentially. The supernatant for each sample was collected respectively after 10 min of centrifugation (12,000 rpm, 4°C). The remaining solid residues of each sample were further extracted once using the same extraction procedure. After centrifugation for 10 min (12,000 rpm, 4°C), the resultant two supernatants were combined and lyophilized following removal of acetonitrile in vacuo. The extracts were then reconstituted into 600 µL of phosphate buffer. Following centrifugation, 550 µL of the supernatant from each extract was transferred into a 5 mm NMR tube.

All ¹H NMR spectra were recorded at 298 K on a Bruker Avance III 800 MHz spectrometer operating at 800.20 MHz for ¹H, equipped with an inverse detection cryogenic probe (Bruker, Biospin, Germany). A standard one-dimensional NMR experiment was recorded using the first increment of NOESY pulse sequence (recycle delay-90°-t₁-90°-t_m-90°-acquisition). Typically, the 90° pulse length was adjusted to approximately 10 µs and t₁ was set to 6 µs. The water signal was suppressed by irradiation during the recycle delay (2 s) and mixing time (100 ms). Sixty-four transients were collected into 32 K data points for each spectrum with a spectral width of 20 ppm. Prior to Fourier transformation, an exponential window function with a line-broadening factor of 0.5 Hz was applied to all free induction decays. To facilitate NMR signal assignments, a range of 2D NMR spectra were acquired and processed as previously described ^{[21, 22]} for selected samples including the ¹H-¹H COSY, ¹H-¹H TOCSY, ¹H J-resolved, ¹H HSQC, and ¹H HMBC spectra.

The ¹H NMR spectral region δ 5.55-0.80 was divided into bins with an equal width of 3.2 Hz using the AMIX software package (V3.8, Bruker-Biospin, Germany) after manual phase- and baseline-corrections as well as calibration to TSP at 0.0 ppm. Regions δ4.96-4.67 and δ2.09-2.07 were removed to avoid the effects of imperfect water suppression and residual acetonitrile signal, respectively. Each bucketed region was then normalized to the sum of total spectral integrals to compensate for the inter-sample differences in sample volume/concentrations.

Multivariate data analysis was carried out on the normalized NMR data using SIMCA-P⁺ software (V.12, Umetrics, Umeå, Sweden). Initially, principal component analysis (PCA) was conducted to generate an overview of group clustering and to identify outliers, using a mean-centered scaling. Subsequently, orthogonal projection to latent structure with discriminant analysis (OPLS-DA) was performed using the unit-variance scaled NMR data as X-matrix and the group information as Y-matrix. To avoid over-interpreting, only two components were calculated for OPLS-DA models with obtained R² and Q² values as initial indicators for model quality. Each of OPLS-DA model qualities was evaluated by 7-fold cross-validation and further ensured by variance analysis of the cross-validated residuals (CV-ANOVA) method with p < 0.05 considered as valid ^[23] The loadings plots from these models were generated using an in-house developed MATLAB script following back-transformation ^[24], where signals were color-coded with correlation coefficients (r) to indicate significantly altered compounds. The color code stands for the absolute value of Pearson’s product-moment correlation coefficients (|r|), which indicates the weight of each variable contributing to the differentiation between groups. Here, the value of |r| greater than 0.602, was considered to be statistically significant (n = 10, p < 0.05).

Nutrient component quantification of the seaweed was carried out by equating the integrals of selected component NMR signals (non-overlapping ones) relative to that of internal reference (TSP) with known concentration. All the obtained nutrient component concentrations were also subjected to classical statistical analyses (one way-ANOVA) using SPSS 13.0 software with a Tukey post-test.

3. Results and Discussion

To have a complete view of nutrient composition of P. yezoensis, a detailed study of ¹H NMR spectra of P. yezoensis extracts obtained from raw material, semi-finished product and finished product was performed. As shown in Figure 1, 32 compounds were identified in the seaweed and its products according to a series of 2D NMR experiments and literature data ^{[19, 25, 26]}. The ¹H and ¹³C signal information of the compounds is tabulated in Table S1. The aqueous extracts of P. yezoensis and its products were dominated by 11 amino acids (isoleucine, leucine, valine, threonine, alanine, glutamate, glutamine, aspartate, asparagine, -alanine, and glycine), 11 carboxylic acids (6-deoxy-ascobate, lactate, 2-hydroxy-5-aminovalerate, 2-oxo-5-aminovalerate, acetate, malate, succinate, citrate, taurine,γ-amino-n-butyric acid, and isethionate), four choline metabolites (choline, choline-O-sulfate, betaine, and betaine aldehyde), four sugars (glucose, sucrose, floridoside, and isofloridoside), laminitol, and dimethylsulphoniopropionate (DMSP). Although these above compounds have been unambiguously assigned, some compounds detected have not been identified in this study (e.g., unknowns in Table S1). In these assigned compounds, a large number of nutrient components such as choline-O-sulfate, betaine, betaine aldehyde, floridoside, isofloridoside, laminitol, and DMSP were reported for the first time in P. yezoensis extracts, apart from amino acids and choline already reported previously ^{[11, 12, 13]}. In fact, the nutrient composition of P. yezoensis is highly similar to the metabolites of P. haitanensis, another species of genus Pyropia, which has been reported recently ^[19]. In addition, some known compounds including eicosapentanoic acid, inositol, porphyosin, and a sulfated galactan ^[5] in P. yezoensis were not observed in this study. The possible reason is that some of them remain in the residue during extraction or their concentrations were below the detection limit of NMR spectrometer.

Figure 1. Three typical 800 MHz ¹H NMR spectra of raw materials (A), semi-finished products (B), and finished products (C) of P. yezoensis. Compared to the chemical shift range at  0.9-4.7, the spectra in the regions  4.9-5.55 were displayed at 4-fold magnification. Resonance assignments are given in Table S1. Keys: 1, isoleucine; 2, leucine; 3, valine; 4, laminitol; 5, 6-deoxy-ascobate; 6, lactate; 7, threonine; 8, alanine; 9, 2-hydroxy-5-aminovalerate; 10, acetate; 11, 2-oxo-5-aminovalerate; 12, glutamate; 13, malate; 14, succinate; 15, glutamine; 16, citrate; 17, β-alanine; 18, aspartate; 19, dimethylsulphoniopropionate; 20, asparagine; 21, γ-amino-n-butyric acid; 22; choline-O-sulfate; 23, isethionate; 24, choline; 25, betaine; 26, taurine; 27, betaine aldehyde; 28, glycine; 29, β-glucose; 30,α-glucose; 31, L-isofloridoside; 32, D-isofloridoside; 33, floridoside; 34, sucrose; 35, sugar and amino acids α-CH resonance; 36, unidentified signal 1; 37, unidentified signal 2

Download as

PowerPoint Slide

Larger image(png format)

Figures index

Veiw figure

View current figure in a new window

View next figure

It is clear from the visual inspection of ¹H NMR spectra (Figure 1 A, and B) that P. yezoensis raw material extracts has higher signals of 6-deoxy-ascobate and choline-O-sulfate, but lower signals of choline, floridoside, and isofloridoside relative to the semi-finished product. The NMR profiling of finished product extracts (Figure 1C) was drastically different from those of above two groups. An increased intensity in the signals of two sugars (glucose and sucrose) were observed in the finished products relative to the other two groups, accompanied by a decreased intensity in the signals of laminitol, 6-deoxy-ascobate, 2-hydroxy-5-aminovalerate, 2-oxo-5-aminovalerate, choline-O-sulfate, and isethionate. These observations indicate that the nutrient composition of seaweed product strongly depends on the processing procedures. To obtain further details on the statistically significant compositional differences between seaweed and its products due to the food processing, multivariate data analyses were performed on the NMR data from three groups of samples.

Table S1. Summary of the nutrient composition identified in the aqueous extracts of P. yezoensis and its products

Download as

PowerPoint Slide

Larger image(png format)

Tables index

Veiw figure

View current table in a new window

View next table

PCA of the normalized NMR data obtained from P. yezoensis raw material and its product extracts was performed to give an overview of the dataset (Figure 2). No outliers were observed for all samples. The raw material, semi-finished product and finished product of P. yezoensis were clearly clustered into three groups with the first two principal components (PC1 and PC2), cumulatively explaining 96.3% of variables in the data set. This suggests that these three groups may have distinct compositional differences due to food processing. Moreover, the nutrient composition of raw materials and semi-finished products is more similar than that of finished products.

To further investigate which nutrient composition was significantly affected by food processing, an OPLS-DA model was constructed with compositional profile data as X-matrix and group information as Y-matrix. The cross-validated scores plots from OPLS-DA (Figure 3, left) showed that samples from three groups of seaweed clearly clustered into respectively separate groups with good model quality indicated by the high values of R²X and Q². The validity of two OPLS-DA models was further evaluated with the CV-ANOVA approach (p < 0.05). Both of these models passed the rigorous test of CV-ANOVA with extremely low r value as listed in Figure 3. Compared with semi-finished product, raw material exhibited a greater variation along the orthogonal axes (Figure 3a, left), probably due to more differentiation among individual raw materials than semi-finished products. OPLS-DA loading plots (Figure 3, right) were obtained to display the significantly altered compounds contributing to the group discrimination. Relative to raw materials, semi-finished products had higher levels of amino acids (isoleucine, leucine, valine, threonine, glutamine, and aspartate), carboxylic acids (lactate, 6-deoxy-ascobate, 2-oxo-5-aminovalerate, taurine and isethionate) as well as choline-O-sulfate. These alterations were also associated with relatively lower concentrations of alanine, choline, isofloridoside, and floridoside. After further processing of semi-finished products, a significant increase in the levels of glutamate, glucose, floridoside, and sucrose was observed in the finished products. A significant decrease was simutaneously observed in the finished products in the levels of amino acids (alanine, glutamine, aspartate, and glycine), carboxylic acids (6-deoxy-ascobate, 2-hydroxy-5-aminovalerate, 2-oxo-5-aminovalerate, taurine, and isethionate), choline metabolites (choline, choline-O-sulfate, and betaine) as well as laminitol. The dynamic changes of representative compounds with the food processing are shown in Figure 4. Their corresponding correlation coefficients, which are above the cutoff value 0.602, are listed in Table S2 in the Supporting Information. The concentrations of some compounds were obtained and subjected to classical statistical analysis (one-way ANOVA) as summarized in Table 1. The semi-quantitative results were consistent with the OPLS-DA ones. For example, the mean concentrations of floridoside and isofloridoside in semi-finished products were, respectively, 0.16 ± 0.02 mg/g and 0.55 ± 0.08 mg/g, an decrease of 86% and 59% from the cases in the raw materials accordingly (Table 1). After the seasoning and roasting of semi-finished products, finished products had the highly elevated sucrose, glucose, and glutamate, up to 38.67 ± 4.91 mg/g, 4.22 ± 0.55 mg/g, and 17.60 ± 1.93 mg/g, respectively. However, they displayed the greatly depleted choline-O-sulfate and 6-deoxy-ascobate, from 10.05 ± 0.66 mg/g and 2.59 ± 0.15 mg/g to 5.93 ± 0.86 mg/g and 1.31 ± 0.41 mg/g, respectively.

Figure 2. (Color online) PCA scores plot for extracts from raw materials (A, star), semi-finished products (B, circles), and finished products (C, diamond) of P. yezoensis. PC1 and PC2 represent 82.6% and 13.7% of the total variance, respectively

Download as

PowerPoint Slide

Larger image(png format)

Figures index

Veiw figure

View current figure in a new window

View previous figure

View next figure

Figure 3. (Color online) OPLS-DA scores (left) and coefficient-coded loadings plots (right) for the models discriminating the raw material (A, star), semi-finished product (B, circle), and finished product (C, diamond) extracts of P. yezoensis. (see Table S1 for compound identification key)

Download as

PowerPoint Slide

Larger image(png format)

Figures index

Veiw figure

View current figure in a new window

View previous figure

View next figure

Figure 4. (Color online) Dynamic alterations of key nutrient components of raw material (A), semi-finished product (B) and finished product (C) extracts of P. yezoensis. The color indicates a correction coefficient as scaled on the right-hand side. The warm colors (e.g., red) denote an increase in the levels of compounds in B and C against the levels in A and B, respectively, and the cool colors (e.g., blue) indicate a decrease. DOA, 6-deoxy-ascobate; OAV, 2-oxo-5-aminovalerate; HAV, 2-hydroxy-5-aminovalerate; COS, choline-O-sulfate

Download as

PowerPoint Slide

Larger image(png format)

Figures index

Veiw figure

View current figure in a new window

View previous figure

Table S2. The significantly altered compounds with corresponding correlation coefficients in the P. yezoensis extracts from raw materials, semi-finished products, and finished products, respectively

Download as

PowerPoint Slide

Larger image(png format)

Tables index

Veiw figure

View current table in a new window

View previous table

The above compositional alterations of P. yezoensis due to processing were significant, although Yoshie and coworkers have shown that there are no significant changes in any of the components during processing ^[27]. This divergence of results may be resulted from the different samples, detection technique, processing method and/or target compounds. Moreover, it is probable that nutrient composition of P. yezoensis is changed by a variety of procedures of food proceeding in this study. For example, the washing and cutting procedures mechanically damage the seaweed tissues, leading to the leakage of some free nutrients such as alanine, isofloridoside, and floridoside. Furthermore, roasting at 160°C may also cause the breakdown of proteins, which contributes to an increase in the levels of amino acids such as isoleucine, leucine, valine, threonine, glutamine, and aspartate in the semi-finished products.

For the finished products, a significant increase in the levels of glutamate, glucose and sucrose is highly correlated with various seasonings including monosodium glutamate, sugar, and soy sauce. It is intriguing to note that soy sauce contains a variety of compounds including amino acids, organic acids, and sugars ^[17], however, the supplement of soy sauce did not lead to the increased levels of these compounds in the finished products. On the contrary, four amino acids (alanine, glutamine, aspartate, and glycine), as well as five carboxylic acids (6-deoxy-ascobate, 2-hydroxy-5-aminovalerate, 2-oxo-5-aminovalerate, taurine, and isethionate) were greatly depleted in it. Hence, it was speculated that roasting may play some roles in causing the aforementioned compositional changes. Nevertheless, this study strongly suggests the great effects of food processing on the composition of P. yezoensis products, which is worthy to be fully considered. This is because all of these nutrient compositional alterations may lead to the changed quality and taste of seaweed products. For example, the score for the essential amino acids such as leucine, isoleucine, and valine is widely used for evaluation of protein quality ^[28]. Taurine, as an organic acid containing an amino group in man, involves in numerous physiological processes such as membrane stabilization, detoxification, and antioxidation ^[29]. Therefore, the food processing-caused alterations of nutrient levels probably highly affect the seaweed quality. In addition, many compounds are saporous substances. Free amino acids specially account for the taste of the seaweed products. For an instance, aspartate and glutamate exhibit interesting properties in flavor development, and glutamate is a main component in the taste sensation of umami ^[30]. Such significantly altered levels of amino acids may lead to the changed flavor of seaweed product.

¹H NMR spectroscopy analysis on P. yezoensis showed that its nutrient composition was largely dominated by 11 amino acids, 11 carboxylic acids, four choline metabolites, and four sugars. The different nutrient concentrations in seaweed extracts varied with processing procedures. Although seasoning unsurprisingly elevated the levels of glucose, sucrose, and glutamate in the finished products, other processing procedures such as washing and roasting may be also responsible for a significant alteration in the levels of amino acids, carboxylic acids, choline metabolites, laminitol, floridoside, and isofloridoside of seaweed from raw materials to finished products. It is therefore essential to take these factors into consideration and make appropriate choices for the quality and taste of seaweed products. Furthermore, this study demonstrates that NMR-based metabolomic strategy is of important values for understanding the effects of food processing on the nutrient composition of seaweed products.

Acknowledgements

This work was supported by grants from Ningbo Agricultural Technologies R & D Project (2012C10027), the National Natural Science Foundation of Ningbo (2014A610183, 2014A610179), Ningbo Marine Algae Biotechnology Team (2011B81007), and K. C. Wong Magna Fund in Ningbo University.

Statement of Competing Interests

The authors have no competing interests.

References

[1]	Asamizu, E, Nakajima, M, Kitade, Y, Saga, N, Nakamura, Y, & Tabata, S. (2003). Comparison of RNA expression profiles between the two generations of Porphyra yezoensis (Rhodophyta), based on expressed sequence tag frequency analysis. Journal of Phycology, 39 (5), 923-930.
	In article		CrossRef
[2]	Tseng, CK. (2004). The past, present and future of phycology in China. Hydrobiologia, 512, 11-20.
	In article		CrossRef
[3]	Santoso, J, Gunji, S, Yoshie-Stark, Y, & Suzuki, T. (2010). Mineral contents of Indonesian seaweeds and mineral solubility affected by basic cooking. Food Science and Technology Research, 12 (1), 59-66.
	In article		CrossRef
[4]	Chan, JC-C, Cheung, PC-K, & Ang, PO. (1997). Comparative studies on the effect of three drying methods on the nutritional composition of seaweed Sargassum hemiphyllum (Turn.) C. Ag.. J Journal of Agricultural and Food Chemistry, 45 (8), 3056-3059.
	In article		CrossRef
[5]	Noda, H. (1993). Health benefits and nutritional properties of nori. Journal of Applied Phycology, 5, 255-258.
	In article		CrossRef
[6]	Zhang, WY, Chen, H, Wang, SJ, Xia, ZH, Fei, XG, & Tan, RX. (2001). An acidic polysaccharide with xylose branches from Porphyra yezoensis. Chinese Science Bulletin, 46 (3), 207-209.
	In article		CrossRef
[7]	Zhou, CS, & Ma, HL. (2006). Ultrasonic degradation of polysaccharide from a red algae (Porphyra yezoensis). Journal of Agricultural and Food Chemistry, 54 (6), 2223-2228.
	In article		CrossRef
[8]	Takahashi, K, Hirano, Y, Araki, S, & Hattori, M. (2000). Emulsifying ability of porphyran prepared from dried nori, Porphyra yezoensis, a red alga. Journal of Agricultural and Food Chemistry, 48 (7), 2721-2725.
	In article		CrossRef
[9]	Hirano, Y, Hattori, M, & Takahashi, K. (2005). Interaction of porphyran with a hydrophobic surface and stabilization of liposomes. Journal of Agricultural and Food Chemistry, 53 (25), 9800-9804.
	In article		CrossRef
[10]	Miyamoto, E, Yabuta, Y, Kwak, CS, Enomoto, T, & Watanabe, F. (2009). Characterization of vitamin B12 compounds from Korean purple lavor (Porphyra sp.) products. Journal of Agricultural and Food Chemistry, 57 (7), 2793-2796.
	In article		CrossRef
[11]	Qu, WJ, Ma, HL, Pan, ZL, Luo, L, Wang, ZB, & He, RH. (2010). Preparation and antihypertensive activity of peptides from Porphyra yezoensis. Food Chemistry, 123 (1), 14-20.
	In article		CrossRef
[12]	Niwa, K, Furuita, H, & Aruga, Y. (2003). Free amino acid contents of the gametophytic blades from the green mutant conchocelis and the heterozygous conchocelis in Porhyra yezoensis Ueda (Bangiales, Rhodophyta). Journal of Applied Phycology, 15, 407-413.
	In article		CrossRef
[13]	Niwa, K, Furuita, H, & Yamamoto, T. (2008). Changes of growth characteristics and free amino acid content of cultivated Porhyra yezoensis Ueda (Bangiales Rhodophyta) blades with the progression of the number of harvests in a nori farm. Journal of Applied Phycology, 20, 687-693.
	In article		CrossRef
[14]	Tarachiwin, L, Ute, K, Kobayashi, A, & Fukusaki, E. (2007). 1H NMR based metabolomic profiling in the evaluation of Japanese green tea quality. Journal of Agricultural and Food Chemistry, 55 (23), 9330-9336.
	In article		CrossRef
[15]	Cho, IH, Kim, YS, & Choi, HK. (2007). Metabolomic discrimination of different grades of pine-mushroom (Tricholoma matsutake Sing) using 1H NMR spectrometry and multivariate data analysis. Journal of Pharmaceutical and Biomedical Analysis, 43 (3), 900-904.
	In article		CrossRef
[16]	Choi, H-K, Yoon, J-H, Kim, Y-S, & Kwon, DY. (2007). Metabolomic profiling of Cheonggukjang during fermentation by 1H NMR spectrometry and principal components analysis. Process Biochemistry, 42 (2), 263-266.
	In article		CrossRef
[17]	Ko, B-K, Ahn, H-J, van den Berg, F, Lee, C-H, & Hong, Y-S. (2009). Metabolomic insight into soy sauce through 1H NMR spectroscopy. Journal of Agricultural and Food Chemistry, 57 (15), 6862-6870.
	In article		CrossRef
[18]	Ye, YF, Zhang, LM, Tang, HR, & Yan, XJ. (2012). Survey of nutrients and quality assessment of crab paste by 1H NMR spectroscopy and multivariate data analysis. Chinese Science Bulletin, 57(25), 3353-3362.
	In article		CrossRef
[19]	Ye, YF, Zhang, LM, Yang, R, Luo, QJ, Chen, HM, Yan, XJ, & Tang HR. (2013). Metabolic phenotypes associated with high-temperature tolerance of Porphyra haitanensis strains. Journal of Agricultural and Food Chemistry, 61 (35), 8356-8363.
	In article		CrossRef
[20]	Xiao, CN, Hao, FH, Qin, XR, Wang, YL, & Tang, HR. (2009). An optimized buffer system for NMR-based urinary metabonomics with effective pH control, chemical shift consistency and dilution minimization. Analyst, 134: 916-925.
	In article		CrossRef
[21]	Xiao, CN, Dai, H, Liu, HB, Wang, YL, & Tang, HR. (2008). Revealing the metabonomic variation of rosemary extracts using 1H NMR spectroscopy and multivariate data analysis. Journal of Agricultural and Food Chemistry, 56 (21), 10142-10153.
	In article		CrossRef
[22]	Dai, H, Xiao, CN, Liu, HB, & Tang, HR. (2010). Combined NMR and LC-MS analysis reveals the metabonomic change in Salvia miltiorrhiza Bunge induced by water depletion. Journal of Proteome Research, 9 (3), 1460-1475.
	In article		CrossRef
[23]	Eriksson, L, Trygg, J, & Wold, S. (2008). CV-ANOVA for significance testing of PLS and OPLS (R) models. Journal of Chemometrics, 22 (11-12), 594-600.
	In article		CrossRef
[24]	Cloarec, O, Dumas, ME, Trygg, J, Craig, A, Barton, RH, Lindon, JC, Nicholson, JK, & Holmes, E. (2005). Evaluation of the orthogonal projection on latent structure model limitations caused by chemical shift variability and improved visualization of biomarker changes in 1H NMR spectroscopic metabonomic studies. Analytical Chemistry, 77 (2), 517-526.
	In article		CrossRef
[25]	Fan, TWM. (1996). Metabolite profiling by one- and two-dimensional NMR analysis of complex mixtures. Progress in Nuclear Magnetic Resonance Spectroscopy, 28, 161-219.
	In article		CrossRef
[26]	Fan, TWM, & Lane, AN. (1996). Structure-based profiling of metabolites and isotopomers by NMR. Progress in Nuclear Magnetic Resonance Spectroscopy, 52, 69-117.
	In article		CrossRef
[27]	Yoshie, Y, Suzuki, T, Shirai, T, & Hirano, T. (1994). Changes in the contents of dietary fibers, minerals, free amino acids, and fatty acids during processing of dried nori. Nippon Suisan Gakkaishi, 60, 117-123.
	In article		CrossRef
[28]	Chen, DW, Zhang, M, & Shretha, S. (2007). Compositional characteristics and nutritional quality of Chinese mitten crab (Eriocheir sinensis). Food Chemistry, 103 (4), 1343-1349.
	In article		CrossRef
[29]	Stapleton, PP, Charles, RP, Redmond, HP, & Bouchier-Hayes, DJ. (1997). Taurine and human nutrition, Clinical Nutrition, 16, 103-108.
	In article		CrossRef
[30]	Kurihara, K. (2009). Glutamate: from discovery as a food flavor to role as a basic taste (umami). The American Journal of Clinical Nutrition, 90 (3), 719S-722S.
	In article		CrossRef