IC4R009-Metabolomics-2013-25056584

Project Title

• Comparative metabolic profiling of pigmented rice (Oryza sativa L.) cultivars reveals primary metabolites are correlated with secondary metabolites

The Background of This Project

• Rice is an important food staple that feeds more than half of the global population. Generally, rice studies are aimed to improve crop breeding and acquire valuable traits. Metabolites present in rice grain have protective activities against human disease following dietary intake, and also have beneficial effects on the immune system. In this study, hydrophilic metabolic profiling (including phenolics) in pigmented rice using GC-time-of-flight (TOF)MS coupled with chemometrics was applied to determine the phenotypic variation and analyze relationships between their contents.

Plant Materials & Treatment

• Six cultivars of black rice (Heugjinjubyeo, BR-1; Heugseolbyeo, BR-2; Heugkwangbyeo, BR-3; Josengheugchalbyeo, BR-4; Heugnambyeo, BR-5; Heughyangbyeo, BR-6) and one of white rice (Hwasungbyeo, WR-1) were used. Previously, the researchers reported the flavonoid and carotenoid contents in these cultivars, which might provide new opportunities for rice breeders and eventually commercial rice growers to promote the production of rice with enhanced nutritional quality (Kim et al., 2010). These cultivars, selected because of their commercial importance in the Korean rice industry, were grown in the same paddy field at the National Institute of Crop Science, Rural Development Administration, Korea in 2010. They were manually hulled and ground to obtain a fine powder using a cyclone mixer mill (HMF-590; Hanil, Seoul, Korea) and a mortar and pestle. The milled rice powders were kept at -80 ℃ prior to extraction. All chemicals used in this study were of analytical grade. Methanol and chloroform, which were used as extraction solvents, were purchased from J.T. Baker (Phillipsburg, NJ, USA). Ribitol and N-methyl-N-trimethylsilyltrifluoroacetamide (MSTFA) were obtained from Sigma Chemical Co. (St. Louis, MO). Methoxyamine hydrochloride was purchased from Thermo Fisher Scientific (Pittsburgh, PA, USA).

Research Findings

• In total, 52 metabolites including 19 amino acids, 13 organic acids, nine sugars, seven phenolic acids, three sugar alcohols, and one amine were detected in black rice (Fig. 1).

Fig. 1. Selected ion chromatogram of metabolites extracted from black rice (cv. Josengheugchalbyeo; BR-3) as MO/TMS derivatives separated on a 30 m × 0.25 mm I.D. fused-silica capillary column coated with 0.25-mm CP-SIL 8 CB low bleed. Peak identification: 1, pyruvic acid; 2, lactic acid; 3, valine; 4, alanine; 5, oxalic acid; 6, glycolic acid; 30 , valine; 7, serine; 8, ethanolamine; 9, glycerol; 10, leucine; 11, isoleucine; 12, proline; 13, nicotinic acid; 14, glycine; 15, succinic acid; 16, glyceric acid; 17, fumaric acid; 70, serine; 18, threonine; 19, b-alanine; 20, malic acid; 21, salicylic acid; 22, aspartic acid; 23, methionine; 24, pyroglutamic acid; 25, 4-aminobutyric acid; 26, threonic acid; 27, arginine; 28, glutamic acid; 29, phenylalanine; 30, p hydroxybenzoic acid; 31, xylose; 32, asparagine; 33, vanillic acid; 34, glutamine; 35, shikimic acid; 36, citric acid; 37, quinic acid; 38, fructose; 380 , fructose; 39, galactose; 40, glucose; 41, syringic acid; 42, mannose; 43, mannitol; 44, ferulic acid; 45, p-coumaric acid; 46, inositol; 440 , ferulic acid; 47, tryptophan; 48, sinapic acid; 49, sucrose; 50, cellobiose; 51, trehalose; 52, raffinose; IS, internal standard (ribitol).

• The quantification data of 52 metabolites normalized based on IS signal intensity were subjected to PCA to identify differences in metabolite profiles among cultivars (Fig. 2). PCA revealed that the two highest-ranking principal components accounted for 56.8% of the total variance within the data set. The PCA results clearly showed the absence of significant variances among the same cultivar. The first principal component, accounting for 36.6% of the total variance, resolved the metabolite profiles of BR-4 and WR-1 cultivars.

Fig. 2. Scores (A) and loading (B) plots of principal components 1 and 2 of the PCA results obtained from metabolite data of seven rice cultivars.

• To examine detailed relationships between the concentrations of the 52 metabolites in rice, we performed Pearson’s correlation analyses and hierarchical clustering analysis on the accessions (Fig. 3). In the studies, there was a significant relationship between shikimic acid, vanillic acid (r ¼ 0.7672, P < 0.0001), and p-coumaric acid (r ¼ 0.8399, P < 0.0001). Likewise, the shikimic acid concentrations were positively correlated with phenylalanine (r ¼ 0.8595, P < 0.0001) in all rice accessions. The shikimate pathway results in aromatic amino acid metabolism and initiation of the phenylpropanoid pathway. Therefore, the metabolome obtained from rice grain could provide a snapshot of the primary and secondary metabolism interface. The 52 metabolites were subjected to HCA using the Pearson correlation analysis, which revealed a few major metabolite clusters (groups). Groups containing phenolic acid were marked by a dotted box. One group contained phenolic acids, such as ferulic acid, sinapic acid, vanillic acid, p-coumaric acid, shikimic acid, leucine, methionine, isoleucine, arginine, asparagine, and raffinose. The other groups, involving p-hydroxybenzoic acid and syringic acid, included aromatic amino acids and sugars. Thus, the HCA revealed that most of the compounds, which could be identified, clustered on the basis of their biochemical nature (e.g. shikimate and phenylpropanoid pathway).

Fig. 3. Correlation matrix of metabolites from seven rice cultivars. Each square indicates the Pearson’s correlation coefficient of a pair of compounds, and the value of the correlation coefficient is represented by the intensity of blue or red colors, as indicated on the color scale.

• All 45 primary metabolites of rice grain samples were divided into 14 training set samples and four test set samples (Fig. 4). The predictive model had a R2Y of 0.912 and RMSEE of 0.4994. The predictive model created in this study had a Q2 of 0.806 and RMSEP of 0.5267. The contribution of variables in the projection could be explained using variables important in the projection (VIP). Variables with VIP values greater than 1 are the most influential to the model (Jumtee et al., 2009). Primary metabolites that have a strong positive correlation with phenolic acid, as shown in the correlation matrix, were important to create a prediction model for rice grain quality (Fig. 5). Phenylalanine, a major amino acid donor for the synthesis of phenolic acid, was found to be most significant in creating a phenolics prediction model for rice grain.

Fig. 4. The PLS predictive model constructed from 18 rice samples as training sets (filled triangle) (A). A predicted result after four samples of test sets (open triangle) were projected to the model (B).

Fig. 5. The influence of variables used to create a total phenolics predictor for rice grain.

Labs working on this Project

• National Academy of Agricultural Science, Rural Development Administration, Suwon 441-707, Republic of Korea

Corresponding Author

• S.-H. Ha: sunhwa@korea.kr