基于液相色谱-质谱代谢组学方法辨别两种泰国籼稻冷榨米糠油

Asian rice，or Oryza sativa L.，is a genus of perennial grass from the Poaceae family. In Thailand，the cultivated area of Khao-Hom-Mali and Khao-Hom-Pathum fragrant rices is almost 51% of overall rice species. First，Khao-Hom-Mali，or jasmine rice，is photoperiod sensitive rice which is planted dominantly in the rain-fed zones in northeast Thailand. Second，Khao-Hom-Pathum is photoperiod insensitive rice which is grown only in central Thailand，where irrigation system is sufficient. Suitable time for rice planting in the rain-fed zones is May to July for northeast，and June to August for central Thailand. In irrigated area，appropriate time for planting is December to February for northeast，and November to April for central Thailand. The Khao-Hom-Mali seed is 7.4 mm in length with alkali spreading value of 6-7，containing amylose of 12%-18%，and 2-acetyl-1-pyrroline （2-AP）around 3.3 ppm，while Khao-Hom-Pathum seed is 7.6 mm in length，with alkali spreading value of 6-7，containing amylose of 15% -16%，and 2-AP around 0.7 ppm ［1］. Because the differences in 2-AP and amylose contents affected its fragrant and flavor，the price of Khao-Hom-Mali seed is 1.5 times more expensive than that of Khao-Hom-Pathum.

While rice seed is known to possess nutritional substances，its bran is also full of valuable nutrients ［2］. Rice bran is a by-product from rice milling which can be utilized to produce rice bran oil （RBO）. There has been global interest in RBO in recent times. It is rapidly growing as an important application in the nutritional industry. This is because RBO can provide benefits to health in its natural state，as it is full of valuable constituents.RBO has been reported to possess sterols，such as phytosterols，tocopherols，tocotrienols，fatty acids，antioxidant γ-oryzanol，and carotenoids［3］. Tocopherols，tocotrienols，and γ-oryzanol exhibit antioxidant and hypocholesterolemic effects［4，5］. In addition，RBO has considerable levels of fatty acids，such as linolenic acid，linoleic acid and oleic acid. All of them show the benefits to reduce the risk of cancer，cardiovascular disease，and inflammation［6，7］. While the price of rice seed varies according to their cultivars，the price of rice bran remains fixed. It is known that the plants in the same species，but different cultivars， naturally produce unequal quantities of compounds depending on genetics and the value of plant species also depends on their nutritional constituents. Thus，in this study，the researchers aimed to use LC-MS based on metabolomics to observe the chemical constituents and discriminate the RBOs from different rice brans，in order to，if possible，preserve the price of rice brans from different cultivars.

Many methods for RBO preparation have been established by both mechanical and chemical processes［8，9］. In Thailand，cold pressed RBO has been developed using the screw compression method. This method has been used in small factories，where local communities have produced commercial RBO. The benefits of cold pressed RBO are its low cost and simple method of preparation. Moreover，cold pressed RBO retains its natural properties and valuable constituents which are favorable for consumption. Many articles have reported the analytical methods of RBO，which have focused only on the determination of specific groups of the compounds ［10-12］. Metabolomics，however，is the comprehensive analysis of all metabolites in a biological system ［13］. A growing popularity of LC-MS based on metabolomics is due to its high throughput and capability to analyze a wide range of molecules. In this article，LC-MS coupling with multivariate data analysis was used as a tool to investigate the cold pressed RBO produced from two different cultivars of two major Thai fragrant rice species. This method can also be used in a quality control process of cold pressed RBO.

1 Experimental conditions

1.1 Materials

Ten samples with five replicates were collected from different sources during May 2013 to November 2013. Khao-Hom-Mali，KDML105，rice bran samples were collected from three sources in Lopburi Province，one source in Chainat Province，and one source in Yasothon Province. Khao-Hom-Pathum，PTT1，rice bran samples were collected from three sources in Chainat Province，one source in Lopburi Province and one source in Nakhon Pathom Province，Thailand. All samples were collected rapidly after milling process. The rice bran was stored in cool container during 2 h shipping before RBO extraction in our laboratory.Linoleic acid （C 18 ∶3），linolenic acid （C 18 ∶2），oleic acid （C 18 ∶1），stearic acid （C 18 ∶0），palmitic acid （C 16 ∶0），and α-tocopherol were purchased from Sigma Aldrich （USA）. Standard γoryzanol was purchased from Tokyo Chemical Industry （Japan）. HPLC grade acetonitrile and methanol were purchased from B＆J （Korea）.

1.2 Standard preparation

Individual stock solutions （1.0 mg／mL）of all standards were prepared in isopropanol and filtered through 0.45 μm membrane filter. Working standards，linoleic acid （10，5，1，0.5 and 0.1 μg／mL），oleic acid （10，5，1，0.5 and 0.1 μg／mL），palmitic acid （10，5，1，0.5 and 0.1 μg／mL），linolenic acid （5，1，0.5，0.1 and 0.05 μg／mL），stearic acid （5，1，0.5，0.1 and 0.05 μg／mL），α-tocopherol （22，11，2.2，1.1 and 0.11 μg／mL），and γ-oryzanol （17，8.5，4.25，2.125 and 1.062 5 μg／mL）were prepared by diluting the corresponding stock solutions with isopropanol for LC-MS analysis.

1.3 Sample preparation

Cold pressed RBO was prepared using the screw compression method. Firstly，the rice bran was strained using sieve mesh number 20. Secondly，the sieved rice bran was pressed with a small size screw press machine which possesses a 2 horse power single phase motor. Thirdly，the crude RBO was filtered through 10 and 2.5 μm Whatman filters，separately in a sterile condition.Khao-Hom-Mali rice bran samples were labeled as RBO01-RBO05， while Khao-Hom-Pathum rice bran samples were labeled as RBO06 -RBO10.The samples were analyzed immediately by LCMS. Cold pressed RBO was accurately weighed to 10 mg. The volume was adjusted to 10 mL in a volumetric flask with isopropanol. The sample was mixed by vortex for 20 s and filtered through a 0.45 μm membrane filter. The experiments were conducted in five replicates.

1.4 LC-MS analysis

The extract was analyzed using a Dionex UltimateTM3000 HPLC coupling with Bruker Amazon SL mass spectrometer. A Poroshell 120 EC-C18 column （150 mm×2.1 mm，2.7 μm）was used for separation. Analyses were performed in linear mode using 0.2% formic acid in acetonitrile and methanol （60 ∶40，v／v，0-25 min）. The column was maintained at 30 ℃with a flow rate of 0.12 mL／min and the injection volume was 5 μL. The mass spectrometer was equipped with an ESI ion source and a quadrupole-ion trap. The system was tuned for optimum sensitivity and resolution using a Bruker ESI tuning mix in both positive and negative ESI modes. LC-MS evaluation was performed using full scan in negative mode recorded on a mass range of m／z 100-1 500 in centroid mode. Capillary voltage was set at 4 500 V and drying gas temperature was 200 ℃with a flow rate of 7.0 L／min. Nebulizer pressure was set at 200 kPa （2 bar）. All compounds were recognized by comparing the parent and fragment ions with the reference standards using the multiple reaction monitoring （MRM ） mode. The constituent amount in cold pressed RBO was quantified basically from the calibration curve. For cycloartenyl ferulate， 24-methylenecycloartanyl ferulate，campesteryl ferulate，and β-sitosteryl ferulate，the relative amount was calculated based on the proportion of it in γ-oryzanol. Data were processed by Compass 1.3 SR2. System suitability was performed using the working standard solution of linoleic acid. Theoretical plate number（N）and precision （RSD）were measured.

1.5 Method validation

The analytical procedure was modified and validated in accordance with the International Conference on Harmonization（ICH）guidelines Q2 （R1）［14］. Before the validation procedure，the highest concentration calibrator was injected into the LC-MS system until a response was obtained. Isopropanol was injected in triplicates to determine the carryover effect.

1.5.1 Linearity

Calibration curves of all reference standards were constructed by plotting peak areas against five concentrations of each reference standard.The calibration curves should show the coefficients of correlation （R2）≥0.999 5.

1.5.2 Limit of quantification

The limit of quantification （LOQ）was determined by means of a serial dilution based on signal-to-noise ratio of 10 ∶1. The LOQ concentration was evaluated by precision of six replicate injections.

1.5.3 Precision

Repeatability （intra-day）and reproducibility（inter-day）precision was optimized in all analyses. A sample solution was used to achieve the intra-day experiment. The repeatability was calculated as the relative standard deviation （RSD）of the results from six injections on the same day.The reproducibility was optimized by comparing the RSDs of the analyzed samples on three consecutive days.

1.5.4 Accuracy

The accuracy was demonstrated by the recovery study，which was carried out by fortifying samples with three concentrations of known quantities of the standard solutions. Prior to fortification，the background levels of linoleic acid，linolenic acid，oleic acid，stearic acid，α-tocopherol，cycloartenyl ferulate，24-methylenecycloartanyl ferulate，campesteryl ferulate，and βsitosteryl ferulatein samples were determined so as to calculate the actual recoveries. The amount of each standard was determined in triplicate and the percentage recoveries were calculated.

1.5.5 Specificity

The specificity was investigated by monitoring the mass of each reference standard using MRM mode. The cut-off selection was set to 27.0% of the precursor mass. With SmartFrag，the amplitude was ramped by starting at 60% of set amplitude，and ended at 180%. The precursor ion was isolated with activation time of 40 ms and m／z width＞4.00.

1.5.6 Robustness

The robustness was evaluated to determine the capacity of a developed method to remain unaffected by small，deliberate variations. By introducing small variation in column temperature，the mean value and RSD were calculated. The robustness was then optimized.

1.5.7 System suitability

System suitability was performed using the working standard solution of linoleic acid. N and RSD were measured.

1.6 Statistical analysis

The analysis of variance was achieved using IBM SPSS statistics 22. Values are expressed as mean±SD. Data were analyzed by Duncan’s multiple range test （DMRT）. The level of statistical significance was taken at p＜0.05.

For multivariate data analysis，MZmine 2.10 was used to preprocess LC-MS data before statistical analysis ［15］. Considering all chromatograms，the peak detection was set to centroid with the noise level of 1×105and m／z tolerance of 0.5 or 20 ppm. The preference for chromatogram deconvolution was set as local minimum search in which values were 1% threshold，0.4 min retention time （RT）range，5% minimum relative height，5×105minimum absolute height，5 minimum ratio of peak top／edge，and 0.3-2 min peak duration range. The representative isotope in a peak list was chosen as the most intense isotope with the maximum charge of 2. In order to process the peak list alignment from different samples，Join aligner was performed with 5% relative RT tolerance. Gap filling and filtering were also set as m／z tolerance of 0.5 （m／z），and minimum peaks in a row of 50，respectively. The data were normalized to average intensity. Integrated intensities of each m／z-retention time pair were achieved for each one of the samples. SIMCA 13 software was used to create the principle component analysis （PCA）of RBO data. PCA was performed to grouping and outlier detection among samples. The variable was traded by Pareto scaling method，and the Hotelling’s T2was used to detect any outliers.

2 Results and discussion

2.1 Cold pressed RBO yield and constituents

Cold pressed RBOs of Khao-Hom-Mali and Khao-Hom-Pathum were extracted by the screw press machine. After filtration，Khao-Hom-Mali RBO was dark brown-yellow color，while Khao-Hom-Pathum RBO was dark brown-greenish color. Both RBOs were clear liquids，and each with a unique odor. Generally the average yield of RBO extraction by screw press method is 4%-8%，depending on rice bran variety，stage of bran，and user skill ［16］. In this study the average yield was 4%-6%. An appropriate LC method was developed. Different ratios of acetonitrile and methanol were tried to determine the appropriate mobile phase. The good separations were obtained in an isocratic mode using acetonitrile and methanol （60 ∶40，v／v）in 25 min. The chromatograms are shown in Fig.1. Moreover，the advantage of LC-MS method is the capability to quantify several compounds at the same RT. In this study，oleic acid and palmitic acid can be accurately analyzed using extracted ion quantification at different molecular masses. Another publication which using HPLC coupling with evaporative light-scattering detector （ELSD）presented a good chromatographic condition for fatty acids determination in O. sativa ［17］，but it is inappropriate for this study since the authors aim to evaluate the quality of RBO based on several types of compounds using simple and short-time analysis in one single run.

2.2 LC-MS method validation

Fig.1 LC-MS chromatograms of cold pressed RBO

For the method validation，the carryover effect of this method was assessed in triplicate runs of blank after run of the highest calibrator. The noise signals in each blank chromatogram were negligible when compared with the LOQ. The calibration curves of all standards were created using five concentrations of each standard and the linearity was determined by means of linear regression analysis. All calibration curves were linear over their concentration ranges with R2≥0.999 5. The lowest reliable amount for quantification of each metabolite was determined. First，for the fatty acids，the LOQ values of linoleic acid，oleic acid，and palmitic acid were 1.0 ng／mL，while the others were 1.5 ng／mL. Second，the LOQ of α-tocopherol was 55.0 ng／mL.Finally，for the γ-oryzanol，the LOQ of cycloartenyl ferulate and 24-methylenecycloartanyl ferulate were 4.25 ng／mL，and the others were 21.0 ng／mL （Table 1）. The precision at LOQ levels of each standard was calculated as RSD in which values were 0.22%，0.37%，0.30%，0.33%，0.21%，0.25%，0.49%，0.43%，0.22% and 0.28% for linolenic acid，linoleic acid，oleic acid，palmitic acid，stearic acid，α-tocopherol，cycloartenyl ferulate， 24-methylenecycloartanyl ferulate，campesteryl ferulate and β-sitosteryl ferulate，respectively.

The precision of the method was assessed by intra-day and inter-day analysis. The method was shown to be reproducible and reliable with both intra-day and inter-day precision values，the RSDs were found to be 0.15% -0.36% and 0.79% -1.94%，respectively. The spiking procedure was performed for accuracy assessment by adding the standards to RBO. Mean recoveries in the range of 100% -102% were observed for all standards（Table 1）.

Table 1 Validation data of the developed LC-MS method

For compound identification，the mass transition was observed both in the reference and sample solutions for linolenic acid（［M-H］-277.3），linoleic acid（［M-H］-279.2），oleic acid（［MH］-281.2），stearic acid（［M-H］-283.5），palmitic acid（［M-H］-255.2），α-tocopherol（［MH］-429.8），cycloartenyl ferulate （［M - H］-601.5），24-methylenecycloartanyl ferulate （［MH］-615.6），campesteryl ferulate （［M - H］-575.5），and β-sitosteryl ferulate （［M - H］-589.6）parent ions to guarantee selectivity and specificity. Each transition had its specific reaction amplitude. Only six compounds showed fragmentation patterns. First，linoleic acid transition was ［M-H］-286.1 to 255.2 with the reaction amplitude of 1.0. Second，α-tocopherol transition was ［M-H］-429.8 to 162.9 with the reaction amplitude of 1.0. Third，cycloartenyl ferulate transition was ［M-H］-601.5 to 586.5 with the reaction amplitude of 0.6. Fourth，24-methylenecycloartanyl ferulate transition was ［M - H］-615.6 to 600.5 with the reaction amplitude of 0.6. Fifth，campesteryl ferulate transition was［M-H］-575.5 to 560.5 with the reaction amplitude of 0.7. Sixth，β-sitosteryl ferulate transition was［MH］-589.6 to 574.5 with the reaction amplitude of 0.6. All compound structures are shown in Fig.2.

Fig.2 Chemical structures of some compounds found in cold pressed RBO

For robustness，the results showed good reliability of analysis with respect to deliberate variation in column temperature （Table 2）. N and RSD of this method were also measured. It was found that N was 14 410 and RSD was 0.23. The values obtained for this method were within the criteria.

2.3 Discrimination between RBOs from two rice cultivars

The proposed method was applied to the determination of fatty acids，α-tocopherol，and γoryzanol in cold pressed RBOs from two rice cultivars （Table 3）. All reference compounds were determined in 10 lots of RBO which were prepared with the same method. In this study，the PCA was used to discriminate the samples.Through PCA of 10 data sets with Pareto scaling，a PCA model with three components was established. All RBO samples were well correlated and summarized by six variables，the scores，explaining 83.2% of the variation. The first component described 54.8% of the variation. Good prediction properties of the model were achieved with a Q2of 78.4%. The score plot of the first two principle components is shown in Fig.3. It was suggested that the RBO produced from different cultivars can be significantly clustered in the direction of the first predictive principle component （X axis）.

Even both rice cultivars in this study are full of nutritional substances such as vitamin B1，B2，niacin，carbohydrates，protein，and minerals［18］，but the price of fragrant rice seed is mainly in positive correlation to the amount of 2-AP because more 2-AP means more fragrant. Khao-Hom-Mali contained 2-AP 4.5 times more thanthat in Khao-Hom-Pathum，then the price of Khao-Hom-Mali seed is more expensive than that of Khao-Hom-Pathum ［19］. On the other hand，there is no regulation to control rice bran price；even the utilities are different. Rice bran can be used directly as animal foods，or be processed as cooking oils，food additives，cosmetic constituents and dietary supplements. Rice bran which used as material for RBO production for human consumption should be in excellent quality and more expensive compared to that used as the animal foods. Good quality RBO should be judged from the types and amounts of active constituents that promote the positive effects to human health.

Table 2 Robustness data for the developed LC-MS method

The purpose of the present study was the metabolites observation which had an obvious impact on clustering tendency of RBOs produced from Khao-Hom-Mali and Khao-Hom-Pathum.From the loading plot in Fig.3，it was found that Khao-Hom-Mali RBO possessed distinctly high levels of linoleic acid，oleic acid，and palmitic acid compared to Khao-Hom-Pathum RBO. There is strong evidence that increasing palmitic acid consumption was highly related to cardiovascular disease promotion ［20］. Moreover，Khao-Hom-Pathum RBO has a more appropriate ratio between ω-6／ω-3 fatty acids，which have a positive effect on cardiovascular system ［21］. From the loading plot，concerning to individual γ-oryzanol，it was found that Khao-Hom-Pathum RBO possessed the molecules of cycloartenyl ferulate，campesteryl ferulate，and β-sitosteryl ferulate.The results were correlated to the contents of metabolites in Table 3，in which Khao-Hom-Pathum RBO contained higher amount of three ferulic acid esters than Khao-Hom-Mali RBO. The γ-oryzanol has been shown to possess antioxidant，anti-inflammatory，anti-tumor，and hypocholesterolemic activities ［22，23］. Therefore，RBO produced from rice bran containing high amount of γ-oryzanol，fewer palmitic acid，and low linolenic acid／linoleic acid ratio should be classified as excellent quality RBO and more suitable for human consumption. Considering within Khao-Hom-Maligroup，it was found that all the ratios of unsaturated fatty acids and γ-oryzanol in RBO from central Thailand were significantly higher than that in RBO from Yasothon Province，which located in northeast Thailand. For Khao-Hom-Pathum，even all samples were collected from central Thailand；there is little variation in γ-oryzanol amount of RBO from different provinces （Table 3）. This is maybe due to the difference in cultivation environment.

Table 3 Contents of fatty acids，α-tocopherol，and γ-oryzanol in ten cold pressed RBOs from Khao-Hom-mali and Khao-Hom-Pathum

Fig.3 （a）Score plot showing ability of PCA to discriminate RBOs from two different cultivars and （b）loading plot showing significant molecules which have an influence in clustering

3 Conclusions

Fatty acids， α-tocopherol， and γ-oryzanol which found in cold pressed RBO possess high nutritional value and various pharmacological activities. Therefore，a new method which can be used to detect these compounds must be developed. It was found that the presented LC-MS method is suitable for the quality control process of cold pressed RBO manufacturing. Moreover，coupling LC-MS with multivariate data analysis revealed that Khao-Hom-Pathum RBO was more appropriate than Khao-Hom-Mali RBO for dietary supplement production since it contained more total γ-oryzanol contents，less palmitic acid，and low linolenic acid／linoleic acid ratio. This finding can support data to increase the value of Khao-Hom-Pathum rice bran. If the government can preserve the price of rice brans from different cultivars，for example，Khao-Hom-Mali and Khao-Hom-Pathum，it will be more profitable to Thai farmers since they can trade Khao-Hom-Pathum paddy in higher price.

AcknowledgementsWe thank all staffs from Sino-Thai Traditional Medicine Research Center who aid on cold pressed rice bran oil preparation.The author thanks K I Tull for assistance with the English in this manuscript.

［1］ Cheaupun K，Wongpiyachon S，Sukwiwat W，et al. Proceeding of Rice and Temperate Cereal Crops Annual Conference，Rayong，2014：397

［2］ Godber J S，Wells J H. Louisiana Agriculture，1994，37（2）：13

［3］ Ghosh M. J Amer Oil Chem Soc，2007，84：315

［4］ Chung Y A，Lee J K. Korean Soc Food Sci Nutr，2003，32（6）：948

［5］ Berger A，Rein D，Schafer A，et al. Eur J Nutr，2005，44：163

［6］ Simopoulos A P. Biomed Pharmacol，2002，56（8）：365

［7］ Sales-Campos H，Souza P R，Peqhini B C，et al. Mini Rev Med Chem，2013，13（2）：201

［8］ Xu Z，Godber J S. J Amer Oil Chem Soc，2000，77（5）：547

［9］ Thanonkaew A，Wongyai S，McClements D J，et al. Food Sci Technol，2012，48（2）：231

［10］ Chen M H，Bergman C J. J Food Comp Anal，2005，18：139

［11］ Kaewkool P. Asian J Food Agric-Ind，2011，4（1）：16

［12］ Lu W，Niu Y，Yang H，et al. Food Chem，2014，148：329

［13］ Fiehn O. Plant Mol Biol，2002，48：155

［14］ ICH-Q2 （R1）. International Conference on Harmonization.Geneva，Switzerland，2005

［15］ Pluskal T，Castillo S，Villar-Briones A，et al. BMC Bioinformatics，2010，11：395

［16］ Sayasoonthorn S，Kaewrueng S，Patharasathapornkul P.Rice Sci，2012，19（1）：75

［17］ Bravi E，Perretti G，Montanari L. J Chromatogr A，2006，1134：210

［18］ Kennedy G，Burlingame B. Food Chem，2003，80（4）：589

［19］ Thai Rice Mills Association. Thai Rice Price. ［2015-04-03］. http：／／www.thairicemillers.com

［20］ Hu F B，Stampfer M J，Manson J E，et al. N Engl J Med，1997，337（21）：1491

［21］ Simopoulos A P. Exp Biol Med，2008，233（6）：674

［22］ Patel M，Naik S N. J Sci Ind Res，2004，63（7）：569

［23］ Ausman L M，Rong N，Nicolosi R J. J Nutr Bioch，2005，16（9）：521