Fast determination of phenolic compounds in rice grains by UPLC-PDA

Subscriber access provided by Washington University | Libraries

New Analytical Methods

Fast determination of phenolic compounds in rice grains by UPLC-PDA: method development and validation Widiastuti Setyaningsih, Irfan E Saputro, Ceferino A. Carrera, Miguel Palma, and Carmelo Garcia-Barroso J. Agric. Food Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jafc.8b05430 • Publication Date (Web): 14 Feb 2019 Downloaded from http://pubs.acs.org on February 14, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 31

Journal of Agricultural and Food Chemistry

Fast determination of phenolic compounds in rice grains by UPLC-PDA: method development and validation Setyaningsih, W.,† Saputro, I. E.,‡ Carrera, C. A.,‡ Palma, M.,*,‡ García-Barroso, C.‡ †Department

of Food and Agricultural Product Technology, Faculty of Agricultural Technology,

Universitas Gadjah Mada, Jalan Flora, Bulaksumur, 55281, Yogyakarta, Indonesia. ‡Department

of Analytical Chemistry, Faculty of Sciences, IVAGRO, University of Cadiz, Campus

de Excelencia Internacional Agroalimentario (CeiA3), Campus del Rio San Pedro, 11510 Puerto Real, Cádiz, Spain. *Corresponding author, (Tel: +34956016360; Fax:+34956016590: E-mail:[email protected])

1 ACS Paragon Plus Environment


Page 2 of 31

ABSTRACT There are several phenolic compounds in rice grains providing benefits for human health. The concentration of phenolic compounds in rice are strongly affected by the polishing steps during rice production. A new sensitive ultra-performance liquid chromatography–ultraviolet-visible spectroscopy method with photodiode array detection protocol has been developed and validated for the quantitation of phenolic compounds in rice grains. Several working variables and two different columns were evaluated. Finally, a less than 3 min analysis time was developed to achieve enough resolution for the simultaneous determination of the 20 most common phenolic compounds in rice. The analytical properties for the separation method produced an adequate sensitivity for all phenolic compounds in the regular range for phenolics in rice: 0.5–100 mg L-1 (R2>0.997) with high precisions for both repeatability and intermediate precisions (CV less than 0.4% and 2.5% for retention time and the area of the peaks, respectively).

KEYWORDS: phenolic compounds, rice grains, Box–Behnken design, UPLC-PDA, desirability function


Page 3 of 31


■ INTRODUCTION Rice (Oryza sativa) is a prominent crop used as a main staple food in many countries around the world.1 Rice in human diet provides mainly calories. However, rice also contains exclusive compounds that are beneficial to human health and these include phenolics.2 Henceforth, numerous studies into phenolic in rice arise extensively thus disclose the composition of phenolics in the grain.3 Phenolic acids and their aldehydes are the most common phenolic compounds in rice grains .4 The phenolic acids are derived from both the hydroxybenzoic acid (C6−C1) and the hydroxycinnamic acid (C6−C3).5 A number of studies in the literature have reported that more than 30 phenolic compounds have been identified in various type of rice cultivars. Table 1 compiles the phenolics identified in foregoing studies in rice grains. According to the data listed in Table 1, the most common phenolics described in rice grains are p-coumaric and ferulic acids followed by vanillic, p-hydroxybenzoic, protocatechuic and caffeic acids. Furthermore, syringic, sinapic and chlorogenic acids were also reported by a minimum of five different studies. Besides the two phenolic groups discussed heretofore, phenolic aldehyde such as protocatechuic aldehyde, p-hydroxybenzaldehyde and vanillin were also determined in rice grain samples. On the basis of these revision, the most common phenolic compounds were selected for this study, specifically 20 compounds including both phenolic acids and phenolic aldehydes. Given their role as potent antioxidants, phenolics yet persist to be fascinating to investigate. Certainly, it is essential to come up further research with a reliable method for their determination, which has been developed in the latest years. Chromatographic methods based on liquid chromatography are the most common used separation methods. Nevertheless, methods based on regular HPLC systems are time-consuming i.e. HPLC-PDA (35 min),6 HPLC–DAD–Q-TOF– MS/MS (40 min),7 HPLC–MS/MS (63 min),8 HPLC-UV (>80 min),9,10 LC-DAD-ESI-MS (120 min),11 whilst in other cases with incomplete separation for rice matrices containing a number of phenolics.12,13 3 ACS Paragon Plus Environment


Page 4 of 31

UPLC has overcome the negative aspect of regular chromatographic columns in highperformance liquid chromatography (HPLC) enabling columns packed with smaller particles (sub-2 μm), then providing an improved chromatographic performance. New type of solid phases has appeared for UPLC based applications, including solid-core technology based columns which have a solid-core of silica in the centre of the solid phase; then the chromatographic separation takes place in the surrounding layer of the porous silica. Although the solid-core particles have a lower surface area than the fully porous particles, the estimated phase ratios for these two columns are similar, therefore resulting in equivalent retention factors. Obviously, applying smaller particles size of the column, as led to a shorter analysis time, UPLC technology dramatically reduces the total volume of mobile phases compared with the conventional liquid chromatography, moreover, higher chromatographic resolutions are produced.14 The UPLC also facilitates a high signal-to-noise ratio thus providing low limits of detection and quantification. Additionally, the injection volume can be considerably reduced with similar sensitivity values.15 The simultaneous study of chromatographic resolution and the total analysis time is usually the most important target during the development of new methods in liquid chromatography.16 It is related also to other working variables such as flow rates and specific gradient times.17 Additionally, composition of the mobile phases, as well as type of the solid phase in the column, were also critical variables in optimizing separations in UPLC.18 In this study, several variables were evaluated for their effects on the chromatographic separation of phenolics from rice grains extracts. The studied variables were: the flow rate and the composition of mobile phases during a gradient elution. The optimization was developed using a Box–Behnken design (BBD).19,20 Because of the optimization procedure involves several responses, it does not allow for the optimization of each one in a separate way, due to a number of solutions equal to the variables under study would be gathered.21 Therefore the desirability function is a good option, as it can be used to get a compromise solution for multiresponse optimization (MRO).22 4 ACS Paragon Plus Environment

Page 5 of 31


Therefore, here the optimization of the simultaneous determination by UPLC of 20 phenolics usually found in extracts from rice grain samples using BBD in conjunction with MRO next to desirability function is presented. Afterwards, the new method has been validated, including its application to both whole and polished rice grain extracts. ■ MATERIALS AND METHODS Chemicals and Reagents. Solvents: ethyl acetate for the extractions, acetic acid and HPLC-grade methanol for the chromatographic analyses were obtained from Merck KGaA (Darmstadt, Germany). Water was obtained from a Milli-Q purification system (Billerica, MA, USA). Standards: Ferulic acid (FER), p-hydroxybenzaldehyde (p-HB), p-hydroxybenzoic acid (p-HBA), protocatechuic aldehyde (PRA), protocatechuic acid (PRO), sinapic acid (SIN), vanillic acid (VAA) and vanillin (VAN), were obtained from Fluka (Buchs, Switzerland). Ellagic acid (ELL) was provided by Sarsynthese (Merignac, France). Caffeic acid (CAF), chlorogenic acid (CHL), pCoumaric acid (p-COU), furfural (FUR), guaiacol (GUA), isovanillic acid (IVA), 5-hydroxymethyl2-furaldehyde (HMF), 5-methylfurfural (MF), quercetin (QUE) and syringic acid (SYR) were obtained from Sigma–Aldrich (St. Louis, MO, USA). Iso-ferulic acid (IFA) was obtained from Extrasynthese (Genay, France). Aqueous methanol 50:50 (v/v) was used as solvent for the standard solutions. After preparation they were stored in a freezer at −32 °C. Samples preparation. Commercial rice was used in this study. Samples were obtained in Spain from regular markets. Aliquots of 20 g were placed in plastic cylinders. Milling process (10 minutes) was done in an Ultraturrax homogenizer (IKA® T25 Digital, Germany). During the milling process, the system was stopped every minute, to prevent high temperatures in the sample. The resulting powder was then homogenized and the sample was kept in a closed container in a refrigerator before used for an analysis. A microwave oven (Milestone-Ethos 1600, Sorisole, Italy) was used for the extractions from rice grains as described by Setyaningsih et al.23 The microwave-assisted 5 ACS Paragon Plus Environment


Page 6 of 31

extraction conditions were: 20 minutes as extraction time, 195 °C as the extraction temperature, 1000 W was the extraction power and 1:10 was the sample-to-solvent ratio. The extract was instantly cooled using an ice bath and subsequently filtered by a 0.22 μm nylon filter before the analysis in the UPLC system. Identification and quantification of phenolic compounds. An ACQUITY UPLC® H-Class system was used for the UPLC analysis. Empower™ 3 Chromatography Data (Waters Corporation, Milford, MA, USA) was the software managing the UPLC system. An ACQUITY UPLC® Photodiode Array (PDA) was used as the detector. For compounds identification, the PDA was operated for the 3D scan mode, collecting 40 points per second from 200 to 400 nm. In the case of compounds quantification, a 2D scan of PDA with collection data rate at 80 points per second at a fixed wavelength providing maximum absorbance of the corresponding compounds (260, 280, and 320 nm) whilst 280 nm was chosen for peak integrations during the optimization process. The identification of chromatographic peaks in rice extracts was performed by comparing chromatographic and spectroscopic properties with those of standards. Spiking procedures were also used. A C18 (Reverse Phase) UPLC Column at a temperature of 47 °C was used for the separations of phenolic compounds in 3.0 μL injected samples. Two different columns, with different solid phases were compared: (1) a Particle-Based (PB) column, Acquity UPLC® BEH (Ethylene Bridged Hybrid, Waters Corporation, Ireland) and a Solid-Core Based (SCB) column, CORTECS UPLC® (silica-based solid-core, Waters Corporation, Ireland). The PB column has 100 mm length; an ID of 2.1 mm and a particle size of 1.7 μm. The SCB column has 100 mm length and I.D. of 2.1 mm and a particle size of 1.6 μm. Two solvents consisting of phase A (2% acetic acid in water) and phase B (2% acetic acid in acetonitrile) were used as the mobile phases. After the analysis, the columns were washed using phase B for 3 min. The equilibration time for the next injection was 3 min. 6 ACS Paragon Plus Environment

Page 7 of 31


Box–Behnken Design and data analysis. Three experimental variables were studied: solvent composition at the starting conditions (% B0, x1); solvent composition at the end of elution (% B1, x2); and flow rate (mL min-1, x3) on the separation of 20 phenolics by two different UPLC columns. The effect of studied experimental variables was evaluated using a Box Behnken design. Ranges studied for the working variables were different, so each variable was initially normalized to the same range (–1 to +1) to guarantee more even response. The ranges of the studied variables and their levels are listed in Table 2. Accordingly, the design constructed of a total of 15 experiments including three center points. Resulting values were used to calculate the effects of both the individual variables and their interactions on the chromatographic responses. The resolutions of the chromatographic peaks in addition to the total time needed for the separation were used as the responses in the experimental design. Peak resolutions (Rs) were calculated as the values obtained for the separation of two consecutive peaks using their average peak width at the base. The value for the retention time of the last eluted peak on the chromatogram was used as the response for analysis run time. Using STATGRAPHICS Centurion XVI (Statpoint Technologies, Inc., USA), a MRO (multiresponse optimization) was conducted with desirability function to obtain the right UPLC conditions for phenolic compounds separation when more than one response is taken into account. A mathematical model was built for each response fitting a second order polynomial function then MRO can be estimated. Method Validation. The new method for the separation of phenolic compounds was validated following the recommendations by ISO 17025 and the ICH Guideline Q2 (R1).24,25 The detection and quantification limits, the range of linearity and the precision, of the method were established. A series of solutions prepared using the standards of the phenolic compounds to reach concentrations from 0.5 to 100 mg L–1 were prepared. Once the regression analysis was applied, the linearity was measured within the studied range to point out that the test results obtained by the method are proportional to the concentration of phenolic compounds. Using the regression results, the limit of 7 ACS Paragon Plus Environment


Page 8 of 31

detection (LOD) and limit of quantification (LOQ) were calculated, specifically, the standard deviation estimated for the response and the slope were used. Repeatability (intra-day) and intermediate precision (extra-day) were used to calculate the precision of the new method. Nine independent analyses of the same samples on the same day were used to establish the value for the repeatability. Three independent analyses on three consecutive days were used to determine the intermediate precision for the new method. Coefficient of variation (CV) of retention time and peak height were used to express the precision. According to the AOAC manual for the Peer-Verified Methods program, the acceptable CV limit is ± 10%.26 ■ RESULTS AND DISCUSSION Data generating for the responses. A standard solution of 20 phenolic compounds was analyzed in the UPLC system using the working conditions in the Box Behnken design. Both columns compared in this study have similar active sites, therefore the order of elution was the same for the standards compounds, wherein polar compounds were eluted first. Following is the order of 20 phenolics sorted by the retention time: 1. HMF; 2. PRO; 3. FUR; 4. PRA; 5. p-OHB; 6. CAF; 7. CHL; 8. VAA; 9. p-HBA; 10. IVA; 11. SYR; 12. MF; 13. VAN; 14. p-COU; 15. ELL; 16. FER; 17. SIN; 18. QUE; 19. IFA; 20. GUA. Hereafter, the phenolic compounds are cited corresponding to the order number in the peak resolution results. Separation factor (S) or resolution (Rs) can be used for the determination of the separation of two adjacent peaks. If different columns are being compared, the use of Rs is preferred versus the S,27 as in this case. Table 3 shows the results for the Rs of phenolic peaks using the working conditions in the BBD on the two different columns (PB and SCB). A complete separation is found for Rs of about 1.5. If the Rs value is 1.0, it is yet acceptable because it means 98% separation. Values below 1.0 means not enough separations28. Henceforth, Rs with values less than 1.0 were included in the optimization process in order to reach higher values. The 14 Rs lower than 1.0 were Rs1-2, Rs4-5, Rs5-6, Rs6-7, Rs7-8, Rs9-10, Rs10-11, Rs11-12, Rs12-13, Rs14-15, Rs158 ACS Paragon Plus Environment

Page 9 of 31 16,


and Rs17-18 for the PB column, while 12 Rs were included for SCB column: Rs1-2, Rs4-5, Rs5-6, Rs9-10,

Rs10-11, Rs11-12, Rs12-13, Rs13-14, Rs14-15, Rs16-17, Rs17-18, and Rs19-20. The total time to elute the last peak was also included into the optimization as the response, since it should be as low as possible, for that reason the flow rate was the working variable to be fitted. The data for run time respects to the retention time of the last eluted peak (GUA). The analysis run time ranging from 1.93 to 3.31 min by BBD using PB column, and from 1.67 to 2.85 min using the SCB column. Both values mean very fast separation methods, therefore the response of total time to elute the last peak was defined to be less important than the lowest Rs. The total number of responses for the optimizations were 15 and 13 responses for PB and SCB column respectively. All responses were considered to be equally important except the lowest Rs. The impact coefficient indicated the importance of the responses for statistical analysis that was given to the responses in the MRO. Values of the impact coefficients, by default, were set to three (STATGRAPHICS Centurion XVI). Nevertheless, depending on some cases, the value can be modified from 1 to 5. Because of different responses and impacts were used, the optimization was developed separately for each UPLC column. Optimization of separation methods. First a model for individual responses per column was generated using the response surface methodology (RSM). Based on the values obtained for each optimized response, the desirability function d(y) was constructed. The MRO approach assumes the response values equal to (y) can be modelled through the d(y), where the desirability ranges from 0 to 1. PB column. A simultaneous optimization for the three working variables affecting the 15 responses was used. Two-factor interactions with seven coefficients were included in the optimization model. A model has been built for each response fitting a second order polynomial function. The resulting equation for the fitted model of Rs1-2 is as follows:



Page 10 of 31

𝑦 = 1.041 + 0.219𝑥1 ―0.030𝑥2 +0.137𝑥3 ―0.090𝑥1𝑥1 +0.087𝑥1𝑥2 ―0.059𝑥1𝑥3 ―0.095𝑥2𝑥2 (1) +0.061𝑥2𝑥3 +0.112𝑥3𝑥3 in this equation, y is the resolution value and xi are the chromatographic variables (x1, %B start; x2, %B end; and x3, flow rate). Analysis of variance (ANOVA) and the significance of the model, were used to test the adequacy and the significance of the model (p 0.05). The R2 statistics provide information about how the model is able to explain the variability of the response studied. Also of interest is the R2 statistic, the values were ranged from 52.24% (Rs13-14) to 98.66% (total analysis time). The resulting model provides the effects by the variables on the responses inside the ranges studied in the BBD. Later, simultaneous optimization of the 15 responses was conducted by means of MRO. It must be noted that the impact used for Rs lower than 1.0 i.e. Rs1-2 and Rs17-18 were set to the highest level (5 impact) while the impact for other 13 responses were set to the intermediate level (3 impact). The optimization by MRO was to achieve the Rs > 1.0 and minimize the total analysis run time < 3 min. To reach the optimum values, the d(y) was plotted as a 3-D mesh plot, as it can be seen in Figure 1. The MRO advises the most convenient values for the studied variables: %B initial (x1, ―0.1724), %B end (x2, ― 0.9935) and flow rate (x3, +0.3945) therefore the values for the working variables have been defined: %B initial (x1, 4.1 %), %B end (x2, 50.2%) and flow rate (x3, 0.64 mL min-1). Using these values, the desirability index for the response variables was 92.56%. Desirability values lower than 1.0 were observed for Rs1-2, Rs4-5, Rs7-8, Rs10-11, and Rs17-18 ranging from 0.63 (Rs4-5) to 0.96 (Rs1-2). However, no further optimization of separation was performed without checking the Rs on different detection wavelengths. By dint of their corresponding maximum wavelengths, the compounds could have an adequate separation. The specific wavelengths producing the maximum absorptions of studied phenolics in the UVVis ranges were ranging from 254 nm (p-hydroxybenzoic acid) to 356 nm (quercetin). For that 10 ACS Paragon Plus Environment

Page 11 of 31


reason, to check the final separation of phenolics, the resulting chromatograms from the standard solution of the 20 phenolic compounds are presented in three-selected wavelengths i.e. 260, 280 and 320nm (Figure 2). By dint of applying different UV wavelengths, some adjacent compounds with severe Rs such as Rs1-2, Rs4-5, Rs7-8, and Rs10-11 could be separated or quantified in such way by applying the spectroscopic adjustment. However, adjacent compounds of p-COU ― VAN (Rs13-14) and SIN ― QUE (Rs17-18) were still not completely separated using the optimized method. SCB column. For this column, a very similar approach that has been applied for the method development and optimization for the PB column was used. The second order polynomial functions were calculated and they were established to build the RSM from the results in the BBD of the studied variables, in this case with 13 responses. The fitted model is presented by the next equation for Rs1-2 : 𝑦 = 0.700 ― 0.115𝑥1 ―0.090𝑥2 +0.194𝑥3 +0.160𝑥1𝑥1 +0.088𝑥1𝑥2 ―0.011𝑥1𝑥3 ―0.008𝑥2𝑥2 (2) ―0.008𝑥2𝑥3 +0.034𝑥3𝑥3 where y is the resolution value and xi are the chromatographic variables (x1, %B at the beginning; x2, %B at the end; and x3, flow rate). The percentage of variation in the response that has been explained by the fitted model, indicated as R2 statistics values, was ranged from 64.19% (Rs13-14) to 99.93% (total analysis time). Unlike the impact setting in PB column, the Rs < 1.0 i.e. Rs1-2, Rs4-5, Rs12-13, Rs16-17 and Rs17-18 were set to the intermediate setting of importance (3 impact) while low importance setting (1 impact) was set for the other 8 responses. The different impact assignment on RSs between SCB and PB columns was due to the different degree of RSs severely wherein the higher the value of RS the less severe the RS. Overall, as the values of RSs produced by DOE using SCB was higher than PB, the setting of the importance of RSs for optimization using SCB could be lower than PB to avoid unnecessary optimization associated with the impact of desirability. The target of optimization by 11 ACS Paragon Plus Environment


Page 12 of 31

MRO was set as higher than 1.0 for Rs and for analysis run time lower than 3 min. Once MRO was performed, the d(y) was plotted as a 3-D mesh plot, which shows the optimum point of the optimization (Figure 3). The MRO suggests the most convenient levels for the studied variables: %B initial (x1, +0.2860), %B end (x2, ― 0.9999) and flow rate (x3, +0.5469) thus the optimal setting of the experimental variables have been determined: %B initial (x1, 6.4%), %B end (x2, 50.0%) and flow rate (x3, 0.61 mL min-1). Using these values, the response variables generate a desirability index of 96.76%. Desirability values lower than 1.0 were observed only for Rs1-2 (0.78) and Rs4-5 (0.63). Because of the adjacent compounds possessed different maximum absorbance on UV-Vis detection, the compounds were completely separated at their corresponding maximum wavelengths. Henceforth, using the SCB under the optimized conditions, 20 phenolic compounds were effectively separated (Figure 4). Comparing the PB and SCB columns. Decreasing particle diameter, the column efficiency is increased; this effect has been one of the main advantage of the UPLC® technology using sub-2 μm particles. In this study, the particle sizes were 1.7 μm and 1.6 μm for PB and SCB columns, respectively. Nevertheless, using lower particle diameters the backpressure is increased, limiting the flow rate29. Therefore, on the BBD, the working range of flow rate for PB was larger than for SCB produce backpressures at the same range for both columns. A very important advantage of the solid-core particle compared to the fully-porous C18 is related to the improvement of the Rs while reducing analysis time, as reported by Song et al

30.

Hence, the number of Rs lower than 1.0 that were defined as responses during the method optimization for SCB was lower than PB applying the same BBD. Furthermore, without any further optimization, the optimized conditions suggested by MRO was successfully applied for the complete separation for 20 phenolics by the SCB column. Henceforth, this UPLC column was chosen and the method was validated. 12 ACS Paragon Plus Environment

Page 13 of 31


Validation of the UPLC method for SCB column. Table 4 summarizes the results of the UPLC method validation. The regression of the calibration graphs produced a high value of correlation coefficients, greater than 0.997 in the range studied. The limit of detection and limit of quantification were obtained from the values for the standard deviation (σ) for the response and the slope (a) . In Table 4, LOD and LOQ values are registered for each phenolic. LOD values ranged from 0.3 to 7.4 mg L−1 while the LOQ values were less than 15 mg L−1 except for GUA (22.5 mg L−1). The high sensitivity made this new method advantageous to determine the low concentration of some phenolics in rice grain samples. Repeatability and intermediate precision were calculated to establish the precision of the method. The CVs for repeatability of the retention time and the signal of area, on average, were 0.33% and 1.08% respectively while the intermediate precision were 0.38% and 2.13% respectively. It was also observed that p-HBA and SIN have the highest value for the repeatability (

Fast determination of phenolic compounds in rice grains by UPLC-PDA

Recommend Documents