Soybean [Glycine max (L.) Merr.] is a popular health nutritious vegetable food widely consumed in Indonesian. The economic value of soybean is very important for Indonesian society because various processed soy-based food products become their daily consumption. Isoflavone (aglycone form of glycosides) such as daidzein (DN) and genistein (GN) is reported to be more responsible for biological activities than glycoside form because aglycones had a good biological activity and are also easy to be absorbed (Izumi et al., 2000). DN and GN have potential protective effects on mammae cancers (Dhananjaya et al., 2012), estrogenic effect (Islam et al., 2008), and protective action against osteoporosis (Morabito et al., 2002).
High-performance liquid chromatography (HPLC) is an analytical technique widely used for the analysis of DN and GN in soybean samples or soybean-based food products (Hong et al., 2011; Magiera and Sobik, 2017; Shao et al., 2011; Yatsu et al., 2016). However, an HPLC analysis needs complex sample preparation and skillful analysis; therefore, fast and reliable analytical techniques based on Fourier transform Infrared (FTIR) spectroscopy methods combined with chemometrics are developed for the analysis of DN and GN in soybeans (Mulsow et al., 2015). FTIR spectroscopy is based on the interaction between infrared radiation and samples to get specific peaks corresponding to the absorption of functional groups present in the analyzed samples (Rohman, 2012).
FTIR spectroscopy coupled with multivariate data analysis (chemometrics) of multivariate calibrations can be used for the qualitative and quantitative analysis of compounds of natural ingredients (Rohman et al., 2014). FTIR spectroscopy-based techniques have been used for the quantification of curcuminoids in Curcuma syrups (Prabaningdyah et al., 2018) and analysis of curcumin and demethoxycurcumin in some extracts of turmeric (Curcuma longa L.) and java turmeric (Curcuma xanthorrhiza) (Lestari et al., 2017; Rohman et al., 2015). Based on this application, the correlation between FTIR spectroscopic and HPLC methods emerged very flexible for the analysis of DN and GN. Using a literature study, there are no reports regarding the employment of FTIR spectroscopy-based techniques for the analysis of DN and GN. For this reason, this study was aimed to optimize and develop FTIR spectroscopy coupled with partial least square (PLS) and principal component regression (PCR) for the analysis of DN and GN. As actual values of DN and GN in soybeans, the validated HPLC method was used.
MATERIALS AND METHODS
The reference standards of DN and GN were purchased from Sigma (Aldrich, St. Louis, MO). HPLC instrument was used for the quantitative analysis of DN and GN. Soybean samples were obtained from Balitkabi (Balai Penelitian Tanaman Aneka Kacang dan Umbi), Malang, East Java, Indonesia.
Preparation of soybean samples
Approximately 2.5 g of soybean samples were extracted with 25 ml of water–methanol (50:50 volume/volume) at room temperature for 3 days with immediate shaking. The mixture was then filtered, and then, 1.0 ml of sample solution was diluted with 9.0 ml of water–methanol (50:50 volume/volume). This solution was subjected to filtration using 0.45-µm polyvinylidene fluoride or polyvinylidene difluoride and finally introduced into HPLC system.
HPLC analytical conditions
Reserved-phase HPLC for the quantification of DN and GN in soybean samples was performed on HPLC waters, using column Sun Fire TMC-18 (150 × 4.6 mm, with internal diameter 5 µm), equipped with guard column Waters Symmetry TMC18 (20 × 4.4 mm, with an internal diameter of 5 µm). The composition of mobile phase consisted of a mixture of methanol and 0.1% acetic acid (53:47 volume/volume) delivered isocratically using a flow rate of 1.0 ml/minute. The volume of injection was 10 µl. For the preparation of standard solutions, 10 mg of reference standards were accurately weighed and then diluted with mobile phase to get stock solution of DN and GN (100 μg/ml). Aliquots of the solution (2, 3, 4, 5, 6, 7, 8, 9, and 10 μg/ml) were prepared and injected into the HPLC equipment. The solutes containing DN and GN were detected by using photodiode array at 254 nm. HPLC running time was 20 minutes with retention times (tR) of DN and GN which were 6.342 and 10.088, respectively.
FTIR spectroscopy analysis
The analysis of samples using FTIR spectroscopy was performed according to Wulandari et al. (2018). The powdered samples were directly placed on an attenuated horizontal total reflectance (Smart iTR™) accessory. The specification of measurement is as follows:
Software: OMNIC ver. 9.7
Wavenumbers of scanning: 4,000–650 cm−1
Number of scanning: 32 scans
Spectral resolution: 8 cm−1
Spectral background: air (environment) spectrum
Model prediction of DN and GN
The prediction of DN and GN was facilitated by multivariate calibrations of partial least-square regression (PLSR) and PCR. The levels of DN and GN obtained during HPLC were used as actual values, and then, the levels of DN and GN were predicted using the variables of absorbance values at the region of optimized wavenumbers. The correlation models for actual values and predicted values were regressed using PLSR.
The software of TQ Analyst from Thermo Fisher Scientific Inc. (Madison, WI) was used for data analysis applying multivariate calibrations including PLS and PCR. Statistical parameters used are as follows:
Coefficient of correlation (R) for correlation between HPLC actual values and FTIR predicted values which indicated the accuracy of the model.
The errors of root mean square error of calibration (RMSEC) and error of prediction (RMSEP) indicating the precision of the model (Miller and Miller, 2005).
RESULTS AND DISCUSSION
Using two models of multivariate calibrations, PLS and PCR were employed for the correlation between HPLC actual values of DN and GN as determined and FTIR predicted values using variables of absorbance values at certain FTIR spectra regions. The HPLC chromatogram obtained during the separation of DN and GN is shown in Figure 1. HPLC conditions used can provide a good system because it has an asymmetry value of 1.2. The relative standard deviation (RSD) values for peak area and peak height were obtained to meet the requirements of maximum RSD value, namely, ≤ 2%. In addition, HPLC used was validated previously by Sulistyowati et al. (2019).
FTIR spectra at the wavenumbers of 4,000–650 cm−1 of soybean pulverized containing DN were overlaid as shown in Figure 2. Each peak at specific wavenumbers (1/λ) in these spectra could be attributed by functional groups present in soybean samples. A wide peak at 1/λ 3,272 cm−1 was the vibration of -OH stretching, associated with hydrogen bond of -OH. Two peaks at 1/λ 2,925 and 2,855 cm−1 were coming from vibrations of CH3- and CH2- functional groups in stretching modes. A peak at 1/λ 1,743 cm−1 originated from carbonyl (C=O) groups (Rohman et al., 2015). The presence of these functional groups indicated the presence of DN and GN in soybean samples.
For quantitative analysis using FTIR spectroscopy, some optimizations were performed by selecting the FTIR spectral treatment and derivatization treatments. Based on the optimization applying wavenumbers’ region and FTIR spectral treatments (normal, first, and second derivatives) and relying on the highest coefficient of correlation (R-value) and lowest values of RMSEC and RMSEP, DN and GN were subjected to quantification using the first derivative spectra at combined 1/λ of 3,600–2,800 cm−1 and 1,500–780 cm−1 with 10 factors. The selection of the wavenumbers of samples is important because the measurements with not suitable wavenumber regions or less informative are able to reduce the performance of PLS/PCR modeling (El-Gindy et al., 2006). Tables 1 and 2 show the performance of multivariate calibrations of PLS and PCR models applying either normal or derivative spectra (first and second) for the quantitative analysis of DN and GN in soybeans together with some statistical performances, consisting of the number of factors, R-values, and error values (RMSEC and RMSEP).
|Figure 1. Chromatogram by HPLC with the SunFireTMC-18 (150 × 4.6 mm, 5 µm): (a) chromatogram of soybeans extract containing DN and GN and (b) chromatogram spiking performed using 1 ppm of DN and GN standard.|
[Click here to view]
|Figure 2. The overlay of FTIR spectra of soybean pulverized containing DN and scanned at midinfrared region (4,000–650 cm−1).|
[Click here to view]
The values of R obtained for the determination of DN using PLS were 0.9943 (calibration model) and 0.9961 (validation model), whereas the error values were 0.865% (calibration) and 0.994% (validation), respectively (Fig. 3). The R-values for the determination of GN were 0.9882 (calibration) and 0.9936 (validation) with the error values of 0.874% (calibration) and 1.01% (validation), respectively (Fig. 4). Figure 5 shows the correlation between HPLC actual results (x-axis) and FTIR calculated values (y-axis) in validation samples which resulted in R2 > 0.99. This indicated that PLS with appropriate spectral treatments is accurate and precise methods for the quantification of DN and GN in soybeans (Sim et al., 2004).
|Table 1. The characteristic performances of PLS for modeling the correlation between HPLC actual values and FTIR predicted values of DN and GN.|
[Click here to view]
|Table 2. The performance of multivariate calibrations of PCR for modeling HPLC actual values and FTIR predicted values of DN and GN.|
[Click here to view]
|Figure 3. PLS model for the relationship between HPLC actual values (x-axis) and FTIR calculated values of DN. (a) Calibration model; (b) residual analysis.|
[Click here to view]
|Figure 4. PLS model for the relationship between HPLC actual values (x-axis) and FTIR calculated values of GN. (a) Calibration model; (b) residual analysis.|
[Click here to view]
|Figure 5. Scatter plot for the relationship between HPLC actual values of DN and GN and FTIR predicted values. (a) Daidzein; (b) GN. R2 = coefficient of determination.|
[Click here to view]
HPLC was used successfully for the analysis of DN and GN in soybean samples, and the concentrations obtained were used for actual values during the prediction of DN and GN in soybean samples. FTIR spectroscopy using the first derivative spectra at combined 1/λ of 3,600–2,800 and 1,500–780 cm−1 coupled PLS model could be used as a rapid and reliable method for the quantitation of DN and GN in soybeans. FTIR spectroscopy offered the reliable method for the analysis of DN and GN in soybeans with the main advantages of being rapid, minimum sample preparation, and being environmental friendly. The developed method could be prolonged to be used in the analysis of soybean-based food products such as soybean milk and soy sauce.
The authors acknowledged the Ministry of Research and Higher Education, Republic of Indonesia, for financial support during this study.
CONFLICT OF INTEREST
Authors declared that they do not have any conflicts of interest.
Dhananjaya K, Sibi G, Mallesha H, Ravikumar KR, Awasthi S. Insilico studies of daidzein and genistein with human estrogen receptor α. Asian Pac J Trop Biomed, 2012; 2:1747–53. CrossRef
El-Gindy A, Emara S, Mostafa A. Aplication and validation of chemometrics-assisted spectrophotometry and liquid chromatography for the simultaneous determination of six-component pharmaceuticals. J Pharm Biomed Anal, 2006; 41(2):421–30. CrossRef
Hong JL, Qin XY, Shu P, Wang Q, Zhou ZF, Wang GK. Comparative study of isoflavones in wild and cultivated soybeans as well as bean products by high-performance liquid chromatography coupled with mass spectrometry and chemometric techniques. Eur Food Res Technol, 2011; 233:869–80. CrossRef
Islam, F., Sparkes, C., Roodenrys, S., dan Astheimer, L., 2008. Short-term changes in endogenous estrogen levels and consumption of soy isoflavones affect working and verbal memory in young adult females. Nutr Neurosci, 11(6): 251–62. CrossRef
Izumi T, Piskula MK, Osawa S, Obata A, Tobe K, Saito M. Soy isoflavone aglycones are absorbed faster and in higher amounts than their glucosides in humans. J Nutrition, 2000; 130:1695–9. CrossRef
Lestari HP, Martono S, Wulandari R, Rohman, A. Simultaneous analysis of curcumin and demethoxycurcumin in Curcuma xanthorriza using FTIR spectroscopy and chemometrics. Int Food Res J, 2017; 24:2097–101.
Magiera S, Sobik A. Ionic liquid-based ultrasound-assisted extraction coupled with liquid chromatography to determine isoflavones in soy foods. J Food Compost Anal, 2007; 57:94–101. CrossRef
Morabito N, Crisafulli A, Vergara C, Gaudio A, Lasco A, Frisina N. Effects of genistein and hormone-replacement therapy on bone loss in early postmenopausal women: a randomized double-blind placebo-controlled study. J Bone Miner Res, 2002; 17(10):1904–12. CrossRef
Mulsow K, Eidenschink J, Melzig MF. FT-IR Method for the quantification of isoflavonol glycosides in nutritional supplements of soy (Glycine max (L.) Merr.). Sci Pharm, 2015; 83(2):377–86. CrossRef
Rohman A, Ramadhani D, Nugroho A. Analysis of curcumin in Curcuma longa and Curcuma xanthorriza using FTIR spectroscopy and chemometrics. Res J Med Plant, 2015; 9:179–86; doi:10.3923/rjmp.2015.179.186 CrossRef
Rohman A, Setyaningrum DL, Riyanto S. FTIR spectroscopy combined with partial least square for analysis of red fruit oil in ternary mixture system. Int J Spectroscopy, 2014; 1-5. CrossRef
Shao S, Duncan AM, Yang R, Marcone MF, Rajcan I, Tsao R. Systematic evaluation of pre-HPLC sample processing methods on total and individual isoflavones in soybeans and soy products. Food Res Int, 2011;44(8) 2425-2434. CrossRef
Sulistyowati E, Martono S, Riyanto S, Lukitaningsih E. Development and validation for free aglycones daidzein and genistein in soybeans (Glycine max (l.) Merr.) using RP HPLC method. Int J Appl Pharm, 2019; 11(2): 138-142. CrossRef
Wulandari R, Sudjadi, Martono S, Rohman A. Liquid chromatography and Fourier transform infrared spectroscopy for quantitative analysis of individual and total curcuminoid in Curcuma longa extract. J App Pharm Sci, 2018; 8(09):107–13. CrossRef
Yatsu FKJ, Koester LS, Bassani VL. Isoflavone-aglycone fraction from Glycine max: a promising raw material for isoflavone-based pharmaceutical or nutraceutical products. Rev Bras Farmacogn, 2016; 26(2):259–67. CrossRef