Modeling and optimizing microwave-assisted extraction
of antioxidants from Thymbra Spicata L. and characterization
of their phenolic constituents
Mustafa Bener1
Received: 16 May 2019 / Revised: 22 August 2019 / Accepted: 16 September 2019 / Published online: 11 October 2019 Ó The Korean Society of Food Science and Technology 2019
Abstract Response surface methodology was used for modeling and optimizing microwave-assisted extraction of antioxidants from Thymbra spicata L. as a factor of tem-perature, extraction time, solvent concentration, and sol-vent-to-solid ratio. The prepared extracts showed maximum antioxidant properties, including total phenolic content (TPC), total antioxidant capacity (TAC), and rad-ical scavenging activity (RSA) at the optimum operating conditions. All models calculated for the three responses that are TPC, TAC, and RSA were noteworthy (p \ 0.0001) and showed a significant relationship between the response and independent parameters. There was a close relationship between the experimental and the predicted values obtained using the proposed method. The phenolic antioxidant profile of Thymbra spicata L. extract was characterized with the UPLC-PDA-ESI–MS/MS sys-tem and rosmarinic acid was found as a major component (1089.2 ± 10.9 mg/100 g-DS). In the future, this opti-mized and modeled MAE method can be applied in food and pharmaceutical industries to effectively extract antioxidants from edible Thymbra spicata L. plant.
Keywords Thymbra spicataL. Antioxidant Microwave-assisted extraction Response surface methodology UPLC-PDA-ESI–MS/MS
Introduction
Under normal physiological conditions, the antioxidants should be in equilibrium with reactive oxygen species (ROS) in the human body. However, under oxidative stress when this balance is broken in favour of ROS, severe damage occurs to the biological macromolecules resulting into several diseases such as Alzheimer, cancer, and coronary heart diseases (Thompson, 1994). Antioxidative enzymes and other compounds with antioxidant properties are found in all organisms and serve to prevent or repair the damage that may occur to macromolecules due to attack by ROS (Ak and Gu¨lc¸in, 2008). The best way to fight these health problems is by consuming foods rich in antioxidants that balance the excess ROS. It has been suggested that natural foods including fruits, vegetables, and spices are so rich in phenolic compounds that are antioxidant in nature (Croft, 1998).
Thymbra spicata L., which belongs to Lamiaceae fam-ily, is mainly grown in some eastern Mediterranean countries; its leaves are dried and taken as herbal tea, spice, and antiseptic agent in folk medicine (U¨ nlu¨ et al., 2009). Thymbra spicata L. has been identified as a rich source of antioxidants such as terpenoids, flavonoids, and iso-prenoids (Hancı et al.,2003). It was reported that Thymbra spicata L. extracts show antioxidant, liver protective, and cholesterol-reducer properties (Akkol et al.,2009; Bozkurt,
2006).
The interest to extract bioactive compounds from com-plex food samples is increasing day by day. Although there are several studies about the extraction of essential oils of Thymbra spicata L. in the literature (Fleisher and Fleisher,
2005; Ozel et al., 2003; Sonsuzer et al., 2004), there are limited studies on the extraction of the antioxidants. In the existing studies, classical heat extraction (Akkol et al., Electronic supplementary material The online version of this
article (https://doi.org/10.1007/s10068-019-00687-5) contains sup-plementary material, which is available to authorized users. & Mustafa Bener
1 Division of Analytical Chemistry, Department of Chemistry, Faculty of Engineering, Istanbul University-Cerrahpasa, Avcilar, 34320 Istanbul, Turkey
2009), soxhlet extraction (Gumus et al., 2011), and hydrodistillation (Dorman et al.,2004) were used to extract the antioxidants from Thymbra spicata L. plant. Compared to other extraction methods, Microwave-Assisted Extrac-tion (MAE) comes into prominence because of its promi-nent advantages such as low solvent requirement, multi-sample processing, automation potentials, and short extraction time (Bener et al.,2016; O¨ zyu¨rek et al.,2014). In MAE like in other solid-solvent extraction processes, the solvent is dispersed into the sample matrix and the com-pound of interest is extracted out of the matrix by the solvent (Camel,2000). In MAE, the process can be done within minutes using polar solvents or non-polar solvents (commonly in water or other polar solvents), and by using a software-based system, the operating parameters can be controlled (Zigoneanu et al., 2008). In MAE, non-polar solvents are generally not used alone, because they do not generate heat under the microwave due to their low dielectric constants and so the extraction efficiency is reduced (Rafiee et al.,2011; Raman and Gaikar,2002). The design requires the system to operate at high temperatures, and the temperature of the system is controlled by a fibre optic sensor or an infrared sensor. To maximize MAE yield, it is necessary to optimize operational factors like temperature, extraction time, solvent concentration, and solvent-to-solid ratio before analysis.
Nowadays, the use of synthetic antioxidants to combat free radicals, which causes oxidative damage, has consid-erably increased. Recent researches have shown that the use of synthetic antioxidants causes vital problems such as cancer and liver damages besides their benefits (Yuan et al.,2008). In this context, the use of natural antioxidants instead of synthetic antioxidants gains importance in combating oxidative damage (Kan et al.,2015). Therefore, the most important sources of natural antioxidants are edible plant extracts. In this study, the possibility of using the MAE method as a rapid and effective method for extracting antioxidants from Thymbra spicata L. was investigated for the first time, furthermore, the operational conditions were optimized. The antioxidant properties which were studied are total phenolic content (TPC), total antioxidant capacity (TAC), and radical scavenging activ-ity (RSA). Optimization and modeling of the MAE method are important to ensure maximum antioxidant recovery from Thymbra spicata L. plant. The main operating factors (temperature, time, solvent concentration, and solvent-to-solid ratio) were optimized by applying the response sur-face method. Another target of this study was to evaluate the effects of operating parameters on TPC, TAC, and RSA of the endemic Thymbra spicata L. extract. Using this proposed optimized MAE based method, antioxidant compounds were obtained from Thymbra spicata L. in a simple, rapid, cost-effective, and efficient manner. Using
ultra-performance liquid chromatography (UPLC)-photo-diode array detector (PDA) and tandem mass spectrometry (MS/MS) system, the phenolic profile of the optimized extract of Thymbra spicata L. was characterized for the first time. Nine phenolic antioxidant compounds were identified and quantified in the optimized extract of Thymbra spicata L. plant. Therefore, the phenolic antiox-idants in Thymbra spicata L. extract have been extensively characterized for the first time in literature in this study. From this study, therefore, it can be concluded that the modeled and optimized MAE method has the potential to be used in the food and pharmaceutical industries to effectively extract antioxidants from Thymbra spicata L. at a low-cost.
Materials and methods
Materials and instrumentations
In this study analytical purity grade chemicals were used and sourced from the following suppliers: neocuproine (Nc), DPPH, the Folin–Ciocalteau reagent, trolox (TR), gallic acid (GA), protocatechuic acid, syringic acid, api-genin, narinapi-genin, rosmarinic acid, quercetin, hesperidin, rutin, carvacrol, formic acid (HCOOH), methanol (MeOH), and ethanol (EtOH) from Sigma-Aldrich (St. Louis, Mo., USA.); ammonium acetate (NH4Ac), copper(II) sulfate, potassium sodium tartrate tetrahydrate, copper(II) chloride dihydrate, sodium carbonate, and sodium hydroxide from Merck (Darmstadt, Germany). Thymbra spicata L. were supplied by Arifoglu Baharat Gida San. Ltd. S¸ti. (Istanbul, Turkey).
Varian CARY Bio 100 UV–Vis spectrophotometer (Mulgrave, Victoria, Australia) was used for the absorption measurements. MAE of the antioxidants from Thymbra spicata L. was done with using Milestone ETHOS ONE microwave extraction system (Shelton, CT, USA) and the temperature was controlled by a fibre optic sensor. Waters Acquity ultra-performance liquid chromatography (UPLC) system coupled with photodiode array detector (PDA) and Xevo TQD mass spectrophotometer (double quadrupole analyzer) equipped with Z-spray ESI (electrospray ioniza-tion) source (Milford, MA, USA) was used for character-ization and quantification of Thymbra spicata L. phenolic antioxidants. Acquity C18 column (100 9 2.1 mm, 1.7 lm) was used for the UPLC analysis.
Microwave-assisted extraction
The air-dried raw Thymbra spicata L. sample was groun-ded in a mill before MAE (particle sizes after milling were 250–355 lm). Microwave-assisted treatment of Thymbra
spicata L. was carried out using a microwave-assisted extraction system at different temperatures (50–100°C), for times (1–10 min), at a solvent concentration (0–100%, ethanol in water) and at solvent-to-solid ratio (5–20 mL/ 0.2 g). The microwave power (0–1500 W) inside the oven was automatically adjusted by the system according to the temperature in closed vessels. Three replicates were per-formed for each extraction. After extraction, the mixture was cooled to room temperature and the extracts filtered through a filter paper, then through 0.45 lm PTFE syringe filters, and kept in a refrigerator at ? 4°C until use.
Determination of total phenolic content (TPC)
The TPC of the Thymbra spicata L. extracts was deter-mined using Folin–Ciocalteau assay (Singleton et al.,
1999). Reagents used in Folin assay were prepared as follow: 2% Na2CO3in 0.1 M NaOH is a Lowry A solution; 0.5% CuSO4in 1% NaKC4H4O6is a Lowry B solution; Lowry C reagent was freshly prepared as a mixture of 50 mL Lowry A solution ? 1 mL Lowry B solution; the Folin–Ciocalteau reagent was diluted with distilled water to a volume ratio of 1:3 prior to use.
In this method, firstly x mL of the sample extract and (1 - x) mL of H2O were added to the test tube. Then, 2.5 mL of Lowry C solution was added, and the mixture was allowed to stand for 10 min. After 10 min, 0.25 mL of Folin reagent was added to the mixture and allowed for 30 min at room temperature. Finally, the absorbance was recorded at 750 nm against a reagent blank. The total phenolic content was expressed as TR equivalent (mmol TR/g-dried sample (DS)) unit based on the standard curve of TR standard.
Determination of total antioxidant capacity (TAC)
The TAC of Thymbra spicata L. extracts was assessed by employing the CUPRAC (Cupric Reducing Antioxidant Capacity) method (Apak et al.,2004). CUPRAC method is based on the reduction of a chromogenic reagent [Cu(II)– Nc] to the highly colored Cu(I)–Nc chelate, and the absorbance measured at 450 nm.
In this method, x ml of sample extract was added to a test tube that contained 3 mL of the reagent mixture (1 mL of 10 mM CuCl2, 1 mL of 7.5 mM Nc, and 1 mL of 1.0 M NH4Ac) and (1.1 - x) mL distilled water. After 30 min, the absorbance of the sample was recorded at 450 nm against a reagent blank. The total antioxidant capacity was expressed as a TR equivalent (mmol TR/g-DS) based on the standard curve of TR standard.
Determination of radical scavenging activity (RSA)
The RSA of the Thymbra spicata L. extracts on DPPH free radicals was determined using the DPPH assay (Sa´nchez-Moreno et al.,1998). In summary, x mL of sample extract was mixed with (2 - x) mL of methanol and 2 mL of 2 9 104M DPPH solution in a test tube. The tubes were stoppered and left in the dark for 30 min. The absorbance at 515 nm was recorded against methanol. The corrected absorbance values were used to calculate the radical scavenging activity of the sample. It was calculated from the following Eq. (1):
DA¼ ADPPH As Að 0Þ ð1Þ
where DA was the corrected absorbance of the sample, ADPPHwas the absorbance of DPPH without the sample, As was the absorbance of the sample with DPPH, A0was the absorbance of the sample without DPPH.
Chromatographic characterization of Thymbra spicata L. phenolic antioxidants
The phenolic characterization of Thymbra spicata L. extracts was performed using UPLC-PDA-ESI–MS/MS system based on a method which was described by C¸ elik et al. (2017) with some modifications. Mobile phases were 0.1% HCOOH in bidistilled H2O (A) and 0.1% HCOOH in MeOH (B). The system was run in gradient mode at flow rate: 0.45 mL/min, sample temperature: 4°C, injection volume: 10 lL, and column temperature: 30°C: at 0. min 90% (A)–10% (B); 2. min 85% (A)–15% (B), curve (6.0); 3. min 80% (A)–20% (B), curve (6.0); 4. min 70% (A)– 30% (B), curve (6.0); 6–10. min 60% (A)–40% (B), curve (6.0); 15. min 30% (A)–70% (B), curve (6.0); 18–20. min 90% (A)–10% (B), curve (6.0). In order to monitor the phenolic antioxidants, the analytical wavelengths of detection were from 400 to 190 nm. The MS ionization conditions were adjusted as follows: cone gas flow rate, 50 L/h; desolvation gas flow rate, 650 L/h; collision energy, 20 V; desolvation temperature, 400°C; and cap-illary voltage, 3 kV. The phenolic antioxidant profile of Thymbra spicata L. was characterized using mass-to-charge (m/z) transitions of precursor and product ions in multiple reaction monitoring (MRM) mode. In order to quantify the phenolic antioxidants of Thymbra spicata L., the calibration curves (peak area vs. concentration graphs) of each phenolic antioxidant were obtained under the described conditions.
Statistical analysis
To analyze and optimize the experimental data, Face-Centered Composite Design (FCCD) was employed using
Design-ExpertÒSoftware Version 11 Trial (Stat-Ease, Inc., Minneapolis, USA). In this study, extraction temperature (X1), extraction time (X2), solvent concentration (ethanol in water; %: v/v) (X3) and the solvent-to-solid ratio (X4) were selected as the independent variables (four factors) and TPC (Y1), TAC (Y2), and RSA (Y3) were selected as the dependent variable (three responses). Based on the experimental data, the factors and levels were determined as given in Supplementary Table S1. These factors and their levels were entered into Design-ExpertÒ Software, and the experimental plan was obtained as presented in Table1. It also shows the values of the responses calcu-lated from experimental data.
The functional relationship between the independent variables and the response can be described as (Barani and Maleki,2011):
Y¼ f Xð 1; X2; . . .; XkÞ þ e ð2Þ
In this FCCD study, the second-order polynomial model was fitted to the experimental data. This model equation can be written as (Barani and Maleki,2011):
Y ¼ b0þ Xk i¼1 bixiþ Xk i¼1 biix 2 i þ Xk1 i¼1 Xk j¼2 bijxixjþ e ð3Þ
where f, response function; Y, the response; X, indepen-dent variable; b0, the regression coefficient constant; bi, bii,
Table 1 FCCD of the independent factors (X1, X2, X3, and X4) for the MAE and experimental results for the TPC, TAC, and RSA Run No Factors
X1 X2 X3 X4 TPC (mmol TR/g-DS) TAC (mmol TR/g-DS) RSA (mmol TR/g-DS)
1 50 1 0 5 0.446 0.326 0.171 2 100 1 0 5 0.746 0.396 0.225 3 50 10 0 5 0.458 0.337 0.185 4 100 10 0 5 0.758 0.411 0.243 5 50 1 100 5 0.238 0.125 0.069 6 100 1 100 5 0.585 0.211 0.125 7 50 10 100 5 0.269 0.104 0.074 8 100 10 100 5 0.593 0.238 0.147 9 50 1 0 20 0.645 0.448 0.241 10 100 1 0 20 0.884 0.513 0.281 11 50 10 0 20 0.651 0.439 0.248 12 100 10 0 20 0.892 0.523 0.293 13 50 1 100 20 0.286 0.112 0.064 14 100 1 100 20 0.515 0.269 0.159 15 50 10 100 20 0.294 0.134 0.077 16 100 10 100 20 0.528 0.285 0.164 17 50 5.5 50 12.5 0.692 0.425 0.259 18 100 5.5 50 12.5 0.864 0.540 0.293 19 75 1 50 12.5 0.956 0.614 0.369 20 75 10 50 12.5 0.967 0.638 0.393 21 75 5.5 0 12.5 0.945 0.583 0.328 22 75 5.5 100 12.5 0.605 0.382 0.249 23 75 5.5 50 5 0.812 0.482 0.291 24 75 5.5 50 20 0.968 0.695 0.368 25 75 5.5 50 12.5 0.984 0.722 0.384 26 75 5.5 50 12.5 0.997 0.690 0.389 27 75 5.5 50 12.5 0.985 0.714 0.396 28 75 5.5 50 12.5 0.984 0.674 0.338 29 75 5.5 50 12.5 0.996 0.735 0.391 30 75 5.5 50 12.5 1.004 0.705 0.383
bij, interaction coefficients; k, the factor number and e, experimental error.
The assays of each extract for each response (TPC, TAC, and RSA) were carried out in triplicate. The rela-tionship between the factors and the responses was estab-lished using the analysis of variance (ANOVA) test in the Design-Expert program.
Results and discussion
Table1 summarizes the effects of extraction temperature (50–100°C), extraction time (1–10 min), solvent concen-tration (0–100%, ethanol in water), and solvent-to-solid ratio (5–20 mL/0.2 g) on the TPC, TAC, and RSA of the Thymbra spicata L. extract obtained using MAE. The TPC values of the extracts were found to range from 0.238 to 1.004 mmol TR/g-DS. As for the TAC values, these ranged between 0.104 and 0.735 mmol TR/g-DS. The RSA values of the extracts varied from 0.064 to 0.396 mmol TR/g-DS with respect to inhibition of DPPH radical.
Modeling and optimization of MAE using RSM
All the determined models corresponding to the responses TPC, TAC, and RSA were notable (p \ 0.0001) and showed a significant relationship between the response and independent factors. Solvent concentration was found to be the most significant operating factor amongst all the responses, on the other hand, extraction time was the least significant factor. Table2illustrates the results after fitting quadratic models to the data. Basing on the results of ANOVA, the contribution of the quadratic model is notable.
The quadratic polynomial models for the coded values of TPC, TAC, and RSA were determined from Eqs.4,5, and 6. As in Table2, using the F-test and p value, the significance of each coefficient was determined. As the absolute F value increases and the p value decreases, the corresponding variables become more significant. To ver-ify the accuracy of the models, lack of fit was tested as given in Table2. TPC¼ 1:5218 þ 0:049795X1þ 4:82 104X2 þ 0:005807X3þ 0:050396X48:89 106X1X2 þ 2:70 106X 1X31:09 104X1X4 þ 6:11 106X2X35:20 105X2X41:21 104X 3X42:88 104X21þ 1:7 104X227:30 105X230:0121X2 4 ð4Þ TAC¼ 1:09901 þ 0:038205X1þ 0:004537X2 þ 0:003648X3þ 0:028160X4þ 3:60 105X1X2 þ 1:20 105X1X3þ 3:10 105X1X4þ 4:72 106X2X3þ 1:30 105X2X45:50 105X3X42:49 104X215:88 104X2 26:20 10 5X2 38:78 10 4X2 4 ð5Þ RSA¼ 0:701520 þ 0:023172X10:004977X2 þ 0:002006X3þ 0:021594X4þ 100 105X1X2 þ 5:70 106X1X3þ 8:67 106X1X41:67 106X2X34:10 105X2X43:20 105X3X4 1:49 104X21þ 5:73 104X223:2 105X237:09 104X24 ð6Þ The models obtained for TPC, TAC, and RSA were found to be significant (p [ 0.05), confirming that the model can sufficiently fit the experimental data. In addi-tion, the predicted values of R2that is 0.9678, 0.9158, and 0.9497 reasonably agreed with the adjusted values of R2 that is 0.9844, 0.9515, and 0.9704 for TPC, TAC, and RSA, respectively. The differences between the two are less than 0.2. The signal-to-noise ratio was determined from the precision. It is desirable to have a ratio greater than 4. The ratio of TPC, TAC, and RSA were 33.256, 18.198, and 24.954, respectively indicating adequate signals. These models can be used to navigate the design space. The MAE operational factors were optimized to obtain maximum responses by using suitable software. The highest TPC (1.04 mmol TR/g-DS; 6.73 g GA/100 g-DS) was obtained under the experimental conditions of X1= 83 °C, X2= 8.3 min, X3= 26.3% and, X4= 17.3 mL/0.2 g; the highest TAC (0.71 mmol TR/g-DS) was obtained under conditions of X1= 79°C, X2= 6.6 min, X3= 29.7% and, X4= 16.5 mL/0.2 g; and the highest RSA (0.40 mmol TR/ g-DS) was obtained under conditions of X1= 79 °C, X2= 7.2 min, X3= 29.4% and, X4= 14.6 mL/0.2 g. In order to obtain maximum antioxidant properties, a study to optimize operational factors was also done. Shan et al. (2005) found that TPC values of Origanum vulgare L., Rosmarinus officinalis L., and Salvia officinalis L., mem-bers of Lamiaceae family, were 10.17, 5.07 and 5.32 g GA/ 100 g-DS, respectively. In the same study, TAC values for Origanum vulgare L., Rosmarinus officinalis L., and Salvia officinalis L. were found to be 1.00, 0.38 and 0.52 mmol TR/g-DS, respectively.
Table 2 ANOVA for the quadratic equations of design expert 8.0.7.1 for the TPC, TAC, and RSA
Sum of squares df Mean square F value P value Prob [ F
Model (TPC) 1.83 14 0.1308 132.08 \ 0.0001 X1-Extraction temperature 0.3163 1 0.3163 319.28 \ 0.0001 X2-Extraction time 0.0007 1 0.0007 0.6663 0.4271 X3-Solvent concentration 0.3506 1 0.3506 353.89 \ 0.0001 X4-Solvent-to-solid ratio 0.0319 1 0.0319 32.22 \ 0.0001 X1X2 0.0000 1 0.0000 0.0162 0.9006 X1X3 0.0002 1 0.0002 0.1840 0.6741 X1X4 0.0067 1 0.0067 6.79 0.0199 X2X3 0.0000 1 0.0000 0.0305 0.8636 X2X4 0.0000 1 0.0000 0.0495 0.8270 X3X4 0.0329 1 0.0329 33.25 \ 0.0001 X12 0.0840 1 0.0840 84.79 \ 0.0001 X22 0.0000 1 0.0000 0.0311 0.8624 X32 0.0868 1 0.0868 87.64 \ 0.0001 X42 0.0120 1 0.0120 12.11 0.0034 Residual 0.0149 15 0.0010 Lack of fit 0.0145 10 0.0014 20.06 0.0020 Pure error 0.0004 5 0.0001 Cor total 1.85 29 Model (TAC) 1.17 14 0.0832 41.62 \ 0.0001 X1-Extraction temperature 0.0487 1 0.0487 24.33 0.0002 X2-Extraction time 0.0005 1 0.0005 0.2507 0.6239 X3-Solvent concentration 0.2487 1 0.2487 124.35 \ 0.0001 X4-Solvent-to-solid ratio 0.0345 1 0.0345 17.25 0.0009 X1X2 0.0003 1 0.0003 0.1320 0.7214 X1X3 0.0035 1 0.0035 1.73 0.2087 X1X4 0.0005 1 0.0005 0.2702 0.6108 X2X3 0.0000 1 0.0000 0.0090 0.9256 X2X4 0.06 9 10-6 1 0.06 9 10-6 0.0015 0.9693 X3X4 0.0068 1 0.0068 3.42 0.0841 X12 0.0626 1 0.0626 31.28 \ 0.0001 X22 0.0004 1 0.0004 0.1838 0.6742 X32 0.0626 1 0.0626 31.28 \ 0.0001 X42 0.0063 1 0.0063 3.16 0.0956 Residual 0.0300 15 0.0020 Lack of fit 0.0276 10 0.0028 5.65 0.0349 Pure error 0.0024 5 0.0005 Cor total 1.20 29 Model (RSA) 0.3408 14 0.0243 68.83 \ 0.0001 X1-Extraction temperature 0.0163 1 0.0163 46.15 \ 0.0001 X2-Extraction time 0.0008 1 0.0008 2.26 0.1533 X3-Solvent concentration 0.0656 1 0.0656 185.63 \ 0.0001 X4-Solvent-to-solid ratio 0.0074 1 0.0074 20.93 0.0004 X1X2 0.0000 1 0.0000 0.0573 0.8141 X1X3 0.0008 1 0.0008 2.30 0.1504 X1X4 0.0000 1 0.0000 0.1195 0.7344 X2X3 2.25 9 10-6 1 2.25x10-6 0.0064 0.9375 X2X4 0.0000 1 0.0000 0.0855 0.7739
Effects of operational factors on MAE of antioxidants from Thymbra spicata L
The 3D response surface plots in Figs.1,2and3illustrate equations (Eqs.4, 5 and 6) and predict the relationship between the different factors and how these factors affect the TPC, TAC, and RSA responses. Figure1shows that the TPC, TAC, and RSA responses have similar trends with respect to solvent concentration. Polar solvents are widely used for the extraction of polyphenols from natural plant samples. Water, ethanol, methanol, acetone, and their acidic or non-acidic aqueous mixtures are the most suit-able extraction solvents (Do et al.,2014; Naczk and Sha-hidi, 2006). In this study, the ethanol–water mixture was chosen as the solvent for the extraction of antioxidants, and the optimum solvent concentration (ethanol in water (%), v/v) was determined. Approximately 30% of the solvent concentration showed the highest efficiency for the extraction of antioxidants due to their low viscosity and also allowed the solvent to penetrate the leaves by swelling of the matrix. Accordingly, while water acts as a plant blowing agent, it is believed that ethanol disrupts the bond between solutes and plant matrices (Wang and Weller,
2006; Zuo et al.,2008). In order to increase the efficiency of extraction by enhancing the penetration of the solvent into plant tissues, the concentration of EtOH should be kept within limit. If the EtOH concentration exceeds the critical value, undesirable factors, such as impurity extraction and protein coagulation, may prevent solvent penetration (Simic´ et al.; 2016). Therefore, amongst all the solvents used the mixture of water and ethanol showed the best performance in extracting the polyphenols. In addition, water has a high dielectric constant, which increases the polarity index of ethanol as a factor (Hemwimon and Pavasant,2007; Spigno et al.,2007).
In addition, Fig.1 shows that TPC, TAC and, RSA of the extracts increased with the MAE temperature up to near 80°C. The closed vessel MAE system allows the
extraction solvent to work at temperatures well beyond the boiling point. Extraction efficiency increases with increasing temperature because the solubility of the ana-lytes in the active sites of the matrix also increases. In addition, while the surface tension and the solvent viscosity decrease with temperature, the solvents have the capacity to dissolve the analytes at higher temperatures improving sample wetting and matrix penetration (Chen et al.,2007; Hemwimon and Pavasant, 2007). As reported in many studies, the increase in working temperature causes a cor-responding increase in the efficiency of the extraction, but structural decomposition occurs in phenolics with antioxi-dant properties at high temperatures (Bener et al., 2013; Wettasinghe and Shahidi,1999; Yilmaz and Toledo,2006). Therefore, any further increase in temperature beyond 80°C leads to a corresponding decrease in the TPC, TAC, and RSA values of the extract which agrees to the infor-mation in the literature. As a result, very high temperatures should not be preferred in the economic and safety points of view and the possible degradations of antioxidant compounds.
As the solvent-to-solid ratio increased, the TPC, TAC, and RSA of the extract increased (Fig.2). This is consis-tent with the principles of mass transfer; the driving force during the mass transfer is the concentration gradient between the solid and the mass of the liquid, which is larger when a higher solvent-solid ratio is used (Cacace and Mazza,2003; Pinelo et al.,2005). Nearly 16 mL/0.2 g was determined to be the optimal solvent-to-solid ratio from the models for TPC, TAC, and RSA. Although the use of a higher solvent volume generally has only a small positive effect on the extraction efficiency, it may be disadvanta-geous in terms of both cost and environmental factors.
All responses slightly increased with extraction time (Fig.3). The mechanism of the MAE process described herein has two major stages; dissolution of soluble com-ponents on surfaces of the plant matrix which occurs first followed by mass transfer of the solute from the plant Table 2continued
Sum of squares df Mean square F value P value Prob [ F
X3X4 0.0023 1 0.0023 6.38 0.0233 X12 0.0226 1 0.0226 63.91 \ 0.0001 X22 0.0003 1 0.0003 0.9859 0.3365 X32 0.0170 1 0.0170 47.95 \ 0.0001 X42 0.0041 1 0.0041 11.66 0.0038 Residual 0.0053 15 0.0004 Lack of fit 0.0031 10 0.0003 0.6893 0.7122 Pure error 0.0022 5 0.0004 Cor total 0.3461 29
matrix into the solvent by diffusion and osmotic processes (I˙lbay et al.,2013). It is known that extended extraction time increases the extraction of polyphenols from natural products. The possible reason for this may be the time needed for the solvent to penetrate into the solid sample, to dissolve the soluble components in the sample, and to release the solvent from the sample with the solute com-ponents (Fan et al.,2008; Gan and Latiff,2011; Lee et al.,
2006). The MAE system offers the opportunity to complete the extraction process in a short time. In this study, the meantime for extraction for the three responses was 7 min. In addition, it was determined that extraction time was the most insignificant factor in the extraction of antioxidant besides the three other factors. In addition, it is clear that
long extraction times are not practical from an economic point of view.
Verification of predictive models
In light of the obtained results, an optimization study was conducted to evaluate the optimal working conditions of the individual response with the aim to obtain high antioxidant yields within the extraction parameters while putting the yield, the feasibility of the experiment, and the suitability to green chemistry into consideration. Supple-mentary Table S2 shows the optimal conditions for each individual response with theoretical and experimental values. Under the described conditions, the experimental Fig. 1 The three-dimensional (3D) surface plot for the (A) TPC, (B) TAC, and (C) RSA of the Thymbra spicata L. extract as a function of solvent concentration and temperature (Solvent-to-solid ratio 12.5 mL/0.2 g and extraction time 5.5 min)
TPC, TAC, and RSA values of the extracts were 1.04, 0.71, 0.40 mmol TR/g-DS, respectively. The experimental val-ues were found to be close to the predicted valval-ues (Sup-plementary Table S2).
Supplementary Figure S1 shows the correlation between the predicted (calculated by the second-order model) and experimental TPC, TAC, RSA values of the Thymbra spicata L. extracts obtained by MAE performed at different operating factors. Consequently, the values predicted from the model were found to be in good agreement with those obtained in the experimental study.
UPLC-PDA-ESI–MS/MS analysis of the optimized extract of Thymbra spicata L.
The phenolic antioxidant profile of the Thymbra spicata L. extract was characterized by UPLC-PDA-ESI–MS/MS
system. Supplementary Table S3 shows the optimal con-ditions for MS/MS parameters (MRM transitions, frag-ments, cone voltage, collision voltage, and retention time). Figure4 shows the total and extracted ion chromatograms of phenolic antioxidants identified in Thymbra spicata L. extract. In addition, the UPLC-PDA chromatogram of phenolics detected in standard mixture solution is sum-marized in Supplementary Figure S2. The peaks were identified by comparing the retention times of the mass and UV–Vis spectra of individual phenolic antioxidants to those of the standards.
The linear calibration curves (chromatographic peak area vs concentration) of the tested phenolic antioxidants were obtained using UPLC-PDA-ESI–MS/MS system. Protocatechuic acid, syringic acid, rutin, hesperidin, ros-marinic acid, quercetin, naringenin, apigenin, carvacrol were identified in Thymbra spicata L. extract. The amounts Fig. 2 The 3D surface plots for the (A) TPC, (B) TAC, and (C) RSA of the Thymbra spicata L. extract as a function of solvent concentration and the solvent-to-solid ratio (Temperature 75°C and extraction time 5.5 min)
of individual phenolic antioxidants identified in Thymbra spicata L. extract by the UPLC-PDA-ESI–MS/MS system are summarized in Supplementary Table S4. Rosmarinic acid (1089.2 ± 10.9 mg/100 g-DS) and carvacrol (461.2 ± 4.5 mg/100 g-DS) were found to be the major phenolic antioxidants in the extract. On the other hand, protocatechuic acid (1.2 ± 0.1 mg/100 g-DS), rutin (1.3 ± 0.1 mg/100 g-DS), and quercetin (1.4 ± 0.1 mg/ 100 g-DS) were found to be the minor phenolic antioxi-dants. The characterized total phenolic content of Thymbra spicata L. was 1827 mg/100 g-DS. Stocker et al. (2004) found that the total phenolic content of Thymus serpyllum L., Ajuga iva L. and Teucrium polium L. extracts, members of the Lamiaceae family as 758, 250 and 700 mg/100 g, respectively by using HPLC system. Askun et al. (2009)
found that the total phenolic content of Salvia fruticosa Mill., Salvia tomentosa Mill., Sideritis albiflora Hub.-Mor., Sideritis leptoclada O. Schwarz & P.H. Davis, and Orig-anum onites L., members of the Lamiaceae family as 1714.4, 4199.5, 1465.2, 1562.8 and 5440.9 mg/100 g, respectively, by using HPLC system. As a result, it can be seen that Thymbra spicata L. as an important source of antioxidant stands out from the other members of the Lamiaceae family.
In this study, the optimum conditions for MAE that yield high antioxidant contents from Thymbra spicata L. were investigated using RSM. The basis of this study was the investigation of the effects of four important parameters on the process of extracting antioxidants. These factors included temperature, extraction time, solvent Fig. 3 The 3D surface plot for the (A) TPC, (B) TAC, and (C) RSA of the Thymbra spicata L. extract as a function of solvent concentration and time (Temperature 75°C and solvent-to-solid ratio 12.5 mL/0.2 g)
concentration (ethanol in water), and solvent-to-solid ratio. All the models calculated for the three responses (TPC, TAC, and RSA) were found to be significant (p \ 0.0001), and show a relationship between the response and the independent parameters. Quadratic models were used to predict all the responses. The solvent concentration was found to be the most significant operational factor amongst the other responses of MAE; on the other hand, extraction time was the least significant one. The optimal conditions were determined to base on individual responses. The results showed that theoretical and experimental values
were close to each other. The detailed phenolic character-ization of the Thymbra Spicata L. extract was then per-formed using UPLC-PDA-ESI–MS/MS system, and nine phenolic antioxidant compounds were identified and quantified. The major phenolic components in Thymbra spicata L. extract were rosmarinic acid (1089.2 ± 10.9 mg/100 g-DS) and carvacrol (461.2 ± 4.5 mg/100 g-DS). The MAE method, which has significant advantages over other extraction methods, was successfully used for the first time to extract antioxidants from Thymbra spicata L., and the process parameters optimized for maximum antioxidant Fig. 4 Total (A: ES-, B: ES?) and extracted (C) ion chromatograms of phenolic antioxidants in Thymbra spicata L. (1: Protocatechuic acid; 2: Syringic acid; 3: Rutin; 4: Hesperidin; 5: Rosmarinic acid; 6: Quercetin; 7: Naringenin; 8: Apigenin; 9: Carvacrol)
properties (TPC, TAC, and RSA). Finally, the simple, cost-effective, fast and highly efficient MAE method proposed in this study will be an important alternative method to extract antioxidants from the Thymbra spicata L. in the food and pharmaceutical industries.
Acknowledgements I thank Istanbul University-Cerrahpasa Appli-cation and Research Center for the Measurement of Food Antioxi-dants for sharing its research infrastructures.
Compliance with ethical standards
Conflict of interest The author states that there is no conflict of interest.
References
Ak T, Gu¨lc¸in, I˙. Antioxidant and radical scavenging properties of curcumin. Chem. Biol. Interact. 174: 27-37 (2008)
Akkol EK, Avcı G, Ku¨c¸u¨kkurt I, Keles¸ H, Tamer U, Ince S, Yesilada E. Cholesterol-reducer, antioxidant and liver-protective effects of Thymbra spicata L. var. spicata. J. Ethnopharmacol. 126: 314-319 (2009)
Apak R, Gu¨c¸lu¨ K, O¨ zyu¨rek M, Karademir SE. Novel total antioxidant capacity index for dietary polyphenols and vitamins C and E, using their cupric ion reducing capability in the presence of neocuproine: CUPRAC method. J. Agric. Food Chem. 52: 7970-7981 (2004)
Askun T, Tumen G, Satil F, Ates M. Characterization of the phenolic composition and antimicrobial activities of Turkish medicinal plants. Pharm. Biol. 47: 563-571 (2009)
Barani H, Maleki H. Plasma and ultrasonic process in the dyeing of wool fibres with madder in presence of Lecithin. J. Disper. Sci. Technol. 32: 1191-1199 (2011)
Bener M, O¨ zyu¨rek M, Gu¨c¸lu¨ K, Apak R. Optimization of microwave-assisted extraction of curcumin from Curcuma longa L. (Turmeric) and evaluation of antioxidant activity in multi-test systems. Rec. Nat. Prod. 10: 542-554 (2016)
Bener M, Shen Y, Apak R, Finley JW, Xu Z. Release and degradation of anthocyanins and phenolics from blueberry pomace during thermal acid hydrolysis and dry heating. J. Agric. Food Chem. 61: 6643-6649 (2013)
Bozkurt H. Utilization of natural antioxidants: Green tea extract and Thymbra spicata oil in Turkish dry-fermented sausage. Meat Sci. 73: 442-450 (2006)
Cacace JE, Mazza G. Mass transfer process during extraction of phenolic compounds from milled berries. J. Food Eng. 59: 379-389 (2003)
Camel V. Microwave-assisted solvent extraction of environmental samples. TrAC-Trends Anal. Chem. 19: 229-248 (2000) Chen Y, Xie MY, Gong XF. Microwave-assisted extraction used for
the isolation of total triterpenoid saponins from Ganoderma atrum. J. Food Eng. 81: 162-170 (2007)
Croft KD. The chemistry and biological effects of flavonoids and phenolic acids. Ann. NY Acad. Sci. 854: 435-442 (1998) C¸ elik SE, Tufan AN, Bekdes¸er B, O¨ zyu¨rek M, Gu¨c¸lu¨ K, Apak R.
Identification and determination of phenolics in Lamiaceae species by UPLC-DAD-ESI-MS/MS. J. Chromatogr. Sci. 55: 291-300 (2017)
Do QD, Angkawijaya AE, Tran-Nguyen PL, Huynh LH, Soetaredjo FE, Ismadji S, Ju YH. Effect of extraction solvent on total phenol
content, total flavonoid content, and antioxidant activity of Limnophila aromatica. J. Food Drug Anal. 22: 296-302 (2014) Dorman HD, Bachmayer O, Kosar M, Hiltunen R. Antioxidant properties of aqueous extracts from selected Lamiaceae species grown in Turkey. J. Agric. Food Chem. 52: 762-770 (2004) Fan G, Han Y, Gu Z, Chen D. Optimizing conditions for anthocyanins
extraction from purple sweet potato using response surface methodology (RSM). LWT-Food Sci. Technol. 41: 155-160 (2008)
Fleisher Z, Fleisher A. Extract analyses of Satureja thymbra L. and Thymbra spicata L. aromatic plants of the holy land and the Sinai Part XVII. J. Essent. Oil Res. 17: 32-35 (2005)
Gan C-Y, Latiff AA. Optimisation of the solvent extraction of bioactive compounds from Parkia speciosa pod using response surface methodology. Food Chem. 124: 1277-1283 (2011) Gumus T, Albayrak S, Sagdic O, Arici M. Effect of gamma
irradiation on total phenolic contents and antioxidant activities of Satureja hortensis, Thymus vulgaris, and Thymbra spicata from Turkey. Int. J. Food Prop. 14: 830-839 (2011)
Hancı S, Sahin S, Yılmaz L. Isolation of volatile oil from thyme (Thymbra spicata) by steam distillation. Nahrung 47: 252-255 (2003)
Hemwimon S, Pavasant P, Shotipruk A Microwave-assisted extrac-tion of antioxidative anthraquinones from roots of Morinda citrifolia. Sep. Purif. Technol. 54: 44-50 (2007)
I˙lbay Z, S¸ahin S, Kırbas¸lar S¸I˙. Optimisation of ultrasound-assisted extraction of rosehip (Rosa canina L.) with response surface methodology. J. Sci. Food Agric. 93: 2804-2809 (2013) Kan Y, Chen T, Wu Y, Wu J. Antioxidant activity of polysaccharide
extracted from Ganoderma lucidum using response surface methodology. Int. J. Biol. Macromol. 72: 151-157 (2015) Lee W, Yusof S, Hamid N, Baharin B. Optimizing conditions for hot
water extraction of banana juice using response surface methodology (RSM). J. Food Eng. 75: 473-479 (2006) Naczk M, Shahidi F. Phenolics in cereals, fruits and vegetables:
Occurrence, extraction and analysis. J. Pharmaceut. Biomed. 41: 1523-1542 (2006)
Ozel MZ, Gogus F, Lewis AC. Subcritical water extraction of essential oils from Thymbra spicata. Food Chem. 82: 381-386 (2003)
O¨ zyu¨rek M, Bener M, Gu¨c¸lu¨ K, Apak R. Antioxidant/antiradical properties of microwave-assisted extracts of three wild edible mushrooms. Food Chem. 157: 323-331 (2014)
Pinelo M, Rubilar M, Jerez M, Sineiro J, Nu´n˜ez MJ. Effect of solvent, temperature, and solvent-to-solid ratio on the total phenolic content and antiradical activity of extracts from different components of grape pomace. J. Agric. Food Chem. 53: 2111-2117 (2005)
Rafiee Z, Jafari SM, Alami M, Khomeiri M. Microwave-assisted extraction of phenolic compounds from olive leaves; a compar-ison with maceration. J. Anim. Plant Sci. 21: 738-745 (2011) Raman G, Gaikar VG. Microwave-assisted extraction of piperine
from Piper nigrum. Ind. Eng. Chem. Res. 41: 2521-2528 (2002) Sa´nchez-Moreno C, Larrauri JA, Saura-Calixto F. A procedure to measure the antiradical efficiency of polyphenols. J. Sci. Food Agric. 76: 270-276 (1998)
Shan B, Cai YZ, Sun M, Corke H. Antioxidant capacity of 26 spice extracts and characterization of their phenolic constituents. J. Agric. Food Chem. 53: 7749-7759 (2005)
Simic´ VM, Rajkovic´ KM, Stojicˇevic´ SS, Velicˇkovic´ DT, Nikolic´ NCˇ , Lazic´ ML, Karabegovic´ IT. Optimization of microwave-assisted extraction of total polyphenolic compounds from chokeberries by response surface methodology and artificial neural network. Sep. Purific. Technol. 160: 89-97 (2016)
Singleton VL, Orthofer R, Lamuela-Ravento´s RM. Analysis of total phenols and other oxidation substrates and antioxidants by
means of Folin-ciocalteu reagent. Methods Enzymol. 299: 152-178 (1999)
Sonsuzer S, Sahin S, Yilmaz L. Optimization of supercritical CO2 extraction of Thymbra spicata oil. J. Supercrit. Fluid. 30: 189-199 (2004)
Spigno G, Tramelli L, De Faveri DM. Effects of extraction time, temperature and solvent on concentration and antioxidant activity of grape marc phenolics. J. Food Eng. 81: 200-208 (2007)
Stocker P, Yousfi M, Djerridane O, Perrier J, Amziani R, El Boustani S, Moulin A. Effect of flavonoids from various Mediterranean plants on the enzymatic activity of intestinal carboxylesterase. Biochimie 86: 919-925 (2004)
Thompson LU. Antioxidants and hormone-mediated health benefits of whole grains. Crit. Rev. Food Sci. 34: 473-497 (1994) U¨ nlu¨ M, Vardar-U¨nlu¨ G, Vural N, Do¨nmez E, O¨zbas¸ ZY. Chemical
composition, antibacterial and antifungal activity of the essential oil of Thymbra spicata L. from Turkey. Nat. Prod. Res. 23: 572-579 (2009)
Wang L, Weller CL. Recent advances in extraction of nutraceuticals from plants. Trends Food Sci. Technol. 17: 300-312 (2006)
Wettasinghe M, Shahidi F. Evening primrose meal: a source of natural antioxidants and scavenger of hydrogen peroxide and oxygen-derived free radicals. J. Agric. Food Chem. 47: 1801-1812 (1999)
Yilmaz Y, Toledo RT. Oxygen radical absorbance capacities of grape/wine industry byproducts and the effect of solvent type on the extraction of grape seed polyphenols. J. Food Compos. Anal. 19: 41-48 (2006)
Yuan JF, Zhang ZQ, Fan ZC, Yang JX. Antioxidant effects and cytotoxicity of three purified polysaccharides from Ligusticum chuanxiong Hort, Carbohyd. Polym. 74: 822-827 (2008) Zigoneanu I, Williams L, Xu Z, Sabliov C. Determination of
antioxidant components in rice bran oil extracted by microwave-assisted method. Bioresour. Technol. 99: 4910-4918 (2008) Zuo Y, Zeng A, Yuan X, Yu K. Extraction of soybean isoflavones
from soybean meal with aqueous methanol modified supercrit-ical carbon dioxide. J. Food Eng. 89: 384-389 (2008)
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
