Optimization of ultrasound-assisted extraction of polyphenols from wheatgrass (Triticum aestivum L.)

Abstract

Conventional extraction techniques require high consumption of available resources and thus are ineffective and expensive, especially at an industrial scale. The aim of the study was to optimize the ultrasound-assisted extraction of polyphenols from fresh wheatgrass (Triticum aestivum L.). The effects of different extraction techniques and solvents were investigated on the yield of extractive substances and antioxidant activity. The ultrasound-assisted extraction technique and ethanol gave the highest yield of extractive substances so that they were used in the optimization studies. The central composite design was employed to find the optimal levels of ethanol concentration, extraction temperature, and extraction time. The total phenolic content was varied in the range of 10.50–15.50 grams of gallic acid equivalents per 100 g of dry weight of plant material (g GAE 100 g⁻¹ dw). The optimal conditions for ultrasound-assisted extraction were: (1) 56% (v/v) ethanol, (2) temperature of 59 °C, and (3) extraction time of 28 min. The results of ANOVA indicated that the highest impact had the extraction temperature on the total phenolic content. The toxic solvents were not used in the developed extraction procedure. The consumption of energy and raw plant material is estimated to be lower by at least 10% compared to conventional techniques.

Electronic supplementary material

The online version of this article (10.1007/s13197-020-04312-w) contains supplementary material, which is available to authorized users.

Introduction

Wheatgrass has become one of the most important dietary supplements because of the presence of polyphenols, amino acids, vitamins (A, B1, C, and E), minerals (iron, calcium, magnesium, iodine, selenium, and zinc), enzymes, and chlorophyll (about 70%) (Bar-Sela et al. 2015; Zendehbad et al. 2014). The fresh and dry wheatgrass can be found in many health food stores. In folk medicine, the wheatgrass extract is used for the treatment of colds, coughs, bronchitis, bacterial infections, chronic skin disorders and joint pain (Byers et al. 2002; Gruenwald et al. 2004). The anticancer (Gore et al. 2017; Shakya et al. 2017; Rajoria et al. 2017), antimicrobial (Kakkar and Dubey 2012), antioxidant (Tandon et al. 2011), and detoxification (Rimple et al. 2016) activities of the wheatgrass extracts are well known because of the presence of polyphenols, flavonoids, and other bioactive compounds (Chon et al. 2009). The wheatgrass extract can be applied as the main component for incorporation in the cosmetical products due to the beneficial effect on the skin, hair, and nails (Singh et al. 2012).

In the literature, several conventional extraction techniques (maceration, Soxhlet, and microwave-assisted extractions) were used for obtaining the wheatgrass extracts (Kakkar and Dubey 2012; Suriyavathana and Roopavathi 2016). In these studies, water, ethanol, methanol, ethyl acetate, n-hexane, chloroform, and petroleum ether were used as the solvents. The conventional extraction techniques require high consumption of available resources and thus are ineffective and expensive. The recent trends in extraction techniques tend to minimize the consumption of energy, toxic solvents, and raw plant material. One of the possibilities for overcoming this problem is the application of modern extraction techniques and mathematical regression models for the optimization of extraction procedures. Ultrasound-assisted extraction (UAE) as one of the modern extraction techniques can offer high reproducibility in the shorter extraction times, easy residual recovery from plant matrix in bound form, simplified manipulation, and reduced consumption of energy (Cui et al. 2019). UAE eliminates the application of toxic solvents while enhancing extraction efficiency. The improvement in the quality of the extract is also possible because thermodegradation of bioactive compounds is prevented due to low extraction temperatures during UAE.

Among the mathematical regression models, a central composite design (CCD) takes a significant place (Mondal and Purkait 2018), which enables process optimization with a lower number of experimental runs. Unlike the conventional technique, the interactions between the defined extraction factors can be estimated using the CCD (Ramandi et al. 2017). During optimization of the process, the CCD offers the global optimal conditions, while the conventional optimization method (one-variable-at-a-time method) can give the local optimal conditions (Jamshidi et al. 2016).

The aim of this study was to model the extraction of polyphenols from wheatgrass. The influence of extraction techniques (UAE and reflux extraction) and solvents polarity (distilled water, ethanol, methanol, and their combination) on the content of extractive substances and antioxidant activity of the extract were considered. The ethanolic solutions and UAE were chosen for further optimization study. Since the extraction process depends on the various factors (Savic et al. 2016), the influence of the ethanol concentration, extraction temperature, and extraction time on the total phenolic content (TPC) was investigated using the CCD. The UAE of polyphenols from wheatgrass using non-toxic solvent has a strong potential for industrial development as an eco-friendly process.

Materials and methods

Reagents and chemicals

Sodium carbonate, 96% (v/v) ethanol, 99.5% (v/v) methanol (Zorka Pharma, Sabac, Republic of Serbia), Folin-Ciocalteu’s reagent (AppliChem, Darmstadt, Germany), gallic acid (97.5%), 2,2-diphenyl-1-picrylhydrazine (DPPH), and butylhydroxytoluene (BHT) (Sigma Chemical, St. Louis, Missouri, USA) were used in this study. All other used chemicals were of analytical grade.

Plant materials

The wheat seeds (Triticum aestivum L.) were purchased in the local market in Leskovac (Republic of Serbia). The wheatgrass was grown in the ambient conditions for 9 days to the height of 15–16 cm. The procedure for the determination of the moisture content in wheatgrass was performed as follows: 2 g of chopped fresh plant material was dried for 2 h in a laboratory oven at 105 °C. The procedure was repeated each 30 min to the constant weight. The determined content of moisture in the plant material was 86.6% (w/w).

Study on extraction kinetics of extractive substances

The impact of solvents and extraction techniques on the content of extractive substances and antioxidant activity were investigated. The distilled water, 96% (v/v) ethanol, 50% (v/v) ethanol, methanol, and 50% (v/v) methanol were used as the solvents. The chopped fresh plant material (10 g) was extracted at the liquid-to-solid ratio of 10 cm³ g⁻¹ by reflux extraction and UAE for 5–80 min. The extractions were performed separately for each extraction time. The extraction temperature of 60 °C was maintained using a water bath at reflux extraction and ultrasonic bath (Sonic Nis, Republic of Serbia) at UAE (Dent et al. 2015). The operating frequency and power of the ultrasonic bath were 40 kHz and 150 W, respectively. The plant material was separated from the liquid by vacuum filtration and evaporated using a rotary vacuum evaporator at 50 °C. The extracts were dried to constant weight in a laboratory dryer at 40 °C. The content of extractive substances (Y) expressed as g per 100 g of dry weight was determined in each collected sample as follows (Eq. 1):

$$Y\left[g100{\text{g}}^{-1}\right]=\frac{m_1\times 100}{m_2}$$ 1

where m₁ is the weight of the extract residue obtained after solvent removal and m₂ is the weight of dry plant material.

The initial content of extractive substances in dry plant material for each used solvent was determined to define the extraction kinetics. The chopped fresh plant material (10 g) was transferred into the flask of 250 cm³ and treated with 100 cm³ of distilled water, 96% (v/v) ethanol, 50% (v/v) ethanol, methanol, or 50% (v/v) methanol for 4 h. The vacuum filtration was used to separate the plant material than liquid extract. The residual plant material was soaked in the fresh solvent (100 cm³) for 2 h, and also filtered under vacuum. The obtained liquid extracts were collected and then evaporated using rotary vacuum evaporator at 50 °C to constant weight.

The kinetics of the extraction process of bioactive substances were modeled using the Ponomaryov (Popov et al. 2016; Paunović et al. 2015) and non-stationary diffusion models (Samadi et al. 2017). The Ponomary and non-stationary diffusion models are depicted in Eqs. 2 and 3, respectively.

$$\frac{q_o-q_i}{q_o}=b+kt$$ 2

$$\frac{q}{q_o}=\left(1-b^{\prime }\right)+k^{\prime }t$$ 3

where q_o is the initial content of extractive substances in the dry plant; q is the content of extractive substances for defined time; b is the washing coefficient in the Ponomary model; b′ is the washing coefficient in the non-stationary diffusion model; k is the specific rate of slow extraction in the Ponomary model; k′ is the slow extraction coefficient in the non-stationary diffusion model. The equation of non-stationary diffusion model (Eq. 3) can be converted to a linear function using logarithm transformation.

Central composite design

Three factors at five levels (− 1.68, − 1, 0, + 1, + 1.68) were analyzed to investigate their impact on the TPC. The ethanol concentration (X₁, %), extraction temperature (X₂, °C), and extraction time (X₃, min) were used as the factors, while TPC was defined as the response. The levels of factors in the form of coded and uncoded units are presented in Table 1.

Table 1.

Uncoded and coded levels of the factors for extraction from wheatgrass

Factors	Coded	Levels
		− 1.68	− 1	0	+ 1	+ 1.68
Ethanol concentration (%, v/v)	X1	33	40	50	60	67
Extraction temperature (°C)	X2	33	40	50	60	67
Extraction time (min)	X3	17	20	25	30	33

In order to model the extraction procedure using this approach, it was necessary to perform 18 experiments. The repetition of the center point was performed to obtain the statistical parameters of the proposed model. The TPC in the extracts were fitted using a second-order polynomial equation (Eq. 4):

$$Y=a_o+a_1{x}_1+a_2{x}_2+a_3{x}_3+a_{11}{x}_1^2+a_{22}{x}_2^2+a_{33}{x}_3^2+a_{12}{x}_1{x}_2+a_{13}{x}_1{x}_3+a_{23}{x}_2{x}_3+\epsilon$$ 4

where are x₁, x₂, x₃—factors; a₀—intercept; a₁, a₂, a₃, a₁₁, a₂₂, a₃₃, a₁₂, a₁₃, a₂₃—regression coefficients, Y—response, and ε—residual.

Statistical analysis

Statistical analysis was performed using the software packages Design Expert 11.0.3.0 (Stat-Ease, Minneapolis, Minnesota, USA). The model significance was estimated based on the analysis of variance (ANOVA) with a 95% confidence level. In statistics, the sum of squares (SS), degree of freedom (df), mean squares (MS), F-values, and p values are commonly used to describe the model (Popov et al. 2016). Based on p value (Prob > F), the statistical significance of the equation terms was confirmed. If this value less than 0.0500, the model terms can be considered as statistically significant. The coefficient of determination (R²), adjusted correlation coefficient (R²_adj), and predicted correlation coefficient (R²_pred) were used to define the goodness of fitting.

Optimization

The extraction procedure was optimized by maximizing TPC using numerical optimization method. The method was based on the development of desirability function for the TPC by adjusting the weighted factor at 1 and the importance of goal at 3.

Determination of total phenolic content

Spectrophotometrically method with Folin–Ciocalteu’s reagent was used to determine the TPC in the plant extracts (Singleton et al. 1999). A series of solutions for determination of gallic acid was prepared in the concentration range of 5–50 μg cm⁻³ from the stock solution (1 mg cm⁻³ prepared in 96% (v/v) ethanol). The samples were obtained by mixing 0.25 cm³ of the standard solution/plant extract, 2.25 cm³ redistilled water, 0.25 cm³ Folin–Ciocalteu’s reagent, and 2.5 cm³ sodium carbonate (7%, w/w). The absorbance was measured at 740 nm after incubation of 90 min. The scanning was performed on the Varian Cary 100 spectrophotometer (Mulgrave, Victoria, Australia) in the quartz cuvettes (1 × 1 cm) at room temperature (22°C). The TPC was expressed as grams of gallic acid equivalents per 100 g of dry weight (g GAE 100 g⁻¹ dw).

DPPH assay

DPPH assay was employed to determine the antioxidant activity of the wheatgrass extracts (Molyneux 2004). The stock solutions of the samples were diluted and a series of different concentrations of the extracts were prepared. The methanolic solution of DPPH radicals (3 × 10⁻⁴ mol dm⁻³) was added per 1–2.5 cm³ of each sample. The negative control was prepared by the addition of 1 cm³ of DPPH solution to 2.5 cm³ of methanol. The methanolic solution of synthetic antioxidant BHT was used as a positive control. The stock solution of BHT (1.0 mg cm⁻³) was diluted and treated with DPPH solution in the same way as the negative control. The samples were incubated for 30 min at room temperature (22°C) in a dark. The absorbance was measured at 517 nm in relation to methanol. The inhibition of DPPH radicals (I_DPPH) expressed in the percentage was calculated according to Eq. 5:

$$I_{\mathit{DPPH}}\left[\%\right]=\frac{A_C-A_S}{A_C}\times 100$$ 5

where A_S—the absorbance of the sample after addition of DPPH radicals and incubation, A_C—the absorbance of negative control.

The antioxidant activity was estimated based on the IC₅₀ value, which was obtained by the interpolation.

Results and discussion

Kinetics of reflux and ultrasound-assisted extraction

The yield of extractive substances, extract composition, and its pharmacological activity depend on the extraction technique, the nature of the solvent, and its polarity (Moller et al. 1999). Because of these reasons, the conventional (reflux) and advanced (UAE) extraction techniques were used for the extraction of bioactive compounds from wheatgrass. Since the increase of solvent’s polarity impacts the yield of extractive substances, the different solvents were used to obtain the extracts with favorable characteristics regarding the composition and content of pharmacologically important compounds. The extraction time and extraction temperature are the parameters that have a significant effect on the extraction of desirable compounds and the reduction of energy costs (d’Alessandro et al. 2012). The initial contents of extractive substances obtained for different solvents are given in Table 2. The contents were decreased according to the following order: 50% (v/v) ethanol > water > 50% (v/v) methanol > methanol > 96% (v/v) ethanol. The variations in the initial content of extractive substances are slight for 50% (v/v) ethanol, water, and 50% (v/v) methanol.

Table 2.

The initial content of extractive substances

Solvent	Content of extractive substances (g 100 g−1 dw)
Water	93.7
96% (v/v) ethanol	84.0
50% (v/v) ethanol	94.4
Methanol	87.7
50% (v/v) methanol	92.4

The extraction process requires the knowledge of velocity extraction, yield of extractive substances and extraction time. The extraction of extractive substances from wheatgrass for different solvents and two extraction techniques (reflux extraction and UAE) was modeled using Ponomaryov and non-stationary diffusion models through the plant material (Supplementary Figure S.1. and Supplementary Figure S.2.).

In these extraction kinetic plots, the two characteristic periods (fast and slow) of extraction can be noticed. The fast extraction period which refers to washing is the result of the dissolution of extractive substances from the surface of the destroyed plant cells. The slow extraction period is due to the diffusion of extractive substances from the plant cells. The fast extraction period (FEP) and extraction efficacy (EE) are given in Table 3. The FEP was 25 min for reflux extraction, while the lowest EE of 62.3% and the highest EE of 75.5% were obtained using 50% (v/v) ethanol and methanol, respectively. Using the UAE, the lowest EE of 79.2% (for 20 min) and the highest EE of 95.2% (for 30 min) was achieved after treatment of plant material with 50% (v/v) methanol and water, respectively. The kinetic parameters, the coefficients of fast (b) and slow (k) extractions, calculated based on the experimental data obtained by linear regression of Ponomaryov and non-stationary diffusion models are also depicted in Table 3. These parameters are depended on the applied kinetic models, solvent type and extraction techniques. The values of k according to the non-stationary diffusion model were higher in comparison with Ponomaryov model, while the values of b were almost the same. The b coefficients had higher values for UAE compared with reflux extraction. These results are in accordance with the impact of the collapse phenomenon of cavitation bubbles on the mass transfer in the washing phase and period of slow extraction.

Table 3.

The values of kinetic parameters according to Ponomaryov and non-stationary diffusion models

Solvent	FEP (min)	EE (%)	Ponomaryov model		Non-stationary diffusion model
			b	k × 103 (min−1)	b	k × 103 (min−1)
Reflux extraction
Water	25	70.0	0.690	0.527	0.689	1.87
96% (v/v) ethanol	25	72.4	0.703	0.755	0.700	2.95
50% (v/v) ethanol	25	62.3	0.606	0.746	0.604	2.10
Methanol	25	75.5	0.738	0.772	0.735	3.48
50% (v/v) methanol	25	70.7	0.692	0.794	0.690	2.97
Ultrasound-assisted extraction
Water	30	95.2	0.940	0.476	0.929	14.15
96% (v/v) ethanol	30	88.5	0.861	0.797	0.850	8.57
50% (v/v) ethanol	30	83.0	0.806	0.863	0.799	5.90
Methanol	25	87.8	0.857	1.200	0.831	15.44
50% (v/v) methanol	20	79.2	0.778	1.090	0.771	6.55

The UAE has a greater effect on the content of extractive substances compared to the other conventional extraction techniques (Vega et al. 2017). The higher extraction efficiency using the UAE was due to the effect of cavitation (Vinatoru et al. 2017). This extraction procedure leads to the better degradation of plant tissue so that the extraction of desirable bioactive compounds from the plant material is easier (Kiani and Sun 2018). The penetration of the solvent through the healthy cells is better, which reflects the increase in the mass transfer. Having this in mind, the UAE can be considered as the technique of choice for extraction of bioactive compounds from wheatgrass. The content of extractive substances was not significantly changed (about 4%) using reflux extraction and UAE for extraction time in the range of 30–80 min. Because of these reasons, the effect of extraction time up to 30 min was considered for further optimization studies using the central composite design.

The antioxidant activity of the extracts

The antioxidant activity of wheatgrass extracts obtained using reflux extraction and UAE was determined by DPPH assay. The antioxidant activity of synthetic antioxidant BHT was also determined to compare with the activity of the extracts. The inhibition of 90% was possible to achieve using the extract concentration of 2 mg cm⁻³. BHT had the ability to inhibit 80% of the free DPPH radicals (Supplementary Figure S.3.).

The lower IC₅₀ indicated the better ability to inhibit the free radicals and the better antioxidant activity of investigated extracts (Supplementary Figure S.4.). For reflux extraction, the ethanolic extract had the highest antioxidant activity, while the activity of the extracts obtained using 50% (v/v) ethanol and 50% (v/v) methanol were significantly lower. In the case of UAE, the extract obtained using 50% (v/v) ethanol had the highest antioxidant capacity, while the aqueous extract had the lowest antioxidant activity. By comparing both extraction techniques and used solvents, it can be noticed that the extract obtained by UAE with 50% (v/v) ethanol had the highest antioxidant activity (IC₅₀ = 0.51 mg cm⁻³). The extract with similar activity (IC₅₀ = 0.49 mg cm⁻³) can be obtained with 96% (v/v) ethanol using reflux extraction.

The results obtained in this study for the determination of antioxidant activity are in accordance with the literature data (Zendehbad et al. 2014; Tandon et al. 2011; Kulkarni et al. 2006; Devi et al. 2017). Some studies investigated the influence of solvent polarity on the antioxidant activity of wheatgrass. Tandon et al. (2011) showed that the ethanolic extract of wheatgrass prepared by maceration had the best antioxidant activity (IC₅₀ = 177.7 µg cm⁻³), while the aqueous extract showed lower activity (IC₅₀ = 646.1 µg cm⁻³). Unlike the antioxidant activity of the aqueous extract, the antioxidant activity of ethanolic extract of wheatgrass depends on the growth period and cultivation conditions of the plant (Kulkarni et al. 2006). The methanolic extract had better antioxidant activity determined by DPPH assay compared with the chloroform extract of wheatgrass (Zendehbad et al. 2014). It was reported that the fresh plant material had a better antioxidant activity for 50% compared with the dry plant material. Devi et al. (2017) reported that the wheatgrass cultivated in interior conditions had a higher antioxidant activity than the plant cultivated under external conditions. These literature results were hard to compare with the results in our study because the antioxidant activity was determined using different assays and the IC₅₀ values were expressed using different units.

Modeling the extraction of polyphenols from wheatgrass

In the screening study, the impact of extraction techniques and solvent polarity on the content of extractive substances and antioxidant activity were investigated. The UAE and diluted ethanol were presented as the suitable extraction parameters for extraction of bioactive compounds from wheatgrass. The extraction time was analyzed slightly higher than 30 min in order to provide saturation in the extraction of desired bioactive compounds. Since the extraction is affected by the great number of parameters, the ethanol concentration (33–67% (v/v)), extraction temperature (33–67 °C), and extraction time (17–33 min) were varied in the optimization study. The CCD was used to obtain the optimal conditions for the UAE of polyphenolic compounds from wheatgrass. The extractions were performed randomly according to the matrix of CCD depicted in Table 4. The response (TPC) was varied in the range of 10.50 to 15.50 g GAE 100 g⁻¹ dw and fitted using a second-order polynomial model.

Table 4.

The matrix of a central composite design for three factors with the total phenolic content in the wheatgrass extracts

Standard order	Run order	X1 (%, v/v)	X2 (°C)	X3 (min)	TPC (g GAE 100 g−1 dw)
8	1	60 (+  1)	60 (+ 1)	30 (+ 1)	15.50
12	2	50 (0)	67 (+ 1.68)	25 (0)	14.98
3	3	40 (− 1)	60 (+ 1)	20 (− 1)	13.20
14	4	50 (0)	50 (0)	33 (+ 1.68)	15.06
18	5	50 (0)	50 (0)	25 (0)	14.54
5	6	40 (− 1)	40 (− 1)	30 (+ 1)	12.46
1	7	40 (− 1)	40 (− 1)	20 (− 1)	10.50
15	8	50 (0)	50 (0)	25 (0)	14.62
9	9	33 (− 1.68)	50 (0)	25 (0)	12.04
6	10	60 (+ 1)	40 (− 1)	30 (+ 1)	14.21
16	11	50 (0)	50 (0)	25 (0)	14.60
4	12	60 (+ 1)	60 (+ 1)	20 (− 1)	15.21
17	13	50 (0)	50 (0)	25 (0)	14.70
2	14	60 (+ 1)	40 (− 1)	20 (− 1)	13.00
13	15	50 (0)	50 (+ 1)	17 (− 1.68)	13.04
11	16	50 (0)	33 (− 1.68)	25 (0)	11.50
10	17	67 (+ 1.68)	50 (0)	25 (0)	15.01
7	18	40 (− 1)	60 (+ 1)	30 (+ 1)	14.33

X₁, ethanol concentration; X₂, extraction temperature; X₃, extraction time

The results of ANOVA for the proposed model is given in Table 5. The linear terms (X₁, X₂, X₃), quadratic terms (X²₁, X²₂, X²₃), and interactions between factors (X₁X₂, X₁X₃, X₂X₃) were presented as a statistically significant at 95% confidence level. The model F-value of 1166.48 was higher than the critical value of 3.39 and implied that the model was statistically significant. The F-value of lack-of-fit of 0.73 was lower than the critical value of 9.01 and implied that the lack-of-fit was insignificant relative to the pure error (0.0138). There was a 73.32% chance that the F-value of lack-of-fit this large could occur due to noise. These results meant that the model was adequate for prediction of TPC.

Table 5.

Analysis of variance (ANOVA) of quadratic response surface model

Source	SS	df	MS	F-value	p value Prob > F
Model	35.07	9	3.90	1166.48	< 0.0001*
X1	11.31	1	11.31	3386.02	< 0.0001*
X2	14.20	1	14.20	4251.29	< 0.0001*
X3	4.67	1	4.67	1397.34	< 0.0001*
X1X2	0.1440	1	0.1440	43.11	0.0002*
X1X3	0.3147	1	0.3147	94.21	< 0.0001*
X2X3	0.3814	1	0.3814	114.17	< 0.0001*
X21	1.86	1	1.86	556.78	< 0.0001*
X22	2.97	1	2.97	887.85	< 0.0001*
X23	0.4950	1	0.4950	148.20	< 0.0001*
Residual	0.0267	8	0.0033
Lack-of-fit	0.0129	5	0.0026	0.5610	0.7332**
Pure error	0.0138	3	0.0046
Corrected total	35.09	17
Std. Dev.	0.0578		R2		0.999
Mean	13.81		R2adj		0.998
C.V. %	0.4186		R2pred		0.997
			Adequate precision		116.7

*Significant

**Insignificant

The R² value of 0.999 for the proposed model was adequate and indicated that 99.9% of the variation in the yield of polyphenols could be explained by the regression model (Table 5). The R²_pred of 0.997 was in reasonable agreement with the R²_adj of 0.998. The signal to noise ratio was estimated based on adequate precision of 116.7. Since this value was higher than 4, it implies that the proposed model could be applied to navigate the design space.

The extraction of polyphenols from wheatgrass can be described using the following second-order polynomial equation in terms of coded factors (Eq. 6):

$$Y=14.61+0.91·X_1+1.02·X_2+0.58·X_3-0.13·X_1·X_2-0.20·X_1·X_3-0.22·X_2·X_3-0.38·X_1^2-0.48·X_2^2-0.20X_3^2$$ 6

The terms that have a positive sign in the given equation effect on the increase of TPC in the plant extract. The influence of the linear terms was decreased in the following order: extraction temperature (X₂), ethanol concentration (X₁), and extraction time (X₃). After the linear terms, the square of extraction temperature (X²₂) had the highest impact on the response. The square of extraction time (X²₃), as well as the interaction between ethanol concentration and extraction time (X₁X₃), and the interaction between extraction temperature and extraction time (X₂X₃) had almost the same effect on the TPC. As can be seen, the lowest impact on the TPC had the interaction between ethanol concentration and extraction temperature.

The normal probability plot of studentized residuals indicated that the residuals were normally distributed because the plotted points were close to the straight line (Supplementary Figure S.5a). Since the Cook’s distance were less than the limit of 1.0, it can be concluded that there were no outliers in the given dataset (Supplementary Figure S.5b). Based on this value, it can be estimated how the regression changes if the case is deleted.

The relationship between factors and response was estimated by analysis of the three-dimensional plots. The effect of ethanol concentration and extraction temperature for extraction time of 25 min is depicted in Fig. 1a. The influence of ethanol concentration was more pronounced at higher temperature levels (> 50 °C). It was expected because the solubility of polyphenols is increased with increasing the ethanol concentrations (He et al. 2016). A similar impact was noticed in the case with increasing extraction temperature because the solubility of polyphenols is better at the higher extraction temperatures (Ghafoor et al. 2009). The interaction between the ethanol concentration and extraction time at the extraction temperature of 50 °C is presented in Fig. 1b. The effect of ethanol concentration can be considered as significant for the shorter extraction times. The increase of ethanol concentration leads to higher TPC in the extracts. Based on the values of regression coefficients in the polynomial equation, it can be concluded that the influence of extraction time on the TPC was lower almost two times compared with ethanol concentration. This fact can be also confirmed based on the shape of the response surface. This effect was particularly expressed at ethanol concentrations higher than 50% (v/v) due to the increase of TPC in the extracts. The interaction between extraction temperature and extraction time for 50% (v/v) ethanol is presented in Fig. 1c. For longer extraction times, a significant effect on the TPC had extraction temperature. The increase of extraction time led to an increase of the TPC, but more significantly at the higher levels of extraction temperature. The detailed analysis of the obtained response surfaces was showed that there were not strong interactions between the analyzed factors.

Fig. 1.

The effects of a ethanol concentration and extraction temperature for 25 min; b ethanol concentration and extraction time at 50 °C; c extraction temperature and extraction time for 50% (v/v) ethanol on the TPC

Optimization of the extraction procedure

The optimal conditions for extraction of polyphenols were obtained using the numerical optimization method. The desirability function was ranged from 0 outside of the limits to 1 at the goal. During optimization, the factors were ranged between the levels − 1 and + 1 corresponding to the limits of factorial design. The optimal conditions for UAE were achieved for 56% (v/v) ethanol, the extraction temperature of 59 °C and an extraction time of 28 min at the liquid-to-solid ratio of 10 cm³ g⁻¹. The TPC in the optimal extract was found to be 15.51 g GAE 100 g⁻¹ dw (2.08 g GAE 100 g⁻¹ fresh plant material), while the predicted value by the proposed regression model was 15.56 g GAE 100 g⁻¹ dw. Based on the obtained values for the TPC of the extract obtained under optimal conditions, it can be calculated that the model error was 0.32%. The low model error is one more confirmation that the prediction ability of the proposed model is adequate. This procedure for extraction of polyphenols from wheatgrass enables to obtain the plant extract, which can be applied in the cosmetic, pharmaceutical, and food industry due to the use of eco-friendly and non-toxic solvents.

Durairaj et al. (2014) prepared the aqueous extract of wheatgrass by Soxhlet extraction for 24 h and determined the TPC of 210.15 µmol GAE g⁻¹ of plant material. Tandon et al. (2011) determined the TPC in the range of 2.44–6.48 mg GAE g⁻¹ of dry extract obtained using water, 70% (v/v) ethanol, 0.1% TCA, sodium acetate buffer pH 5, and potassium phosphate buffer pH 7.4. They obtained that the ethanolic extract had the highest TPC (6.48 mg GAE g⁻¹ of dry extract) compared to other obtained extracts. The TPC of aqueous extract (3.08 mg GAE g⁻¹ of dry extract) was almost two-fold lower in concerning the ethanolic extract. Kulkarni et al. (2006) prepared the ethanolic and aqueous extracts of wheatgrass grown under different conditions for 6, 7, 8, 10 and 15 days. The TPC increased proportionally with the time of wheatgrass cultivation and was the highest for 15 days. The TPC of 0.699 mmol GAE 100 g⁻¹ of fresh wheatgrass was obtained using ethanol, while the TPC of 0.331 mmol GAE 100 g⁻¹ of fresh wheatgrass was obtained using water. Zendehbad et al. (2014) reported that the TPC of methanolic and chloroform extracts of wheatgrass were 44% and 12%, respectively. Niroula et al. (2019) treated wheatgrass using methanol by maceration and defined the TPC of 1290.51 mg GAE 100 g⁻¹ of dw. These literature data indicates that the TPC of wheatgrass can be varied depending on the cultivation conditions, used solvents, and used extraction techniques. It can be concluded that the TPC in our study is manifold higher with the results obtained by Kulkarni et al. (2006) and Niroula et al. (2019), but almost two-fold lower compared with Durairaj et al. (2014). Despite this fact, the proposed procedure in our study is acceptable from an economic point of view, because extraction time is much shorter than Soxhlet extraction (24 h).

Cost and scaling-up approach

The reduced cost in terms of time and energy gives priority to the proposed UAE technique. The obtained results demonstrated that the potential use of ultrasounds is promising for extraction even at an industrial scale. UAE can be performed to treat much larger quantities of wheatgrass by using existing large-scale ultrasound reactors which are already in use in the chemical industry and could be easily modified and used for UAE. Among other advanced extraction techniques (e.g., supercritical fluid extraction, pressurized solvent extraction, or accelerated solvent extraction), the application of UAE has increased in recent years. The reasons for that are the existence of several disadvantages of conventional and other novel extraction techniques, such as high energy use, high capital investment, high CO₂ rejection, and consumption of toxic solvents.

Conclusion

In pharmaceutical production, solvents are used in large amounts per mass of final products. As such they define a significant part of the environmental performance of an extraction process and at the same time have a big impact on cost, safety and health issues, especially when toxic solvents, such as chloroform or petroleum ether are used. In this study, a UAE technique that uses ethanol as an environmentally preferable solvent was developed and optimized. The effects of different extraction techniques and solvents on the content of extractive substances, extraction kinetics, and antioxidant activity of wheatgrass were investigated. Based on these results, it was concluded that the UAE and diluted ethanol represent the extraction parameters that can be used for further optimization study using the CCD. The extraction kinetics models indicated that the extraction time can be analyzed within 30 min because after that time the saturation of extracted bioactive compounds was achieved. In the optimization study, the CCD with three factors was successfully employed to model the UAE of polyphenols. The ANOVA results showed that the ethanol concentration, extraction temperature, and extraction time had significant positive effects on the polyphenols extraction. The optimal conditions for extraction of polyphenols were achieved for 56% (v/v) ethanol, the extraction temperature of 59 °C, and extraction time of 28 min at the liquid-to-solid ratio of 10 cm³ g⁻¹. Under the optimal conditions, the TPC was found to be 15.51 g GAE 100 g⁻¹ dw. The extraction temperature and extraction time were optimized to maximize phenolic content and thus reduce consumption of both energy and raw plant material that is of crucial importance for scale-up of the extraction process.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

DPPH

2,2-diphenyl-1-picrylhydrazine

BHT

Butylhydroxytoluene

UAE

Ultrasound-assisted extraction

CCD

Central composite design

TPC

Total phenolic content

FEP

Fast extraction period

EE

Extraction efficiency

ANOVA

Analysis of variance

GAE

Gallic acid equivalents

dw

Dry weight of plant material

df

Degree of freedom

SS

Sum of squares

MS

Mean squares

Std. dev.

Standard deviation

Compliance with ethical standard

Conflict of interest

The authors declare that they have no conflict of interest.