Synchronous extraction and separation of corn silk bioactive substance and the prediction of partition efficiency

2.1. Materials

UCON (average M_n ∼3,900) was obtained from Sigma-Aldrich Co., Ltd. (Shanghai, China). Sulfuric acid (GR, 95.0∼98.0%), acetonitrile (HPLC grade), potassium carbonate (GR, 99.0%), sodium hydroxide (GR, 99.0%), potassium hydrogen phosphate (AR, 99.0%), aluminum chloride (AR, 99.0%), ammonium hydrogen phosphate (AR, 99.0%), ammonium dihydrogen phosphate (GR, 99.5%), sodium nitrite (AR, 99.0%), potassium oxalate (GR, 99.5%), and potassium tartrate (AR, 99.0%) were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Rutin standard samples were obtained from the Chinese National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China). Analytical-grade phenol and glucose anhydrase were purchased from Shanghai Macklin Biochemical Co., Ltd. (Shanghai, China). 1,1-diphenyl-2-picrylhydrazyl (DPPH, 97%), Tris-HCL (pH 8.0), FeSO₄ (0.1 mol/L), H₂O₂ (30 wt.% in H₂O), salicylic acid (AR, 99.5%), and pyrogallol (AR) were obtained from Aladdin Reagent Co., Ltd. (Shanghai, China).

2.2. Apparatus

The ultrasonic cell disruption instrument used for sample pre-treatment was provided by Shanghai Hanuo Instrument Co., Ltd. (HN99-IID, China). An ultraviolet-visible (UV-Vis) spectrophotometer was purchased from SHIMADZU Co., Ltd. (UV2500, Japan). The thermostatic water bath used to control the ATPS temperature was purchased from Zhuocheng Instrument Technology Co., Ltd. (SYP, China). The vacuum freezing dryer used to dry the corn silk was purchased from Shanghai Tianfeng Industrial Co., Ltd. (TF-FD-27, China). The centrifuge was provided by Xiangtan Xiangyi Instrument Co., Ltd. (Cence-H1650, Xiangtan, China). The analytical balance used for weighing was purchased from Mettler Toledo Technology (China) Co., Ltd. (Balance XPR106DUHQ/AC, China), and its uncertainty was $\pm 1.0\times {10}^{-7}$ kg.

2.3. Preparation of corn silk samples

From late July to early August, mature corn silks were handpicked from a pollution-free farm in Siping. A yellow corn variety was used. The corn silks were rinsed and dried under shade for 72 hrs, and the temperature was controlled at 20 ± 2°C. The corn silks were placed in a grinder to obtain the sample powder. In a beaker, 50 mL of a 10% UCON solution and 50 g of sample powder were added and completely mixed. The corn silk UCON solution was subjected to wall-breaking treatment using an ultrasonic cell disruption instrument. The corn silk UCON solution was placed into separate plastic centrifuge tubes after 30 mins of cell disruption. The centrifugal time and rotation speed were 10 mins and 1200 rpm, respectively. Subsequently 30 mL of a 10% UCON solution was added to the residue after the supernatant fluid was collected. The residue was completely mixed, and wall-breaking treatments were performed for 30 mins. The residue mixture was centrifuged at a rotation speed of 1200 rpm for 10 mins. The supernatant fluid was collected. The two supernatants were mixed together and filtered through a 0.45 μm membrane. The filtrate was a corn silk extract solution, denoted as the UCON-based corn silk extract. This extract solution was moved into brown bottles, and was placed in the refrigerator at 4°C.

2.4. Procedure

The UCON solution (0.3 g·mL^-1), UCON-based corn silk extract, water, and potassium tartrate (0.25 g·mL^-1) were added into a colorimetric tube with a certain ratio. The total volume was 10 mL. The colorimetric tubes were in a three-min oscillation and then placed in a thermostatic water bath for 15 mins. At this point, there were two phases in the colorimetric tubes. The volumes of the two phases were recorded, and then they were separated. The concentrations of flavonoids and polysaccharides in the two phases and UCON-based corn silk extract were determined.

The UCON-rich phase was then diluted by adding distilled water. The thermostat water bath which was maintained at 65°C temperature was used to control the formation of the two phases. The top phase was isolated to determine flavonoid concentration when the solution formed two new phases. The extraction efficiency and purity of flavonoids were determined [51]. The recovery efficiency of UCON was also determined.

The salt-rich phase was then transferred to a dialysis tube (8000-14,000 Da). Small molecular impurities, including salts, were removed by dialysing in distilled water. After 24 hrs, the dialyzed solution was concentrated using rotary evaporation. According to the ratio of five-time volume, 90% ethanol was added to the concentrated solution. The solution was well blended and placed for 12 hrs at 4°C to precipitate polysaccharides. Sediments were collected by centrifugation. The collected sediment was dissolved in water. The polysaccharides were resedimented in 80% ethanol, and the sediment was obtained by centrifugation. Anhydrous ethanol was used to wash the sediment, which was then freeze-dried to obtain corn silk polysaccharide. The purity of the corn silk polysaccharide was determined [16].

In our previous study [52], the phenol-sulfuric acid method was used to determine polysaccharide concentration where glucose anhydrase was used as the standard. This method was used for the determination of corn silk polysaccharides. Flavonoid concentration was measured using the reported method [53]. Rutin was taken as the standard. In the purity and antioxidant experiments, the flavonoid concentration was measured by high performance liquid chromatography with ultraviolet detection (HPLC-UV). The chromatographic separation was carried out on an analytical reversed-phase column (4.6 × 250 mm, 5 μm, Agilent TC-C18 column). 10 µL of sample solution was injected. The column temperature was at 25°C. The mobile phase comprised methanol (A) and 1.0 % acetic acid solution (B). The ratio of A and B was as following: 3:7 (0-8 min), 3:7-5:5 (8-13 min), 5:5 (13-20 min), 5:5-6:5 (20-25 min), 6:5-3:4 (25-35 min), 3:4-1:0 (35-45 min), 1:0- 3:7 (45-50 min), 3:7 (50-60.0 min). The flow rate was 1.0 mL·min^-1. UV detection wavelength was 280-350 nm [54]. The entire experimental procedure was shown in Figure 1.

Close

Figure 1.

Technology roadmap.

Export to PPT

2.5. Determination of parameters

The partition efficiency (PE_n) was used to measure the separation performance of ATPE for the flavonoids (n=1) and corn silk polysaccharides (n=2), and was defined by the following Eq. (1):

where $C_n$ and V_n are the flavonoid concentration and volume of the upper phase respectively, when n = 1; C_n and V_n are the corn silk polysaccharide concentration and volume of the bottom phase, respectively, when n = 2. C_s and V_s are the concentrations of flavonoids or corn silk polysaccharides in the initial system and the volume of the initial system, respectively; n=1 represents flavonoids, and n=2 represents corn silk polysaccharides.

The extraction efficiency (EE_n) represents the ratio of the extracted materials to corn silk and is defined by the following Eq. (2):

where m represents the mass of materials, n=1 represents flavonoids, n=2 represents corn silk polysaccharides, and t represents corn silk.

The purity of the flavonoids and corn silk polysaccharides was defined using the following Eq. (3):

where C and V represent the concentration and volume of materials in the purity test, respectively, m represents the total mass, n=1 represents flavonoids, and n=2 represents corn silk polysaccharide.

The proposed ANN and RSM models were evaluated using the mean relative percent deviation (MRPD) expressed as follows in Eq. (4):

where ‘n’ and ‘i’ were the total amount of data and the data number in the dataset, respectively, y was the response, ‘c’ was the computational data, and ‘e’ was the experimental data.

2.6. Antioxidant activity

The antioxidant activities of the flavonoids and corn silk polysaccharides were studied. The 2,2-Diphenyl-1-picrylhydrazyl (DPPH) free radical scavenging activities of the flavonoids and corn silk polysaccharides were evaluated using the reported method [17]. 2.5 mL DPPH solution was added to 1.0 mL of flavonoids or corn silk polysaccharide solution (0.025-0.175 mg·mL^-1). They were then completely mixed. Absorbance (A_i) was recorded at 517 nm after 30 mins. The absorbance of the reaction without the sample was denoted as A_c, and the absorbance of anhydrous ethanol was denoted as A_b. The clearance rate was calculated as follows in Eq. (5):

The superoxide radical scavenging abilities of the flavonoids and corn silk polysaccharides were determined using the reported method [55]. 1.0 mL of flavonoids or corn silk polysaccharide solution (0.025-0.175 mg·mL^-1) was added to seven tubes, which included 3.0 mL ultrapure water and 4.2 mL Tris-HCL (50 mmol·L^-1, pH 8.0). The mixture was shaken and then placed at 25°C for 30 mins. The mixture was shaken after adding 0.3 mL pyrogallol (3 mmol·L^-1). The absorbance was measured at a wavelength of 325 nm every 30 seconds, resulting in a total of 10 measurements. The rate of absorbance change (F_x) with time was calculated. The change rate (F₀) was obtained without the sample. The clearance rate was calculated as follows in Eq. (6):

The hydroxyl radicals inhibitory activities of the flavonoids and corn silk polysaccharides were quantified using the reported method [20] with minor modifications. Salicylic acid ethanol (2.0 mL) (9 mmol·L^-1) and FeSO₄ (2.0 mL) (9 mmol·L^-1) were added to 1.0 mL flavonoids or corn silk polysaccharide solution (0.025-0.175 mg·mL^-1). Then, 2.0 mL H₂O₂ (8.5 mmol·L^-1) was added to the mixture. The tubes were placed at 35°C for 20 mins. The absorbance (A_x) was determined at 510 nm, and the absorbance of the reaction without the sample was denoted as A_y. The absorbance of the distilled water was denoted as A₀ (Eq. 7).

2.7. Statistical analysis

The range of conditions for the multifactor experiment was obtained by analyzing the results of the single-factor test. The factorial design of the multifactor experiment was created using Design-Expert 10.0, according to the RSM. The Central Composite Design (CCD) method was used to perform the experimental design, and there were three factors with five levels, as described next. The first independent factor was the UCON concentration, represented by X₁; the second factor was the K₂C₄H₄O₆ concentration, represented by X₂; and the third factor was the system temperature, represented by X₃. Regression analysis was performed using Design-Expert 10.0, and the analysis model used was a quadratic equation (Eq. 8) as follows:

where X_i and X_j are the factors, and the values of ‘i’ and ‘j’ are 1 (concentration of UCON), 2 (concentration of K₂C₄H₄O₆), and 3 (system temperature), respectively; Y is the response; β₀ is the intercept; β_i, β_ii, and β_ij are the regression coefficient for linear, quadratic, and interaction, respectively; The validity of the model was evaluated based on statistical significance, which was achieved using an F-test.

2.8. ANN modeling

The extraction and separation processes were simulated using the developed backpropagation ANN model, as shown in Figure 1. The input layer had three artificial neurons, which represented three factors in the RSM. The output layer had two artificial neurons; the extraction efficiencies of flavonoids (Y₁) and the extraction efficiencies of corn silk polysaccharides (Y₂). The number of artificial neurons was determined by testing in the hidden layer (Figure S1).

The input and output variables were normalized in this ANN model. The three commonly used methods for unification are as follows in Eq. (9) to (11):

where X represents the original data, $\text{X′}$ represents unified values, $X_{max}$ is the original maximum value, and $X_{min}$ is the original minimum value. The training parameters of this ANN model have been listed in Table S1. The best sets of weights were obtained using a cross-validation technique to prevent overtraining.

3.1. Ultrasonic UCON-extraction efficiency

3.2. Choice of salt in UCON-based ATPS

UCON-based corn silk extract was obtained from corn silk via cell disruption extraction. A UCON-based ATPS was then used to separate flavonoids and corn silk polysaccharides. Based on the phase formation mechanism of ATPS, we selected 13 salts to construct the ATPS. Six UCON-based ATPS containing different salts were used to separate flavonoids and corn silk polysaccharides because the mechanisms of extraction and separation in the ATPS were mainly salting-out and rule of similarity. The reason for the choice of six types of UCON-based ATPS was that these six salts have excellent performances. The partition efficiencies of the UCON-based APTSs have been shown in Figure S2. K₂CO₃-UCON-ATPS, K₂HPO₄-UCON-ATPS, and K₂C₄H₄O₆-UCON-ATPS showed superior abilities in the separation of flavonoids and corn silk polysaccharides. We used K₂C₄H₄O₆-UCON-ATPS as the separation system because K₂C₄H₄O₆ was degradable in this study. Furthermore, the degradation components have no adverse effects on the environment.

3.3. Material distribution in UCON-K₂C₄H₄O₆ ATPS

3.4. The multi-factor experiment and RSM modeling

3.4.1. The regression model

The design of a multifactor experiment with three factors at five levels was performed using RSM. The multifactor experimental data have been provided in Table S4. Multiple regression analysis was used to analyze the response function (Eq. 8). The significance of each response function coefficient was evaluated. For the two objects, the final partition efficiency prediction models (Eqs. 12 and 13) can be obtained as follows:

In the above prediction equations, Y₁ and Y₂ represent the partition efficiency of the two objects; X₁, X₂, and X₃ are the same as those in Eq. (4). For these three equations, the R-square values were 0.7520 and 0.8052. We attempted to obtain better results. Therefore, the quadratic equation was replaced with the quartic equation used in our previous study, and the R-square values can reach 0.9568 and 0.9656, respectively. However, the quartic equation as a prediction equation yields very poor prediction results. The analysis shows that the quartic equation over-fit the experimental data. Therefore, the quadratic equation was used as the regression model.

3.4.3. The combined effect of three factors

Response surface plots were used to show the combined effect of the factors on the partition efficiency, as shown in Figure 2. From this figure, the optimum ranges for the two variables were obtained. The optimal values of the two responses can be determined based on Figure 2 and Table S4. The observations were as follows: the partition efficiencies of the flavonoid were 92.62% when the UCON concentration, K₂C₄H₄O₆ concentration, and temperature were 0.15 mg·mL^-1, 0.0875 mg·mL^-1, and 35°C, respectively. The partition efficiency of corn silk polysaccharide was 88.59% when the UCON concentration, K₂C₄H₄O₆ concentration, and temperature were 0.12 mg·mL^-1, 0.0625 mg·mL^-1, and 42°C, respectively. A higher concentration of K₂C₄H₄O₆ was beneficial for the partition efficiencies of the flavonoid, but the partition efficiency of corn silk polysaccharide decreased when the concentration of K₂C₄H₄O₆ was over 0.08 mg·mL^-1. Considering the balance between two objects and mild conditions, the optimal experimental conditions were: UCON concentration was 0.15 mg·mL^-1, K₂C₄H₄O₆ concentration was 0.07 mg·mL^-1, and temperature was 35°C. Under these conditions, the partition efficiencies of the two objects reached 91.18% and 88.67%, respectively.

Close

Figure 2

(a-f ) The 3D response surface plot for the interactive effect of variables, while other variables are set at the fixed value.

Export to PPT

3.5. The ANN model and prediction

3.5.1 Construction of ANN model

Three typical training algorithms, the Levenberg-Marquardt Algorithm, the Scaled Conjugate Gradient Algorithm, and the Bayesian Regularization Algorithm were used to form the ANN structure. The performances of the three algorithms were evaluated, and the results have been shown in Figure S7. The Levenberg-Marquardt Algorithm was used as the training algorithm for this ANN model. The input and output variables were normalized using three equations (Eqs. 9 to 11), and the regularized and original data were used to form the ANN model. Their performances have been listed in Table S6. The ANN model trained on the dataset normalized by Eq. (9) exhibited superior performance compared to those listed in Table S6. Therefore, Eq. (9) was used to normalize the input and output variables in this study. The hidden layer structure was the most important in the ANN model. The number of hidden neurons was varied from 3 to 45 in Step 3 to determine the optimal structure. The regression R values and epochs for the ANN model containing different hidden cells have been shown in Figure S8. Figure 3 shows the regression R values of the proposed ANN Model in the training, validation, test, and total datasets. The iterative time and MSE for each epoch have been shown in Figure S9.

Supplementary Figure 7

Download

Supplementary Table 6

Download

Supplementary Figure 8

Download

Supplementary Figure 9

Download

Close

Figure 3.

The regression R values of the proposed ANN Model in (a) the regression R values in training dataset, (b) the regression R values in validation dataset, (c) the regression R values in test dataset, (d) the regression R values in total dataset.

Export to PPT

3.6. UCON recycling and cyclic test

The UCON-rich phase was transferred to a new test tube after ATPS formation. The UCON-rich phase was then diluted by adding water. The thermostat water bath which was maintained at 65°C of temperature was used to control the two phases forming. To determine the optimal recovery efficiency, the amount of added water was increased from 0.5 mL to 3.0 mL in 0.5 mL steps. On addition of, 2.0 mL water, the recovery efficiency reached its maximum value.

The recovered UCON was used as the main forming material in subsequent extraction experiments. The results of the three cycle tests have been presented in Table S8. It was found that the loss of UCON in each cycle test was about 0.4 mL from the table. This loss primarily resulted from the quantity needed for detection and unavoidable losses during the transfer process. Additional EOPO was used to compensate for this shortfall in the subsequent ATPS extraction system. After the three extractions, the recovery of flavonoids increased gradually because flavonoid residues were present during each extraction. In each of these experiments, the recovery rate of flavonoids in the ATPS test was slightly higher than that in the UCON recovery experiment, and this was due to the wastage of flavonoids.

3.7. Validation of method

Two objects in seven groups of raw materials with different weights were extracted under optimum conditions. The partition and extraction performances have been listed in Table S9. The proposed method exhibited excellent partition and extraction performances in the range of raw material mass of 10-2000 g.

3.8. Purity and activity

The purities of the flavonoids and corn silk polysaccharides were determined under the optimum conditions. The purity of the flavonoids was between 98.19% and 99.27%, and that of the corn silk polysaccharide was between 98.46% and 99.73% (Table S10).

The antioxidant activities of flavonoids and corn silk polysaccharides were studied by evaluating their radical scavenging activities against DPPH, superoxide, and hydroxyl. The clearance rates have been shown in Figure 4. For the inhibitory on hydroxyl radical, the IC₅₀ values of flavonoids and corn silk polysaccharides were 0.087 mg·mL^-1 and 0.096 mg·mL^-1. It was reported that the scavenging effect of Vitamin C on hydroxyl radical was between 50% and 60% at 1.63 mg·mL^-1. For the inhibitory effects on DPPH radical, the IC₅₀ values of flavonoids and corn silk polysaccharide were 0.069 mg·mL^-1 and 0.025 mg·mL^-1, respectively. For the inhibitory on superoxide radical, the IC₅₀ values of flavonoids and corn silk polysaccharide were 0.073 mg·mL^-1 and 0.054 mg·mL^-1. Comparing the reported IC₅₀ values of corn silk polysaccharide against DPPH (33.5 ug·mL^-1) and hydroxyl (102.7 ug·mL^-1) [16], and the IC₅₀ values of polysaccharide against superoxide (0.06 mg·mL^-1) [57], the corn silk polysaccharide extracted by this method performed better. The DPPH activities of the flavonoids from astragalus membranaceus stems and leaves was given as IC₅₀ values (0.149 mg·mL^-1) [53]. The IC₅₀ value of flavonoids from kosteletzkya virginica root on superoxide radical was 0.078 mg·mL^-1 [55]. Thus, the flavonoids from corn silk in this work have better antioxidant activity.

Close

Figure 4.

The antioxidant activities of the flavonoids and corn silk polysaccharides against DPPH, superoxide, and hydroxyl.

Export to PPT

3.9. Comparison