The optimization of the reverse micelles extraction of protein from grape seeds was carried out using response surface methodology (RSM). Based on the Plackett-Burman design and steepest ascent, CTAB concentration, pH, NaCl concentration and crude protein concentration were selected as the most extract conditions. Subsequently, the optimum combination of the selected factors was investigated by the Box-Behnken design. The final condition of extraction optimized with RSM was CTAB concentration 39 mmol/L, pH 5.6, NaCl concentration 0.01 mol/L, and crude protein concentration 2.1 mg/mL. The forward extraction yield of 82.3% in triplicate under optimal extraction condition was obtained.
The grape is one of the major fruit crops worldwide and its harvest is about 60 million tonnes per year [
Therefore, treatment and disposal of winery waste are serious environmental problems and winery waste must find another use other than as animal feed or as fertilizers.
Grape seeds are the primary main byproducts of viniculture and fermentation. In the last few years, increased attention has been focused on industrial wastes that are rich source of polyphenolic compounds, flavonoids, protein and oil [
Grape seeds have relatively high content of protein (13% - 18%), which can be extracted by conventional procedures such as solvent extraction and isoelectric precipitation [
The reverse micelles extraction is a novel separation technology with prospect for separating bio-product. Reverse micelles are aggregates of surfactant molecules spontaneously in non-polar solvents. The aggregates of surfactant molecules contain an inner core of water molecules and are dispersed in a continuous organic solvent medium. The bio-molecules can be transferred from the aqueous phase to the polar core of reverse micelles without loss in activity [
Grape seeds were obtained from Palieri grape cultivar. Ceryl-trimethyl-ammo- nium bromide (CTAB), sodium chloride were purchased from the 6th Chemical Reagent Factory of Tianjin, Chin. Other materials used in this study were of analytical grade.
Grape seeds were selected manually and cleaned to remove contaminants. Grape seeds were milled using a small scale hammer mill (FZ-102, Hebei province, China ), and the resulting flour was sieved through a 200-mesh screen. Grape seeds power was defatted with n-hexane for 10 h and air-dried at room temperature (about 20˚C) by Soxhlet extraction. The power was kept in polyethylene bags and stored at 4˚C until used.
Defatted grape seeds power was soaked by 0.2 mol/L citric acid-sodium hydrogen phosphate buffer solution at pH 6.0 for 1 h. The solution and residue were isolated by a centrifuge at a rolling speed of 4000 rpm and 4˚C for 10 min. The crude protein was collected.
The reversed micelles systems were formed by Ceryl-trimethyl-ammonium bromide (CTAB), methenyl trichoride and butyl alcohol. The aqueous solutions were crude protein after centrifugation. Sodium chloride was added to the aqueous solution to adjust the ionic strength. For the forward extraction, equal volumes of the reverse micellar systems (the organic solution) and aqueous solution were mixed in a test tube in a reciprocating shaker bath for various time periods and temperatures. The mixture was then centrifuged at 1500 g for 5 min to separate the two phases. The aqueous phase was then taken for analysis. All the experiments were carried out in duplicate.
Protein concentration in water phase was determined by UV-Vis spectrophotometer (LabTech UV-2100, Beijing ). BSA was used as standard, and the results were expressed as BSA equivalents. The forward-extraction efficiency was calculated as follows.
Forward-extractionefficiency ( Y % ) = totalproteininthesupernatant − totalproteininaqueoussolution totalproteininthesupernatant × 100 %
Plackett-Burman design, an efficient technique for forward extraction conditions optimization [
After selecting the most important factor affecting the forward extraction yield in the screening study, the steepest ascent method was used to construct a line through the center of the design [
Once critical factors were identified via screening, a Box-Behnken design for the most important independent variables (CTAB concentration, pH, NaCl con- centration, crude protein concentration). Each at three levels with three replicates at the centre points was employed to fit a polynomial model:
Y i = β 0 + ∑ β i X i + ∑ β i i X i 2 + ∑ β i j X i X j (1)
where Yi is the predicted response, XiXj are input variables which influence the response variable Y; β 0 is the offset term; β i is the ith linear coefficient; β i i is the iith quadratic coefficient and β i j is the ijth interaction coefficient. Design expert package (version 7.0, Stat-Ease, Inc., Minneapolis, MN, USA) was used for the experimental design and regression analysis of the data obtained.
Based on the earlier studies, a total of seven variable conditions were analyzed for their effect on forward extraction using a Plackett-Burman design. The variables chosen for the present study were CTAB concentration, pH, extraction temperature, alkyl alcohol than, extraction time, NaCl concentration and crude protein concentration. All the variables were denoted as numerical factors and investigated at two widely spaced intervals designated as −1 (low level) and +1 (high level). The yield of forward-extraction, determined for each experimental design was shown in
Y = 49.6633 + 14.715 X 1 − 9.66167 X 2 − 10.0383 X 6 + 8.98 X 7 (2)
| Run | The forward extraction yield (%) | |||||||
|---|---|---|---|---|---|---|---|---|
| 1 | 40 | 8 | 20 | 4:1 | 20 | 0.04 | 1 | 39.36 |
| 2 | 20 | 8 | 40 | 2:1 | 20 | 0.04 | 2 | 32.21 |
| 3 | 40 | 4 | 40 | 4:1 | 10 | 0.04 | 2 | 72.5 |
| 4 | 20 | 8 | 20 | 4:1 | 20 | 0.02 | 2 | 35.58 |
| 5 | 20 | 4 | 40 | 2:1 | 20 | 0.04 | 1 | 19.36 |
| 6 | 20 | 4 | 20 | 4:1 | 10 | 0.04 | 2 | 45.02 |
| 7 | 40 | 4 | 20 | 2:1 | 20 | 0.02 | 2 | 82.3 |
| 8 | 40 | 8 | 20 | 2:1 | 10 | 0.04 | 1 | 29.3 |
| 9 | 40 | 8 | 40 | 2:1 | 10 | 0.02 | 2 | 84.25 |
| 10 | 20 | 8 | 40 | 4:1 | 10 | 0.02 | 1 | 19.31 |
| 11 | 40 | 4 | 40 | 4:1 | 20 | 0.02 | 1 | 78.56 |
| 12 | 20 | 4 | 20 | 2:1 | 10 | 0.02 | 1 | 58.21 |
*X1: CTAB concentration (mmol/L); X2: pH; X3: Extraction temperature (˚C); X4: alkyl alcohol than (V:V); X5: Extraction time (min); X6: NaCl concentration (mol/L); X7: Crude protein concentration (mg/mL).
| Source | Sum of Squares | Degree of freedom | Mean Square | F Value | Prob > F |
|---|---|---|---|---|---|
| Model | 5895.451 | 4 | 1473.863 | 15.16151 | 0.0015 |
| X1 | 2598.375 | 1 | 2598.375 | 26.72928 | 0.0013 |
| X2 | 1120.174 | 1 | 1120.174 | 11.52314 | 0.0115 |
| X6 | 1209.218 | 1 | 1209.218 | 12.43913 | 0.0096 |
| X7 | 967.6848 | 1 | 967.6848 | 9.9545 | 0.0160 |
| Residual | 680.4755 | 7 | 97.21079 | ||
| Cor Total | 6575.926 | 11 |
R2 = 0.896.
The multiple correlation coefficient (R2) of this first-order model is 0.896, which means that 89.6% of the data variation can be evaluated by the model. However, the difference between the adjusted R2 value (83.7%) and the predicted R2 value (69.6%) revealed that a first-order model is not an adequate mathematical equation for demonstrating the relationship between the significant independent variables and the response. Therefore, a second-order model should be employed for further investigation.
It can be seen from Equation (2) that all the significant factors, except pH (X2), NaCl concentration (X6) had a positive sign. Therefore, increasing their value would result in an increase in the level of forward-extraction efficiency. Further statistical analysis revealed that the difference between the means of the center point and factorial trials in this design was significant (P < 0.05). This indicated that the optimum levels for forward-extraction efficiency would be beyond the experimental ranges chosen for the Plackett-Burman design. Therefore, the steepest ascent method should be used. All the other insignificant variables were neglected and optimum combinations of these four were further analyzed by a steepest ascent design.
The steepest ascent method was used to construct a line through the center of the design, due to the contribution obtained by Plackett-Burman first-order equation. Consequently, some experiments were implemented along this line with defined intervals, and the response at each point was measured. If a maximum value is found, that point could be employed as the center point for the following optimization experimental design. These results are summarized in
Experiments were carried out in duplicates to arrive at an optimum combination of the four conditions above using Box-Behnken design. Based on the results of steepest ascent experiments,
Y = 82.244 + 0.1917 A + 3.9 B − 0.895 C − 1.3483 D − 1.7225 A B − 1.285 A C + 2.2575 A D + 1.45 B C + 5.3575 B D + 1.015 C D − 6.6749 A 2 − 4.9474 B 2 − 4.019 C 2 − 5.145 D 2 (3)
1) Analysis of variance (ANOVA) for the extraction yield of grape seeds protein
The analysis of variance (ANOVA) was conducted to test the significance of the fit of the second-order polynomial equation for the experimental data as shown in
| Run | CTAB concentration (mmol/L) | pH | NaCl concentration (mol/L) | crude protein concentration (mg/mL) | The forward extraction yield (%) |
|---|---|---|---|---|---|
| 1 | 30 | 6 | 0.03 | 1.5 | 55.49 |
| 2 | 33 | 5.8 | 0.025 | 1.7 | 64.02 |
| 3 | 36 | 5.6 | 0.02 | 1.9 | 73.24 |
| 4 | 39 | 5.4 | 0.015 | 2.1 | 78.68 |
| 5 | 42 | 5.2 | 0.01 | 2.3 | 71.45 |
| 6 | 45 | 5 | 0.005 | 2.5 | 68.47 |
| Trial No. | Coded variablesa | Uncoded variables | The forward extraction yield (%) | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| A | B | C | D | A | B | C | D | Experimental | Predicted | |
| 1 | −1 | −1 | 0 | 0 | 33 | 5 | 0.015 | 2.1 | 65.23 | 64.81 |
| 2 | 1 | −1 | 0 | 0 | 45 | 5 | 0.015 | 2.1 | 69.2 | 68.64 |
| 3 | −1 | 1 | 0 | 0 | 33 | 5.8 | 0.015 | 2.1 | 75.5 | 76.05 |
| 4 | 1 | 1 | 0 | 0 | 45 | 5.8 | 0.015 | 2.1 | 72.58 | 72.99 |
| 5 | 0 | 0 | −1 | −1 | 39 | 5.4 | 0.01 | 1.9 | 76.1 | 76.34 |
| 6 | 0 | 0 | 1 | −1 | 39 | 5.4 | 0.02 | 1.9 | 73.24 | 72.52 |
| 7 | 0 | 0 | −1 | 1 | 39 | 5.4 | 0.01 | 2.3 | 70.9 | 71.61 |
| 8 | 0 | 0 | 1 | 1 | 39 | 5.4 | 0.02 | 2.3 | 72.1 | 71.85 |
| 9 | −1 | 0 | 0 | −1 | 33 | 5.4 | 0.015 | 1.9 | 72.28 | 73.84 |
| 10 | 1 | 0 | 0 | −1 | 45 | 5.4 | 0.015 | 1.9 | 66.23 | 69.71 |
| 11 | −1 | 0 | 0 | 1 | 33 | 5.4 | 0.015 | 2.3 | 70.23 | 66.63 |
| 12 | 1 | 0 | 0 | 1 | 45 | 5.4 | 0.015 | 2.3 | 73.21 | 71.53 |
| 13 | 0 | −1 | −1 | 0 | 39 | 5 | 0.01 | 2.1 | 71.22 | 71.72 |
| 14 | 0 | 1 | −1 | 0 | 39 | 5.8 | 0.01 | 2.1 | 79.23 | 76.62 |
| 15 | 0 | −1 | 1 | 0 | 39 | 5 | 0.02 | 2.1 | 64.55 | 67.03 |
| 16 | 0 | 1 | 1 | 0 | 39 | 5.8 | 0.02 | 2.1 | 78.36 | 77.73 |
| 17 | −1 | 0 | −1 | 0 | 33 | 5.4 | 0.01 | 2.1 | 69.5 | 70.97 |
| 18 | 1 | 0 | −1 | 0 | 45 | 5.4 | 0.01 | 2.1 | 74.23 | 73.92 |
| 19 | −1 | 0 | 1 | 0 | 33 | 5.4 | 0.02 | 2.1 | 71.3 | 71.75 |
| 20 | 1 | 0 | 1 | 0 | 45 | 5.4 | 0.02 | 2.1 | 70.89 | 69.56 |
| 21 | 0 | −1 | 0 | −1 | 39 | 5 | 0.015 | 1.9 | 78.3 | 74.96 |
| 22 | 0 | 1 | 0 | −1 | 39 | 5.8 | 0.015 | 1.9 | 73.25 | 72.04 |
| 23 | 0 | −1 | 0 | 1 | 39 | 5 | 0.015 | 2.3 | 60.2 | 61.55 |
| 24 | 0 | 1 | 0 | 1 | 39 | 5.8 | 0.015 | 2.3 | 76.58 | 80.06 |
| 25 | 0 | 0 | 0 | 0 | 39 | 5.4 | 0.015 | 2.1 | 80.2 | 82.24 |
| 26 | 0 | 0 | 0 | 0 | 39 | 5.4 | 0.015 | 2.1 | 82.56 | 82.24 |
| 27 | 0 | 0 | 0 | 0 | 39 | 5.4 | 0.015 | 2.1 | 83.26 | 82.24 |
| 28 | 0 | 0 | 0 | 0 | 39 | 5.4 | 0.015 | 2.1 | 82.9 | 82.24 |
| 29 | 0 | 0 | 0 | 0 | 39 | 5.4 | 0.015 | 2.1 | 82.3 | 82.24 |
aA: CTAB concentration (mmol/L), B: pH, C: NaCl concentration (mol/L), D: Crude protein concentration (mg/mL).
this study less than 0.01 indicate model terms are very significant. Among model terms, B, BD, A2, B2, C2, D2 are significant with a probability of 99%. P-values greater than 0.05 indicate the model terms are not significant. Here the R2 value was 91.21%, which could explain 91.21% variability of the response. It indicates a good agreement between experimental and predicted values and implies that the mathematical model is very reliable for protein extraction field in the present study. At the same time, a very low value 3.3 of coefficient of the variation (CV) clearly indicated a very degree of precision and a good deal of reliability of the
| Factors | Sum of squares | df | Mean square | F-value | p-value |
|---|---|---|---|---|---|
| Model | 856.6741 | 14 | 61.19101 | 10.37369 | <0.0001 |
| A-A | 0.4408 | 1 | 0.440833 | 0.074734 | 0.7886 |
| B-B | 182.52 | 1 | 182.52 | 30.94255 | <0.0001 |
| C-C | 9.6123 | 1 | 9.6123 | 1.62957 | 0.2225 |
| D-D | 21.8160 | 1 | 21.81603 | 3.698464 | 0.0750 |
| AB | 11.8680 | 1 | 11.86803 | 2.011982 | 0.1779 |
| AC | 6.6049 | 1 | 6.6049 | 1.119726 | 0.3079 |
| AD | 20.3852 | 1 | 20.38523 | 3.4559 | 0.0842 |
| BC | 8.41 | 1 | 8.41 | 1.425744 | 0.2523 |
| BD | 114.8112 | 1 | 114.8112 | 19.4639 | 0.0006 |
| CD | 4.1209 | 1 | 4.1209 | 0.698615 | 0.4173 |
| A2 | 289.0022 | 1 | 289.0022 | 48.99444 | <0.0001 |
| B2 | 158.7693 | 1 | 158.7693 | 26.9161 | 0.0001 |
| C2 | 104.8199 | 1 | 104.8199 | 17.77007 | 0.0009 |
| D2 | 171.6984 | 1 | 171.6984 | 29.10796 | <0.0001 |
| Residual | 82.58144 | 14 | 5.898674 | ||
| Lack of Fit | 76.83792 | 10 | 7.683792 | 5.351277 | 0.0601 |
| Pure Error | 5.74352 | 4 | 1.43588 | ||
| Cor Total | 939.2555 | 28 |
experimental values.
Response surface plots are shown in
The graph shown in
In order to optimize processing conditions of grape seeds protein extraction , the first partial derivatives of the regression model were equated to zero according to A, B, C and D. From the model, optimum conditions for grape seeds protein extraction were prepared as follows: CTAB concentration 38.84 mmol/L, NaCl concentration 0.01 mol/L, crude protein concentration 2.12 mg/mL. The pH of the aqueous phase was 5.58. Under such conditions, the yield of forward extraction process was predicted to be 83.06%.
To ensure the predicted result was not biased toward the practical value, experiment rechecking was performed by using these modified optimal conditions: CTAB concentration 39 mmol/L, NaCl concentration 0.01 mol/L, crude protein concentration 2.1 mg/mL. The pH of the aqueous phase was 5.6. A mean value of 82.3% (N = 3) was obtained from real experiment. The results of analysis confirmed that the response model was adequate for reflecting the expected optimization, and the model of Equation (3) was satisfactory and accurate.
The data presented in this article demonstrate the feasibility of the forward extraction of protein from grape seeds by reverse micelles. Based on the Plackett- Burman design and steepest ascent, response surface methodology (RSM) was used to estimate and optimize the experimental variables: CTAB concen- tration, pH, NaCl concentration and crude protein concentration. The optimal forward extraction conditions for grape seeds protein by reverse micelles were determined as follows: CTAB concentration 39 mmol/L, NaCl concentration 0.01 mol/L, crude protein concentration 2.1 mg/mL; the pH of the aqueous phase was 5.6. Under these conditions, the forward extraction yield of grape seeds protein was 82.3%, which was closed with the predicted yield value. The data presented in this article demonstrate the feasibility of the forward extraction of protein from grape seeds by reverse micelles. Reverse micelles extraction was an efficient method compared to conventional solvent extraction. These results demonstrated the successful extraction of protein with Reverse micelles extraction, providing potential benefits for industrial extraction of protein from grape seeds.
This work has been supported by the Gansu provincial innovation foundation for technology based firms (No. 1047GCCG001), and by President Foundation of Hexi University (No. XZ2014-29).
The authors declared that they have no conflicts of interest to this work. We declare that we do not have any commercial or associative interest that re- presents a conflict of interest in connection with the work submitted.
Zhang, X.F., Hou, Y.Y., Zhang, F.Q. and Luo, G.H. (2017) Protein Extraction from Grape Seeds by Reverse Micelles: Optimization of the For- ward Extraction. Open Access Library Jour- nal, 4: e3376. https://doi.org/10.4236/oalib.1103376