Concentration of pistachio hull extract antioxidants using membrane separation and reduction of membrane fouling during process

Abstract The purpose of this study was to concentrate the polyphenolic compounds in pistachio hull extract as a rich source of natural antioxidants using the membrane process and the reduction of membrane fouling during the process. After the optimization of the membrane performance includes chamber pressure and stirring rate by monitoring flux and membrane fouling, pectinase and tannase enzymes were compared in order to reduce fouling. Pectinase showed a better potential in decreasing membranous fouling. Enzyme concentration was optimized, and treatment with 17.4 (U) was selected. The permeate obtained from optimized membrane condition and enzyme level treatment was enriched in total phenol (120.31 ± 0.35 mgGAE/g) and flavonoid (34.54 ± 0.09 mgCE/g), while the amount of anthocyanin was not remarkable.

There is a possibility to concentrate phenolic compounds from the pistachio hull extract via membrane separation due to a difference between the molecular weights of the main ingredients of pistachio hull, such as polyphenolic compounds, protein, and pectin.
Membrane-based processes, including UF 1 , have been widely used in various industrial fields. It is among the conventional technologies that are utilized prior to or after the extraction process in order to separate macromolecules from smaller compounds in a physicochemical and nondestructive way (Liu et al., 2014). UF has been broadly used to treat fruit juices to obtain fractions enriched in phenolic. It has been recently reported that ultrafiltration and nanofiltration can be suited for the concentration and recovery of bioactive compounds from juices and byproducts (Avram et al., 2017;Conidi & Cassano, 2015;Conidi, Cassano, Caiazzo, & Drioli, 2017).
One of the most problematic issues that can interfere in the efficiency of ultrafiltration is membrane fouling. It is caused by the deposition of rejected material such as colloidal particles on the surface and within the pores of the membrane; results in flux decline and change in terms of membrane selectivity. Different methods to reduce membrane fouling include physical (e.g., ultrasound, sponge balls, back pulsing), biological (e.g., enzymatic treatment), and chemical (acids, alkalis, disinfectants, detergents) techniques (Saha & Balakrishnan, 2009). Thus, this problem can be overcome by the enzymatic treatment (by the usage of hemicellulases, phenol oxidase, and, in particular, pectinases) of the sample in which the colloidal particles are degraded before ultrafiltration (Kilara & Van Buren, 1995).
The main aim of this research is to concentrate and increase the amount of polyphenolic compounds in the pistachio green hull extract using the membrane system and to reduce membrane fouling during the process by the use of enzyme.

| Material
Pistachio green hulls (the Ahmadaghaei variety) were obtained from the Kerman Agricultural Research Center of Iran. The hulls were dried and ground, and then, a fraction that was sieved through a 10-mesh sieve and retained on a 40-mesh sieve was selected and stored in a freezer at −20°C until extraction. All the chemicals were analytical grade and obtained from the Sigma-Aldrich Company Ltd.
(Gillingham, UK) and Merck (Darmstadt, Germany) and were used without further purification.
In this regard, water was used to extract phenolic compounds from the pistachio hull as a solvent. Therefore, 1 g of milled hull was subjected to the extraction, using a liquid-to-solid ratio of 1:15, during 8 hr and at 25°C (Goli et al., 2005;Rajaei et al., 2010).

| Membrane operation system
Ultrafiltration of the pistachio hull extract was carried out in the batch dead-end filtration plastic cell with a maximal capacity of 50 ml and an effective filtration surface area (A) of 13.4 cm 2 (Amicon 8050, Merc Millipore, USA). A polysulfone membrane with a molecular weight CUT-OFF SIZE (MWCO) of 100 kDa (GR-40 pp, Alfa Laval) was used for filtration. The operational parameters, which may affect membrane separation performance include chamber pressure (1, 2.5, and 4 bar) and stirring rate (50, 150, and 250 rpm), were optimized. The pressure was prepared using a pump that was connected to the nitrogen container. The system also consisted of a magnetic stirrer, whose speed was controllable.
Prior to filtration, the membrane was washed by distilled water and then the filtration process of the pistachio hull extract was started. After filtration, distilled water was again filtered through the fouled membrane for the determination of membrane fouling. The most common way to evaluate membrane fouling included comparing water flux through origin and used membranes under the same operating conditions (Shi, Tal, Hankins, & Gitis, 2014).
where flux 1 is the water flux of the membrane after sample filtration at a given TMP and flux 0 is the water flux through the virgin membrane at the same TMP (Shi et al., 2014). The filtrated flux was calculated using Equation 2: where Q is the volume of the permeate (in L), A is the active membrane surface area (in m 2 ), and t is the time taken for permeate filtration (in hr).

| Enzymatic treatment of extract samples
In this regard, four levels of pectinase enzyme (5, 10, 15, and 25 μl/ ml) were added to the 25 ml extract. Therefore, the treatments showed concentration levels (U) of 5. 8, 11.6, 17.4, and 29.1, respec-tively. Then, the solutions were incubated at 50°C for 30 min in a thermoshaker system (120 rpm). The treatments were labeled P-1, P-2, P-3, and P-4, respectively. The control sample did not receive any enzyme treatment, but 5 ml of the sample was taken for further analysis and the remaining 20 ml was injected into the aforementioned membranous system and filtering process was continued till 15 ml of the extract passed through the membranous system. The flux of the pistachio hull aqueous extraction was monitored while filtering. The membranous fouling and the amounts of different phenol compounds, such as total phenol, total tannin, and antioxidant activity of extract, were measured in initial feed, permeate, and retentate through the following methods. Finally, the optimal dose of the enzyme was determined.
(1) % Fouling = flux 1 − flux 0 flux 0 , In order to compare the performance of the tannin enzyme with the performance of pectinase in terms of reducing the membrane fouling of this enzyme, 10 mg/g of dry material of the extract was added to 25 ml of the extract (concentration (U): 5/14). Then, the resulting mixture was incubated at 37°C for 2 hr in a thermoshaker system (120 rpm; T-1 sample).

| Determination of total phenolic compounds
The total phenolic compounds were determined using a modified version of the Folin-Ciocalteu colorimetric method (Waterhouse, 2001

| Antioxidant activity
2.6.1 | DPPH˙ assay DPPH (2, 2-diphenyl-1-picrylhydrazyl) radical-scavenging activity was performed according to Blois (1958), with certain modifications. Different concentrations of the sample were mixed with 2.7 ml of 0.1 mM methanolic solution of DPPH radicals. The reaction mixture was vortexed and then incubated for 30 min at room temperature. The absorbance was read at 517 nm against a blank. The scavenging ability was calculated using the following equation: The IC 50 value is the effective concentration at which the DPPH radicals were scavenged by 50% and were obtained by interpolation from linear regression analysis (Hashemi, Aminlari, & Moosavi-Nasab, 2014).
The radical cation solution was prepared by the reaction of 7.4 mM ABTS˙ and 2.6 mM potassium persulfate solutions (in equal quantities), after 16 hr incubation at 22°C in a dark place. The ABTS + solution was then diluted by methanol to obtain a solution with the absorbance of 1.00 ± 0.02 units at 734 nm. 200 ml of each sample was added to 4 ml of the ABTS + solution, and after reacting for 2 hr in a dark place, the absorbance was measured at 734 nm. Ascorbic acid was used as standard and results were expressed in mg AA 2 equivalents/mg phenolic.

| Determination of total tannin
The total tannin was determined by measuring the nontannin phenols (NTP) and the precipitation of tannins using insoluble polyvinyl pyrrolidone (PVPP). 100 mg of insoluble polyvinylpyrrolidone was weighted, and 1 ml of distilled water and then 1 ml of extract were added and vortexed. The mixture was kept at 4°C for 15 min, vortexed again, and then centrifuged (3,000 × g) for 10 min; then, the supernatant was collected. The phenolic content of the supernatant was measured by the Folin-Ciocalteu reaction, and this was accepted as the NTP. Total tannins were calculated as the difference between total phenol and nontannin phenols (Makkar, Blummel, & Becker, 1995). Gallic acid was used as the standard.

| Determination of total flavonoid content
The total flavonoid content was determined according to the colorimetric method described by Heimler, Vignolini, Dini, and Romani (2005). In 1.25 ml of distilled water, 250 μl of sample extract was mixed; then, 75 μl of 5% NaNO 2 solution was added. After 6 min, 150 μl of 10% AlCl 3 ·6H 2 O solution was added and allowed to stand for another 5 min before adding 0.5 ml of 1 M NaOH. The mixture was brought to 2.5 ml with distilled water and then vortexed. The absorbance was immediately read against the blank at 510 nm.
Catechin was used as the standard.

| Statistical analysis
Experimental data were analyzed using the analysis of variance (ANOVA), and significant differences among the means from triplicate analyses at p < 0.05 were determined by Duncan's test using the SPSS 19 software.

| RE SULTS AND D ISCUSS I ON
About 40.7 g of aqueous extract was obtained per 100 g of green pistachio hull powder. The results of measuring the physicochemical characteristics of the aqueous extract of green pistachio hull are presented in Table 1, which conveys similar results on pistachio. Rajaei et al. (2010) reported that the total phenol of the PGH aqueous extract was 49.32 mg of GAE 3 /g sample and EC 50 of DPPH˙ was 2.53 ± 0.02 μg phenolic/ml. In 2016, Barreca et al. reported that the contents of total phenol, total flavonoids, and proanthocyanidins of the PGH methanol extract were 11.7 ± 0.48 (μM GAE), 0.688 ± 0.0197 (mg QE 4 /g fresh weight), and 0.177 ± 0.004 (mg of cyanidin chloride equivalents/g of fresh water), respectively.
As the first step, the prepared aqueous extract, was filtered using the 100 kD membrane under different conditions, including container pressure and mixer rotation, and (as mentioned in the previous section) its physiochemical parameters were calculated.
As presented in Figure 1, among the membrane-filtered samples, the highest and the lowest amount of total phenol was observed in the permeate and the retentate of the sample filtered under pressures of 2.5 and 1 bar, and mixer speeds of 250 and 50 rpm, respectively. It is obvious that the level of phenolic compounds for all the samples in the retentate was more than the permeate after filtration; this could have happened due to membranous fouling during the process. Generally, phenolic compounds have lower molecular size than 100 kD, but owing to the presence of complicated compounds such as pectin in the pistachio hull extract, they can be trapped between pectin branches or they can also face difficulties in finding a way and passing through the membrane owing to its fouling.
In order to examine the membrane fouling behavior and to select the condition which generates the least fouling at the time of filtering the extract through the filter, the percentage of membranous fouling and the flux of samples through the membrane were measured using the methods mentioned in materials and methods. These results are presented in Table 2.
The highest level of membranous fouling was observed when the sample was filtered through the membrane under 1 bar of pressure and a mixing speed of 50 rpm. The best condition for filtering the extract, according to the maximum content of polyphenolic compounds and the antioxidant potential of permeate, was a pressure of 2.5 bar and a mixing speed of 250 rpm.
The aqueous extract of the green pistachio hull includes macromolecular polymeric compounds such as pectin, which could act as a membrane fouling agent. In 2017 Chaharbaghi, Khodaiyan, & Hosseini reported the green pistachio hull as a valuable source of pectin. They showed that the maximum yield of pistachio green hull pectin obtained at the optimal conditions was 22.1 ± 0.5% (Chaharbaghi et al., 2017). In order to root out the problem of fouling in the current study, two tests were designed. In the second test, 96% ethanol (3:1 v/v), pectinase enzyme, and tannase enzyme were used for the sedimentation and the decomposition of pectin as well as the decomposition of tannin (as a large molecule that can play a role in the fouling). After filtering the treated samples through the membrane, the total phenol contents of the permeate and the retentate were compared with the content of the sample (Table 3).
The results suggest that pectinase is a more effective enzyme for reducing membranous fouling in comparison with tannase enzyme. This is probably due to the largeness of pectin molecules and their derivation from the pistachio extract. The total phenol in the sample treated with alcohol was reduced in comparison with the control sample. It seems that the underlying reason for this issue might be the concurrent deposition of phenolic compounds stuck between pectin branches and pectin molecules.
Based on the obtained results, one could conclude that the treatment of the extract by pectinase enzyme could be influential in reducing membranous fouling during the filtration process. Moreover, the concentration of the pectinase enzyme was optimized. In this regard, the effect of the pectinase enzyme of four concentrations was compared with the control sample. The physiochemical characteristics of extracts treated with different enzyme concentrations were determined after passing them through a 100-kDa membrane and under-optimized conditions (2.5 bar pressure; 250 rpm mixing). Tables 4 and 5  order to increase the concentration of the phenolic compound in the retentate stream.

As shown in
Linear regression between the antioxidant capacity (DPPH˙ and ABTS + ) and total phenolic content and total flavonoid content in the permeate of the enzyme-treated sample, in optimal conditions, is shown in Figure 3. The correlation coefficients of total phenols with DPPH and ABTS + were 0.982 and 0.967, while the correlation coefficients of total flavonoids with DPPH and ABTS assays were 0.955 and 0.833, respectively. It was observed that the radical-scavenging activity of the permeate of enzyme-treated extract showed a better correlation with total phenols rather than flavonoids. It could be estimated that in the pistachio hull aqueous extract, the nonflavonoid phenols such as phenolic acids are more responsible for antioxidant activity than flavonoids. Kumar, Sandhir, and Ojha (2014) reported  by Barreca et al. (2016). The amount of total flavonoid in the permeate of enzyme-treated sample was increased from 29.09 ± 1.53 to 34.54 ± 0.09 mg CE/gDW after passing through the membrane,  The antioxidant activity of enzyme-treated and crude extracts, expressed in DPPH˙ and ABTS˙+ assays, is shown in Figure 5.
Unstable free radical of DPPH is scavenged by antioxidants, as a result of electron donating and changes to a stable radical which has absorption at 517 nm. IC 50 shows an effective concentration of extract needed for inhibiting 50 percent of free DPPH˙ radicals (Barreca et al., 2016;Goli et al., 2005). For the determination of IC 50 , the best line of different concentration of the extract was used. In this regard, Figure 5a shows that 2.38 ± 0.8 ppm of the permeate of enzyme-treated filtered extract of green pistachio hull is needed for controlling 50 percent of the free DPPH˙ radicals, which demonstrates the excellent antioxidant activity, higher than the permeate of crude-filtered extract. Although the retentate of crude-filtered extract has a higher level of phenolic compounds from control and permeate, it shows a less antioxidant activity, probably due to the pro-oxidant activity of its impurities. It can also suggest that the type of phenolic compounds is more effective on antioxidant activities than its quantity (Rajaee et al., 2010).
The radical cation-scavenging activities of crude-and enzymetreated extracts, expressed as ascorbic acid equivalents, are shown in Figure 5b. permeate of crude extract was approximately equal to the control sample (unfiltered crude extract).
The results suggest that after treating the pistachio hull aqueous extract with the optimal level of pectinase enzyme, more amounts of phenolic compounds pass through the membrane due to the pectin decomposition and membrane fouling reduction and membrane concentration leads to separate a permeate part, rich in phenolic compounds and antioxidant activity.

| CON CLUS ION
The aqueous extract of pistachio green hull is a rich source of polyphenol compounds with antioxidant properties. The use of the polysulfonic membrane with 100 MWCO is an effective way to increase the concentration of the polyphenol compounds in the extract. To achieve this goal, conditions of the membrane process (pressure and agitation) were optimized and the best performance

CO N FLI C T O F I NTE R E S T
None declared.

E TH I C A L S TATEM ENTS
This study does not involve any human or animal testing. F I G U R E 5 Comparison between the antioxidant activity of control, retentate and permeate of pectinase enzymetreated and crude extracts, filtered in optimal condition, expressed in (a) DPPH assay and (b) ABTS assay. Capital letters show significant differences among similar columns (p < 0.05). Small letters show significant differences among three columns of treatments (p < 0.05)