Oil seeds (flaxseed, grape seed, apricot kernel) were obtained from a local supplier in Konya, Turky. Reagents for measurements were supplied by Sigma-Aldrich (St. Louis, MO, USA) and Merck (Darmstadt, Germany). Water was of ultrapure milli-Q grade. The commercial enzyme was a complex enzyme preparation of cellulases, pectinases and hemicellulolytic enzymes produced by Aspergillus aculeatus (SEBMax Olive, Advanced Enzyme Technologies Ltd., India). All the remaining chemicals used for experiments were at least analytical grade. The hydrolyzed or conventionally treated samples were directly subjected to pressing without any roasting or moisture adjustment. The control sample was considered as the sample produced by the conventional method. Neither buffer solution nor enzyme was applied.

Oil was obtained by pressing 1000 g of seeds with a screw press (Karaerler NF 500, Turky ) (nozzle diameter of 5 mm, shaft screw diameter of 33 mm, rotation speed of 50 rpm) at oil flowing temperatures below 50 °C (cold pressing).

The oil seeds were ground in a laboratory scale grinder (220 V, 1-30 kg seeds/hour capacity, 1.5 kW, 9000 rpm) to a particule size of 1-2 mm ground. Enzymes were applied by using a buffer solution (0.1 M of aqueous NaH₂PO₄, pH adjusted to 6 with 0.5 M NaOH or phosphoric acid) containing 10 g commercial enzyme mixture. 100 mL of this solution was sprayed onto 1 kg of ground seeds. Then the seeds were incubated at 60 ^oC for 3 hours. Enzyme solution was applied at a final enzyme concentration of 1 g enzyme per 100 g seeds. Pressing time was approximately 5 min. The oil extracted was stored in amber glass bottles. The pressing procedure was carried out in three replicates and the average values were reported. The percentage oil yield was obtained by calculating the proportional amount of oil obtained from 100 g of seed (Ezeh et al., 2016).

The determination of free fatty acids (FFA), peroxide value, and p-Anisidine values (pAV) of the extracted oil was carried out according to the standard AOCS official methods Cd 3d-63 and Cd 8b-90, respectively. Instrumental color was measured by a colorimeter [Minolta Chroma meter CR 400 (Osaka, Japan)]. The chromaticity coordinates a^* and b^* measures red-green and yellow-blue, respectively, depending on the negativity and positivity of the values (Criado et al., 2004).

The fatty acid composition was determined following the ISO standard (ISO, 1990; ISO, 2000). One miligram of oil was dissolved in 1 mL of n-heptane and 50 μg of sodium methylate, and the tube was shaken vigorously for 1 min at room temperature. 100 μL of water were added and the tube was centrifuged at 4500 g for 10 min and the lower aqueous phase was removed. Then the solution was mixed with 50 μL of HCl (1 mol with methyl orange) and the lower aqueous phase was rejected. About 20 mg of sodium hydrogen sulphate (monohydrate, extra pure) were added, and after centrifugation at 4500 g for 10 min, the top n-heptane phase was used. A Shimadzu GC-2010 Plus/FID/HS-20, with a capillary column, SP 2-4111 was also used (100 m long, 0.25 mm ID, film thickness 0.2 μm). The temperature program was as follows: from 140 °C (for 5 min.); heated to 240 °C (4° C/min) and held for 20 min; injector 250 °C, detector 260 °C; carrier gas 36 cm/s hydrogen; split ratio 1:50; detector gas 30 mL/min hydrogen; 300 mL/min air and 30 mL/min nitrogen; manual injection volume was 1 μL. The peak areas were computed by the integration software, and the results were calculated as weight percent by direct internal normalization.

The contents of tocopherol (α-, β-, γ- and δ-forms) in the oil were determined by Shimadzu LC-20A HPLC system equipped with a fluorescence detector and Lichrosorb Si 60 column (250 × 4.0 mm, 5 μm) (Supelco, USA). The emission and excitation wavelengths were set at 330 and 290 nm, respectively. The oil samples were dissolved in n-hexane at 50 mg/mL. A 20 μL volume of the loaded sample on the column was eluted with n-hexane/2-propanol (99.5/0.5, v/v) isocratically at 0.8 mL/min. The tocopherol contents were determined by comparing to calibrated standard curves. For the identification of tocopherols, the peak retention times were compared against absolute/ pure tocopherol compounds (α-, β-, γ- and δ -tocopherols) (IUPAC, 1987).

7.5 g oil were weighed and dissolved in cyclohexane up to a final volume of 25 mL. Carotenoid contents were calculated from the absorption spectra of the oils by using the specific extinction values. Absorption at 470 nm and 670 nm were recorded for carotenoid (mg lutein / kg oil) and chlorophyll (mg pheophytin a / kg oil) fractions, respectively (Minguez-Mosquera et al., 1991).

The antioxidant activity of the oil (total fraction) and the hydrophilic and lipophilic fractions were assessed by the scavenging activity of 1,1-diphenyl-2-picrylhydrazyl (DPPH) (Tuberoso et al., 2007; Teixeira et al., 2013). 1 g of oil was dissolved in 2 mL of 80% methanol, the solution was agitated for 30 min at room temperature and centrifuged at 700 g for 10 min to separate the methanol phase as hydrophilic fraction, while ethyl acetate was mixed with the remaining solution as lipophilic fraction.

The oil was diluted in ethyl acetate alone, without fractionation as total fraction. Antiradical activity was expressed as percent inhibition of DPPH. For the antiradical activity measurement, a methanolic solution (100 µL) of the extracts from different fractions of the oil was placed in a cuvette and 0.5 mL of a methanolic solution of DPPH (50 mg DPPH/100 mL MeOH) were added. After 20 min incubation in the dark and at ambient temperature (27 ^oC), the resultant absorbance was recorded at 517 nm (Shimadzu Co., Ltd., Kyoto, Japan). Antiradical activity was expressed as percent inhibition of DPPH (Roginsky and Lissi, 2005).

The methanolic extracts prepared for the antioxidant analysis were used for the determination of total phenolics. The method described by Rigane et al., (2011) was performed. The results were expressed as mg of gallic acid per kg of oil.

All parameters analyzed were determined in triplicate. Statistical analysis was performed with the SPSS v. 16 statistical software (SPSS Inc., Chicago, IL, USA). “General linear model multivariate analysis” was used to evaluate oil seed and pretreatment depended differences regarding the parameters analyzed. Significant differences between mean values were evaluated by using the Duncan’s new multiple range test (P<0.05).

The yield values forof flaxseed, apricot kernel and grape seed oils are presented in Table 1. Enzyme application showed the most significant increase in oil yield (22.75% increase) in the apricot seed sample. The pre-treatment of mixed enzyme formulation on apricot kernel resulted in 47.33% oil recovery which was 14.22% more than that obtained without enzymes as reported in the study of Bisht et al. (2015). Although the enzyme application provided a small increase in oil recovery from grape seed, it was not statistically significant. In a previous study, an enzymatic treatment for 9 hours at 50% moisture along with pre-heating, grape seed oil extraction yield was raised by 59.4% in comparison to the yield obtained from the control without enzymes (Tobar et al., 2005). The lower increase in oil yield achieved in the present study compared to that of Tobar et al. (2005) can be attributed to the longer incubation time (9 hours) and moisture level applied in their study. We tried to keep these conditional parameters as low as we could in order to suppress the formation of free acidity and oxidation. Soto, Chamy and Zúñıga (2004) also reported that with the use of an enzyme mixture, borage oil yield was 84% in comparison to 77.7% from control samples.

Contrary to expectations, this application significantly reduced oil yield from flaxseed. It is thought that the applied enzyme mixture is not suitable for the flaxseed cell membrane structure and also created a negative result by affecting the moisture ratio. Because the enzymes were applied in a buffer solution which increased the moisture ratio of ground seeds to a level that could not be ignored even 60 ^oC, 3 hours incubation led to a loss in the moisture level to some extent. This is a well-known effect of moisture which gives rise to the differences between the components of hydrophobic (van der Waals forces) and hydrophilic (hydrogen and ionic bonding) matrixes (Mustafa and Turner, 2011). In flaxseed, enzyme pre-treatment probably reduced the difference between these two fractions and, on the contrary, showed an effect in favor of binding (emulsion). Enzymes affect the release of oil depending on the seed composition. The oil found inside the plant cells (oil bodies, 0.6-2 µm) is a triglyceride matrix which is compromised of a monolayer of phospholipids linked together with proteins. Proteins give stability to this matrix due to the steric hindrance and electronegative repulsion of hydrophobic proteins on the surface of the oil bodies. Due to the structural properties of protein in the flaxseed, it is thought to be more stable with the addition of enzymes, unlike apricot kernel and grape seed.

On the other hand, in order to achieve a higher amount of oil extraction, enzyme preparations should have a broad spectrum of activity to disrupt the cell wall structure. The commercial enzyme preparate used in this study did not contain protease, which seemed inevitable for flaxseed pressing. Long et al. (2011) reported that when a mixture of cellulose, pectinase, and hemicellulose enzymes were applied, the flaxseed oil yield was 73.9%, higher than the oil yield from the application of each enzyme individually. These higher oil yield values reported by Long et al. (2011) are of course due to the application of a method completely different from our study (ultrasound-assisted aqueous enzymatic method) to extract the oil. In addition, it is known that the dissolution of proteins, at a pH far from the isoelectric point, disrupts the matrix and enables oil release (Oyinlola et al., 2004). This triggers the idea that a commercial mixture of enzymes along with buffer solution carrying it might have formed a pH close to the isoelectric point in ground flaxseed. A physical effect can also be considered in this case, because products from the degraded materials can negatively affect enzymes to act on their substrates (Ezeh et al., 2016). Particle size is another factor thought to be effective on the release of oil. The physical effect of enzymatic treatment is also involved, which was demonstrated by smaller particle size with a degraded surface of cotyledon cells of enzymatically treated soybeans (Sineiro et al., 1998). Hence, the disruption of walls and cytoplasm leads to a slow and gradual lose in the cellular and sub-cellular structure of cell tissues.

The enzyme pre-treatment duration of hemp seeds was up to 6 hours in study by Latif and Anwar (2009). Nosenko et al. (2019) demonstrated that using a different kind of proteolytic enzyme for pumpkin seed pre-treatment resulted in an increase in destroyed cell quantity from 3 to 10.4%. In particular, the acid proteases were the most effective enzymes in increasing destroyed cell content in oily material. The increase in oil extractability by means of enzyme pre-treatment was attributed to the breakdown of the protein network surrounding the lipid bodies and also supports the degradation of the complex lipoprotein molecules into simpler lipid and protein molecules, thereafter enhancing the release of oil (Murphy, 1993; Tzen and Huang, 1992).

Regarding the low oil content in grape seeds (determined between 10-11% in the present study), in previous studies, oil yields between 7-20% were reported and the authors pointed out that high amounts of oil remained in the cake from cold pressing (Matthäus, 2008; Özcan et al., 2017; Özcan and Aljuhaimi, 2017).

The enzyme application was shown to increase free fatty acidity (%) values (Table 1). Although it showed the lowest free fatty acidity value, the highest increase was in apricot oil. Moisture had a big effect on this because the carrier buffer solution which enables the enzyme to be added homogeneously increased the moisture content in the ground seed. Similarly, it has been reported that the acidity of oil increases with increasing moisture content in canola seed (Patil and Singh, 2017). Enzyme treatment also led to an increase in free fatty acids in borage oil from 5.56 to 6.55% compared to the control (Soto et al., 2007). An increase in the acid value (mg KOH/g) of apricot seed oil proportionate to the increase in enzyme concentration was reported by Bisht et al. (2015).

Free fatty acids in oil are formed as a result of hydrolysis in the presence of moisture but the non-enzymatic reaction only occurs at high temperatures. Lipase in the seeds may also have increased the hydrolysis reaction which leads to the occurence of free fatty acid during grinding and extraction. The main reason for a high acid value was probably the prolonged activity of native seed enzymes, especially lipases, since the optimum temperature and pH for the lipases of various origin ranges between 30 and 80 °C, and between 4.5 and 11, respectively (Barros et al., 2010).

Peroxide values were found between 0.80-3.00 meq O₂/kg oil. The peroxide values for enzyme-treated samples were lower for flaxseed and apricot oils than that of the control, but it caused a slight increase in grape seed oil which was not statistically significant. In addition, the high peroxide value for grape seed oil was remarkable. Patil and Sing (2017) reported that the moisture content in the seeds was responsible for the high peroxide values of the oil and that the high moisture content facilitated the oxidation of oil, which resulted in rancidity. Akinoso et al. (2010) also explained the high peroxide values for sesame oil with the same arguments. The more free fatty acids and primary oxidation products in aqueously extracted oil than cold-pressed oil reported by Konopka et al. (2016) might also show the influence of paste moisture on the oxidative stability of pressed oil. Grape seeds are obtained as waste from wine or molasses production, and grape seeds are naturally in contact with water during processing. With the application of enzymes on flaxseed and grape seeds, p-anisidine values decreased insignificantly, while the decrease was significant in the case of apricot kernel.

Grape seed oil showed important changes in color. It is understood that enzyme pre-treatment caused a significant decrease in the a* value and increase in b* value (lower redness and blueness) compared to the control.

There was a slight but significant increase in the stearic acid percentage of flaxseed and apricot kernel oils (Table 2). With respect to enzyme pre-treatment, generally there was no significant change in the concentration of fatty acids among the studied oil samples. In addition, a significant increase in the amount of grapeseed omega-3 content can be mentioned. Increases in the sum of SFA and PUFA were determined in the oils of enzyme pre-treated seeds. However, there were negligible decreases in the U/S ratios.

Carotenoid contents showed ambiguous results with regards to enzyme pre-treatment. Enzyme pre-treatment did not reveal a significant change in carotenoid content of flax seed. However, it caused a significant increase in apricot kernel and a significant decrease in grape seed in terms of carotenoid content.

The ratio between the individual representatives of tocopherols in the cold-pressed seed oils practically did not differ. When tocopherols were examined, the most prominent effect of enzyme treatment was on β-tocopherol. The contents of β-tocopherol in flax and apricot seed oils were lower as a result of enzyme application. However, while the β-tocopherol content increased in grape seed oil, at the same time, the α-tocopherol content decreased, although a significant change occurred only in grape seed oil in terms of this tocopherol homologue.

Total phenolics in flaxseed and grapeseed oil exhibited similar changes with enzyme treatment. The phenolic contents in these oil samples showed non-significant decreasing results, but the total phenolic content in apricot oil was almost doubled. Oilseed extracts generally contain high levels of phenolics and have shown excellent antioxidant activities in both vivo and vitro (Schmidt et al., 2003). However, depending on whether the phenolics are free or bound, only a small proportion can pass into the oil, and most phenols remain in the meal. In recent years, many studies have been conducted in order to increase the phenol content in seed oils (Schmidt and Pokorný, 2005).

Enzyme application did not show a significant effect in terms of antioxidant activity in grape seed oil. In fact, the effect of enzyme application in terms of antioxidant activity from the total fraction was not found statistically significant for all oil seed samples. In this regard, we can say that the effect of enzyme application varied according to the oilseed. That is, hydrophilic and lipophilic-induced antioxidant activity decreased in flaxseed oil. On the contrary, antioxidant activities from these fractions increased by up to 100% in the case of apricot seed oil. Considering the seed grain structure, there is a hard exocarp in flax and grape seed compared to apricot. Therefore, we are of the opinion that the enzymes can penetrate the tissues of apricot kernel more effectively. Here, we see that the DPPH-RSA analysis carried out where the oil is directly used as the sample does not reflect antioxidant capacity sufficiently. Because, according to the DPPH-RSA analysis results we performed directly in oil, enzyme application resulted in a slight increase, but this increase level was not even statistically significant. Phenolic compounds were also affected by enzyme pre-treatment in a similar manner to the DPPH-RSA results. The phenolic content of apricot kernel oil obtained by enzyme application is almost twice (x1.68) that of the control. In contrast to this situation, there was only a small decrease in the contents in phenolic substances in flax and grape seed oils, which were statistically insignificant. On the other hand, the results also revealed the direct relationship between total phenolic content and antioxidant capacity from hydrophilic and lipophilic fractions but not with the total antioxidant capacity determined in the oil. Symoniuk, Ratusz, Ostrowska-Ligęza, and Krygier (2018) reported that none of the quality characteristics affected the oxidative stability of cold-pressed oils, as determined by the Rancimat test. They associated the induction times of oils with their fatty acid compositions.

Pearson correlation coefficients among the concentrations of bioactive components, antioxidative activitiy percentages as variables and oxidation parameters (p-anisidine value and peroxide value), are given in Table 3. The results showed that correlation relationships changed according to the oilseed source. In terms of peroxide value, a negative correlation was observed only with γ-tocopherol in flaxseed oil (r = -59.2); whereas apricot and grape seed oils showed negative correlations between peroxide number and hydrophilic/lipophilic antioxidant capacities, total phenolics and total carotenoid contents. In addition, a negative correlation was found for grape seed oil with α-tocopherol (r = -69.2). These variables correlated with much higher ratios [r values between (-80.4) - (-97.5)] in apricot kernel oil compared to grape seed and flaxseed oils.

p-Anisidine value was negatively correlated with total carotenoids for flaxseed oil (r = -70.3), to total antioxidant capacity for grapeseed oil (r = -79.7). A high negative correlation (r = -92.2) could be seen between p-anisidine value and δ-tocopherol for grapeseed oil. For apricot seed oil, there were negative poor correlations between total antioxidant capacity and total carotenoids with p-anisidine value.