Sunflower seeds were purchased at a local market in Umuarama (Paraná - Brasil). Alcalase^® 2.4L FG (endo-protease that hydrolyzes most peptide bonds within a protein) with an activity of 2.4 U·mL^-1 (Unit defined by the hydrolysis of casein to produce 1 mmole of tyrosine per minute at pH 7.5 and 37 °C), was provided by LNF Latino Americana. The reagents used to adjust the pH were sodium hydroxide (Nuclear, 95%) and Chloridric acid (Nuclear, 40%) and n-hexane was used to determine the non-lipid fraction in the free oil (Panreac, Castellar del Vallès, Barcelona). The solvents ethanol (95%, Anhydrol, Diadema, São Paulo, Brazil) and n-hexane (Panreac, Castellar del Vallès, Barcelona) were used in the oil recovery tests. The fatty acid profile was determined using methanol (≤ 99.9%, Panreac, Castellar del Vallès, Barcelona), sodium hydroxide (≤ 97%, Anidrol, Diadema, São Paulo, Brazil), a boron trifluoride-methanol solution (BF₃, B1252, Sigma-Aldrich, St. Louis, MO, USA) and heptane (Neon, Suzano, São Paulo, Brazil). The contents of phytosterols, tocopherols and free fatty acids were determined using N,O-Bis (trimethylsilyl) trifluoroacetamide with 1% trimethylchlorosilane (BSTFA, 15238, Sigma-Aldrich, St. Louis, MO, USA), heptane (Neon, Suzano, São Paulo, Brazil) and 5-α-Cholestane (≥ 97.0%, C8003, Sigma-Aldrich, St. Louis, MO, USA) and methyl heptadecanoate (≥ 99.0%, 51633, Sigma-Aldrich, St. Louis, MO, USA) as internal standards. Acylglycerols were quantified by external standardization using monoolein (purity ≥ 99%), 1,3-diolein (purity ≥ 99%) and glyceryl trioleate (purity ≥ 99%), purchased from Sigma-Aldrich (Saint Louis, Missouri, United States).

Sunflower seeds with moisture 4.3% ± 0.20 (determined in an oven at 105 °C) were crushed using a household blender and then the particles obtained were classified (set of Tyler-type sieves, Bertel) and a fraction with an average diameter of 0.725 mm was used in the extractions.

In the experiments which evaluated thermal pre-treatment effect of the seeds, before crushing and granulometric classification, the methodology described by Tian et al. (2019) was adopted. Thus, whole seeds were immersed in distilled water in the ratio of 1:3 (w/v). After 3 hours of immersion at room temperature, the excess water was removed and the seeds (150 g) were spread in a thin layer under a metal sieve. The sieve was put in an oven with air circulation (Marconi, Model MA035) at 120 °C for 60 minutes.

The maximization of enzymatic oil extraction from sunflower seeds with the enzyme Alcalase^® 2.4L FG was carried out using the Box-Behnken experimental design with three factors, three levels and four central points. The values of the three variables evaluated were: temperature (A) 40, 50 and 60 °C; pH (B) 7.0, 8.0, and 9.0 and enzyme concentration (C) 1, 5 and 9% (v/v). The three values for the variables corresponded to the levels: -1 (low), 0 (central point) e +1 (high), respectively. The values adopted for the independent variables took into account the Novozyme (2019) indications, which report temperature and pH range for optimal activity of the Alcalase^® 2.4L FG enzyme at 30 to 65 °C and pH 7.0 to 9.0, respectively. Regarding the enzyme concentration evaluated (in relation to the extraction medium volume), it was based on the enzymatic aqueous oil extractions performed by Jiang et al. (2010), Ribeiro et al. (2016) and Meng et al. (2018).

The enzymatic extraction was carried out in Erlenmeyer flasks (125 mL) the crushed sunflower seeds (10 g) and distilled water (40 mL), in the mass ratio of 1:5 (g of seeds/g of water), proportions used according to the study of Aquino et al. (2019). In sequence, the pH was adjusted according to the condition to be evaluated with NaOH solution (1 mol·L^-1) and enzyme added in the concentration of the test. Flasks were put in an orbital shaker (Marconi, model MA 830/A) at 180 rpm, for 5 hours with controlled temperature according to the experimental design.

After extraction, the free oil was recovered using the steps and conditions described by Aquino et al. (2019). Therefore, the suspensions had pH adjusted to 5.0 and the flasks were incubated for 1 hour under shaking 180 rpm at 25 °C. Then, the samples were stored overnight in a refrigerator (Consul, 340) at 4 °C. After that, the samples were centrifuged twice at 2700 rpm for 15 minutes at room temperature and four phases were formed (solid, aqueous, emulsion and free oil). The free oil in the upper phase was transferred to a petri dish and kept in an oven until reaching constant weight. To assess the free oil content in the upper phase, the determination of the non-lipid fraction in the sample was performed as described by Rodriguez et al. (2021), obtaining 4.7% ± 0.71 of non-lipid compounds from this phase. The free oil yield (FOY) was calculated by Equation 1:

The analysis of variance (ANOVA) was used to evaluate the effects of independent variables on the dependent variable (free oil yield). The experimental data were adjusted to a second-order polynomial mode, using the Statistica^® 8.0 software. The generalized model used is expressed in Equation 2:

Where β₀, β_i, β_ii and β_ij are the regression coefficients (β₀ = constant term; β_i = linear term; β_ii = quadratic term; β_ij = linear interaction term) and Y is the response variable (free oil yield - FOY) observed in the experiments. X_i and X_j are independent variables: temperature, pH and enzyme concentration.

The experimental conditions provided the maximum FOY in the evaluated experimental range as determined by Equation 2. In this condition, verification experiments (triplicate) were carried out to evaluate the predictive capacity of Equation 2 and the experimental results were submitted to the Student’s t-test (Excel^®, 2016) to estimate differences between the experimental values and the predictions. In this same experimental condition, experiments were carried out in triplicate to verify the efficiency of the solvent recovery of the oil from the phases. 5 mL of solvent (ethanol or n-hexane) were added in the step where four phases were formed, and the flasks were shaken to ensure homogenization and in the sequence centrifuged at 2700 rpm for 15 minutes. This procedure was performed three times for each solvent. The oil with solvent was transferred to a petri dish, and kept in an oven until reaching constant weight.

The influence of the thermal pretreatment of sunflower seeds was evaluated in the experimental conditions which give maximum FOY, as obtained from Equation 2. The extractions (destructive) were performed in triplicate at the times 60, 120, 180, 240, 300 and 360 minutes for seeds with and without treatment.

A gas chromatograph coupled to the mass spectrometer (GC-MS) (Shimadzu, model CG-2010 Plus, Tokyo, Japan) and equipped with an automatic injector (Shimadzu, model AOC-20i, Tokyo, Japan) was used to analyze the fatty acid profile and contents in phytosterols, tocopherols and free fatty acids.

For the fatty acid composition, the oil was previously prepared according to the procedure described by Stevanato and Silva (2019). 1.5 mL of a 0.5 mol·L^-1 methanolic sodium hydroxide solution were placed in a test tube. The tube was shaken vigorously and heated in a thermostatic bath (Nova Ética, model 314/8, Piracicaba, São Paulo, Brazil) at 100 °C for 100 min. Subsequently, 2 mL of derivative agent BF₃ were added and the tube was subjected to heating again for 5 min. A 5-mL aliquot of heptane was added to the test tube and after phase separation the supernatant was collected and sent for analysis. The analytes were separated in a capillary column DB-Wax™ (Shimadzu, 30m × 0.25mm × 0.25 μm, Tokyo, Japan), using helium as carrier gas (1.0 mL·min^-1). The temperature of the injector, the ionic source and the interface were 250, 260 and 250 °C, respectively. The initial temperature of the column was 80 °C, which was elevated to 180 °C at a rate of 10 °C·min^-1, and then elevated again to 240 °C at 4 °C·min^-1, remaining constant for 2 min at this temperature. Fatty acid methyl esters (FAMEs) were identified using the NIST Spectrum Library Spectrum Library (version 2014). The FAME quantification was performed from the normative area of the chromatographic peaks, using the percentage of the relative area of each peak in relation to the sum of all peaks.

The contents of phytosterols, tocopherols and free fatty acids were determined using BSTFA/TMSC as derivatizing agent, following the method described by Stevanato and Silva (2019). A SH-Rtx-5MS™ capillary column (Shimadzu, 30m × 0.25mm × 0.25 μm, Tokyo, Japan) was used to elute the compounds. The injection temperature was 280 °C. The oven was initially operated at 150 °C with an increase in temperature to 230 °C (10 °C·min^-1), then the temperature was increased again to 280 °C (15 °C·min^-1), and kept constant for 25 min. The identification was carried out as previously described and quantification was performed by internal standardization using 5-α-cholestane (5 mg·mL^-1) and methyl heptadecanoate (5 mg·mL^-1) as reference standards.

The acylglycerol composition was determined on a gas chromatograph (Shimadzu, GC-2010 Plus, Tokyo, Japan) equipped with flame ionization detector (FID), on-column injector (Shimadzu, Tokyo, Japan) and capillary column Zebron™ ZB-5HT inferno (Phenomenex, 10 m×0.32 mm×0.10 m, Torrance, CA, USA). The chromatographic conditions were previously described by Stevanato and Silva (2019). Chromatographic areas generated from the standard solutions of triolein (0.1 to 3.1 mg·mL^-1), diolein (0.03 to 2.5 mg·mL^-1), and monolein (0.05 to 2 mg·mL^-1) were plotted against the concentration to obtain the line equations (R² > 0.99).

The oxidative stability of the oil was determined using the Professional Rancimat Biodiesel equipment (Metrohm, model 823, Herisau, Switzerland). The samples (2.5 ± 0.003 g) were exposed to an air flow of 20 L·h^-1 at a constant temperature of 130 °C. The secondary oxidation products were transferred to the measuring vessel containing 50 mL of distilled water. The induction period was automatically determined by the equipment, and measured from the increase in thermal conductivity of the distilled water.

To verify the influence of heat pre-treatment on sunflower seeds and oil quality, the results were evaluated by analysis of variance (ANOVA) and Tukey’s test with a significance level of 5% (α=0.05), using the Statistica^® 8.0 software. All treatments and analyses were performed at least in duplicate (n=4).

The experimental condition and the free oil yield (FOY) obtained from each experimental condition, adopted in the experimental design, are presented in Table 1. Based on the results from this table, it can be verified that FOY varied from 7.02 to 15.59% and to identify the influence of each variable and its interactions on the response variable, the coded variable was adjusted to a second-order polynomial equation as expressed in Equation 3:

Table 2 presents the ANOVA results which were used to validate the second-order polynomial model (Equation 3) adjusted to the experimental data, as well as to evaluate the influence of each variable on the response. The results showed that the model was significant (p < 0.05) only for the linear effects of the three variables and quadratics of the temperature variable (A). For the interactions among them, the model was significant only for the interaction of the varying temperature (A) and enzyme concentration (C). These results are shown by high values for F and low values for p.

According to the ANOVA data, F_calc (123.28) was higher than F_tab (3.33); thus the polynomial model was valid in relation to the experimental data. The coefficient of determination (R²) and the adjusted coefficient of determination (R²Adj) of the model were 0.984 and 0.976, respectively, which indicates a high degree of correlation between the observed and predicted values.

In order to assess the adequacy of the predicted model, diagnostic graphs were generated, and are shown in Figure 1. Figure 1a shows good agreement between the experimental and predicted data, since the data are close to the straight line, which indicates a satisfactory fit. The graph of normal probability of the residues (Figure 1b) confirms the assumption of normality of the residues. This fact is characterized by the position of the points near the straight line, indicating that the errors are normally distributed. In the graph of raw residuals versus predicted values (Figure 1c), it can be seen that most of the residues were randomly distributed around zero, showing that there is no pattern of behavior between them and that the variance was constant. Thus, the information presented in Figure 1 confirms the adequacy of the model, as well as the validity of its predictions.

The chemical composition of the oil obtained from sunflower seeds with and without heat pre-treatment obtained with 3 and 4 hours extraction time, respectively, is shown in Table 3. Oleic, linoleic, palmitic and stearic acids were the four main fatty acids present in the oil, with a predominance of oleic. According to Codex Alimentarius standards, sunflower oil obtained from EAE is classified as mid-oleic acid, whose content in this fatty acid is in the range of 43.1-71.8% (Codex Alimentarius, 2019). The heat pre-treatment of the seeds did not influence the fatty acid profile of the oil, which showed high levels of monounsaturated fatty acids (~48%) and polyunsaturated fatty acids (~40%); while the saturated fatty acid content was low (~12%). The higher proportion of oleic acid than linoleic acid is an advantage to the quality of the oil. Smith et al. (2007) showed that sunflower oil with high oleic acid content has better oxidative and thermal stability than regular sunflower oil with high linoleic acid content (71.6%).

As shown in Table 3, four phytosterols (campesterol, stigmasterol, γ-sitosterol and β-sitosterol) and one tocopherol (α-tocopherol) were identified in sunflower oil. Among the phytosterols, β-sitosterol was the major component, representing on average 67.19% of the total composition. The high concentration of phytosterols in the oil can promote anti-lipid and hypolipidemic effects when ingested (Dai et al., 2013). In addition, phytosterols have antioxidant activity based on electron transfer and also have the ability to scavenge free radicals (Liu et al., 2019), attenuating lipid oxidation. The α-tocopherol present in the samples can also increase the oxidative stability of the oil due to its antioxidant capacity, which interrupts the chain reaction of free radicals (Liu et al., 2021). The heat pre-treatment slightly reduced (~10.54%) the total phytosterol content, due to the lower levels of stigmasterol, γ-sitosterol and β-sitosterol quantified in the oil. This effect can be attributed to the thermal degradation caused by heating during the pre-treatment. It is known that phytosterols undergo oxidation when subjected to heating, as reported by Chen et al. (2020) who studied the thermo-oxidative stability of soy germ phytosterols and reported that heating the oil to 120 °C for 60 min promoted a loss of ~8% in these phytosterols, which is in accordance with this study. However, α-tocopherol was not influenced by seed pre-treatment. The values obtained for phytosterol content were higher than those reported using the conventional method with n-hexane (Aquino et al., 2019) after 8 hours’ extraction.

Sunflower oil had a low free fatty acid content (>0.6%), which indicates the absence of hydrolytic reactions responsible for causing rancidity and decomposing triglycerides (Goszkiewicz et al., 2020). Sunflower oil contains an average of ~92.11% total acylglycerols, which shows that aqueous enzymatic extraction showed high selectivity to this oil. The thermal pre-treatment did not modify the composition of acylglycerols, indicating that the heating time of the seeds was insufficient to hydrolyze the triacylglycerides into smaller components (MG and DAG).

The oil obtained in the present study had a longer period of induction compared to the studies by Ghosh et al. (2019) (0.56 h) and Ramos et al. (2020) (1.47 h), who evaluated the oxidative stability of sunflower oil at 130 °C. The high oxidative stability of sunflower oil can be attributed to the presence of active compounds, such as phytosterols and tocopherols. In addition, the mild conditions applied in the oil extraction of the oil can contribute to resistance to thermal oxidation. The oils obtained from seeds with and without heat treatment showed similar thermal stability. This result was expected, since susceptibility and oxidative resistance is mainly affected by the chemical composition of the oil and in this study the oil obtained from raw and pre-treated seeds showed similar compositions.