Effects of Processing Conditions on the Simultaneous Extraction and Distribution of Oil and Protein from Almond Flour

The enzyme-assisted aqueous extraction process (EAEP) is an environmentally friendly strategy that simultaneously extracts oil and protein from several food matrices. The aim of this study was to investigate the effects of pH (6.5–9.5), temperature (45–55 ◦C), solids-to-liquid ratio (SLR) (1:12–1:8), and amount of enzyme (0.5–1.0%) on the extraction and separation of oil and protein from almond flour using a fractional factorial design. Oil and protein extraction yields from 61 to 75% and 64 to 79% were achieved, respectively. Experimental conditions resulting in higher extractability were subsequently replicated for validation of the observed effects. Oil and protein extraction yields of 75 and 72% were achieved under optimized extraction conditions (pH 9.0, 50 ◦C, 1:10 SLR, 0.5% (w/w) of enzyme, 60 min). Although the use of enzyme during the extraction did not lead to significant increase in extraction yields, it did impact the extracted protein functionality. The use of enzyme and alkaline pH (9.0) during the extraction resulted in the production of more soluble peptides at low pH (5.0), highlighting possible uses of the EAEP skim protein in food applications involving acidic pH. The implications of the use of enzyme during the extraction regarding the de-emulsification of the EAEP cream warrant further investigation.


Introduction
The intrinsic composition of almonds, which has long been recognized as nutritious and healthy, has enabled its use in a wide range of applications such as nutritive snacks, bakery, and confectionery ingredients, and as a feedstock to produce almond milk, yogurt, and oil [1][2][3]. The California almond industry accounts for approximately 80% of the world's almond supply and nearly 100% of the almond supply in the United States, with an estimated production of 2.2 billion pounds for 2017 [2,4].
Major downstream processing challenges in the fractionation of oil-bearing materials into added-value compounds (oil, protein, and carbohydrates) are usually associated with the simultaneous attainment of high extraction yields and adequate product functionality [5,6]. In general, oil-bearing materials can be extracted by the use of screw pressing or by solvent extraction. While the former leads to the production of specialty oils, extraction yields (70-90%) are not as high as the ones obtained by The liquid phase was placed in a separatory funnel and allowed to settle overnight at 4 °C. After settling, the liquid phase was separated into three fractions (skim-protein-rich fraction, cream-oil-rich fraction, and free oil). Insoluble, skim and cream fractions were stored at −20 °C for subsequent analysis. Extractions were performed in triplicate for each enzyme.
Skim, cream, insoluble, and starting material were evaluated for oil, protein, and solids content. Total oil and protein extraction yields, which represent all oil and protein extracted from the almond flour, and the distribution of the extracted compounds among the fractions generated by the EAEP were calculated as described by Souza et al. [16]. Total oil extraction yield (TOE), oil distribution in the fractions (free oil yield, oil yield in the cream, oil yield in the skim, and oil yield in the insoluble), total protein extraction yield (TPE), and protein distribution in the fractions (protein yield in the cream, protein yield in the skim, and protein yield in the insoluble) were determined according to Equations (1), (2), (3) and (4), respectively.  The liquid phase was placed in a separatory funnel and allowed to settle overnight at 4 • C. After settling, the liquid phase was separated into three fractions (skim-protein-rich fraction, cream-oil-rich fraction, and free oil). Insoluble, skim and cream fractions were stored at −20 • C for subsequent analysis. Extractions were performed in triplicate for each enzyme.
Skim, cream, insoluble, and starting material were evaluated for oil, protein, and solids content. Total oil and protein extraction yields, which represent all oil and protein extracted from the almond flour, and the distribution of the extracted compounds among the fractions generated by the EAEP were calculated as described by Souza et al. [16]. Total oil extraction yield (TOE), oil distribution in the fractions (free oil yield, oil yield in the cream, oil yield in the skim, and oil yield in the insoluble), total protein extraction yield (TPE), and protein distribution in the fractions (protein yield in the cream, protein yield in the skim, and protein yield in the insoluble) were determined according to Equations (1)-(4), respectively. TOE (%) = 100 − Oil (g) in the insoluble f raction Oil (g) in the almond f lour × 100 (1) Oil distribution in each f raction (%) = Oil (g) in each f raction * Oil (g) in the almond f lour × 100 (2) Processes 2019, 7, 844 4 of 19 TPE (%) = 100 − Protein (g) in the insoluble f raction Protein (g) in the almond f lour × 100 (3) Protein distribution in each f raction (%) = Protein (g) in each f raction * Protein (g) in the almond f lour × 100 (4) where fraction * corresponds to cream, skim or insoluble. The enzyme resulting in higher oil and protein extraction yields and/or better distribution of the extracted compounds (i.e., higher amount of oil in the cream or in the form of free oil, which would lead to a skim fraction with a lower oil content) was selected to further evaluate how extraction conditions such as pH, amount of enzyme, temperature, and SLR influence overall extractability and distribution of oil and protein from almond flour in the EAEP.

Effects of Extraction Variables on the Enzyme-assisted Aqueous Extraction Process (EAEP) of Almond Flour
Understanding the effects of processing variables is of key importance when aiming to reduce the number of experiments and improve process efficiency and productivity [17]. An experimental design was used to evaluate how operational parameters (pH, amount of enzyme, temperature, and SLR ratio) can affect the extraction and separation of oil and protein from almond flour. Because of the high number of variables, a fractional factorial design (2 4-1 , plus three central points, totaling 11 experiments) was used for the evaluation of the effects of the variables on the responses.
The effects of the independent variables (pH: 6.5-9.5, temperature: 45-55 • C, SLR: 1:12-1:8, and amount of enzyme: 0.5-1.0%) were evaluated on the extraction and distribution of oil and protein from almond flour ( Table 1). The variable levels used in the experimental design were selected based on preliminary tests as well as on existing literature data [15,18]. Dependent variables (i.e., evaluated response) were total oil extraction yield (TOE) (%), oil yield in the cream (OYC) (%), oil yield in the skim (OYS) (%), oil yield in the insoluble (OYI) (%), free oil yield (FOY) (%), total protein extraction yield (TPE) (%), protein yield in the cream (PYC) (%), protein yield in the skim (PYS) (%) and protein yield in the insoluble (PYI) (%). Initial evaluation of the extraction kinetics was performed at the central point (pH 8.0, 50 • C, 1:10 SLR and 0.75% (w/w) of enzyme), with time points ranging from 1 to 5 h, to assess the best extraction time range to be evaluated during the fractional design. Based on the extraction kinetics data at the central point, which demonstrated no significant increase in oil and protein extractability when reaction time increased from 1 to 5 h, extractions were performed for 40 min (to verify the possibility of achieving high extraction yields at reaction time < 1 h) and 3 h (evaluated as a conservative approach) for each experimental condition established by the experimental design.
Best experimental conditions identified by the factorial design were validated in triplicate for the enzyme assisted extraction process (EAEP) along with the aqueous extraction process (AEP) which was used as a control (same conditions without the use of enzyme). Approximately 50 g of the AEP and EAEP skim fractions obtained in the experimental validation of the best extraction conditions were freeze-dried on a benchtop freeze dryer VirTis-BenchTop™ "K" Series (SP-Scientific, Gardiner, NY, USA) and stored at -20 • C for subsequent functionality assays.

Proximate Analysis
Oil, protein, and dry matter contents were determined in all fractions: skim, insoluble, cream, and almond flour (starting material). Oil content was determined by using the acid hydrolysis Mojonnier method (AOAC method 922.06) [19], protein content by using the Dumas combustion method and a nitrogen conversion factor of 5.18 (Vario Max Cube, Elementar Analysensysteme GmbH, Langenselbold, Germany), and total solids (dry matter) by weighing after drying the samples in a vacuum oven (AACC Method 44-40) [20]. Extraction yields were expressed as percentages of each component in each fraction relative to the initial amounts in the almond flour according to Equations (1) to (4). All analyses were conducted in duplicate for each sample and a mass balance was provided for all extracted compounds.

Degree of Hydrolysis
The degree of hydrolysis (DH) of AEP and EAEP skim fractions was evaluated by the o-phthaldialdehyde (OPA) method described by Nielsen et al. [21]. Briefly, 400 µL of skim was added to 3 mL of freshly prepared OPA reagent. The mixture was vortexed and let stand for 2 min at room temperature before measuring the absorbance at 340 nm. A 0.9516 meqv/L L-serine solution was used as standard. A blank solution was prepared with distilled water instead of sample and used as the reaction control. Protein quantification (%) was determined by the Dumas combustion method (nitrogen conversion factor of 5.18) and the DH was determined as described below (Equation (5)): where h is the number of hydrolyzed bonds and h tot is the total number of peptide bonds per protein equivalent (7.58 eq/kg for almond protein [22]).

Low Molecular Weight (MW) Polypeptide Profile Characterization of AEP and EAEP Skim Proteins by SDS-PAGE
SDS-PAGE was used to determine the low MW protein profile of AEP and EAEP skims obtained under experimental validation of best extraction conditions. Skim fractions were mixed with (1:1, v/v) Laemmli solution, vortexed and placed in a water bath (95 • C, 5 min), as described by Laemmli [23]. A Tris-HCl buffer (25 mM Tris, 192 mM glycine, 0.1% SDS, pH 8.3) (Bio Rad, Hercules, CA, USA) was used as a running buffer. Electrophoretic separation of proteins was performed by loading 30 µg of protein/well onto a precast 12% acrylamide gel (CriterionTM TGX Precast Gels, Bio Rad, Hercules, CA, USA). Electrophoretic separation was carried out at 200 V at room temperature for 1 h. A low MW range SDS-PAGE standard (14.4-97.4 kDa) (Bio Rad, Hercules, CA, USA) was used. Relative quantification and polypeptide distribution were performed using a Gel DOCTM EZ Imager system and Image Lab software (Bio-Rad, Hercules, CA, USA).

AEP and EAEP Skim Protein Solubility
Protein solubility of the AEP and EAEP freeze-dried skim fractions obtained by experimental validation of best extraction conditions was determined as described by Rickert et al. [24] with few modifications. A total of 10 mL of a 1% (w/v) skim solution diluted in DI water was placed in a 30 mL beaker and the pH of the protein solution was adjusted to 5.0 and 9.0 by adding 1 M HCl or 1 M NaOH solution. The dispersions were stirred for 1 h at room temperature and then centrifuged at 10,000 × g at 20 • C for 10 min. The protein content of the supernatant was measured using the Biuret method. The protein quantification was performed using a standard curve ranging from 0 to 18 mg/mL of bovine serum albumin (R 2 = 0.997). The total protein content was measured after solubilizing the samples in a 1 M NaOH solution [25]. All samples were analyzed in triplicate. The solubility (%) was determined as follows (Equation (6)): Protein in the supernatant (mg /mL) Total protein (mg /mL) × 100 (6)

Statistical Analyses
Replicates of each measurement were analyzed by using Analysis of Variance (ANOVA) with generalized linear models from the SAS system (version 9.4, SAS Institute Inc., Cary, NC, USA). One-way ANOVA was used for evaluating the effects of reaction time on oil and protein extraction yields at the central point condition. Two-way ANOVA was used for evaluating the effects of different extraction conditions at different reaction times during the experimental validation of best extraction conditions. Multiple comparisons of least-square means were made by Tukey's adjustment with level of significance set at p < 0.05. Multivariate statistical analyses were performed using the Protimiza Experimental Design ® Software (Protimiza Experimental Design, Campinas, Brazil) and the significance of the effects was determined by the t-test and p-value, which was set at p < 0.05.

Enzyme Selection for the Enzyme-assisted Extraction Process of Almond Flour
Enzymatic hydrolysis has shown to increase the extraction of oil and protein from many oil-bearing materials, the magnitude of which depending on the type and source of the enzyme used and on the treatments to which the material has been previously subjected to (e.g., mechanical disruption and heat treatments) [12,18,26]. The performance of NP2M and NPL in regards to total oil and protein extractability and distribution of the extracted compounds in each phase generated by the process is shown in Figure 2A

AEP and EAEP Skim Protein Solubility
Protein solubility of the AEP and EAEP freeze-dried skim fractions obtained by experimental validation of best extraction conditions was determined as described by Rickert et al. [24] with few modifications. A total of 10 mL of a 1% (w/v) skim solution diluted in DI water was placed in a 30 mL beaker and the pH of the protein solution was adjusted to 5.0 and 9.0 by adding 1 M HCl or 1 M NaOH solution. The dispersions were stirred for 1 h at room temperature and then centrifuged at 10,000 × g at 20 °C for 10 min. The protein content of the supernatant was measured using the Biuret method. The protein quantification was performed using a standard curve ranging from 0 to 18 mg/mL of bovine serum albumin (R 2 = 0.997). The total protein content was measured after solubilizing the samples in a 1 M NaOH solution [25]. All samples were analyzed in triplicate. The solubility (%) was determined as follows (Equation (6)):

Statistical Analyses
Replicates of each measurement were analyzed by using Analysis of Variance (ANOVA) with generalized linear models from the SAS system (version 9.4, SAS Institute Inc., Cary, NC, USA). Oneway ANOVA was used for evaluating the effects of reaction time on oil and protein extraction yields at the central point condition. Two-way ANOVA was used for evaluating the effects of different extraction conditions at different reaction times during the experimental validation of best extraction conditions. Multiple comparisons of least-square means were made by Tukey's adjustment with level of significance set at p < 0.05. Multivariate statistical analyses were performed using the Protimiza Experimental Design ® Software (Protimiza Experimental Design, Campinas, Brazil) and the significance of the effects was determined by the t-test and p-value, which was set at p < 0.05.

Enzyme Selection for the Enzyme-assisted Extraction Process of Almond Flour
Enzymatic hydrolysis has shown to increase the extraction of oil and protein from many oilbearing materials, the magnitude of which depending on the type and source of the enzyme used and on the treatments to which the material has been previously subjected to (e.g., mechanical disruption and heat treatments) [12,18,26]. The performance of NP2M and NPL in regards to total oil and protein extractability and distribution of the extracted compounds in each phase generated by the process is shown in Figure 2A,B.  The use of proteases to assist the extraction did not lead to significant increment in TOE and TPE yields (Figure 2A,B), compared to the control (without enzyme). Similar oil extraction yields were observed when using both proteases to assist the extraction compared with the control (73-76 vs. Processes 2019, 7, 844 7 of 19 74%) ( Figure 2A). However, a significant increase (p < 0.05) in free oil yield was observed when using both proteases compared with the control. Free oil yield increased from 0.7 ± 0.3% (control) to 8 ± 1% and 6 ± 4% when NP2M and NPL were used, respectively ( Figure 2A). Higher free oil yield when using proteases during the extraction is likely the result of the action of the proteases on the cream protein, which can be further hydrolyzed thus improving the release of the entrapped oil in the cream. The higher amount of free oil obtained when using both proteases is evidenced by the slightly lower oil yield in the cream (58-61%) compared with the control (68%), although this difference was not statistically significant at p < 0.05. Reduced oil yield in the skim (6-9%) was observed for all treatments, not being statistically different at p < 0.05 ( Figure 2A).
As observed for TOE, similar protein extraction yields were obtained by the AEP and EAEP. Protein extraction yields of 64 ± 3 and 69 ± 3% were achieved when NPL and NP2M were used to assist the extraction, respectively ( Figure 2B). It is worth mentioning that the use of proteases during the extraction led to a significantly lower protein yield in the cream fraction compared with the control (2 vs. 3%), which might affect the resistance of the EAEP cream towards subsequent de-emulsification (i.e., the release of the entrapped oil from the emulsion). Because the use of NP2M resulted in higher free oil yield and slightly higher protein yield in the skim as evidenced by the reduced amount of protein in the cream, this enzyme was selected for the subsequent study where a broader range of reaction parameters was investigated.

Effects of Extraction Parameters on the Enzyme-assisted Aqueous Extraction Process (EAEP) of Almond Flour
A preliminary evaluation of the extraction kinetics at the central point (pH 8.0, 50 • C, 1:10 SLR, and 0.75% of enzyme) was performed with the objective of identifying the appropriate reaction time range to be evaluated in the fractional factorial design. Extractions were performed for 1, 2, 3, 4, and 5 h. The use of pH 8.0 in the central point, instead of pH 9.0 as used during the enzyme-screening step, had the objective of avoiding the use of excessive alkaline pH (above 10) in the + 1 level in the fractional factorial design. As observed in Figure 3A,B, increasing extraction time from 1 to 5 h did not significantly improve oil and protein extraction yields at p < 0.05. Based on these results, the effects of the variables ranging from levels −1 and +1 on extraction yields of oil and protein were evaluated at 40 min and 180 min (3 h). Reaction time of 40 min was selected to evaluate if similar extraction yields could be achieved at shorter reaction times (<1 h), while 3 h was selected as a conservative strategy to further confirm the trend observed. The use of proteases to assist the extraction did not lead to significant increment in TOE and TPE yields (Figure 2A,B), compared to the control (without enzyme). Similar oil extraction yields were observed when using both proteases to assist the extraction compared with the control (73-76 vs. 74%) ( Figure 2A). However, a significant increase (p < 0.05) in free oil yield was observed when using both proteases compared with the control. Free oil yield increased from 0.7 ± 0.3% (control) to 8 ± 1% and 6 ± 4% when NP2M and NPL were used, respectively ( Figure 2A). Higher free oil yield when using proteases during the extraction is likely the result of the action of the proteases on the cream protein, which can be further hydrolyzed thus improving the release of the entrapped oil in the cream. The higher amount of free oil obtained when using both proteases is evidenced by the slightly lower oil yield in the cream (58-61%) compared with the control (68%), although this difference was not statistically significant at p < 0.05. Reduced oil yield in the skim (6-9%) was observed for all treatments, not being statistically different at p < 0.05 ( Figure 2A).
As observed for TOE, similar protein extraction yields were obtained by the AEP and EAEP. Protein extraction yields of 64 ± 3 and 69 ± 3% were achieved when NPL and NP2M were used to assist the extraction, respectively ( Figure 2B). It is worth mentioning that the use of proteases during the extraction led to a significantly lower protein yield in the cream fraction compared with the control (2 vs. 3%), which might affect the resistance of the EAEP cream towards subsequent deemulsification (i.e., the release of the entrapped oil from the emulsion). Because the use of NP2M resulted in higher free oil yield and slightly higher protein yield in the skim as evidenced by the reduced amount of protein in the cream, this enzyme was selected for the subsequent study where a broader range of reaction parameters was investigated.

Effects of Extraction Parameters on the Enzyme-assisted Aqueous Extraction Process (EAEP) of Almond Flour
A preliminary evaluation of the extraction kinetics at the central point (pH 8.0, 50 °C, 1:10 SLR, and 0.75% of enzyme) was performed with the objective of identifying the appropriate reaction time range to be evaluated in the fractional factorial design. Extractions were performed for 1, 2, 3, 4, and 5 h. The use of pH 8.0 in the central point, instead of pH 9.0 as used during the enzyme-screening step, had the objective of avoiding the use of excessive alkaline pH (above 10) in the + 1 level in the fractional factorial design. As observed in Figure 3A,B, increasing extraction time from 1 to 5 h did not significantly improve oil and protein extraction yields at p < 0.05. Based on these results, the effects of the variables ranging from levels −1 and +1 on extraction yields of oil and protein were evaluated at 40 min and 180 min (3 h). Reaction time of 40 min was selected to evaluate if similar extraction yields could be achieved at shorter reaction times (<1 h), while 3 h was selected as a conservative strategy to further confirm the trend observed.  SLR, 0.75% enzyme). No significant increment in oil extractability was observed when reaction time increased from 40 min to 3 h. Total oil extraction yields of 71 ± 4% and 73 ± 1% were achieved at 40 and 180 min, respectively ( Figure 4A). In addition to achieving high extraction yields, the form in which the extracted oil is present (i.e., as free oil or entrapped in the cream fraction) is of great relevance when considering subsequent recovery and utilization of the extracted oil. At 40 min, higher free oil yields (8-16%) were observed at experiments #2 (pH 9.5, 45 • C, 1:12 SLR, 1.0% enzyme), #3 (pH 6.5, 55 • C, 1:12 SLR, 1.0% enzyme), and #4 (pH 9.5, 55 • C, 1:12 SLR, 0.5% enzyme) ( Figure 4E). Overall, free oil yield decreased when extraction time increased from 40 to 180 min, corresponding to an increase in the amount of oil in the cream fraction ( Figure 4B) and reduction in the amount of oil yield in the skim ( Figure 4C). These results indicate that emulsion formation was favored with increasing reaction time, thus leading to less free oil and more oil entrapped in the cream fraction. Oil extraction yields and its distribution among the fractions generated for each experimental condition evaluated in the fractional factorial are described in Figure 4A-E. Higher oil extraction yields were observed for experiments performed at the central point conditions (#9-11) (pH 8.0, 50 °C, 1:10 SLR, 0.75% enzyme). No significant increment in oil extractability was observed when reaction time increased from 40 min to 3 h. Total oil extraction yields of 71 ± 4% and 73 ± 1% were achieved at 40 and 180 min, respectively ( Figure 4A). In addition to achieving high extraction yields, the form in which the extracted oil is present (i.e., as free oil or entrapped in the cream fraction) is of great relevance when considering subsequent recovery and utilization of the extracted oil. At 40 min, higher free oil yields (8-16%) were observed at experiments #2 (pH 9.5, 45 °C, 1:12 SLR, 1.0% enzyme), #3 (pH 6.5, 55 °C, 1:12 SLR, 1.0% enzyme), and #4 (pH 9.5, 55 °C, 1:12 SLR, 0.5% enzyme) ( Figure 4E). Overall, free oil yield decreased when extraction time increased from 40 to 180 min, corresponding to an increase in the amount of oil in the cream fraction ( Figure 4B) and reduction in the amount of oil yield in the skim ( Figure 4C). These results indicate that emulsion formation was favored with increasing reaction time, thus leading to less free oil and more oil entrapped in the cream fraction. , and free oil (%) (E). Experimental conditions: #1 (pH 6.5, 45 °C, 1:12 SLR, 0.50% enzyme (w/w*)); #2 (pH 9.5, 45 °C, 1:12 SLR, 1.0% enzyme (w/w)); #3 (pH 6.5, 55 °C, 1:12 SLR, 1.0% enzyme (w/w)); #4 (pH 9.5, 55 °C, 1:12 SLR, 0.50% enzyme (w/w)); #5 (pH 6.5, 45 °C, 1:8 SLR, 1.00% enzyme (w/w)); #6 (pH 9.5, 45 °C, 1:8 SLR, 0.50% enzyme (w/w)); #7 (pH 6.5, 55 °C, 1:8 SLR, 0.50% enzyme (w/w)); #8 (pH 9.5, 55 °C, 1:8 SLR, 1.00% enzyme (w/w)); #9, 10 and 11 (pH 8.0, 50 °C, 1:10 SLR, 0.75% enzyme (w/w)); * w/w, weight of enzyme/weight of almond flour. , and free oil (%) (E). Experimental conditions: #1 (pH 6.5, 45 • C, 1:12 SLR, 0.50% enzyme (w/w*)); #2 (pH 9.5, 45 • C, 1:12 SLR, 1.0% enzyme (w/w)); #3 (pH 6.5, 55 • C, 1:12 SLR, 1.0% enzyme (w/w)); #4 (pH 9.5, 55 • C, 1:12 SLR, 0.50% enzyme (w/w)); #5 (pH 6.5, 45 • C, 1:8 SLR, 1.00% enzyme (w/w)); #6 (pH 9.5, 45 • C, 1:8 SLR, 0.50% enzyme (w/w)); #7 (pH 6.5, 55 • C, 1:8 SLR, 0.50% enzyme (w/w)); #8 (pH 9.5, 55 • C, 1:8 SLR, 1.00% enzyme (w/w)); #9, 10 and 11 (pH 8.0, 50 • C, 1:10 SLR, 0.75% enzyme (w/w)); * w/w, weight of enzyme/weight of almond flour. The estimated effects of the variables evaluated (pH, temperature, SLR, and enzyme) on oil extraction yields as well as on the distribution of the extracted compounds among the fractions are shown in Table 2. The estimated effects of the extraction parameters evaluated on the extraction and separation of oil from almond flour are shown in Table 2. Overall, total oil extraction was not significantly affected by the reaction parameters in the range evaluated, regardless of the extraction time evaluated. According to the effects described in Table 2, reduced oil yield in the skim could be achieved by the use of higher SLR (1:8) (effect value −3.89, p < 0.05) and lower pH values (effect value of 2.6, p < 0.05), at 40 min of extraction. The curvature analysis revealed that the curvature for TOE and OYS at 40 min of extraction time was significant and positive, indicating that higher yields were observed around the central point conditions. For all responses evaluated, the experiments performed at the central point conditions (experiments #9-11) presented a low relative standard deviation (1 to 8%), indicating good reproducibility of the experimental data.
Protein extraction yields and its distribution among the fractions generated by the process (skim, cream, insoluble) for all experimental conditions evaluated in the fractional factorial are described in Figure 5A-D.
around the central point conditions. For all responses evaluated, the experiments performed at the central point conditions (experiments #9-11) presented a low relative standard deviation (1 to 8%), indicating good reproducibility of the experimental data.
When evaluating the effects of the use of enzyme on protein extractability, it can be observed that, except by the higher protein yield in the cream at 40 min (p < 0.05), increasing the amount of enzyme from 0.5 to 1.0% did not result in a significant increase in protein extractability nor shifted the protein distribution among the fractions, regardless the extraction time used (Table 3). Within the range evaluated, our results indicate that the use of 0.5% enzyme is sufficient to achieve high protein extraction yields.
Increasing the slurry pH from 6.5 to 9.0 did not significantly alter protein extractability, regardless of the extraction time evaluated. Taking into account that the enzyme used in this study is stable at pH values ranging from 5.5 to 9.0 and that almond proteins have a higher solubility at higher pH [29,30], increased protein extractability at higher pH would be expected. In addition to its effect on overall extractability, pH can also impact the extracted protein functionality [31], as described in Section 3.4. Therefore, extraction yields should not be the only response evaluated when selecting the best extraction conditions. At 180 min, the amount of protein in the cream fraction was significantly affected by pH (−1.02), the negative effect indicating that when pH increases from 6.5 to 9.0, the protein yield in the cream decreases. Considering the well-known ability of proteins to interact with oil and to form stable emulsions [32], it becomes necessary to understand the effects of the cream composition regarding its resistance to subsequent de-emulsification, which is beyond the scope of this work.
Increasing the extraction temperature from 45 to 55 • C did not significantly increase TPE, regardless of the extraction time evaluated. While a positive trend was observed when using higher temperatures at 40 min, a negative trend was observed at 180 min, which could be due to possible enzyme or protein denaturation when exposed at 55 • C for longer times. However, these effects were not statistically significant at p < 0.05. Our results are in agreement with the ones presented by Esteban et al. [28] which reported that a temperature of 50 • C increased the solubility of almond protein while the use of higher temperatures (50-75 • C) could promote partial coagulation of the protein which would, in turn, reduce its solubility.
Overall, the effects observed indicate that lower values of temperature, SLR (higher amount of water) and amount of enzyme could result in the production of a cream fraction with a lower protein content at 40 min. Since proteins can act as strong emulsifiers [33], a lower protein concentration in the cream might be desirable when considering the subsequent de-emulsification step to which the cream needs to be subjected to release the entrapped oil. The same conditions leading to reduced protein yield in the cream could be exploited to achieve high protein extraction yields at 40 min at any pH value in the range evaluated.
Overall, the effects described in Tables 2 and 3 indicate that all processing variables (pH, temperature, SLR, and enzyme) could be used at their lowest level (−1.0) to achieve high oil and protein extraction yields at 40 min. However, the use of a higher SLR (1:10 or 1:8) could enable the production of a cream fraction with lower protein content and a skim fraction with lower oil content. In order to further confirm the observed effects of the extraction parameters evaluated, extraction conditions were replicated as described in the section below.

Experimental Validation of the Best Extraction Conditions
Because increasing reaction time from 40 to 180 min did not result in higher extractability, the validation of the best conditions identified in the fractional factorial design was performed at 20, 40 and 60 min to maximize process productivity. Based on the estimated effects (Tables 2 and 3) and preliminary conditions used during the enzyme screening, two experimental conditions were selected for the validation experiments, including a control experiment for each treatment (AEP, without enzyme). Experimental conditions and treatments are described as the following: (i) Condition 1: pH 6.5, 1:8 SLR, 45 • C and 0.5% of enzyme for the EAEP1, and without enzyme for the AEP1. (ii) Condition 2: pH 9.0, 1:10 SLR, 50 • C and 0.5% of enzyme for the EAEP2, and without enzyme for the AEP2.
Extraction yields and distribution of the extracted compounds for the EAEP 1 and EAEP 2 , with their respective controls, are shown in Tables 4 and 5. Total oil extraction did not significantly increase when extraction time increased from 20 to 60 min for EAEP 1 (pH 6.5, 0.5% NP2M, 1:8 SLR, and 45 • C) (56 ± 2% vs. 61 ± 2 %) ( Table 4). The use of a more alkaline pH (9.0), an intermediate SLR (1:10), and a slightly higher temperature (50 • C) (EAEP 2 ) led to higher oil extraction yields compared to the ones obtained in EAEP 1 (75% for EAEP 2 vs. 61% EAEP 1 ). However, similar oil extraction yields were observed within each condition, with or without enzyme use, indicating that the use of enzyme did not play a key role in improving oil extraction yields compared with the other parameters evaluated such as pH. Overall, increasing reaction time from 20 to 60 min significantly increased total oil extraction for the EAEP 2 (62 vs. 75%).
The highest oil yield in the skim (7.8 ± 2.0%) was observed for EAEP 1 (pH 6.5, 0.5% NP2 M, 1:8 SLR, and 45 • C) at 60 min, being statistically higher than the 4.0% observed for the control experiment (AEP 1 ). Slightly lower oil yields in the skim (4.7 and 4.3%) were achieved for EAEP 2 (pH 9.0, 0.5% N2 MP, 1:10 SLR, and 50 • C) and AEP 2 , respectively. It is worth mentioning that the separation of the fractions (skim, cream, and free oil) at lower pH (6.5) is challenging compared with the use of pH 9.0, which might be related to the higher oil content of the skim fractions produced by the EAEP 1 . Overall, cream fractions with higher oil yields were obtained from the EAEP 2 (63-66%) compared with the EAEP 1 (51-47%). Similar free oil yields (3-6%) were obtained, regardless of the treatment used.
Protein extraction yields increased from 55 ± 5 to 68 ± 5 % when reaction time increased from 20 to 60 min for the EAEP 1 (Table 5), being significantly higher than the 55 ± 4% obtained for the control experiment at 60 min (AEP 1 ). Overall, high protein extraction yields were achieved at shorter reaction times for the EAEP 2 compared with the EAEP 1 (70 vs. 55% at 20 min and 72 vs. 63% at 40 min). For the EAEP 2 and its control (AEP 2 ), increasing extraction time from 20 to 60 min did not significantly increase protein extraction. Our results evidence the key role of extraction pH on overall protein extractability. At lower pH (6.5), where protein solubilization is not as favored as compared with a more alkaline pH (9.0) [31], the use of enzyme has a positive effect on improving extraction yields compared with the control. However, at alkaline pH (9.0), where almond protein solubilization is favored, the use of enzyme does not play a key role in overall protein extractability. Based on the validation experiments, high protein and oil extraction yields (70%) could be achieved in short reaction time when using pH 9.0, 1:10 SLR, and 50 • C. While protein extraction yields of approximately 70% could be achieved at 20 min, similar oil extraction yields would require 40-60 min of extraction.
The comparison of our data with the ones in the literature is challenging due to differences in the starting materials used and lack of studies focusing on the simultaneous extraction of oil and protein from almond flour. Oil extraction yields obtained in our study (75% for EAEP 2 and 70% for AEP 2 ) are similar or higher to the ones reported by Sharma and Gupta [14], where oil extraction yields ranging from 37-40% were reported for the AEP (pH 4-9, 18 h reaction time) and 75-77% for the EAEP process (40 • C, 4-18 h, and use of 416 U of the enzyme Protizyme TM). Balvardi et al. [13] reported a 77.8% oil extraction recovery from almond flour when extractions were performed at pH 5.76, 50 • C, reaction time of 4 h, and a mixture of proteases and cellulases was used at 1% level. However, in their study oil recovery yields were calculated in relation to the ones obtained using hexane extraction, while herein extraction yields were calculated in relation to the initial amount of oil present in the almond flour that can be extracted, making difficult to make a direct comparison of the results from both studies. No protein extraction yields were reported for both studies. The simultaneous extraction of oil and protein from the almond press cake has been recently reported by Souza et al. [16]. In their study, oil extraction yields of 42 and 50% were achieved for the AEP and EAEP, respectively. In regards to protein extraction, 70 and 75% of the almond cake protein was achieved for the AEP and EAEP, respectively. However, it is worth mentioning the use of almond press cake in their study and almond flour in ours.  1 2.7 ± 0.4 a 2.7 ± 0.6 a 3.6 ± 0.5 a EAEP 2 1.6 ± 0.1 a 1.5 ± 0.1 a 1.6 ± 0.1 a AEP 2 1.8 ± 0.2 a 1.9 ± 0.2 a 2.0 ± 0.3 a Means with different lowercase letters indicate statistical difference at p < 0.05. Statistical significance differences were denoted by different letters, with the letter "a" being assigned to the highest value EAEP 1 : pH 6.5, 0.5% of enzyme, 1:8 SLR, and 45 • C; EAEP 1 control (AEP 1 ): pH 6.5, 1:8 SLR, and 45 • C; EAEP 2 : pH 9.0, 0.5% of enzyme, 1:10 SLR, and 50 • C; EAEP 2 control (AEP 2 ): pH 9.0, 1:10 SLR, and 50 • C.
Overall, our results demonstrate that experimental validation of the best conditions identified in the fractional factorial design enabled the identification of processing parameters leading to high extraction yields of oil and protein at very short extraction time compared with existing data in the literature (40-60 min vs. 4-18 h). However, because similar extraction yields could be achieved at different reaction conditions (acidic and alkaline pH, with and without enzyme), although at the expense of longer reaction times, the impact of the best extraction conditions on the extracted protein solubility was evaluated.

Molecular Weight Polypeptide Profile, Degree of Hydrolysis (DH) and Solubility of AEP and EAEP Skims
The use of proteases during the extraction process leads to the breakdown of peptide bonds thus resulting in increased concentration of primary amines which corresponds to an increase in the degree of hydrolysis (DH) [34]. The functionality of the protein hydrolysate is tied to the nature and the composition of peptides generated during hydrolysis [35]. Therefore, understanding how extraction conditions impact the DH and protein functionality becomes necessary to better select processing conditions that will lead to the production of proteins with desired technological functions. Figure 6 shows the impact of the extraction conditions evaluated during the experimental validation of the AEP and EAEP on the DH and protein profile of AEP and EAEP skims. As expected, no changes in the DH of AEP skim proteins were observed (2%) when reaction time increased from 20 to 60 min for Conditions 1 and 2 ( Figure 6). However, extraction time had a significant impact on the DH for the EAEP skim. The DH of the EAEP 1 (pH 6.5, 0.5% enzyme, 1:8 SLR, 45 • C) and EAEP 2 (pH 9.0, 0.5% enzyme, 1:10 SLR, 50 • C) skim proteins increased from 18 to 25% and 7 to 12% when reaction time increased from 20 to 60 min, respectively. The higher DH of the EAEP 1 skim is likely due to the higher enzyme activity at pH 6.5 (95% of its maximum activity), compared with its activity at pH 9.0 (50% of its maximum activity, as described by the manufacturer). Our results are in agreement with the literature where the use of enzyme for a longer time, until a certain extent, resulted in higher DH [11,16,34].
the DH and protein functionality becomes necessary to better select processing conditions that will lead to the production of proteins with desired technological functions. Figure 6 shows the impact of the extraction conditions evaluated during the experimental validation of the AEP and EAEP on the DH and protein profile of AEP and EAEP skims. As expected, no changes in the DH of AEP skim proteins were observed (2%) when reaction time increased from 20 to 60 min for Conditions 1 and 2 ( Figure 6). However, extraction time had a significant impact on the DH for the EAEP skim. The DH of the EAEP1 (pH 6.5, 0.5% enzyme, 1:8 SLR, 45 °C) and EAEP2 (pH 9.0, 0.5% enzyme, 1:10 SLR, 50 °C) skim proteins increased from 18 to 25% and 7 to 12% when reaction time increased from 20 to 60 min, respectively. The higher DH of the EAEP1 skim is likely due to the higher enzyme activity at pH 6.5 (95% of its maximum activity), compared with its activity at pH 9.0 (50% of its maximum activity, as described by the manufacturer). Our results are in agreement with the literature where the use of enzyme for a longer time, until a certain extent, resulted in higher DH [11,16,34]. Figure 6. Impact of the extraction conditions evaluated during the experimental validation of the AEP and EAEP on the degree of hydrolysis (%) of skim proteins. Data were analyzed by one-way ANOVA followed by Tukey's post-hoc test. Different letters indicate a significant difference between samples at p < 0.05. Condition 1: pH 6.5, 0.5% enzyme, 1:8 SLR, and 45 °C for the EAEP1 and pH 6.5, 1:8 SLR, and 45 °C for the AEP1. Condition 2: pH 9.0, 0.5% enzyme, 1:10 SLR, and 50 °C for the EAEP2 and pH 9.0, 1:10 SLR, 50 °C for the AEP2. Figure 6. Impact of the extraction conditions evaluated during the experimental validation of the AEP and EAEP on the degree of hydrolysis (%) of skim proteins. Data were analyzed by one-way ANOVA followed by Tukey's post-hoc test. Different letters indicate a significant difference between samples at p < 0.05. Condition 1: pH 6.5, 0.5% enzyme, 1:8 SLR, and 45 • C for the EAEP 1 and pH 6.5, 1:8 SLR, and 45 • C for the AEP 1 . Condition 2: pH 9.0, 0.5% enzyme, 1:10 SLR, and 50 • C for the EAEP 2 and pH 9.0, 1:10 SLR, 50 • C for the AEP 2 .
The DH values described in Figure 6 are in agreement with the SDS-PAGE peptide profile of AEP and EAEP skim proteins (Figure 7). Similar electrophoretic profile was observed for AEP skim proteins, regardless of the extraction condition used. This trend is in agreement with the constant DH of these samples (2%). AEP skim proteins are mainly composed of two polypeptide fragments (40 kDa (25%, α-subunit) and 24 kDa (30%, β-subunit)) which correspond to the subunits of amandin, the major protein accounting for 65-70% of extractable almond protein [36] (Figure 7). Our results are in agreement with the ones reported by Wolf and Sathe [29] and Derbyshire et al. [37], which demonstrated that amandin (62 to 66 kDa subunits) could be converted into acidic (20 kDa) and basic (40 kDa) polypeptides in the presence of mercaptoethanol. The EAEP skim protein profile evidences increased protein hydrolysis as extraction time increased from 20 to 60 min for both conditions evaluated (from 18 to 25% for EAEP 1 and from 7 to 13% for EAEP 2 ). At 60 min of extraction, the EAEP 1 showed the highest amount of peptides < 14.4 kDa (80%), indicating that the major polypeptides were broken down into smaller fractions. demonstrated that amandin (62 to 66 kDa subunits) could be converted into acidic (20 kDa) and basic (40 kDa) polypeptides in the presence of mercaptoethanol. The EAEP skim protein profile evidences increased protein hydrolysis as extraction time increased from 20 to 60 min for both conditions evaluated (from 18 to 25% for EAEP1 and from 7 to 13% for EAEP2). At 60 min of extraction, the EAEP1 showed the highest amount of peptides < 14.4 kDa (80%), indicating that the major polypeptides were broken down into smaller fractions. Figure 7. SDS-PAGE protein profile of the skims from AEP and EAEP process. Condition 1: pH 6.5, 0.5% enzyme, 1:8 SLR, and 45 °C for the EAEP1 and pH 6.5, 1:8 SLR, and 45 °C for the AEP1. Condition 2: pH 9.0, 0.5% enzyme, 1:10 SLR, and 50 °C for the EAEP2 and pH 9.0, 1:10 SLR, 50 °C for the AEP2.
Since extraction conditions such as pH, enzyme use, and reaction time can significantly affect protein functionality [5,16], we determined how the extraction conditions evaluated during the experimental validation of the AEP and EAEP (Conditions 1 and 2) affected the solubility of the AEP and EAEP skim proteins. With the goal of identifying possible applications for the extracted protein, skim protein solubility was assessed at acidic pH (5.0, isoelectric point for almond protein [37]) and alkaline pH (9.0, extraction pH) ( Figure 8A,B). EAEP1 and AEP1 skim proteins (Condition 1) had very similar solubilities at pH 5.0 ( Figure 8A), suggesting no increment in protein solubility when using enzyme and extraction pH 6.5. However, the use of enzyme and alkaline pH (9.0) during the extraction (Condition 2) resulted in the production of a skim fraction with increased solubility at pH 5.0 ( Figure 8B). EAEP2 skim protein solubility was significantly higher than the solubility of AEP2 skim proteins (45 vs. 23%), regardless of the extraction time evaluated. Higher solubility of EAEP2 skim proteins at pH 5.0, where almond protein solubility is unfavorable [35], is likely related to the enzymatic release of smaller and more soluble peptides [5,34,38]. As shown in the SDS-PAGE gel, nearly complete hydrolysis of amandin, an insoluble protein under acidic pH, was observed in the EAEP skim for both conditions studied. The hydrolysis Figure 7. SDS-PAGE protein profile of the skims from AEP and EAEP process. Condition 1: pH 6.5, 0.5% enzyme, 1:8 SLR, and 45 • C for the EAEP 1 and pH 6.5, 1:8 SLR, and 45 • C for the AEP 1 . Condition 2: pH 9.0, 0.5% enzyme, 1:10 SLR, and 50 • C for the EAEP 2 and pH 9.0, 1:10 SLR, 50 • C for the AEP 2 .
Since extraction conditions such as pH, enzyme use, and reaction time can significantly affect protein functionality [5,16], we determined how the extraction conditions evaluated during the experimental validation of the AEP and EAEP (Conditions 1 and 2) affected the solubility of the AEP and EAEP skim proteins. With the goal of identifying possible applications for the extracted protein, skim protein solubility was assessed at acidic pH (5.0, isoelectric point for almond protein [37]) and alkaline pH (9.0, extraction pH) ( Figure 8A,B). of amandin into smaller peptides could explain the higher solubility of the EAEP skim at acidic pH compared with the AEP skim. Our results are in agreement with the literature where the use of proteases resulted in the production of proteins with higher solubility at lower pH [5,16,34,38]. However, despite the higher extent of hydrolysis of EAEP1 skim proteins (Condition 1), compared with EAEP2 skim proteins (Condition 2), reduced solubility was observed for EAEP1 skim proteins. Reduced solubility of EAEP1 skim proteins is likely the result of excessive degree of hydrolysis which might have resulted in the exposure of hydrophobic groups initially buried in the core of the protein [35,39]. This trend, however, was not observed at pH 9.0, where almond protein solubility is generally favored [31]. AEP and EAEP skim protein solubilities were very similar at pH 9.0 for both extraction conditions (98% for Condition 1 vs. 99.5% Conditions 2, at 60 min of extraction time). Our results are in agreement with those reported by Souza et al. [16] in which significantly higher solubility of EAEP skim proteins extracted from almond cake was observed at pH 5.0, compared with the solubility of AEP skim proteins. Similar solubilities were also reported for AEP and EAEP skim proteins at pH 9.0.

Conclusions
The use of a fractional factorial design and subsequent validation of the effects of important reaction parameters on the extraction and separation of oil and protein from almond flour led to high Figure 8. Impact of extraction conditions evaluated during the experimental validation of the AEP and EAEP on the solubility of AEP and EAEP skim proteins at acidic and alkaline pH. Condition 1 (A): pH 6.5, 0.5% enzyme, 1:8 SLR, and 45 • C for the EAEP 1 and pH 6.5, 1:8 SLR, and 45 • C for the AEP 1 . Condition 2 (B): pH 9.0, 0.5% enzyme, 1:10 SLR, and 50 • C for the EAEP 2 and pH 9.0, 1:10 SLR, 50 • C for the AEP 2 . Data were analyzed by a two-way ANOVA followed by Tukey's post-hoc test. Different letters indicate a significant difference between samples at p < 0.05. EAEP 1 and AEP 1 skim proteins (Condition 1) had very similar solubilities at pH 5.0 ( Figure 8A), suggesting no increment in protein solubility when using enzyme and extraction pH 6.5. However, the use of enzyme and alkaline pH (9.0) during the extraction (Condition 2) resulted in the production of a skim fraction with increased solubility at pH 5.0 ( Figure 8B). EAEP 2 skim protein solubility was significantly higher than the solubility of AEP 2 skim proteins (45 vs. 23%), regardless of the extraction time evaluated. Higher solubility of EAEP 2 skim proteins at pH 5.0, where almond protein solubility is unfavorable [35], is likely related to the enzymatic release of smaller and more soluble peptides [5,34,38]. As shown in the SDS-PAGE gel, nearly complete hydrolysis of amandin, an insoluble protein under acidic pH, was observed in the EAEP skim for both conditions studied. The hydrolysis of amandin into smaller peptides could explain the higher solubility of the EAEP skim at acidic pH compared with the AEP skim. Our results are in agreement with the literature where the use of proteases resulted in the production of proteins with higher solubility at lower pH [5,16,34,38]. However, despite the higher extent of hydrolysis of EAEP 1 skim proteins (Condition 1), compared with EAEP 2 skim proteins (Condition 2), reduced solubility was observed for EAEP 1 skim proteins. Reduced solubility of EAEP 1 skim proteins is likely the result of excessive degree of hydrolysis which might have resulted in the exposure of hydrophobic groups initially buried in the core of the protein [35,39]. This trend, however, was not observed at pH 9.0, where almond protein solubility is generally favored [31]. AEP and EAEP skim protein solubilities were very similar at pH 9.0 for both extraction conditions (98% for Condition 1 vs. 99.5% Conditions 2, at 60 min of extraction time). Our results are in agreement with those reported by Souza et al. [16] in which significantly higher solubility of EAEP skim proteins extracted from almond cake was observed at pH 5.0, compared with the solubility of AEP skim proteins. Similar solubilities were also reported for AEP and EAEP skim proteins at pH 9.0.

Conclusions
The use of a fractional factorial design and subsequent validation of the effects of important reaction parameters on the extraction and separation of oil and protein from almond flour led to high extraction yields with a significant reduction of processing time. Although similar oil and protein extraction yields could be achieved at pH values ranging from 6.5-9.0 and SLR ranging from 1:8-1:10, the use of a more alkaline pH (9.0) resulted in significant reaction time reduction and improved separation and functionality of the extracted compounds. Experimental validation of best extraction conditions (pH 9.0, 0.5% of NP2M, 1:10 SLR, 50 • C, 60 min) resulted in 74.5% of oil and 72.3% of protein extraction. The use of enzyme to assist the extraction did not lead to significant increments in oil and protein extraction yields compared with the control (without enzyme), evidencing the major role of other processing parameters such as pH, SLR, and reaction time. However, the use of enzyme and alkaline pH (9.0) during the extraction resulted in the production of more soluble proteins for applications at acidic pH (5.0), compared when not using enzyme or using pH 6.5 during the extraction. At alkaline pH (9.0), high solubility was observed for both AEP and EAEP skim proteins (95%), regardless of the extraction conditions used. The higher concentration of soluble hydrolyzed peptides in the EAEP skim produced under alkaline extraction conditions could enable its use for specific food or feed applications involving acidic pH. However, it is important to highlight the increase in processing costs due to the use of enzyme to assist the extraction. Subsequent studies evaluating the use of reduced amounts of enzyme (<0.5%) during the extraction or enzyme recycling from the cream de-emulsification to the extraction are needed to minimize enzyme costs. In addition, the implications of the use of enzyme during the extraction on other functionality properties as well as on the resistance of the cream fraction produced by the EAEP process towards subsequent de-emulsification warrant further investigation.