Protein isolate of α-lactalbumin (Biopure-lactoalbumin^TM) was provided by Davisco Food International Inc. (Le Sueur, MN, USA). Alcalase was provided by Novozymes Biopharma US Inc (Alcalase^Ò 2.4 L, Proteinase from Bacillus Licheniformis, Subtilisin A). Buffer salts Na₂HPO₄ (Mallinckrodt) and NaH₂PO₄ came from J. T. Baker. Folin reagent was purchased from Sigma Aldrich (St. Louis, MO). For antioxidant assays: 2,20-azinobis-(3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt (ABTS), 6-hydroxy-2,5,7,8-tetramethylchroman-2-acid (Trolox), fluorescein (FL) disodium salt and 2,20-azobis(2-methylpropionamidine) dihydrochloride (AAPH) were obtained from Sigma Aldrich (St. Louis, MO), and potassium persulphate was from J. T. Baker. Histidil-hipuril-leucine (HHL) was purchased from Sigma Aldrich (St. Louis, MO) for performing angiotensin converting enzyme (ACE)-inhibition assay. To carry out digestion studies, α-amylase, pepsin and pancreatin came from Sigma Aldrich (St. Louis, MO).

To carry out cell studies, High-glucose Dulbecco’s modified Eagle medium (DMEM) with L-glutamine and pyruvate (Phenol red-DMEM), High-glucose Dulbecco’s modified Eagle medium without L-glutamine neither pyruvate (Phenol red-free DMEM), Dulbecco’s phosphate-buffered saline (DPBS) ± Ca²⁺ and Mg²⁺, Hank’s Balanced Salt Solution (HBSS), penicillin-streptomycin mixture, MEM non-essential amino acid and foetal bovine serum (FBS) were purchased from Life Technologies (Villebon-sur-Yvette, France). For cell metabolic activity determination (MTT assay), 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl tetrazolium bromide (MTT) and dimethyl sulfoxide (DMSO) were purchased from Sigma.

Optimization of hydrolysis process was carried out as in Fernández-Fernández et al. [10] where a composite central design was used [23] based on a response surface model, full factorial design. Briefly, arrangement of seven hydrolysates was generated. Enzymatic hydrolysis reaction was evaluated by determining ABTS and ORAC-FL values, as the response variables, with the variation of two factors as independent variables, enzyme:substrate ratio (r) (% w/w) and time (t) (minutes), at two levels each (0.0050% and 0.1000% w/w, 0 and 60 minutes, respectively), and three repeats of the central point (0.0525% w/w and 30 minutes). The equation for the proposed model of response variables ABTS and ORAC-FL (Yi) is shown in Equation (1):

Yi = β₀ + β₁r + β₂t + β_1,2r × t + ε (1)

where β₀ is the regression coefficient for the intercept point; β₁ and β₂ are the linear regression coefficients; β_1,2 is the regression coefficient for the interaction between the independent variables (factors r and t); and ε is the variable error. Model parameters were calculated with Statgraphic Plus version 5.1 program by multiple linear regression (MLR).

Enzymatic hydrolysis reaction was performed using Alcalase enzyme (1.158 mg/mL, enzymatic activity ≥ 2.4 AU/g) with 8% (w/V) of protein isolate (α-LA) in phosphate buffer solution 100 mM pH 7 and was incubated in a water bath at 30˚C with agitation of 150 rpm. After incubation time (as in Section 2.2), reaction was stopped by heating at 100˚C for 10 minutes. Ultrafiltration was carried out using an Amicon membrane (cut-off of 3 kDa, Merk Millipore) to separate peptides of low molecular weight (LMW) from high molecular weight (HMW). For each hydrolysate, two separate fractions were obtained: LMW and HMW fractions. From each of the seven samples, two fractions were obtained resulting in 14 samples (7 LMW and 7 HMW). Samples were frozen, lyophilized and stored at −20˚C for subsequent analysis.

Protein content determination was performed by Lowry method [24] using solutions of 0.3 mg/mL of each hydrolysate fraction prepared in phosphate buffer 10 mM pH 7.4. For solubility determination 10 mg/mL solutions of each hydrolysate fraction were prepared in the same phosphate buffer by 15 minutes of gentle magnetic agitation followed by centrifugation at 10,000 g for 10 minutes then determination of protein content in the supernatant by Lowry method.

SE-HPLC analysis was performed as described by Molina Ortiz [25] using a Shimadzu, SPD-20A detector and LC-10AT pump equipment detecting at 280 nm. Hydrolysates fractions were eluted in a Molecular Exclusion Column BioSep-Sec 2000 with an isocratic flow of 1 mL/min and phosphate buffer 50 mM (pH 6.8) 0.5% of SDS as mobile phase. Each sample was prepared in mobile phase in a concentration of 2 mg/mL. Hydrolysis percentage was determined by quantifying α-LA in each hydrolysate with α-LA calibration curve.

Selected sample with greater antioxidant capacity, sample 4 LMW was submitted to in vitro digestion (D 4 LMW) by the gastrointestinal digestion simulation model described by Hollebeeck et al. [17] . In vitro digestions were performed in 50 mL Erlenmeyer flasks in water bath at 37˚C and 200 rpm of agitation. This simulation model has three successive stages. First stage corresponds to salivary stage in which α-amylase stock solution (90 units/mL, 0.43 mL) and 10 mL of phosphate buffer 10 mM pH 6.9 were added to 200 mg of 4 LMW lyophilized hydrolysate. Incubation took place in a water bath at 37˚C for 5 minutes and 200 rpm of agitation. Stage two corresponds to gastric stage in which pepsin (71.2 units/mL) freshly prepared in HCl 0.1 M was added to the same phosphate buffer brought to pH 2.0 with HCl 1 M (22.73 mL final volume, 12.3 mL of volume added in the second stage). Incubation was carried out in the same conditions for 90 minutes. Third stage corresponds to duodenal stage in which pancreatin and bile extract (9.2 mg pancreatin and 55.2 mg bile extract/mL final concentrations) were added as well as the same phosphate buffer brought to pH 7.0 with NaHCO₃ 0.1 M, (30.09 mL final volume, 7.36 mL of volume added in the third stage). Incubation was carried out in the same conditions for 150 minutes. After finishing stage three, the enzymes were inactivated by heating in a water bath at 90˚C for 10 minutes. Then, digestion samples were centrifuged at 10,000 rpm for 10 minutes and supernatant was separated, frozen and lyophilized for further analysis.

Antioxidant capacity was determined by electron transfer (ET) and hydrogen atom transfer (HAT) methods ABTS and ORAC-FL, respectively. ABTS was performed based on the method described by Re et al. [26] with some modifications [10] measuring absorbance at 734 nm. Dose-response curves were constructed to calculate IC₅₀ from the curve % Inhibition vs. Protein (mg/mL) (from 0.25 to 5 mg/mL of protein). A logarithmic function was obtained in order to calculate the corresponding protein concentration which is able to inhibit 50% of ABTS radicals. Inhibition percentage was calculated according to Equation (2):

% Inhibition = A control − A antioxidant A control (2)

where A_control is the absorbance of 3 mL of ABTS in buffer with 30 μL of buffer and A_antioxidant is the absorbance of 3 mL of ABTS in the same phosphate buffer with 30 μL of Trolox or sample.

ORAC-FL was performed as described by Ou et al. [27] modified by Dávalos, Bartolomé and Gómez-Cordovés [28] . In this assay a fluorescent probe (FL) is oxidized by an oxygen radical generator (AAPH) lowering its fluorescence intensity. Fluorescence measurements were displayed at 485 nm and 520 nm of excitation and emission wavelengths, respectively, at 37˚C for 104 minutes in the equipment Varioskan^Ò Flash. Briefly, hydrolysates fractions and reagents were prepared in phosphate buffer 75 mM pH 7.4. Each well had a final volume of 200 μL: 120 μL of 1.17 mM fluorescein solution (70 nM final concentration), 60 μL of AAPH (12 mM final concentration) and 20 μL of antioxidant substance (Trolox or sample). All samples’ solutions were prepared at least in duplicate and each one of the preparations was tested at least in triplicate. The area under the curve (AUC) of Fluorescence vs Time was calculated according to Equation (3):

AUC = 1 + ∑ i = 1 i = 104 f i / f 0 (3)

where f₀ is the fluorescence at 10 minutes of incubation at 37˚C and f_i is the fluorescence measured every minute, for 104 minutes. Curves of Fluorescence vs Time were normalized to the curve of the blank calculating the net AUC as the difference between AUC_antioxidant (Trolox or sample) minus AUC_blank. Trolox calibration curve (net AUC_Trolox vsnTE (μmol TE, Trolox equivalents)) was constructed in order to calculate samples antioxidant capacity (μmol TE/mg of protein). Besides punctual measurements, IC₅₀ values were obtained by constructing the curve nTE vs [Protein] (mg/mL) to obtain a logarithmic function from which it could be calculated the concentration of protein correspondent to 50% inhibition of peroxyl radicals.

Antihypertensive activity was determined as described by Cushman and Cheung [29] modified by Kim et al. [30] which consists of evaluating the inhibition of angiotensin converting enzyme (ACE). Briefly, samples were prepared as described by Fernández-Fernández et al. [10] in a concentration of 5 mg/mL. Assay buffer (borate buffer 0.2 M and NaCl 2 M, pH 8.3), mili Q water and HHL 5 mM was added to the sample. Eppendorf tubes were incubated at 37˚C, 700 rpm, for 5 minutes, and ACE was added in cold (0 - 10 mU). Same tubes were incubated at 37˚C, 700 rpm, for 30 minutes and then incubated at 90˚C for 10 minutes. Finally, colour reagent and potassium phosphate buffer 0.2 M pH 8.3 were added, samples were centrifuged at 20˚C, 6000 rpm, for 10 minutes and supernatant absorbance was measured at 382 nm. ACE inhibition percentage was calculated as shown in Equation (4):

% ACE Inhibition = 100 ∗ [ 1 − ( A S − A 0 S A max − A 0 max ) ] (4)

where A_s is the absorbance of sample with ACE, A_0s is the absorbance of sample without ACE, A_max is the absorbance in the absence of sample and A_0max is the absorbance without sample and ACE.

Cell metabolic activity (or cell viability; MTT assay) of TC7-cells was determined after 1 h incubation in the presence of HBSS (control), the hydrolysate or the in vitro digestion of the hydrolysate that presented greater antioxidant capacity, sample 4 LMW and its digestion (D 4 LMW), respectively.

TC7-cells (passage 42 - 47) were routinely grown according to Benzaria et al. [31] [32] using 75 cm² cell culture flasks in phenol red-DMEM culture medium supplemented with penicillin-streptomycin, MEM non-essential amino acids and heat-inactivated FBS. TC7-cells were seeded in sterile 12-well Trans well plates with Thin Cert inserts at a density of 1.25 × 10⁵ cells/well and cultivated at 37˚C and 8% CO₂, 100% RH (relative humidity) for 17 days until cell-confluence was reached as assessed by transepithelial electrical resistance (TEER) measurement before deposing the samples on the cells.

Samples were prepared in HBSS at concentrations of 1, 5 and 10 mg/mL protein. For the hydrolysate (4 LMW) and its gastro-intestinal digestion (D 4 LMW), 26.35 and 61.26 mg of dry powder were weighted for 1 mL of HBSS, corresponding to 20 mg/mL of protein for both samples (75.9% and 32.65% of protein, respectively). A volume of 500 μL sample solutions were deposed on the cells and incubated for 1 h at 37˚C and 8% CO₂, 100% RH. After incubation, the apical TC7-culture media in the cell-wells were taken out and MTT assay was performed by adding 500 µL MTT reagent (0.15 mg/mL in FBS-free phenol red-free DMEM) to TC7-cells [31] [32] . MTT is reduced into Formazan^Ò by a succinate deshydrogenase in living cells. The cell ability to reduce MTT provides an indication of mitochondrial integrity, and therefore of cell metabolic activity or cell viability. After 3 h incubation with MTT, determination was carried out by taking out MTT reagent from the cell-wells and by adding 500 μL of DMSO to each well with subsequent incubation at 37˚C for 30 minutes for cell lysing and Formazan^Ò recovering.

Amounts of 100 µL lysate were then transferred into 96-well plates to measure Formazan^Ò absorbance at 570 nm in a microplate reader, after half diluting with DMSO. Viability percentage was calculated by taking the absorbance value of the control (HBSS) as 100%.

All the measurements were determined at least in triplicate. Results were expressed as mean values ± standard deviation. One-way analysis of variance (ANOVA) and pos-hoc Tukey test was applied to determine significant differences between values (p < 0.05). Statistical analysis was done using Infostat v. 2015 and Statgraphic Plus v. 5.1 programs.

Protein content and hydrolysis percentage were determined in the low molecular weight (LMW) fractions of samples 1 to 7 with the exception of sample 3, where hydrolysis conditions led to gelation. Gelification probably took place be due to a combination of factors such as: lower enzyme to protein ratio; lower hydrolysis rate; time and temperature of hydrolysis. For this reasons results of sample 3 are not included in Table 1. It could be observed that protein content measurements tended to increase with factors time of hydrolysis and r (enzyme to substrate ratio, see Table 1). As expected, for ultrafiltrated LMW hydrolysates, hydrolysis percentages determined by SE-HPLC were approximately 0 for samples with 0 minutes of reaction; no significant differences (p < 0.05) were found between samples 5, 6 and 7 which have the same reaction conditions. Sample 4 LMW had no significant differences in hydrolysis percentage compared to samples

Results are expressed as the means ± SD (n = 6). ANOVA analysis was made by column using Tukey test. Means in the same column with different letters state significant differences (p < 0.05).

5, 6 and 7 stating that short peptides resulting from hydrolysis were separated correctly by ultrafiltration.

Other authors [33] treated whey protein isolate with chymotrypsin, pepsin and trypsin reaching lower percentages of hydrolysis. Alcalase hydrolysates appear to be more hydrolyzed than those of digestive enzymes, possibly because of its unspecificity [33] producing a great quantity of short peptides with bioactive properties [12] [34] . Nevertheless, digestive enzymes could enhance hydrolysis improving antioxidant capacity [10] . LMW hydrolysate from sample 4 to 7 showed similar % hydrolysis but in all cases higher than their non-ultrafiltrated counterparts which showed values below 31 [10] .

ABTS and ORAC-FL assays were the selected methods for evaluating the relationship between the factors enzyme:substrate ratio and time of reaction with the antioxidant capacity. The coefficients obtained from multiple linear regression analyses are shown in Table 2. ABTS and ORAC-FL response variables did not show similar behaviour. The coefficient values of both response variables showed a positive effect on the factor time of reaction and only a positive effect on factor enzyme:substrate ratio for ORAC-FL. The factor enzyme:substrate ratio was discarded for ABTS because of showing a p-value higher than 0.01 (non-significant). As to the model, for ABTS R² was close to 1 so it can be deducted that the variation of the antioxidant capacity with the factors fit the model adequately (p < 0.01). However, in the case of ORAC-FL R² was lower than for ABTS but the model is adequate (p < 0.01). Figure 1 shows surface plots of the antioxidant capacity by ABTS and ORAC-FL methods (response variables), as a function of enzyme:substrate ratio (0.0050% - 0.1000% w/w) and of time (0 - 60 minutes). Both plots show similar tendencies augmenting the

r: enzyme:substrate ratio; t: time; R²: regression coefficient; p: p-value for the unfit of the model (coefficients were p < 0.01).

antioxidant capacity with time, presenting a maximum for the conditions of sample 4 LWM (0.1000 w/w and 60 minutes). Sample 4 LMW antioxidant value was 1.574 ± 0.060 and 1.636 ± 0.076 μmol TE/mg of protein, compared to α-LA 0.191 ± 0.007 and 0.159 ± 0.011 μmol TE/mg of protein, for ABTS and ORAC-FL antioxidant capacity, respectively. Sample 4 LMW had 8.2 and 10.2 times more antioxidant capacity for ABTS and ORAC-FL, respectively, than α-LA. This sample values of antioxidant capacity differed greatly from other samples, showing increased antioxidant power. In our previous work, sample 4 presented 1.015 ± 0.042 and 1.495 ± 0.114 μmol TE/mg of protein in which case sample 4 LMW shows 1.55 and 1.09 times more antioxidant capacity than sample 4 (ABTS and ORAC-FL values, respectively). These values were not surprisingly higher than those of sample 4 for ORAC-FL value but for ABTS value antioxidant capacity increased, meaning separation by ultrafiltration seems to favor the concentration of peptides with ET mechanism for the neutralization of radicals. In addition, IC₅₀ values were obtained for sample 4 LMW (0.805 ± 0.035 and 0.065 ± 0.004 mg/mL of protein for ABTS and ORAC-FL, respectively) and α-LA (15.732 ± 0.256 and 0.223 ± 0.014 mg/mL of protein for ABTS and ORAC-FL, respectively). With these values 19.5 and 3.4 times more α-LA than sample 4 LWM to neutralize 50% of ABTS and peroxyl radicals, respectively. The latter confirms that sample 4 LWM had a higher antioxidant capacity which increases with the percentage of hydrolysis [34] ; this was confirmed by SE-HPLC (Section 3.1). Comparing sample 4 values before and after ultrafiltration it is clear that the process beneficiates the separation and concentration of the peptides that present higher antioxidant capacity (significant differences, p < 0.05), evidencing the importance of using enzymatic hydrolysis as a strategy for increasing the value of whey proteins. This suggests that short peptides liberated during hydrolysis and concentrated by ultrafiltration were responsible for the antioxidant capacity as in the work of Contreras et al. [3] . These values were similar to those of other enzymes using longer time of reaction [3] [33] , establishing Alcalase is more

efficient than other enzymes, attaining powerful antioxidant hydrolysates in less time of reaction and strengthening antioxidant capacity by obtaining short peptides through ultrafiltration.

Regarding antihypertensive properties measured in vitro, sample 4 LMW presented 22% of ACE inhibition percentage with respect to Captopril. This percentage is similar to the obtained for sample 4 without ultrafiltration, as reported by Fernández-Fernández et al. [10] (no significant differences). This could be explained by the fact that the peptides hydrolysis was not enough to release shorter peptides from α-LA which are known to be responsible for ACE inhibitory activity [5] . Studying the effect of digestion on sample 4 LMW by in vitro simulation to demonstrate better inhibition was also observed.

As described in section2.5, in vitro digestion of α-LA (D α-LA, control), sample 4 (D 4) and 4 LMW (D 4 LMW) were performed in order to evaluate the effect of digestion conditions on antioxidant capacity. Characterization of each digestion was performed by determining protein content and percentage of hydrolysis by HPLC analysis (as described in Section 2.4), and antioxidant capacity determination by ABTS and ORAC-FL assays described in Section 2.6. Percentages of protein content were 50.98% ± 1.80%, 41.48% ± 0.52% and 32.65% ± 1.79%, for the digestion of α-LA, sample 4 and 4 LMW, respectively. Hydrolysis percentages of the digestion of α-LA, sample 4 and 4 LMW were 35.59% ± 0.22%, 41.72% ± 0.13% and 80.89% ± 0.44%, respectively (Table 3).

According to the results obtained, it can be said that the increasing percentage of protein content of in vitro digestions resulted in diminished hydrolysis percentage. Hydrolysis percentage of the samples tends to increase with digestion related to samples without digestion (Table 3). In vitro digestion of sample 4 (D 4) showed higher hydrolysis percentage than α-LA digestion (D α-LA) but less than its respective LMW fraction. In vitro digestion of sample 4 LMW (D 4 LMW) showed 80% of hydrolysis (2 times fold compared to the digestion of sample 4), meaning ultrafiltration enhances hydrolysis rendering short peptides more available to digestive enzymes.

As to antioxidant capacity, α-LA had an ABTS value of 0.191 ± 0.007 µmol TE/mg of protein and for its in vitro digestion 0.619 ± 0.023 µmol TE/mg of protein increasing 3.2 times fold the antioxidant capacity (Figure 2). For both samples, 4 and 4 LMW, in vitro digestion enhanced antioxidant capacity but less than for α-LA. Sample 4 presented an ABTS value of 1.015 ± 0.042 which increased to 1.115 ± 0.049 µmol TE/mg of protein for its in vitro digestion (p < 0.05). Similarly, sample 4 LMW presented values of 1.574 ± 0.060 and 1.743 ± 0.086 µmol TE/mg of protein, respectively, with significant differences as well.

For ORAC-FL assay, α-LA presented values of 0.156 ± 0.011 µmol TE/mg of protein without in vitro digestion and 1.819 ± 0.141 µmol TE/mg of protein for

the in vitro digestion, showing a huge improvement on antioxidant capacity due to in vitro digestion (11.6 times fold). For sample 4, antioxidant capacity increased 1.4 times fold from 1.495 ± 0.114 (non-digested) to 2.046 ± 0.113 µmol TE/mg of protein after in vitro digestion showing less improvement than for α-LA. Similar results were observed for sample 4 LMW which presented a value of 1.636 ± 0.076 and for its in vitro digestion 2.542 ± 0.245 µmol TE/mg of protein (1.5 times fold).

In both cases (ABTS and ORAC-FL), samples 4 and 4 LMW did not show great improvements on antioxidant capacity which could be explained by the fact that these had already have enzymatic hydrolysis, releasing short peptides with antioxidant properties. Thus, in vitro digestion surely enhanced hydrolysis of all samples, increases being more evident for α-LA. Overall, samples with in vitro digestion showed increased hydrolysis as well as for non-digested 4 and 4 LMW (p < 0.05) with the associated improvement of antioxidant capacity making encrypted peptides more bioaccessible by in vitro digestion. These results agree with those of Adjonu et al. [35] , Ranamukhaarachchi et al. [36] , Saavedra et al. [21] and Tavares et al. [37] that highlighted the importance of enzymatic hydrolysis as the reason why antioxidant capacity increases due to release of encrypted short peptides release from α-LA native sequence. In our study, enzymatic hydrolysis was enhanced by gastrointestinal tract enzymes, improving antioxidant capacity. Thus, we aimed at understanding whether the digested hydrolysate provokes cell metabolic activity improvement. This would state that short peptides could be able to produce beneficial effects on intestinal epithelium. For this reason, cell metabolic activity on intestinal cells was measured.

MTT assay was performed on TC7-cells after 1 h incubation in the presence of sample 4 LMW and its in vitro digestion (D 4 LMW) counterpart standing for the hydrolysates and the in vitro digestion of the hydrolysates that presented the best antioxidant capacity, respectively. Results in Figure 3 show that cell metabolic activity of non-digested hydrolysates in concentrations 1, 5 and 10 mg/mL of protein did not differ from the control (HBSS) (p < 0.05) stating 4 LMW is not cytotoxic for intestinal cells at the tested concentrations. Conversely, digested hydrolysates presented increments at 5 mg/mL of protein implying an increase in the absorption of digested peptides (D 4 LMW) compared to non-digested peptides (4 LMW). Concentration of 10 mg/mL (4 LMW) did not differ from control samples but D 4 LMW decreased cell metabolic activity compared to concentration of 5 mg/mL. Moreover, concentration of 20 mg/mL (D 4 LMW) presented less cell metabolic activity than control meaning that

Concentrations higher than 5 mg/mL may overcharge cell metabolism and diminish cell metabolic activity (data not shown). Clearly, D 4 LMW had a positive effect on cell metabolism up to concentrations of 5 mg/mL. For non-digested 4 LMW, concentrations tested seemed not to beneficiate cell metabolism. This could be explained by the fact that sample 4 LMW contain longer peptides unable to enter the intestinal cells and increase cell metabolic activity (for 1 h time of exposure). Digestion appears to enhance bioaccessibility of peptides within 4 LMW structure with the consequent increase in the bioavailability. In other research, Xie et al. [38] studied the bioavailability of low molecular weight peptides of casein hydrolysates using Alcalase enzyme after simulated gastrointestinal digestion. In the latter work they found that digested Alcalase hydrolysates were bioavailable because of the low molecular weight peptides. Moreover, low molecular weight fraction was the most bioavailable of the fractions being in accordance with the increase detected in the present study for the D 4 LMW that suffered more hydrolysis and generated shorter peptides more easily absorbed. In addition, low molecular weight fraction was also found to present the highest antioxidant capacity after simulated gastrointestinal digestion and absorption in TC7-cells for which D 4 LMW could be able to exert its antioxidant effect after absorption or over intestinal cells helping to maintain an antioxidant environment in the lumen of the intestine. Overall, these results state that D 4 LMW enhanced cell metabolism showing relevance for intestinal cells health. Further research is being carried out by our team regarding the incorporation of the isolated 4 LMW and D 4 LMW peptides into dairy foods using spraydrying and lyposomes as carriers. In vivo studies could provide data related to actual bioavailability of the isolates.