Free amino groups were measured by the 2,4,6-trinitrobenzene sulfonic acid (TNBS) method. HD was calculated [27].

Freeze-dried samples were dispersed in sample buffer (0.0625 mol/L Tris-HCl, 2% SDS, 10% v/v glycerol, pH = 8.8) and centrifuged before loading in the gel. Separating and stacking gels (120 and 40 g/L acrylamide, respectively) were used. Runs were carried out in a Mini Protean II Dual Slab Cell (Bio-Rad, California, USA) at room temperature, applying a constant current (30 mA/gel). Gels were stained with Coomasie Brilliant Blue R-250 (1 g/L) [28].

Suspensions (20 mg/mL) of freeze-dried samples in PBS (1.5 mmol/L KH₂PO₄, 138 mmol/L NaCl, 3 mmol/L KCl, 8.1 mmol/L Na₂HPO₄, pH = 7.4) were prepared by agitation at 500 rpm (1 h, 37°C) (Termomixer, Eppendorf, Hamburg, Germany) followed by centrifugation (10,000× g, 10 min, room temperature, Hermle, Labortechnik GmbH, Germany). The supernatants were separated, forming the B_s, DB_s, and RB_s fractions, and the soluble protein concentration was determined by the Lowry method [29].

PBS-soluble fractions were analyzed in an ÄKTA purifier (GE Healthcare, Illinois, USA) using a molecular exclusion column. Superdex Peptide 10/300 GL (GE Healthcare) (exclusion limit: 10 kDa; separation range: 0.1–7 kDa) was calibrated with blue dextran (exclusion volume V_o = 7.60 mL), aprotinin (6.5 kDa), vitamin B12 (1.35 kDa) and hippuric acid (0.18 kDa) (log MM = 4.84–3.30 × K_av, where K_av = (V_e – V_o) / (V_t – V_o) (V_e: elution volume; V_o: void volume; V_t: total volume, 24 mL)). Samples were filtered by a 0.45 µm nylon filter, 200 µL of sample was loaded, and eluted with PBS buffer at 0.5 mL/min. Detection at 210 nm was performed.

Freeze-dried B, DB, and RB were extracted according to a previous optimization in our laboratory [30]. Ultrasound-assisted extraction (UAE) was performed in a VCX 750 ultrasonic processor (Sonics & Materials Inc., Newtown, USA) by using 60:40 ethanol/water as solvent and the following conditions: 15 min, 40% amplitude. The extractions were carried out in an ice bath to avoid excessive increases in temperature, which was maintained in all cases below 42°C. The extracts were centrifuged (Hermle, 5 min, 10,000× g, room temperature). The supernatants were evaporated (30°C, Concentrator plus/Vacufuge^® plus, Eppendorf, Hamburg, Germany) and solubilized in PBS buffer, obtaining the B_e, DB_e, and RB_e fractions.

TPC of B_e, DB_e, and RB_e was determined using the Folin-Ciocalteu method [31]. A standard curve was made with gallic acid (0–0.06 mg/mL) in PBS buffer. Results were expressed as mg gallic acid equivalent (GAE)/mL and as mg GAE/g sample in dry base (d.b.).

For the determination of the qualitative and quantitative profiles of the PCs, dry extracts were dissolved in the initial mobile phase of the chromatographic method at the time of analysis. Separation and determination of PCs was performed in a high-performance liquid chromatography coupled with diode-array and fluorescence (HPLC-DAD-FLD) detectors (Dionex Ultimate 3000 system, Dionex Softron GmbH, Thermo Fisher Scientific Inc., Germering, Germany) and a reversed-phase Kinetex C₁₈ column (3.0 mm × 100 mm, 2.6 mm; Phenomenex, Torrance, California, USA). The software Chromeleon 7.1 was used to control the run and to process the data. The list of PC determined, chromatographic, and detection conditions were those reported by [32], with little modifications [30].

ORAC was measured following the previous method [33]. The fluorescence intensity (λ_exc: 485 nm, λ_em: 535 nm) was read every minute in a SYNERGY HT– SIAFRT™ multidetection microplate reader (Biotek Instruments, Agilent, California, USA) to obtain the fluorescein-decay curve and the area under the curve (AUC). A blank without AAPH was included, and the % scavenging calculated as: %ROO· scavenging = [(AUC_S – AUCN_C) / (AUC_B – AUC_NC)] × 100; where S: sample, B: blank, NC: negative control. Scavenging % was plotted versus concentration, and the concentration that inhibits 50% of radicals (IC₅₀) was obtained. Trolox (6.25–75 μmol/L) was used as a reference compound.

ABTS^•+ was measured according to [34]. Trolox (0.05–0.20 mmol/L) was used as the reference compound. The % inhibition of the radical was calculated as [(Abs_B0 – Abs_S15) – (Abs_B0 – Abs_B15) × 100] / Abs_B0; where Abs_B0 and Abs_B15 refer to the absorbance of the blank at 0 and 15 min, respectively, and Abs_S15 refers to the absorbance of the sample at 15 min. The IC₅₀ value was obtained.

Direct soluble fraction (DB_d) was prepared by solubilization of lyophilized DB in milliQ water, homogenization (Thermomixer Eppendorf 37°C, 1 h, 500 rpm), and centrifugation (10,000× g, 10 min). To reduce the bile salt content of this fraction and also of the ethanol extract, they were treated with cholestyramine resin (10% w/v), homogenized (Eppendorf Thermomixer, 25°C, 1 h, 500 rpm), and centrifuged (10,000× g, 20 min), obtaining DB_b and DB_eb fractions, respectively. The supernatant was filtered (0.22 µm) [35].

Caco-2 TC7 cells (passages 51–52) used were from the American Type Culture Collection (ATCC, Manassas, VA, USA; ATCC performs the STR identification (https://www.atcc.org/products/htb-37#detailed-product-information)). Cells are regularly monitored for characteristic morphology and confirmed to be free of mycoplasma contamination (LookOut^@ Mycoplasma PCR Detection Kit, Sigma Aldrich MP0035, St. Louis, MO, USA). Also, we work in a way that prevents cross-contamination between cell lines. The growth medium was DMEM (pH = 7.4) supplemented with 15% w/v FCS, 4.5 g/L glucose, 13 mL/L PenStrep, 2 g/L NaHCO₃, 0.5 g/L gentamicin, and 1% w/v NEAA. All cell culture experiments were done at 37°C with a 5% CO₂ atmosphere. For each determination, cell growth and assay conditions were previously optimized in our lab [16, 36]. DB_b and DB_eb fractions in different dilutions (direct, 1/5, 1/10, and 1/20 were analyzed). Bioaccessible fractions obtained from DF (DF_b) and DI (DI_b) were also analyzed for comparison.

Cells were seeded onto 96-well plates with a density of 2.5 × 10⁴ cells/well and incubated for 48 h to confluence. Cytotoxicity of samples was quantified in terms of lactate dehydrogenase (LDH) release to the extracellular medium [37]. Cells were incubated with 100 μL of the sample for 3 h, and the LDH-P UV Unitest kit (1521351, Wiener Lab, Rosario, Argentina) was used. LDH activity (λ: 340 nm) was determined in the supernatants using the microplate reader as: LDH (U/L) = (ΔA/min) × f (f = 4.921, 25°C). The %LDH release was expressed as (LDH / LDH_total) × 100, where LDH_total corresponded to the cellular death positive control (DMEM with 3% v/v Triton X-100) [38].

Intracellular ROS content in Caco-2 TC7 cells (2.5 × 10⁴ cells/well) cultured in 96-well plates, treated with 20 μmol/L DCFH-DA, sample, and finally 500 μmol/L H₂O₂ was determined according to [35]. Fluorescence (λ_exc: 485 nm, λ_em: 528 nm) was measured during 3 h (37°C). Control systems were analyzed: C1: H₂O₂-induced oxidation level (buffer, no antioxidant), C2 (baseline, no H₂O₂, no antioxidant), C3: intrinsic fluorescence (no DCFH-DA, no sample).

The polypeptide profile of sample B was examined by SDS-PAGE and compared with those of F and I, which has been previously analyzed by Cipollone and Tironi [17]. Several bands (1, 3, 4, 6, 7, 9, 10, 11, 13, 15, and 16) showed reduced intensity in B (Figure 2B) relative to F and I (Figure 2A). These included convicilin (band 3), legumin (band 4), and both the acidic and basic subunits of legumin (bands 7 and 13). In contrast, bands 2, 5, and 8—corresponding to vicilin and its subunits—remained unchanged. In addition, B showed the appearance of new molecular species (not present in F or I) such as some of high molecular weight indicated at the top of the lane (circled in Figure 2B) and probably others of lower molecular mass (especially in the < 30 kDa range, and bands 12 and 14—primarily albumins) that enter the gel, which could decrease the resolution of the bands and produce a “spraying” effect. The observed changes may be attributed to: 1. Protein loss during the preparation of B, which involves alkaline solubilization and filtration; for example, band 10 (< 30 kDa) also decreased in sample I (Figure 2A). 2. Aggregate formation, including aggregates that remain soluble in the electrophoresis sample buffer, as indicated by the material accumulating at the top of the gel in B. The presence of insoluble aggregates that fail to enter the gel cannot be excluded.

After the SGID process, the SDS-PAGE profile of DB (Figure 2B) showed the absence of detectable bands (> 14 kDa), including the aggregates observed in sample B. Although the presence of remaining proteins in quantities undetectable by Coomassie-Blue staining (< 50 ng) cannot be ruled out, this result suggests that the proteins and peptides originally present in B were extensively hydrolyzed by the digestive enzymes. This outcome is consistent with the degree of proteolysis (HD) measured by the TNBS method, which was 49 ± 4%. This value was higher than those recorded for DF and DI (36% and 37%, respectively).

Protein contents in the PBS-soluble fractions (B_s and DB_s) were measured (Table 3). The low protein solubility observed in sample B may be due to the formation of aggregates between polypeptides or with gums, which become insoluble upon centrifugation. SGID caused a significant increase (P < 0.05) in both soluble protein content (Table 3) and protein solubility in DB_s.

The PBS-soluble fractions were analyzed by size-exclusion FPLC (Superdex Peptide 10/300 GL column). In B_s, a marked reduction in the V_o peak (> 10 kDa) was detected compared with F_s, along with decreases in peaks 2, 3, and 4 (Figure 3A), indicating a loss of soluble molecules during the production of B. A new peak corresponding to very low molecular-weight species (peak 5) also emerged in B_s. Relative to I_s, B_s displayed a narrower molecular-weight distribution in peak 1 (> 10 kDa) and a higher proportion of low-molecular-weight species (peaks 3, 4, and 5). DB_s (Figure 3B) clearly reflected enzymatic activity during SGID, evidenced by the reduction in peak 1 and the appearance of new peaks with molecular weights < 6.5 kDa.

Following SGID, a direct soluble fraction (DB_d) was also obtained. Considering that bile salts are known to exhibit cytotoxic and oxidative effects [16, 45], a treatment with cholestyramine was performed to yield the so-called “bioaccessible” fraction (DB_b). As shown in Table 3, protein concentration presented no significant difference (P < 0.05) between DB_d and DB_s. However, DB_b exhibited a notable decrease in soluble protein content, accompanied by a reduction of compounds across the molecular weight range in the gel filtration chromatograms (Figure 3C). Specifically, peak 1 was minimal in DB_d and absent in DB_b; additionally, DB_b displayed a lower proportion of molecules with MW < 6.5 kDa and the disappearance of molecules smaller than 0.17 kDa (peak 2).

An initial assessment of the antioxidant activity of B_s, DB_s, DB_d, and DB_b was carried out using two approaches: the ORAC and ABTS^•+ scavenging assays, with IC₅₀ values determined (Table 3). Using the ORAC method, B_s showed an IC₅₀ value comparable to those reported for F_s and I_s (0.24 ± 0.08 and 0.31 ± 0.02 mg protein/mL, [26]), suggesting that alterations in polypeptide composition of B did not affect this activity. In contrast, the SGID process led to a tenfold reduction (P < 0.05) in IC₅₀, indicating the release of bioactive compounds, such as previously inactive peptides. This enhancement was greater than that observed for DF_s (IC₅₀ = 0.078 ± 0.008 mg protein/mL, approximately threefold higher than F_s) and DI_s (IC₅₀ = 0.068 ± 0.001 mg protein/mL, approximately 4.5-fold higher than I_s) [26].

No significant differences (P > 0.05) were observed between the IC₅₀ values of DB_s and DB_d. However, treatment of DB_d with cholestyramine to obtain the DB_b fraction resulted in a significant reduction (P < 0.05) of nearly 50% in ORAC antioxidant potency. A similar pattern was observed for the DI_b and DF_b. Notably, the IC₅₀ value of DB_b was approximately six times lower than that of DF_b (0.28 ± 0.09 mg protein/mL) and 2.5 times lower than that of DI_b (0.12 ± 0.01 mg protein/mL), indicating a higher antioxidant (ORAC) capacity of the aqueous soluble fraction of the DB. This trend aligns with the findings of [16], who reported higher ORAC values for amaranth beverage digests compared to the corresponding flour and protein isolate. SGID also led to an increase in antioxidant activity as measured by the ABTS method. DB_s exhibited an IC₅₀ value approximately one-tenth that of B_s. Moreover, although treatment with cholestyramine showed a tendency to reduce potency, the change was not statistically significant (P > 0.05) (Table 3).

Cellular assays were carried out to determine the capacity of aqueous soluble DB to inhibit intracellular ROS. First, the cytotoxicity on Caco2-TC7 cell cultures of DB_b and RB_b was evaluated.

As noted earlier, bile salts have the potential to induce cytotoxicity and promote cellular oxidation. For DB_b, cytotoxicity was influenced by sample dilution; however, none of the tested concentrations caused significant LDH release, suggesting minimal impact on cell integrity (Table 4). In the case of RB_b, cytotoxicity also varied with dilution, with LDH release remaining low—comparable to that observed for undiluted DB_b—at a 1/20 dilution.

The intracellular ROS levels were measured after exposing cells to the bioaccessible fractions, using H₂O₂ as an oxidative inducer [36]. In the basal control system (C2), ROS levels were half of those in C1, meaning that H₂O₂ treatment roughly doubled the basal cellular ROS content. Pretreatment with RB_b, DF_b, DI_b, and DB_b caused a substantially higher increase in fluorescence compared to C1. This phenomenon has been previously reported [16] and is likely related to the presence of bile acids/salts, which can induce ROS production in non-polarized Caco-2 cells [45]. To calculate the % inhibition for the different dilutions of DB_b, the corresponding dilutions of RB_b were used as the maximum oxidation control, and the percentage of viable cells in each sample was also considered. Similar to DF_b and DI_b, DB_b exhibited a concentration-dependent inhibition of intracellular ROS based on the soluble polypeptide/peptide content within the tested range (Table 4). From the dose-response curve, the estimated IC₅₀ values were 0.3 mg protein/mL for DB_b, 0.26 mg protein/mL for DI_b, and 0.62 mg protein/mL for DF_b. These results suggest that DB_b has a neutralizing potency comparable to DI_b and greater than that of DF_b.

The TPC content of the B_e extract was 110 ± 1 µg GAE/mL, corresponding to 2.4 mg GAE/g d.b. (Table 5), higher than those previously obtained for F_e and I_e (0.68 and 0.98 mg GAE/g d.b. [30]).

HPLC-DAD-FLD profiling of B_e revealed some differences compared to F_e and I_e (Table 6). Compared with F_e, B_e exhibited higher levels of OH-tyrosol, phenolic acids, and stilbenes, while flavonoid concentrations were similar. In contrast, when compared with I_e, B_e showed increased levels of OH-tyrosol, phenolic acids, and flavonoids, but a reduction in total stilbene content. All PCs detected in F_e and I_e—namely OH-tyrosol, procyanidin B1, (+)-catechin, trans-resveratrol, genistein, gallic acid, p-coumaric acid, and syringic acid—were present at higher concentrations in B_e. Moreover, B_e contained additional compounds not detected in F_e or I_e, including sinapic acid and the flavonoids fisetin, quercetin, quercetin-3-galactoside, procyanidin B2, genistin, and (–)-epigallocatechin gallate (Table 6).

After SGID, DB_e exhibited a significant (P < 0.05) and marked increase in TPC compared to B_e (Table 5). The TPC reached 324 µg GAE/mL (18 mg GAE/g d.b.), surpassing the values previously reported for DF_e and DI_e (5.04 and 13.5 mg GAE/g d.b., respectively [30]). HPLC-DAD-FLD analysis of DB_e confirmed that its quantified PC content was higher than that of B_e. A similar trend was observed for DI_e, whereas DF_e showed a decrease in PC content compared to F_e. Several PCs present in B_e were not detected in DB_e; specifically, all phenolic acids were lost after SGID except for p-coumaric acid, which remained at a reduced concentration. With regard to flavonoids, quercetin-3-galactoside, genistein, genistin, quercetin, and fisetin were absent in DB_e, whereas kaempferol-3-glucoside, naringenin, hesperetin, and myricetin were newly detected in DB_e but were not present in B_e. These results suggest a possible release of some compounds during SGID as well as different behaviors of the different compounds. The flavonoids have different chemical structures, which is associated to their interaction/bond with the food matrix and their abilities to be released during the digestion process. Thus, bioaccessibility is related to how the non-extractable polyphenols are released or not from each matrix during the specific conditions of in vitro digestion.

Following cholestyramine treatment, TPC decreased significantly (DB_eb, Table 5). TPC was also measured in RB_e and RB_eb. RB_e showed a value of 22 ± 2 µg GAE/mL, which is approximately 15 times lower than that observed in DB_e (Table 5). This low value likely reflects the presence of other Folin-reactive compounds, such as amino acids or peptides, rather than polyphenols, due to the animal origin of all digestion reagents used. In RB_eb, no Folin-reactive compounds were detected.

The antioxidant activities of B_e, DB_e, and DB_eb were evaluated using both the ORAC method and the ABTS^•+ assay. While SGID led to an increase in TPC, the ORAC activity of these compounds remained unchanged, as no significant differences (P > 0.05) were observed between the IC₅₀ values of B_e and DB_e (Table 5). A similar pattern had previously been reported for F_e and DF_e [30]. Likewise, DB_eb showed no significant difference (P > 0.05) compared to DB_e. In contrast, ABTS^•+ scavenging showed a significant increase (P < 0.05) in antioxidant activity following SGID, which was unaffected by cholestyramine treatment (Table 5). A comparable trend had been partially observed for DI_e, which exhibited higher activity than I_e, while no significant differences were noted between I_e and DI_eb [30].

The cytotoxicity of DB_eb and RB_eb was assessed based on %LDH release (Table 7). Similar to the water-soluble fraction, RB_eb exhibited high cytotoxicity (74%), likely due to residual bile salts remaining after cholestyramine treatment, which decreased upon dilution of the sample. No TPC was detected in this fraction. DB_eb showed a substantial but lower cytotoxicity (47%), which also declined with sample dilution. These findings indicate that the cytotoxicity of the bioaccessible fractions is primarily attributable to residual digestion reagents and is mitigated due to the presence of PCs (or other bioactive components) in DB_eb, consistent with observations reported for DF_e and DI_e [29].

DB_eb demonstrated a significant inhibition of intracellular ROS (73%, Table 7), although it exhibited a relatively high cytotoxicity of 47%. In comparison, DF_eb showed slightly different results, with a lower TPC (141 µg GAE/mL), a higher cytotoxicity of 56%, and a ROS inhibition of 66%. On the other hand, DI_eb presented a higher PC concentration (356 µg GAE/mL), lower cytotoxicity (33%), and a strong intracellular ROS inhibition of 83%. The IC₅₀ values were 30.1 µg GAE/mL for DB_eb, 42 µg GAE/mL for DF_eb, and 17 µg GAE/mL for DI_eb.