3.2. Preparation of Fermented Flours in Bioreactor
Taking into account the information obtained in the preliminary tests, natural fermentation was carried out in a bioreactor. Two different systems were studied:
Test 1. The fermented flour FF1 was obtained by using a 36.4%
w/
w flour dispersion and incubating it for 24 h at 30 °C. The evolution of the pH was recorded, and the value dropped from 6.2–6.3 (t = 0) to a final value (24 h) of 4.75 (
Table 2). Thus, pH values decreased to a greater extent than in tube fermentation in the same conditions (
Table 1). It should be noted that this dispersion presented a high viscosity and, in consequence, was difficult to stir.
Test 2. The fermented flour FF2 was obtained by using a lower flour concentration (14.3%
w/
w) and incubating it for 24 h at 37 °C to achieve a greater fluidity of the dispersion and easy agitation. The pH value achieved after 20 h of incubation was 4.75 (similar to that reached at 24 h by FF1), and after 24 h, it was about 4.4 (
Table 2), showing a slightly higher decline (
p < 0.05) than in FF1. These results suggest an increase in the fermentation rate under conditions of a lower flour/water ratio. Sáez et al. [
26] reported a pH decrease from 6.30–6.43 to 4.80–4.83 after the first back-slopping for the natural fermentation (24 h at 37 °C) of different varieties of beans in 1 g/mL flour/water dispersions.
A microbiological screening of the non-fermented and fermented samples using nutrient agar (NA) (total mesophilic aerobic bacteria), YGC (fungi and yeasts), and MRS with a selection factor to observe the growth of LABs was performed. The microbiological counts of samples before fermentation (t = 0) were in a range of 4.2 to 4.5 log CFU/g in both NA and MRS, with no evident growth on the YGC medium (
Table 2). These results are comparable to the previously reported value of 4.6 log CFU/g for unfermented chickpea flour [
27] and are slightly higher than those obtained by Rizzello et al. [
28] for Faba bean flour (3.6 log CFU/g). After fermentation, counts of around 9 log CFU/g (NA and MRS) were registered for FF1 and FF2 (
Table 2). The bacterial colonies that grew in the MRS medium presented smooth edges, a white color, and a creamy appearance. When observed under the microscope, these colonies presented chained coccus-type and bacilli morphologies. According to
Bergey’s Manual, Gram staining (positive) and the catalase test (negative) carried out on different colony-forming units indicated that these colonies could be presumptively identified as LABs.
Generally, sourdough contains a variable number of LABs, ranging from 7 to 9 log CFU/g [
29]. Values of around 8 log CFU/g have been reported for different beans’ sourdoughs after 6 days of fermentation [
26]. No growth was observed in the YGC medium, indicating undetectable counts of yeasts after the fermentation process in the lower dilution performed. At this point, it is worth mentioning that tests 1 and 2 were carried out at different times, which forced us to use different batches of peas (different harvest years and different storage times). However, the microbiological counts were similar, both for the initial microbiota and for the fermented samples. In this way, according to our experience and bibliographic data, the counts of microorganisms (LABs, yeasts) after fermentation do not present major differences between different harvests. However, it is very likely that there is a difference between the type and proportion of constituent microbial species, an aspect that will be studied in greater depth soon.
3.4. Changes in the Protein Fractions of Fermented Flours
As in the case of 10 mL (tube) fermentations, partial proteolysis was evidenced by the increment in the HD value. Although FF1 and FF2 presented similar HD values, we can remark that the HD value doubled in the case of FF1 and increased by 5 times in the case of FF2 with respect to the corresponding initial values (
Table 4). In this way, there was a greater level of protein hydrolysis in the case of FF2.
A complementary test was carried out to determine whether endogenous proteases from pea seeds could be activated by the drop in pH, producing proteolysis. The mobilization of storage proteins in germinating seeds is initiated by endo-proteases that convert water-insoluble storage proteins into soluble peptides. Most of the plant proteases are neutral or alkaline, and there are few acid proteases (pH optimum: 2–3) widely distributed in the plant seeds [
34]. In the case of cereals, the comparison of wheat and rye sourdoughs and chemically acidified doughs indicated that primary proteolysis is mainly attributable to endogenous proteases [
35]. To evaluate this, the pH of a flour dispersion (14.3%
w/
w) was lowered with 2 mol/L HCl to the final value obtained in the fermentations (4.4), and the degree of protein hydrolysis was measured (TNBS method). A very low value (close to 0), even lower than those registered for flour dispersions in water before fermentation, was obtained. According to this, no activation of endogenous proteases was evidenced. In agreement, Akhtaruzzaman et al. [
36] extracted proteases from seven overnight-imbibed leguminous seeds and found that the alkaline proteases involved in all seeds were more potent than the acidic proteases. In consequence, proteolysis in the fermented samples would be the product of the action of proteases from microorganisms. LAB strains displayed a wide range of proteolytic activities [
27].
Comparing the results obtained for 10 mL (tube) fermentations and those in the reactor under the same conditions (36.4%
w/
w, 24 h, and 30 °C), there was no significant difference (
p > 0.05) in the protein solubility values (52 and 56%, respectively,
Table 1 and
Table 4). In all cases, the solubility decreased in the fermented samples with respect to the non-fermented ones, which could be due to the formation of aggregates during the fermentation process, as will be discussed next.
The changes in the peptide/polypeptide profile produced by fermentation were analyzed by glycine-SDS-PAGE. The glycine-SDS-PAGE profiles of F1 and F2 (
Figure 1) showed a great variety of polypeptides between 14 and 97 kDa. It was possible to detect bands tentatively belonging to linoleate 9S-lipoxygenase (band 1, 93 kDa), alpha-dioxygenase (band 2, 77 kDa), convicilin (an important storage protein in peas, 70 kDa, band 3), legumin subunits (59 kDa, band 4), free acidic (40 kDa, band 7) and basic (band 13, 20 kDa) legumin subunits, and vicilin subunits (53 kDa, band 5; 34 kDa, band 9: pea vicilin is heterogeneous, so variable polypeptides could be produced by different gene coding), and band 6 (probably alpha-galactosidase, 45 kDa); bands 11 and 12 (28 and 25, probably subunits/polypeptides of albumin-2), and bands 14 to 17 (20–14 kDa) would correspond to albumins. The recognition of the pea polypeptides was carried out according to Ma et al. [
37].
After fermentation, a decrease in the intensity of all bands was observed, being more evident for 93 kDa (band 1) and for bands corresponding to MW < 40 kDa. Also, an increase in high-MW molecules that did not enter the gel could be observed in fermented samples, suggesting the presence of aggregates that remain even in the presence of SDS and urea (
Figure 1). The formation of aggregates could explain the decrease in solubility observed in fermented samples (
Table 4); this fact can be at least partially explained since pea proteins have their minimum solubility at the isoelectric point (between 4 and 5), coinciding with the final pH value in fermented flours. Band 10 (31 kDa, which could correspond to the anti-nutritional factor lectin [
37]) appeared much more intense in samples F2 and FF2 than in F1 and FF1, while band 12 (25 kDa, which could include Kunitz-type trypsin inhibitor-like 2 protein [
37]) has a higher intensity for F1 and FF1 with respect to F2 and FF2 (
Figure 1). Beyond these differences between the two dispersions, the intensity of these bands decreased after fermentations, suggesting a diminution in the mentioned anti-nutritional factors. The reduction in the color intensity of several bands after fermentation could be associated with polypeptide diminution due to proteolytic activity. Byanju et al. [
38] also observed this pattern of band discoloration in pea, lentil, and soybean flours after fermentation with
L. plantarum and
Pediococcus acidilactici. Also, the fermentation of pea flour with three LABs (
Pediococcus pentosaceus,
Lactococcus raffinolactis, and
L. plantarum) resulted in similar patterns of the Coomassie brilliant blue-stained gels, which were not very different from the extract of the unfermented flour, except for the disappearance of some high-molecular-weight bands [
39].
The peptide/polypeptide composition of the soluble fractions of non-fermented and fermented flours was analyzed by gel filtration chromatography using a Superdex 30 column (optimal separation in the range for MW < 10 kDa) in order to evaluate low-MW peptides. As expected, the chromatograms of F1 and F2 (
Figure 2A) were similar. Fermentation caused an increase in molecules smaller than 6.5 kDa in both conditions (FF1 and FF2). However, some differences between FF1 and FF2 could be described: peak 2 (MW > 6.5 kDa) decreased more in the case of FF1, while peaks 1 (MW > 10 kDa), 3 (1.5–0.8 kDa), 5 (0.47–0.18 kDa), and 6 (0.18–0.08 kDa) increased more in the case of FF2 (with respect to the non-fermented flour), showing a greater occurrence of small molecules in FF2.
According to the electrophoresis and FPLC analyses, fermentation produced some minor changes in the protein profile of the pea flour related to the appearance of aggregates and soluble proteolytic fragments with MW < 2 kDa, with some differences between the two fermentation conditions assayed. These results, together with the registered proteolysis degree, showed that the LAB strains present in the fermented flours produced moderate proteolysis of the pea proteins.
3.5. Effect of Fermentation on Protein Fraction Bioaccessibility (SGID) and Antioxidant Activity
After SGID, as expected, HD significantly increased (
p < 0.05) for all samples (
Table 4). However, the values were significantly greater (
p < 0.05) when the flour was previously fermented (FF1D and FF2D with respect to F1D and F2D, respectively), indicating that the fermentation process improved the proteins’ gastrointestinal digestion. FF1D presented a significantly greater (
p < 0.05) HD value than FF2D. However, the HD value of FF1D was 7.5 times greater than that of F1, while the HD value of FF2D was 13 times greater than that of F2, showing a greater proportion of proteolysis in the second case (
Table 4). SDS-PAGE (
Figure 1) showed that in the samples subjected to SGID, most of the polypeptides disappeared, with the appearance of some bands, such as 18 (51 kDa), 19 (43 kDa), and 20 (a broad band of about 35 kDa), and partially remaining bands of MW < 25 kDa (legumin subunits and albumins) for all the digests (F1D, F2D, FF1D, and FF2D). In this way, some pea polypeptides resisted gastrointestinal digestion. This fact has been previously observed when the gastrointestinal digests of flours and protein isolates from two pea varieties were analyzed [
1]. Ma et al. [
37] reported that a pea protein hydrolysate obtained by the action of a mixture of trypsin, chymotrypsin, and peptidase presented a reduction in most of the bands present in the raw pea profile but with the persistence of bands with MWs between 10 and 30 kDa. Some differences could be detected among digests, mainly in the molecules generated by SGID. The intensity of bands 18 and 20 was greater for digests from non-fermented flour (F1D and F2D), while the intensity of band 19 was greater for digests from fermented flour (FF1D and FF2D) (
Figure 1). In analyzing the effect of fermentation on the subsequent SGID, the electrophoretic profiles showed a lower intensity in some of the remaining bands in the fermented meals, in agreement with the highest HD values obtained for digests of fermented flours. Partial proteolysis due to the fermentation process made the sequences more susceptible to further degradation by the digestive enzymes, as has been previously reported [
40].
In analyzing the composition of the PBS-soluble fractions of gastrointestinal digests, gel filtration chromatograms (
Figure 2B) showed that the peaks corresponding to the exclusion volume (>10 kDa) decreased with respect to the undigested samples and significantly increased the number of molecules smaller than 6.5 kDa in all digested samples. Similar behavior has been previously reported for flours and protein isolates of two pea varieties and their corresponding digested samples [
1]. Considering each particular peak, only minor differences in the area were observed among the four digests. Peak 8 (0.4 to 8 kDa) constituted the greatest modification after gastrointestinal digestion and presented the highest area in the four digests, representing about 60–63% of the total area. Peak 1 (the remaining MW > 10 kDa molecules) accounted for around 30–33% of the area, with F1D presenting the highest value and FF1D the lowest one. Peak 5 (0.47–0.18 kDa) represented about 2.5 to 4%, and peak 6 (< 0.18 kDa) between 2.5 and 3%.
The antioxidant activity of the PBS-soluble fractions of non-fermented and fermented pea flour before and after SGID was evaluated. The ORAC assay method measures the scavenging capacity against peroxyl radicals (generated from AAPH at 37 °C) by the oxidative degradation of fluorescein [
40]. Dose–response curves for ORAC (ROO· scavenging % versus peptide concentration) were obtained, and IC
50 values were calculated (
Table 4). The ORAC activity was significantly (
p < 0.05) increased by the fermentation process, with a diminution in IC
50 values of 2.5 times for FF1 with respect to F1 and 2.7 times for FF2 with respect to F2, with a significantly (
p < 0.05) lower IC
50 value for FF2 (
Table 4). Also, the HORAC assay was performed, in which the oxidative degradation of fluorescein is caused by hydroxyl radicals generated by the Fenton reaction [
41]. Dose–response curves presented a linear fitting in this case. There was no significant difference (
p > 0.05) between the IC
50 values of non-fermented and fermented flours in any fermentation condition (
Table 4), indicating that fermentation had no effect on this activity.
SGID produced a significant increase (
p < 0.05) in ORAC activity in the case of both F1D and FF1D, with increases of about 4 and 9 times with respect to F1, respectively. F2D and FF2D also presented a significant increase (
p < 0.05) in ORAC activity with respect to the initial sample (F2), with a potency increment of 5 times. FF2D presented an IC
50 value that was slightly (but significantly) lower than that of FF1D (
Table 4). The SGID process produced an increase in antioxidant HORAC potency since the IC
50 values were reduced by half, without a significant difference between the different digests (
p > 0.05). Based on these results, we can conclude that the natural fermentation of pea flour produced an increase in ORAC activity associated, in principle, with the release of peptides, but had no noticeable effect on HORAC activity. The difference in the sensitivity and in the mechanisms of action related to these two methods could explain the differences in the behavior of fermented flours.
Taking into account the previous results and some practical considerations related to the ease of agitation and dispersion, it was decided to continue studying the flour fermented in condition 2 (14.3%
w/
w, 24 h, and 37 °C). In order to learn more about the distribution of molecules that contribute to the antioxidant activity of these samples, fractions of different MWs from F2, FF2, F2D, and FF2D were separated by FPLC gel filtration, after which their peptide concentrations and ROO· scavenging activities were determined using the ORAC test (
Figure 3). In F2, as expected, the fractions with the greatest polypeptide concentrations were those with MW > 10 kDa (fractions 1 to 9). These fractions presented ROO· scavenging activity (40–60%); however, fractions 23–26 (MWs between 0.29 and 0.59 kDa) presented the highest activities (66 to 81%,
Figure 3A), but low or non-detectable concentrations of peptides (
Figure 3B). According to the MWs of these fractions, they could involve peptides of between 3 and 5 amino acids, although the presence of other components, such as phenolic compounds, cannot be ruled out, all of which would present significant ORAC activity.
After fermentation (FF2), fractions 1 to 8 (>10 kDa) decreased their polypeptide concentrations (
Figure 3A) and their ORAC activity (
Figure 3B). Also, fractions 23 to 26 (0.29–0.59 kDa) had diminished ROO· scavenging activity, while several fractions in the ranges of 0.75 to 4 kDa (fractions 15–22) and 0.18–0.23 kDa (fractions 28–29, 65–74%) increased it (
Figure 3B). In this way, the increment in ORAC activity registered after the fermentation of pea flour could be mainly related to the appearance of molecules in the ranges of 0.75–4 kDa and 0.18–0.3 kDa with improved ROO· scavenging. Most of the studies involving the formation of bioactive peptides by fermentation were carried out with LABs, which possess a complex system of proteases and peptidases [
42]. As reported by Venegas-Ortega et al. [
43], the differences found in LAB proteinases explain the variety of bioactive peptides produced, even when the same protein matrix is used. In a previous work [
44], nine
Lactobacillus strains were evaluated for their ability to grow in a pea seed protein-based medium and to hydrolyze purified pea proteins to produce peptides with antioxidant activity. Two strains,
Lacticaseibacillus rhamnosus BGT10 and
Lacticaseibacillus zeae LMG17315, exhibited strong proteolytic activity against pea proteins. These authors showed that the antioxidant activity (DPPH assay) of the fraction with MW < 10 kDa increased after 12 h of fermentation with
L. rhamnosus BGT10. This fraction presented antioxidant activity in different assays, and when performing a separation by ion exchange chromatography, they showed that a low-abundance sub-fraction of basic peptides presented the highest activity.
The SGID process (F2D and FF2D) produced increases in the peptide concentrations of all fractions with MW < 3 kDa (
Figure 3A) and increments in the ROO· scavenging % for almost all fractions with MW < 6.5 kDa (
Figure 3B). F2D showed the higher scavenging % values (41–87%) for fractions between 0.14 and 4 kDa (fractions 15–29). FF2D presented higher scavenging values with respect to F2D in almost all fractions greater than 4 kDa (<45% scavenging) and in some of the fractions in the ranges of 2 to 0.3 kDa (18–26) and less than 0.10 kDa (<40% scavenging), and both digests presented their maximum ROO· inhibition in fractions of around 0.14–0.18 kDa (28 and 29, probably free amino acids), being 84 and 87% for FF2D and F2D, respectively (
Figure 3B). These results also showed some differences in the molecular composition of the gastrointestinal digest of non-fermented and fermented pea flours.
3.6. Effect of Fermentation on PC Bioaccessibility (SGID) and Antioxidant Activity
Given the importance that PCs have in antioxidant activity, whether fermentation modified the content of PCs was evaluated on 60% ethanol extracts of F2 and FF2. The fermentation process significantly (
p < 0.05) increased (about 3 times) the TPC measured by the Folin–Ciocalteu method (
Table 5).
Gan et al. [
45] reported that natural fermentation increased the TPC in most legumes, especially in the mottled cowpea, where it increased by about 80%. Xiao et al. [
46] performed extractions with different solvents (80% methanol, 80% ethanol, 80% acetone, and water) of fermented mung bean, and in all of them, the TPC increased with respect to the non-fermented samples. These authors suggested that the chemical structures, polarities, and solubilities of the mung bean PCs were significantly influenced by the fermentation process. The PC profile of FF2 was analyzed by HPLC-DAD-FLD and compared to that of F2 (
Table 6).
The PC composition of yellow pea flour has been previously studied in our lab (Cipollone et al., under revision), with (−)-epigallocatechin (a flavan-3-ol) and polydatin (stilbene) as the major PCs. Several changes were observed after fermentation. Increments in ellagic, rosmarinic, and especially caffeic acids but diminutions in gallic (a hydroxybenzoic acid),
p-coumaric, and ferulic acids (hydroxycinnamic acids) were observed. OH-tyrosol was not detectable after fermentation. Among the flavonoids (the majority in the flour), only (−)-epicatechin and naringenin (flavanone) decreased, while the flavanone hesperitin, the flavan-3-ols (−)-epigallocatechin and (+)-catechin, and the flavonols rutin (quercetin-3-O-rutinoside), quercetin-3-glucoside, kaempferol-3-glucoside, and quercetin-3-glucoside increased; quercetin, which was not found in the flour, appeared in FF2 (
Table 6). The total amounts of HPLC-DAD-FLD-detected PCs increased after fermentation, mainly due to a flavonoid family increment. Dueñas et al. [
47] carried out fermentations of cowpea flour, both spontaneous and with
L. plantarum ATCC 14,917 (48 h, 37 °C); both fermentations modified the contents of PCs but in a different manner. They found—as in our case—that the fermentation gave rise to the appearance of some PC compounds not detected in raw flour, such as quercetin. That was explained by the fact that pH lowering could activate some enzymes that hydrolyze quercetin glycosides, thus yielding quercetin. Lactobacillaceae possess a broad spectrum of enzymatic activities for the biotransformation of dietary PCs that could have participated in the change in the PC profile previously described. Esterases, reductases, and decarboxylases would participate in the conversion of hydroxycinnamic and hydroxybenzoic acids. In addition, LABs contain glycosyl hydrolases that seem to be dedicated to the hydrolysis of glycosides of plant secondary metabolites, such as glycosylated flavonoids, although little is known about the substrate specificity of these enzymes [
48].
When analyzing the antioxidant activity of the ethanolic extracts, it was observed that the ABTS activity significantly decreased (
p < 0.05) after fermentation (
Table 5). In addition, fermentation did not have an effect on ROO· scavenging since FF2 presented a similar (
p > 0.05) IC
50 value for ORAC to that of F2 (
Table 5). This behavior was different from that recorded for the fractions soluble in PBS, in which ORAC activity increased after fermentation (
Table 4). Gel filtration FPLC chromatograms (
Figure 4) of the ethanol extracts showed that both F2 and FF2 presented molecules with MWs in a broad range (<10 kDa), but they did not contain the larger polypeptides (>10 kDa) that appeared in the peak corresponding to the exclusion volume (unlike the fractions soluble in PBS,
Figure 2). As in the PBS-soluble fractions, the increment in molecules lower than 2 kDa was evident after fermentation (
Figure 4).
After SGID, an increase in the TPC was observed in F2D and FF2D with respect to their non-digested samples, being greater when the flour had been previously fermented (
Table 5). According to this, Ketnawa and Ogawa [
49] reported an increase in TPC values after subjecting fermented soybeans to an SGID process. SGID produced several changes in the PC profile of F (Cipollone et al., under revision). The gastrointestinal digest of fermented flour (FF2D) presented higher contents of some phenolic acids than F2D, such as gallic, syringic, caffeic, and rosmarinic acids (
Table 6). However, the greatest difference was found in flavonoids, whose content was much lower in FF2D since compounds from the flavan-3-ol family (catechins and procyanidin) were not found. These results suggest that, after fermentation, these compounds were more available for the modifications that can occur during the gastrointestinal digestion process, such as the instability of catechins at neutral pH [
50] and of procyanidin at gastric acid pH [
51]. SGID produced significant decreases (
p < 0.05) in the IC
50 values of both digests, without significant differences between them (
Table 5). It also led to an increase in ABTS activity in the case of FF2, with both digests showing similar IC
50 values. Thus, although fermentation produced modifications in the PC profile of F, these did not translate into important changes in ORAC and ABTS activities after SGID. Sancho et al. [
52] measured the antioxidant activity in the methanol extracts of raw red and black beans before and after digestion and reported that there was no significant difference in the ABTS values, and there was only a difference in the extract of black beans when measured by the ORAC method.
It is important to note that, although the total content of PCs detected by HPLC-DAD-FLD was much lower in the case of FF2D, its TPC determined by Folin was somewhat higher than for F2D. In addition, both digests presented higher TPC but lower HPLC-detected PCs than non-digested samples (
Table 5 and
Table 6). These facts suggested that other substances reactive to Folin were present in the extracts. To analyze this, gel filtration FPLC of the ethanol extracts was performed. After SGID, the presence of molecules with MW < 6.5 kDa strongly increased, and to a much lesser extent, molecules with MW > 10 kDa also increased (
Figure 4); the latter presented a much lower abundance than in the case of the fractions soluble in PBS (
Figure 2). These analyses demonstrated the presence of other kinds of compounds in the ethanol extracts, such as peptides and amino acids, with higher abundances in FF2D. Therefore, the antioxidant activity of these ethanol extracts showed the contributions of both PCs and peptides and free amino acids that could be solubilized in the extraction conditions.