2.1. Computed Molecular Properties
The first part of the study was intended to predict molecular properties for the fifty (50) plant compounds grouped as: quercetin derivatives (
Table 1), kaempferol derivatives (
Table 2), myricetin derivatives (
Table 3), luteolin derivatives (
Table 4), apigenin derivatives (
Table 5), flavanone and flavan derivatives (
Table 6), isoflavone derivatives (
Table 7), hydroxycinnamic acid derivatives (
Table 8) and hydroxybenzoic acid derivatives (
Table 9); their chemical formulas (as optimized structures) are available in the
supplementary material, Figures S1–S9. Further, the optimized structures, molecular properties and chemical-physical features of the fifty test compounds were obtained [
41,
42,
43,
44]: e.g., area, volume, polar surface area (PSA and TPSA), ovality, polarizability, dipole moment, water–octanol partition coefficient (log
P and milog
P), one descriptor related to the flexibility of molecules and the number of rotatable bonds (nrotb), respectively, as well as the RO5 (rule of the five) parameter following Lipinski’s criteria for estimating the feasible oral bioactivity of a drug/compound [
45,
46].
Given the high share of oral bioavailability of a compound, the results of DFT calculations (
Table 1,
Table 2,
Table 3,
Table 4,
Table 5,
Table 6,
Table 7,
Table 8 and
Table 9) were plotted as milog
P along nrotb, as shown in
Figure 1,
Figure 2,
Figure 3,
Figure 4,
Figure 5,
Figure 6,
Figure 7,
Figure 8 and
Figure 9.
As expected, area (A) and volume (V) of all the series of tested compounds varied in the same trend as the molecular weight (MW), while polar surface area (PSA) stood out independently, depending on the number and configuration of hydroxyl groups in the specific compound. PSA values were due to the van der Waals surface of all nitrogen and oxygen atoms and any hydrogens attached to these electronegative atoms. PSA reflected the presence of oxygen sp
3 or sp
2 and increased mainly with the increase of hydroxyl groups, which are very abundant in polyphenols and their glycosides, but also carbonyl and methoxy groups, and their disposal on the molecule’s skeleton; it can be noticed that hydrogen and hydroxyl ions are directly responsible for the chemical and physical stability and degradation of an active compound (e.g., the active ingredient in a drug), being involved in specific acid-base catalysis [
47], and are considered during preformulation step assays of drug candidates. In addition, the PSA value is important when estimating the rate at which molecules can go through hydrophilic or hydrophobic media; therefore, it influences the bioavailability of a compound. In this context, correlative to the fifty plant polyphenols studied, most indicated a predictable PSA dynamic; the PSA value rose from flavonol aglycone to flavonol (poly)glycoside and showed an increasing magnitude from the 3-
O-rhamnoside to 3-
O-glucoside and 3-
O-galactoside derivatives. Flavone derivatives were less predictable concerning the PSA dynamic, and luteolin derivatives performed, once again, a particular behavior compared to other flavonoid derivatives, also noticed in the case of antioxidant activity, i.e., resistance to hydrolysis and high difficulty in the process of chemical derivatization. Flavanones and isoflavones also showed an increasing magnitude with an increasing number of hydroxyl groups. In the specific case of phenolic acids, the PSA value also rose from aglycone to its esterified/condensed forms and showed an increasing magnitude with the increasing number of hydroxyl groups. In terms of TPSA (predicted with Molinspiration), a similar dynamic with PSA was observed, and the same behavior was noticed for luteolin derivatives. TPSA represents the topological surface area calculated as a sum of the fragment-based contribution. The available surface area influences the intermolecular contact; therefore, the compact shape of the molecules provides a smaller available surface area for intermolecular interactions and weaker dispersion forces. Interactions with amino acid residues of the target proteins occur within the accessible area of ligands too.
The ovality index is a more complex molecular descriptor associated with the effective molecular shape of molecules obtained as area, volume and PSA from the three-dimensional, space-filling model; it represents the deviation from spherical shape, where 1.0 corresponds to spherical top molecules and values greater than (>) 1.0 indicate deviation from the sphere. Generally, the values of ovality index increase with increasing linearity of the molecule. In the present study, the smallest deviations from the spherical shape were shown by quercetin (1.40), kaempferol (1.38), myricetin (1.41), luteolin (1.38), apigenin (1.37), naringenin (1.4), biochanin A (1.40), quinic acid (1.28), p-coumaric acid (1.29) and salicylic acid (1.21). These data were added to another computer-assisted drug design (CADD) study of 16 flavonoids compounds which concluded that 5,7-dihydroxy flavonoid compounds were the best trypsin and trypsin-like enzyme inhibitors; quercetin, myricetin and morin had the best structural configuration due to their suitably located hydroxyl groups and their planar configuration [
47].
Complementary predictive data, for better understanding of the electronic structure of compounds, are given by dipole moment and polarizability (alpha polarizability parameter), related to aspects of the electronic, vibrational structure and bonding [
48]. The logarithm of the water–1-octanol partition coefficient (Log
P) of compounds, an indication of lipophilicity, is used in the pharma field to predict properties and transport behavior of molecules. Water mimics the aqueous cellular or extracellular media, while 1-octanol is employed as an organic model for the lipid biological membrane; yet this organic solvent has some limitations due to the presence of its free hydroxyl group, water inclusion (4%,
v/v) and hydrogen bonding capability. Its use can attenuate the crossing membrane ability for compounds able to form H bonds and considered to have high hydration potential. However, Log
P is an accessible and widely used descriptor for predicting lipophilicity in pharma screenings. In Spartan software, Log
P values are estimated by employing the widely used atomic contribution method of Ghose, Pritchett and Crippen [
49]; developed by a Molinspiration methodology, miLog
P is a sum of fragment-based contributions and correction factors. Finally, a zero or single violation of Lipinski’s criteria (RO5 parameter) is assigned to feasible, orally active drugs [
50].
Thereby, the computational analysis revealed that 33 of the 50 phytocompounds tested were in good agreement with Lipinski’s rule. Generally, the lead compounds (aglycones) of each flavonoid series had no Lipinski’s violation: e.g., Q, K, L and M (1 violation). Interestingly, apigenin derivatives (A) were all feasible compounds excepting the rutinoside derivative; non-substituted flavanones (N, H), as well as isoflavones and all flavan and hydroxycinnamic acid derivatives, showed zero or one Lipinski’s violations. However, the molecules that remained feasible following the double analysis, RO5 and log
P, were as follows: apigenin and apigenin-7-
O-rhamnoside, naringenin, hesperetin, genistein, daidzin, biochanin A and formonetin in the flavonoid series and all hydroxycinnamic acids and all hydroxybenzoic acids excepting the condensed form of ferulic acid, namely ellagic acid. Veber’s rule [
51] supplements Lipinski’s filter by introducing limitation to polar surface area (PSA) values (no larger than 140 Å
2) and to the number of rotatable bonds (recommended less than 10) for good oral bioavailability. Furthermore, the flexibility of a molecule (estimated by the nrotb parameter) plays an important role in establishing interactions within the amino acids from the active binding site of the enzyme; no rotatable bond on the structure indicates rigid molecules. Relative to the fifty test compounds, the computational analysis indicated that the more flexible compounds were those containing rutinoside residues (nrotb = 6), glucoside and galactoside residues (nrotb = 4) and rhamnoside residue (nrotb = 3); curcumin (nrotb = 8), rosmarinic acid (nrotb = 7), chlorogenic acid and its isomers (nrotb = 5) and sinapic acid (nrotb = 4) were also flexible molecules. The number of rotatable bonds and molecular flexibility also provide clues as to whether the compound crystallizes or not [
52]; increased flexibility means a lower tendency to crystallization [
53,
54]. Furthermore, acknowledged as a measure of the hydrophilicity of an orally administered compound and, at the same time, a measure of the molecule flexibility, milog
P analysis in correlation with nrotb generates a bioavailability scale for a series of compounds studied. From a computational point of view “reduced molecular flexibility, as measured by the number of rotatable bonds, and low polar surface area or total hydrogen bond count (sum of donors and acceptors) are found to be important predictors of good oral bioavailability, independent of molecular weight” [
51]. In addition, a planar conformation of a molecule results in a better interaction with serum albumin and, therefore, in a better bioavailability in humans [
55].
Applied to the present study, the computational analysis on milog
P, along with nrotb (
Figure 1,
Figure 2,
Figure 3,
Figure 4,
Figure 5,
Figure 6,
Figure 7,
Figure 8 and
Figure 9), suggested the following correlations between the bioavailability of the fifty compounds in the series: flavonols and flavones subclasses’ bioavailability generally increased with a decreasing number of hydroxyl groups at C and B rings (the summed milog
P values of similar derivatives in each series decreased from the myricetin to kaempferol series and from luteolin to apigenin series, explained by the decrease in the number of hydroxyl groups at flavan core); flavan, flavanone and isoflavone derivatives generally had a better bioavailability than flavone and flavonol derivatives; in the flavonol series, the best bioavailability values were shown by aglycones followed by the -3-
O-rhamnoside, -3-
O-galactoside, -3-
O-glucoside and -3-
O-rutinoside series; in the flavone series, the best bioavailability was revealed by aglycones followed by -6/8-
C-glucoside, -7-
O-glucoside, -5-
O-glucoside and -7-
O-diglycoside; rutinoside and, generally, diglycoside derivatives revealed a lower bioavailability than monoglycoside derivatives, while rhamnoside derivatives largely showed the best bioavailability among the monoglycosides studied; genistein derivatives showed the best bioavailability values among the isoflavone series tested. In the phenylcarboxylic acid series, good bioavailability results were noticed in the hydroxybenzoic acid series; in the hydroxycinnamic acid series, aglycone compounds, especially curcumin, but also rosmarinic acid in comparison with other esterified homologues, appeared to have the best ability to pass the cell membranes in humans.
For further comparison with in vitro and in vivo data, it should be first noted that previous pharmacological studies demonstrated that the bioavailability of the plant compounds administered orally is the result of numerous biological processes; for example, the bio-accessibility of the active compounds, their intestinal and hepatic metabolism in correlation with their transformation by gut microflora, the nature of conjugates after hepatic metabolization and their plasma kinetics and binding to albumin, as well as their absorption at the level of the target cell, their accumulation in the specific tissues and urinary and biliary excretion.
In this way, a comparison with bioavailability data resulting from computational in silico studies could be a useful tool for better understanding plant compounds and herbal-derived medicines but also food products in relation to their health benefits and limits [
56,
57,
58]. Previous clinical data revealed a low bioavailability for most of the plant compounds investigated; it was stated that about 5 percent of the daily oral intake of polyphenols is absorbed and metabolized at the level of the intestine and liver. For the most part, vegetal polyphenols in glycoside and esterified forms are generally thought to be degraded by the intestinal microflora and largely excreted as feces, excepting small quantities which are absorbed and metabolized at the level of intestine and liver [
59]; as general rule, polyphenol aglycones proved to have a better bioavailability at the level of digestive system in humans (due to their lipophilic character), but they are present only in very small quantities in natural vegetal sources. Therefore, the glycoside forms and esterified derivatives are the predominant polyphenol compounds in the human diet.
In the specific case of flavonoids, the type and the number of units in glycoside moiety both play a crucial role in a polyphenol compound’s bioavailability; for example, studies revealed that, while glucoside derivatives are largely absorbed and metabolized at the level of small intestine, galactosides, rhamnosides, arabinosides, xylosides and glucuronic acid derivatives, such as polyglycoside derivatives (e.g., rutinoside), are metabolized at the level of the colon. The process is assisted by the bacterial hydrolases from the microbiota: the only ones that can cut esters and release aglycones from their glycoside moieties [
58]. Another parameter that influences the bioavailability of the plant compounds is their affinity for human serum albumin [
58,
59]. Studies also showed that, even if the lipophilic compounds pass more successfully through the cell membranes, high hydrophilicity of a molecule increases the probability of binding to albumin; quercetin aglycone indicated the best interaction with albumin, explained by its planar conformation [
56]. Concerning phenylcarboxylic acid derivatives’ bioavailability in vivo, data indicated that hydroxybenzoic acid derivatives are generally of low nutritional interest and, therefore, less studied; gallic acid is the most studied compound in the series, and it was proved to have high bioavailability in humans [
58,
59]. Hydroxycinnamic acid derivatives are of more nutritional interest, especially due to their transformation (e.g., hydrolysis to aglycones and multiple other isomerization transformations) during the sterilization, fermentation or freezing processes of fruits and vegetables. Studies revealed that, while the free aglycones are rapidly absorbed from the small intestine [
60,
61], the esterified forms (e.g., chlorogenic acid and its isomers) have a much lower bioavailability in humans [
62,
63,
64], in the most part being metabolized by the intestinal hydrolases from microflora [
65]. Finally, the high bioavailability noticed in vivo in the particular case of quercetin-4′-
O-glucoside/spireoside from onions, of quercetin-3-
O-glucoside/isoquercitrin from apples and of quercetin-3-
O-galactoside/hyperoside from the St. John’s Wort herb was not sustained by the in silico computational study; these exceptions are explained by a cumulus of metabolic particularities and solvent effects, including the existence of specific intestinal hydrolases and the copresence of alcohol [
58,
66] or of pectins, surfactants and bitter compounds [
67]. Similarly assigned as having good bioavailability in in silico studies, naringenin flavanone (found in citrus) can specifically increase the bioavailability of numerous xenobiotics in vivo by modulating the cytochrome P450 enzyme family function [
68,
69].
In summary, in silico results confirmed in vitro and in vivo data regarding the high bioavailability of soy isoflavones and better bioavailability of free aglycones in comparison with esterified and glycosylated forms. The computational study also revealed a high bioavailability for flavanones naringenin and hesperitin, apigenin and kaempferol derivatives and catechin and epicatechin flavan derivatives; curcumin, rosmarinic acid and salicylic acid were also revealed to have high bioavailability in the computational approach. In comparison, in vivo studies indicated the high bioavailability of caffeic and ferulic acids, while proanthocyanidols and gallocatechins (from green tea) were ranked last [
58]; however, some clinical data proved anthocyanins (flavan derivatives) are fully absorped in humans [
70]. Rhamnoside derivatives, the predominant polyphenolic compounds found in plant-derived products and the human diet [
71], which were reported to have the lowest bioavailability by in vivo studies [
72], were revealed to have the higher bioavailability values (miLog
P/nrotb comparison) among the flavonoid series in the computational approach.
2.1.1. Predicted Bioactivity
Table 10 gives the bioactivity scores predicted with Molinspiration software (Slovensky Grob, Slovak Republic:
https://www.molinspiration.com (accessed on 2 June 2021) of the fifty test compounds towards six cell modulators, namely, G-protein-coupled receptors (GPCRs), ion channel modulators, kinase inhibitor, nuclear receptor, protease inhibitor and enzyme inhibitor activity. A high bioactivity score suggests a greater probability of a test molecule being active against a selected target. According to similar predictive studies [
73,
74], active molecules exhibit a bioactivity score of more than 0, moderately active between −5.0 and 0.0 and inactive less than −5.0. The general behavior and the compounds in the series with the highest activity (stimulatory activity) were, therefore, analyzed as follows.
Overall, comparative analysis of the fifty test compounds on the six cell modulators indicated good to moderate bioactivity scores; a remarkable similitude between the flavonoid series was also observed (see the
supplementary material plotting their comparative activity areas,
Figures S10–S15), the differences between the compounds in the series mostly being the intensity of the effects on the cell modulator.
Since G-protein-coupled receptors (GPCRs) belong to a large family of signaling proteins which mediate the cellular responses to numerous external molecules (ligands), such as hormones, cytokines, neurotransmitters and various metabolites, 34% of FDA-approved drugs target the 108 members of this family of cell surface receptors [
75]. According to the literature data [
76,
77,
78], GPCRs are involved in numerous physiological processes, including reactions upon the visual, gustatory and smell senses, behavioral and mood regulation, immune system regulation, autonomic nervous system transmission (responsible for the control of blood pressure, heart rate and digestive processes), cell density sensing, homeostasis modulation, tumor cell growth and metastasis and also hormone binding (through cAMP–kinase stimulation), thus, allowing the transcription processes in cells. The computational analysis carried out on the fifty polyphenolic compounds generally indicated moderate activity for flavonoid subclasses, the most active compounds against GPCR function being flavan derivatives, catechin and epicatechin, respectively (+0.41). Phenylcarboxylic acid derivatives were less active than flavonoid derivatives, apart from caffeic acid derivatives (chlorogenic and rosmarinic acid esters), which were shown to have higher potency (+0.29); compounds which less significantly affected the GPCRs’ activity were salicylic acid followed by vanillic acid > gallic acid > p-coumaric acid > caffeic acid > ferulic acid > sinapic acid > ellagic acid > quinic acid.
Ion channels are pore-forming membrane proteins which allow ions (e.g., calcium, potassium, sodium, chlorine) to pass through the channel pore, thus, principally controlling the flow of the ions across the cell membrane and, therefore, the electrolyte balance of the body and the cell volume. However, they are also involved in body cell signaling activity through ligand-gated ion channel signalization molecules, 5-HT3 (5-hydroxytriptamine receptor mediates neuronal depolarization and excitation), GABBA (gamma amino butyric acid is the major inhibitory neurotransmitter in the brain), glutamate (the most important neurotransmitter in the nervous system) and nicotinic receptor (helping the transmission of outgoing signals from the sympathetic and parasympathetic system to the whole body) [
79,
80,
81]. The computational analysis indicated almost identical behavior in the case of GPCRs’ function: a generalized, moderate activity of the compounds tested and the same amplified activity against ion channel activity of flavan derivatives and caffeic acid derivatives. The less active compounds against ion channel activity were flavonoid rutinosides, isoflavones, flavanones and phenylcarboxylic acid aglycones. It must be noted that ion channel function controls every aspect of the digestion process (e.g., fluid secretion and absorption, motility and visceral sensitivity), irritable bowel syndrome manifestations being mostly driven by the altered ion channel expression and function [
82]. These data are helpful in selecting the most appropriate plant-derived drug therapy in the situation of a susceptible patient.
Kinases are some of the most important enzymes in developing human body physiology since they catalyze the transfer of a phosphate group through which the high energy of an ATP molecule is donated as a phosphate group to a substrate molecule; this process is critical in all aspects of cell metabolism in prokaryotic processes, for example, cell signaling, protein regulation and cellular transport, all secretory processes and many other biological processes [
83]. Related to the fifty compounds studied, all flavonoid compounds, particularly the aglycone forms in the series, indicated inhibitory potency against the activity of kinases (e.g., quercetin 0.28, myricetin 0.28, luteolin 0.26 and kaempferol 0.21); flavanone and isoflavone derivatives, similar to phenylcarboxylic acid aglycones (e.g., salicylic acid, vanillic acid, p-coumaric acid and gallic, caffeic and quinic acids), exhibited the weakest ability to influence the activity of human kinases.
The nuclear receptor superfamily comprises transcriptional factors (proteins) involved in thyroid/steroid hormones sensing. They can directly interact with the DNA molecule (by binding condensed chromatin templates); therefore, they control the gene expression and corresponding cell (embryo and adult) development, homeostasis and metabolism [
84,
85]. Their natural ligands are in the series of lipophilic substances (e.g., vitamins A and D) [
86]; in the series of phytocompounds studied, the most augmented activity against nuclear receptor function (measuring from 0.57 to 0.74) was registered in the case of flavan, catechin and epicatechin and caffeic acid derivatives (e.g., chlorogenic acid, isochlorogenic acid, neochlorogenic acid > catechin, epicatechin > rosmarinic acid). Quercetin (0.36), hesperitin (0.38) and luteolin (0.39) also showed real inhibitory potency; the compounds with the weakest activity against nuclear receptor activity were in the series of hydroxybenzoic acid derivatives (e.g., salicylic acid, vanillic acid and gallic acid).
Protease inhibitors are basically compounds that can bind proteolytic enzymes (namely proteases) and block their function in the body. Digestion and healing wounds are two major examples of biological process which cannot be achieved without the activity of proteases, and HIV inhibitors are an example of potential use and applicability [
87]. Related to the digestion process, studies indicated that patients taking protease inhibitors started to manifest important side effects such as “⋯new or exacerbated cases of diabetes or hyperglycemia, hemolytic anemia, spontaneous bleeding in hemophiliac patients, and changes in body composition.” [
88]. All plant compounds tested indicated a moderate protease inhibitor activity (bioactivity scores less than 0), apigenin derivatives (from 0.01 to 0.04), catechins (+0.26) and caffeic acid derivatives (from 0.15 to 0.27) proving the highest scores.
Enzyme inhibitor activity is likely the parameter with the highest negative potential upon the digestion process, aside from ion channel inhibition. As shown in
Table 10, all the polyphenolic compounds studied acted as enzyme inhibitors, the usually naturally occurring glycosylated and esterified forms being occasionally more active than aglycones forms. Computed as having a high bioactivity score in the investigated flavonoid series (0.43), an additional concern comes from quercetin-4′-
O-glucoside (spireoside from onion), which is known to have the highest bioavailability in humans (counted at about 42% from the ingested weight). Aside from this, the most frequent polyphenol compounds in the foods to show the biggest bioactivity scores (0.62 and 0.47) were chlorogenic acid esters and flavan derivatives, together confirming the antinutritional potential of polyphenols in humans.
In support of these findings, studies on 21 flavonoid compounds indicated that luteolin, luteolin-7-
O-glucoside, amentoflavone and daidzein were the most powerful alpha-glucosidase and alpha-amylase inhibitors, even stronger than acarbose [
89]. Other studies on 14 compounds of plant origin indicated six phenolics with certain inhibitory activity upon trypsin activity, active at concentration values (IC
50) ranging from 3.7 to 15.4 µM; they were silybin (3.7 µM), hypericin (4.5 µM), sennoside A and B (6.1 and 10.6 µM), hyperoside (14.5 µM) and quercetin (15.4 µM). Studies also demonstrated that a glycoside chain in position 3 of the flavan core led to high inhibitory potency [
90]. Furthermore, a computer-assisted drug design (CADD) study upon 16 flavonoid compounds revealed that 5,7-dihydroxy derivatives were the best trypsin and trypsin-like enzyme inhibitors, quercetin, myricetin and morin compounds having the best structural configuration due to their suitably located hydroxyl groups and planar configuration as well. The specific compounds and their IC
50 (µM) values were as follows: quercetin (10 µM), myricetin (15 µM), morin (27 µM), galangin (36 µM), isorhamnetin (40 µM), fisetin (46 µM), kaempferol (60 µM), acacetin (28 µM), apigenin (40 µM), baicalein (55 µM), 7,8-dihydroxyflavone (657 µM), chrysin (>1000 µM), 6/7-di-hydroxyflavones (>1000 µM), naringenin (484 µM) and biochanin A (134 µM) [
47]. It was observed that quercetin had the lowest IC
50 value in the series, meaning it had inhibitory potency upon the digestive protease enzymes at doses similar or lower than that of most beneficial activities in vitro.
Proving this, in vitro studies aiming to assess the defending activity of quercetin against cell death (endothelial cells, human skin fibroblasts and keratinocytes) induced by intracellular peroxides generated by buthionine sulfoximine (an irreversible inhibitor of glutathione synthesis) indicated that the protective effect of quercetin is manifested at EC
50 values between 30 and 40 µM [
91]. Studies regarding the antiarthritic, anti-inflammatory and antioxidant activity of nine South African plants used traditionally to treat arthritis [
92], also revealed IC
50 values from 11.89 to 53.78 µg/mL. Other studies [
93] regarding the in vivo results (animal models) of the antidiabetic, anti-inflammatory, antioxidant, antimicrobial, anti-Alzheimer’s, antiarthritic, cardiovascular and wound-healing effects of quercetin administered orally in rats and mice indicated the following data: antidiabetic activity occurs at 10–100 mg/kg body, anti-Alzheimer activity at 10–50 mg/kg body, antiarthritic activity at 30 mg/kg body, antimicrobial effects at 5–30 mg/kg body, liver protection at 100 mg/kg body, antioxidant effects at 30 mg/kg body and protective cardiovascular effects up to 1.5 g/kg body; at the same time, in vitro results on different types of cancer cell indicated an inhibitory activity in the interval 5–50 µM. In summary, since the inhibitory activity of quercetin on the digestive enzymes (computed at IC
50 = 10 µM) occurs at lower concentrations than that necessary for any other beneficial effect (most of values being over 10 µM in vitro and 10 mg/kg body in vivo), the food products and, especially, plant-based drugs based on quercetin and its derivatives could have this potential negative side effect on the digestion at humans.
Using the specialized database [
94],
Table S1 summarizes the mean content (mg per 100 g/mL) of the studied compounds in food products (e.g., cereals, fruits, vegetables, spices and herbs), highlighting the richest vegetal sources reported. The achieved analysis indicated that the sub(sub)classes of plant polyphenols in the present study mostly fell in the range of 0–50 mg per 100 g of product. The vegetal sources over 50 mg per 100 g of product were in the interest area for inhibitory potency, but the conclusion could be drawn by also considering the frequency and the amount of the product used in the daily diet of humans. In this way, in the series of flavonoid derivatives, flavan-3-ols were emphasized as the dominant polyphenolic compounds in the food products, while caffeic acid derivatives were likely the dominant bioactive compounds in the series of phenolic acid derivatives. It can be concluded that, excepting cocoa, chocolate, chestnut, coffee, tea (black, green), plum and berry products, which can each bring between 250 and 500 mg of a polyphenol subclass per day, all other food products do not reach high concentrations of specific polyphenol compounds in daily food. Thus, the main concern regarding plant polyphenol inhibitory activity on the digestive enzymes in humans is basically through the consumption of commercial products usually recommended in doses up to 1000 mg per day. In conclusion, the recommendation to supplement the usual diet with plant-derived products should be strictly made by specialists.
Finally, it is remarkable that birds developed several defense mechanisms to counteract plant polyphenols’ harmful, antinutritional effects, specifically by developing an alkaline pH gut, by secreting high contents of surfactants to decrease the polyphenols’ affinity in the intestine and by the presence of a peritrophic membranes and mucus able to absorb tannins after that excreted in the feces [
95].
2.1.2. Principal Component Analysis of the Test Parameter
Principal component analysis (PCA) is a statistical tool for the identification of linear combinations of the variables which account for certain proportions of the variance of the set of variables. The selection is based on the eigenvalues of the dispersion matrix of variables. The principal components are associated with decreasing eigenvalues and, therefore, share the amount of variance. Usually, the first few principal components account for virtually all the variances. PCA also represents the pattern of similarity of the observations and the variables by displaying them as points in maps [
96,
97,
98,
99]. The predicted bioactivity score data in
Table 10 were processed by the PCA XLSTAT extension of Excel. The PCA correlation matrix (
Table 11) showed a good correlation (r = 0.946) between GPCR and protease inhibitor parameter and a moderate correlation between GPCR and nuclear receptor (r = 0.799), enzyme inhibitor and nuclear receptor (r = 0.797) and enzyme inhibitor and protease inhibitor (r = 0.835), respectively.
Figure 10 and
Table 12 are related to the eigenvalues which reflect the quality of the projection from the F-dimensional initial (F = 6) to a lower number of dimensions. From
Table 12 it can be observed that the first eigenvalue equaled 4.528, representing 75.464% of the total variability. Each eigenvalue corresponds to a factor and each factor to one dimension. A factor is a linear combination of the initial variables, and all the factors are uncorrelated (r = 0). The eigenvalues and the corresponding factors are sorted by descending order of how much of the initial variability they represent (converted to %). Specifically, the first two factors allowed 87.33% of the initial variability of the data.
The correlation circle (
Figure 11) below, on the axes F1 and F2, shows a projection of the initial variables in the factors space. When two variables are far from the center (as depicted for GPCR and nuclear receptor and enzyme inhibitor and nuclear receptor), if they are close to each other, they are significantly positively correlated (r close to 1); conversely, as observed for kinase and ionic channel, when they are almost orthogonal to each other, they are not correlated (r close to 0).
The correlation circle is also useful in interpreting the meaning of the axes. In this case, the horizontal axis is linked with GPCR, protease and enzyme inhibitor; as proof, in the squared cosines of the variables F1–F5 (
Table 13), the greater the squared cosine, the greater the link with the horizontal axis.