Phenolic Acids from Lycium barbarum Leaves: In Vitro and In Silico Studies of the Inhibitory Activity against Porcine Pancreatic α -Amylase

: Nowadays, bioactive compounds from vegetable food and waste are of great interest for their inhibitory potential against digestive enzymes. In the present study, the inhibitory activity of methanolic extract from Lycium barbarum leaves on porcine pancreas α -amylase has been studied. The α -amylase inhibitory activity of the constituent phenolic acids was also investigated. The leaves were extracted by ultrasound-assisted method, one of the most e ﬃ cient techniques for bioactive extraction from plant materials, and then the phenolic acids were identiﬁed by Accurate-Mass Quadrupole Time-of-Flight (Q-TOF) Liquid Chromatography / Mass Spectrometry (LC / MS). Chlorogenic and salicylic acids were the most abundant phenolic acids in L. barbarum leaf extract. The inhibitory e ﬀ ect against α -amylase, determined for individual compounds by in vitro assay, was higher for chlorogenic, salicylic, and ca ﬀ eic acids. L. barbarum leaf extract showed an appreciable α -amylase inhibitory e ﬀ ect in a concentration-dependent manner. Docking studies of the considered phenolic acids into the active site of α -amylase suggested a conserved binding mode that is mainly stabilized through H-bonds and π - π stacking interactions.


Introduction
Despite the large number of synthetic pharmaceuticals, interest into natural products is increasing. The agricultural field offers a great opportunity thanks to the large quantities of byproducts of plant food processing, a promising source of biologically active compounds. In this context, food waste such as leaves are a cheap and available source of naturally-occurring bioactives, potentially useful for the development of functional foods and food supplements [1,2]. Starting from plant materials, numerous and different processes are used to obtain single bioactive substances or compounds belonging to the same chemical category. The recovery of high added-value components from food wastes can be carried out by traditional or innovative technologies. Among the latter, ultrasound-assisted extraction (UAE) can be considered a very successful and efficient extraction process for bioactive extraction [2].
Previous studies have reported the phytochemical composition of Lycium spp. berries and the availability of their bioactive compounds, including carotenoids, polyphenols, phytosterols, and phenolic acids [3][4][5][6]. Lycium leaves are interesting byproducts with health properties, due to the presence of polyphenols [7,8]. Nowadays, these bioactives are considered functional ingredients in the composition of different phytopharmaceuticals and dietary supplements due to numerous biological properties, among which are antioxidant, anti-inflammatory, and anti-hypertensive [9]. In addition to these activities, it has been reported that chlorogenic acid (CHA) shows an inhibitory effect on α-amylase [10]. This enzyme, together with other digestive enzymes (e.g., α-glucosaccharase, transglycosylase), is involved in the in vivo hydrolysis of starch into glucose, which, transported into blood, leads to postprandial hyperglycaemia. In particular, α-amylase is the key digestive enzyme that acts on starch to release dextrin, disaccharides, or other low molecular-weight carbohydrates, which will be then broken down into glucose [11]. Since high blood glucose levels can be associated with diabetes or obesity, the inhibition of α-amylase might be an effective way to reduce starch digestibility and thereby relieve postprandial glycaemia [11].
It has been reported in the literature that the extracts from leaves of various origin possessed inhibitory activity against α-amylase [12][13][14]. Abdelli et al. [15] evaluated the inhibitory activity against α-amylase of tannic acid, catechin, gallic acid, quercetin, and epicatechin by in silico approach, revealing that tannic acid can be used as alternative drug for the regulation of postprandial hyperglycaemia. Demir et al. [16] reported that genistein, tangeretin, pelargonidin, formononetin, and delphinidin showed antidiabetic properties. L. barbarum leaves also showed inhibition activity against various enzymes, among which cholinesterase, α-amylase, and α-glycosidase [8].
In this work, L. barbarum leaves were extracted with methanol by UAE and the phenolic acid profile was investigated by ultra-high-performance liquid chromatography quadrupole-time-of-flight tandem mass spectrometry (UHPLC-Q-TOF MS/MS) analysis. The identified phenolic acids were studied for α-amylase inhibitory activity by in vitro approach. The objective of this research was also to assess a putative binding mode of active phenolic acids to human pancreatic α-amylase. To the best of our knowledge, this is the first time that the evaluation of the inhibitory effect of individual phenolic acids, identified in the leaf extract, against porcine pancreatic α-amylase is studied.

Plant Sample
Intact fresh leaves from L. barbarum, harvested in 2019 in Umbria (central Italy), were dried to constant weight in a ventilated oven (Binder, Series ED, Tuttlingen, Germany) at 40 • C for 72 h. Then the dried leaves were ground to obtain a fine powder and stored in amber glass vial at room temperature in the dark, until extraction.

Ultrasound-Assisted Extraction (UAE) of L. barbarum Leaves
Dried L. barbarum leaves (330 mg) were extracted with pure methanol (20 mL) using an ultrasonic bath (mod. AU-65, ArgoLab, Carpi, Italy) with ultrasonic power of 180 W. The extraction was carried out for 30 min at 45 • C, then the extract was filtered and collected in amber glass vials until the further analysis. The extraction was repeated three times.

Determination of Total Phenol Content (TPC)
The TPC was determined spectrophotometrically according to the method previously reported [17], based on redox reaction of phenols with Folin−Ciocalteu's reagent. Lambda 20 spectrophotometer (PerkinElmer, Inc.; Waltham, MA, USA) was used to measure the absorbance at 765 nm. The TPC was expressed as mg of gallic acid equivalents (GAE) per gram of dry leaves (mg GAE/g).

Free Radical-Scavenging Activity Using DPPH Assay
The measurement of the scavenging ability of phenolic extract towards the stable DPPH radical was performed by DPPH assay, using the procedure previously reported [18]. An aliquot (3.9 mL) of a 0.0634 mM DPPH solution in methanol (95%) was added to L. barbarum leaf extract (100 µL) and vortexed for about 5-10 s. After 30 min, the change in the absorbance was measured at 515 nm using the spectrophotometer reported in 2.4 Paragraph. The DPPH antiradical capacity was expressed as mg Trolox Equivalents (TE) per gram of dry leaves (mg TE/g).

Free Radical-Scavenging Activity Using ABTS Assay
ABTS assay was carried out following the procedure described by Urbani et al. [19]. The ABTS stock solution was prepared by mixing 7 mM ABTS and 2.45 mM of potassium persulphate as the oxidant agent. The sample extract (200 µL) was added to ABTS +· solution (1800 µL) and after 10 min the absorbance at 734 nm was measured using the spectrophotometer reported in 2.4 Paragraph. The ABTS antiradical capacity was expressed as mg TE/g.

Ferric Reducing Antioxidant Power (FRAP) Assay
The reducing capacity of leaf extracts was determined using a modified method of the FRAP assay [20]. Freshly prepared FRAP reagent (2.0 mL) was mixed with L. barbarum extract (100 µL) and distilled water (900 µL), then the sample was left away from light at room temperature. After 30 min the absorbance at 593 nm was measured using the spectrophotometer reported in 2.4 Paragraph. Distilled water was used as blank. The FRAP value was expressed as mg TE/g.

α-Amylase Inhibition Activity Assay
The inhibitory activity of α-amylase was carried out according to the procedure reported by Wu et al. [21]. L. barbarum extract solutions were mixed with α-amylase and maintained at 30 • C for 10 min. Then, soluble starch solution was added as substrate and the hydrolysis was carried out at 30 • C. After 3 min, DNSA reagent was added, and the mixture was placed in a boiling water bath at 80 • C for 15 min. Afterwards, the solutions were diluted with water, and then the absorbance at 540 nm (Abs extract) was measured using a Lambda 20 spectrophotometer. For each concentration of the L. barbarum extract, a blank incubation was prepared by replacing the enzyme solution with distilled water, to correct for absorbance due the L. barbarum extract (Abs blank extract). A control incubation, representing 100% enzyme activity, was carried out by replacing L. barbarum extract with the vehicle, and Abs control was measured. The same test was also carried out without the enzyme and Abs blank control was registered. The α-amylase inhibitory activity was calculated as percentage inhibition using the following formula: where ∆Abs control = Abs control − Abs blank control (2) ∆Abs extract = Abs extract − Abs blank extract (3) The concentration of the extract able to inhibit the α-amylase activity by 50% (IC 50 ) was calculated by regression analysis. Single phenolic acids were also tested for α-amylase inhibitory activity and their IC 50 values were determined as above described. Acarbose was also analyzed as a known α-amylase inhibitor.

Analysis of Phenolic Acids by Q-TOF-LC/MS
The ultra-high-performance Accurate-Mass Q-TOF-LC/MS analysis was performed using an Agilent Technologies (Santa Clara, CA, USA) 1200 Infinity Series LC coupled with an Agilent Technologies 6540 UHD Accurate-Mass Q-TOF-LC/MS. This device was equipped with an electrospray ionization Agilent Technologies Dual Jet Stream ion source (Dual AJS ESI). Chromatographic separation was carried out with an Agilent Infinity lab Poroshell 120 EC-C18 (3 × 100 mm, 2.7 µm) column. The injection volume was 10 µL. The mobile phase consisted of 0.1% formic acid in water milli-Q (solvent A) and acetonitrile (solvent B), at a flow rate of 0.5 mL/min, with the following gradient: 0-10 min, 5% B; 10-13 min, 95% B; 13-15 min, 95% B. The Q-TOF-MS conditions were the following: drying gas flow (N 2 ), 12.0 L/min; nebulizer pressure, 45 psi; gas drying temperature, 370 • C; capillary voltage, 3500 V; fragmentor voltage,110 V; skimmer voltage 65 V and octopole RF peak, 750 V. Dual AJS ESI interface was used in negative ionization mode and negative ions were acquired in the range of 100-1100 m/z for MS scans, and 50-600 m/z for auto MS/MS scans, at a scan rate of 5 scans/s for MS and 3 scans/s for MS/MS, respectively. Automatic acquisition mode MS/MS were carried out using the following collision energy values: m/z 20 eV, m/z 30 eV, and 40 eV. Internal mass correction was enabled using two reference masses at 121.0509 and 922.0098 m/z. Instrument control and data acquisition were performed using Agilent MassHunter Workstation software B.08.00. All the MS and MS/MS data of the validation standards were integrated by MassHunter Quantitative Analysis B.10.0 (Agilent Technologies).

Molecular Docking
The crystal structure of human pancreatic α-amylase (PDB ID: 5KEZ) was downloaded from the Research Collaboratory for Structural Bioinformatics Protein Data Bank (RCSB PDB; https://www. rcsb.org/) [22], removing the peptide inhibitor from the active site. The optimization procedure was performed using the Protein Preparation Wizard tool implemented in Maestro (Maestro, Version 12.1, Schrödinger, LLC, New York, NY, USA, 2019). In particular, this procedure included the refinement of the hydrogen bond network that was performed using the default setting. The OPLS3e force field was used for the energy minimization of the protein structure, restraining atomic coordinates of heavy atoms. Atomic coordinates of chemical structures were retrieved from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/) [23]. LigPrep (v.8.4) was then used to add hydrogens, and Epik (Epik, Version 2.5, Schrödinger, LLC, New York, NY, USA, 2013) to generate tautomeric and protonation states at pH 7.0 ± 2.0. Next, energy minimization of chemical structures was carried out with the OPLS3e force field until reaching an energy gradient convergence criteria of 0.05 kJ/Å mol. Docking simulations were performed using Glide (v.8.4) [24,25]. Specifically, a grid box was defined with its center located on the center of mass of binding site residues in the reference structure 5KEZ. These residues included Ala128, Pro130, Tyr131, Ser132, Tyr151, Asp197, Lys200, His201, Glu233, Ile235, Asp300, and His305. The inner grid box was sized 10 × 10 × 10 Å. Docking studies were carried out using the standard precision (SP) method and the Gscore scoring function. The top ten scored binding poses were stored for each molecule. The binding mode for each compound was selected in terms of the best docked pose showing the highest number of ligand/target interactions (electrostatic interactions, hydrogen bonds, π-stacking/cation interactions, hydrophobic contacts) and lowest Gscore (kcal/mol). Experimental or calculated acidic constants (pKa) of phenolic acids were retrieved from DrugBank [23,26].

Statistical Analysis
Each analytical procedure was carried out in triplicate and the results, reported on dry leaves, were expressed as mean ± standard deviation (SD). For data analysis Microsoft Excel 2016 (Microsoft Corporation, Redmond, WA, USA) was used.

Characterization of L. barbarum Leaf Extract and α-Amylase Inhibition of Constituent Phenolic Acids
In this study, L. barbarum leaf extract was obtained by nonconventional UAE using methanol as solvent. This extraction process was chosen on the basis of the results obtained in a recent paper [7], which showed that UAE is a very successful and efficient technique for extracting phenols from L. barbarum leaves, also in comparison with microwave-assisted extraction. UAE methanolic extraction yield (20.39% ± 1.08 SD) was calculated taking into consideration the weight of the extract residue obtained after solvent removal and the initial weight of leaf powder [7]. The obtained residue was first characterized by evaluating the phenol content and antioxidant activity, then the phenolic acid profile was determined by Q-TOF-LC/MS analysis. The total phenolic content, determined according to the Folin−Ciocalteu method, was 7.75 mg GAE/g ± 0.43 SD; similar findings have been previously reported in other studies regarding Italian and Romanian L. barbarum leaves [7,8]. Mocan et al. reported values of 11.14 mg GAE/g dw (dry weight) and 11.98 mg GAE/g dw for Bigligeberry and Erma cultivars, respectively, while for wild-growing L. barbarum leaves a lower amount (2.49 mg GAE/g dw) was reported [8].
The evaluation of the antioxidant activity, based on different complementary assays, is a routine analysis commonly carried out to characterize vegetable extracts. The antioxidant capacity of L. barbarum leaf extract was measured using three spectrophotometric methods, DPPH and ABTS assays determining antiradical activity, and FRAP assay determining reducing power; the results are shown in Figure 1.

Statistical Analysis
Each analytical procedure was carried out in triplicate and the results, reported on dry leaves, were expressed as mean ± standard deviation (SD). For data analysis Microsoft Excel 2016 (Microsoft Corporation, Redmond, WA, USA) was used.

Characterization of L. barbarum Leaf Extract and α-Amylase Inhibition of Constituent Phenolic Acids
In this study, L. barbarum leaf extract was obtained by nonconventional UAE using methanol as solvent. This extraction process was chosen on the basis of the results obtained in a recent paper [7], which showed that UAE is a very successful and efficient technique for extracting phenols from L. barbarum leaves, also in comparison with microwave-assisted extraction. UAE methanolic extraction yield (20.39% ± 1.08 SD) was calculated taking into consideration the weight of the extract residue obtained after solvent removal and the initial weight of leaf powder [7]. The obtained residue was first characterized by evaluating the phenol content and antioxidant activity, then the phenolic acid profile was determined by Q-TOF-LC/MS analysis. The total phenolic content, determined according to the Folin−Ciocalteu method, was 7.75 mg GAE/g ± 0.43 SD; similar findings have been previously reported in other studies regarding Italian and Romanian L. barbarum leaves [7,8]. Mocan et al. reported values of 11.14 mg GAE/g dw (dry weight) and 11.98 mg GAE/g dw for Bigligeberry and Erma cultivars, respectively, while for wild-growing L. barbarum leaves a lower amount (2.49 mg GAE/g dw) was reported [8].
The evaluation of the antioxidant activity, based on different complementary assays, is a routine analysis commonly carried out to characterize vegetable extracts. The antioxidant capacity of L. barbarum leaf extract was measured using three spectrophotometric methods, DPPH and ABTS assays determining antiradical activity, and FRAP assay determining reducing power; the results are shown in Figure 1. The results obtained for the DPPH and ABTS radical-scavenging assays were 9.39 mg TE/g and 11.28 mg TE/g, respectively, while 8.25 mg TE/g was obtained by FRAP assay. The lower value obtained for DPPH assay in respect to ABTS assay can be attributed to the steric hindrance effects and the chemical characteristics of the phenolic compounds. Mocan et al. [8] reported different antioxidant values obtained by ABTS assay for leaves of two cultivars of Romanian L. barbarum and from wild-growing plants. Other authors studied the antioxidant activity of another plant source of chlorogenic and neochlorogenic acids, plum fruits of three different species, and found lower ABTS  Values of in vitro antioxidant activities of L. barbarum leaf extracts, evaluated by DPPH, ABTS, and FRAP assays.
The error bars represent the standard deviation (n = 3).
The results obtained for the DPPH and ABTS radical-scavenging assays were 9.39 mg TE/g and 11.28 mg TE/g, respectively, while 8.25 mg TE/g was obtained by FRAP assay. The lower value obtained for DPPH assay in respect to ABTS assay can be attributed to the steric hindrance effects and the chemical characteristics of the phenolic compounds. Mocan et al. [8] reported different antioxidant values obtained by ABTS assay for leaves of two cultivars of Romanian L. barbarum and from wild-growing plants. Other authors studied the antioxidant activity of another plant source of chlorogenic and neochlorogenic acids, plum fruits of three different species, and found lower ABTS values [27]. In another paper, Mocan et al. [28] studied the antioxidant capacity of L. barbarum and L. chinense leaves from Romania using DPPH, ABTS, and hemoglobin ascorbate peroxidase activity inhibition assays and found higher values for L. chinense than L. barbarum leaves. ABTS values similar to those obtained in this work were obtained for other vegetable waste, such as olive leaves harvested in different season from different Olea europaea cultivar [1]. However, it is important to highlight that it is not always possible to make comparisons with antioxidant data reported in the literature, in particular for DPPH and FRAP data, since the results of the same assay can be expressed with different measure unit. In this regard, it is known that antioxidant activity of leaf extracts is strongly dependent on the harvesting season, cultivar, and extraction conditions [1,20]. In this study, a Q-TOF-LC/MS analysis was carried out to investigate the qualitative and quantitative profile of phenolic acids in the UAE methanolic extract of L. barbarum leaves, considering that the main objective of this paper was the in vitro and in silico study of the inhibitory activity of this category of phenolic compounds against porcine pancreatic α-amylase. Figure 2 shows the chemical structures of the main phenolic acids identified in the methanolic extract of L. barbarum leaves, while Table 1 shows their content. values [27]. In another paper, Mocan et al. [28] studied the antioxidant capacity of L. barbarum and L. chinense leaves from Romania using DPPH, ABTS, and hemoglobin ascorbate peroxidase activity inhibition assays and found higher values for L. chinense than L. barbarum leaves. ABTS values similar to those obtained in this work were obtained for other vegetable waste, such as olive leaves harvested in different season from different Olea europaea cultivar [1]. However, it is important to highlight that it is not always possible to make comparisons with antioxidant data reported in the literature, in particular for DPPH and FRAP data, since the results of the same assay can be expressed with different measure unit. In this regard, it is known that antioxidant activity of leaf extracts is strongly dependent on the harvesting season, cultivar, and extraction conditions [1,20]. In this study, a Q-TOF-LC/MS analysis was carried out to investigate the qualitative and quantitative profile of phenolic acids in the UAE methanolic extract of L. barbarum leaves, considering that the main objective of this paper was the in vitro and in silico study of the inhibitory activity of this category of phenolic compounds against porcine pancreatic α-amylase. Figure 2 shows the chemical structures of the main phenolic acids identified in the methanolic extract of L. barbarum leaves, while Table 1 shows their content.  Data are reported as mean ± SD of three independent measurements (n = 3) and are expressed on dry leaves.
CHA was the main phenol compound (358.34 μg/g), followed by salicylic acid (239.02 μg/g). These two molecules are also reported in two systematic Lycium reviews [29,30]. CHA was previously described as dominant in the leaves of wild and cultivated L. barbarum [8,28,31,32], but also in L. chinense leaves [28,32]. Caffeic, chlorogenic, p-coumaric, and vanillic acids were also  Data are reported as mean ± SD of three independent measurements (n = 3) and are expressed on dry leaves.
CHA was the main phenol compound (358.34 µg/g), followed by salicylic acid (239.02 µg/g). These two molecules are also reported in two systematic Lycium reviews [29,30]. CHA was previously described as dominant in the leaves of wild and cultivated L. barbarum [8,28,31,32], but also in L. chinense leaves [28,32]. Caffeic, chlorogenic, p-coumaric, and vanillic acids were also certainly identified by Inbaraj et al. [31]. It is widely reported in the literature that the phenolic profile of leaves may strongly change according to the type of plant, seasonal conditions, and extraction method [1,7,20].
In order to evaluate the capacity of the L. barbarum extract to inhibit the α-amylase, and, therefore, the potential hypoglycemic effect, an in vitro assay based on the spectrophotometric evaluation of the hydrolytic activity of porcine pancreatic α-amylase was carried out. The linear regression equations, obtained plotting α-amylase % inhibition versus concentration of the standard phenolic acids and UAE methanolic extract are reported in Table 2. The R 2 values and the IC 50 values are shown in the same Table. The IC 50 value for Acarbose was 0.1 mg/mL, while Min and Han [33] reported a value of 0.45 mg/mL. The results revealed that the percentage of α-amylase inhibition increased for chemical standards in a dose-dependent manner, with a linear trend, as also reported in other papers [10,34], with good R 2 values (from 0.9987 of CHA to 0.9999 of sinapic acid). Higher α-amylase inhibitory activity was found for CHA, salicylic, and caffeic acids than the other constituents. On the other hand, the inhibitory activity of the L. barbarum methanolic extract was lower than the individual phenolic acids, probably due to the complexity of the chemical composition of vegetable matrices, as leaves. However, comparable results were reported by other authors [35], which found an IC 50 value of 23.7 mg/mL for ethanolic extract of dried Garcinia schomburgkiana leaves. Antidiabetic properties of phenolic compounds were reported in other papers, regarding phenolic compounds from leaves [8], fruits [36], and plants [37]. It has been reported that natural compounds, such as phenolics, may bind by hydrogen bonding the amino acid residues in the active sites of digestive enzyme. In this way, the hydrolysis reaction of digestive enzymes on carbohydrates may be inhibited and consequently slowed down their absorption [21]. Only few data on enzyme inhibition potential of Lycium leaves are reported, as for example Mocan et al. reported similar results for amylase inhibitory activity (0.24-0.26 mmol Acarbose equivalents/g extract) for L. barbarum considered samples [8].
The investigation of the amylase inhibitory activity of the individual phenolic acids showed that some molecules (chlorogenic acid, CHA, 2; caffeic acid, 4; salicylic acid, 3) had inhibitory activity against the key carbohydrate digestive enzyme, while other molecules (syringic acid, 1; p-coumaric acid, 6; vanillic acid, 5; sinapic acid, 7; vanillin, 8) showed lower inhibition potency. These results confirmed that CHA (2) has high inhibitory activity against α-amylase. It has been reported that this should be due to its ability to form quinones, rather than semiquinones [34]. Sinapic and caffeic acids are less active than CHA (2), as also reported by Funke and Melzig [34]. Notably, in this work the inhibition mechanism of the main phenolic compounds toward this key digestive enzyme was investigated by molecular docking analysis, in order to provide a better understanding of how the considered phenolic acids inhibit α-amylase activity.

Docking Studies
In order to infer a putative binding mode of active phenolic acids (1)(2)(3)(4)(5)(6)(7)(8) to the enzyme, docking studies were carried out using the human pancreatic α-amylase that shares 90% of sequence similarity with the porcine isoform ( Figure 3). inhibition mechanism of the main phenolic compounds toward this key digestive enzyme was investigated by molecular docking analysis, in order to provide a better understanding of how the considered phenolic acids inhibit α-amylase activity.

Docking Studies
In order to infer a putative binding mode of active phenolic acids (1)(2)(3)(4)(5)(6)(7)(8) to the enzyme, docking studies were carried out using the human pancreatic α-amylase that shares 90% of sequence similarity with the porcine isoform ( Figure 3). In particular, the crystal structure of the enzyme in complex with a peptide inhibitor (pdb code: 5KEZ) as reference structure was used [38]. Firstly, CHA (2, calcpKa = 3.3) was docked into the enzyme and the resulting binding mode is shown in Figure 4. CHA forms hydrogen bonds between the three hydroxyl groups of the cyclohexane ring and the side chains of Asp197, Glu233, and Asp300. An additional hydrogen bond is observed between the para hydroxyl group of the caffeic acid moiety and the side chain of Glu240, whereas the phenyl ring engages Lys200 with a -cation interaction (Figure 4).
In particular, the crystal structure of the enzyme in complex with a peptide inhibitor (pdb code: 5KEZ) as reference structure was used [38]. Firstly, CHA (2, calc pKa = 3.3) was docked into the enzyme and the resulting binding mode is shown in Figure 4. CHA forms hydrogen bonds between the three hydroxyl groups of the cyclohexane ring and the side chains of Asp197, Glu233, and Asp300. An additional hydrogen bond is observed between the para hydroxyl group of the caffeic acid moiety and the side chain of Glu240, whereas the phenyl ring engages Lys200 with a π-cation interaction ( Figure 4). Caffeic acid (4, exppKa = 4.6) is a structural fragment of CHA. The docking study of this compound into the crystal structure of human pancreatic α-amylase yields a binding mode wherein the carboxylic group engages the side chain of Lys200 with a salt bridge interaction, while making a hydrogen bond with the backbone amide group of Ile235 ( Figure 5A). The shift of the molecule within the active site of the enzyme with respect to the ester fragment of CHA is driven by the electrostatic interaction of the free carboxylic group with Lys200.
Differently from CHA, such a shift places the meta hydroxyl moiety in a complementary position for the formation of a hydrogen bond with Glu233. p-Coumaric acid (6, calcpKa = 3.8) lacks the meta hydroxyl moiety while keeping the para hydroxyl group of caffeic acid. As a result of the docking study, this compound adopts a binding mode akin to caffeic acid, keeping the interactions with Lys200 and Ile235 while losing the hydrogen bond with Glu233 ( Figure 5B). This binding mode is also found in the docking study of sinapic acid (7, calcpKa = 3.4) to human pancreatic α-amylase, wherein a further -stacking interaction is also observed between the substituted aromatic moiety and His201 ( Figure 5C).  Caffeic acid (4, exp pKa = 4.6) is a structural fragment of CHA. The docking study of this compound into the crystal structure of human pancreatic α-amylase yields a binding mode wherein the carboxylic group engages the side chain of Lys200 with a salt bridge interaction, while making a hydrogen bond with the backbone amide group of Ile235 ( Figure 5A). The shift of the molecule within the active site of the enzyme with respect to the ester fragment of CHA is driven by the electrostatic interaction of the free carboxylic group with Lys200. Caffeic acid (4, exppKa = 4.6) is a structural fragment of CHA. The docking study of this compound into the crystal structure of human pancreatic α-amylase yields a binding mode wherein the carboxylic group engages the side chain of Lys200 with a salt bridge interaction, while making a hydrogen bond with the backbone amide group of Ile235 ( Figure 5A). The shift of the molecule within the active site of the enzyme with respect to the ester fragment of CHA is driven by the electrostatic interaction of the free carboxylic group with Lys200.
Differently from CHA, such a shift places the meta hydroxyl moiety in a complementary position for the formation of a hydrogen bond with Glu233. p-Coumaric acid (6, calcpKa = 3.8) lacks the meta hydroxyl moiety while keeping the para hydroxyl group of caffeic acid. As a result of the docking study, this compound adopts a binding mode akin to caffeic acid, keeping the interactions with Lys200 and Ile235 while losing the hydrogen bond with Glu233 ( Figure 5B). This binding mode is also found in the docking study of sinapic acid (7, calcpKa = 3.4) to human pancreatic α-amylase, wherein a further -stacking interaction is also observed between the substituted aromatic moiety and His201 ( Figure 5C).  Differently from CHA, such a shift places the meta hydroxyl moiety in a complementary position for the formation of a hydrogen bond with Glu233. p-Coumaric acid (6, calc pKa = 3.8) lacks the meta hydroxyl moiety while keeping the para hydroxyl group of caffeic acid. As a result of the docking study, this compound adopts a binding mode akin to caffeic acid, keeping the interactions with Lys200 and Ile235 while losing the hydrogen bond with Glu233 ( Figure 5B). This binding mode is also found in the docking study of sinapic acid (7, calc pKa = 3.4) to human pancreatic α-amylase, wherein a further π-stacking interaction is also observed between the substituted aromatic moiety and His201 ( Figure 5C).
Since a fraction of vanillin (8, exp pKa = 7.4) exists in the anionic form at the experimental pH of the inhibition assay (pH = 6.9), this compound was docked using both the neutral form and negatively charged form. These forms adopt a binding mode that is markedly affected by the electrostatic interaction with Lys200 ( Figure 6). In the anionic form, the hydroxylate group of vanillin is placed nearby Lys200, forming a salt bridge interaction with this residue and a hydrogen bond with the backbone amide group of Ile235. A further hydrogen bond is observed between the methoxy group and the side chain of Lys200, whereas a π-stacking interaction is found between the imidazole ring of His201 and the aromatic moiety of vanillin. No specific interactions are observed for the aldehyde group of the anionic form of vanillin. In the neutral form, only the π-stacking interaction with His201 is kept, while the aldehyde group engages the backbone carbonyl group of Ile235 with a hydrogen bond. Since a fraction of vanillin (8, exppKa = 7.4) exists in the anionic form at the experimental pH of the inhibition assay (pH = 6.9), this compound was docked using both the neutral form and negatively charged form. These forms adopt a binding mode that is markedly affected by the electrostatic interaction with Lys200 ( Figure 6). In the anionic form, the hydroxylate group of vanillin is placed nearby Lys200, forming a salt bridge interaction with this residue and a hydrogen bond with the backbone amide group of Ile235. A further hydrogen bond is observed between the methoxy group and the side chain of Lys200, whereas a -stacking interaction is found between the imidazole ring of His201 and the aromatic moiety of vanillin. No specific interactions are observed for the aldehyde group of the anionic form of vanillin. In the neutral form, only the -stacking interaction with His201 is kept, while the aldehyde group engages the backbone carbonyl group of Ile235 with a hydrogen bond. Akin to the anionic form of vanillin, vanillic acid (5, exppKa = 4.5) adopts a binding mode featured by a conserved salt bridge interaction with Lys200 and a -stacking interaction with His201 ( Figure  7A). A further hydrogen bond is established between the para hydroxyl group of this compound and Glu233.  Akin to the anionic form of vanillin, vanillic acid (5, exp pKa = 4.5) adopts a binding mode featured by a conserved salt bridge interaction with Lys200 and a π-stacking interaction with His201 ( Figure 7A). A further hydrogen bond is established between the para hydroxyl group of this compound and Glu233.
Two additional analogues of vanillic acid are salicylic acid (3, exp pKa = 2.97) and syringic acid (1, calc pKa = 3.9). Docking studies of these compounds reveal conserved binding modes in which both compounds engage Lys200 and His201 with salt bridge and π-stacking interactions, respectively ( Figure 7B,C). A further hydrogen bond interaction is observed between the backbone amide group of Ile235 and carboxyl group of salicylic acid and syringic acid. At odds with vanillic acid, no hydrogen bond is made by the para hydroxyl group of syringic acid, which is due to the steric shield provided by the two methoxy groups in meta positions of the benzene ring. Overall, results of docking studies allow drawing relationships among chemical structure, pKa property, binding mode and inhibition activity of tested compounds. Specifically, the conserved salt bridge or π-cation interaction between the anionic carboxylate or aromatic ring and Lys200 is found important for the inhibition activity. Indeed, vanillin (8, IC 50 = 10.8 mg/mL) shows a significant drop of activity being in part present as neutral form at the experimental pH for its poor acidic constant ( exp pKa = 7.4). The higher inhibition activity of CHA (2, IC 50 = 0.5 mg/mL) with respect to caffeic acid (4, IC 50 = 3.5 mg/mL) suggests that the binding interaction of the quinic acid moiety of CHA is important for the inhibition potency, making the hydrogen bond interactions between its three hydroxyl groups and the acidic residues Asp197, Glu233, and Asp300. Likewise, the hydrogen bond interaction of the meta hydroxyl group of caffeic acid (4) with Glu233 favors the inhibition activity against pancreatic α-amylase. Indeed, removal or methylation of this group determines a decrease of the inhibition activity in p-coumaric acid (6, IC 50 = 5.6 mg/mL) and sinapic acid (7, IC 50 = 8.3 mg/mL), respectively. This trend is also observed in the shorter homologues of caffeic acid (4), namely vanillic acid (5, IC 50 = 6.2 mg/mL) and syringic acid (1, IC 50 = 6.0 mg/mL). Conversely, salicylic acid (3, IC 50 = 1.8 mg/mL) keeps a similar inhibition activity to caffeic acid, albeit its ortho hydroxyl group is not involved in any hydrogen bond interaction with the enzyme binding site. Hence, the better inhibition activity of salicylic acid (3, IC 50 = 1.8 mg/mL) than vanillic acid (5, IC 50 = 6.2 mg/mL) and syringic acid (1, IC 50 = 6.0 mg/mL) may likely be explained by the lowest acidic constant of this compound (3, exp pKa = 2.97) that favors a stronger electrostatic salt bridge interaction with Lys200. Figure 6. Binding mode of vanillin (8) to human pancreatic α-amylase according to its anionic (A; Gscore = −3.70 kcal/mol) and neutral (B; Gscore = −5.59 kcal/mol) form. Salt bridge interaction is depicted with a continuous blue/red faded line. Hydrogen bonds are shown with magenta arrows. -Stacking interaction is plotted with a green capped line. Polar and hydrophobic contacts between ligand and enzyme residues are displayed with blue and green faded tick lines, respectively.
Akin to the anionic form of vanillin, vanillic acid (5, exppKa = 4.5) adopts a binding mode featured by a conserved salt bridge interaction with Lys200 and a -stacking interaction with His201 ( Figure  7A). A further hydrogen bond is established between the para hydroxyl group of this compound and Glu233.

Conclusions
In the present work, for the first time several phenolic acids, identified in L. barbarum leaf extract, have been studied for their inhibitory activity against porcine pancreatic α-amylase by in vitro and in silico approaches. The results obtained by in vitro assay indicated that phenolic acids had interesting α-amylase inhibitory activity and showed that chlorogenic acid as the most active compound, followed by caffeic and salicylic acids with one-fold lower inhibition activity. The high sequence similarity between porcine and human isoforms of the enzyme (90%) suggests that such results may apply in both species. Results of docking studies identifies the interaction with Lys200 as important for the inhibition activity of tested extracts from L. barbarum leaves. Although further studies are needed to investigate the nutraceutical potential of these natural extracts for the management of hyperglycemia related disorders, findings of this study support the α-amylase inhibitory properties of phenolic acids, commonly represented in plant extracts.