Exploration of human pancreatic alpha-amylase inhibitors from Physalis peruviana for the treatment of type 2 diabetes

Abstract Type 2 Diabetes (T2D), a chronic metabolic disorder characterized by persistent hyperglycemia, accounts for ∼90% of all types of diabetes. Pancreatic α-amylase is a potential drug target for preventing postprandial hyperglycemia and inhibiting T2D in humans. Although many synthetic drugs have been identified against pancreatic α-amylase, however, reported several side effects, and plant-derived natural products are less explored against T2D. This study tested 34 flavonoids derived from the plant Physalis peruviana against the human pancreatic α-amylase (HPA) using in silico computational approaches such as molecular docking and molecular dynamics simulation approaches. Schrödinger, a drug discovery package with modules applicable for molecular docking, protein-ligand interaction analysis, molecular dynamics, post-dynamics simulation, and binding free energy calculation, was employed for all computational studies. Four flavonoids, namely, Chlorogenic acid, Withaperuvin F, Withaperuvin H, and Rutin, were picked based on their docking score ranging between −7.03 kcal/mol and −11.35 kcal/mol compared to the docking score −7.3 kcal/mol of reference ligand, i.e. Myricetin. The molecular dynamics analysis suggested that all flavonoids showed considerable stability within the protein’s catalytic pocket, except chlorogenic acid, which showed high deviation during the last 15 ns. However, the interactions observed in initial docking and extracted from the simulation trajectory involved > 90% identical residues, indicating the affinity and stability of the docked flavonoids with the protein. Therefore, all four compounds identified in this study are proposed as promising antidiabetic candidates and should be further considered for their in vitro and in vivo validation. Communicated by Ramaswamy H. Sarma


Introduction
Type 2 Diabetes (T2D) is a metabolic disease due to insulin deficiency caused by pancreatic b-cell dysfunction and insulin resistance in target organs (Chatterjee et al., 2017).Over the past few decades, obesity, physical inactivity, and highcalorie diets have been linked to a rise in type 2 diabetes patients worldwide.According to the International Diabetes Federation, the number of diabetic people is expected to increase from 425 million in 2017 to 629 million by 2045.Among them, the proportion of the T2D population is higher in most countries (Laakso, 2019) As of 2030, global spending on diabetes is projected to reach USD 490 billion, up from USD 376 billion in 2010 (Alam et al., 2021;Zhang et al., 2010).The delay in diabetes treatment may cause failure and dysfunction of multiple organs resulting into several other consequences such as cardiovascular and neurodegenerative diseases, retinopathy, nephropathy, and neuropathy (American Diabetes Association, 2014; Beseni et al., 2019).
The T2D can be prevented and managed by reducing the rise of dietary glucose levels, which can be done by inhibiting the digestion of dietary carbohydrates in the gut by inhibiting the digestive enzyme, i.e. a-amylase (Ogunyemi et al., 2023).The term 'cure' is not applicable to T2D yet and it has been well documented that it has potential for recurrence (Kalra et al., 2021).Several studies have targeted the a-amylase enzyme to identify synthetic (Alqahtani et al., 2019;el Khatabi et al., 2020;Imran et al., 2017;Rahim et al., 2020) and natural inhibitors (Blahova et al., 2021;Ogunyemi et al., 2020Ogunyemi et al., , 2021Ogunyemi et al., , 2023;;Yan et al., 2020) and therefore have been considered as a potential therapeutic target for preventing T2D.The enzyme a-amylase is released predominantly in saliva and pancreas (Stiefel & Keller, 1973) which hydrolyze glycosidic bonds present in complex polysaccharides to convert them into simpler sugars.The digestion of carbohydrates initiates in the mouth by the salivary a-amylase secreted by the salivary glands, followed by further digestion by pancreatic a-amylase secreted by pancreas in the duodenum (the first part of the small intestine).Consequently, decreasing postprandial hyperglycemia (hyperglycemia associated with diet) will be an effective therapeutic approach (Chen et al., 2006;Chiasson, 2006).Therefore, pancreatic a-amylase catalyzes the initial hydrolytic reactions converting complex polysaccharides (starch) into maltose which cleaved into glucose, is considered a key therapeutic target for treating T2D.A well-characterized active site region can be found on the surface of the 57 KDa pancreatic a-amylase protein (HPA), which consists of 512 amino acids (Whitcomb & Lowe, 2007).The pancreatic a-amylase, a potential antidiabetic target selected in this study, consists of domains A, B, and C, where domains A and B form a catalytic cleft forming a substrate binding site.The amino acid residues, Asp 197 and Asp 300 , are found to be critical catalytic residues in the catalytic site (Figure 1).Unlike domains A and B, domain C is unnecessary and has been unexplored for its function (Whitcomb & Lowe, 2007).In this study, the active catalytic site of the pancreatic a-amylase has been targeted for identifying potential antidiabetic drug molecules.The previously identified inhibitors of the pancreatic a-amylase active site have been reported with several side effects such as abdominal discomfort, diarrhea, and vomiting.Therefore, the roar of time is the identification of a safer and more potent drug for preventing and managing T2D.In the treatment of diabetes, plant-derived chemicals, particularly flavonoids, are preferred and are recommended because they are safe, highly effective, and generally cheaper than synthetic drugs.Flavonoids also reduce blood sugar and lipid levels (Al-Ishaq et al., 2019), are antioxidants, and are antidiabetic (Bai et al., 2019;Sarian et al., 2017) Therefore, a small library of flavonoids derived from the plant Physalis peruviana has been screened against HPA in this study to identify a potential drug candidate.Physalis peruviana attracted our attention due to its long medicinal history against various human diseases related to skin, eye, immune system, endocrine, and metabolism, which has been reviewed by Kasali et al. (2021).
Additionally, the plant species under the Physalis genus have been reported to have antitumor, anti-inflammatory, anti-antioxidant, hyperglycemic and antidiabetic properties.They are also rich in essential minerals and vitamins (Fatemeh et al., 2018;Shenstone et al., 2020;Singh et al., 2019).Recent studies have reported that Physalis peruviana-derived physapruin A (PHA) inhibited the proliferation of cells in breast cancer and induced oxidative stress mediated apoptosis (Yu et al., 2021), calyces extract reduced insulin resistance and oxidative stress in diabetic mice (Valderrama et al., 2022), magnolin showed antiproliferative potential against pancreatic cancer (Sayed et al., 2022), and P. peruviana pulp prevented liver inflammation and insulin resistance in skeletal muscles of obese mice (Pino-de la Fuente et al., 2020).In this study, the four flavonoid compounds (Chlorogenic acid, Withaperuvin F, Withaperuvin H, and Rutin) were identified from thirty-seven Physalis peruviana-derived flavonoids based on their docking score.The dynamic stability of best-docked poses containing the target protein (HPA) and the ligands was analyzed using molecular dynamics (MD) simulation.Following MD simulation, the binding free energy was also calculated using the molecular mechanic generalized Born surface area calculation (MMGBSA).Moreover, the flexibility and rigidity experienced by the target protein during MD simulation in the presence of docked ligands were also analyzed by principal component analysis (PCA).In conclusion, all four flavonoids showed significant binding affinity within the active catalytic site of the protein and exhibited high dynamic stability with the HPA protein.

Protein and ligand selection
Using PDB ID 4GQR, the crystal structure of human pancreatic a-amylase (HPA) was downloaded in PDB format from the protein data bank (PDB) database (Burley et al., 2017).The resolution of the three-dimensional HPA was 1.20 Å and predicted by the X-ray diffraction method (Williams et al., 2012).The downloaded HPA consists of a single chain with a length of 496 amino acids and is complexed with its native ligand Myricetin.The native ligand Myricetin interacts with the essential residues, Tyr 62 , Gln 63 , Asp 197 , Glu 233 , and Asp 300 , of the HPA catalytic site shown in Figure 1.
The compounds used in this study for molecular docking against the HPA catalytic site were flavonoids derived from the plant Physalis peruviana.A total of 34 flavonoid compounds were sourced from a plant database known as IMPAT (Indian Medicinal Plants, Phytochemistry and Therapeutics) (https://cb.imsc.res.in/imppat/)(Mohanraj et al., 2018) (Table 1).

Molecular docking and ADMET analysis
The initial experiment conducted in this study was the molecular docking of collected flavonoids against the HPA catalytic site, for which the target protein and the ligands were prepared using the Protein Preparation Wizard (PrepWizard) and LigPrep Module of Schr€ odinger suite, respectively (Madhavi Sastry et al., 2013).The protein was prepared by adding the missing residues and assigning the formal and partial charges to the structure.The native ligands and the co-crystallized water molecules were deleted from the protein structure to allow the docking of new ligands into the active catalytic pocket.A docking grid with centers at 13.33 (X), 15.11 (Y), and 39.56 (Z) regions were prepared to surround the catalytic site residues forming interactions with the native ligand Myricetin.Along the X-, Y-, and Z-axis, the inner and outer boxes with the size of 10 � 10 X 10 Å and 30 � 30 X 30 Å, respectively, were set by employing the grid generation tool of Schr€ odinger suite allowing the free movement of ligands docked with the catalytic site of the protein.The catalytic residues, Asp197 and Asp300, along with other active residues of the catalytic site interacting with Myricetin (Figure 1), were used for constructing the grid.The flavonoid's three-dimensional structure was downloaded from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/)(Kim et al., 2016).Using the default settings of the LigPrep module in the Schr€ odinger suite (Schr€ odinger, 2018), the ligands were prepared by generating their tautomeric confirmations with OPLS3e force field and EPIK state penalty at pH 7.0 ± 2.0.Following molecular docking, the drug-likeness analysis for all four selected compounds was done using an online server, i.e. swissADME (Daina et al., 2017).The canonical smiles for all the compounds were obtained from the PubChem database and inserted into the swissADME server to predict the medicinal properties.

Molecular dynamics (MD) and post-MD simulation
A 100 nanosecond (ns) Molecular Dynamics (MD) simulation program was run for analyzing the dynamic stability, and intermolecular interactions of the selected protein-ligand docked complex using the freely available academic Desmond provided by Schr€ odinger (Bowers et al., 2006;Schr€ odinger Release 2020-4: Desmond, 2020).Using the system builder tool, the simulation system was prepared by placing the docked complex in an orthorhombic box of size 10 Å � 10 Å � 10 Å.The TIP4P solvent and Na þ /Cl -counter ions were added for solvation and neutralization, respectively, of the complete MD simulation system, followed by the minimization under default settings.The neutralized and minimized system was then simulated at 310 K temperature and 1.01325 bar pressure under the OPLS-2005 force field with default settings in Desmond.To analyze the global fluctuations in the protein and ligand structures, the Root Mean Square Deviation (RMSD), and for local fluctuations at the residue/atom level, Root Mean Square Fluctuation (RMSF) values were extracted from the 100 ns MD simulation trajectory of each respective docked complex.Additionally, the protein-ligand interactions were extracted and analyzed with the help of the Simulation Interaction Diagram (SID) using the Desmond-Maestro 2018.4 module.

Principal component analysis
For Principal Component Analysis (PCA) (Jolliffe, 2011), the trajectory generated from the MD simulation was converted into Bio3D compatible format and implemented in the 'R' Program (Grant et al., 2006;Skjaerven et al., 2014).The initial coordinates of molecules were used as a reference, and other poses generated during the simulation were compared to find the eigenvector corresponding to the orthogonal Principal Components (PCs) of the molecule's motion.

End-point binding free energy calculation
The end-point binding free energy of HPA protein with bound flavonoid was estimated using the molecular Figure 1.The three-dimensional crystal structure of human pancreatic a-amylase complexed with its native ligand Myricetin.The interacting catalytic residues have also been displayed (Williams et al., 2012).6,14,17-trihydroxy-10,13-dimethyl-1-oxo-6,7,8,9,11,12,15,16octahydrocyclopenta[a]phenanthren-17-yl]ethyl]-4,5-dimethyl-2,3-dihydropyran-6-one mechanics-generalized Born surface area (MMGBSA) (Ylilauri & Pentik€ ainen, 2013).For this estimation, the last ten poses were extracted from the 100 ns MD simulation trajectory for each docked complex under the OPLS3e force field using default settings of Prime MM/GBSA modules in Schr€ odinger (Schr€ odinger, 2020).The MM/GBSA calculation is based on molecular mechanics (MM) force field, generalized Born model, and solvent accessibility method to obtain free energies (Sahakyan, 2021).The MM component corresponds to the system's internal energy consisting of bonded and nonbonded interactions between protein and ligand.The bonded interactions include dihedral, bond, and angle energy, whereas the non-bonded interactions include electrostatic and van der Waal interactions.The solvent molecules and ions were deleted from the extracted poses and binding free energy (DG) was calculated (Bajrai et al., 2022;Kumar et al., 2022) using the following equation.
where DG bind stands for binding free energy, DG complex stands for the free energy of the complex, and DG receptor and DG ligand stand for the free energy of receptor and ligands, respectively.

Molecular docking and interaction analysis
The docking of 34 flavonoids derived from the Physalis peruviana plant resulted in the identification of four topmost flavonoids based on their high negative docking score, namely, Chlorogenic acid (-8.3 kcal/mol), Withaperuvin F (-7.76 kcal/mol), Withaperuvin H (-7.03 kcal/mol), and Rutin (-11.35 kcal/mol) (Figure 2).Except for Withaperuvin H (-7.03 kcal/mol), all docked flavonoids showed a docking score more negative than the control ligand, i.e.Myricetin (-7.3 kcal/mol), indicating their higher affinity towards the protein binding site compared to control ligand.The structural details for all selected flavonoids are compared with each other, and with control ligand are provided in Table 1.Among all selected flavonoids, Rutin generated three poses with docking scores of À 11.35 kcal/mol, À 10.92 kcal/mol, and À 10.48 kcal/mol.Likewise, Chlorogenic acid generated two poses with docking scores of À 10.42 kcal/mol and À 8.37 kcal/mol (Table 2).Furthermore, Withaperuvin F and Withaperuvin H generated only a single pose with docking scores of À 7.76 kcal/mol and À 7.03 kcal/mol, respectively.Therefore, they were directly considered further for intermolecular interaction analysis.In the case of Rutin and Chlorogenic acid, only the best poses (with highly negative docking score) from each docked complex, viz., Rutin with À 11.35 kcal/mol and Chlorogenic acid with À 10.42 kcal/mol, were selected.Followed by the pose selection, the three-dimensional and two-dimensional (3D and 2D) interaction figures were prepared (Figure 3), followed by their intermolecular interaction analysis (Table 3).The 3D and 2D figures were generated using the Free Maestro v12.3 version's Graphic User Interface.
Intermolecular interaction (IMI) analysis is performed to observe molecular interactions between the target protein and docked flavonoids.This analysis revealed the formation of various non-covalent interactions such as hydrogen bonds (H-Bond), hydrophobic and hydrophilic (Polar) interactions, and positive and negative interactions, along with the glycine interactions between the protein and ligands in each docked complex (Table 3).The IMI analysis of the docked complex containing Chlorogenic acid revealed the formation of a total of three H-Bonds (Thr 163 , Glu 233 , Gln 63 ), seven hydrophobic interactions (Leu 165 , Ala 198 , Trp 58 Trp 59 , Tyr 62 , Val 107 , Leu 162 ), four polar interactions (Gln 63 , His 101 Thr 163 , His 299 ), one positive (Arg 195 ) and three negative interactions (Asp 197 , Glu 233 , Asp 300 ), and two glycine interactions (Gly 104 , Gly 164 ).In the case of the docked complex containing Withaperuvin F, the IMI analysis revealed the formation of two H-Bonds (Asp 197 , Glu 233 ), nine hydrophobic interactions (Ile 235 , Ala 307 , Trp 58 , Trp 59 , Tyr 62 , Tyr 151 , Leu 162 , Leu 165 , Ala 198 ), five polar interactions (Gln 63 , His 101 Thr 163 , His 201 , His 299 ), two positive (Arg 195 , Lys 200 ), three negative (Asp 197 , Glu 233 , Asp 300 ), and one glycine (Gly 306 ) interactions.The IMI analysis of docked complex containing Withaperuvin H revealed that it consisted of two H-Bonds (Tyr 151 , and Asp 300 ), eight hydrophobic interactions (all the residues present in Withaperuvin F docked complex, except Ala 198 ), four polar interactions (all the residues present in Withaperuvin F docked complex, except His 299 ), along with two positive (Lys 200 , and His 305 ) two negative (Asp 197 , and Asp 300 ) and two glycine (Gly 306 and Gly 308 ) interactions.Among all the flavonoids, Rutin was the one forming most hydrogen (four) and hydrophobic (ten) interactions along with additional p-p stacking and p -cation bonds, therefore carrying the highest negative docking score.The p-p stacking and p -cation bonds were formed with the same residue, i.e.His305 of the target protein.Rutin formed hydrophobic interactions with Trp 59 , Gln 63 , Glu 233 , and Asp 300 residues, and H-Bonds with Trp 58 Trp 59 , Tyr 62 , Tyr 151 , Leu 162 , Leu 165 , Ala 198 , Ile 235 , Ala 307 , and Trp 357 resides.In addition, the IMI analysis was also done for the control ligand, i.e.Myricetin and it revealed the formation of three H-Bonds (Asp 197 , Asp 300 , and Asp 356 ), six hydrophobic interactions (Trp 58 , Trp 59 , Tyr 62 , Leu 162 , Leu 165 , and Trp 357 ), along additional interactions such as two polar (Gln 63 , His 101 ), two positive (Arg 303 , His 305 ), three negative (Asp 197 , Arg 300 , Arg 356 ), and also one p-p stacking (Trp 59 ) bond.Concluding the overall IMI analysis results suggested that all identified flavonoids occupied the active catalytic pocket and interacted with similar amino acid residues of the target protein.However, three flavonoids (Chlorogenic acid, Withaperuvin F, and Rutin) showed the highest negative docking score (> À 7.76 kcal/mol) compared to the control ligand (-7.3 kcal/mol), indicating their higher affinity with the target protein.

ADMET analysis
The compounds proposed as potential drug candidates must possess high biological activity with a less toxic nature.Therefore, critical parameters such as absorption, distribution, metabolism, and excretion (ADME) are suggested to validate the proposed drug molecules.All four compounds, i.e. chlorogenic acid, Rutin, Withaperuvin F, Withaperuvin H were found to be the non-inhibitors of several cytochromes such as CYP2D6, CYP1A2, CYP2C19, CYP2C9, CYP2D6, and CYP3A4, which are essential for the metabolism of drugs and xenobiotics.Additionally, the selected compounds lacked the bloodbrain barrier (BBB) permeability.Except for Withaperuvin F, all three compounds were predicted to exhibit lower gastrointestinal absorption.Rutin showed three, whereas the other three compounds showed one violation of Lipinski's rule of five.All four compounds showed violations for other rules as well, such as Ghose, Veber, Egan, and Muegge.However, the druglikeness rules are not necessary to be followed by the natural compounds as the cells distinguish the bioactive compounds through active transport (Lipinski, 2004;Macarron, 2006).In addition, several other parameters related to drug-likeness were also predicted for all four compounds (Table S1).Moreover, the ADME analysis suggested the selected bioflavonoids against the Human Pancreatic Alpha-Amylase with ideal medicinal properties.

Molecular Dynamics simulation trajectory analysis
Molecular Dynamics (MD) simulation has been considered a paradigm for decades in the initial stage of drug  identification, which predicts the dynamicity of protein and ligand in protein-ligand docked complex in a defined simulation interval (Durrant & McCammon, 2011;Salo-Ahen et al., 2020).This study employed MD simulation on each of the selected docked complexes to understand the stability of protein and ligand with each other with respect to time.To understand the dynamic behavior of the docked complexes, several essential parameters were extracted from the respective 100 ns simulation trajectories, such as (i) the transitional modifications experienced by the docked complex during the simulation on (Figure 4), (ii) root mean square deviation (RMSD) values for protein and ligand (Figure 5), and (iii) root mean square fluctuation (RMSF) values for protein residues and ligand atoms (and Figure 6).The transitional modification during the 100 ns simulation run helped us understand the overall protein and ligand structural modifications by comparing the first (before simulation) and last docked poses.The RMSD analysis helped us understand the global protein and ligand structure fluctuation at the global level.The RMSF analysis helped us understand the local fluctuations occurring at protein residual and atomic ligand levels.All these analyses were also compared with the reference complex, docked with the native ligand, Myricetin (control).

Transitional modification during the simulation
The last pose from the respective 100 ns trajectory of each docked complex was drawn out to compare it with the first pose (resulting from initial docking) to observe the transitional modification in the ligand and protein structure with the effect of the simulation.Impressively, none of the docked phytochemicals vacated the protein pocket but was slightly dislocated from their initial binding site.The change in ligand orientation induced substantial conformational modifications in the protein's native structure, suggesting the potential of selected phytochemicals to affect the protein's actual function.Similar results were also observed in the case of the reference docked complex, i.e.Pancreatic a-amylase-Myricetin (Figure 4).

Protein and ligand RMSD analysis
Following the analysis of first-last pose transitional modifications for all the docked flavonoids, the structural deviation in protein and ligand structure was analyzed by observing their RMSD values (Y-axis) which were calculated for all succeeding frames (poses) compared to the initial pose as a  --reference frame with respect to time, in this case, 100 ns (Xaxis).The RMSD value under similar conditions was also calculated for the complex docked with reference ligand, i.e.Myricetin, and the resulting simulation trajectory was compared to the trajectory obtained from each of the complexes docked with selected flavonoids (Chlorogenic acid, Withaperuvin F, Withaperuvin H, and Rutin) (Figure 5).The target protein, i.e.HPA, in the case of all the complexes, showed the highest stability and steadiness compared to the reference docked complex, as the protein Ca atoms in each of the docked complexes showed the RMSD < 2 Å and exhibited trajectory exactly parallel to the X-axis (Figure 5a-d).However, the trajectory for the protein Ca atom from the reference docked complex showed minor deviation and did not maintain the parallel appearance as shown in the case of selected flavonoids but remained under the RMSD of < 2.5 Å (Figure 5e).
Likewise, the stability of the protein-fit-ligands (Chlorogenic acid, Withaperuvin F, Withaperuvin H, and Rutin) was also calculated.The Chlorogenic acid exhibited a highly stable trajectory with minor fluctuations with overall RMSD below 3 Å for more than 85% of the total simulation trajectory.During the last 15 ns, the ligand jumped up to RMSD of 11 Å but tended to come down at �6 Å. Considering Chlorogenic acid's extreme stability till 85 ns with RMSD < 3 Å, the structural deviation in the last 15 ns cannot be reasoned for its instability in the protein binding pocket.Thus, in this case, the longer simulation time (200 ns or 500 ns) might give a better idea about the stability of this ligand.Among all four flavonoids, Withaperuvin F exhibited the most stable trajectory with an overall RMSD value < 2 Å, indicating its extremely high stability within the binding pocket of the target protein.The flavonoid Withaperuvin H also showed an impressively stable trajectory which stabilized at the RMSD of 4 Å and continued until the end of the 100 ns simulation.Despite the highest docking score (-11.35kcal/mol), Rutin exhibited the most deviated trajectory compared to all three flavonoids.However, more than 90% of the trajectory remains in the considerable range of RMSD value, i.e. < 5 Å.Impressively, the trajectory for all the ligands was found highly stable compared to the trajectory of the reference ligand, i.e.Myricetin with RMSD > 5 Å.

Protein and ligand RMSF analysis
The RMSF was measured for the target protein and extracted from 100 ns MD simulation trajectories.Unlike RMSD, which tells the story of global structural minima, RMSF is plotted to measure the confirmational jump for each amino acid residue that causes the molecular motions in the protein structure.The RMSF value for the protein residues in all complexes was recorded with respect to time to be in the acceptable range where the N-terminal region showed high stability with RMSF value < 1 Å compared to the higher fluctuation at the C-terminal end with RMSF value > 1.5 Å ().Few protein residues around the residue index 350 showed higher peaks with RMSF > 3 Å, in the case of complex docked with the ligands, Chlorogenic acid (a) and Withaperuvin F (b), indicating the residues fluctuated most during the simulation, however, these peaks also observed in case of the reference docked complex.The blue region represents the alpha-helical region, whereas the red region represents the beta-strand of the protein.The green bars represent the residues interacting with the ligands, where Herein, the RMSD values for protein was computed in terms of Ca atoms, whereas for ligand, it was computed in terms of protein-fit-ligands for all the respective docked complexes from 100 ns MD simulation trajectories.
the interacting residues are observed between the residue index of 50 and 380.Except few, most interacting residues showed the highest stability with an RMSF value below 1.5 Å which might be responsible for the steadiness and stability of all the ligands as well, resulting in less structural deviation with RMSD < 4 Å for 90% of the trajectory in each docked complex (Figure 5).Hence, Hence, these acceptable RMSF values for protein structure and residue-ligand contact mapping suggested the structural stability of human pancreatic a-amylase complexes with flavonoids compared to native ligand, i.e.Myricetin of crystal structure during 100 ns simulation.
Similarly, the RMSF value was also calculated for the selected flavonoids fitted into the binding pocket of pancreatic a-amylase with respect to the simulation time of 100 ns (Figure 6).In all the docked complexes, the RMSF value for all ligand atoms was in the acceptable deviation range (< 3.5 Å), contributing to the lower RMSD for all the selected ligands during MD simulation.The lower RMSF for the protein residues and ligand atoms in all the docked complexes suggested the overall structural stability of protein and selected ligands compared to the reference ligand, i.e.Myricetin, during the 100 ns MD simulation.

Protein-Ligand contact profiling
The dynamic stability of protein and ligand in a docked complex can be understood by observing the interactions between the protein and ligand during the MD simulation.The protein-ligand contacts were extracted from the 100 ns MD simulation trajectory of each docked complex to understand the dynamic stability with respect to time.Notably, all flavonoids showed interaction with at least one of the catalytic residues, i.e.Asp 197 or Asp 300, during the simulation process, indicating their interaction stability and affinity within the catalytic site of HPA protein compared to reference complex HPA-Myricetin.Interestingly, catalytic residues and other interacting residues observed during simulation were also noted in the initially docked complexes (Table 3).
In the HPA-Chlorogenic acid docked complex, Trp 59 , Tyr 62 , His 101 , and Leu 165 formed hydrophobic interaction during the 100 ns simulation period, where Trp 59 and Leu 165 showed interaction for more than 50% of the total simulation time.The residue Tyr 62 showed hydrophobic interaction for only �20% of the total simulation.Four residues, Gln 63 , His 11 , Thr 163 , and Asp 197 , formed H-Bonds, in which Gln 63 and Asp 197 interacted for more than 50%, whereas Thr 163 interacted for only �20% of the total simulation time.His101 formed both H-Bond as well as hydrophobic interaction during the simulation.The ionic interaction was also observed for a short period formed by Asp 197 .Water bridges were also observed significantly during the simulation formed by Gln 63 , Ala 106 , Thr 163 , and Asp 197 (Figure 7a).Moreover, the residues showing interaction for more than 30% of the total simulation time were also extracted from the 100 ns MD simulation trajectory.Asp 197 and Gln 63 formed H-Bod, His 101 formed p-p stacking, Trp 59 and Leu 165 formed hydrophobic interactions, and Gln 63 and Thr 163 formed water bridges (Figure 8a).The catalytic residue (Asp 197 ) actively participated in interactions, indicating Chlorogenic acid's stability within the protein active site.Interestingly, all these residues forming interaction during the simulation were also observed in the initially docked complex (Table 3).
In HPA-Withaperuvin F docked complex, residues Trp 59 and Tyr 62 formed hydrophobic interaction for more than 50% of the total simulation time.One H-Bond was formed by the Gln 63 for �50%, whereas Asp197 formed two H-Bonds during the simulation.Additionally, several residues (Leu 162 , Thr 163 , Asp 197 , Lys 200 , Glu 233 ) were involved in forming water bridges (Figure 7b).The interaction extracted from 30% of the simulation trajectory showed residues, Gln 63 forming H-Bond for 51%, whereas Asp 197 forming two H-Bonds for 96% and 93% of the total simulation time.Trp 59 and Tyr 62 formed hydrophobic interactions, and residues Lys 200 and Glu 233 formed water bridges for 54% and 66% of the total simulation time (Figure 8b).The catalytic residue Asp 197 showed significant interaction with the ligand during the simulation, indicating the affinity and stability of Withaperuvin F within the active catalytic site of the HPA protein.
In HPA-Withaperuvin H docked complex, mainly hydrophobic interactions and H-Bonds were observed during the simulation.Tyr 62 , Leu 162 , Leu 165 , and Ile 235 formed hydrophobic interactions, where only Tyr 62 interacted for more than 45% of the total simulation time.Three residues formed, Glu 233 , His 299 , and Asp 300, formed H-Bonds for more than 30% of the total simulation time (Figure 7c).These residues were also observed in extracted interactions at 30% of the total 100 ns simulation interval (Figure 8c).One of the catalytic residues, Asp300, interacted with the ligand during the simulation, indicating the affinity and stability of Withaperuvin H within the catalytic pocket of the HPA protein.
During the simulation, the HPA-Rutin docked complex showed significantly higher interactions among all four docked complexes.Trp 58 , Trp 59 , Tyr 62 , Tyr 151 , Leu 162 , Leu 165 , and His 305 formed hydrophobic interactions, where Trp 58 , Trp 59 and His 305 interacted for more than 50% of the total simulation time.Gln63, Thr163, Asp300 and asp356 residues formed H-Bonds, where Gln 63 and Thr 163 interacted for more than 40% of the total simulation time.Residues Trp 59 and Tyr 151 formed both hydrophobic and H-Bonds during the simulation.Several residues (Thr163, His201, Asp300, and Asp356) also formed water bridges for a significant simulation time.During the simulation, the catalytic residue Asp300 interacted with the ligand via forming H-Bond and water bridge (Figure 7d).The catalytic residue Asp 300 formed H-Bond for a shorter period and formed a water bridge during the simulation.The interaction profiles extracted at 30% of the total simulation period showed that residues Thr 163 and Asp 356 formed H-Bond for 63% and 47% of the total simulation time.Trp 59 and His 305 formed p-p stacking for 51% and 45% of the total simulation period (Figure 8d).
Likewise, the reference docked complex HPA-Myricetin, was also analyzed for their interaction profiles to compare it with the selected docked complex.Here, the resides Trp 58 , Gln 63 , Arg 303 , His 305 , and Asp 356 formed H-Bonds, whereas residues Trp 59 , His 305 , and Trp 357 formed hydrophobic interactions.However, only the residues Trp 58 , Trp 59 , His 305 , and Asp 356 interacted for more than 40% of the total simulation period.All these residues observed during the simulation were also found in the initially docked complex (Figure 7e).Interestingly, the catalytic residues were present in the initially docked complex but were not observed in interactions during the simulation.The interaction profiles extracted at 30% of the total simulation period showed that residues Trp 59 and His 305 formed p-p stacking for 67% and 40%, whereas Trp 58 and Asp 356 formed H-Bonds for 72% and 49% of the total simulation period (Figure 8e).
In the case of all the docked complexes, the conservancy of these residues between the interactions observed from the initially docked complex and the interactions extracted from the trajectory of the respective docked complex indicates the stability of selected flavonoid ligands within the catalytic site of HPA.Additionally, all ligands also interacted with one of the catalytic residues (Asp197 and Asp300) of target HPA, suggesting the affinity of all ligands with the protein catalytic site.

Principal Component analysis
Principal Component Analysis (PCA) in drug discovery is generally used to analyze the modifications and confirmational changes occurring in proteins on binding with the ligand during the MD simulation (de Vivo et al., 2016).The PCA of the reference docked complex was also performed to compare it with the selected docked complex.During simulation, the protein exhibits high dimensional motions, which can only be analyzed by extracting the principal dimensions (primary essential components) to understand if the protein in the presence of a bound ligand shows flexibility or rigidity (Jolliffe & Cadima, 2016;Kitao, 2022).Therefore, in this study, three principal components (PCs), PC1, PC2, and PC3, were extracted from the simulation trajectory of each docked and analyzed in Figure 9.The sum of extracted PCs accounted for 34.99%, 32.82%, 30.91%, 27.84%, 40.38% of the total confirmational variation in protein structure docked with Chlorogenic acid, Withaperuvin F, Withaperuvin H, Rutin, and Myricetin (control Inhibitor), respectively.The top three PCs were plotted pairwise (PC1 vs PC2, PC2 vs PC3, and PC1 vs PC3) to represent the confirmational modifications in the protein structure with simulation progression (Lange & Grubm€ uller, 2006;Papaleo et al., 2009;van Aalten et al., 1995).The clustering graphs show the variation in the confirmational distributions, where each dot in a cluster stands for confirmation of the respective complex.The different colors represent the protein's conformational motions, and blue and red represent the degree of variation.In contrast, colors changing from blue to red via white represent the periodic fluctuations between these confirmations.The separation of blue and red clusters represents the flexibility of the protein during ligand binding, whereas the overlapping of colors indicates the rigidity of the protein (Bajrai et al., 2022).The analysis of clustering profiles obtained from all the respective docked complexes has shown the separation between blue and red color clusters, which corresponds to the flexibility of the protein, except for the PC2-PC3 cross profiling of chlorogenic acid (Figure 9a 2 ) and Rutin docked complex (Figure 9d 2 ).However, the other cross profiles (PC1-PC2 and PC1-PC3) in both cases (Figure 9a 1 , a 3 and d 1 , d 3 ) produced separated clusters for blue and red regions suggesting the flexibility of protein but slightly less compared to the other two docked complexes.The Scree plot with eigenvalue on the X-axis and proportion of variation on the Y-axis also supported the respective clustering analysis.The Eigen fraction is the variance percentage for the mean square positional variations as a function of 20 eigenmodes.Among Scree plots generated from all the docked complexes, the first three PCs did not show a steep decline in the case of the chlorogenic acid (Figure 9a 4 ), and Rutin docked complex (Figure 9d 4 ).The steeper the line produced by the first three PCs, the higher will the flexibility of the protein (Bharadwaj et al., 2021).Moreover, the higher distribution of clusters represents the higher flexibility and high confirmational changes in the protein structure and overlapping of clusters indicates the rigidity of the protein structure with less conformational changes.Therefore, the proteins in the complexes docked with Chlorogenic acid had slightly less flexible than other complexes resulting into the high deviation in simulation trajectory during the last 30 to 40 ns.In contrast other flavonoids showed stability during the entire period of simulation and formed higher number of contacts due to flexibility and confirmational changes in protein.

End-Point binding free energy calculation
Following the MD simulation and PCA, the binding affinity and binding free energy of the selected between the protein and selected flavonoid ligands from all the respective docked complexes were evaluated.The last ten poses were extracted from the 100 ns simulation trajectory of each docked complex, i.e.HPA-Chlorogenic acid, HPA-Withaperuvin F, HPA-Withaperuvin H, and HPA-Rutin at the regular interval of 10 ns to calculate the binding free energy against the binding free energy of reference docked complex, i.e.HPA-Myricetin (Figure 10).The binding free energy of Rutin was comparatively highly negative (-53.29 kcal/mol) than the control ligand (-49.67 kcal/mol), as well as other selected flavonoids (Table 4).Additionally, other dissociation energy parameters related to coulomb energy (DG Bind Coulomb ), covalent interactions (DG Bind Covalent ), H-Bonds (DG Bind H-Bond ), hydrophobic interactions (DG Bind Lipo ), Van der Waal forces (DG Bind vdw ) and ligand strain energy, were also calculated.Among all the dissociation energy components, DG Bind Coulomb , DG Bind Lip , and DG Bind vdw exhibited substantial support in the stability.In contrast, other parameters were not highly significant in terms of stability and instability of the respective docked complexes.In conclusion, all selected flavonoids showed significant binding affinity and dynamic stability by forming hydrophobic and van der Waal interactions with the catalytic residues of the human pancreatic a-amylase.

Conclusion
In silico molecular docking of 34 Physalis Peruviana derived Flavonoids resulted in the identification of the four best candidates with high negative docking scores, viz., Chlorogenic acid (-10.42 kcal/mol), Withaperuvin F (-7.76 kcal/mol), Withaperuvin H (-7.03 kcal/mol), and Rutin (-11.35 kcal/mol) compared to the reference ligand inhibitor, i.e.Myricetin (-7.3 kcal/mol).The docking and interaction analysis revealed that all selected flavonoids interacted with either one of the catalytic (Asp 197 and Asp 300 ) or other accessory residues of the protein's catalytic site.The most stable trajectory was exhibited by Withaperuvin F (< 2.0 Å RMSD), followed by chlorogenic acid (< 3.0 Å RMSD) for more than 80% of the trajectory, Withaperuvin H (< 5.0 Å RMSD), and Rutin (< 5.0 Å RMSD) with fluctuations throughout the trajectory.Finally, the end-point binding free of Rutin (-53.29 kcal/mol) was higher than the control docked complex (-49.67 kcal/mol), and for other docked complexes, it was in a considerable range.Based on docking, simulation and binding free energy analysis, all four flavonoids showed significant affinity and stability with the catalytic site of pancreatic a-amylase and therefore suggested as potential antidiabetic compounds.

Figure 2 .
Figure 2. 2D structure of selected flavonoids and native ligand.

Figure 5 .
Figure 5. RMSD plots for the HPA-flavonoids docked complex (a) Chlorogenic acid, (b) Withaperuvin F, (c) Withaperuvin H, (d) Rutin, (e) Native Ligand-Myricetin.Herein, the RMSD values for protein was computed in terms of Ca atoms, whereas for ligand, it was computed in terms of protein-fit-ligands for all the respective docked complexes from 100 ns MD simulation trajectories.

Table 2 .
Docking score and structural details of selected flavonoids along with their PubChem and IMPAT ID.

Table 3 .
Intermolecular interactions were observed for the selected flavonoids against the Pancreatic alpha-amylase within 4 Å around the docked flavonoid in the respective binding pocket.

Table 4 .
Calculated energy components and net binding free energies (kcal/mol) value for HPA-Flavonoid docked complex against the reference docked complex snapshots collected from the respective 100 ns MD simulation trajectories.