Barosmin against postprandial hyperglycemia: outputs from computational prediction to functional responses in vitro

Abstract Previously, barosmin has been demonstrated to possess anti-diabetic action. However, its effect to inhibit α-amylase and α-glucosidase, including glucose utilization efficacy, has yet to be revealed. Hence, the current study attempted to assess the efficiency of barosmin in inhibiting the α-amylase, α -glucosidase, and dipeptidyl peptidase 4 enzymes, including glucose uptake efficacy. Molecular docking and simulation were performed using AutoDock Vina and Gromacs respectively followed by gene ontology analysis using the database for annotation, visualization, and integrated discovery. Further, in vitro enzyme inhibitory activities and glucose uptake assay were performed in L6 cell lines. Density functional theory analysis detailed mechanistic insights into the crucial interaction sites of barosmin of which the electron-dense region was prone to nucleophilic attack (O-atoms) whereas hydroxyl groups (-OH) showed affinity for electrophilic attacks. Barosmin showed good binding affinity with α-amylase (-9.2 kcal/mol), α-glucosidase (-10.7 kcal/mol), and dipeptidyl peptidase 4 (-10.0 kcal/mol). Barosmin formed stable nonbonded contacts with active site residues of aforementioned enzymes throughout 200 ns molecular dynamics simulation. Further, it regulated pathway concerned with glucose homeostasis i.e. tumor necrosis factor signaling pathway. In addition, barosmin showed α-amylase (IC50= 95.77 ± 23.33 µg/mL), α-glucosidase (IC50= 68.13 ± 2.95 µg/mL), and dipeptidyl peptidase 4 (IC50= 13.27 ± 1.99 µg/mL) inhibitory activities including glucose uptake efficacy in L6 cell lines (EC50= 12.46 ± 0.90 µg/mL) in the presence of insulin. This study presents the efficacy of the barosmin to inhibit α-amylase and α-glucosidase and glucose uptake efficacy in L6 cell lines via the use of multiple system biology tools and in vitro techniques. Communicated by Ramaswamy H. Sarma


Introduction
Postprandial hyperglycemia is defined as high plasma glucose levels after ingesting and is caused by a range of factors such as meal timing and composition, carbohydrate content, insulin and glucagon output (Hiyoshi et al., 2017) and is characterized by hyperglycemic spikes.Uncontrolled hyperglycemia may trigger diabetes mellitus and a variety of consequences, including organ failure, as well as other causes, such as obesity, hypertension, and endothelial dysfunction, which can contribute to diabetic complications (Chaudhary et al., 2021;Dwivedi et al., 2021;Hiyoshi et al., 2017;Meza et al., 2019).
Two enzymes a-amylase and a-glucosidase are involved in the degradation of the polysaccharides into monosaccharides and trigger postprandial hyperglycemia (Khanal & Patil, 2020;Chaudhary et al., 2022) which is later absorbed via glucose transporters.Glucose is absorbed by the sodium-glucose cotransport mechanism after carbohydrate degradation into monosaccharides through absorptive cells (Gromova et al., 2021).In addition, glucagon-like peptides play an important role in minimizing postprandial hyperglycemic spikes (Hanssen et al., 2020).However, dipeptidyl peptidase-4 (DPP4) degrades glucagon-like peptides and also inhibits peripheral glucose utilization (Singh, 2014).Sitagliptin prevents DPP4 action (Makrilakis, 2019) to degrade the incretin hormones like glucagon-like peptide-1 (GLP-1).Further, sitagliptin was reported to improve glycemic control and decrease postprandial glucose levels in subjects with type 2 diabetes (Sakura et al., 2016).Metformin reduces blood glucose by enhancing insulin sensitivity, lowering intestinal glucose absorption, and decreasing hepatic glucose synthesis (Rena et al., 2017).Metformin also reduces postprandial glucose levels by up to 30% and has been proven to improve insulin sensitivity and reduce the risk of long-term complications such as cardiovascular disease and neuropathy (Aroda & Ratner, 2018).
Barosmin (also called diosmin, Figure 1), a flavone glycoside, is a natural bioactive present in multiple citrus fruits (Gerges et al., 2022) and has been reported for its potent anti-oxidant (Feldo et al., 2018) .However, its action on postprandial hyperglycemia is a bit limited.Therefore, the current study employed both dry-lab and wet-lab trials to assess the effectiveness of barosmin against the aforementioned enzymes as well as the potency of glucose utilization against postprandial hyperglycemia.

Computational evaluation of barosmin
In silico molecular docking was performed to assess the binding affinity of the barosmin with a-amylase, a-glucosidase, and DPP4 enzymes.The binding affinity of barosmin was compared with clinically approved, a-amylase/aglucosidase/DPP4 inhibitors i.e. acarbose and sitagliptin.This process involved 3 different steps which are detailed below and were performed as explained earlier (Khanal, Patil, Bhandare, Dwivedi, et al., 2022;Patil et al., 2022).After molecular docking, the complexes were subject to ligandprotein stability studies using aqAfdsagromacs.Further, gene ontology analysis was performed to identify the barosminmodulated pathways in response to glucose metabolism.

Density functional theory calculation of barosmin
In this study, density functional theory (DFT) based calculation was performed using Gaussian 09 (Frisch & Clemente, 2009) software and GaussView (Dennington et al., 2007) visualization program.DFT-based calculations (Feizi-Dehnayebi et al., 2021a, 2021b, 2021c, 2022a) were performed to obtain the minimized geometry and the minimized ground state structure of barosmin is obtained by employing Becke's three-parameter Lee-Yang-Parr hybrid (Lee et al., 1988) functional and the 6-31 þ G (d, p) basis set (B3LYP/6-31 þ G (d, p)).The quantum chemical reactivity descriptor such as molecular electrostatic potential (MEP) map was computed for barosmin to identify potential electrophilic and nucleophilic sites in the compound.MEP is the visual representation of relative reactivity at different sites of molecules on their electron density (Feizi-Dehnayebi et al., 2022b).
Ligand-protein docking: Both compounds were docked to each target using AutoDock Vina (Samdani & Vetrivel, 2018) within the grid size of the center (x¼ À 8.  (x ¼ 36.30, y ¼ 50.27, z ¼ 37.16) and size (x ¼ 26.33, y ¼ 30.42, z ¼ 27.37) for DPP4.The system exhaustiveness was set to 8, docking was executed and generated 9 conformations.Further, the ligand with the least binding energy and rootmean-square deviation (RMSD) was considered to visualize the interaction in Discovery Studio 2019.

Molecular dynamic simulation
A molecular dynamic simulation was performed using Gromacs ver.2022.1 (Van Der Spoel et al., 2005; https://www.gromacs.org/).Charmm36-jul2021 forcefield was used to generate the topology of the protein via the pdb2gmx module of gromacs.The proteins were solvated in a three-point water model in a dodecahedron box with a size of 1 nm 3 .The ligand topology was created using the official CHARMM general force field server (Srinivasa et al., 2022; https:// cgenff.umaryland.edu/).The charge in the model system was neutralized by appending sodium and chloride ions as per requirement.Energy minimization was performed through the steepest descent integrator with a verlet cutoff scheme of a maximum of 50,000 steps till the protein achieved the least energy confirmation.Further, the system was equilibrated using canonical (NVT) and isobaric (NPT) for 100 ps.V-rescale thermostat was applied to maintain standard temperature and pressure at 300 K. Similarly, the C-rescale algorithm was implemented to maintain constant pressure at 1 bar.Each complex was subjected to molecular dynamics (MD) run for 200 ns; the coordinates and energies were saved at every 20 ps.The trajectories generated were analyzed using built-in gromacs utilities.Further, analysis of results was performed using the built-in module of gromacs and other external free tools as per requirement.

Molecular mechanics Poisson-Boltzmann surface area analysis
The Molecular Mechanics Poisson-Boltzmann Surface Area (MMPBSA) analysis was performed using the gmx_MMPBSA module (Kumari et al., 2014) to analyze different energy contribution parameters like Vander Waals and electrostatic molecular mechanics energy, total solvation energy, and total relative binding energy.The MMPBSA run was performed for 100 frames from a total of 10,000 frames with an interval of 100.The Poisson Boltzmann calculations were performed using an internal PBSA solver in a sander.The MMPBSA_ana module and other external tools were used to visualize the results obtained from the gmx_MMPBSA run (Bhandare et al., 2019;Bhandare & Ramaswamy, 2018;Charla et al., 2023;Kumari et al., 2014).

Principal component analysis
The principal component analysis (PCA) is a statistical method widely used effectively in biomolecular simulations to extract significant collective motions.It aids to explore the protein folding/unfolding associated with domain motions.Principal Components (PCs) are obtained by orthogonal transformation of the coordinates of protein conformation (Hess, 2000).In the present study, the PCA was performed by generating the covariance matrix followed by the diagonalization of the matrix using the Gromacs embedded tools such as 'g_covar' and 'g_anaeig' to explore the principal modes governing the dynamics of the biomolecules.

Dynamic cross-correlation matrix analysis
The dynamic cross-correlation matrix (DCCM) quantifies the degree of correlation between pairs of atoms in terms of their motions, either positive or negative by evaluating the magnitude of all pairwise cross-correlation coefficients.We focused on analyzing each element of the DCCM, denoted by C ij .A value of C ij equal to 1 indicates that the atoms i and j are positively correlated, meaning they undergo similar motions with the same period and phase.A value of C ij equal to 0 indicates no correlation between the motions of the atoms i and j.Conversely, a value of C ij equal to À 1 indicates that the fluctuations of i and j are negatively correlated, meaning they undergo opposing motions with the same period and phase (Bhandare & Ramaswamy, 2018;Khanal, Patil, Bhandare, Patil, et al., 2022).

Gene set enrichment analysis of barosmin-regulated targets
Initially, the barosmin-regulated targets were retrieved from the DIGEP-Pred (Lagunin et al., 2013; http://www.way2drug. com/ge/) by querying the SMILES of barosmin at pharmacological activity greater than pharmacological inactivity.The proteins involved in diabetes (ID: C0011860) were retrieved from the DisGeNET (Piñero et al., 2017; https://www.disgenet.org/search) database.The barosmin-regulated proteins involved in diabetes were identified and enriched in DAVID functional annotation bioinformatics microarray analysis (Huang et al., 2007; https://david.neiferfgov/) which were evaluated using Bonferroni and Benjamin tests to trace the Kyoto Encyclopedia of Genes and Genomes pathways and 3 gene ontology terms to identify affected cellular components, and various molecular function, and biological processes.

Experimental pharmacology
After retrieving the computational data, an approach was made to validate the findings using suitable experimental methods.Herein, we utilized in vitro a-amylase, a-glucosidase, and DPP4 inhibitory activities and glucose uptake assay.Later we performed the in vitro cytotoxicity assays in L6 cell lines to assess the potentiality of barosmin covering the wide range of inhibitory concentration (IC 50 ) of hereby experiments. .

a-amylase and a-glucosidase inhibitory activity of barosmin
a-amylase and a-glucosidase inhibitory activity of barosmin (Highmedia RM8293-100G) was performed in 6 different concentrations (10 to 320 mg/mL in geometric series) using starch and p-NPG as a substrate respectively (Khanal & Patil, 2020).Controls were set in the absence of inhibitor and acarbose as negative and positive respectively.Experiments were performed on 3 replicates.Later, enzyme inhibitory activity (%) was calculated as (1À (As/Ac)) �100, where, As is the absorbance in the presence of the test and Ac is the absorbance of control.

In vitro DPP4 inhibitory assay
DPP4 inhibitory activity of barosmin was performed using chromogenic substrate Gly-Prop-nitroanilide as a substrate.Briefly, the 5 different concentrations of barosmin (6.25 to 100 mg/mL in geometric series) and sitagliptin were prepared in dimethyl sulfoxide and diluted with tris buffer (pH 8.0, 50 mM).The solutions were incubated with DPP4 (0.05 U/mL).The hydrolysis of the substrate was recorded at 405 nm (Singh et al., 2020).Later, enzyme inhibitory activity (%) was calculated as (1À (As/Ac)) �100, where, As is the absorbance in the presence of the test and Ac is the absorbance of control.

In vitro glucose uptake assay in L6 cell lines
A glucose uptake assay was performed in L6 cell lines (National Centre for Cell Science, Pune India).Initially, the cells were grown in 6 well plates and incubated (37 � C, 48h) in a CO 2 incubator.After the formation of confluent monolayer, the culture was renewed (Dulbecco's Modified Eagle Medium consisting of 0.2% bovine serum albumin) and again incubated (37 � C, 18 h).After incubation, the media was discarded and washed with Krebs ringer phosphate buffer.The cells were then treated with 6 different concentrations (3.125 to 100 mg/mL in geometric series) of barosmin and metformin in the presence of insulin followed by the addition of glucose (1 M) and incubated (30 min).The remaining amount of glucose was quantified from supernatant (glucose oxidase and peroxidase method, Erba Glucose Kit) and the percentage of glucose uptake was calculated as the difference between the initial and final glucose content in the incubated medium (Gupta et al., 2009;Yap et al., 2007).

Statistical analysis
The half-maximal IC 50 and half-maximal effective concentration (EC 50 ) were calculated using the linear regression curves using GraphPad Prism (https://www.graphpad.com/)ver 5.0.All the data were presented in mean ± SD.Similarly, docking data were analyzed using binding energy, the number of hydrogen bond interactions, and the residues involved in it.Likewise, the Kyoto Encyclopedia of Genes and Genomes (KEGG) data were analyzed using the false discovery rate (FDR), gene count, and strength of KEGG terms.

Density functional theory calculation
The DFT-based calculation has been employed to obtain the minimized geometry of barosmin using Gaussian09 software (Figure 1a).In addition, the MEP surface map for barosmin was computed (Figure 1b).MEP is of utmost importance in terms of intermolecular interactions as well as to predict the reactive centers (sites for electrophilic and nucleophilic attacks) in the molecule.The electrostatic potential in MEP surface plots can be understood with the assistance of different colors with the sequential increase from red, orange, yellow, green, and blue color.The red region indicates the most negative potential (electron-rich centers) prone to nucleophilic attacks, the blue region indicates the most positive potential (electron-deficient centers) prone to electrophilic attacks, and the green color indicates the zero potential.As demonstrated (Figure 1b), the red region is located over O5, O7, and O8 from the group 1 and O12, O13, and, O14 from the group 2 (groups 1 and 2 are shown in dark blue) indicating more electron density and an electronegative region prone to nucleophilic attack.The sites with more positive potential indicated with blue color are located at H59, H61, H62, H63, H64, and, H72 from hydroxyl groups which show affinity for electrophilic attacks.

Molecular docking reveals binding affinity of
barosmin with a-amylase, a-glucosidase, and DPP4 Barosmin possessed a higher binding affinity (binding energy ¼ À 9.2 kcal/mol, 5 hydrogen bonds, Gln63  1).Similarly, the twodimensional interaction of each ligand with respected targets along with hydrogen bonds and hydrophobic interactions are presented in supplementary file 1.

Barosmin/acarbosea-glucosidase complex
The RMSD for the barosmin-a-glucosidase complex and backbone displayed minor fluctuations of �1 Å for the first 80 ns where it reached the highest value at �90 ns.Thereafter the RMSD of both the complex and backbone were stable with a difference of �0.5 Å.However, the RMSD for the acarbose-a-glucosidase complex was observed to be unstable throughout the MD run.The root-mean-square fluctuation (RMSF) analysis displayed fluctuation in the range of �0.25 to � 1.00 nm.The residue Ile1814 possessed the highest RMSF value of 0.90 nm for the barosmin-a-glucosidase complex whereas Ser1366 possessed the highest RMSF value of 1.10 nm for the acarbose-a-glucosidase complex.The radius of gyration (RoG) displays minor fluctuations (�0.2 Å) for the barosmina-glucosidase complex; the RoG was unstable for the first �25 ns and thereafter remained stable up to 100 ns.Further, a slight decrease in the RoG was observed and continued to be stable throughout the run whereas, the RoG for the acarbose-a-glucosidase complex was in the range of �2.9 nm to �3 nm and was stable after 100 ns of MD run.Solvent-accessible surface (SASA) area displayed fluctuations in the range of �340nm 3 to 360 nm 3 for the a-glucosidase -barosmin complex whereas there was an increase in SASA for the a-glucosidase -acarbose complex.A maximum of 7 hydrogen bonds were formed in a barosmin-a-glucosidase complex in which a constant 3 hydrogen bonds were observed till �100 ns of MD run.Thereafter there was a decrease in the number of hydrogen bonds to 1, till the next �30 ns.Further, thereafter there was a minimum of 2 hydrogen bonds throughout the simulation (Figure 3).

Barosmin/acarbose-a-amylase complex
The RMSD of barosmin-a-amylase complex displayed stable RMSD throughout the MD run with fluctuations within �1.5Å for the complex and the backbone whereas, the RMSD for acarbosea-amylase complex displayed high fluctuations in the range of �6 Å after �75 ns of MD run.The residue Asn350 displayed the highest RMS fluctuation of 0.83 nm and residues Asn152 and Tyr151 displayed an RMSF of 0.77 nm and 0.76 nm respectively.Similarly, the RoG for the barosmina-amylase complex possessed a fluctuation of less than 1 Å which is less than the fluctuation displayed by the acarbose-a-amylase complex (>1 Å).SASA for the barosmin complex was in the range of 190 to 220 nm 3 and in the acarbose-a-amylase complex, the SASA was slightly more (�5 nm) than that of barosmin.A maximum of 8   hydrogen bonds were formed between a-amylase and barosmin at the start of the MD run for the first �20 ns forming 4-8 hydrogen bonds.However, the number of hydrogen bonds decreased thereafter and 2-3 hydrogen bonds were visible for the remaining MD run (Figure 4).

Barosmin/sitagliptin-DPP4 complex
The backbone RMSD for both barosmin-and sitagliptin-DDP4 complexes displayed the least fluctuations of �2 Å and �1.55 Å, respectively throughout the 150 ns MD simulation.Similarly, complex RMSD was also found to be stable throughout the 150 ns production run (�2.5 Å for both complexes).The RMSF analysis was performed to check the residue-wise fluctuation.The residues Ile236 and Val254 displayed large fluctuation during simulation (�5 Å) whereas, the residues involved in ligand binding showed the least fluctuation (�1.4 Å) for the complex.For the sitagliptin-DDP4 complex, the RoG value was found to decrease from 27.4 Å to 27 Å at �120 ns indicating the closure of the binding pocket and formation of a stable complex.This was also evidenced by a decrease in the SASA at �120 ns whereas no fluctuation was observed in the barosmin-DPP4 complex.The RoG and SASA were found to be uniform throughout the 150 ns production run.A maximum of 6 hydrogen bonds were displayed by the sitagliptin-DDP4 complex; a constant number of 3 hydrogen bonds were visible till �150 ns of MD run.Similarly, a maximum of 10 hydrogen bonds was displayed by the barosmin-DDP4 complex and a constant 4 hydrogen bonds were formed throughout the simulation (Figure 5).

MMPBSA analysis of a-glucosidase complex analysis for barosmin and acarbose
MMPBSA analysis for a-glucosidase-barosmin and a-glucosidase-acarbose complex displayed the total relative binding energy of 12.10 and 4.87 kcal/mol for barosmin and acarbose  respectively (Table 2).The total energy decomposition analysis revealed Tyr1251, Ile1280, Trp1355, Arg1510, and His1584 to contribute majorly in favor of simulation whereas, Asp1317, 1420, 1526, and 1555 were majorly against the simulation.In addition, Tyr1355 displayed the least À 2.48 kcal/mol energy contributions, however, Asp1420 possessed the highest energy contribution (2.18 kcal/mol).The total energy contribution by barosmin with a-glucosidase was À 9.39 kcal/mol (Table 2).The contribution energy plot highlighting the importance of the binding pocket residues in stable complex formation is presented in Figure 6.

MMPBSA analysis of a-amylase complex for barosmin and acarbose
MMPBSA analysis for a-amylase-barosmin and acarbose complex displayed total relative binding energy of 7.70 and 4.18 kcal/mol for the complexes with barosmin and acarbose respectively (Table 2).The total energy decomposition analysis revealed the residues Pro54, Trp59, 357, Val107, and Leu165 to be majorly in favor of the simulation, whereas, Asp300 and 356 were not in favor of the simulation.Typ59 possessed the least energy contribution (-2.81 kcal/mol) and Asp300 possessed the highest energy contribution (1.2 kcal/mol).The total energy contribution by barosmin with a-amylase was À 7.35 kcal/mol.

MMPBSA analysis of DPP4 complex for barosmin and sitagliptin
MMPBSA analysis for barosmin and sitagliptin DPP4 complex possessed total free binding energy of À 23.36 and À 111.30 kcal/mol (Table 2).For the barosmin-DPP4 complex, the total energy decomposition analysis revealed the amino acid Ile629 (-12.66 kcal/mol) to contribute significantly to ligand interaction during simulation whereas, Asp545 and Lys554 were not in favor of the simulation.In the sitagliptin-DPP4 complex, Asp326, Glu464, Asp515, Glu608, Ile629, Glu660, and Trp59 possessed favorable interaction during simulation whereas, Arg140, 429, and 596 were not in favor of the simulation.

PCA of the ligand-protein complex to evaluate the collective motion
Additional analysis was performed over the obtained trajectory to gain detailed insights into the dynamics expressed by various complexes.PCA helps to understand the evolution of various conformations during simulation while DCCM provides insights into the overall collective motion.The complexes a-glucosidase-barosmin and DPP4-barosmin showed the evolution of clusters in the conformational spaces that range from À 7 to 6 and À 5 to 5 compared to acarbose with a-glucosidase (-10 to 6) and sitagliptin with DPP4 (-4 to 3).a-amylase-barosmin complex showed greater diversity and conformational  spaces range from À 4 to 5 and a-amylase-acarbose showed À 4 to 3 (Figure 7 a-c).The eigenvalues of the a-glucosidasebarsomin complex were 13 and 17 for a-glucosidase-acarbose representing acarbose with larger diversity in its structure and undergoing significant conformational changes at the secondary structural level.Similarly, the eigenvalue of a-amylase-barosmin and DPP4-barosmin complex was found to be 4.5 and 4 respectively and for acarbose and sitagliptin complexed with a-amylase and DPP4 was 3 and 2.5 respectively representing the barosmin to possess greater structural diversity and significant conformational changes in the secondary structure of a-amylase and DPP4 (Figure 7 d and e).

Dynamic cross-correlation matrix of the ligandprotein complex
The maximum region in the complex of barosmin and acarbose with a-glucosidase showed a positive correlation mainly at the binding pocket region.This indicates that the observed closure movement at the binding pocket region plays a crucial role in promoting stable complex formation.In the barosmin and acarbose with a-amylase complexes, acarbose showed a higher positive correlation from residues 50 to 350 compared to barosmin.Likewise, in barosmin and sitagliptin complex with DPP4, barosmin showed a positive correlation primarily at the residues of 70 to 240 and 340 to 540 region which was observed weaker in sitagliptin (Figure 8).

In vitro a-amylase inhibitory activity of barosmin
There was concentration-dependent inhibition of the a-amylase by barosmin ranging from the log concentration from (1.0-2.5)mg/mL including acarbose.The IC 50 for barosmin and acarbose were (95.77 ± 23.33) and (58.36 ± 3.30) mg/mL respectively.The IC 50 of acarbose was significantly lower (p < 0.05) than barosmin, indicating that it was 1.64 times more potent than the test drug (Figure 11).

In vitro a-glucosidase inhibitory activity of barosmin
Similar to the a-amylase inhibition, there was concentrationdependent a-glucosidase inhibition by barosmin ranging from the log concentration (1.0-2.5)mg/mL including      Experiments were performed in triplicates.The above data are expressed in mean ± SD.The IC 50 of barosmin and sitagliptin was found to be (13.27± 1.99) and (9.62 ± 0.20) mg/mL respectively.� p < 0.01 compared to sitagliptin.Data were analyzed using a one-tailed unpaired t-test.The EC 50 of barosmin and metformin was found to be (12.46 ± 0.90) and (7.14 ± 0.34) mg/mL respectively.� p < 0.001 compared to barosmin.Data were analyzed using a one-tailed unpaired t-test.

In vitro DPP4 inhibitory activity of barosmin
There was concentration (log concentration 0.8-2 mg/mL)dependent DPP4 inhibition by both barosmin and sitagliptin.

In vitro glucose uptake assay in L6 cell lines
There was a concentration-dependent (log concentration (0.50-2.00) mg/mL) increase in glucose uptake with barosmin incubation in L6 cells in the presence of insulin.The EC 50 of the barosmin and metformin was found to be (12.46 ± 0.90) and (7.14 ± 0.34) mg/mL respectively for glucose uptake efficacy.The EC 50 of the metformin was significantly lower (p < 0.001) compared to barosmin proving it more potent (1.74 times) than barosmin (Figure 14).

In vitro cytotoxicity assay of barosmin in L6 cell lines
Barosmin possessed concentration-dependent cytotoxicity in L6 cell lines.The IC 50 of the barosmin and TBHT was found to be (275.4± 47.06) and (44.57± 6.472) mg/mL respectively.The IC 50 of the TBHT was significantly low (p < 0.05) compared to barosmin proving to more cytotoxic (5.85 times) than barosmin (Figure 15).

Discussion
The effectiveness of barosmin against postprandial hyperglycemia was assessed in the current investigation.In this regard, the higher-end approach was implemented to assess the a-amylase, a-glucosidase, and DPP4 inhibitory activity performed using both computational and pharmacological approaches.In addition, the effectiveness of barosmin in promoting glucose uptake was also investigated.Additionally, the cytotoxicity profile of barosmin concentrations used to inhibit enzymes and increase glucose utilization was assessed.It was discovered that these concentrations were less cytotoxic than others since the IC 50 of barosmin in the MTT experiment was substantially greater compared to enzymes inhibitory and glucose uptake.Moreover, this study provided the idea relating to the management of postprandial hyperglycemia which may be achieved by targeting multiple proteins that are directly concerned with glucose homeostasis.
In this study, initially, we targeted the a-amylase and a-glucosidase enzymes that are concerned with hydrolyzing the polysaccharides (Khanal et al., 2020).Polysaccharides are hydrolyzed or degraded into monosaccharides and get absorbed into the systemic circulation resulting in hyperglycaemic spikes.Hence, inhibition of a-amylase and a-glucosidase checks the conversion of polysaccharides to monosaccharides.To evaluate the efficacy of barosmin on these enzymes, initially, we performed molecular docking followed by molecular dynamics simulation including related analysis of ligand-protein complex.Herein, the binding affinity of the barosmin was greater towards a-glucosidase compared to acarbose which was further supported by in vitro enzyme inhibition.Additionally, the study also evaluated the DPP4 inhibitory activity of barosmin.Through incretin release, DPP4 inhibitors aid in improving b-cell function, physiology, and mass.As a result, after eating a meal, incretin triggers a continuous release of insulin to lower blood sugar levels, which is a sign of increased b-cell activity (Singh et al., 2021).This effect enhances the insulin signaling pathway and also stimulates peripheral glucose utilization.
In-silico molecular docking predicts the binding affinity of a ligand with a macromolecule.In the present study, we observed barosmin to have a better binding affinity (-9.2 kcal/mol) with a-amylase compared to acarbose (-8.1 kcal/mol) predicting it to be more potent than acarbose.However, in the in-vitro study, we observed acarbose (IC 50 ¼ 58.36 mg/mL) to be more potent than barosmin (IC 50 ¼ 95.77 mg/mL), and the reason for this opposite finding needs to be further investigated.Similarly, there were inverse results of binding energy and inhibitory constant of both molecules towards DPP4.In addition, we observed barosmin to possess better binding affinity (-10.7 kcal/mol) compared to acarbose (-8.1 kcal/mol) with a-glucosidase suggesting barosmin to be more potent than acarbose.This prediction was further supported by a significantly lower (p < 0.01) IC 50 of barosmin (68.13 mg/mL) than acarbose (80.28 mg/mL) indicating to be more potent (1.17 times) than acarbose.
Further, simulation studies revealed complexes with barosmin to possess greater stability and interaction with respective proteins (a-glucosidase and a-amylase).At the end of the simulation study, the parameters provided include RMSD, the RoG, SASA, the number of hydrogen bonds, and RMSF.A simulation was performed for both complexes i.e. test (barosmin) and standard (acarbose and sitagliptin) for 200 ns to understand binding empathy at time-bound stability among the complexes.It is perceptible from the RMSD plot that the barosmin-a-glucosidase complex possessed greater stability (fluctuation< 1Å) as compared to the acarbose (fluctuation > 4Å) representing structural free form.The residue-wise fluctuation can be evident from the RMSF plot that similar residues took part in the interaction with barosmin and acarbose.The RoG was calculated based on the varied masses calculated to the root mean square distance from the center of the complex.The RoG plot for the barosmin complex displayed less than 0.5 Å deviations throughout the MD run.This represents the folding, shape, and capability during each frame of the simulation.The solvent assessable surface area displays fluctuation between 340 to 360 nm 3 .Fluctuations in the SASA may be due to the hydrogen bonds which were forming and deforming throughout the MD run.However, the barosmin complex possessed greater stability compared to acarbose.The number of hydrogen bonds was constant for a time interval of 95 ns with a maximum number of 7 bonds; throughout the simulation, bonds were broken and formed.We also analyzed the MMPBSA and the total energy contribution of each residue, Trp1355 possessed the least energy contribution indicating the residue to be in favor of the interaction.
Similarly, from the RMSD plot, it was displayed that the barosmin-a-amylase complex possessed greater stability (fluctuation < 1.5Å) compared to the acarbose (fluctuation > 4 Å) representing structural free form.The residue-wise fluctuation can be evident from the RMSF plot in which similar residues took part in the interaction with ligands barosmin and acarbose, Asn350 displayed maximum fluctuation for the barosmin complex.The RoG plot for the barosmin complex presented a deviation of less than 1 Å throughout the simulation.This represents the folding, shape, and capability during each frame of the complex.The solvent assessable surface area displayed fluctuation between 190 to 220 nm 3 .However, the barosmin complex possessed greater stability compared to the acarbose.The number of hydrogen bonds was formed and deformed throughout the simulation with a maximum number of 8 bonds.The MMPBSA and total energy contribution of each residue identified Trp59 to possess the least total energy contribution indicating its favorability toward the interaction.
In addition, previously it was predicted that enzyme inhibitors acting in the gastrointestinal tract may get absorbed into the systemic circulation and display peripheral effects (Khanal & Patil, 2019).Over this speculation, we initially predicted the probability of barosmin for human intestinal absorption (Shen et al., 2010) and Caco-2 permeability (Pham The et al., 2011).Herein we observed 0.6344 and 0.8957 probability for human intestinal absorption and Caco-2 permeability respectively.Therefore, we further evaluated the efficacy of barosmin for glucose utilization in L6 cell lines.Since DPP4 metabolites (Deacon, 2018) glucagon-like peptides (GLPs), during postprandial hyperglycaemic spikes, we investigated the binding affinity of barosmin with DPP4 and we observed its binding affinity by forming the stable complex.However, the in vitro effectiveness was comparatively lower than that of sitagliptin.This controversial result could be due to the different conformation changes and formation of hydrogen bond interaction with different active site residues.However, to validate the stability of sitagliptin and barosmin with DPP4, we performed 150 ns of MD simulation.The RMSD of sitagliptin in complex with DPP4 was lower compared to barosmin i.e. �2.6 and 2.8 Å.Also, the RoG and SASA plot for the sitagliptin complex was decreased after 110 ns which indicates higher compactness and stable complex formation.From the MD results, it is evidenced that sitagliptin may bind stably to DPP4 compared to barosmin.The results obtained from PCA and DCCM analysis are also complementing our observations from structural stability parameters and intermolecular interactions reported during 150 ns MD simulation.Density functional theory calculation performed over barosmin reveals the possible binding mode and its important site that would highlight the region responsible for nucleophilic and electrophilic attack during the interactions.Oxygen atoms were identified to be more prone to nucleophilic attack while hydroxyl groups showed affinity towards electrophilic attack.
In response to postprandial hyperglycaemic spikes, insulin is released from the pancreas and it stimulates glucose to efflux into the skeletal muscle.Hence, to assess this efficacy, we performed the gene ontology analysis.Herein, we observed the regulation of the TNF signal that is concerned with glucose homeostasis (Chen et al., 2018).Previously, barosmin has been reported to ameliorate TNF signals (Bakr et al., 2020) and also modulate other inflammatory mediators (Zaragoz et al., 2020).In addition, TNF signals are negative regulators of insulin signaling and inhibit glucose utilization in peripheral tissues (Hotamisligil et al., 1994).Therefore, we performed glucose uptake efficacy of barosmin in rat L6 cell lines.There was a concentration-dependent increase in glucose uptake in the presence of insulin reflecting its insulin secretagogue activity though the IC 50 was significantly higher (p < 0.001) compared to metformin.This suggests barosmin may not limit within the gastrointestinal tract to inhibit a-amylase and a-glucosidase but may get absorbed into the systemic circulation and promote glucose utilization in various tissues by modulating TNF signals.To an extent, we observed the barosmin's efficacy to promote glucose uptake which might be the outcome of TNF signals suppression.However, other molecular biology techniques ought to be used to support this theory.
In summary, we point out the efficacy of barosmin to regulate glucose homeostasis and attenuating postprandial hyperglycemia by inhibiting carbohydrate hydrolyzing enzymes (Figure 16).In addition, using both computational and in vitro inhibition activity, we demonstrated its efficacy to inhibit the serine protease DPP4, which is involved in the degradation of GLP-1.Moreover, we could only observe the glucose uptake efficacy of barosmin; however, the current study lacks the reason behind it.Since our computational finding shows the TNF signaling pathway regulation; probably barosmin may inhibit its signals and enhance insulin sensitivity including glucose transporters regulation (Bakr et al., 2020;Hotamisligil et al., 1994) which needs to be further studied.

Conclusion
In the present study, we used various molecular modeling approaches including molecular mechanics and quantum mechanics to evaluate the efficacy of barosmin against postprandial hyperglycemia whose outcomes were further confirmed with in vitro enzyme inhibitory activities and glucose uptake assay in L6 cell lines.The findings were moreover based on the binding affinity and efficacy of the molecule to affect the intrinsic function using both dry-and wet-lab data.The results show that barosmin could be an effective molecule to deal with hyperglycaemic spikes.However, further studies are required to evaluate enzyme kinetics and gene expression studies for TNF signaling pathways in L6 cell lines and are the perspective of the present findings.

Figure 2 .
Figure 2. Interaction of barosmin with (a) a amylase, (b) a-glucosidase, and (c) DPP4, acarbose with (d) a-amylase, and (e) a-glucosidase and sitagliptin with (f) DPP4.Ligands are represented in balls and sticks.The pink and green zone of the cloud represents the hydrogen bond donor and acceptor region respectively.11111 represents the active site.

Figure 7 .
Figure 7. Principal component analysis of ligand-protein complexes.The collective motions of barosmin and standard compounds with (a) a-glucosidase (PDB: 3TOP), (b) a-amylase (PDB: 4W93), and (c) DPP4 (PDB: 6B1E) using projections of MD trajectories on two eigenvectors corresponding to the first two principal components PC1 and PC2.The first 50 eigenvectors were plotted v/s eigenvalues for barosmin and standard compounds with (d) a-glucosidase (e) a -amylase, and (f) DPP4.

Figure 10 .
Figure 10.Top 10 gene ontology terms triggered by barosmin (a) cellular components, (b) molecular function, and (c) biological process.Barosmin was identified to affect different cellular components like the nucleus, cytosol, neoplasm, extracellular space, and peroxisomal matrix.Barosmin may possess receptors, enzymes, transcription coactivators, and zinc bindings.It also possessed transcription factor activity.Further, it has a response to estradiol, hypoxia, and drugs.It had a molecular function in monocyte chemotaxis, signal transduction and regulation of the MAPK Cascade, angiogenesis, gene expression, and transcription.

Figure 11 .
Figure 11.a-amylase inhibitory activity of the barosmin.(a) concentration-dependent a-amylase inhibition and (b) comparison of IC 50 for barosmin and acarbose.Experiments were performed in triplicates.The above data are expressed in mean ± SD.The IC 50 of barosmin and acarbose was found to be (95.77± 23.33) and (58.36 ± 3.30) mg/mL respectively.� p < 0.05 compared to barosmin.Data were analyzed using a one-tailed unpaired t-test.

Figure 12 .
Figure 12. a-glucosidase inhibitory activity of the barosmin.(a) concentration-dependent a-glucosidase inhibition and (b) Comparision of IC 50 for barosmin and acarbose.Experiments were performed in triplicates.The above data are expressed in mean ± SD.The IC 50 of barosmin and acarbose was found to be (68.13± 2.95) and (80.28 ± 3.18) mg/mL respectively.� p < 0.01 compared to barosmin.Data were analyzed using a one-tailed unpaired t-test.

Figure 13 .
Figure 13.DPP4 inhibitory activity of the barosmin.(a) concentration-dependent DPP4 inhibition and (b) a comparison of IC 50 for barosmin and sitagliptin.Experiments were performed in triplicates.The above data are expressed in mean ± SD.The IC 50 of barosmin and sitagliptin was found to be (13.27± 1.99) and (9.62 ± 0.20) mg/mL respectively.� p < 0.01 compared to sitagliptin.Data were analyzed using a one-tailed unpaired t-test.

Figure 14 .
Figure14.Glucose uptake efficacy of the barosmin vs metformin in the presence of insulin.(a) Concentration-dependent glucose uptake efficacy and (b) comparison of EC 50 for barosmin and metformin.Experiments were performed in triplicates.The above data are expressed in mean ± SD.The EC 50 of barosmin and metformin was found to be (12.46 ± 0.90) and (7.14 ± 0.34) mg/mL respectively.� p < 0.001 compared to barosmin.Data were analyzed using a one-tailed unpaired t-test.

Figure 15 .
Figure 15.Cytotoxicity (MTT) assay of barosmin vs TBHP in L6 cell lines.(a) Concentration dependent cytotoxicity and (b) comparison of IC 50 for barosmin and metformin.Experiments were performed in triplicates.The above data are expressed in mean ± SD.The IC 50 of barosmin and TBHT were found to be (275.4± 47.06) and (44.57± 6.472) mg/mL respectively.� p < 0.05 compared to barosmin.Data were analyzed using one-tailed unpaired t-test.

Table 1 .
Binding affinity of ligands, hydrogen bond count, and amino acid residues with respective targets.

Table 2 .
Individual components contributing to the binding free energy of each ligand with respective targets.
All the data are presented in mean ± SEM (n ¼ 100); � Standard, DVDWAALS: Vander Waals molecular mechanics energy; DEEL: Electrostatic molecular mechanics energy; DEPB: Polar contribution to the solvation energy; DGTotal: Total relative binding energy.