We followed the procedure outlined by Ojo et al. (2022) to evaluate the ability of the extracts to scavenged DPPH radicals. For the comparative analysis, ascorbic acid served as the reference standard. The percentage of DPPH inhibition was determined by employing the following formula: % D P P H = A b s c o n t r o l − A b s s a m p l e A b s c o n t r o l   x   100

The standard procedures of (Ojo et al., 2022; Ahmad et al., 2023) with a slight modification were used to measure the ability of EFSFL to reduce ferric ions. We measured the absorbance at 593 nm and we reported the outcome as µmol Fe (II)/g of the powder’s dry weight using the FeSO4 standard curve for calculation.

This study followed the standard protocol (Ahmad et al., 2023) to determine the α-amylase inhibitory potential of EFSFL. To begin, we made a fresh preparation of enzyme, comprising 5 units per milliliter, in pH 6.7 ice-cold PBS with a concentration of 20 mM and including 6.7 mM NaCl. Then, 250 µL of the enzyme was combined with inhibitors (acarbose or EFSFL) at varying concentrations (excluding a blank sample), and the mixture was incubated at 37°C for 20 min. We subsequently added a starch solution with a concentration of 0.5% (w/v), and the mixture was incubated for an additional 15 min at 37°C. Immediately after the DNS reagent was added, the mixture was mixed and placed in a water bath at 100°C for 10 min. Finally, the absorbance was read at 540 nm.

The effect of EFSFL on intestinal α-glucosidase activity was evaluated using a technique described by (Bouslamti et al., 2023), which quantified the glucose generated by sucrose breakdown. To perform the assay, 100 µl of sucrose (50 mM), 1000 µl of phosphate buffer (50 mM; pH = 7.5), and 100 µl of α-glycosidase enzyme solution were prepared as the test solution (10 I.U.). Control (distilled water), positive control (acarbose), or EFSFL were all added to this mixture at varying concentrations. We read the absorbance of the ultimate solution at 500 nm.

We carried out the inhibition of DPP-IV activity in 96-well ELISA plates. Evogliptin was used as the standard inhibitor. In each well, 30 μl of Tris-HCl buffer solution, 100 μL of enzyme DPP-IV, 100 μl of various concentrations (15.625, 31.25, 62.5 125, 250, 500, and 1000 μg/ml) of the fraction or standard, and 50 μL gly-pro-pnitroanilide as the substrate, were added. After adding the mixture, they incubated it for 30 min at 37°C, and then the absorbance was read by microplate readers at 405 nm (Bharti et al., 2012).

We calculated %inhibition using the formula: %   I n h i b i t i o n = D P P − I V   a c t i v i t y   w i t h   f r a c t i o n D P P − I V   a c t i v i t y   w i t h o u t   f r a c t i o n   X   100

We conducted the procedure for the HRBC membrane stabilization assay following established protocols (Hogan et al., 2023). In summary, 2 ml reaction mixtures were prepared by combining varying amounts of EFSFL or a standard diclofenac drug with a 10% suspension of red blood cells. We kept these mixtures for 30 min at a temperature of 56 °C. Afterward, they were subjected to centrifugation at a speed of 2500 revolutions per minute (rpm) for 5 min. The resulting liquid above the sediment, known as the supernatant, was read for its absorbance at a wavelength of 560 nm.

We carried out this assay to measure the inhibition of protein denaturation by following the procedure described by Hogan et al. (2023). The test sample included different concentrations of EFSFL and/or standard diclofenac, with a 10 µl 1% solution of bovine serum albumin. This mixture was heated to a temperature of 55°C for 30 min and left to cool. The turbidity of the samples was measured at a wavelength of 660 nm, and the extent to which the protein denaturation was prevented was calculated as the percentage of inhibition.

The proteinase inhibitory assay was carried out using the approach outlined in (Hogan et al., 2023). The procedure involved incubating 1 mL of the extract with a reaction mixture containing 2 ml of Tris-HCl buffer and 0.06 mg trypsin at 37°C for 5 min. Then, 0.8% (w/v) casein was added to the reaction mixture and allowed to incubate for 20 min. The reaction was stopped by adding 2 ml of 70% perchloric acid, and the resulting supernatant was measured for absorbance at 210 nm after centrifugation.

The study utilized healthy male Wistar rats, weighing between 200 and 250 g each, obtained from the Anatomy Department at Bowen University in Iwo, Nigeria. Before being put to death using ketamine, the rats underwent an overnight fasting period. The livers were then extracted and blended in a 50 mM phosphate buffer solution containing 1% Triton X-100. For ex-vivo research purposes, we collected the supernatants in plain tubes after centrifugation at 15,000 rpm and 40°C. The study was approved under a specific identification number (BUAC/BCH/2023/0001A), and the protocols approved by Bowen University’s institutional animal ethics committee were followed when caring for the rats.

The protocol defined by Ojo et al. (2022) was employed to induce liver injury ex vivo in experimental rats. To perform this, the organ supernatant, which had different concentrations of EFSFL (varying between 31.25 μg/ml to 1000 μg/mL), was mixed with 200 μL. We added 100 μL of a solution containing 0.1 mM FeSO₄. We then placed the resulting mixture in a 37°C environment for 30 min to allow for biochemical analysis. In the normal control, only the organ supernatant was used in the reaction mixture, while the negative control comprises the tissue supernatant and FeSO₄.

We assessed the CAT activity assay of EFSFL with slight modifications to the method described by Ojo et al. (2022). Tissue samples containing different concentrations of EFSFL were mixed with 780 μl of 50 mM phosphate buffer, followed by the addition of 300 μl of 2 M H₂O₂. For 3 minutes, we read the absorbance at 240 nm at 1-min intervals.

We used the method provided in (Ajiboye et al., 2019) to determine SOD activity. To summarize, we mixed 170 μl of diethylenetriaminepentaacetic acid and 15 μl of the incubated sample in a test tube. Then, we added 15 μl of 6-hydroxydopamine to the solution and gently shook it. Finally, we measured the solution at 492 nm for 3 min with a 1-min interval.

Based on the protocol depicted in (Ajiboye et al., 2019), the tissue lysates, with a volume of 600 μl, were treated to remove proteins by adding 600 μl of a solution containing 10% TCA. After 10 min of centrifuging the mixture at 3500 rpm, 500 μl of the sample was transferred to a new test tube. Next, 100 μl of Ellman reagent was added to the sample, and the mixture was allowed to incubate at a temperature of 25°C for 5 min. We subsequently measured the absorbance of the solution at a wavelength of 415 nm.

The ability of EFSFL to inhibit lipid peroxidation by following the method outlined in reference (McIntyre et al., 2019) They carried out the process by taking 100 μl of tissue lysates that contained varying concentrations of EFSFL. We added 1000 μl of 0.25% thiobarbituric acid, 100 μl of 8.1% SDS, and 375 μl of 20% acetic acid to it. We boiled the mixture at 95°C for an hour in a water bath. After cooling it to room temperature, they measured its absorbance at 532 nm.

The retrieval of protein structures was from the Protein Data Bank (http://www.rcsb.org) for the deposited three-dimensional structures of human dipeptidyl peptidase IV (DPP IV) complexed with evogliptin (PDBID: 5Y7K), human pancreatic α-amylase (HPA) complexed with acarbose (PDBID: 1B2Y), and human α-glucosidase complexed with acarbose (HG) (PDB ID: 3TOP). The existing ligands and water molecules were removed from all the crystal structures while missing hydrogen atoms were added using MGL-AutoDock Tools (ADT, v1.5.6) (Morris et al., 2009).

The retrieval of Structure Data Format (SDF) of acarbose (reference inhibitors) and 13 phytocompounds identified by GC-MS analyses of EFSFL were downloaded from the PubChem database (www.pubchem.ncbi.nlm.nih.gov) before their conversion to PDB chemical format using Open Babel (O'Boyle et al., 2011). Non-polar hydrogen molecules were merged with the carbons, while the polar hydrogen charges of the Gasteiger-type were assigned to atoms. Furthermore, ligand molecules were converted to dockable PDBQT format with the help of AutoDock Tools.

The virtual screening docking protocol was validated by aligning the docked poses of the native ligands (acarbose and evogliptin) with the extracted co-crystallized ligand from both proteins, which had the lowest binding affinity from the initial docking. We calculated the RMSD using Discovery Studio Visualizer, BIOVIA, 2020.

The active site targeted molecular docking with the reference inhibitors and the GC-MS identified compounds against DPP IV, HPA, and HG was performed using AutoDock Vina in PyRx 0.8 (Trott and Olson, 2010). For the docking analysis, the ligands were imported, and energy minimization was accomplished using Open Babel (O'Boyle et al., 2011) incorporated into PyRx 0.8. The Universal Force Field (UFF) and conjugate gradient descent were employed as the energy minimization parameter and optimization algorithm, respectively. Although other parameters were left at their default values, the binding site coordinates of the target enzymes are shown in Supplementary Table S1, and the molecular interactions were viewed using Discovery Studio Visualizer version 16.

For ensemble-based docking, the DPP IV, HPA, and HG apoenzymes were subjected to a 50 ns simulation of molecular dynamics. The MD trajectory obtained was also used in cluster analysis. GROMACS 2019.2 and GROMOS96 43a1 forcefield (Lee et al., 2016; Lee et al., 2017; Lee et al., 2020) were used for the analysis. We generated protein and ligand topology files using the Charmm GUI (Lee et al., 2016; Lee et al., 2020). The solvation system, periodic boundary conditions, physiological conditions, minimization of the systems, and equilibration in a constant number of atoms, constant pressure, and constant temperature (NPT) used in the simulation are similar to those in our previous report (Gyebi et al., 2021; Ogunyemi et al., 2021; Ogunlakin et al., 2023; Ogunyemi et al., 2023). Velocity rescales and Parrinello-Rahman barostat were used to maintain the temperature and pressure at 310 K and 1 atm, respectively. We used a 2-femtosecond time step with a leap-frog integrator. Each system underwent a 100 ns simulation, with snapshots taken every 0.1 nanosecond and totaling 1000 frames for each system. From the MDs trajectories, the RMSD and RMSF were computed and presented as Supplementary Material.

A representative conformation that was obtained from the generated cluster from the clustering of the 500 ns MD trajectories of the unbound enzymes was employed for ensemble-docking studies. We performed the MD simulation trajectory clustering using TTClust V 4.9.0 (Lee et al., 2020). The systems were automatically clustered using the Python TTClust package, which uses the elbow approach to establish the ideal number of clusters before generating a representative frame for each cluster.

Using the AutoDock Vina program (Trott and Olson, 2010), the phytocompounds identified by GCMS were docked to the various representative conformers of DPP IV (4), HPA (2), and HG (4). The average binding affinities of the phytochemicals to each of the proteins were computed from the binding affinities of the phytochemicals to each of the conformations of the enzymes. The average binding energies of the compounds for the two targets were calculated and then their affinities were scored. The Discovery Studio Visualizer, BIOVIA 2020, was used to see how the lead chemicals interacted with one another on a molecular level.

The top two phytochemicals of each protein were then analyzed for ADMET filtering and drug-likeness using a variety of descriptors. SwissADME was used for the drug-likeness analysis using Lipinski filtering methods (http://www.swissadme.ch/index.php) on the webserver, while the anticipated toxicity, distribution, metabolism, and absorption (ADME/tox) study was analyzed with the SuperPred webserver (http://lmmd.ecust.edu.cn/admetsar1/predict/). The SDF file and the canonical SMILES of the compounds were downloaded from the PubChem database or copied from ChemDraw to calculate the ADMET properties using default parameters.

EFSFL efficiently blocked α-amylase in a concentration-dependent manner and had an IC₅₀ value of 307.02 ± 4.25 μg/ml compared with those of the standard (Figure 2A; Table 1). Acarbose revealed an 87.94% inhibitory property of α-amylase (IC₅₀ 121.79 ± 2.26 μg/mL) (Figure 2A; Table 1).

EFSFL efficiently inhibited α-glucosidase at all concentrations tested and an IC₅₀ value of 215.51 ± 0.47 μg/ml (Figure 2B; Table 1). Acarbose revealed an 89.68% inhibitory property of α-glucosidase (IC₅₀ 189.79 ± 0.67 μg/ml) (Figure 2B; Table 1).

Figure 2C shows the DPP-IV inhibitory activity of EFSFL. With an IC₅₀ value of 380.94 μg/mL (Table 1), EFSFL inhibited DPP-IV in a concentration-dependent manner. The EFSFL DPP-IV inhibitory activity, however, was less potent than that of the standard DPP-IV inhibitor, evogliptin (IC₅₀ = 211.35 μg/ml).

Our data revealed that the EFSFL demonstrated HRBC membrane stabilization potential (IC₅₀ = 319.85 ± 4.54 μg/mL) in a concentration-dependent manner. The standard drug, diclofenac, showed 89.68% membrane stabilization potential (IC₅₀ 3.63 ± 1.32 μg/mL) (Figure 3A; Table 1).

Our analysis also revealed concentration-dependent increases in the inhibition of protein denaturation (IC₅₀ = 72.75 ± 11.06 μg/mL) by EFSFL. The standard drug, diclofenac, also exhibited protein denaturation inhibition (IC₅₀ = 59.13 ± 12.40 μg/mL) (Figure 3B; Table 1).

Our analysis revealed that EFSFL exhibited significant proteinase inhibitory activities ((IC₅₀ = 296.08 ± 11.47 μg/mL). The standard drug, diclofenac, also exhibited proteinase inhibitory activities (IC₅₀ = 154.62 ± 4.29 μg/mL) (Figure 3C; Table 1).

Figure 4A showed a significant (p < 0.05) reduction in CAT activities in the liver tissues of FeSO₄-induced animals. Treatment with EFSFL concentration-dependently (p < 0.05) increased the activity in a significant manner, with the most pronounced activity found in the 1000 μg/ml treated group.

SOD activities were significantly (p < 0.05) reduced in the liver tissues of FeSO₄-induced animals (Figure 4B). Treatment with EFSFL dose-dependently (p < 0.05) elevated SOD activity in a manner close to that of the normal group.

GSH levels were drastically (p < 0.05) decreased in the liver tissues of FeSO₄-induced animals (Figure 4C). In contrast, EFSFL-treated groups significantly (p < 0.05) increased the levels of GSH in a manner close to the normal group, with the most pronounced effect found in the 1000 μg/ml treated group.

The MDA level was notably increased in the liver tissues of FeSO₄-induced animals (Figure 4D). In contrast, EFSFL-treated groups significantly (p < 0.05) reduced the level of malondialdehyde to near normal, with the most striking effect found in the 1000 μg/ml treated group.

The protocol to be used was validated to predict the precision, reliability, and accuracy of the docking protocol (Ogunyemi et al., 2023) before the docking of the GCMS-identified compounds to the target proteins. The generated docked poses of the reference compounds having the least energetic conformation were superimposed on the native ligand that was co-crystallized. After the superimposition, the root-mean-square deviation (RMSD) was computed. The RMSD, or acarbose complexed with 5y7K and 3top, was 3.5651 and 0.4341 Å respectively. The low RMSD shows that the docking protocol was suitable for the docking of phytochemicals (Supplementary Figure S2).

Table 3 shows the docking affinities of the 13 GCMS-identified phytochemicals from EFSFL and the reference molecule (acarbose) against DPP IV, HPA, and HG. Based on the lowest binding energies, binding poses, and interactions in the catalytic site, the top two ranked phytocompounds for each enzyme were selected for interactive analysis (Table 4). The two top docked phytocompounds to the DPP IV are alpha-caryophyllene and piperine, with binding affinity values of −7.8 and −7.8 Kcal/mol, respectively, compared to the reference inhibitor (evogliptin) (−7.8 Kcal/mol). Alpha-caryophyllene and piperine were the top-ranked phytocompounds for HG, with binding affinities of −9.6 and −8.9 Kcal/mol, respectively, compared to the reference inhibitor (acarbose), which had a binding affinity of 10.6 Kcal/mol. For HPA, the top-ranked phytocompounds were piperine and benzocycloheptano[2,3,4-I,j]isoquinoline, 4,5,6,6a-tetrahydro-1,9-dihydroxy-2,10-dimethoxy-5-methyl, with binding affinities of −7.8 and −7.9 Kcal/mol, compared to acarbose, which had a binding affinity of −8.3 Kcal/mol. The results from the initial docking studies showed that alpha-caryophyllene and piperine had high multi-target binding tendencies (Table 4).

The interaction of the reference compound and two top-docked phytochemicals with amino acids of the catalytic residues of DPP IV, HPA, and HG is represented in Table 5. The interaction of respective ligand groups with residues of the enzymes was majorly hydrophobic, with a few H-bonds below (less than 3.40 Å). Top-docked compounds were oriented in the active site of DPP IV and interacted with the amino acid to which we docked the reference compound. The interaction between alpha-caryophyllene and DPP IV was stabilized by a hydrogen bond with Ser209 and a hydrophobic contact with Phe357, while piperine formed two hydrogen bonds with Tyr662 and Tyr547 and pi-alkyl hydrophobic Tyr666, Arg358, and Arg356 (Figure 5).

Although the orientation of acarbose in the binding site of HPA was stretched into the five subsites, the top docked phytocompounds piperine and 4-hydroxy-3-methylacetophenone were docked into the −3 and −1 subsets of the α-amylase. Piperine did not establish hydrogen bonds like the reference compounds, but it did interact with the catalytic residues at the hydrophobic gate of α-amylase, which are Trp-59, Tyr62, and His299. BCQDM made several hydrogen bonds with the catalytic residues, including Asp197, Glu233, Arg195, Ala198, and His305; a pi-sigma contact with Leu165; and carbon hydrogen contacts with Asp197 and Glud233 (Figure 6). The piperine formed one hydrogen bond with Lys1460 and Arg1510, pi-pi stacking, and pi-pi-T-shaped stacking with Phe1427 and Trp1369 of HG. The 1-piperoyl moiety made pi-alkyl contact with Trp1355 and alkyl contact with Ile1280 and Tyr1251. While the bond between alpha-caryophyllene and HG with pi-alkyl contact with Phe1560 and Trp1355 (Figure 7).

From the trajectories obtained from the MDS analysis of the dipeptidyl peptidase IV, α-amylase, and α-glucosidase enzymes, the RMSD and RMSF plots were calculated to measure the extent of fluctuation during the simulation period (Supplementary Figures S3, S4). We obtained 1000 conformers of the 1000 frames from the MD simulation trajectories using TTclust to provide 4, 2, and 4 clusters for dipeptidyl peptidase, α-amylase, and α-glucosidase, respectively. The Supplementary Material shows the dimensions of the cluster, which are the number of individual frames that make up the clusters, the number of frames of the representative conformation, and the spread, which is the average distance among the conformations that make up the cluster. From these clusters, representative conformations were selected. An ensemble docking was performed by docking the phytochemicals to the representative structures of the various conformers. We calculated the mean and standard deviation for each enzyme of the phytochemical docking scores with minimal energy for each conformation (Figure 8). The results of the ensemble-based docking analysis further confirmed piperine as the phytochemical with the highest binding affinities to DPP IV (−7.65 ± 0.92 Kcal/mol), α-amylase (−8.15 ± 0.50 Kcal/mol), and α-glucosidase (−7.45 ± 0.468 Kcal/mol), while benzocycloheptano[2,3,4-I,j]isoquinoline, 4,5,6,6a-tetrahydro-1,9-dihydroxy-2,10-dimethoxy-5-methyl was the second top docked phytochemical to DPP IV (−7.95 ± 1.06 Kcal/mol), α-amylase (−8.14 ± 0.05 KKcal/mol), and α-glucosidase (−7. 1 ± 0.14 Kcal/mol) respectively. The representative cluster with the lowest binding affinities to the best-docked phytochemicals was selected for interactive analysis and is presented in Figure 9.

Predictive drug-likeness and ADMET (absorption, distribution, metabolism, excretion, and toxicity) filtering studies were conducted on the two hit compounds that were obtained from the ensemble-based docking analysis. The results are shown in Supplementary Tables S2, S3. The two top-docked phytochemicals, piperine and BCQDM, fulfilled the requirement for the four filters (Lipinki, Veber, Ghose, and Egan), hence they are predicted to have favorable druggable properties.

Piperine and BCQDM were further subjected to predictive ADMET analysis. The substantial gastrointestinal absorption of piperine and BCQDM suggests high bioavailability. Both piperine and BCQDM demonstrated the ability to cross the blood-brain barrier, which is a crucial characteristic of medications used for neurotherapeutic purposes (Supplementary Table S2). Piperine and BCQDM were predicted to be negative substrates of the p-glycoprotein with very high plasma protein binding tendencies. A variety of molecular cytochrome P450 descriptors were employed to investigate the effects of lead metabolites on the biotransformation of drugs in the liver. It was concluded that these descriptors would not be inhibited by piperine and BCQDM. Lead compounds were neither mutagenic, carcinogenic, nor likely to cause skin sensitivity, according to the prediction analysis. The projected LD₅₀, half-life, and clearance rate of the lead phytochemicals fell within an acceptable range (Supplementary Table S2; Figure 10).