Hypoglycemic and hepatoprotective effects in adult zebrafish (Danio rerio) of fisetinidol isolated from Bauhinia pentandra: In vivo and in silico assays

Abstract Diabetes mellitus is a chronic metabolic disorder that has been increasing drastically around the worldwide. It is important to emphasize that although many drugs are commercially available to treat diabetes, many of them have shown a number of adverse effects. Therefore, search for new antidiabetic agents is of great interest, and natural products, especially those obtained from plants sources, may be an alternative to available drugs. This study reports the in vivo and in silico evaluation of the hypoglycemic activity of fisetinidol. The conformational analysis confirmed that the fisetinidol compound possesses two valleys in the potential energy curve, showing a stable conformer on the global minimum of the PES defined by the dihedral angle θ (C6-C7-O-H) at 179.9°, whose energy is equal to zero. In addition, fisetinidol has shown promise in glycemic control and oxidative stress caused by hyperglycemia induced by high sucrose concentration, causing hypoglycemic and hepatoprotective effects in adult zebrafish. ADMET studies showed that fisetinidol has high passive permeability, low clearance and low toxic risk by ingestion, and computational studies demonstrated that fisetinidol complexes in the same region as metformin and α-acarbose, which constitutes a strong indication that fisetinidol has the same inhibitory mechanisms of α-acarbose and metformin. Communicated by Ramaswamy H. Sarma.


Introduction
Diabetes mellitus is a chronic metabolic disorder that has been increasing drastically around the worldwide (Ogurtsova et al., 2017). It is characterized by disorders in carbohydrates, proteins and fat metabolism caused by complete or partial insufficiency of insulin secretion and/or insulin action, which results in hyperglycemia (Diabetes, 2012). Additionally, the hyperglycemia leads to severe complications such as angiopathy, retinopathy, neuropathy, deficiency in the antioxidant defense system, and lipid profile disorders (Shirali et al., 2013) and various studies report that there is a correlation between oxidative stress-induced hyperglycemia and progression of complications in diabetic patients ( € Ozkaya et al., 2011). It is important to emphasize that although many drugs are commercially available to treat diabetes, many of them have shown a number of adverse effects (Maurya et al., 2021). Therefore, search for new antidiabetic agents is of great interest, and natural products, especially those obtained from plants sources, may be an alternative to available drugs.
It is important to highlight that the fisetinidol ( Figure 1) is a flavonoid found in different Bauhinia species de Oliveira Monteiro et al., 2021;de Sousa et al., 2016;G ois et al., 2017) and a previous study showed that fisetinidol, when evaluated as an alpha-glucosidase inhibitor, had superior activity than acarbose, which was used as a positive control (de Oliveira Monteiro et al., 2021).
To support the hypoglycemic effects of fisetinidol, which showed promising results as an alpha-glucosidase inhibitor, this study reports the in vivo and in silico evaluation of the hypoglycemic activity of fisetinidol.

Isolation and structural determination of fisetinidol
Fisetinidol was isolated from the the ethanol extract of Bauhinia pentandra stems according to the methodology described by da . 1 H and 13 C NMR (1 D and 2 D) spectra were performed on Bruker Avance DPX-300 spectrometer, operating at 300 for 1 H NMR, and 75 MHz for 13 C NMR. The spectra were measured in CD 3 OD and the chemical shifts (d) are expressed in ppm. Silica gel 60 (70-230 mesh, Merck) was used for column chromatography. Thin layer chromatography (TLC) was performed on precoated silica gel polyester sheets (kieselgel 60 F 254 ,0.20 mm,Silicycle,Quebec,Canada), and the spots were visualized by UV detection and/or heating after spraying with vanillin/perchloric acid/EtOH solution.

Conformational analysis
Conformational analysis was carried to obtain the potential energy surface (PES) on the rotation of the dihedral angles of the four hydroxyl groups of the fisetinidol molecule, and the conformers of the most stable conformations, those whose energy are equal to zero and the values of the rotational barriers of highest energy of these conformers were obtained. The input molecule was drawn using the Avogadro 1.2.0 (Hanwell et al., 2012), and the geometrical optimization calculation was done at the B3LYP/6-311þþG(d,p) computational level (Becke, 1992;Lee et al., 1988) in the gas phase using the Gaussian 09 software (Frisch et al., 2009).

Zebrafish
Zebrafish (Danio rerio) (age from 90 to 120 days; 0.4 ± 0.1 g, 3.5 ± 0.5 cm), of both sexes, was purchased at a local store (Fortaleza, CE, Brazil). The animals were kept in a glass aquarium (30 Â 15 Â 20 cm) of 10 L (n ¼ 3/L), at a temperature of 25 ± 2 C, in light-dark cycles of 24 h with chlorinated water (ProtecPlusV R ) and air pump with submerged filters, at a temperature of 25 C and pH 7.0, circadian cycle of 10-14 h (light/ dark). The fish received feed (SpirulinaV R ) ad libitum 24 h before the experiments. Before drug applications, the animals were anesthetized in ice water and after the experiments, the animals were sacrificed by immersion in ice water (0 and 3 C) for 1 min until loss of opercular movements. The work was approved by the Ethics Committee on the Use of Animals of the State University of Cear a (CEUA-UECE; n 04983945/2021), in accordance with the Ethical Principles of Animal Experimentation.  there was a change in the animal's motor coordination or possible visible side effects in behavior, either by sedation and/or muscle relaxation. Zebrafish ZFa (n ¼ 6/group) were treated orally (v.o) with 20 lL of fisetinidol solutions at doses (4, 20 and 40 mg/Kg) and vehicle (DMSO 3%). After 1 h of treatments, the animals were added to Petri dishes, containing the same water as the aquarium, marked with quadrants and analyzed for locomotor activity by counting the number of crossing lines for 5 min. Animals that did not receive treatments (Naive) were considered as baseline (100% of locomotor activity) and the percentage of locomotor activity was calculated (%AL).

Toxicity 96 h
The acute toxicity study was carried out against adult zebrafish according to the Organization for Economic Cooperation and Standard Development Method (OECD, 1992) to determine the DL 50 -96h. Dead individuals were removed immediately. In the first experiment, the animals (n ¼ 6/group) were orally treated with the fisetinidol sample at doses (4; 20 and 40 mg/Kg; 20 mL). As a negative control, 3% DMSO (20 lL; i.p) was used. After 24, 48, 72 and 96 hours, the values obtained with the number of dead ZFa were submitted to statistical analysis, estimating the lethal dose to kill 50% (LD 50 ) ZFa through the mathematical method Trimmed Spearman-Karber with intervals of 95% confidence (Arellano-Aguilar et al., 2015).

Sucrose-induced hyperglycemia (83.25 mM) in adult zebrafish
The induction of hyperglycemia in adult zebrafish was performed based on the methodology of (Ranjan & Sharma, 2020) with modifications. The animals (n ¼ 6/group) were immersed in sucrose solution (83.25 mM/L) in dechlorinated water in 5 L glass aquariums for seven days. The sucrose solution was changed every day at the same time. On the 8th day, the animals were removed from the sucrose solution and kept until the 11th day in dechlorinated water. (There was no death of animals during the induction of hyperglycemia). The groups of animals (n ¼ 6/group) that spent the seven days submerged in sucrose were pretreated orally (using a 20 mL automatic pipette) after a 12 h fast for four days (at the same time) with fisetinidol (40 mg/Kg; 20 ml). Other groups of animals (n ¼ 6/group) were treated with controls-Acarbose (300 mg/Kg; 20 mL; positive control), Metformin (200 mg/Kg; 20 mL; positive control), or 3% DMSO (Drug thinner; Negative control). A group of untreated zebrafish was included (Naive). Blood glucose levels were measured on the fourth day after four hours of treatments. Before blood collection, the animals were euthanized (by inducing hypothermia with 0 and 3 C ice and water), then blood glucose readings were taken by placing a glucometer test strip (Active, Accu Check) directly on the docked tail of each animal (Capiotti et al., 2014).
2.4.4. Reactive oxygen species (ROS) levels in hyperglycemic zebrafish liver Oxidative stress was assessed in animals treated for four days as previously described. The animals (n ¼ 6/group) that were euthanized for the glycemic test had the liver removed. In the assay, 2 0 ,7 0 -dihydrodichlorofluorescein diacetate (DCHF-DA) was used. Liver tissues from two animals, in triplicate, were macerated in Tris-HCl-EDTA followed by centrifugation (10,000 g Â 10 min). The supernatant was collected (50 mL) and mixed with 5 mL of DCHF-DA. The oxidation of DCHF-DA to fluorescent dichlorofluorescein (DCF) was measured for the detection of ROS. The fluorescence intensity emission of the DCF was recorded at 520 nm (with excitation of 480 nm) 2 h after the addition of DCHF-DA to the sample. Protein concentration was determined using the method of UV light spectrophotometry at 280 nm through a standard curve of bovine serum albumin (BSA) (Loetchutinat et al., 2005).

Preparation and optimization of binders
Fisetinidol molecule was geometrically optimized using the Density Functional Theory (DFT) method at B3LYP/6-311þþG(d,p) computational level (Becke, 1992;Ditchfield et al., 1971;Lee et al., 1988) in the gas phase using the Gaussian 09 software (Frisch et al., 2016). From the optimized geometry, the Molecular Electrostatic Potential was computed at the same level of theory, and the isosurface was rendered (isovalue ¼ 0.01) with the Gabedit 2.5.0 software (Allouche, 2011).

General docking procedures
For molecular docking studies, the C-terminal subunit maltase-glucoamylase enzyme (CtMGAM) was selected. CtMGAM is involved in the production of glucose in the human lumen and is considered a promising target for the study of drug development for type 2 diabetes (Ren et al., 2011). To obtain the three-dimensional coordinates of CtMGAM, data available in the Protein Data Bank-RCSB repository (https://www.rcsb. org/) were used. The structure of CtMGAM (PDB-codex 3TOP) was solved by X-ray diffraction, with the crystal obtained with a resolution of 2.88 Å (R-Value Free: 0.284, R-Value Work: 0.218 and R-Value Observed: 0.222 respectively) , being classified as a hydrolase/hydrolase inhibitor of the Homo sapiens organism, being expressed in the Komagataella pastoris system (Ren et al., 2011). CtMGAM was co-crystallized with acarbose inhibitor (PRD_900007) (reference inhibitor).
To carry out the molecular docking simulations, Lamarkian Genetic Algorithm-LGA was chosen, implemented in the AUTODOCK Vina code (Trott & Olson, 2010). To determine the simulation space, the grid box was centralized in order to encompass all protein chains. The grid box was centered on coordinates À45.828, 21.487 and 17.927 for the x, y and z axes respectively and had size parameters 126 Å(x), 82 Å(y) and 124 Å(z).
As criteria for preparing the protein structure, water molecules were removed, while Gasteiger charges and essential hydrogen atoms were added (Yan et al., 2014). The preparation was performed using the ADT-AutoDocktools code (Morris et al., 2009).
To obtain a larger data set, all simulations (docking and redocking) were performed 50 independent simulations, being possible to obtain 20 poses per simulation. The exhaustiveness equal 64 criterion was used to improve the partial refinement of individual coupling calculations (Marinho et al., 2020). The protein structure was kept rigid (Nguyen et al., 2017), while all binding and twists of the ligands were adjusted to rotate.
For statistical validation of the simulations, redocking procedures were performed and the RMSD (Root Mean Square Deviation) values were evaluated, having as parameter of choice of the Best Pose values less than 2 Å (Yusuf et al., 2008). To assess the stability of the complex (protein/ligand) the affinity energy (DGÞ (Equação 1), was used as parameter (Equation 1), which has ideality parameters values below À6.0 kcal/mol (Shityakov & Foerster, 2014). To assess the affinity of the ligand for the protein, the value of the inhibition constant (K 1 ) calculated by Equation 2 was used (Kadela-Tomanek et al., 2021).
where DG is the binding free energy in KJ.mol À1 , R is the gas constant, 8.32 J.mol À1 K À1 , T is the absolute temperature, 298 K and Ki is inhibition constant. To assess the strength of the H-bond, the values of the distances between the donor and recipient atoms were used, being classified as strong, the interactions that were between 2.5 Å and 3.1 Å, average those that were between 3.1 and 3.55 Å, and weak those with a distance greater than 3.55 Å (Imberty et al., 1991). For a validation of the simulations, the redocking technique of the co-crystallized ligand, acarbose (PubChem CID: 445421) was performed. In addition to acarbose, metformin (PubChem CID: 4091) and with the co-crystalized ligand acarbose (PubChem CID: 445421) were used, which was submitted to the same criteria and conditions for simulating the flavonoid.

Molecular dynamics
The simulations of molecular dynamics (MD) were realized through GROMACS (GROningen MAchine for Chemical Simulation) 2020.4 package software 2020.4 (Van Der Spoel et al., 2005) implemented with the CHARMM27 force field (Mackerell et al., 2000). The SwissParam server (Daina et al., 2017) was used to obtain parameters for the fisetinidol ligand. One box of the simulation was created containing the 3TOP protein and fisetinidol. Posteriorly, this box was solvated with water molecules described by the TIP3P model (Jorgensen et al., 1983) and additional 56 ions of sodium to neutralize the system. The geometry of the system was realized through the steepest descent algorithm (Arfken et al., 2013) with an energy tolerance of 10 kJ mol À1 nm À1 and step size of 10 À4 nm. The step equilibrium was divided into two parts and simulate with 1 ns for each part, the first one was realized with the ensemble NVT through of V-rescale thermostat (Bussi et al., 2007)

The chemical space of protein-ligand binding sites
The hydrogen bond donor atoms were mapped by the algorithms of the MarvinSketch v21.12 program, ChemAxon (https://chemaxon.com/products/marvin) and evaluated according to Lipinski's 2015 "rule of five" (Lipinski, 2016) for the favorable physicochemical space of ligand-receptor interaction, while the two-dimensional structure was submitted to the identification of interference substructures test from the PAINS library (Baell & Holloway, 2010) and to the similarity test of the bioactivity dataset with biological target classes from the Molinspiration (https://www.molinspiration. com/) and SwissTargetPrediction (http://www.swisstargetprediction.ch/) platforms.

Analysis of drug-likeness and ADMET profile
The physicochemical properties were calculated using the program MarvinSketch v21.12, ChemAxon (https://chemaxon. com/products/marvin) and applied to the drug-likeness filters of Lipinski's "rule of five" (Lipinski, 2004) and Veber's rule (Veber et al., 2002) embedded in the SwissADME platform (http://www.swissadme.ch/) and then transformed into druglikeness scores by the Central Nervous System Multiparameter Optimization algorithm (CNS MPO) from Pfizer (Wager et al., 2016) to correlate the viable physicochemical space of fisetinidol as a drug with the alignment of absorption, distribution, metabolism and excretion (ADME) taxes provided for in the consensual analysis of the SwissADME (http://www.swissadme.ch/) and pkCSM (http:// biosig.unimelb.edu.au/pkcsm/) platforms. The parameters of hepatotoxicity, Ames mutagenicity and acute oral toxicity (LD 50 ) were estimated through the similarity test with reactive molecular fragments from the ProTox-II platform (https://tox-new.charite.de/protox_II/index.php?site=home).

Statistical analysis
Results were expressed as mean values ± standard error of the mean for each group of 6 animals. After confirming the normal distribution and homogeneity of the data, the differences between the groups were submitted to analysis of variance (one-way ANOVA), followed by the Tukey test. All analyzes were performed with GraphPad Prism v software. 8.0. The level of statistical significance was set at 5% (p < 0,05).

Structural determination of fisetinidol
The analysis of the spectroscopic data (Table 1) of the fisetinidol allowed us to suggest the molecular formula C 15 H 14 O 5 , indicating nine degrees of unsaturation. The structural characterization of this compound was defined through some important signs that allowed the identification of the compound as a flavan-3-ol type flavonoid. The HSQC spectrum of the fisetinidol ( Figure S4, Supplementary Material) made it possible to make a direct association between the hydrogens and carbons. The analysis of the HMBC spectrum of fisetinidol ( Figure S5

Conformational analysis
The relaxed potential energy surface (PES) scan of the fisetinidol molecule calculated at the B3LYP/6-311þþ (d,p) level for the dihedral angle h (C6-C7-O-H) with step variation of 15 is shown in Figure 2(a).

Effect of fisetinidol on zebrafish locomotor activity
As a result, it was observed that fisetinidol did not cause motor impairment in the ZFa, as there was no reduction in Table 1. Spectroscopic data of (-)-fisetinidol.
HSQC HMBC   the number of line crossings in the petri dish by the animals, a result significantly similar to the naïve or vehicle group (p < 0.05) (Figure 3).

Toxicity 96 h
Through the zebrafish model, the assessment of drug toxicity and side effects can be performed, being a significant pre-filter for the early choice of safer drugs during their discovery process (Garcia et al., 2016). In this context, the adult zebrafish was used as an animal model to assess the acute toxicity of fisetinidol. As a result, it was found that the sample was not toxic against ZFa up to 96 h of analysis (LC 50 > 40 mg/Kg).

Sucrose-induced hyperglycemia (83.25 mM) in adult zebrafish
After seven days of zebrafish immersed in sucrose at 83.25 mM, a significant persistence ( ÃÃÃ p < 0.001 vs Naive) of hyperglycemia was observed in animals after four days of sucrose withdrawal (Control: 96.5 ± 26.6 mg/dL vs Naive: 44.3 ± 6.8 mg/dL. The persistence of sucrose-induced hyperglycemia was significantly reduced ( Ã p < 0.05 vs Control) by fisetinidol, an effect similar to the positive control groups (Acar-Acarbose and Met-Metformin) which reduced ( ÃÃ p < 0.01; ÃÃÃ p < 0.001 vs control, respectively) hyperglycemia at basal levels in animals (Figure 4).

Reactive oxygen species (ROS) levels in hyperglycemic zebrafish liver
Sucrose-induced hyperglycemia increased ( ÃÃ p < 0.01 vs Naive) the levels of reactive oxygen species in the animals' livers. This oxidative stress was significantly reduced ( Ã p < 0.05 vs Control) by fisetinidol, an effect similar to the positive control groups (Sugar-Acarbose and Met-Metformin; Ã p < 0.05 vs Control) ( Figure 5).

Molecular docking
Initially, the RMSD (Root-Mean-Square Deviation) was used for statistical validation of the results of the complex formation simulations and choice of the best pose. The RMSD is calculated based on the measurement of the average distance between the atoms of the two ligands, with validation criteria being values close to 2 Å. All simulations performed (docking and re-docking) presented RMSD values lower than 2 Å, highlighting the best pose of the CtMGAM complex, which presented rmsd in the order of 1.713 Å (Table 2). From the best pose choices based on the RMSD, the affinity energy of the complexes for the ligands was evaluated, where again the fisetinidol-CtMGAM complex can be highlighted, which presented an energy in the order of À8.9 kcal/ mol. The metformin-CtMGAM complex presented energy in   the order of À5.2 kcal/mol. In the re-docking simulations, the acarbose-CtMGAM complex was in the order of À7.9 kcal/ mol. The fisetinidol-CtMGAM, acarbose-CtMGAM and metformin-CtMGAM complexes showed inhibition constants in the order of 2972 Â 10 À7 , 1608 Â 10 À6 and 1536 Â 10 À4 respectively. Based on the Ki value, it was possible to calculate the pKi, obtaining values of 6.52, 5.79 and 3.81 for the fisetinidol -CtMGAM, acarbose-CtMGAM and metformin-CtMGAM complexes respectively (Table 2). Regarding the interactions involved in the formation of the protein-binding complex, it was possible to identify that the fisetinidol-CtMGAM complex is formed by hydrophobic interactions involving the side chains of residues Trp 1355 A

Molecular dynamics
The root mean square deviation (RMSD) between the receptor and the ligand is obtained to analyze the interval time that the complex studied reached equilibrium. The 3TOP-FIS complex present in Figure 6 showed an increase of your RMSD at the beginning of simulation until 40 ns, however, the system reached the equilibrium in the time interval of 65-100 ns of simulation.
3.9. ADMET study 3.9.1. The chemical space of protein-ligand binding sites The calculated physicochemical properties are expressed in Table 3 and were applied to the criteria of Lipinski's rule for evaluating the favorable physicochemical space of fisetinidol   for ligand-receptor interaction. The results show that the molecule has four hydrogen donors responsible for strong hydrogen interactions with the C-terminal subunit maltaseglucoamylase (CtMGAM). Associated with this result, the Pan Assay Interference Structure (PAINS) filter showed a structural alert that characterizes the catechol substructure as a chemically reactive region (Figure 7(A)). This data corroborates the molecular docking simulation performed, since this molecular fragment acts as an electrophilic agent and contributes with two hydrogen donations with the residues of Asp 1279 A  and His 1584 A (d ¼ 2.08 and 3.11 Å, respectively). In addition, bioactivity prediction done on the Molinspiration platform showed that bioactivity scores > 0.2 estimate a good inhibitory activity of G protein-coupled receptors (GPCRs), enzymes and nuclear receptors (Table 4).

Analysis of drug-likeness and ADMET profile
The physicochemical properties of fisetinidol applied to the chemical space of ligand-receptor interaction were applied to Lipinski's "rule of five" to characterize its chemical space as a drug-like compound. The results show that the substance did not violate any of the criteria established by the rule (Molecular weight 500 g/mol, logP 5, H-bond acceptors 10 and donors 5), in addition to meeting the oral bioavailability criteria reported by the rule of Veber (TPSA 140 Å 2 and rotatable bonds 10) as properties that relate unsaturation, molecular size and H-bonds (Table 5). A visual inspection of the bioavailability radar in Figure 7(B) shows that the substance is within the ideal limits of lipophilicity (LIPO), size (SIZE), polarity (POLAR) solubility (INSOLU) and molecular flexibility (FLEX) that allow its therapeutic use as an oral drug, while higher than ideal unsaturation suggests that the substance tends to have low lipophilicity. In the graph of Figure 7(C) it is possible to observe the change in the solubility of fisetinidol with the change in pH. This analysis corroborates the calculated pKa value for the most acidic species formed (9.09), suggesting a decrease in lipophilicity at high pH levels, which tends to a small variation between physiological pH levels (around pH 7.4), where the calculated value of logD in the order of 2.09 suggests a balance between lipophilicity and solubility, which guarantees a good passive permeability of the molecule. The alignment of the physicochemical properties was converted into a score by the Central Nervous System Multiparameter Optimization (CNS MPO) algorithm, as shown in Figure 7(D), where it is possible to observe that the amount of hydrogen donors makes the substance's cerebral permeability impracticable and unlikely and may present a limitation in its bioavailability. However, a CNS MPO score >4 indicates that the substance has a good drug-likeness alignment, a relationship that is easily associated with the bioavailability radar in Figure 7(B), where the substance satisfies most of the criteria explored by the rules. The pharmacological properties of absorption, distribution, metabolism, and excretion and the toxicity profile (ADMET) are represented in Figure 7(E) and in the Table 5. Through the consensual ADME prediction of the SwissADME and pkCSM platforms, it is possible to observe that fisetinidol has a good passive permeability (Papp) in intestinal Caco-2 cells, which guarantees a predicted high gastrointestinal (GI) absorption of 93.12% and has its limited bioavailability as it is a substrate of P-glycoprotein (P-gp). The volume of distribution in the order of 1.01 L/kg (logVD) indicates that the substance has an excellent tendency to distribute among tissues, but without access to the CNS (logPS < À3) ( Table 5). The prediction corroborates the logP and TPSA parameters, which indicate that fisetinidol is in a physicochemical space where permeability at the blood-brain barrier (BBB) is unlikely (Figure 7(E)). Furthermore, the substance did not inhibit any of the cytochrome P450 enzyme isoforms evaluated (1A2, 2C19, 2C9, 2D6 and 3A4), indicating a good metabolic stability, which guarantees a low clearance rate of the free fraction in the hepatic system (logCL int,u ¼ 0.097 mL/min/kg). Through the similarity test with the dataset of the ProTox-II platform, it is possible to determine, with similarity pattern of 98.89% and reliability of 72.9%, that the lethal dose (LD 50 ) for predicted acute oral toxicity of 10000 mg/kg indicates that there is no risk of toxicity due to ingestion of the substance (toxicity class 6), while it is possible to notice a low incidence of hepatotoxic and mutagenic risk (reliability > 60%) (Table 5).

Discussion
The 1 H NMR spectrum ( Figure S1, Supplementary Material) of the fisetinidol showed signals in the region between d H 6.29-6.85 (Table 1), confirming the presence of hydrogens bonded to aromatic ring carbons. It is worth highlighting the signals at d H 6.85 (1H; d; J ¼ 8.4 Hz) and d H 6.76 (1H; d; J ¼ 8.4 Hz) related to ortho positioned hydrogens, in addition to signals at d H 6.83 (1H; d; J ¼ 1.9 Hz) and d H 6.29 (1H; d; J ¼ 1.9 Hz) assigned to hydrogens in meta position. Signals were also observed at d H 6.71 (1H; dd; J ¼ 8.2 and 1.6 Hz) and d H 6.34 (1H; dd; J ¼ 8.2 and 2.3 Hz), which according to the magnitudes of the coupling constants, they were assigned to hydrogens in ortho and meta positions. Additionally, the spectrum showed a doublet at d H 4.65 (1H; d; J ¼ 7.1 Hz) and a triplet at d H 4.01 (m) indicating the presence of hydrogens bonded to oxygenated carbons. This spectrum also revealed two doublets that indicated the presence of benzylic hydrogens in d H 2.87 (1H; dd; J ¼ 15.7 and 4.9 Hz) and d H 2.68 (1H; dd; J ¼ 15.7 e 8.0 Hz).
In the 13 C NMR spectrum ( Figure S2, Supplementary Material) of the fisetinidol it was possible to detect the presence of 14 spectral lines, with three more unprotected signals at d C 157.99, d C 156.27 and d C 146.37 associated with oxygenated sp 2 carbons. It is noteworthy that among the signals corresponding to non-hydrogenated carbons, the signal at d C 146.37 was related to two carbon atoms due to its intensity. The comparative analysis of the 13 C NMR and DEPT spectra ( Figure S3, Supplementary Material) allowed the identification of six non-hydrogenated carbons (C), eight methine carbons (CH) and one carbon methylene (CH2), as shown in Table 1.
The HMBC spectrum ( Figure S5, Supplementary Material) showed the correlations of the A-ring hydrogens allowing to fix the position of the hydroxyl groups. The hydrogen at d H 6.29 (H-8) correlated two bonds ( 2 JCH) with the carbons at d C 157.99 (C-7) and d C 156.27 (C-9). Correlations were also observed at 2 JCH and 3 JCH of hydrogen at d H 6.34 (H-6) with carbons at d C 157.99 (C-7) and d C 112.70 (C-10), respectively. Additionally, hydrogen at d H 6.85 (H-5) correlated to three bonds with carbon at d C 33.24 (C-10) ( Figure  8(A)). The positions of the hydroxyl groups in the B ring were defined by correlating the three and two hydrogen bonds at d H 6.87 (H-2 0 ) with the carbons at d C 83.10 (C-2) and d C 146.37 (C-3 0 ), respectively; correlation of two hydrogen bonds at d H 6.76 (H-5 0 ) with carbon at d C 146.37 (C-4 0 ); correlations of three hydrogen bonds at d H 6.87 (H-2 0 ) and d H 6.71 (H-6 0 ) with carbon at d C 83.10 (C-2). Furthermore, hydrogen at d H 4.65 (H-2) correlated to 2 JCH with carbon at d C 146.37 (C-3 0 ) (Figure 8(B)). In the spectrum, correlations in the central ring were also observed. The hydrogens at d H 2.68 (H-4) and d H 2.87 (H-4) correlated two bonds ( 2 JCH) with the carbons at d C 68.94 (C-3) and d C 112.70 (C-10). Additionally, hydrogen at d H 4.65 (H-2) correlated two and three bonds with carbons at d C 68.94 (C-3) and d C 33.24 (C-4), respectively. (Figure 8(C)). In the 1 H-1 H COSY spectrum ( Figure S6 (Figure 8(D)). On the basis of these spectroscopic data (Table 1), the compound was identified as 2-(3 0 ,4 0 -dihydroxyphenyl)-chroman-3,7-diol, commonly known as (-)-fisetinidol (Figure 1).
The conformational analysis of the fisetinidol was carried out with the purpose of interpreting and predicting the physicochemical properties through its various conformations. The information on conformational analysis allows for knowledge and detailed descriptions of equilibrium geometries. This information is important in hypoglycemic effect, as these conformers indicate how the functional groups of the fisetinidol are oriented, discovering relevant aspects of how the molecule can interact with CtMGAM receptor, since the most stable conformation must be in greater numbers during the process of interaction with the receptor.
The conformational analysis confirmed that the fisetinidol possesses two valleys in the potential energy curve, showing a stable conformer on the global minimum of the PES defined by the dihedral angle h (C6-C7-O-H) at 179.9 , whose energy is equal to zero. Figure 2(b) shows the stable conformer on the global minimum of the PES, with the red plane containing the C7 and C3 carbons, which are linked to hydroxyl groups, and the blue plane containing the hydroxyl groups, which are linked to the C3 0 and C4 0 carbons. The angle / between these planes is 60.73 . Therefore, the stable conformer is fully nonplanar. The PES scan on the dihedral angles r .57 kcal/mol at h ¼ À90.1 , 3.33 kcal/mol at r ¼ 8.1 , 6.76 kcal/mol at c ¼ 81.2 , and 6.80 kcal/mol at u ¼ 78.8 .
The effect of hyperglycemia induced by high sucrose concentration on zebrafish learning and memory was investigated (Ranjan & Sharma, 2020). The same procedure was done inducing hyperglycemia in adult zebrafish by immersion in 83.25 mM sucrose for seven days. With the removal of sucrose after four days, hyperglycemia persisted after 12 h  of fasting in the animals (control group-3% DMSO) ( Figure  3). The increase in blood glucose is a result of impaired insulin secretion by b cells and the demonstration of a false compensation of insulin sensitivity in peripheral tissue (Oyelaja-Akinsipo et al., 2020). The basal level of fasting blood glucose in zebrafish is approximately 50-75 mg/dL (J€ orgens et al., 2012) and in this study, fisetinidol reduced sucrose-induced hyperglycemia to basal levels after 4 days of treatment (Figure 4), a result similar to that obtained in this study with positive controls (acarbose and metformin) and untreated group (Naive) after 12 h of fasting ( Figure 4). KKAy mice that received a diet with polyphenols extracted from the bark of Acacia meansii rich in fisetinidol had their body weight, glucose and plasma insulin suppressed (Ikarashi et al., 2011;Kashiwada et al., 2021), a result similar to that obtained in this study with the isolated fisetinidol, demonstrating the hypoglycemic effect of fisetinidol.
In this study, fisetinidol also demonstrated a hepatoprotective effect on oxidative stress induced by hyperglycemia in adult zebrafish, similarly to the effect of acarbose and metformin (positive controls) ( Figure 5), demonstrating its promising effect in the control of metabolic syndromes such as diabetes and stress oxidative. The hapatoprotective effect of fisetinidol can be justified by the fact that polyphenols are involved in human defense mechanisms, particularly against oxidative stress (Pandey & Rizvi, 2009). In this way, they prevent or treat certain oxidative-based human pathologies.
Using the molecular docking technique showed that fisetinidol has a more favorable energy than acarbose and metformin, as it presented a lowest affinity energy (À8.9 kcal/ mol) (Duong et al., 2020). In addition, it is possible to infer that fisetinidol will need a lower concentration to inhibit CtMGAM, as it presented a lower ki, in the order of 2.972 Â 10 À7 and a higher pKi value (6.52) when compared to the values of metformin and the co-crystallized inhibitor (acarbose) (Kadela-Tomanek et al., 2021).
When analyzing the formation of the complexes, it is possible to observe that the fisetinidol-CtMGAM complex, is formed mainly by H-bonds, the interactions can be explained through the analysis of the Molecular Electrostatic Potential (MEP) of fisetinidol molecule (Figure 9). It can be noted that oxygen atoms from the hydroxyl groups can make hydrogen bonds with the amino acids residues of the target protein (ASP-1279, ASP-1526, ARG-1510, and HIS-1584). In addition, comparing the interaction region, it is possible to observe that fisetinidol complexes in the same region as metformin and acarbose, interacting with the same residues, this fact is a strong indication that fisetinidol has the same inhibitory mechanisms of acarbose and of metformin on CtMGAM ( Figure 10).
Molecular dynamics showed the 3TOP-FIS complex with the Interaction Potential Energy (IPE) value of À196.634 kJ mol À1 (Figure 11). The high value of IPE for this complex is explained by the presence of hydrogen bonds, these intermolecular forces contribute with 11-60 kJ mol À1 (Steed et al., 2012) in the interaction potential energy and an of most force influential in the molecular recognition (Dong & Davis, 2021).
Several filters have been improved from Lipinski's "rule of five" to improve the ideal physicochemical space for a substance to be considered a new begging, preserving the ideal characteristics of ligand-receptor interaction (Lipinski, 2016). The results indicated that, the four hydrogen donor hydroxyls, with a strong contribution from the catechol group, identified as a chemically reactive substructure by the PAINS filter , are responsible for the strength of the hydrogen interaction between fisetinidol and the residues of Asp 1279 A, Arg 1510 A, Asp 1526 A, His 1584 A. In addition, it was possible to observe that the logD value calculated in the order of 2.09 suggests that fisetinidol has a good balance between solubility and lipophilicity, which favors its passive permeability. The warning associated with the hydrogen donor atoms suggests that penetration of the substance into the CNS is impracticable or unlikely, since this number of hydrogen donors increases the molecule's polarity. However, a CNS MPO score ¼ 4.95 indicates favorable pharmacokinetics. The CNS MPO score obtained for fisetinidol, close to the ideal predicted by the rule, corroborates the likely inhibition activity of GPCRs and enzymes. In these results, it was possible to observe that fisetinidol has 25.6% of its interactions with GPCRs and 14.6% with enzymes, in addition to having affinities with other transporters, including ion transport channels, as shown in the graph in Figure 12.
The BOILED-Egg graph of Figure 7(E) shows that fisetinidol is in a physicochemical space of low logP and high TPSA, where its pharmacokinetic activity is it is based on a high passive intestinal permeability (P app ) and does not tend to penetrate the BBB, due to the likely risk of efflux by P-gp back to the gastrointestinal tract, limiting its bioavailability in the systemic circulation. Fisetinidol's CNS MPO >4 score also favors its low metabolic clearance (logCL int,u ¼ 0.097 mL/min/ kg), due to the free molecular fraction in the hepatic system, resulting from the good metabolic stability due to non-inhibition of the evaluated CYP450 isoforms (1A2, 2C19, 2C9, 2D6 e 3A4) (Wager et al., 2010a(Wager et al., , 2010b. In addition, it was possible to observe the low susceptibility of fisetinidol to be a CYP450 substrate, not resulting in liver damage caused by potentially toxic biotransformed structures, while the cactechol structure identified as chemically reactive can form a covalent bond with macromolecules such as DNA, however, testing negative for mutagenic risk to the host. Finally, it was possible to classify the substance as belonging to toxicity class 6, where the calculated value of LD 50 > 5000 mg/kg suggests minimal risk of toxicity by ingestion.

Conclusion
Fisetinidol has shown promise in glycemic control and oxidative stress caused by hyperglycemia induced by a high concentration of sucrose in adult zebrafish. Molecular docking and dynamics study showed that compound has a greater affinity than acarbose and metformin, interacting in the same region with the same residues of these drugs, indicating that molecule maybe has the same inhibitory mechanisms on CtMGAM. The ADMET prediction results showed the fisetinidol has high passive permeability, low clearance and low toxic risk by ingestion. The results obtained confirm the ethnopharmacological information that different species of the genus Bauhinia can be used for their hypoglycemic effects as an adjuvant in the control of Diabetes Mellitus and other related syndromes of this pathology.