A Comprehensive Approach to Derivatization: Elemental Composition, Biochemical, and In Silico Studies of Metformin Derivatives Containing Copper and Zinc Complexes

The current study was designed to synthesize, characterize, and screen the molecular and biological activities of different metformin derivatives that possess potent antidiabetic potential with minimal side-effects. Metformin-based derivatives containing the metal complexes Cu II (MCu1–MCu9) and Zn II (MZn1–MZn9) were generated using aromatic aldehydes and ketones in a template process. The novel metal complexes were characterized through elemental analysis, physical state, melting point, physical appearance, Fourier-transform infrared (FTIR) spectroscopy, UV/visible (UV/Vis) spectroscopy, 1H nuclear magnetic resonance (NMR) spectroscopy, and 13C-NMR spectroscopy. Screening for inhibitory activity against the enzymes α-amylase and α-glucosidase, and molecular simulations performed in Schrödinger were used to assess the synthesized derivatives’ biological potential. Met1, Met2, Met3, and Met8 all displayed activities that were on par with the reference in an enzymatic inhibition assay (amylase and glucosidase). The enzyme inhibition assay was corroborated by molecular simulation studies, which also revealed a competitive docking score compared to the gold standard. The Swiss ADME online web server was utilized to compute ADME properties of metformin analogues. Lipinski’s rule of five held true across all derivatives, making it possible to determine the percentage of absorption. Metformin derivatives showed significant antidiabetic activities against both targeted enzymes, and the results of this work suggest that these compounds could serve as lead molecules for future study and development.


Introduction
In recent decades, Schiff bases have shown their importance in the industrial and pharmacological fields due to their novel properties. Derivatives of Schiff bases are also important in studying coordination chemistry. Synthesis of novel Schiff bases is of prime interest to scientists in the discovery of medicinal drugs [1,2]. Metformin is an oral drug used to reduce blood glucose levels in patients suffering from non-insulin-dependent diabetes mellitus (NIDDM) [3]. It improves insulin sensitivity, thereby lowering the resistance to insulin that is common in NIDDM [4]. Its history is mainly linked with a herbal plant (Galega officinalis also known as goat's rue) that is native to European plants and mainly used in lowering blood glucose levels; the plant extract is rich in guanidine. Different derivatives of guanidine (other than metformin) were mainly used between the 1930s as antidiabetic agents but discontinued in the market due to their toxic effects toward vital organs and the invention of insulin [5]. Therapeutic approaches to control hyperglycemia include increasing the level of insulin or decreasing insulin resistance [6]. Another pharmacological approach that is implemented in the clinics to manage postprandial hyperglycemia is the inhibition of the α-glucosidase enzyme [7]. However, some undesirable side-effects are also associated with the prolonged use of these inhibitors such as gastrointestinal side-effects, bloating, and diarrhea. Alternatives with fewer side-effects and cost-effective treatments are desired for the treatment of diabetes [8,9]. In the early 1940s, metformin was discovered and used as an antimalarial and to treat influenza, although it was observed during testing that it also causes a reduction in blood glucose levels. The antidiabetic use of metformin was first time reported by Jean Steme (a French physician) in 1957 [10].
The field of metallo-pharmaceutics has received extensive attention recently. This background has led to the development of a wide range of metal-containing drugs that exhibit notable antibacterial, anticancer, and antidiabetic properties [11][12][13]. Many researchers have synthesized different derivatives of metformin due to the increased glucose-lowering effect among others. Schiff bases are versatile and of a plastic nature; they form the majority of the ligand complexes and show a wide variety of biological effects. Metal complex chemistry containing Schiff-based ligands formed by aldehydes/ketones with amines exhibits significant biological effects, expressing the same features as metalloporphyrin with respect to the electronic and catalytic effects that mimic enzymatic effects. In the past few decades, the catalytic activity of metal complexes has been highlighted [14]. In the past few years, Schiff base metal complexes of metformin have revealed anticancer effects [15]. The strongest correlation with diabetes and cancer has been observed in breast, pancreas, and endometrial tumors [16]. Copper complexes of metformin have been synthesized, showing great levels of antidiabetic and antioxidant defense effects [17,18]. Various metals such as copper and zinc have been used to form metal complexes with metformin [19,20]. In this study, we generated different copper and zinc metal complexes with metforminbased derivatives synthesized using an aromatic aldehyde and ketone through a template reaction [21].

Characterization of Metabolites
The characterization schemes of all the metformin metabolites and metal complexes are presented in Schemes S1-S3 (Supplementary Materials). While, IR and NMR spectras are shown in Figures

In Silico ADME Studies
Swiss ADME can help to determine different physicochemical properties of novel tested compounds such as molecular weight, hydrogen bond donors (HBDs), hydrogen bond acceptors (HBAs), partition coefficient (log P). and solvent-accessible surface area within the desired range of acceptance, with the results depicted in Tables 1-3 and represented in Figures 1-3. The detailed ADME predictions are shown in Table S1 (Supplementary Materials).

α-Glucosidase
The α-glucosidase enzyme inhibition potential of metformin derivatives is presented in Tables 4-6. The graphical depiction of the results is shown in Figures S1-S3

Molecular Docking Studies
A total of 27 ligands were docked to the enzymes. The interactions of the receptors α-amylase and α-glucosidase with the ligands are shown in Figures 4-8, while the binding affinities and binding forces of the ligands and enzyme are reported in Tables 10 and 11.

Three-Dimensional Coordination Geometries
The 3D coordination geometries and poses were established to show the 3D coordination of all metformin derivatives and metal complexes with respect to different interaction sites (Table S4, Supplementary Materials).

Discussion
Most drugs fail at the clinical trial stage or are withdrawn from the market due to the negative properties of ADME (absorption, distribution, metabolism, and excretion) and unavoidable side-effects. The in silico prediction of ADME properties can reduce the effort and time needed by investigators to make models for every new derivative in the development of a lead compound [22]. Swiss ADME can help to determine the different physicochemical properties of novel tested compounds such as molecular weight, hydrogen bond donors (HBDs), hydrogen bond acceptors (HBAs), partition coefficient (log P), and solvent-accessible surface area within a desired range of acceptance. All novel derivatives of metformin had the ability to cross the gastrointestinal layer, and only one derivative (MET-7) could cross the blood-brain barrier (BBB). All novel derivatives exhibited log S under the predicted value (−6.5 to 0.5). All derivatives followed Lipinski's rule of five and were freely soluble in aqueous media. Therefore, according to the overall physicochemical prediction, most derivatives possessed drug-likeness behavior. No derivatives of copper and zinc could cross the blood-brain barrier, and they had low affinity for gastrointestinal absorption.
All derivatives showed a positive effect of enzyme inhibition on amylase and glucosidase. Met8 showed good enzyme inhibition when compared to the parent drug, as well as the standard. Met1, Met2, and Met3 showed comparable activity against glucosidase and amylase. The metal complexes with copper showed a linear graph, whereas zinc derivatives showed abrupt inhibition values.
Docking studies were performed to evaluate the binding interaction and binding energy between the receptors and small organic molecules. Docking results were further

Three-Dimensional Coordination Geometries
The 3D coordination geometries and poses were established to show the 3D coordination of all metformin derivatives and metal complexes with respect to different interaction sites (Table S4, Supplementary Materials).

Discussion
Most drugs fail at the clinical trial stage or are withdrawn from the market due to the negative properties of ADME (absorption, distribution, metabolism, and excretion) and unavoidable side-effects. The in silico prediction of ADME properties can reduce the effort and time needed by investigators to make models for every new derivative in the development of a lead compound [22]. Swiss ADME can help to determine the different physicochemical properties of novel tested compounds such as molecular weight, hydrogen bond donors (HBDs), hydrogen bond acceptors (HBAs), partition coefficient (log P), and solvent-accessible surface area within a desired range of acceptance. All novel derivatives of metformin had the ability to cross the gastrointestinal layer, and only one derivative (MET-7) could cross the blood-brain barrier (BBB). All novel derivatives exhibited log S under the predicted value (−6.5 to 0.5). All derivatives followed Lipinski's rule of five and were freely soluble in aqueous media. Therefore, according to the overall physicochemical prediction, most derivatives possessed drug-likeness behavior. No derivatives of copper and zinc could cross the blood-brain barrier, and they had low affinity for gastrointestinal absorption.
All derivatives showed a positive effect of enzyme inhibition on amylase and glucosidase. Met8 showed good enzyme inhibition when compared to the parent drug, as well as the standard. Met1, Met2, and Met3 showed comparable activity against glucosidase and amylase. The metal complexes with copper showed a linear graph, whereas zinc derivatives showed abrupt inhibition values.
Docking studies were performed to evaluate the binding interaction and binding energy between the receptors and small organic molecules. Docking results were further characterized with biological studies. All novel synthesized derivatives were simulated with the active pockets of α-glucosidase and α-amylase. After evaluating the docking scores (Tables 7 and 8), only four active compounds (Met1, Met2, Met3, and Met8) were selected for deep insight into different interactions.
All synthetic derivatives were analyzed by molecular simulation studies to identify binding interactions in the active pocket site of the enzyme. The radiographic structures of human pancreatic α-amylase (PDB: 5EOF) were selected as models for this purpose. The molecular docking results showed that several interactions were available between the derivatives and active pocket residues. The top four active derivatives were Met1, Met2, Met3, and Met8, showing multiple additional hydrogen bond hydrophobic interactions with residues Tyr62, Leu165, Lys200, and Ile235. In addition to the hydrophobic interactions, multiple hydrogen bonds were also observed. The metal bond order was considered as zero order [23]. Therefore, the force field method was able to adequately handle the coordinate bonds of metal complexes. All derivatives had a nitrogen moiety to build a hydrogen bond with residues Asp197, Glu233, and Asp300. However, an additional hydrogen bond was formed by the hydroxyl group of the derivatives with Trp59. Additional interactions such as salt bridges were seen in derivatives of positively charged nitrogen atoms on residues Asp197 and Glu233. Derivative Met1 exhibited pi stacking with the phenyl group of His201. Zinc complex derivatives showed different types of interactions in the pocket site such as hydrophobic interactions, pi-pi stacking, positive hydrogen bond, and metal bonding. MZn1, MZn2, MZn3, and MZn8 exhibited hydrophobic interactions with Val175, Ala169, and Trp134. MZn2 established a positive hydrogen bond with Trp134. Copper metal derivatives exhibited a positive hydrophobic interaction with copper metal, with the benzene ring being mainly responsible for the hydrophobic interactions. MCu1, MCu2, MCu3, and MCu8 established hydrogen bonds with copper via the active pocket residues Tyr131 and Tyr174. MCu8 exhibited a strong metal bond with copper metal and with residue Asp135.
The simulated binding of compounds resulted in several polar and nonpolar contacts with our target protein (2ZEO). Although the compounds were present in the active pocket site, some minor changes were observed. The nitrogen groups formed hydrogen bonds with active residue Asn61, Tyr63, Asp199, and Asp326, although the active pocket featured a few hydrophobic residues, e.g., Trp49, Val100, Asp199, Ala200, Leu285, and Val383. The novel derivatives established pi-pi stacking of Met1 and Met3 with the phenyl rings of residues Tyr63 and His325, as well as salt bridges between the positively charged nitrogen moieties present in the structure via residue Asp326, while Met2 formed a salt bridge with Arg411 of the hydroxyl group present in the structure.
In vitro studies showed that the Met8 derivative exhibited both amylase and glucosidase activity, while the docking studies revealed that derivative Met2 showed the highest docking scores for both enzymes. The α-glucosidase activity of derivative Met8 showed good activity in both in vitro and in silico docking studies. The metal complex with copper showed better activity than zinc complexes. The α-amylase activity results and docking studies revealed that MZn8 and MCu8 showed good activity.

Synthesis of Metformin Schiff Base
Metformin was dissolved in 10 mL of methanol, and the methanolic solution was reacted with aromatic aldehyde/ketone added dropwise. Drops of acetic acid were added, and the mixture was refluxed with continuous stirring over 4 h. Synthesized products were recrystallized by washing with methanol. Scheme 1 represents copper and zinc metal complexes with functional groups depicted in Table 12.

Copper Complexes of Metformin Derivatives (Met1-Met9)
The complexes were prepared via the reaction of CuSO 4 ·5H 2 O and ligands (Met1-Met9) in methanol. The methanolic solution (10 mL) of CuSO 4 ·5H 2 O (0.748 g; 3 mmol) was added dropwise to the methanolic solution (20 mL) of ligand (6 mmol) with constant stirring for 6 h. The solution was then concentrated to 15 mL volume and left to stand for 48 h. The precipitate was filtered with filter paper, and then dried in a desiccator for 24 h.

Zinc Complexes of Metformin Derivatives (Met1-Met9)
The complexes were prepared via the reaction of ZnSO 4 ·5H 2 O and ligands (Met1-Met9) in methanol. Methanolic solution (10 mL) of ZnSO 4 ·5H 2 O (0.748 g; 3 mmol) was added dropwise to the methanolic solution (20 mL) of ligand (6 mmol) with constant stirring for 6 h. The solution was then concentrated to 15 mL volume and left to stand for 48 h. The precipitate was filtered with filter paper, and then dried in a desiccator for 24 h.

In Silico ADME Prediction
Swiss ADME, an online ADME prediction tool, was applied in the present study to predict the drug likeness and physicochemical properties of novel derivatives of metformin [24]. The structure of derivatives was translated into canonical SMILES format and then submitted to the Swiss ADME tool, which provides free access to predict different properties of compounds, e.g., drug likeness, Lipinski's rule of five, gastrointestinal absorption, blood-brain barrier, and total polar surface area [25].

α-Amylase
All derivatives of metformin were analyzed for serum amylase inhibitory activity using the Yukihoko Hara method [26]. Three test tubes were labeled as blank (B), test (T), and control (C). Then, 2.5 mL of phosphate buffer (pH 6.8) was added to each test tube. Next, 1 mL of starch and 1 mL of 2 N sodium chloride solution were added to each test tube. The test tubes were incubated at 37 • C for 10 min. After incubation, 0.5 mL of derivatives was added to test tube T, followed by 0.2 mL of enzyme (α-amylase). All contents were mixed well, and then the tube was incubated at 37 • C for 10 min. In the control test tube, normal saline (2 N), starch solution (1 mL), and enzyme solution (0.2 mL) were added before incubating at 37 • C for 10 min. In the blank test tube, only 5.7 mL of distilled water was added. Dinitro-salicylic acid (0.2 mL) was added to all test tubes. The contents were mixed well, and the test tubes were kept in a boiling water bath for 15 min. The intensity of reddish orange was read at 540 nm. The percentage inhibitory action of serum amylase was calculated using the following formula: Percentage inhibition = OD of control − OD of test tube/OD of control × 100.

α-Glucosidase
The α-glucosidase-inhibitory activity of all derivatives of metformin was tested using the method previously described [27]. Three test tubes were marked as test (T), blank (B), and control (C). Then, 200 µL of substrate (starch) and enzyme (α-glucosidase) was added to test tubes T and C. The contents were mixed well, and then the test tubes were incubated at 37 • C for 30 min. After incubation, trichloroacetic acid was added to all test tubes, which were then centrifuged for 10 min. Free glucose was available after the reaction with anthrone reagent (1.5 mL). The contents in the test tubes were boiled for 15 min, and the absorbance was read at a wavelength at 640 nm. The percentage inhibition was calculated using the following formula: Percentage inhibition = OD of control − OD of test tube/OD of control × 100.

Molecular Docking
Docking studies of novel derivatives of metformin were performed using the Glide module of Schrodinger software 4.6 [28,29]. The selected target proteins (amylase and glucosidase) were downloaded from the Protein Data Bank (RCSB) [30]. Novel derivatives of metformin were drawn using Chemdraw12.0 PerkinElmer software.

Ligand Preparation
All ligands used in the docking studies were drawn in ChemDraw 12.0, and the structure was optimized for bond and structure alignment. Then, the structure was imported into the Schrodinger software workspace for energy minimization using the OPLS4 (version 9.03) (Optimized Potentials for Ligand Simulations) (44) force field in LigPrep wizard [31]. Maestro (12.8) was then used to hydrogens, assign bond orders, and convert the 2D ligand structure to 3D.

Protein Preparation
The Protein preparation wizard in Maestro was used to minimize the selected proteins from RCSB. All nonprotein molecules were removed without interfering with the already attached ligand, which became the basis for the grid box generation. As a first step, charges were equally distributed, and polar hydrogens were added using Epik at pH 7.0 ± 2.0. The heat state was generated, and the protein structure was refined. Water molecules were removed before the protein was subjected to docking analyses.

Receptor Grid Generation
The receptor grid was generated using the co-crystallized ligand. The ligand center was selected for grid box creation, and the van der Waals radius for the receptor was scaled at 1.00 Å with a partial charge of 0.25. For α-amylase (5E0F), the grid box coordinates were x −7.20, y 5.69, z −23.42 with a box size of 20 Å. For α-glucosidase (2ZE0), the grid box coordinates were x 10.39, y 4.11, z 16.74 with a box size of 20 Å.

Docking and Analysis
Docking studies were carried out using the Glide wizard with the proteins and ligands prepared as described above. In the final step of docking, in extra precision mode, the RMSD (root-mean-square deviation) value was restrained within 0.46 Å and the atomic charge was restrained within 0.15 for the receptor and all derivatives. When the docking process was complete, the interactions were observed using XP visualizer and Pymol (1.5.7).

Conclusions
This study focused on the synthesis, characterization, design, and screening of many metformin derivatives with the potential to effectively treat diabetes while causing only a small number of adverse effects. It was determined whether any of the compounds had the ability to inhibit the enzymes α-amylase and α-glucosidase. Only derivatives 1, 2, 3, and 8 exhibited any significant activity during the amylase and glucosidase tests. Each derivative included a benzene ring with a replacement. The presence of a substituted benzene ring in the meta position, a hydroxyl group, and an ether linkage in the para position provided derivative 8 with the maximum activity. This was also due to the presence of a hydroxyl group. When two benzene rings were included, a reduction in activity was observed, e.g., in derivatives 7 and 9. Docking experiments were conducted on all compounds against the human pancreatic amylase and glucosidase enzymes, and the results showed a variety of interactions, including hydrogen bonds, pi-pi stacking, and salt bridges. Accordingly, we were able to identify novel metformin derivatives with the potential to serve as lead compounds in subsequent studies for the development of effective inhibitors.

Data Availability Statement:
The data used to support the findings of this study are included within the article.