Thiocarbohydrazone and Chalcone-Derived 3,4-Dihydropyrimidinethione as Lipid Peroxidation and Soybean Lipoxygenase Inhibitors

Abstract

The potential of the 4,6-diphenyl-3,4-dihydropyrimidine-2(1H)-thione (abbreviated as KKII5) and (E)-N′-benzylidenehydrazinecarbothiohydrazide (abbreviated as DKI5) compounds as possible drug leads is investigated. KKII5 and DKI5 are synthesized in high yield of up to 97%. Their structure, binding in the active site of the LOX-1 enzyme, and their toxicity are studied via joint experimental and computational methodologies. Specifically, the structure assignment and conformational analysis were achieved by applying homonuclear and heteronuclear 2D nuclear magnetic resonance (NMR) spectroscopy (2D-COSY, 2D-NOESY, 2D-HSQC, and 2D-HMBC) and density functional theory (DFT). The obtained DFT lowest energy conformers were in agreement with the NOE correlations observed in the 2D-NOESY spectra. Additionally, docking and molecular dynamics simulations were performed to discover their ability to bind and remain stabile in the active site of the LOX-1 enzyme. These in silico experiments and DFT calculations indicated favorable binding for the enzyme under study. The strongest binding energy, −9.60 kcal/mol, was observed for dihydropyrimidinethione KKII5 in the active site of LOX-1. ADMET calculations showed that the two molecules lack major toxicities and could serve as possible drug leads. The redox potential of the active center of LOX-1 with the binding molecules was calculated via DFT methodology. The results showed a significantly smaller energy attachment of 2.8 eV with KKII5 binding in comparison to DKI5. Thus, KKII5 enhanced the ability of the active center to receive electrons compared to DKI5. This is related to the stronger binding interaction of KKII5 relative to that of DK15 to LOX-1. The two very potent LOX-1 inhibitors exerted IC₅₀ 19 μΜ (KKII5) and 22.5 μΜ (DKI5). Furthermore, they both strongly inhibit lipid peroxidation, namely, 98% for KKII5 and 94% for DKI5.

Introduction

Chalcones,^1,2 in particular, flavonoids, are natural products that display a broad spectrum of activities such as anticancer,³ antibacterial,⁴ antifungal,⁵ antimalarial,⁶ antidiabetic,⁷ anti-inflammatory,⁸ lipoxygenase inhibition,^9,10 and antioxidant.¹¹ Representatives of this class of compounds are already used in clinical practice or are undergoing clinical trials, including metochalcone,¹² sofalcone,¹³ isoliquiritigenin,¹⁴ xanthohumol,¹⁵ hesperidin, and methylchalcone.^16,17

Thiocarbohydrazones are an important, though less studied, class of molecules possessing biochemical, pharmaceutical, and industrial applications.¹⁸ Of utmost importance is their metal-chelating ability.¹⁹

3,4-Dihydropyrimidines are important intermediates in the synthesis of bioactive molecules like natural products,²⁰ functional materials like polymers,²¹ and adhesives.²² In particular, 4,6-diaryl-3,4-dihydropyrimidine-2(1H)-thione derivatives were recently proven to possess potent cytotoxic activity.²³⁻²⁵

Combining anti-inflammatory activity with antioxidant capacity^26,27 in one drug turns out to be important for some therapies, including for COVID-19.²⁸ Toward this end, in the present study, thiocarbohydrazone DKI5 and chalcone-derived 3,4-dihydropyrimidinethione KKII5 were rationally designed through molecular docking and dynamics in silico and toxicity predictions, synthesized and evaluated for their antioxidant capacity as well as anti-inflammatory²⁹ agents through their ability to inhibit soybean lipoxygenase.

Lipoxygenases (LOXs) constitute an important iron-containing enzyme family. They catalyze hydroperoxidation of polyunsaturated fatty acid derivatives, containing a cis,cis-1,4-pentadiene moiety (e.g., arachidonic and linoleic acids).^30,31 Their role in many acute and chronic diseases such as inflammation, cancer, asthma, allergy, and stroke has been well established.³² Due to their potential as therapeutic targets, there is great interest for the discovery and synthesis of novel and potent LOX inhibitors. LOX inhibitors exert their mechanism of action as (i) redox inhibitors or antioxidants by interfering with the redox cycle of hydroperoxidation, (ii) iron chelator agents that directly form a complex with iron core of the enzyme, and (iii) nonredox competitive inhibitors.³³ Most of the redox inhibitors exert their action by using their radical scavenging property.

The rational design was achieved through molecular docking, molecular dynamics, density functional theory (DFT) and in silico pharmacokinetics, pharmacodynamics, and toxicity studies. Their promising profile led to the synthesis and structure identification of their conformational properties. Finally, their in vitro activity was evaluated. This strategy led to the identification of two of the most potent soybean LOX-1 and lipid peroxidation inhibitors. To the best of our knowledge, this is the first study leading to a thiocarbohydrazone and a 3,4-dihydropyrimidinethione as LOX-1 and lipid peroxidation inhibitors. The steps described above are illustrated in Scheme 1.

Scheme 1. Overall Diagram of This Research Work.

Results and Discussion

Molecular Binding to the Active Site of Soybean LOX-1

Induced fit docking was utilized to study the possible binding mode of the two compounds in the active site of LOX-1. DKI5 binds strongly to the active site of LOX-1. The ΔG_bind value has been calculated to be −7.14 kcal/mol. The molecule forms three hydrogen bonds with the residues of the cavity, which contributes to the strength of the binding. In particular, two H-bonds are formed between the hydrogen atoms of the thiourea functional group with GLN495, and one H-bond is observed with one of the hydrogen atoms of the amine functional group and the water molecule in the first coordination sphere of an Fe cation. In addition, a π–π stacking interaction is observed between the aromatic ring of the molecule and indole group of TRP500 (Figure 1).

Figure 1.

Interactions of DKI5 with LOX-1 in 3D (left) and 2D (right) docking poses.

In addition, KKII5 binds strongly to the active site of LOX-1 indicating ΔG value of −9.60 kcal/mol. KKII5 interacts with HIS499 of the cavity by forming one π–π stacking interaction between the phenyl ring of the molecule and the imidazole ring of HIS499. Furthermore, two aromatic hydrogen bonds are formed between H9 and H12 (see Figure 2) with the residues GLN495 and ILE839, respectively.

Figure 2.

Interactions of KKII5 with LOX-1 in 3D (left) and 2D (right) docking poses.

Molecular Dynamics

Molecular dynamics simulations have been applied to assess the stability of the formed complexes and predict a more accurate conformation that the two aforementioned molecules adopt inside the active site of LOX-1. The docking poses described above (docking poses) have been used as the initial systems for the MD simulations.

As described in the diagram above, both enzyme–inhibitor simulations have been converged, since it is observed that, after the first 30 ns, the RMSD (root mean square deviation) of the protein does not present any significant changes until the completion of the simulation. In particular, the protein in the “enzyme–KKII5” complex has been stabilized at ∼2.11 Å, while the “enzyme–DKI5” complex has been stabilized at ∼1.97 Å (Figure 3).

Figure 3.

RMSD of the protein for the “enzyme–KKII5” complex (black line) and for the “enzyme–DKI5” complex (purple line) throughout the whole simulation (200 ns). The RMSD value was calculated based on the initial docking pose (first snapshot of the MD), which is used as a reference snapshot.

Molecular Dynamics Simulations of the “Enzyme–KKII5” Complex

According to the MD studies, KKII5 remains bound to the active site of the enzyme during the production time. The conformations that the inhibitor adopts inside the cavity do not present significant changes in comparison to the initial docking pose (reference pose), and the orientation of the molecule remains similar to the first MD snapshot. In particular, the first phenyl ring is oriented toward HIS499, while the second is toward TRP500. Moreover, the key interactions that are formed between the molecule and the residues of the cavity were preserved from the docking studies throughout the whole simulation. The π–π stacking interaction that was formed between the aromatic ring and the imidazole group of HIS499 in the first coordination sphere of iron was preserved for 31% of the simulation time, while the same type of interaction observed between the second phenyl ring and the indole group of TRP500 was for 55% of the simulation time. Furthermore, one hydrogen bond between the hydrogen atom of the thiopyrimidine ring and the oxygen atom of LEU546 was also preserved for 19% of the simulation time and another H-bond between the second hydrogen of thiopyrimidine ring and the oxygen atom of GLN495 (36% of the simulation time) (Figure 4).

Figure 4.

(Left) Protein–ligand interaction fraction shown by bars. The values are normalized over the trajectory period of 200 ns. (Right) Protein–ligand interactions. H-bonds are illustrated with purple dotted lines and π–π interactions with a solid green line.

To depict these results and present a more accurate pose of the ligand inside the LOX’s cavity, a trajectory cluster analysis was performed based on ligands’ conformations in the active site. Hence, the conformation that governed during the simulation in terms of time lasted for 49% of the simulation time. This statistically predominant conformation is illustrated in the figure below, and the RMSD calculated between the pose that derived from cluster analysis in comparison to the pose derived by docking calculations is equal to 1.79 Å (Figure 5).

Figure 5.

Three-dimensional pose of the active site of the “LOX–KKII5” complex derived after trajectory cluster analysis based on the ligands’ conformations. This pose was adopted for 49% of the simulation time (200 ns).

Molecular Dynamics Simulations of the “Enzyme–DKI5” Complex

The inhibitor remains bound to the active site of the enzyme during the whole simulation time. The conformations that the inhibitor adopts inside the cavity do not change significantly in comparison to the initial docking pose regarding the orientation of the molecule, and the phenyl ring is oriented toward TRP500, while the diazole group is oriented toward the residues in the iron’s first coordination sphere. Moreover, the key interactions that were observed during the docking studies between the molecule and the residues of the cavity were preserved in the MD simulation. The only minor change that was observed is that the whole molecule was slightly displaced away from the iron in the pose derived by the MD studies. The amine group is now in vicinal proximity with HIS499, in comparison to its proximity to the water molecule from the docking pose. In particular, the observed π–π stacking interaction between the phenyl ring and the indole group of TRP500 was present for 85% of the simulation time. Furthermore, two hydrogen bonds between the hydrogens of the thiopyrimidine ring and the oxygen atom of GLN495 were present for 97 and 54% of the simulation time, respectively. In addition, the amine group of the molecule forms one H-bond with GLN495 (51% of the simulation time). Finally, an additional hydrogen bond between the hydrogen of the pyrimidine group and HIS499 is observed (Figure 6).

Figure 6.

(Left) Protein–ligand interaction fraction shown by bars. The values are normalized over the trajectory period of 200 ns. (Right) Protein–ligand interactions. H-bonds are illustrated with purple dotted lines and π–π interactions with a solid green line.

We performed trajectory cluster analysis based on the ligands’ conformations to depict these results and present a more accurate pose of the ligand inside the LOX’s cavity after the MD studies. In particular, the conformation that governed the simulation in terms of time lasted for 32% of the whole simulation time. This statistically predominant conformation is illustrated in the figure below, and the RMSD calculated between the pose that was derived from cluster analysis in comparison to the pose derived by docking calculations is equal to 0.009 Å (Figure 7).

Figure 7.

Three-dimensional pose of the active site of the “LOX–DKI5” complex derived after trajectory cluster analysis based on ligands’ conformations. This pose was adopted for 32% of the simulation time (200 ns).

MM/GBSA Calculations

MM/GBSA calculations were performed to the most favored energetic complexes extracted from Desmond trajectory clustering. MM/GBSA calculations are considered to be an adequate method to rank compounds targeting the same enzyme and to discover the key interactions that govern the binding of the compounds to the active site. In the table below, the binding energies of the two molecules under study are illustrated. According to the ΔG_bind values shown in the table, both compounds bind strongly to the active site of soybean LOX-1, which complements the results that have been already obtained by docking and MD simulations. In particular, for the most prominent cluster of the “LOX-1–KKII5” complex (49% of the simulation time), ΔG_bind is calculated to be −39.58 kcal/mol, while the most prominent cluster of the “LOX-1–DKI5” complex (32% of the simulation time), the ΔG_bind value is a little lower (−37.81 kcal/mol). Hence, we could conclude that the inhibitors present binding with similar strength to the active site of LOX-1, and thus we expect they will also illustrate similar IC₅₀ values in the in vitro experiments as it is observed.

Table 1. ΔG_bind Values Derived from MM/GBSA Calculations Based on the Statistically Predominant Cluster of MD Simulations.

	ΔGbind (kcal/mol)	ΔGbind Coulomb	ΔGbind H-bond	ΔGbind Lipo	ΔGbind packing	ΔGbind Solv GB	ΔGbind vdW
KKII5	–39.58	–9.31	–0.62	–28.76	–3.13	54.54	–52.31
DKI5	–37.81	–17.21	–1.11	–16.28	–1.13	35.36	–37.46

Although the two compounds have similar ΔG_bind values, they show considerable different interactions. DK15 shows increased Coulomb interactions presumably due to the additional amino group in its structure. KK15 is apparently more lipophilic (logP = 3.01 average of the five values using SwissADME) than DK15 (logP = 1.12 average of the five values using SwissADME) and shows stronger lipophilic and van der Waals interactions.

ADMET Calculations

The possible toxicity was predicted for both molecules under study by using SwissADME,³⁴ pkCSm,³⁵ and preADMET³⁶ computational tools (platforms). The compounds were predicted as nontoxic, and their structural properties reveal that they could serve as prominent drugs. For that reason, we proceeded to further analyze both (Tables 1S, 2S, and 3S).

Synthesis

The synthesis of studied compounds is illustrated in Schemes 1 and 2. Benzaldehyde reacted with thiocarbohydrazide in warm ethanol to give thiocarbohydrazone derivative DKI5 (Scheme 2).

Scheme 2. DKI5 Synthesis.

3,4-Dihydropyrimidine KKII5 was prepared by cyclocondensation of chalcone KKI1 (obtained from the reaction of benzaldehyde with acetophenone) with thiourea in the presence of NaOH with ultrasonication at room temperature (Scheme 3).

Scheme 3. KKII5 Synthesis.

Structure Assignment

Scheme 4 shows the two compounds under study. The numbering will be used for the NMR structural assignment.

Scheme 4. Structures of (Left) DKI5 and (Right) KKII5 Compounds.

The structure assignment of DK15 was initiated using the readily assigned H-6, which resonates at 8.01 ppm. Its signal gives integration of two protons, and it was the most shielded protons. Through 2D HSQC (Figure S3A), the H-6 shows ¹J_C–H coupling with the C6, and therefore, C6 is assigned unambiguously at 143 ppm. Η-8 is identified easily through 2D-NOESY (Figure S1A) due to its spatial proximity with H-6. Through 2D-NOESY, H-9 and H-10 are identified at 7.40 and 7.42 ppm. Protons attached with nitrogen are identified at 4.88, 9.82, and 11.43 ppm. Through a two-dimensional heteronuclear single quantum coherence (HSQC) spectrum of KKI15, all of the carbons bearing protons were identified. The quaternary and carbonyl carbons were identified through 2D-heteronuclear multiple bond coherence (HMBC) (Figure S4A). Specifically, H-9 shows a ³J_C–H coupling with C-7, and H-2 shows a ²J_C–H coupling with C-3.

The experimental data were carried out in DMSO solvent as it simulates the amphoteric environment and is suitable for the observation of NOE effects.^37,38 The same procedure was performed for KKII5.

Using ¹H–¹⁵N (HSQC) and (HMBC)³⁹ experiments, we identified all of the nitrogen atoms on both molecules (Figures S1C, S2C and S3C).

The identification strategy is shown in the Supporting Information (Figures S1D and S2D). Proton spectra of ΚΚII5 and DKI5 are shown below (Figure 8). Furthermore, Table 2 with the chemical shifts of these compounds is shown below.

Figure 8.

¹H NMR spectra in DMSO-d₆ of DKI5 (a) and KKII5 (b). The spectra were recorded on a Bruker AC 400 MHz spectrometer at 25 °C.

Table 2. Assignment of the Experimental ¹H NMR Spectra of DKI5 (Top) and KKII5 (Bottom) in DMSO-d₆.

position (DKI5)	1Η	NOESY	HSQC	HMBC	13C
1	4.88	H-2/H-4/H-6/H-8/H-9/H-10
2	9.82	H-1/H-4		C-3
3				H-2/H-4	176.11
4	11.43	H-1/H-2/H-6		C-3/C-6
5
6	8.015	H-1	C-6	H-4/H-8/C-7/C-8/C-9	143
7				H-6/H-8/H-9	135
8	7.83		C-8	H-6/H-9/H-10/C-6/C-7/C-9/C-10	127.85
9	7.4		C-9	H-6/H-8/H-9/H-10/C-7/C-8/C-9/C-10	128.9
10			C-10	H-8/H-9/C-8/C-9	130

position (KKII5)	1Η	NOESY	HSQC	HMBC	13C
1				H-14/H-2	175.63
2	9.65	H-4		C-1/C-3
3				H-2	134.82
4				H-5/H-6	133.79
5	7.32	H-8	C-5	C-4/C-6	128.50
6	7.32	H-5/H-4	C-6	H-7/H-5	128.50
7	7.32	H-6	C-7	H-6/H-5	128.50
8	5.22	H-5/H-11	C-8	H-9	101.65
9	4.93	H-11	C-9	C-8/C-10	55.10
10				H-11/H-8	144.55
11	7.16	H-9	C-11	C-12/C-13	126.35
12	7.16	H-11/H-13	C-12	C-13/C-11	126.35
13	7.16	H-12	C-13	C-12/C-11	126.35
14	8.91	H-9		C-1/C-9

DFT Conformational Analysis of DKI5 and KKII5

The lowest in energy structures of DKI5 (a) and KKII5 (b) are shown below. As these are in harmony with 2D-NOESY spectra, thye are considered as the most probable conformations for KKII5 and DKI5 (see Figure 9).

Figure 9.

Lowest in energy conformations derived from DFT calculations for (a) DKI5 and (b) KKII5.

The characterization of the compounds by ¹H NMR and ¹³C NMR indicates one predominant isomeric form in DMSO for each compound. The planes of phenyl groups Β και C in KKII5 are perpendicular to each other, and A and B are coplanar to each other, having the potential to form a better interaction with the active center of LOX-1.

The spatial correlations that are observed in 2D-NOEs are shown in Figure 10.

Figure 10.

Correlations that were observed in 2D-NOESYs for (a) DKI5 and (b) KKII5.

Redox Potential

Τhe binding of DKI5 and KKII5 in the active site of LOX-1 (see Figures 5 and 7) was calculated via DFT (B3LYP/6-311+G(d,p)) methodology. The redox potential of the LOX-1 active center with the two binding molecules was calculated.

Charges on iron were calculated with Mulliken and NBO analyses. Both predict the same charges (see Table 3). DFT calculations showed that KKII5 interacts strongly with the amino acids in the active center since it forms smaller bond distances than DKI5: 1.91 Å for KKII5 in contrast to 1.95 Å for DKI5. A result of the stronger (shorter) bonds there is an enhanced charge transfer from KKII5 to Fe³⁺ compared to that with DKI5. Note that about 0.9 electrons are transferred from ligands to Fe³⁺ when DKI5 is binding, while 1.7 electrons are transferred from ligands to Fe³⁺ when KKII5 is binding. On the contrary, similar charge transfer occurred from ligands to Fe²⁺, i.e., 0.6 e^– (DKI5) and 0.8 e^– (KKII5). It is interesting that the binding of the molecules significantly affects the redox potential of the active center of LOX-1. The energy gain for the electron attachment when KKII5 is binding is significantly smaller by 2.8 eV than that when DKI5 is binding. This means that KKII5 enhances the ability of Fe³⁺ to receive electrons compared to DKI5; i.e., the active center is further stabilized. This is related to the strong binding interaction of KKII5 to LOX-1.

Table 3. Charges on Fe and Redox Potential for Binding DKI5 and KKII5.

DKI5	Mulliken	NBO	Εea(Fe3+ → Fe2+) = –ΔΕ(attach)
Fe3+	2.05	2.09	9.6 eV
Fe2+	1.47	1.43

KKII5	Mulliken	NBO	Εea(Fe3+ → Fe2+) = –ΔΕ(attach)
Fe3+	1.32	1.32	6.8 eV
Fe2+	1.41	1.21

In Vitro Biological Assays

Soybean isozyme LOX-1 was used in our in vitro experiments that reveals homology to mammalian lipoxygenase.^40,41 This enzyme uses free fatty acids as substrates⁴² and presents maximal activity at pH 9.0. This is achieved by producing a conjugated diene absorbing at 234 nm attributed the conversion of linoleic acid into to 13-hydroperoxylinoleic acid.⁴³ Both tested biologically compounds present moderate inhibitory activity against the enzyme of LOX-1. The obtained results were compared with the reference compound nor-dihydroguaeretic acid (NDGA) (Table 4). The calculated IC₅₀ values of both compounds are lower in comparison to those of the compound NDGA. Thus, both compounds appear to act as less potent inhibitors. LOXs contain a “non-heme” iron per molecule in the enzyme active site. Iron has a high-spin Fe²⁺ in the native state and high-spin Fe³⁺ in the activated state. A relationship between LOX inhibition and the ability of the inhibitors to reduce Fe³⁺ at the active site to the catalytically inactive Fe²⁺ has been suggested by some studies. This inhibition is related to their ability to reduce the iron species in the active site to the catalytically inactive ferrous form. Furthermore, several LOX inhibitors have been found to be excellent ligands for Fe³⁺.⁴⁴

Table 4. In Vitro Inhibition of LOX as IC₅₀ Values and % Inhibition of Lipid Peroxidation (%ILP) at 100 μΜ.

code	LOX IC50 (μΜ)	% ILP
DKI5	22.5	94
KKII5	19	98
NDGA	0.45
Trolox		92

2,2-Azobis(2-amidinopropane) hydrochloride AAPH, a water-soluble azo compound, was utilized as a free radical initiator to follow oxidative changes of linoleic acid to conjugated diene hydroperoxide. In the AAPH assay, the highly reactive alkylperoxyl radicals are intercepted mainly by a hydrogen atom transfer (HAT) from the antioxidant.⁴⁵ Effective HAT agents are compounds with high hydrogen atom donating ability. More specifically, these are molecules that show low heteroatom–H bond dissociation energies or molecules from which hydrogen abstraction leads to sterically hindered radicals as well as molecules from which abstraction of hydrogen leads to C-centered radicals stabilized by resonance. We compared the antilipid peroxidation activities of both compounds in vitro in comparison to a well-known antioxidant, i.e., Trolox (Table 4). The experimental conditions resemble cellular lipid peroxidation due to the activity of the undertaken radicals. Both tested compounds present high antilipid peroxidation activity compared to the reference compound Trolox. It appears that LOX inhibition is related with the antilipid peroxidation activity. Lipophilicity does not appear to affect the antioxidant activity.

Both compounds significantly inhibit LOX. Between them, KKII5 was more potent followed by DKI5. The higher lipophilicity of KKII5 compared to that of DKI5 led to higher inhibitory activity. The reported literature confirms the positive role of lipophilicity for LOX inhibitors.⁴⁶

The majority of the literature records that LOX inhibitors are antioxidants, inhibitors of lipid peroxidation, or free radical scavengers.⁴⁶

Experimental Section

General

The reagents are of highest purity from Aldrich, Fluka, and Acros, greater than 98%. Thin layer chromatography was performed on 0.25 mm silica gel plates (E. Merck silica gel 60F254), and the plates were UV visualized. NMR spectra were recorded on a Bruker 400 Avance instrument. A MSQ 400 Surveyor Finnigan instrument was used for mass spectra.

Induced Fit Docking

Induced fit docking has been performed to discover the possible binding of the two molecules (DKI5 and KKII5) in the active site of LOX-1. This enzyme has been selected as a possible target since chalcone and thiocarbohydrazone substructures could serve as potent inhibitors of the enzyme, according to our bibliographic research.^47,48 Furthermore, the Swiss Target webtool (http://www.swisstargetprediction.ch/) revealed LOX as a potent target for both two molecules. The crystal structure used for the in silico studies was PDB ID 5T5V.⁴⁹ The in silico calculations were performed using protein preparation wizard module available in Schrödinger Suites to prepare the crystal structure. Since LOX is a metalloprotein, several computational challenges had be addressed. Thus, the quantum effects associated with the presence of an Fe³⁺ cation in LOX’s active site were determined with the “create zero-order bonds to metals” module of the Schrödinger’s Maestro molecular modeling platform. In this module, the bonds to metals are broken and new zero-order bonds between metals and nearby atoms are added. Furthermore, the formal charges are corrected to constrain the X-ray acquired coordination geometry.

The two compounds were sketched in Schrödinger’s Maestro⁵⁰ molecular modeling platform and minimized using Macromodel⁵¹ and DFT calculations. LigPrep was used to prepare the 3D models, taking into account their stereochemistry features. During the ligand preparation, the “add metal binding states” option of the Epik module of LigPrep was chosen. If this option was not selected, the binding would fail as it would be characterized by high energy state penalties. The geometries were optimized with MacroModel and during the structure relxation process chiral centers were checked that retained their proper chiralities. The force field OPLS2005⁵² was used for minimization and proper treatment of their protonation states at physiological pH (∼7.4). Hammett and Taft methods have been implemented in conjunction with an ionization tool to generate chemically sensible 3D models. Ligand structures were furthermore rigorously minimized by MacroModel with water as the solvent and OPLS2005 as the force field, using a conjugate gradient (CG) method with a threshold of 0.01 kcal/mol. The derived minimized structures were used as input to a mixed-torsional/low-sampling conformational search forced to avoid the modification of the input chiralities. Conformational search generated several conformers for each molecule that were energetically ranked. The most favored conformation was used as input for the docking calculations.

The calculations were performed using the induced fit docking (IFD) method available in Schrödinger Suites. The ligand was docked in the 5 energetically favored conformations generated by Macromodel. Prior to the free binding calculations, protein preparation constrained refinement was applied in the Glide docking stage. Trimming side chains automatically (based on B-factor) and Prime refinement of the protein side chains were applied, and the docking process was accomplished by Glide/XP. During the docking process, the dielectric constant of the active site was set to 80 and the waters in the crystallographic data were preserved.

Molecular Dynamics

The MD simulation system was set up with SPC/E modeled waters surrounding the drug–protein complex and neutralized with Na⁺ and Cl^– ions until the experimental salt concentration of 0.150 M NaCl was reached. The N-terminus of the protein was capped by an acetyl group, whereas the C-terminus remained uncapped since it is part of the protein’s active site. The OPLS2005 force field was used to model all protein–ligand interactions, and the long-range electrostatics were treated with the particle mesh Ewald method (PME)^53,54 and a grid spacing of 0.8 Å. van der Waals and short-range electrostatic interactions were smoothly truncated at 9.0 Å. During the calculations, the temperature was kept constant using the Nosé–Hoover thermostat,⁵⁵ while the Martyna–Tobias–Klein method⁵⁴ was used to control the pressure. Calculations were performed using periodic conditions and a simulation box with dimensions of 10.0 × 10.0 × 10.0 A. The equations of motion were integrated using the multistep RESPA integrator⁵⁶ with an inner time step of 2 fs for bonded interactions and nonbonded interactions within a cutoff of 9 Å. An outer time step of 6.0 fs was used for nonbonded interactions beyond the cutoff. Each system was equilibrated using the default protocol provided by Desmond.⁵⁷ The system was relaxed initially with Brownian dynamics simulation in the NVT ensemble at T = 310 K with restraints on solute heavy atoms. Before starting with the production phase, the system was left to relax in the NPT ensemble with no restraints for 1.0 ns. The production phase of the MD simulation was set to 200 ns. This time is sufficient to obtain adequate sample size to analyze the complexing of the drug in the enzyme active site cavity.

The MD simulations were run in workstations using the GPU implementation of the MD simulations codes and the statistics of simulation were evaluated based on the RMSD convergence of the protein backbone Cα atoms and the RMSD of the ligand.

All MD simulations were performed three times to verify the reproducibility of the results, and the trajectory was analyzed with the Desmond and VMD trajectory analysis tools.

MM/GBSA Calculations

MM/GBSA was used to determine free binding energy of the complexes of the two molecules with their target proteins. Specifically, the most favored interactions contributing more to the formation of the complexes were calculated using the Prime module of Maestro. In particular, the MM/GBSA calculations were applied to three most favored energetically ligand–protein complexes extracted from MD trajectory cluster analysis. VSGB solvation was used to describe protein flexibility.⁵⁸

Synthesis

(E)-N′-Benzylidenehydrazinecarbothiohydrazide, DKI5(59)

To a stirred solution of benzaldehyde (0.14 mL, 1.4 mmol) in ethanol 70% (90 mL) was added thiocarbohydrazide (0.29 g, 2.8 mmol) followed by the addition of 1 drop of acetic acid. The reaction mixture was heated to 80 °C for 4 h. Then the reaction mixture was left to stand at room temperature, filtered, washed with ethanol/water 1:1, and recrystallized from ethanol to give DKI5 as an off-white solid in 75% yield. R_f (PE/EtOAc: 2/1) = 0.73.

4,6-Diphenyl-3,4-dihydropyrimidine-2(1H)-thione, KKII5(60)

To a stirred solution of NaOH (68 mg, 1.70 mmol) in EtOH (3.5 mL) were added E-chalcone (0.35 g, 1.70 mmol) and thiourea (0.13 g, 1.70 mmol), and the mixture was sonicated for 1 h at room temperature. Then it was poured in ice–water, and the solid formed was filtered, washed with H₂O, EtOH/H₂O 1:1, and dried to give KKII5 as an off-yellow solid in 97% yield. R_f (PE/AcOEt 9.5/0.5) = 0.71.

Structure Assignment

The two molecules have been identified using a 400 MHz NMR Bruker Avance spectrometer operated at 25 °C. The pulse sequences used were obtained from the library of the spectrometer. Data processing including sinebell apodization, Fourier transformation, phasing symmetrization, and plotting were performed using TopSpin Bruker Data Analysis and MestreNova⁶¹ software packages.

Quantum Mechanics Calculations: DFT Calculations

Conformational analysis of the DKI5 and KKII5 molecules was performed employing DFT⁶² at the B3LYP^63,64/6-311+G(d,p)⁶⁵ level of theory. All compounds were optimized to find the lowest minima. All calculations were performed using DMSO solvent employing the polarizable continuum model (PCM).⁶⁶⁻⁶⁸

Furthermore, the binding of DKI5 and KKII5 in the active site of LOX-1 was calculated via DFT(B3LYP/6-311+G(d,p)) methodology. The geometries of the LOX-1 active center with the binding molecules, obtained via MD calculations, were used as the initial guess for the DFT calculations. The complexes included about 200 atoms, and they were geometry optimized considering iron is complexed with two oxidation states +2 and +3 to check the effect of the oxidation state on the binding, the electron transfer, and the redox potential of LOX-1. The charges of the iron center and ligands were calculated using two population analyses, i.e., Mulliken and NBO.⁶⁹ Calculations and visualization were carried out via Gaussian 16.⁷⁰

In Vitro Studies

UV–vis spectra, for the in vitro tests, were obtained on a Shimadzu 1700Spec UV double-beam spectrophotometer (PerkinElmer Corporation Ltd., Lane Beaconsfield, Bucks, UK). Soybean lipoxygenase, sodium linoleate, and 2,2-azobis(2-amidinopropane)dihydrochloride (AAPH) were purchased from Sigma Chemical, Co. (St. Louis, MO, USA).

The in vitro assays were performed at a concentration of 100 μM (a 10 mMstock solution in DMSO was used, from which several dilutions were made). The IC₅₀ values of the samples studied were calculated using at least triplicates, and the standard deviation of absorbance did not exceed 10% of the mean. The compounds prior to be recorded were diluted in 0.1% DMSO under sonification in an appropriate buffer in several dilutions. Statistical comparisons were made using the Student t test. A statistically significant difference was defined as p < 0.05.

Inhibition of Linoleic Acid Peroxidation

In vitro study was performed as described by Hadjipavlou et al.⁴⁸ Ten microliters of the 16 mM sodium linoleate solution were added to the UV cuvette containing 0.93 mL of a 0.05 mM phosphate buffer, pH 7.4, and prethermostated at 37 °C. Addition of 50 μL of a 40 mM free radical initiator AAPH solution initiated at 37 °C the oxidation reaction. Oxidation was monitored at 234 nm and was performed in the presence of aliquots (10 μL) without the use of antioxidants. Lipid oxidation was achieved and recorded in the presence of the same level of DMSO which served as a negative control. Trolox was used as the appropriate standard (positive control).

Soybean Lipoxygenase Inhibition Study

The in vitro study was performed as described by Hadjipavlou-Litina et al.⁴⁸ The compounds were incubated at room temperature with sodium linoleate (0.1 mM) and 0.2 mL of enzyme solution (1/9.10^–4 w/v in saline), and the conversion of sodium linoleate to 13-hydroperoxylinoleic acid at 234 nm was calculated quantitatively. Nor-dihydroguaeretic acid NDGA (IC₅₀ = 0.45 μM) was used as a standard (positive control). To determine their IC₅₀ values we have used six ascending concentrations (0.01–100 μM). To assess the repeatability and accuracy of the experiments they were repeated with the same experimental conditions in six duplicate replicates. The calculated results are reported from the mean of the six experiments. It was found that the standard deviation was less than 10% (a = 0.01). Prior to the experiments, a blank determination was used as a negative control.

Conclusions

Two thiosemicarbazone analogues, the 4,6-diphenyl-3,4-dihydropyrimidine-2(1H)-thione (KKII5) and (E)-N′-benzylidenehydrazinecarbothiohydrazide (DKI5), were synthesized here in high yield up to 97% using known methodology. Their unambiguous structures were studied utilizing 1D and 2D homonuclear and heteronuclear NMR, 2D-NOESY, and DFT methodology.

The low energy conformers of the two compounds have been used for molecular docking experiments in LOX-1 enzyme. The promising favored binding energies of both molecules triggered our interest to evaluate them with an in vitro test. MD simulations showed that both molecules are stabilized in their binding cavity. The experimental results were also promising, showing that both compounds have considerable inhibitory activity (IC₅₀ was 22.5 μΜ for DKI5 and 19.5 μΜ for KKII5).

The redox potential of the active center of LOX-1 with the binding molecule was calculated via DFT methodology. It is interestingly found that the energy gain for the electron attachment in the active center of LOX-1 is significantly smaller when KKII5 is binding than when DKI5 is binding, by about 3 eV, meaning that KKII5 enhanced importantly the ability of the active center to receive electrons comparing to DKI5. This is related to the strong binding interaction of KKII5 to LOX.

Finally, our ADMET studies showed that the two molecules lack of considerable toxicity and can be used for future studies as drug leads.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.2c07625.

• 2D-NOESY-NMR spectra of DKI5, 2D-NOESY-NMR spectra of KKII5, 2D HSQC ¹⁵N–¹H spectra of KKII5, and structure assignment of DKI5 (Figure S1); spectra were recorded in DMSO-d₆ on a Bruker AC 400 MHz spectrometer at 25 °C; ¹³C NMR spectra of DKI5, ¹³C spectra of KKII5, 2D-HMBC ¹⁵N–¹H spectra of DKI5, and structure assignment of KKII5 (Figure S2); spectra were recorded in DMSO-d₆ on a Bruker AC 400 MHz spectrometer at 25 °C; 2D-HSQC-NMR spectra of DKI5, 2D-HSQC spectra of KKII5, and 2D-HSQC ¹⁵N–¹H spectra of DKI5 (Figure S3); spectra were recorded in DMSO-d₆ on a Bruker AC 400 MHz spectrometer at 25 °C; 2D-HMBC-NMR spectra of DKI5 and 2D-HMBC spectra of KKII5 (Figure S4); spectra were recorded in DMSO-d₆ on a Bruker AC 400 MHz spectrometer at 25 °C (PDF)

The authors declare no competing financial interest.