In-silico assay of a dosing vehicle based on chitosan-TiO₂ and modified benzofuran-isatin molecules against Pseudomonas aeruginosa

Geometrical optimization

The structures of the studied molecules see Fig. 1, were optimized using the density functional method (DFT) (Arai et al., 2007), though the software Gaussian (G09) (Rappoport & Furche, 2010), using the experimental-correlated generalized gradient approximation hybrid functional proposed by Zhao & Truhlar, (2008) (M06-2x), with a Pople basis set with polarizing and diffuse functions (6-31+g (d, p)) (Yang, Parr & Pucci, 1984; Geerlings et al., 2014). The frequencies of all optimized molecules were calculated to verify that they were a minimum at the potential energy surface (PES).

Fukui condensed functions

Fukui condensed functions were used (Fukui & Pullman, 1980), Eqs. (1)–(3), to calculate the reactivity index of the molecules, as well as to know the reactive sites (areas where the molecules would interact more easily with another). These functions are based on the formalism of the conceptual DFT; in the present work, we obtained them from the analysis of natural orbitals (NBO) (Fukui & Pullman, 1980; Yang, Parr & Pucci, 1984; El Ouafy et al., 2020).

(1) $${\mathit{f}}_{\mathit{j}}^{+}\equiv {\mathit{q}}_{\mathit{j}}\left(\mathit{N}+1\right)-{\mathit{q}}_{\mathit{j}}\left(\mathit{N}\right)$$

(2) $${\mathit{f}}_{\mathit{j}}^{-}\equiv {\mathit{q}}_{\mathit{j}}\left(\mathit{N}\right)-{\mathit{q}}_{\mathit{j}}\left(\mathit{N}-1\right)$$

(3) $${\mathit{f}}_{\mathit{j}}^0\equiv {\displaystyle \frac{{\mathit{q}}_{\mathit{j}}\left(\mathit{N}+1\right)-{\mathit{q}}_{\mathit{j}}\left(\mathit{N}-1\right)}{2}}$$

Molecular docking and pharmacophoric model

The binding interactions between ligands (studied molecules) and a macromolecular target (typically proteins), whose structure is experimentally known, were analyzed. The technique involves locating the most likely binding between the ligand and the receptor with lower energy (the lower the energy, the stronger the binding) and the suitable molecular binding site (Huggins et al., 2019; Lin, Li & Lin, 2020).

The targets were selected considering their biochemical importance of them when interacting with ligands, promoting bacterial inhibition. The cartesian coordinates of the targets (Hfq and LpxC) were obtained from the protein data bank with PDB codes: 5I21 (Roy et al., 2010; Lekontseva et al., 2020) and 4J3D (Hale et al., 2013), respectively.

Water molecules and coupled ligands were deleted for both proteins, and only the cofactor responsible for their antibiotic action was taken as a reference. For the calibration of each protein the quadratic deviation values (RMSD) were evaluated and considered an RMSD less than 1 Å, specifically 0.5580, and 0.0017 for 4J3D, and 5I21 proteins respectively. Chimera software was used to add charges, remove solvents, and correct the structure to be analyzed (Yang & Chen, 2004).

In silico molecular coupling was performed between the targets described above and the 10 molecules studied (Fig. 1), as well as with the biomaterials studied: TiO₂ nanotube model (NT) and chitosan model (QT). To compare the ten studied molecules were added some control molecules (Fig. 2): two commercial drugs (C-Cip: ciprofloxacin, and C-Azi: azithromycin, Figs. 2A, and 2B respectively); two benzofuran derivatives (Budhwani, Sharma & Kalyane, 2017), see Figs. 2C, and 2D; and two isatine derivatives (Chemchem et al., 2020), see Figs. 2E, and 2F. The molecular docking was carried out through the MolDock Score (Thomsen & Christensen, 2006) implemented in the molegro virtual docker package (MVD) (Yang & Chen, 2004).

This study compared which biomaterial could be introduced to the selected target to understand the best binding mode of a possible antibiotic-carrying nanovector. We evaluated the above by first coupling the NT, then converting NT into a “cofactor” of the protein (in other words, considering it as part of the protein) so that the software would consider it part of the complete system and continue with the analysis. QT coupling on the said white-NT system. With this, we could corroborate how the presence of NT affects this type of nanodevice and the energy contribution due to the presence of said NT.

Converting the NT as a “cofactor” was also performed for the QT to couple the NT subsequently. Once this target-NT-QT or target-QT-NT system, the ten studied molecules were coupled to evaluate the effect of the NT-QT or QT-NT type nanotransporter on a Pseudomonas target.

Finally, was proposed a pharmacophore model for each studied target considered the docking interactions. This study was made using the ZincPharmer (Koes & Camacho, 2012).

Condensed Fukui functions

All the studied molecules were optimized following the described method. The most susceptible zones of electrophilic, nucleophilic, and radical attacks were computed using the Fukui condensed functions. The carbonyl carbon linked to the benzofuran moiety is a high-reactivity atom for a nucleophilic attack for the designed molecules. The carbonyl carbon in the isatin fragment presents the most electrophilic-attack site. We highlight that these results are essential due to the fact that they can predict the most suitable sites of interaction of the benzofuran-isatin designed molecules with the NT and chitosan; see Table 1 and Fig. 3.

Molecular docking and pharmacophoric model

The docking was made in Hfq and LpxC targets, which are proteins in P. aeruginosa. Note that the main cavities of these targets were evaluated, see the Supplemental File, but this section only shows the better-docked cavity.

Figure 4 depicts the studied molecules docked with Hfq. Table 2 shows the ligand efficiency for the ten designed molecules, the referencing molecules, the model of QT, NT, and the commercial drugs. Table 2 shows the ligand efficiency for the above molecules with the free target in the second column. The third column depicts the target with the NT linked, in other words, the system Target-NT is a simulation of the first introduction of NT in the target after testing the studied molecules. The four to six columns present similar behavior to the third, denoting the influence of QT, NTQT, and QTNT when this is docked in the target.

In this order, Table 2 shows that when introduced QT in the protein, see column four, the ligand efficiency of the systems is higher, which means a less interaction ligand-target. However, when QT is docked first in the protein and after NT is introduced into the Target-NT system, see column six, the LE is higher, representing a badly ligand-target interaction. As column five depicts, the same occurs when NT is docked first and after QT is introduced in the Target-QT system.

Table 3 shows the principal interaction energies for the target-NT assay. The cause that MOL E interacts better than the other designed molecules is not clearly explained. In this order, Fig. 5A shows the hydrogen bond interactions between MOL-E and Hfq-NT (target-NT), and Fig. 5B shows the electrostatic interactions.

In the case of hydrophobic surfaces, were analyzed the MOL-A in the Hfq, see Fig. 6A, and the MOL-E in the Hfq-NT system, see Fig. 6B.

Regarding the docking in LpxC, Table 4 shows that the influence of NT is similar to Hfq protein due to the presence of NT (coupling the designed molecules in the LpxC-NT). Above promotes a higher LE in molecules BRA, MOL-A, MOL-B, MOL-C, MOL-E, MOL-F, MOL-G, MOL-J (Figs. 7A and 7B), and QT than designed molecules coupled on free target (LpxC), see Table 4 column two and three.

Analyzing the specific interactions of the studied molecules with the LpxC-NT (Table 5), the behavior is similar to Hfq protein, not being determinant Vdw, Electro of H-bond interactions to the ligand-target whole interaction.

Moreover, analyzing the hydrophobic surfaces of LpxC, Fig. 8A shows MOL-I docked into the free main cavity of LpxC, and Fig. 8B shows the interaction of MOL-J with the LpxC-NT system.

To conclude, the present study made a pharmacophoric model for each studied protein (Figs. 9 and 10). This study was based on the ligand interactions and proposed the main kind of fragments to try to inhibit this protein. Note that the green spheres depict the hydrophobic fragments, purple spheres are the aromatic-hydrophobic fragments, blue sphere is the hydrophilic fragments, and orange spheres are the hydrogen bond donor fragments.

Condensed Fukui function

In regard to condensed Fukui functions analysis, MOL A shows an electrophilic behavior with a value of 0.337 in the C29 and nucleophilic with 0.022 in C18. A radical-attack zone is less, with a value of 0.138 in C20. The reactive zones for (B–D) are in the C28, F-with around 0.337.

Molecule A tends to react with an electrophilic attack on the C29 atom (atom 28 in Molecule B–D). This zone corresponds to the structure of isatin in position 2 with a secondary carbonyl group. A possible radical attack occurs at the C20 position for this molecule, which is a secondary carbonyl linked to isatin and benzofuran. This behavior is similar in molecules (B–D).

Molecular docking and pharmacophoric model

On the other hand, the influence of introducing the first NT in the target Hfq is benefic for the ligand-target interaction, such as with BRB, C-CIP, IRA, and MOL E, which the LE diminishes, which means a better ligand-target interaction. Despite MOL E not presenting better interaction than the referencing molecules (see Figs. 4 and 5, BRA, BRB, Budhwani, Sharma & Kalyane, 2017; IRA, and IRB, Chemchem et al., 2020), when it is docked with the presence of NT (Table 2, second column), their ligand efficiency is better. The ligand-target interactions beat the values for the commercial drugs (C-AZI and C-CIP), with a value of −4.321 kcal/mol. Note that there is a reported interaction between Hfq with TiO₂ NPs, but the importance of this as a nanocarrier is not highlighted (Anupama et al., 2019).

In regard to Table 2 results and analyzing the structures depicted in Fig. 1, Mol E presents two Flours substituting the benzene moiety, conferring this possible better interaction. MOL-E interacts specifically with Ser 6 and Gln 41 residues in terms of hydrogen bond interactions, see Fig. 5A. In the case of electrostatic interactions, Fig. 5B highlights that the NT confers higher attractive electrostatic interactions with MOL-E, also interacting more with the oxygens of the molecule.

Concerning hydrophobic interactions, the free-designed molecule (MOL-A) is introduced into the cavity, interacting more with the hydrophilic zone of MOL-A with the hydrophilic surfaces of Hfq (Fig. 6A). In the case of MOL-E (docked in the Hfq-NT system), the NT is introduced in the bottom of the cavity, interacting with a hydrophilic surface of Hfq. This first interaction promotes the designed molecule (MOL-E) docket out of the cavity, specifically on a hydrophilic surface (Fig. 6B).

On the other hand, in the case of the docking in LpxC, note that the ligand with the higher LE is the MOL-I (LE of −4.941 kcal/mol), and in the case of the LpxC-NT system, is MOL-I (LE of −4.908 kcal/mol). This means that the influence of NT in the target is favorable to the ligand-target interaction. Considering the structure of MOL-J, see Fig. 1, is the small of all the designed systems, which is a determinant of making a better LE. Despite the determinants of H-bond and Electro interactions, Fig. 7A shows that Met 62 and Gly 192 interact via H-bond with MOL-J, and the Zn co-factor is determined in the electrostatic interactions of MOL-J, see Fig. 7B.

Moreover, analyzing the hydrophobic surfaces of LpxC, Fig. 8A shows MOL-I docked into the free main cavity of LpxC, which interacts mainly with hydrophobic surfaces of the protein and covers the surface of the studied cavity. Figure 8B shows the interaction of MOL-J with the LpxC-NT system, which exposes the influence of NT on the system. This figure clearly shows that NT is introduced in the deep zone of the cavity, promoting a similar interaction but with higher LE.

To conclude the present study, the pharmacophoric model was formulated. Figure 9 shows the pharmacophoric model for the Hfq protein of P. aeruginosa, demonstrating that six fragments are necessary to dock some molecules better to this protein. Specifically, it needs two hydrophobic fragments at both sites of the designed molecule (green spheres), followed by two aromatic-hydrophobic fragments (purple spheres). Describing the model’s center is necessary for one hydrophilic (blue spheres) and other hydrogen bond donor fragment (orange sphere).

In the case of the pharmacophoric model for LpxC protein, it needs nine fragments to be docked with some designed molecules, see Fig. 10. The first two are similar to the Hfq model, being two hydrophobic (green spheres) sites at the sites of the molecule. The following fragments are also two aromatic-hydrophobic sites (green spheres). The left side of the molecule needs two small hydrophilic fragments, and the molecule’s center should need two hydrogen bond donor fragments (orange spheres) and one hydrophilic site (blue sphere).