Homology modeling and structural validation
The crystal structure of cytochrome b subunit of the cytochrome bc1 complex (QcrB) in M. tuberculosis is not available in the Protein Data Bank database (PDB). As such, the SWISS-MODEL template library was searched with BLAST (close homologs) and HHBlits (remote homologs) as mentioned previously for evolutionary related structures matching the target sequence [22]. The results of the searching revealed a ubiquinol-cytochrome C reductase QcrB structure of a functional obligate respiratory supercomplex from Mycobacterium smegmatis (PDB: 6HWH) as the closest template [23], and it shares 99.5% query coverage and 82.56% identity with Mycobacterium tuberculosis QcrB (Fig. 3a). Therefore, the homology model of M. tuberculosis QcrB subunit was built with GMQE score of 0.91 and QMEAN of − 3.86 which suggests good quality and reliability (Fig. 3b). Local appraisals of model quality in terms of the QMEAN scoring function with respect to residue number and as a global score in relation to PDB structure set of high-resolution (Z score) was generated from the workspace.
The graph of the predicted local similarity to target versus the residue number of the predicted 3D structure of the modeled protein was plotted so as to estimate the local quality of the residues (Fig. 4a). It was observed that most of the residue scores are close to 1, signifying that the predicted model has a good local quality estimate, while residues that score lower than 0.6 are said to be of low quality. Figure 4 b shows a plot of normalized QMEAN score against protein size which relates the model quality scores of an individual model to values obtained for experimental structures of similar size. The structure of the modeled protein was found within the range in comparison with the non-redundant set of PDB structures which affirms its reliability. In theory, the dihedral angle (ω) of the protein backbone is limited to the amide bond (C and N) planarity in addition to the hybridization of atomic orbitals involved which resulted in a resonance structure with a permanent dipole and double binding sites. The dihedral angle values ѱ(Psi) and ϕ(Phi) are controlled by the steric hindrance between the side-chain atoms and neighboring peptide bonds. The Ramachandran map is an efficient approach to visualize favored regions for backbone dihedral angles ѱ(Psi) against ϕ(Phi) of amino acid residues. It involves plotting the ѱ(Psi) scores on the y-axis and the ϕ(Phi) on the x-axis with angle spectrum ranging from − 180º to + 180º to predict the secondary structure and possible conformation of the peptide.
Figure 4 c shows the Ramachandran plot of the homology model of QcrB protein for all non-glycine or non-proline residues generated by the Swiss-Model workspace. The colored contour (green) represents the allowed and favored regions. However, the plot revealed 93.25% Ramachandran favored, and the most dominant amino acid residues are predicted to be contained in right-handed alpha-helices as secondary structure.
In silico molecular docking results
The in silico molecular docking simulation hunts the best binding modes or the conformation of the ligand with the active residues of a protein target [21]. Furthermore, the assessment of docking output depends on the highly negative magnitude of binding affinity (lowest binding energy) portraying the best conformation of the ligand in the active pockets of the target. As stated earlier, the homology-modeled QcrB protein was used as the receptor to perform in silico docking computational studies on a series of thirty-five (35) IPA compounds.
Figure 5 depicted the binding affinity scores of the best ligand’s binding modes ranged from − 6.5 to − 10.1 kcal/mol which affirms their resilient potency as experimentally reported by Wang et al. [5]. In addition, the ligands were predicted to have better binding affinity values than isoniazid (standard drug) with a binding score of − 6.0 kcal/mol. Thus, ligand numbers 2, 7, 22, 26, and 35 with higher binding affinity values (− 9.60, − 9.80, − 10.10, − 10.00, − 10.00 kcal/mol) among others were selected as the best to assess their residual interactions by using the Discovery Studio software.
Ligand 2 formed ten (10) hydrophobic interactions with 9 amino acid residues as shown in Fig. 6a. The imidazo[1,2-a] pyridine fragment of ligand 2 formed five (5) pi-alkyl interactions with Leu 65, Leu 166, Pro 167, Phe 69, and Arg 111 at different distances, while the bromo substituent (-Br) attached to the benzene formed three (3) pi-alkyl interaction with Phe 55, Leu 58, and Phe 121 residues. Also, delocalized Π-electron of the benzene ring interacted with Leu 58 to form pi-sigma hydrophobic and pi-anion interaction with Glu 153 (electrostatic) respectively. Other noticeable interactions include C–H bond with His 216 and Gly 160, unfavorable donor-donor interaction with Arg 111, and halogen interaction between fluoro substituent of imidazo[1,2-a] pyridine fragment and His 216 accordingly.
Ligand 7 formed thirteen (13) major hydrophobic interactions in which the –Cl (chloro) substituent of imidazo[1,2-a] pyridine moiety formed three (3) alkyl and pi-alkyl hydrophobic with different residues (Leu 58, Ala 118, and Phe 121), and its delocalized Π-electrons formed amide-pi stacked and pi-alkyl interaction with Leu 58 and Pro 221 respectively. The bromo substituent (−Br) attached to the benzene formed four (4) pi-alkyl interaction with Ile 100, Phe 69, Ala 97, and Pro 167 residues while delocalized Π-electron in the benzene formed a pi-pi T-shaped interaction, pi-sigma, and pi-alkyl interactions with Phe 69, Leu 65, and Leu 166 accordingly. Other interactions comprise one (1) C–H bond with His 114 and two (2) pi-anion electrostatic interactions with Glu 159 as in Fig. 7a.
Ligand 22 formed two (2) C–H bond interactions with the residue His 159 and one (1) van der Waals interaction with Ile 217. Also, ligand 22 formed fifteen (15) hydrophobic interactions which encompass pi-sigma with Leu 65 while pi-alkyl and alkyl hydrophobic involved Ala 97, Pro 167, Ile 100, Leu 166, Arg 111, Leu 59, Val 63, Phe 69, Pro 221, His 114, Leu 58, Ala 118, and Leu 65 at different interaction distances as elucidated in Fig. 8a.
Ligand 26 formed two (2) conventional H-bonds where –NH of imidazo[1,2-a] pyridine fragment interacts with His 240 and also –NH of acetamide fragment interacts with His 128 different distances respectively. In addition, ligand 26 also formed fourteen (14) hydrophobic interaction including pi-pi T-shaped, pi-sigma, pi-alkyl, and alkyl interaction types with residues such as Leu 152, Phe 121, Ile 228, Ala 156, Ile 125, Ile 224, Ala 51, His 128, Val 235, Ala 137, and Ile 132. Other interactions include three (3) C–H bonds with His 231, Gly 48, and a halogen interaction involving the fluoro substituent and Asn 145 amino acid residue as shown in Fig. 9a.
Ligand 35 formed one (1) conventional H-bond due to the residual interaction between –NH of acetamide fragment and Ala 385 amino acid. Furthermore, ligand 35 also formed thirteen (13) hydrophobic interactions including pi-pi T-shaped, amide-pi-stacked, pi-sigma, pi-alkyl, and alkyl interaction types with residues such as Phe 133, Met 126, Leu 129, Ile 386, Try 389, Ala 385, Val 345, Leu 348, Tyr 352, and Ile 381. Halogen residual interaction was also formed between fluoro substituent of benzene and Leu 355 as elucidated in Fig. 10a.
Furthermore, receptor surfaces (hydrophobicity and hydrogen bonds) were created to have insight into the inner workings of the QcrB modeled receptor. The hydropathy index of the ligand protein was measured as the hydrophobicity surfaces which show the hydrophilic or hydrophobic properties of the amino acid side chains shown in Figs. 6b, 7b, 8b, 9b, and 10b. In addition, the hydrophobicity of the amino acid residue surfaces are colored from blue for hydrophilic to brown for hydrophobic accordingly. For H-bond surfaces, H-bond donors are colored as magenta surfaces while H-bond acceptors are colored in green as shown in Figs. 6c, 7c, 8c, 9c, and 10c. The binding modes of the docking output (Figs. 6, 7, 8, 9, and 10) revealed a high amount of hydrophobic/π interactions such as π–alkyl, π–sigma, and π–π type of interactions in conjunction with the presence of some hydrogen bonds and other noticeable interactions that contributed to the ligand-protein complex stability in the binding pockets of the modeled protein.