Unveiling structural components of dibenzofuran-based MMP-12 inhibitors: a comparative classification-dependent analysis with molecular docking-based virtual screening and molecular dynamics simulation

Abstract

The implication of matrix metalloproteinase-12 (MMP-12) in various major disorders including cancer, COPD, cardiovascular disorders, and neurological diseases makes it a potential target for drug discovery. Contemplating the significance of MMP-12, a number of MMP-12 inhibitors were designed, synthesized and tested throughout the world but the non-selective nature of most of those molecules can lead to adverse drug interactions. In contradiction, the dibenzofuran (DBF) and dibenzothiophene (DBT) derivatives showed highly potent and selective MMP-12 inhibition. Therefore, to identify the prime molecular and structural attributes that are affecting the MMP-12 inhibitory activity, the linear discriminant analysis (LDA), Bayesian classification, recursive partitioning, and SARpy analysis were performed to extract the prime attributes of these DBFs and DBTs affecting MMP-12 inhibition. These studies suggested that substructures like isopropyl carboxylic acid, 5-methyl furan, 1,2,4-oxadiazole, and DBT moieties can impart moderate to high contribution for MMP-12 inhibition. Importantly, the outcomes of the current studies were also in agreement with our regression-based study performed earlier. Furthermore, the molecular docking-mediated virtual screening of DBT and DBF analogs of the ChEMBL database demonstrated the viability of other DBT and DBF analogs to become potential MMP-12-selective inhibitors. The molecular dynamics (MD) simulation study of hit molecules also showed the potential of the combination of phosphonic acid ZBG and DBF P1' substituent for effective anchoring/binding at the MMP-12 active site. Therefore, the findings may help in the discovery and designing of novel MMP-12 inhibitors that may be used for the treatment of various pathological diseases including cancer and COPD.

Supplementary information: The online version contains supplementary material available at 10.1007/s40203-024-00296-z.

Keywords: Classification-based molecular modeling; MMP-12; Molecular docking; Molecular dynamics simulation; Virtual screening.

Fig. 1

Implication of MMP-12 in different disease conditions

Fig. 2

Schematic representation of dataset preparation for classification-based model development

Fig. 3

A The ROC plot for the training set and the test set according to the LDA model; B the ROC plot for the training set and the test set as per the Bayesian classification model

Fig. 4

Good substructural fragments (G1-G20) along with their Bayesian score obtained after Bayesian classification modeling

Fig. 5

Structure of some potent MMP-12 inhibitors containing good Bayesian substructural fragments (green color represents good substructural fingerprints)

Fig. 6

Bad substructural fragments (B1-B20) along with their Bayesian score obtained after Bayesian classification modeling

Fig. 7

Structure of some less effective MMP-12 inhibitors containing bad Bayesian substructural fragments (magenta color represents bad substructural fingerprints)

Fig. 8

A Schematic representation of decision tree 1 after recursive partitioning B Radar plot of the training set performance for the decision tree 1 C Radar plot of the test set performance for the decision tree 1

Fig. 9

Structure of the important fragments contributing to MMP-12 inhibition observed in Recursive partitioning study

Fig. 10

SARpy active rule sets having these structural alerts

Fig. 11

Four dibenzofuran analogs screened as final hit molecules for MMP-12 inhibition

Fig. 12

RMSD plots for A CHEMBL146676 and B CHEMBL357226, RMSF plots (green lines indicates prime residues interacted with complexed ligand) for C CHEMBL146676 and D CHEMBL357226 for 100 ns MD simulation study

Fig. 13

The binding mode and interactions of CHEMBL146676 in A 0 ns, B 50 ns, and C 100 ns; the binding mode and interactions of CHEMBL357226 in D 0 ns, E 50 ns, and F 100 ns

Fig. 14

A Overall interaction, B interaction fraction, C interaction frequency, and D numberof interactions between amino acid residues of MMP-12 and ligand for 100 ns MD simulation study of CHEMBL146676-MMP-12 (PDB ID: 1RMZ) complex

Fig. 15

A Overall interaction, B interaction fraction, C interaction frequency, and D numberof interactions between amino acid residues of MMP-12 and ligand for 100 ns MD simulation study of CHEMBL357226-MMP-12 (PDB ID: 1RMZ) complex

Fig. 16

Important substructures obtained from classification-based molecular modeling studies