Computers Hunt Bacteria’s Weak Spots: New Multi-Target Compounds Show Promise Against Staphylococcus aureus

Staphylococcus aureus is one of the most formidable pathogens in modern medicine, responsible for everything from minor skin infections to deadly pneumonia, sepsis, and bloodstream infections that claim lives worldwide. What makes this bacterium particularly dangerous is its remarkable ability to develop resistance to nearly every antibiotic deployed against it, including methicillin and other beta-lactam drugs. Even vancomycin, long considered the last line of defense, has encountered resistant strains. Now, a research team from Bhupal Nobles’ University and Jamia Hamdard in India has reported a computationally driven strategy that may help tip the scales back in favor of medicine, identifying candidate compounds designed to attack the bacterium at multiple essential points simultaneously rather than relying on a single vulnerable target.

The study, published in Discover Chemistry, employed a polypharmacological approach known as multiple-target-directed ligands, or MTDLs. The underlying logic is elegant: conventional antibiotics that inhibit a single bacterial enzyme impose intense selective pressure, allowing pathogens to acquire resistance through relatively simple genetic changes. By contrast, a compound that disrupts several essential pathways at once distributes that pressure across interconnected biological networks, theoretically making it far harder for the bacterium to evolve escape routes. The team selected five enzymes that are indispensable for S. aureus survival: MurB, which drives peptidoglycan biosynthesis for the cell wall; DNA gyrase B, required for DNA replication and transcription; dihydrofolate reductase and dihydropteroate synthase, both crucial for folate synthesis; and tyrosyl-tRNA synthetase, essential for protein synthesis.

To find compounds capable of engaging all five targets, the researchers turned to structure-based virtual screening, a technique that considers the entire three-dimensional architecture of an enzyme’s active site when searching for matching ligands. This contrasts with ligand-based screening, which relies on matching a limited number of pharmacophoric features, typically four or five, derived from known inhibitors. The team downloaded the MolMall database of 22,401 molecules and prepared each enzyme from high-quality crystal structures deposited in the Protein Data Bank, selecting entries with a resolution of 2.5 angstroms or better, no reported mutations, and a bound co-crystallized ligand. Using the Maestro molecular modeling suite from Schrödinger, the proteins were protonated, energy-minimized, and fitted with receptor grids centered on their native ligands: FAD for MurB, and the co-crystallized inhibitors 07N, Q27, HH2, and 662 for the remaining enzymes.

A critical step before any large-scale screening was validation of the docking protocol itself. The researchers extracted each co-crystallized ligand from its enzyme and re-docked it into the prepared binding site, then superimposed the docked pose on the original crystallographic orientation. The root-mean-square deviation of heavy atoms remained within acceptable limits, and the key interactions between ligands and active-site residues were consistent with reported binding features, confirming that the computational pipeline could reliably reproduce experimentally observed binding modes. Only then did the team unleash the full virtual screening workflow, passing the ligand library through successive high-throughput, standard-precision, and extra-precision docking stages with retention thresholds of 50, 40, and 30 percent respectively, progressively narrowing the pool to the most promising candidates.

The screening produced 191 compounds with favorable docking profiles across all five enzymes. After applying computational filters for predicted absorption, distribution, metabolism, and excretion properties, along with Lipinski’s rule of five, Jorgensen’s rule of three, and estimates of binding free energy using the molecular mechanics/generalized Born surface area method, fifteen multitarget hits were shortlisted. Six of these were physically procured through MolMall’s molecular exchange program and sent to the CSIR-Central Drug Research Institute in Lucknow for in vitro antibacterial testing against the ESKAPE pathogen panel, a collection of bacteria notorious for antibiotic resistance, using the broth microdilution method with levofloxacin as a reference agent.

The experimental results were strikingly selective. Four of the tested compounds, designated MolMall 643, 644, 10988, and 17782, exhibited inhibitory activity against S. aureus, while showing limited effect against the other ESKAPE pathogens. Two compounds, MolMall 643 and MolMall 17782, emerged as the most active, achieving minimum inhibitory concentrations of 4 micrograms per milliliter against the reference strain. Computationally, these two compounds also displayed their most favorable binding profiles against MurB, with docking scores of minus 8.884 and minus 6.781 kilocalories per mole and MM/GBSA binding free energies of minus 97.801 and minus 72.404 kilocalories per mole respectively, suggesting that the cell wall biosynthesis enzyme may be a principal contributor to their antibacterial effect.

To verify that the docking-predicted binding modes were physically plausible over time, the researchers subjected the MurB complexes of the two most active compounds to 100-nanosecond molecular dynamics simulations using the Desmond engine. The systems were solvated, neutralized, and brought to physiological salt concentrations before equilibration and production runs. Throughout the simulations, both protein-ligand complexes remained stable, with root-mean-square deviation values confined below 3 angstroms, and both ligands stayed within the MurB active site for the entire duration. MolMall 643 sustained a hydrogen bond with Gly81, while MolMall 17782 interacted with Val199 through a hydrogen bond and formed halogen bonds with Ser143 and Asn80. Notably, these residues also participate in binding the native FAD cofactor, indicating that the synthetic ligands occupy the same functional pocket as the enzyme’s natural partner.

The flexibility analysis revealed subtle but informative details about how the compounds adapt within the binding pocket. MolMall 643 showed localized fluctuations at the sulfur atom of its thiazolidinone moiety and at several positions on its piperidine ring, while MolMall 17782 displayed increased mobility at the meta- and para-chloro substitutions on its phenyl ring. These localized movements did not disrupt overall binding stability, and the authors suggest that such limited flexibility may actually facilitate adaptive fitting within the MurB active site, a phenomenon sometimes described as induced fitting that allows ligands to optimize their contacts without detaching.

The researchers are careful to delineate the boundaries of their findings. No target-specific enzyme inhibition assays were performed, meaning the proposed interactions with the five enzymes remain computationally predicted rather than experimentally confirmed. The compounds should therefore be regarded as experimentally validated antibacterial agents against S. aureus with computationally predicted multitarget activity. The molecular basis for the limited activity against Gram-negative organisms remains unclear, resistance frequency studies were not conducted, and any advantage of the multitarget strategy in delaying resistance remains theoretical at this stage. Lead optimization and comprehensive toxicity assessment were also beyond the scope of the current hit-identification campaign.

Nevertheless, the study demonstrates a complete and rational workflow that connects computational prediction with experimental validation, moving from a database of more than 22,000 molecules to two compounds with measurable antibacterial activity and dynamically stable binding modes. By targeting enzymes involved in cell wall biosynthesis, DNA replication, folate metabolism, and protein synthesis simultaneously, the approach explores pathway-level inhibition rather than dependence on a single molecular interaction. As multidrug-resistant S. aureus continues to complicate clinical management globally, such structure-based multitarget screening pipelines may offer a practical blueprint for discovering the next generation of antibacterial starting points, providing a foundation for subsequent structural optimization and mechanistic investigation.

Subject of Research: Structure-based identification of multitarget inhibitors of essential Staphylococcus aureus enzymes with in vitro antibacterial validation

Article Title: Structure-based identification of potential inhibitors of essential Staphylococcus aureus enzymes with in vitro antibacterial screening

Article References: Singhal, L., Kumar, N., Sharma, C. S., Akhter, M., & Shrivastava, A. (2026). Structure-based identification of potential inhibitors of essential Staphylococcus aureus enzymes with in vitro antibacterial screening. Discover Chemistry, 3(1), Article 461. https://doi.org/10.1007/s44371-026-00919-8

Image Credits: AI Generated

DOI: 10.1007/s44371-026-00919-8

Keywords: Staphylococcus aureus, multiple-target-directed ligands, polypharmacology, antibacterial resistance, structure-based virtual screening, MurB, molecular docking, molecular dynamics simulations, MM/GBSA, antimicrobial drug discovery, ESKAPE pathogens, minimum inhibitory concentration