An Optimized Workflow for the Discovery of New Antimicrobial Compounds Targeting Bacterial RNA Polymerase Complex Formation

Abstract

Bacterial resistance represents a major health problem worldwide and there is an urgent need to develop first-in-class compounds directed against new therapeutic targets. We previously developed a drug-discovery platform to identify new antimicrobials able to disrupt the protein-protein interaction between the β' subunit and the σ⁷⁰ initiation factor of bacterial RNA polymerase, which is essential for transcription. As a follow-up to such work, we have improved the discovery strategy to make it less time-consuming and more cost-effective. This involves three sequential assays, easily scalable to a high-throughput format, and a subsequent in-depth characterization only limited to hits that passed the three tests. This optimized workflow, applied to the screening of 5360 small molecules from three synthetic and natural compound libraries, led to the identification of six compounds interfering with the β'-σ⁷⁰ interaction, and thus was capable of inhibiting promoter-specific RNA transcription and bacterial growth. Upon supplementation with a permeability adjuvant, the two most potent transcription-inhibiting compounds displayed a strong antibacterial activity against Escherichia coli with minimum inhibitory concentration (MIC) values among the lowest (0.87-1.56 μM) thus far reported for β'-σ PPI inhibitors. The newly identified hit compounds share structural feature similarities with those of a pharmacophore model previously developed from known inhibitors.

Keywords: RNAP holoenzyme assembly; antibiotics; bacterial RNA polymerase (RNAP); bacterial transcription inhibitors; drug-discovery workflow; protein–protein interaction (PPI); protein–protein interaction inhibitor; yeast Bioluminescence Resonance Energy Transfer (yBRET).

Figure 1

X-ray structure of the E. coli RNA polymerase σ⁷⁰ holoenzyme complex (PDB id 4YG2 [11]) with an expanded view of the β’ subunit–σ⁷⁰ PPI; binding sites of rifampicin and fidaxomicin are also shown. Rifampicin and fidaxomicin structures are from PDB structures 1YNN [20] and 7L7B [21], respectively.

Figure 2

Outline of the drug-discovery workflow applied to the identification of small-molecule inhibitors of RNAP β’–σ⁷⁰ interaction, endowed with antimicrobial activity. The number of compounds that passed each step is indicated. Candidate hits were further characterized with the indicated assays, as described in the following sections.

Figure 3

Antibacterial activity determined at increasing growth times and at different concentrations of compounds 5, 7, and 20 (E. coli) and compounds 13 and 19 (B. subtilis) as indicated. Data are the mean of three replicates.

Figure 4

Transcription inhibition capacity. (A) In vitro transcription activity was measured under single-round conditions in the presence of a fixed concentration (100 µM) of the indicated hit compounds and of reference compound 4. (B) Dose–response curves determined for a selected subset of compounds. In vitro transcription activity is reported as percent relative to the DMSO vehicle. Data are the mean of four replicates; error bars represent SD.

Figure 5

The most potent β’–σ interaction inhibitors belonging to (a) the pyrido-pyrrolo-isoquinoline (7), (b) the benzofuran (9), and (c) the pyrido-indole (19) series are shown, superposed to the reference indolyl-urea inhibitor 4 (cyan carbons) fitted on the pharmacophore model previously devised for β’–σ interaction inhibitors [18]. The pharmacophore model comprises hydrophobic (green), aromatic (orange), and hydrogen bond donor (light blue) and acceptor (red) features characterized by a tolerance of 2 Å (gray sphere). The features highlighted with a colored contour of the tolerance region represent the sites matched by each of the new inhibitors.