Combinatorial Libraries As a Tool for the Discovery of Novel, Broad-Spectrum Antibacterial Agents Targeting the ESKAPE Pathogens

Abstract

Mixture based synthetic combinatorial libraries offer a tremendous enhancement for the rate of drug discovery, allowing the activity of millions of compounds to be assessed through the testing of exponentially fewer samples. In this study, we used a scaffold-ranking library to screen 37 different libraries for antibacterial activity against the ESKAPE pathogens. Each library contained between 10000 and 750000 structural analogues for a total of >6 million compounds. From this, we identified a bis-cyclic guanidine library that displayed strong antibacterial activity. A positional scanning library for these compounds was developed and used to identify the most effective functional groups at each variant position. Individual compounds were synthesized that were broadly active against all ESKAPE organisms at concentrations <2 μM. In addition, these compounds were bactericidal, had antibiofilm effects, showed limited potential for the development of resistance, and displayed almost no toxicity when tested against human lung cells and erythrocytes. Using a murine model of peritonitis, we also demonstrate that these agents are highly efficacious in vivo.

Figure 1.. Screening the scaffold ranking library for antibacterial activity against the ESKAPE pathogens.

Compound mixtures were assayed against the ESKAPE pathogens using a micro broth dilution assay. Data is presented as a stacked scaled score, which is determined by dividing 100 μM (the maximum concentration tested) by the individual doses tested. Each library is given a scaled score for each pathogen, and these are then stacked to determine the library with the broadest activity, at the lowest concentration.

Figure 2.. Deconvolving the antibacterial activity of the bis-cyclic guanidine library.

The bis-cyclic guanidines were systematically synthesized into a positional scanning library containing 110 samples (shown in supplemental table S2). These were fixed at: A = the R1 (42 samples); B = R2 (26 samples); or C = R3 (42 samples) position. For example, the first sample in A, is an approximate equal molar mixture of 1,092 compounds. The 1,092 compounds contain hydrogen fixed in the R1 position and all 1,092 combinations of the 26 R2 and 42 R3 functionalities. Similarly, the first sample in B is 1,764 compounds generated from fixing R2 with S-methyl and utilizing all 1,764 combinations of the 42 R1 and 42 R3 functionalities. The height for each color of individual bars is determined by dividing 100 μM (the maximum concentration tested) by the individual MIC for each agent. Libraries are then given a scaled score for each pathogen, and these are stacked to determine the library with the broadest activity, at the lowest concentration.

Figure 3.. Assessing the antibacterial activity of individual bis-cyclic guanidines synthesized based on library SAR data.

Fifty-four individual compounds were synthesized for testing against the ESKAPE pathogens. 1–27 were generated based on SAR data from ESKAPE testing with the combinatorial libraries; 28–54 were included as they were predicted to be significantly less active based on PSL data. Data is presented as stacked, scaled scores, with the height for each color of individual bars determined by dividing 100 μM (the maximum concentration tested) by the individual MIC for each agent. Compounds are then given a scaled score for each pathogen, and these are then stacked to determine which have the broadest activity, at the lowest concentration. Note data is generated using “crude” compounds (see Materials and Methods Section for details).

Figure 4.

Lead bis-cyclic guanidine compounds.

Figure 5.. Computational Exploration of Physicochemical Properties.

Each of the 54 compounds (1–54) are compared against each of the remaining 53 compounds for differences in potency (Y-axis both left and right panel) and molecular representation (Physicochemical Properties: X-axis left panel; Radial: X-axis right panel). Each pair is represented by a dot. In this way a pair of compounds with similar activity potencies and physicochemical properties will be shown by a dot in the upper right hand quadrant of the left panel. The dots are colored by activity of the most potent compound in a pair, using a continuous color of: Grey (no activity), Yellow (low activity), Orange (moderate activity), and Red (high activity). Shown below the panels are structures for two such pairs. The pair in the left location on both panels (19–35) is identified by open blue circles, whilst the pair in the right location (2–32) is indicated by open black circles. Under each structure is the total activity value used for each compound, as well as the three physicochemical values (MW, AlogP, and RB) associated with a given agent.

Figure 6.. Bis-cyclic guanidines are bactericidal but not bacteriolytic.

A. Time kill studies were performed using MRSA and the front runner agents (at MIC concentrations), alongside positive (4 μM lysostaphin, 0.001% Benzalkonium chloride (BA), 0.001% Benzethonium chloride (BC), and 2.0% Sodium dodecyl sulfate (SDS)), and negative (200 μM Doxycyline (Doxy)) control agents. Shown is the optical density of cells relative to starting values from three independent experiments. Error bars are shown ±SEM. B: Cell viability of all samples after the 120 min experiment. Compounds were removed by centrifugation and washing of cells, followed by serial dilution and enumeration. Percent recovery was determined by comparison to no drug (ND) controls.

Figure 7.. Cytotoxicity of lead agents.

Shown is the survival of A549 cells measured using an MTT assay with all five lead agents (A-E). Data is presented as percent recovery compared to vehicle only controls. Error bars are shown ±SEM, from at least three independent experiments; MICs are denoted by grey coloring. A solid black line is shown for IC50 value determination. Hemolytic capacity towards human erythrocytes was also measured using the lead agents (F). Data is shown as percent hemolysis compared to positive (1% Triton-X100 (T), 100% hemolysis) controls. Lead agents were added at a concentration of 10 μM. Error bars are shown ±SEM, from at least three independent experiments. A solid black line is shown at 1% hemolysis.

Figure 8.. Exploring Adaptive Tolerance by ESKAPE Pathogens to Front Runner Agents.

ESKAPE pathogens were serially passaged for eight days in fresh liquid media (changed every 24h), with the concentration of compound increased 2-fold each day. Shown are the increases in MIC observed over time. Ef = E. faecium; Ec = E. cloacae.

Figure 9.. Front runner bis-cyclic guanidines are efficacious during in vivo infection.

Mice were I.P. infected with a lethal dose of S. aureus. After 1h, they were then injected with either front-runner bis-cyclic guanidines (at 2 × MIC), vancomycin (positive control, at 5 × MIC and 10 × MIC) or vehicle alone (negative control). Mice were then monitored for five days, and the significance of mortality measured using a log rank and chi square test with 1-degree of freedom. * = p > 0.05, ** = p > 0.01.

Scheme 1.. Synthetic scheme of bis-cyclic guanidines.

a) 5% DIEA/DCM; b) Fmoc-Lys(Boc)-OH, DIC, HOBt, DMF; c) 20% Piperidine/DMF; d) R1COOH, DIC, HOBt, DMF; e) 55% TFA/DCM; f) Boc-AA(R2), DIC, HOBt, DMF; g) R3COOH, DIC, HOBt, DMF; h) BH3-THF, 65oC, 96 hours; i) Piperidine, 65°C, 24 hours; j) CNBr, DCM; k) HF, anisole, 0°C