An Optimized Synthetic-Bioinformatic Natural Product Antibiotic Sterilizes Multidrug-Resistant Acinetobacter baumannii-Infected Wounds

Abstract

The antibiotic paenimucillin A was originally identified using a culture-independent synthetic-bioinformatic natural product (syn-BNP) discovery approach. Here we report on a bioinformatics-guided survey of paenimucillin A analogs that led to the discovery of paenimucillin C. Paenimucillin C inhibits the growth of multidrug-resistant (MDR) Acinetobacter baumannii clinical isolates, as well as other Gram-negative bacterial pathogens. In a rat cutaneous wound model, it completely sterilized MDR A. baumannii wound infections with no sign of rebound. Mechanistic studies point to a membrane-associated mode of action that results in leakage of intracellular contents. IMPORTANCE Natural product-inspired antibiotics have saved millions of lives and played a critical role in modern medicine. However, the emergence of drug-resistant pathogens is outpacing the rate at which new clinically useful antibiotics are being discovered. The lack of a means to combat infections caused by multidrug-resistant (MDR) Acinetobacter baumannii is of particular concern. The sharp increase in cases of MDR A. baumannii infections in recent years prompted the CDC (https://www.cdc.gov/drugresistance/biggest_threats.html) and WHO (http://www.who.int/medicines/publications/global-priority-list-antibiotic-resistant-bacteria/en/) to list this pathogen as a "serious threat" and "critical pathogen," respectively. Here we report a new antibiotic, paenimucillin C, active against Gram-negative bacterial pathogens, including many clinical isolates of MDR A. baumannii strains. Mechanistic studies point to membrane disruption leading to leakage of intracellular contents as its antibacterial mode of action. Paenimucillin C sterilizes MDR A. baumannii infections in a rat cutaneous wound model with no sign of rebound infection, providing a potential new therapeutic regimen.

Keywords: Acinetobacter baumannii; antibiotics; syn-BNP; wound infections.

FIG 1

Overview of the syn-BNP approach. (a) syn-BNPs are synthesized based on a bioinformatic analysis of biosynthetic gene clusters found in sequenced bacterial genomes and assayed for bioactivities. (b) As any single bioinformatic prediction may not perfectly represent the natural product encoded by a gene cluster, additional bioinformatic analyses are used to expand the structural diversity to explore around each primary bioactive hit.

FIG 2

The original syn-BNPs (paenimucillin A [1F] and B [1W]) were synthesized according to Minowa and NRPSPredictor2, two algorithms commonly used for prediction of natural product structures from NRPS biosynthetic gene clusters. In this study, we systematically explored the influence of the N-terminal substituent because it is known to influence the antibacterial activity of peptides and the bioinformatic algorithms we used were not designed to provide detailed predictions of this substituent. “>” denotes a MIC of >128 μg/ml (the highest concentration tested).

FIG 3

Rat cutaneous wound model study. In the tolerability assessment (a), vehicle and 4F both showed weight loss within the expected range typical of administration of analgesia (buprenorphine). The graph shows the average data tested on two independent wounds. In the efficacy assessment (b), the rats were infected on day 0 with multidrug-resistant A. baumannii AB5075 (500 CFU). The infected wounds were then treated twice daily, starting at 0.5 h postinoculation, with vehicle, tetracycline (10 mg/ml), or 4F (10 mg/ml). 4F completely sterilized the wound, with no rebound throughout the study. The graph shows the average data from four independent wounds per treatment per time point; error bars represent ±1 standard deviation.

FIG 4

Mechanism of action studies using paenimucillin C (4F). (a) Among strains resistant to 4F, genome sequencing identified common mutations in three genes (pla1, pld, and pgsA) within the cell membrane phospholipid metabolism network. Abbreviations: CDP-DAG, CDP diacylglycerol, CL, cardiolipin; G3P, glycerol-3-phosphate; LPA, lysophosphatidic acid; PA, phosphatidic acid; PS, phosphatidylserine. Panels b and c show results from permeability assays using DIBAC₄ (b) and SYTOX (c), fluorescent dyes for probing membrane depolarization and pore formation, respectively. Polymyxin and Triton X-100 serve as positive controls for membrane depolarization and pore formation, respectively.