Title: Understanding the Nephrotoxic Origin of Lipopeptide Antibiotics
Abstract: Unexpected nephrotoxicity and hepatotoxicity are leading causes of antibiotic attrition. In vivo preclinical models are good predictors of antibacterial efficacy but often poor predictors of safety, including nephrotoxicity. Improved predictive translational models for meaningful assessment of structure-toxicity relationships are therefore desirable. As part of our efforts to develop octapeptin, a novel lipopeptide antibiotic class with broad-spectrum Gram-negative activity, we sought to compare the predictive utility of different in vitro and in vivo models of nephrotoxicity. This work forms part of CARB-X’s Portfolio Acceleration Tools (PATs); an initiative designed to streamline antibiotic development by identifying developmental hurdles faced by many drug developers.
Methods: A series of polymyxin lipopeptides were benchmarked against several in vivo and in vitro preclinical models of nephrotoxicity. Using mouse and rat models we ranked compounds according to histopathology scoring, urinary biomarker expression, and kidney/plasma exposure. The lipopeptides were also tested against primary (HRPTEpiC) and immortalised (HK-2) proximal tubule epithelial cells using high content analysis and ATP luminescence. Polymyxin, a known nephrotoxicant, was used as a baseline comparator.
Results: In vitro data confirmed that polymyxin was cytotoxic against proximal tubule epithelial cells. However, repeated subcutaneous administration of polymyxin to mice and rats did not elicit consistent and reproducible histological damage, and levels of excreted urinary biomarkers considered predictive of nephrotoxicity were often variable. Furthermore, polymxin analogues currently under advanced clinical development or previously withdrawn due to toxicity were often indistinguishable from polymyxin itself, confounding its use as a baseline comparator. When given as a single subcutaneous dose in mice, some lipopeptides, including polymyxin, did not accumulate in the kidneys whereas others persisted even after 72 h. However, the propensity for lipopeptides to accumulate in the kidneys was not necessarily indicative of increased nephrotoxic risk. Collectively we demonstrate that the origins of lipopeptide nephrotoxicity is multifactorial, and that the variability observed in rodent-based nephrotoxicity models may limit their predictive value.