Project Objectives

• Extend an established model of human airway tissue, which has working cilia, produces mucus, and forms a barrier, to three bacteria on the WHO priority pathogen list that cause serious lung infections: Pseudomonas aeruginosa, Klebsiella pneumoniae, and Acinetobacter baumannii.
• Measure how up to eight widely used antibiotics, from six drug classes, kill four to five strains of each species, both in standard lab broth and on the tissue.
• Track how well each antibiotic passes from beneath the tissue, which stands in for the bloodstream, into the mucus on its surface, where lung infections take hold.
• Compare the results with existing animal and patient data using computational modelling, led by Uppsala University, against success criteria set before the experiments begin.
• Map where bacteria survive antibiotic treatment within the tissue, such as inside cells or in biofilms, and compare these patterns with lung samples from patients.
• Publish standardised protocols, datasets, and model code openly so other labs and companies can use the approach.

3Rs Impact

• Involves no new animal experiments. Every comparison draws on existing data, including published studies, rodent datasets from the international COMBINE consortium, and industry data where available.
• Tests directly whether human tissue gives results that match those from animal studies, and for which types of drug and measurement. That is the evidence needed before the model can replace animal studies in practice.
• Where the model performs well, it could take over mouse studies in early antibiotic development, including testing drugs against multiple strains of each species, which multiplies the number of animal studies needed.
• Even partial success is useful, because it defines the specific contexts where animals are no longer needed.
• Standardised, openly published protocols and training give other labs and companies a practical route to adopting the method.

Background

At the Biozentrum, University of Basel, Dr Anna Leoni Swart has developed a model of the human airway lining, grown from human airway cells on a permeable membrane with its upper surface exposed to air. Like real airways, it produces mucus, moves it with beating cilia, and forms a protective barrier. Developed within the NCCR AntiResist research consortium, the model has already revealed how Pseudomonas aeruginosa breaks through the airway lining by invading mucus-producing cells. Because antibiotics can be added beneath the tissue and bacteria on top, it shows both how a drug reaches an infection and how the bacteria respond.

Antibiotic development still relies heavily on animal studies, mostly in mice, to establish how a drug reaches the site of infection and how well it kills bacteria there. Each candidate must be tested against several strains of each species, which multiplies the number of studies. Mouse lungs also differ from human lungs in anatomy, immune response, and how drugs are handled, so results do not always translate to patients. A human tissue model could address both problems, but only if there is evidence that its results match what happens in animals and in patients.

This project provides that test. The team is using antibiotics already in clinical use, for which extensive animal and patient data exist, so the tissue’s predictions can be checked against known outcomes. Early experiments with one strain and one antibiotic are encouraging. Bacteria grew and died more slowly on the tissue than in broth, as they do in animals, and the drug reached the mucus at a concentration similar to that reported in human lungs.

Professor Lena Friberg’s group at Uppsala University leads the modelling, and scientists at Roche contribute an industry perspective on strain selection and testing standards. The outcome will show where the model can replace animal studies in antibiotic development, and where it cannot yet.