Deciphering the impact of the cystic fibrosis lung microenvironment on antibiotic activity

Promovendus/a: Sara Van den Bossche

Promotor(en): Prof. dr. Aurélie Crabbé, Prof. dr. Tom Coenye

Chronic airway infections with Pseudomonas aeruginosa are associated with a higher morbidity and
mortality in cystic fibrosis (CF) patients. Despite aggressive antibiotic treatment, P. aeruginosa can reside
and evolve in CF airways. As in vitro antibiotic activity against P. aeruginosa and treatment efficacy in CF
patients poorly correlate, treatment choice is typically empiric and/or based on previous patient
experiences rather than on antibiotic susceptibility test (AST) results. The striking contrast between the
simplicity of standard in vitro AST and the complex and personalized nature of the CF lung
microenvironment, consisting of both host and microbial factors, is believed to explain (at least partially)
this discordance. Hence, the present thesis aims at deciphering the impact of key CF lung
microenvironmental factors on antibiotic activity against P. aeruginosa and at developing in vitro models
that could more closely predict treatment efficacy.
As P. aeruginosa isolates from one patient sample often differ in phenotype (including antibiotic
resistance) and AST is typically performed on only a few isolates, the first part of this thesis evaluated the
impact of intrasample diversity on AST outcome. To this end, 15 CF patients provided one sputum sample
and 30 P. aeruginosa isolates were collected per sample and subjected to AST for three common
antibiotics (aztreonam, ceftazidime and tobramycin). Intrasample diversity in antibiotic resistance for at
least one antibiotic was found in 80% of patient samples and strongly correlated to categorical
disagreement (i.e. classification of isolates into different susceptibility categories). Given the clear impact
of intrasample diversity on AST outcome, pooling isolates before AST was evaluated as a possible
approach to lower diversity in AST outcome. Pooling an increasing number of isolates gradually decreased
diversity, and pooling nine isolates abolished diversity in AST outcome for all patient samples. Also, in
most patient samples, pooling nine isolates decreased the level of categorical disagreement and did not
decrease the frequency of resistance detection.
In the second part of this thesis, we aimed to assess whether the lung microbiota is an influencer of
antibiotic activity. Hence, the antibiotic activity of four common antibiotics (tobramycin, colistin,
ciprofloxacin and aztreonam) was evaluated in patient-specific microbiota-conditioned medium (MCM).
To generate this patient-specific MCM, the sputum microbiota of ten individual CF patients was isolated
and cultured in vitro in conditions similar to CF airways (i.e. low oxygen levels and synthetic cystic fibrosis
sputum medium) and the cell-free supernatant was harvested. The composition of the microbiome was
compared before and after in vitro culturing through metagenomics and the metabolites secreted in the
MCM were identified using a metabolomics approach. The most abundant bacteria in the original sputum
sample remained present after in vitro culturing. Patient-derived MCM impacted antibiotic activity against
P. aeruginosa in a patient-specific manner, with antibiotic- and strain-specific effects. Clustering patient
samples, based on the composition of the in vitro cultured microbiome, identified three clusters with a
distinct microbiome profile. Each cluster exhibited a potentiating effect on a single antibiotic. Comparing
the abundance of metabolites between clusters revealed multiple candidate potentiators of tobramycin,
colistin or ciprofloxacin.
In the last part of this thesis, we took some important first steps in developing airway epithelial cell models
that could be used in infection studies to evaluate the antibiotic activity against P. aeruginosa. Firstly, 3-
dimensional in-vivo-like aggregates of bronchial epithelial cell lines were grown in a rotating wall vessel
bioreactor and the expression of differentiation markers was evaluated through immunostaining followed
by confocal laser scanning microscopy. Secondly, as evaluating epithelial cell viability is indispensable for
in vitro experiments on host-microbe interactions, a commonly used viability assay, the lactate
dehydrogenase (LDH) assay, was assessed for its suitability. Bacterial interference with LDH activity was
found for four out of eight tested lower airway bacteria and the mechanism behind this interference was
elucidated. Finally, a modified protocol was developed that circumvented bacterial interference with the
LDH assay.
Altogether, the findings of this thesis broaden our knowledge on the impact of the (patient-specific) lung
microenvironment in CF on antibiotic activity and provide us with strategies to incorporate the (patientspecific) microenvironment in in vitro models.