Probing the essential genome of Burkholderia cenocepacia K56-2 with CRISPRi to identify bioactive compounds with novel mechanisms of action(s)
Loading...
Date
2024-08-16
Authors
Rahman, A S M Zisanur
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
A significant bottleneck in the discovery of novel antibiotics is the poor understanding of
the mechanism of action as it impedes compound prioritization for further development. Chemical-genetics,
which links bioactive compounds with their cellular targets, offers a streamlined
approach to antibiotic discovery. Integrating empirical investigations with machine learning (ML)
algorithms enables rapid exploration of large compound libraries in silico to identify promising
antimicrobial compounds.
The objective of this thesis was to develop an approach that increases the probability of
early antimicrobials discovery together with the determination of their mechanism of action. I
began by validating that a ML model could effectively learn from large training data and rapidly
screen compound libraries to predict bioactivity. Therefore, a machine learning model trained with
a previous high throughput screening (HTS) dataset was employed for bioactivity prediction on
large compound libraries. The ML-approach yielded a 30-fold increase in hit rate compared to the
conventional HTS approach and led to the identification of PHAR(A), a growth inhibitory
compound with novel chemical scaffold.
Because antibiotics usually target essential gene products, my project focused on probing
bacterial essential genome to explore a broader antibiotic target space. I constructed and
characterized an arrayed essential gene mutant library in Burkholderia cenocepacia K56-2 using
CRISPR interference (CRISPRi). By analyzing the clonal growth parameters of the individual
mutants and their depletion from a pool, an optimized CRISPRi-mediated pooled library of
essential genes (CIMPLE) was created. Upon exposure of CIMPLE to a bacterial growth inhibitor
or a novel class, PHAR(A), changes in fitness of the individual mutants were detected by
CRISPRi-Seq. We identified peptidyl-tRNA hydrolase as the putative cellular target of PHAR(A).
Overall, this thesis demonstrates the sensitivity and specificity of our strategy in
uncovering the mechanisms of action of novel antimicrobials against targets that otherwise would
have eluded a conventional drug discovery pipeline. It underscores how combining in silico
approaches with experimental studies can accelerate the discovery of novel antimicrobials. This
work can be extended to generate large chemical-genetic interaction profiles to train a ML platform
to predict the bioactivity and mechanism of action of active compounds.
Description
Keywords
Essential Genome, CRISPRi, CRISPRi-Seq, Chemical-genetics, Machine learning, Burkholderia cenocepacia K56-2, Antibiotic discovery