Design and Discovery of Novel Bacteriocins

Abstract

Since the discovery of Penicillin in 1928, global demand for antibiotics has continued to grow year after year. Despite our reliance upon them, few new classes of antibiotics have been discovered since the 1980’s. Our existing arsenal of therapeutics have been systematically overused and misused, driving the emergence of Antimicrobial Resistance (AMR). The rise of AMR, that is, pathogens which are resistant to conventional antibiotic treatments, poses a significant threat to human health which, if not addressed, is predicted to contribute to the deaths of millions around the world annually by the year 2050. Novel therapeutics and a globally concerted policy change are urgently required to avoid entering a post-antibiotic era, where simple medical procedures and common infections could become life threatening. This study sought to address this threat in some part by the discovery of novel antimicrobial molecules which could serve to replace or augment the use of conventional antibiotics. The subjects of this thesis, 4-Helix Bundle (4-HB) Bacteriocins, are a small group of potent antimicrobials produced within the bacterial communities which share a common topology and physicochemical properties, and the few reported examples have demonstrated promising therapeutic potential.
A bioinformatic approach was taken to identify putative 4-HB Bacteriocins in silico from the vast public protein databases which have been populated in recent years, encompassing over 580 million sequences. A total of 234 naturally occurring candidates were identified using a combination of sequence-based and structure-based homology search strategies, dramatically increasing our knowledge of 4-HB Bacteriocins over the small numbers that had been previously reported. Additionally, methods for the generation of non-natural sequences were also explored, including the development of a Variational Autoencoder (VAE) model which was able to rapidly produce over 90 synthetic sequences which replicated the physicochemical properties and structural fold of the 4-HB Bacteriocin. Furthermore, the putative functionality of these molecules was experimentally validated. A Cell-free Protein Synthesis platform for the screening of 4-HB Bacteriocin candidates was developed and tested alongside an orthogonal Solid Phase Peptide Synthesis approach and the false negative rate of each was compared. Ultimately an efficient screening strategy composing sequential screening using both methods was proposed. Finally, the VisABLE® recombinant expression platform was used to isolate high-expressing Pichia pastoris clones expressing 22 of the targets, and isolation of 4 candidates in their native form was demonstrated.
This work has led to the discovery of over 400 novel 4-HB Bacteriocins summarised in Supplementary Table 22. Data included here validates the use of mining of public databases and the use of generative AI methods for Bacteriocin discovery. Promising antimicrobial activity was demonstrated against several clinically relevant target organisms, including Vancomycin Resistant Enterococcus faecalis, Acinetobacter baumannii and the yeast Candida albicans. Further investigation of the promising molecules identified here is expected to lead to the development of much needed, novel antimicrobial therapies.

Awarding Institution(s)

University of Plymouth

Award Sponsors

Ingenza, Royal Commission for the Exhibition 1851

Supervisor

Mathew Upton, Sabine Lengger

Document Type

Thesis

Publication Date

2025

Embargo Period

2028-09-30

Deposit Date

2025-10-03

Creative Commons License

Creative Commons Attribution-NonCommercial 4.0 International License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License

This document is currently not available here.

This item is under embargo until 30 September 2028

Share

COinS