Identifying antibiotic precursors by screening genetically encoded cyclic peptide libraries

Abstract

The spread of antibiotic resistant pathogens is an increasing problem, giving rise to a need for new antibiotics. To this end, an innovative high-throughput drop out screening strategy to produce cyclic peptides intracellularly, with the goal of identifying antibiotic molecules, has been developed. This approach uses split-intein circular ligation of peptides and proteins (SICLOPPS) libraires, in combination with next generation sequencing. As a proof of principle, an SXXXXX (where X is any of the 20 canonical amino acids) SICLOPPS library was built in E. coli. Up to 3.2 million different cyclic peptides were produced in the E. coli cell-line (one library member per bacterium), which only grow if the specific cyclic peptide does not inhibit normal cellular function. All DNA was then collected from the surviving cells, and deep sequenced along with the initial library to identify the missing sequences, which are potentially lethal to bacteria. A set of python scripts were developed to deconvolute the cyclic peptide sequences and to find common motifs in these cyclic peptides. Screening of a focused library of active tetrapeptide motifs led to the identification of a series of potential antibacterial candidates containing the conserved motif (S)MDIK. Expression of a cyclic peptide containing the active motif as a SICLOPPS construct was shown to inhibit bacterial growth. Furthermore, a hit peptide and a scrambled control were synthesised by solid phase peptide synthesis (SPPS) and tested in a zoning assay. Here, only the hit showed growth inhibition, whilst the scrambled cyclic peptide showed no activity. Furthermore, the active sequence motif could be confirmed via in vivo alanine-scanning, where sequences lost their activity when any one of the amino acids within the motif was mutated to an alanine. This suggests that the identified sequence has an antimicrobial effect which is sequence specific, and it may be a potential lead for a novel antibiotic (target). For target identification, a mutant generation screen with internally synthesised cSMDIKG was successfully carried out. Two mutant strains 1-1 and 1-3 were obtained and whole genome sequencing was performed alongside control strains. For strain 1-1, a single point mutation in a non-coding region was identified, whilst strain 1-3 contained a deletion of 7 genes, 6 of which belonged to the glc operon. Further analysis of these strains revealed that intein expression in strain 1-1 was strongly decreased. Investigation of the deleted genes of 1-3 as single gene knockout strains showed similar growth advantages under selective pressure as for the parental 1-3 strain. This might point towards those deleted genes being part of a targeted pathway of cSMDIKG. Further analysis is necessary to confirm this finding. As a second approach to target identification, a pull down with biotinylated cyclic peptide was performed with the aim to identify any potential binding partners. This is currently undergoing protein identification via proteomic analysis. In conclusion, a new pipeline for high throughput screening and motif analysis of genetically encoded cyclic peptide libraries was developed and successfully applied to identify a potential antibiotic precursor. The set of experiments presented can easily be adjusted to perform similar screens in more clinically relevant species. Furthermore, the software developed in this work can and will be applied for any subsequent screens involving SICLOPPS libraries.

More information

Identifiers

ORCID for Ali Tavassoli: orcid.org/0000-0002-7420-5063

ORCID for Matthias Baud: orcid.org/0000-0003-3714-4350

Catalogue record

Export record