Bioinformatics Platform Designs Therapies Against Multi-Drug Resistant Bacteria
By Allison Proffitt
April 22, 2021 | A University of Colorado Boulder team has published details of their Facile Accelerated Specific Therapeutic (FAST) platform in Nature Communications Biology, a platform designed to address antibiotic resistance. The FAST platform develops effective peptide nucleic acid (PNA) therapies against multi-drug resistant bacteria within a week, the authors write, by targeting non-traditional pathways/genes of bacteria, then performs in-situ synthesis, validation, and efficacy testing.
The rise of multi-drug resistant bacteria is worldwide public health issue, and resistance is developing faster than new antibiotics. “The large majority of drugs in clinical development for priority pathogens—as defined by the World Health Organization—do not introduce new classes or targets and are not pathogen-specific, thus presenting higher risk of rapid bacterial adaptation,” write the authors from the lab of Associate Professor Anushree Chatterjee in the paper (DOI: 10.1038/s42003-021-01856-1)
Peptide nucleic acids (PNA) present an antisense strategy that offer advantages in stability, binding strength, and mismatch discrimination compared to similar technologies, the authors write, but PNAs do have serious drawbacks, notably tedious design and screening processes. The team developed FAST to be a “semi-automated” platform to speed that design and screening process.
The FAST platform process takes about a week. First, the genome sequence of the target pathogen is input into the PNA Finder Toolbox, a bioinformatics tool that designs PNA candidate sequences and flags any off-target activity in less than 10 minutes. Over the next four days or so, promising candidates are synthesized using high throughput, automated solid-phase synthesis chemistry. Purification of the PNA product is done using HPLC and validation is performed through LC–MS. Candidates are tested in parallel on a panel of multidrug-resistant (MDR) clinical isolates to validate the toolbox’s predictions—testing candidates both alone and in combination with other antibiotics. Results can cycle back to the PNA Finder Toolbox to improve upon specificity predictions. Finally, promising PNAs are delivered to treat intracellular infections of bacteria.
The PNA Finder Toolbox is the heart of the FAST platform, combining custom Python 3.7 scripts with the alignment and analysis programs Bowtie 2, SAMTools, and BEDTools to deliver two functions: Get Sequences and Find Off-Targets. Get Sequences creates a library of PNA candidate sequences by targeting mRNA translation start codons for the pathogen genomes entered by user, and flags solubility and self-complementarity issues, as well as a protein interaction network analysis via the STRING database. With the output of Get Sequences, the Find Off-Targets function searches for incidental inhibitory alignments by searching a non-target genome for highly similar sequences—1-bp mismatch or less—within a 20-nucleotide range of the start codon.
“The PNA Finder toolbox is a dynamic component and, as more PNA antibiotics are tested, will incorporate feedback into its algorithm to improve the PNA target suggestion list based on weighted parameters correlated with the PNA’s predicted efficacy,” the authors write.
The researchers tested their platform against five MDR Enterobacteriaceae clinical isolates: two Escherichia coli isolates, two Klebsiella pneumoniae isolates, and one Salmonella enterica serovar Typhimurium isolate.
Using the PNA Finder Toolbox, the team ran both Get Sequences and Find Off-Targets in several cycles, choosing PNAs effective against all three reference genomes and eliminating PNAs expected to have low solubility or exhibit self-complementarity. Their first library contained of 260 PNA candidates targeting essential genes and 3,524 candidates targeting non-essential genes within E. coli. The final list included 71 essential and 243 non-essential target gene candidates. The process took under 10 minutes.
After synthesis and purification, they tested their PNAs on MDR clinical isolates that all showed phenotypic and genotypic resistance. “PNA treatments demonstrated high selectivity to their target strains. Of the 45 total treatments, 34 were predicted to have homology to the clinical isolate tested and eleven did not. Seven out of eleven PNA treatments without sequence homology to the clinical isolate showed no growth reduction; the remaining four showed less than 10% growth inhibition,” the authors write. “Of the 34 clinical isolate-PNA pairs predicated to have homology, 28 showed significant growth inhibition, with six monotherapy treatments reducing growth by more than 97%.”
“The final obstacle for PNA antibiotic viability that the FAST platform seeks to address is the molecules’ poor uptake into mammalian cells to treat mammalian intracellular infections,” the authors said. The FAST platform, therefore, uses a probiotic delivery system: a Type III secretion system (T3SS), a gram-negative bacterial machine for invasion of the eukaryotic host cell combined with a lysis switch to release the therapeutic. The delivery system, they say, proved to be an effective solution for the delivery of PNA to treat intracellular infections.