Harnessing long-read sequencing for antibiotic discovery - Allison Guitor

Allison Guitor (Wright Lab, McMaster University) began by introducing the urgent nature of antibiotic resistance and the discovery of new antibiotics. With antibiotic resistance on the rise, she noted that many organisations have listed pathogens for which novel antibiotics are now required for treatment. She highlighted Gram-negative bacteria of critical priority listed by the WHO: Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacteriaceae species, all of which have developed carbapenem resistance and the latter of which is also resistant to third-generation cephalosporins.

One of the many challenges in the fight against antibiotic-resistant bacteria, Allison explained, is the natural physiology of Gram-negative pathogens: few antibiotics are able to enter Gram-negative bacterial cells, rendering any compounds that cannot pass through the cell membrane as useless. Allison introduced how she and her team are looking to nature to find compounds that organisms have evolved to produce that can pass through this cell wall. Presenting a timeline of antibiotic discovery, starting with the isolation of Penicillin F from mould in 1928, Allison highlighted how the majority of available antibiotics are derived from natural products. In the 1990s, she noted, the discovery of new antibiotics from natural sources was considered exhausted, so research shifted towards synthetics. However, Allison and her colleagues aim to continue to drive forwards the search for novel antibiotics from environmental sources: "in our lab, we're looking into digging up new antibiotics".

Allison introduced the Wright Actinomycete Collection (WAC): 11360 environmental strains obtained from soil, from the Actinomycete phylum, which is particularly good at producing compounds that could represent novel antibiotics. Identifying and isolating these compounds can be achieved through a number of methods. Traditionally, bacteria are screened for their ability to produce novel compounds, producing plates of "natural product libraries"; these can then be tested to see whether they are active against bacteria. A common challenge in antibiotic discovery, Allison explained, is that previously discovered antibiotics are often re-discovered; this replication can be targeted by knocking out genes that produce already-discovered antibiotics.

Allison described how genome sequencing presents "untapped potential" for this application, through sequencing and identification of biosynthetic gene clusters that may be capable of producing as-yet undiscovered antibiotics. a 2014 survey from NCBI found that Streptomycetales are a rich source of potential new antibioitcs, featuring a per-genome biosynthetic gene cluster average of 21.6. Initially, the team used short-read sequencing to assemble Streptomyces species genomes. However, these genomes present many challenges for traditional sequencing technologies, making their assembly difficult: they are GC-rich, highly repetitive, feature terminal inverted repeats and have high genome instability. The short-read assemblies generated were comprised of 300-500 contigs, making analysis of the biosynthetic gene clusters very difficult to elucidate, especially where they were split across different contigs; Allison showed an example bandage plot of a highly fragmented assembly and the resulting gene structure prediction.

The team then decided to use the MinION device to sequence these complex genomes using long nanopore reads, without the need for amplification. The environmental samples were barcoded without PCR using the Native Barcoding Kit or Rapid Barcoding Kit, prepared for sequencing with the Ligation Sequencing Kit, and sequenced in multiplex on the MinION Mk1b. The resulting data was basecalled with Guppy, analysed via an in-house pipeline and assembled with Unicycler.

Allison displayed the results of two bacterial genome assemblies: the first using only short reads, the second after the addition of long nanopore reads to form a hybrid assembly. Whilst the short-read assembly was highly fragmented and not contiguous, the hybrid approach enabled assembly of the bacterial chromosome into a single contig, enabling the bacterial gene cluster to be resolved - "we were able to completely resolve this and move forward with this project." She demonstrated how long nanopore reads consistently improved both the contiguity and contig length N50 of the Streptomyces sp. genome assemblies. After candidate biosynthetic gene clusters are identified from the hybrid assemblies, Allison described how homologous recombination can then be used to then be used to transfer these to an optimised host; nanopore sequencing can then be used to confirm successful integration of the biosynthetic gene clusters. She showed how her colleague designed a heterologous host for this process; a transferred biosynthetic gene cluster exhibited "great activity" in this superhost. Allison then showed disc arrays demonstrating activity of the produced compounds against bacteria.

Allison concluded her talk by displaying the team's final workflow, harnessing long-read sequencing for antibiotic discovery. She highlighted how this method enables the production of high-quality genomes, from which biosynthetic gene clusters can be predicted and prioritised; these can then be selected for heterologous expression in an optimised host, hopefully enabling their goal for the future: the production and purification of new antibiotics.