Comprehensive prediction of secondary metabolite structure and biological activity from microbial genome sequences

Michael A Skinnider,Robyn J Maclellan,My P T Cao,Mathusan Gunabalasingam,Norman Spencer,Andrew L H Webster,Q Huy To,Nathan A Magarvey,Chris A Dejong,Annabelle Pfeifle,Chad W Johnston,Nishanth J Merwin,Haoxin Li,Dan Peter Wallace,Michael R M Ranieri,Agata M Kieliszek

doi:10.1038/s41467-020-19986-1

Abstract

Novel antibiotics are urgently needed to address the looming global crisis of antibiotic resistance. Historically, the primary source of clinically used antibiotics has been microbial secondary metabolism. Microbial genome sequencing has revealed a plethora of uncharacterized natural antibiotics that remain to be discovered. However, the isolation of these molecules is hindered by the challenge of linking sequence information to the chemical structures of the encoded molecules. Here, we present PRISM 4, a comprehensive platform for prediction of the chemical structures of genomically encoded antibiotics, including all classes of bacterial antibiotics currently in clinical use. The accuracy of chemical structure prediction enables the development of machine-learning methods to predict the likely biological activity of encoded molecules. We apply PRISM 4 to chart secondary metabolite biosynthesis in a collection of over 10,000 bacterial genomes from both cultured isolates and metagenomic datasets, revealing thousands of encoded antibiotics. PRISM 4 is freely available as an interactive web application at http://prism.adapsyn.com.

Highlights

Novel antibiotics are urgently needed to address the looming global crisis of antibiotic resistance
To evaluate the accuracy of PRISM 4, we assembled a comprehensive set of 1281 biosynthetic gene clusters (BGCs) with known products from public databases and extensive literature curation, subject to multiple rounds of manual review by a team of natural products chemists to correct errors in chemical structures or the boundaries of deposited nucleotide sequences (Methods)
To quantify the similarity of predicted structures to the true cluster products, we calculated the Tanimoto coefficient[13] (Tc) between real and predicted structures from each cluster, a measure of chemical similarity that reflects the fraction of substructures shared between the two molecules, and compared these to predicted and true structures from random BGCs pairs (Methods)

Summary

Introduction

The biosynthetic pathways responsible for the production of these molecules have been honed over long evolutionary time scales in order to provide microbes with competitive advantages in their natural environments[2] These pathways are encoded within the genomes of the producing organisms, and comparative genomics studies have suggested a wealth of novel antibiotics encoded in the genomes of both culturable and unculturable organisms that remain to be discovered[3,4,5]. PRISM 4 includes 1772 hidden Markov models (HMMs) and implements 618 in silico tailoring reactions in order to predict the chemical structures of 16 different classes of secondary metabolites, making it a comprehensive resource to link microbial genome sequence information to the natural antibiotics encoded within (Fig. 1c, Supplementary Table 1, and Supplementary Data 1)

Methods

Results

Conclusion