Practical applications of genomics to natural product discovery and biosynthesis

Natural products, generally defined as metabolites of biotic origin, have a long history as medicines and have remained a fruitful source of new drug leads for multiple diseases in the modern age. Arguably the largest asset of natural products, and a pitfall of synthetic compound libraries, is the remarkable diversity of structure and chemical group functionality that discloses them as Nature’s handiwork and is critical for the bioactivities that many of them possess. Despite this precedent, characterization of new chemical entities is met with a significant challenge: rediscovery of known compounds. For compound discovery to continue translating to useful, practical chemical matter, it is essential that strides be made to streamline and augment current methodology to prioritize novelty. Modern genomics provides a potential resolution to these issues: as the financial and technical hurdles for whole-genome sequencing are reduced, annotating the biosynthetic capabilities of sequenced organisms becomes more accessible and reliable. It follows that gene clusters can be analyzed before undertaking efforts to isolate and characterize compounds, minimizing costs and resources. However, this approach is dependent upon a robust understanding of gene function as it translates to natural product structure, and assignment of biosynthetic enzyme/cluster function is drastically outpaced by the availability of genome sequences, leading to an excess of genomic data that cannot be optimally exploited. We describe the current state of these challenges as it relates to the discovery of new natural products in Chapter 1, and developments in “reverse genetics” techniques which seek to address these shortcomings. Our group has developed a platform to automate analysis of biosynthetic gene clusters and structure prediction, described in Chapter 2, which we applied to an understudied family of natural products. RODEO (Rapid ORF Description and Evaluation Online) was designed as a practical tool for high-throughput analysis of natural product gene clusters, and initially focused on lasso peptides, where our application of hidden Markov models, heuristic and sequence motif analysis, and machine learning revealed new scaffolds. This genome-mining approach enabled by RODEO prioritized strains with unique biosynthetic clusters and predicted lasso peptides, resulting in the isolation of six new compounds with unique structural features including one with a new ribosomal peptide post-translational modification and another with an unprecedented “handcuff” topology. In Chapter 3, a family of bioinformatically-identified prenyltransferases is investigated that has members in several bacterial pathogens. Starting from genomic data, these terpenoid synthase enzymes are biochemically characterized and found to be functionally divergent from previously identified examples. Sequence analysis suggests residues which are believed to influence activity, and mutagenesis was used to provide a starting point for correlating residues to catalysis. The enzyme family responsible for formation of azol(in)e heterocycles in ribosomal peptide natural products is revealed in Chapter 4 to have a new biosynthetic role. Characterization of the bottromycin heterocyclases suggest an instance of Nature using a common amide-backbone activation mechanism for different reactions - in this case a macroamidine heterocyclization. Substrate tolerance and recognition are explored for the unique substrate, which contains bipartite peptide regions for modification and binding by the enzymes. More broadly, the genomic distribution of atypical heterocyclases are investigated, indicating a high probability of discovering more examples of new biosynthetic chemistry and consequently, new natural product scaffolds. In Chapter 5, we extend this approach to a different area of YcaO biosynthetic enzymes which are implicated to not perform heterocyclizations at all, but instead appear to utilize this mechanism to induce thioamidation, an uncommon and divergent post-translational modification. Genetic evidence is found which supports the “TfuA-associated YcaOs” as a subfamily capable of thioamidation of ribosomal natural products. Whole genome sequencing of “orphan” natural products containing thioamides but no associated gene clusters revealed consistent presence of both TfuA-YcaO genes. The co-occurrence of the TfuA-YcaO pair were then used to guide screening for new natural products containing thioamides and aid in the structural identification of a novel thiopeptide with this post-translational modification.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Chemistry

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2018

Author dc:creator

• Schwalen, Christopher Joseph

Contributors dc:contributor

• Mitchell, Douglas A.
• van der Donk, Wilfred A.
• Oldfield, Eric
• Metcalf, William W.

• Natural product
• Biosynthesis
• YcaO
• Lasso peptide
• Genomic

Statement dc:rights

• Copyright 2017 Christopher J. Schwalen

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/99245

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/99245