This research leverages computational tools such as Hidden Markov Models (HMMs) and CluSeek to unearth natural product enzymes in soil bacteria (meta)genomes. It focuses on iminosugar natural products, which are sugar mimicking compounds in which the endocyclic oxygen is replaced by a nitrogen atom. Because of their similarity to sugars, they are an attractive class of glycosidase inhibitors and are currently marketed for the treatment of diabetes and certain lysosomal storage disorders. Soil bacteria produce a significant repertoire of these iminosugars, however, little is known about their biosynthetic enzymes. They have been overlooked by traditional computational tools such as antiSMASH due to their small chemical structure. After identifying natural producers of iminosugars, their genomes were sequenced and mined for enzyme candidates using personalized Hidden Markov Models. Discovered enzymes are subjected to functional validation through heterologous expression using Golden Gate-assembled expression plasmids. In addition, the recently developed CRISPR-Cas3 and the CRISPomyces-2 systems for unconventional Streptomyces iminosugar producers are being used to knock out enzyme candidates to confirm their functionality. Furthermore, the recently launched CluSeek tool is used to search for new iminosugar producers. From this output, putative iminosugar gene clusters from three different strains are being tested for functional characterization. Finally, a metagenomics approach is applied to uncover new microbial iminosugar enzymes. The metagenomic sampling workflow, eDNA extraction and long-read sequencing using Nanopore technologies were optimized. In previous work, we demonstrated widespread misclassification within Streptomyces through genome-based taxonomic analysis 1. Importantly, fragmentation during eDNA extraction remains a major bottleneck for long-read sequencing and thus for the assembly into metagenome assembled genomes. Therefore, attention was given to increasing the N50 to improve metagenome assembly, amounting to an N50 of 12.46 kb. This metagenomic data was further analyzed using personalized Hidden Markov Models and pBLAST to identify novel iminosugar biosynthesis enzymes.
Rop et al. (Thu,) studied this question.