Strain-specific proteogenomics accelerates the discovery of natural products via their biosynthetic pathways

Abstract

The use of proteomics for direct detection of expressed pathways producing natural products has yielded many new compounds, even when used in a screening mode without a bacterial genome sequence available. Here we quantify the advantages of having draft DNA-sequence available for strain-specific proteomics using the latest in ultrahigh-resolution mass spectrometry for both proteins and the small molecules they generate. Using the draft sequence of Streptomyces lilacinus NRRL B-1968, we show a >tenfold increase in the number of peptide identifications vs. using publicly available databases. Detected in this strain were six expressed gene clusters with varying homology to those known. To date, we have identified three of these clusters as encoding for the production of griseobactin (known), rakicidin D (an orphan NRPS/PKS hybrid cluster), and a putative thr and DHB-containing siderophore produced by a new non-ribosomal peptide sythetase gene cluster. The remaining three clusters show lower homology to those known, and likely encode enzymes for production of novel compounds. Using an interpreted strain-specific DNA sequence enables deep proteomics for the detection of multiple pathways and their encoded natural products in a single cultured bacterium.

Fig. 1

Workflow of “genome-enabled” PrISM, an extended form of a proteomics-based approach to discover expressed gene clusters in microbial systems. From left to right, the workflow entails microbial culture, striation of the proteome according to mass by SDS-PAGE, in-gel digestion followed by LC-MS/MS and searching of peptide fragment ion data against the assembled genome of a specific microbial strain. The structural elements of a molecule are determined using the predicted substrates for domains in the assembly line and the rules of co-linear biosynthesis. This informs a targeted metabolomics-based approach to identify the small molecule being produced.

Fig. 2

Intact mass spectrum of novel NRPS compound without (top) and with (bottom) the addition of 2,3-d₂ labeled threonine (inset). The isotope pattern is consistent with the addition of three threonine monomers.

Fig. 3

Draft structure of a new catecholate siderophore and its high-resolution MS² analysis. Peaks highlighted in red are consistent with the predicted fragmentation of this compound. The table below shows the explanation given for the most abundant signals in the tandem mass spectrum.

Fig. 4

High-resolution tandem mass spectrum confirming the presence of rakicidin D. Peaks highlighted in red are consistent with the expected fragment ions from rakicidin D.

Fig. 5

A) Map of contig containing putative biosynthetic genes of rakicidin D. NRPS and PKS genes are shown in blue, while transport genes are shown in orange. rakG and rakD, which correspond with the most striking features of rakicidin D, are colored in green and red, respectively. B) Bioinformatic prediction of the domain architecture of NRPS and PKS genes in the biosynthetic cluster. Proteins identified using MS-based proteomics are highlighted in red. C) Structure of rakicidin D. The methyl group of the sarcosine substituent is circled in green, corresponding to the NRPS protein RakG which contains an A-domain specific to glycine as well as an N-methyltransferase domain. The hydroxyl group of the hydroxyasparagine monomer is circled in red, corresponding to RakO, a putative asparagine oxygenase.