Despite the wide availability of antibiotics, infectious diseases remain a leading cause of death worldwide.1 In the absence of new therapies, mortality rates due to untreatable infections are predicted to rise more than 10-fold by 2050. Natural products (NPs) made by cultured bacteria have been a major source of clinically useful antibiotics. In spite of decades of productivity, the use of bacteria in the search for new antibiotics was largely abandoned due to high rediscovery rates.2, 3 As only a fraction of bacterial diversity is regularly cultivated in the laboratory and just a fraction of the chemistries encoded by cultured bacteria is detected in fermentation experiments, most bacterial NPs remain hidden in the global microbiome. In an effort to access these hidden NPs, we have developed a culture-independent NP discovery platform that involves sequencing, bioinformatic analysis, and heterologous expression of biosynthetic gene clusters (BGCs) captured on DNA extracted from environmental samples (eDNA). Here, we describe the application of this platform to the discovery of the malacidins, a distinctive class of antibiotics that are commonly encoded in soil microbiomes but have never been reported in culture-based NP discovery efforts. The malacidins are active against multidrug-resistant (MDR) pathogens, sterilize MRSA skin infections in an animal wound model, and did not select for resistance under our laboratory conditions.

Lichens, which are defined by a core symbiosis between a mycobiont (fungal partner) and a photobiont (photoautotrophic partner), are in fact complex assemblages of microorganisms that constitute a largely untapped source of bioactive secondary metabolites. Historically, compounds isolated from lichens have predominantly been those produced by the dominant fungal partner, and these continue to be of great interest for their unique chemistry and biotechnological potential. In recent years it has become apparent that many photobionts and lichen-associated bacteria also produce a range of potentially valuable molecules. There is evidence to suggest that the unique nature of the symbiosis has played a substantial role in shaping many aspects of lichen chemistry, for example driving bacteria to produce metabolites that do not bring them direct benefit but are useful to the lichen as a whole. This is most evident in studies of cyanobacterial photobionts, which produce compounds that differ from free living cyanobacteria and are unique to symbiotic organisms. The roles that these and other lichen-derived molecules may play in communication and maintaining the symbiosis are poorly understood at present. Nonetheless, advances in genomics, mass spectrometry and other analytical technologies are continuing to illuminate the wealth of biological and chemical diversity present within the lichen holobiome. Implementation of novel biodiscovery strategies such as metagenomic screening, coupled with synthetic biology approaches to reconstitute, re-engineer and heterologously express lichen-derived biosynthetic gene clusters in a cultivable host, offer a promising means for tapping into this hitherto inaccessible wealth of natural products.

In this study, we compare biosynthetic gene richness and diversity of 96 soil microbiomes from diverse environments found throughout the southwestern and northeastern regions of the United States. The 454-pyroseqencing of nonribosomal peptide adenylation (AD) and polyketide ketosynthase (KS) domain fragments amplified from these microbiomes provide a means to evaluate the variation of secondary metabolite biosynthetic diversity in different soil environments. Through soil composition and AD- and KS-amplicon richness analysis, we identify soil types with elevated biosynthetic potential. In general, arid soils show the richest observed biosynthetic diversity, whereas brackish sediments and pine forest soils show the least. By mapping individual environmental amplicon sequences to sequences derived from functionally characterized biosynthetic gene clusters, we identified conserved soil type–specific secondary metabolome enrichment patterns despite significant sample-to-sample sequence variation. These data are used to create chemical biogeographic distribution maps for biomedically valuable families of natural products in the environment that should prove useful for directing the discovery of bioactive natural products in the future.

Complex microbial ecosystems contain large reservoirs of unexplored biosynthetic diversity. Here we provide an experimental framework and data analysis tool to facilitate the targeted discovery of natural-product biosynthetic gene clusters from the environment. Multiplex sequencing of barcoded PCR amplicons is followed by sequence similarity directed data parsing to identify sequences bearing close resemblance to biosynthetically or biomedically interesting gene clusters. Amplicons are then mapped onto arrayed metagenomic libraries to guide the recovery of targeted gene clusters. When applied to adenylation- and ketosynthase-domain amplicons derived from saturating soil DNA libraries, our analysis pipeline led to the recovery of biosynthetic clusters predicted to encode for previously uncharacterized glycopeptide- and lipopeptide-like antibiotics; thiocoraline-, azinomycin-, and bleomycin-like antitumor agents; and a rapamycin-like immunosuppressant. The utility of the approach is demonstrated by using recovered eDNA sequences to generate glycopeptide derivatives. The experiments described here constitute a systematic interrogation of a soil metagenome for gene clusters capable of encoding naturally occurring derivatives of biomedically relevant natural products. Our results show that previously undetected biosynthetic gene clusters with potential biomedical relevance are very common in the environment. This general process should permit the routine screening of environmental samples for gene clusters capable of encoding the systematic expansion of the structural diversity seen in biomedically relevant families of natural products.

Recent bacterial (meta)genome sequencing efforts suggest the existence of an enormous untapped reservoir of natural-product-encoding biosynthetic gene clusters in the environment. Here we use the pyro-sequencing of PCR amplicons derived from both nonribosomal peptide adenylation domains and polyketide ketosynthase domains to compare biosynthetic diversity in soil microbiomes from around the globe. We see large differences in domain populations from all except the most proximal and biome-similar samples, suggesting that most microbiomes will encode largely distinct collections of bacterial secondary metabolites. Our data indicate a correlation between two factors, geographic distance and biome-type, and the biosynthetic diversity found in soil environments. By assigning reads to known gene clusters we identify hotspots of biomedically relevant biosynthetic diversity. These observations not only provide new insights into the natural world, they also provide a road map for guiding future natural products discovery efforts.DOI: http://dx.doi.org/10.7554/eLife.05048.001

PPTases (phosphopantetheinyl transferases) are of great interest owing to their essential roles in activating fatty acid, polyketide and non-ribosomal peptide synthetase enzymes for both primary and secondary metabolism, as well as an increasing number of biotechnological applications. However, existing techniques for PPTase characterization and development are cumbersome and technically challenging. To address this, we have developed the indigoidine-synthesizing non-ribosomal peptide synthetase BpsA as a reporter for PPTase activity. Simple co-transformation allows rapid assessment of the ability of a PPTase candidate to activate BpsA in vivo. Kinetic parameters with respect to either CoA or BpsA as variable substrate can then be derived in vitro by continuously measuring the rate of indigoidine synthesis as the PPTase progressively converts BpsA from its apo into holo form. Subsequently, a competition assay, in which BpsA and purified carrier proteins compete for a limited pool of CoA, enables elucidation of kinetic parameters for a PPTase with those carrier proteins. We used this system to conduct a rapid characterization of three different PPTase enzymes: Sfp of Bacillus subtilis A.T.C.C.6633, PcpS of Pseudomonas aeruginosa PAO1, and the putative PPTase PP1183 of Ps. putida KT2440. We also demonstrate the utility of this system for discovery and characterization of PPTase inhibitors.

In molecular evolutionary analyses, short DNA sequences are used to infer phylogenetic relationships among species. Here we apply this principle to the study of bacterial biosynthesis, enabling the targeted isolation of previously unidentified natural products directly from complex metagenomes. Our approach uses short natural product sequence tags derived from conserved biosynthetic motifs to profile biosynthetic diversity in the environment and then guide the recovery of gene clusters from metagenomic libraries. The methodology is conceptually simple, requires only a small investment in sequencing, and is not computationally demanding. To demonstrate the power of this approach to natural product discovery we conducted a computational search for epoxyketone proteasome inhibitors within 185 globally distributed soil metagenomes. This led to the identification of 99 unique epoxyketone sequence tags, falling into 6 phylogenetically distinct clades. Complete gene clusters associated with nine unique tags were recovered from four saturating soil metagenomic libraries. Using heterologous expression methodologies, seven potent epoxyketone proteasome inhibitors (clarepoxcins A-E and landepoxcins A and B) were produced from these pathways, including compounds with different warhead structures and a naturally occurring halohydrin prodrug. This study provides a template for the targeted expansion of bacterially derived natural products using the global metagenome.T he advent of cost-effective high-throughput sequencing and an increasingly sophisticated understanding of bacterial secondary metabolite biosynthesis have led to two important revelations with respect to the search for new natural products: first, that the biosynthetic potential of most cultured bacteria, as judged by the number of biosynthetic gene clusters (BGCs) observed in sequenced genomes, is far greater than previously estimated (1, 2); second, that the number of bacterial species in most environments is at least 100× greater than the number of species that is readily cultured (3, 4). These observations suggest that conventional "phenotype-first" natural products isolation approaches have only examined a small fraction of earth's bacterial biosynthetic potential.There are now a number of genomic search engines available that allow researchers to rapidly scan microbial whole genome sequences for BGCs encoding new natural products (5-7). Unfortunately, the large DNA contigs that these search strategies require as input are not readily available from complex metagenomes. In response to the need for a more robust metagenomic search strategy, our group recently developed an informatics platform called eSNaPD (8, 9) (environmental Surveyor of Natural Product Diversity) with the specific aim of facilitating sequence-guided discovery of new bacterial natural products from complex metagenomes (Fig. 1).The eSNaPD software is designed to bioinformatically assess short DNA sequences that have been amplified from environmental metagenomes by degenerate PCR targeting c...

Non-ribosomal peptide synthetase (NRPS) enzymes form modular assembly-lines, wherein each module governs the incorporation of a specific monomer into a short peptide product. Modules are comprised of one or more key domains, including adenylation (A) domains, which recognise and activate the monomer substrate; condensation (C) domains, which catalyse amide bond formation; and thiolation (T) domains, which shuttle reaction intermediates between catalytic domains. This arrangement offers prospects for rational peptide modification via substitution of substrate-specifying domains. For over 20 years, it has been considered that C domains play key roles in proof-reading the substrate; a presumption that has greatly complicated rational NRPS redesign. Here we present evidence from both directed and natural evolution studies that any substrate-specifying role for C domains is likely to be the exception rather than the rule, and that novel non-ribosomal peptides can be generated by substitution of A domains alone. We identify permissive A domain recombination boundaries and show that these allow us to efficiently generate modified pyoverdine peptides at high yields. We further demonstrate the transferability of our approach in the PheATE-ProCAT model system originally used to infer C domain substrate specificity, generating modified dipeptide products at yields that are inconsistent with the prevailing dogma.