COMMENTARY

Atlas for drug discoveryPierre Stallfortha and Jon Clardyb,1aJunior Research Group ‘Chemistry of Microbial Communication,’ Leibniz Institute for NaturalProduct Reserach and Infection Biology, Hans-Knöll-Institute, 07745 Jena, Germany; andbDepartment of Biological Chemistry and Molecular Pharmacology, Harvard Medical School,Boston, MA 02115

Reports of increasing antibiotic resistance andsagging drug discovery rates are appearingwith increasing frequency, and many warnthat we are living on the wrong side of anantibiotic peak—a period in which ever-fewer new antibiotics are being discoveredat ever-increasing costs (1). In response,efforts to discover new antibiotics and otherdrugs have taken many forms, includinglooking at formerly productive sources thatwere thought exhausted, much as rising oilprices led to new extraction techniques forabandoned oil fields. For several decades, smallmolecules produced by soil bacteria were ourmost important source of new drugs, as rep-resented by the antibiotics erythromycin andvancomycin, the anticancer agents bleomycinand mitomycin, and the immunomodulatorscyclosporin and rapamycin (2). In the last twodecades, almost all pharmaceutical companieshave abandoned bacterially based drug dis-covery because it seemingly fits poorly withthe high-throughput screening and medicinalchemistry approaches that define the indus-try’s favored discovery paradigm, and becauseits high rate of rediscovering previously knowncompounds indicated that it was unlikely toyield new drugs (3). More recently, genomicrevelations have dramatically altered our viewof soil bacteria. The revelations were of twosorts: (i) most bacteria (∼99%) cannot be cul-tured under typical laboratory conditions, and(ii) even those that can be cultured produce onlya fraction (∼10%) of the small molecules en-coded in their genomes (4–6). These dual short-falls in culturing and expression pointed to abonanza of potentially useful small moleculesremaining to be discovered and led to innova-tive technical approaches to finding them. InPNAS, Charlop-Powers et al. (7) use culture-independent DNA sequencing on a broadenvironmental scale to systematize where dif-ferent types of bacterially produced small mol-ecules are likely to be found. Their “chemical-biogeographic survey” provides insights intomicrobial ecology along with a practical guidefor microbial genome prospectors.The potential scale of such a survey is stag-

gering. Bacteria are the most diverse organisms

on the planet, and a single gram of soil cancontain between a thousand and a milliondistinct species according to a pioneeringsurvey (8). Species definition in bacteria hasits whimsical aspects, but it usually meansthat a characteristic DNA sequence from onestrain has, or does not have, a specified sim-ilarity to the sequence from another strain. Ifthe similarity is high enough, say 98%, thenthe two strains are said to belong to the

Charlop-Powers et al.use culture-independentDNA sequencing ona broad environmentalscale to systematizewhere different types ofbacterially producedsmall molecules arelikely to be found.same operational taxonomic unit (OTU). Re-searchers typically use the 16S ribosomalRNA gene sequence, a highly conserved se-quence whose utility was established decadesago (9). Researchers isolate DNA from anenvironment—from desert soil to the humangut—and use PCR to amplify selected se-quences from the sample for sequencing.These selected and amplified fragments ofDNA are called amplicons. Amplicons of 16SrRNA have been widely used to create bio-geographic surveys of soil bacteria—a catalogof what bacteria are found in different envi-ronments (9, 10).

Biosynthesis-Based Genotyping of SoilMicrobiomesCharlop-Powers et al. used a similar ap-proach to answer a different question: wherecan we find genes that are likely to encodethe production of useful molecules? Answer-ing this question requires deciding on whatgenes, really what conserved DNA sequenceson the genes, to look for. The authors selecttwo important and well-studied biosynthetic

pathways: the polyketide (PK) pathway andthe nonribosomal peptide (NRP) pathway,which assemble either carbon chains or aminoacid chains from smaller subunits. These twopathways, either singly or in combination,make all of the examples mentioned earlier.The specific genes that were selected arenecessary elements in both pathways, namelythe genes that encode the proteins that selectthe smaller subunits. This necessary restric-tion has consequences, and antibiotics suchas fosfomycin—whose biosyn-thesis followsa pathway different from PK or NRP path-ways—would be overlooked. However, theselimitations are obvious and could easily beaddressed in future studies.To ensure a faithful representation of geo-

graphically resolved biosynthetic richness, theauthors collect environmental DNA samplesfrom more than 90 different locations. Thesesamples served as templates for the prepara-tion of ketosynthase (KS) and adenylation(AD) amplicons. Although these conditionsintroduced amodest amplification-dependentbias, the resulting amplicons closely mirrorthe variety of AD and KS domains found inthe metagenome of the soil-dwelling bacteria.Subsequent 454 pyrosequencing of theseamplicons allowed a classification systembased on biosynthetic OTUs to be con-structed. The OTU-based classification ofenvironments was linked to environmentalsoil data including pH, moisture, and mineralcomposition among others gathered at thecollection site, and the two data sets—bio-synthetic OTUs and environmental parame-ters—could be cross-analyzed. A principalcomponent analysis revealed three principaltypes of soils with different biosynthetic po-tential. For example, arid soil environmentsshowed the largest biosynthetic potential.

Biosynthetic Repertoires of Soil TypesThe rich data set provided by this chemical-biogeographic survey provides two in-teresting insights. The first is that bacteriain similar environments produce, or at leasthave amplicons that suggest they wouldproduce, similar small molecules. A famous

Author contributions: P.S. and J.C. wrote the paper.

The authors declare no conflict of interest.

See companion article on page 3757.

1To whom correspondence should be addressed. E-mail: jon_clardy@hms.harvard.edu.

www.pnas.org/cgi/doi/10.1073/pnas.1400516111 PNAS | March 11, 2014 | vol. 111 | no. 10 | 3655–3656

COMMEN

TARY

generalization of microbial diversity notesthat “Everything is everywhere, the environ-ment selects.” (11) Soil variables, especiallypH, largely control the environmental selec-tion (10). These soil variables presumablyselect for bacteria with primary metabolismsthat match a given environment. The presentstudy demonstrates clearly that a plausible,but previously unproven, consequence of theenvironmental determinism of microbialbiogeographic diversity is reflected in thesecondary metabolism—the similar bacteriafound in similar soils will produce similarcompounds. This extreme environmentaldeterminism in which soil variables controlboth biological and chemical diversity hasno correlate in the biogeography of mac-roorganisms. Arid soils in geographicallyseparated regions will, for example, hostdistinctly different plants that will producedifferent small molecules. Bacteria do notproduce small molecules so that we canmine them for drugs; the molecules mustconfer some survival benefit. However, theecological roles of these bacterially pro-duced small molecules are poorly known,and their discovery represents a challengefor microbial ecology.

A second conclusion, that arid soils havegreater biosynthetic potential than pine forestsoils or brackish sediments, might seem puz-zling at first. SelmanWaksman largely initiatedthe Golden Age of Antibiotics and won the1952 Nobel Prize in Physiology or Medicinefor his discovery of antibiotics, especially strep-tomycin, from swampy soils in New Jersey.The stream of antibiotics from Waksman’slaboratory, which all came from actinomycetesfound in these soils, reflects his pioneeringspirit that allowed him to succeed even withsampling suboptimal environments.Finally, Charlop-Powers et al. (7) pro-

vide a treasure map for drug prospec-

tors, and it will be interesting to see howfiner-grained geographic and molecular ana-lyses expand on this work. The chemical-bio-geographic biodiversity approach could alsobe useful in exploring complex environmentssuch as the human microbiome. We haveincreasingly good maps of what bacteria oc-cupy different regions (and metabolic envi-ronments) of the human body. Will there bea similar correlation of chemistry and envi-ronment? This work also highlights the needfor new techniques to convert genomic datainto molecules, which is perhaps the greatestchallenge facing genome-based drug discoveryefforts.

1 Payne DJ, Gwynn MN, Holmes DJ, Pompliano DL (2007) Drugs forbad bugs: Confronting the challenges of antibacterial discovery. NatRev Drug Discov 6(1):29–40.2 Nett M, Ikeda H, Moore BS (2009) Genomic basis for naturalproduct biosynthetic diversity in the actinomycetes. Nat Prod Rep26(11):1362–1384.3 Baltz RH (2008) Renaissance in antibacterial discovery fromactinomycetes. Curr Opin Pharmacol 8(5):557–563.4 Kaeberlein T, Lewis K, Epstein SS (2002) Isolating “uncultivable”microorganisms in pure culture in a simulated natural environment.Science 296(5570):1127–1129.5 Winter JM, Behnken S, Hertweck C (2011) Genomics-inspireddiscovery of natural products. Curr Opin Chem Biol 15(1):22–31.6 Challis GL (2008) Genome mining for novel natural productdiscovery. J Med Chem 51(9):2618–2628.

7 Charlop-Powers Z, Owen JG, Reddy BVB, Ternei MA, Brady SF

(2014) Chemical-biogeographic survey of secondary metabolism in

soil. Proc Natl Acad Sci USA 111:3757–3762.8 Torsvik V, Øvreås L, Thingstad TF (2002) Prokaryotic diversity—

magnitude, dynamics, and controlling factors. Science 296(5570):

1064–1066.9 Pace NR (1997) A molecular view of microbial diversity and the

biosphere. Science 276(5313):734–740.10 Fierer N, Jackson RB (2006) The diversity and biogeography

of soil bacterial communities. Proc Natl Acad Sci USA 103(3):

626–631.11 de Wit R, Bouvier T (2006) ‘Everything is everywhere, but, the

environment selects’: What did Baas Becking and Beijerinck really

say? Environ Microbiol 8(4):755–758.

3656 | www.pnas.org/cgi/doi/10.1073/pnas.1400516111 Stallforth and Clardy