Chemistry & Biology

Perspective

Metagenomic Approachesfor Exploiting Uncultivated Bacteriaas a Resource for Novel Biosynthetic Enzymology

Micheal C. Wilson1 and Jorn Piel1,*1Institute for Microbiology, Eidgenossische Technische Hochschule (ETH) Zurich, Wolfgang-Pauli-Strasse 10, 8093 Zurich, Switzerland*Correspondence: jpiel@ethz.chhttp://dx.doi.org/10.1016/j.chembiol.2013.04.011

Most biologically activemicrobial natural products are known from strains that can be isolated and cultivatedin the laboratory. However, the genomics era has revealed that cultured bacteria represent a mere fraction oftotal estimated bacterial biodiversity. With the development of community genomics, termedmetagenomics,the uncultivated majority became accessible for functional analysis. Through metagenomic studies, novelbiocatalysts and biosynthetic pathways are being discovered at a pace previously not possible using tradi-tional molecular biology techniques. Additionally, the study of uncultivated bacteria has provided valuableinsights into previously overlooked biocatalysts from cultured strains. This perspective highlights recent dis-coveries frommetagenomics of uncultivated bacteria and discusses the impact of those findings on the fieldof natural products.

IntroductionWith the advent of cultivation-independent methods for study-

ing microbial diversity (e.g., 16S ribosomal RNA phylogeny),

we now know that bacteria inhabit nearly every ecological niche

imaginable: from the deepest oceans to arid deserts to the gut

of nearly every multicellular organism. They can exist as free-

living organisms or form symbiotic or pathogenic relationships

with eukaryotic hosts or other prokaryotes. Yet, despite the

ubiquity of bacteria and our extensive effort to study them, the

vast majority of microbial biodiversity has not been functionally

characterized (Wu et al., 2009). This is largely due to current lim-

itations in cultivation technologies but also to the sheer number

of estimated bacterial taxa. For instance, approximately 70% of

all known bacterial phyla do not have a single cultured represen-

tative (Achtman and Wagner, 2008). This reality is both exciting

and somewhat daunting considering the extremely prosperous

relationship between cultivated bacteria and industrial applica-

tions. Specialized bacterial strains, their produced proteins,

and small molecules are used in industries involved in food pro-

cessing, biofuel production, cosmetics, materials chemistry,

and pharmaceutical drug discovery and production. In fact,

bacteria are likely the best hope for discovery of novel antibi-

otics to combat the ever growing prevalence of antibiotic-resis-

tant pathogens (Wright, 2012). Moreover, greater than 10,000

biologically active metabolites with therapeutic properties

against cancer, HIV, inflammation, and many more human dis-

eases are known from cultured actinomycetes alone, most of

which are produced by a single genus (Streptomyces) (Berdy,

2005). Full genome sequencing of cultured bacteria also brings

light to the fact that, in most cases, the isolated secondary me-

tabolites of even well-studied producers barely scratch the sur-

face of the biosynthetic potential encoded within their DNA

(Bentley et al., 2002; Udwary et al., 2007). Extrapolating what

we know from these prolific cultured strains, we can hypothe-

size that the potential for discovery of novel biocatalysts and

636 Chemistry & Biology 20, May 23, 2013 ª2013 Elsevier Ltd All righ

biosynthetic gene clusters from the uncultured majority is

massive.

In the last decade or so, the field of metagenomics has proven

to be a fruitful means for accessing the biosynthetic machinery of

uncultured bacteria, thus circumventing traditional molecular

methods that rely on cultivation. Metagenomics is the analysis

of DNA from a mixed population of organisms and initially

involved the cloning of either total or enriched DNA directly

from the environment (eDNA) into a host that can be easily culti-

vated (Handelsman, 2004; Miao and Davies, 2009). More

recently, advances in next generation sequencing (NGS) tech-

nologies allow isolated eDNA to be sequenced and analyzed

directly from environmental samples (Shokralla et al., 2012). Us-

ing a variety of functional or bioinformatic screening methods,

novel biocatalysts and small molecule biosynthetic genes can

be identified at an increasing pace. Metagenomics in industrial

applications has yielded numerous enzyme biocatalysts used

in the production of important commodities, such as biofuels

(Xing et al., 2012) and detergents (Lorenz and Eck, 2005). In

the field of natural product drug discovery, metabolites or puta-

tive pathways for most known and some previously unreported

structural families have been isolated from soils (Chang and

Brady, 2011; Feng et al., 2011), sponges (Freeman et al., 2012;

Piel et al., 2004a), tunicates (Rath et al., 2011; Schmidt et al.,

2012), insects (Piel, 2002), and many more ecological habitats

using metagenomic methods. In many cases, the challenge still

remains to link biosynthetic genes to products; however, several

examples of successful heterologous expression or in vitro char-

acterization of biosynthetic genes have been reported (Brady

et al., 2001; Brady and Clardy, 2005a; Long et al., 2005). For

detailed reviews, see Piel (2011), Nikolouli and Mossialos

(2012), and Brady et al. (2009). Excitingly, research on unculti-

vated bacteria has shown that unprecedented biosynthetic en-

zymes and transformations can be found in these organisms at

high frequency. In this perspective, we discuss strategies for

ts reserved

Figure 1. General Metagenomic WorkflowGraphical schematic summarizing the stepsinvolved in a typical metagenomic workflow.

Chemistry & Biology

Perspective

accessing the biosynthetic potential of uncultivated bacteria and

highlight case studies of novel biochemistry discovered by these

methods.

Metagenomics: A Powerful Tool for AccessingUncultured BacteriaA powerful tool for investigating the biosynthetic potential of an

uncultivated bacterial population involves the construction of

an eDNA library in a suitable host (Figure 1). This begins with

the isolation of high molecular weight eDNA, for which several

protocols have been developed from a variety of different envi-

ronments (Gurgui and Piel, 2010; Liles et al., 2008; Ouyang

et al., 2010). Enrichment of specific cell populations is common

prior to eDNA isolation (Hildebrand et al., 2004; Jiao et al., 2006;

Lenk et al., 2012). This strategy is often used for samples that

contain mixtures of eu- and prokaryotic cells, such as marine

sponges and other invertebrates containing symbiotic bacteria

(Bewley et al., 1996; Piel et al., 2004a). Each population can

Chemistry & Biology 20, May 23, 2013

then be targeted for analysis separately.

As with genomic library construction of

single organisms, Escherichia coli is the

most common host; however, it is some-

times necessary to use an alternate DNA

recipient for properties such as codon

bias and precursor supply capability

when considering function-based

screening protocols (Brady et al., 2009;

Craig et al., 2009, 2010; Piel, 2011;

Wang et al., 2000). The choice of vector

can also determine the success or failure

of a particular project. Cosmid and fosmid

vectors are the most commonly used for

library construction, because they can

accept fragments from 35–45 kb, can

easily be transfected into E. coli to pro-

duce millions of clones, and are relatively

stable (single copy fosmids are more sta-

ble than cosmids). Larger insert libraries

using bacterial artificial chromosomes

(BACs) have also been successful (Gilles-

pie et al., 2002; Long et al., 2005) but are

less common, due to the challenge of

isolating DNA fragments of sufficient

sizes from eDNA. Hundreds to millions

of clones can be maintained in a 96 or

384 well format, as liquid pools (Rondon

et al., 2000), or in a low-melting-point

agarose three-dimensional (3D) matrix

(Gurgui and Piel, 2010), depending on

the number of clones required to obtain

ample coverage for screening purposes.

Once the library is constructed, it can be

subjected to functional or homology-

based screening protocols.

A common method for identifying discrete biocatalysts or me-

tabolites synthesized by small to medium-sized gene clusters

(i.e., smaller than the vector insert size) is to screenmetagenomic

library clones for modified phenotypes or specific enzymatic

function. This method is prevalent for the identification of indus-

trially relevant enzymes, such as cellulases, lypases, esterases,

and many more (Reyes-Duarte et al., 2012). Phenotypic screens

based on color, zones of inhibition on bacterial (Rondon et al.,

2000) or fungal (Chung et al., 2008) lawns, hemolytic activity on

blood agar plates (Gillespie et al., 2002), or the use of specific re-

porter genes (Guan et al., 2007; Owen et al., 2012) are employed

for identifying functional biocatalysts. A number of natural prod-

ucts and their biosynthetic machinery have been identified

through functional screening of soil metagenomes, including tur-

bomycin A (1) and B (2) (Gillespie et al., 2002), terragines (3)

(Wang et al., 2000), fatty acid derivatives (Brady et al., 2002;

Brady and Clardy, 2003), and several aromatic polyketides

(Feng et al., 2011; Figure 2). Intracellular reporters based on

ª2013 Elsevier Ltd All rights reserved 637

Figure 2. Natural Products WhoseBiosynthetic Pathways Were Identifiedthrough Metagenomics or Single-CellIsolation from a Metagenomic SampleThe turbomycins (1 and 2) and terragine A (3)pathways were discovered from soil meta-genomes, the patellazole A (4) and bryostatin 1 (5)pathways are from symbiotic bacteria associatedwith marine invertebrates, and the apratoxin A (6)pathway was identified by single-cell analysis of acyanobacterium.

Chemistry & Biology

Perspective

fluorescence or pigment production have been used to identify

novel quorum-sensing molecules (Williamson et al., 2005) and

phosphopantetheinyl transferases involved in polyketide and

nonribosomal peptide biosynthesis (Owen et al., 2012). In these

screens, the metagenomic library host carries a reporter gene

(e.g., GFP), usually located on a plasmid, that is activated

when a specific target from the eDNA is expressed. Functional

screens are advantageous because they do not require any pre-

vious knowledge of gene sequences and are perhaps the best

way to identify novel biochemical mechanisms.

Alternatively, gene homology-based screens rely on using

sequence similarity to previously known biosynthetic genes or

biocatalysts. Though limited in their utility for discovery of

completely novel enzymatic reactions, homology-based

methods do not require expression of biosynthetic genes for

detection. PCR or hybridization techniques are the most com-

mon strategies for identifying, quantifying, and cataloging genes

of interest frommetagenomic libraries, especially in the early ge-

nomics era. Typically, degenerate primers designed from highly

638 Chemistry & Biology 20, May 23, 2013 ª2013 Elsevier Ltd All rights reserved

conserved regions of biosynthetic targets

are used to PCR amplify gene fragments

from total or enriched eDNA. From these

fragments, specific PCR primers can be

generated for screening a metagenomic

clone library. This is a common strategy

to identify modular polyketide or nonribo-

somal peptide biosynthesis in a metage-

nomic sample, since these pathways

are more difficult to detect in functional

screens compared to individual bio-

catalysts or smaller pathways (e.g., type

II polyketide synthase [PKS] systems).

This is due to the fact that these pharma-

ceutically relevant natural products are

synthesized by gene clusters that can

easily exceed the insert size limitations

of fosmid or even BAC vectors, making

it often necessary to isolate multiple

clones or perform genome walking in

order to obtain the complete gene clus-

ter. Thus, heterologous expression of

modular PKS or nonribosomal peptide

synthetase (NRPS) pathways can usually

only be attempted after initial identifica-

tion using homology-based screens.

As high-throughput sequencing and

assembly technologies improve and pri-

ces decrease, large-scale sequencing of enriched and metage-

nomic DNA is becoming more common for homology-based

gene discovery. These NGS platforms, including 454 Life Sci-

ences pyrosequencing, Illumina, single molecule sequencing,

and others have revolutionized the way metagenomic (and tradi-

tional genomic) data are obtained. Enormous amounts of

sequencing reads can be generated within hours using the latest

platforms. An unfortunate disadvantage with NGS is that assem-

bly becomes very challenging, due to short read lengths associ-

ated with most platforms. PKS and NRPS pathways from

metagenomes can be particularly challenging to assemble

because of the highly conserved and repetitive organization of

these genes. Additionally, complex systems (e.g., animal-micro-

bial associations and soil) can contain several hundred to thou-

sand different species of bacteria and with numerous PKS and

NRPS genes. Nonetheless, this challenge can be addressed to

some degree by using paired-end libraries, deeper sequencing,

or enrichment of the producing bacteria, and gap closing using

PCR. These strategies have been used with NGS to identify

Chemistry & Biology

Perspective

PKS and NRPS clusters from a number of cultured strains (John-

ston et al., 2013; Liu et al., 2010) as well as the patellazole (4)

(Donia et al., 2011a) and bryostatin (5) (Sudek et al., 2007)

biosynthetic gene clusters from uncultured bacteria associated

with marine invertebrates.

An important limitation to the use of metagenomics for com-

plex communities is that biosynthetic genes can rarely be

ascribed to a specific producer, as clones carrying both biosyn-

thetic genes and phylogenetic markers are almost nonexistent in

clone libraries. Single-cell genomics has emerged as a tool to

address this dilemma and reducemetagenomic complexity. Sin-

gle cells can be isolated from complex microbial mixtures via mi-

crofluidic encapsulation, micromanipulation, fluorescence-acti-

vated cell sorting (FACS), or microdissection (Blainey, 2013).

The genome of one cell is then amplified by multiple displace-

ment amplification (MDA) to generate enough DNA for

sequencing or PCR screening (Kvist et al., 2007). For example,

single-cell isolation and sequencing was used to identify the

elusive apratoxin (6; Figure 2) pathway from the filamentous cy-

anobacteria, Lyngbya bouillonii, by separating the cyanobacteria

from its associated heterotrophic bacteria (Grindberg et al.,

2011). FACS and MDA were used to sequence the partial

genome of a sponge-associated bacterium from the unusual

candidate phylum ‘‘Poribacteria’’ and linked the bacterium to a

new class of PKS genes (Siegl et al., 2011). Differentiating the ge-

netic make-up of a single organism from a complex assemblage

may give hints to the nutritional requirements necessary for culti-

vation experiments. Many obligate symbionts, however, are

likely to remain unculturable, as they often require host-specific

nutrients or cell-cell interactions in order to grow.

Ideally, the identification of a natural product biosynthetic

gene cluster will subsequently lead to heterologous expression

of the entire cluster and production of its compound. However,

expression of whole pathways from uncultured or slow-growing

organisms remains a major bottleneck in the functional charac-

terization natural product pathways and production of their prod-

ucts. This is especially true for large PKS and NRPS clusters that

can span several cosmid or fosmid clones and can incorporate

precursors or cofactors from external metabolic pathways not

present in the host. Additional challenges to successful heterol-

ogous expression, including codon bias, promoter recognition,

host toxicity, product yield, and host versatility, must be resolved

when considering any experiment. In this case, genomic infor-

mation can assist in choosing an appropriate host organism for

heterologous expression of desired genes. Additional tech-

niques to address these challenges are steadily advancing. For

example, transformation-assisted recombination (TAR) in yeast

has been used to assemble type II PKS clusters fromoverlapping

clone inserts derived from soil eDNA (Kim et al., 2010). Genetic

engineering of broad range host strains has also led to the suc-

cessful expression of cryptic pathways (Busch and Hertweck,

2009; Komatsu et al., 2013).

By applying these methods to various habitats, uncultivated

bacteria are increasingly being recognized as a rich source for

unprecedented biosynthetic enzymology. In the following sec-

tions, wewill provide several examples that illustrate the diversity

of novel pathways and biochemistry encountered in these or-

ganisms. It should be pointed out that many of these features

were later also found in cultured bacteria and are thus not entirely

Chemistry & B

restricted to uncultivated organisms. However, the high fre-

quency at which biosynthetic novelty is encountered in the still

limited number of published metagenomic studies suggests a

significant skew toward unusual pathways in environmental pro-

karyotes and thus an exciting discovery potential.

N-to-O Exchange in EnamidesA series of saturated andmonosaturated long-chainN-acyl-tyro-

sines (7 to 8; Figure 3A) were the first novel biologically active

natural products discovered through functional screening of

metagenomic libraries (Brady andClardy, 2000). Later, a broader

range ofN-acyl-amino acids were identified along with enol ester

derivatives (9; Figure 3A; Brady et al., 2002; Brady and Clardy,

2005b). Sequencing and transposon mutagenesis of clones pro-

ducing these compounds led to the identification of a novel N-

acyl-amino acid synthase and, subsequently, to the fatty acid

enol ester biosynthetic gene cluster (Brady et al., 2002; Brady

and Clardy, 2000). Of particular note in this pathway is the un-

precedented N,O-acyltransferase, FeeH, which, through a yet

unknown biochemical mechanism, facilitates the enamine-to-

enol ester conversion (Figure 3A).

Isocyanide BiosynthesisBy functionally screening a soil-derived metagenomic E. coli li-

brary for antibiotic activity against Bacillus subtilis, an isocya-

nide-functionalized indole (10; Figure 3B) was discovered (Brady

and Clardy, 2005a). The genes isnA and isnB located on the

cloned eDNA region were shown to be responsible for biosyn-

thesis. Through stable isotope feeding studies, the indole was

shown to derive from tryptophan. IsnA is an unprecedented iso-

nitrile synthase that performs the C–N condensation, while IsnB

is a decarboxylase (Figure 3B). Extensive feeding studies re-

vealed that the carbon atom of the isonitrile group originates

from a single excised carbon (C2) of ribulose-5-phosphate

(Brady and Clardy, 2005c). To explore this remarkable transfor-

mation further, 12 additional isnA-related sequences were

recovered by homology-based screening of 400,000 soil-

derived cosmid clones, and the resulting clusters were heterolo-

gously expressed in E. coli (Brady et al., 2007). Re-evaluation

of the pvc biosynthetic gene cluster from Pseudomonas

aeruginosa, which contains the isnA homolog pvcA, showed

that themajor product of the pathway was the isocyanide natural

product paerucumarin (11; Figure 3C) and not pyoverdine, as

previously proposed (Clarke-Pearson and Brady, 2008).

Aromatic Polyketide SkeletonsPKSs are responsible for the biosynthesis of some of the most

structurally and functionally diverse bioactive natural products,

many of which are employed pharmaceutical drugs. Evolution-

arily related to fatty acid synthases, PKSs similarly catalyze the

successive Claisen condensation and modification of small

acyl units. There are three main classes of PKS enzymes or

enzyme complexes (type I, II, and III), which can be further

divided into subtypes based on their gene architecture and

whether they function iteratively or modularly. Type II PKSs

generate most of the aromatic polyketides found in bacteria

(Brady and Clardy, 2000), of which tetracyclines and daunoru-

bicin are well-known examples. While a myriad of functionaliza-

tion patterns exist among these substances, the diversity of their

iology 20, May 23, 2013 ª2013 Elsevier Ltd All rights reserved 639

Figure 3. Small Molecules and Proposed Biosynthesis Determined through Functional Screening and Heterologous Expression ofMetagenomic eDNA(A) Proposed biosynthesis of long-chain N-acyltyrosines ([7] and [8]) and their enol ester derivatives (9).(B) Proposed biosynthesis of isocyanide-functionalized indoles (10).(C) Structure of paerucumarin (11).(D) Representative aromatic polyketides isolated from heterologous expression of eDNA from soil metagenomes.

Chemistry & Biology

Perspective

carbon skeletons is restricted to just a few basic types. Thus, the

discovery of novel skeletons is relatively rare for this substance

class. To explore uncultivated bacteria as a resource of new ar-

omatic polyketides, clones from soil metagenomic libraries

carrying characteristic ketosynthase b (KSb) genes of type II

PKSs were identified using degenerate primers. Positive over-

lapping clones were reassembled by TAR and expressed in

Streptomyces (Feng et al., 2010, 2011; King et al., 2009). From

these studies, two new pentacyclic type II PKS core skeletons

were discovered (12 and 13; Figure 3D) along with derivatives

of two compound families whose core skeletons are extremely

rare among cultured strains (14 and 15; Figure 3D). The novel

pentacyclic product, erdacin (12) is predicted to arise from the

coupling of two smaller but similarly derived bicyclic octaketide

intermediates and has potent antioxidant activity (King et al.,

2009). The other novel pentacyclic polyketide (13) is proposed

to derive from the rearrangement of a common tetracyclic inter-

mediate. Among the rare skeletons isolated were the antibiotic

fasamycins (14), which are active against methicillin-resistant

Staphylococcus aureus and vancomycin-resistant Enterococci

by inhibiting FabF during type II fatty acid elongation (Feng

et al., 2012). The moderately antibacterial fluostatins (15) repre-

sent another rare structural family derived from the TAR-based

assembly and heterologous expression of two clones from a

640 Chemistry & Biology 20, May 23, 2013 ª2013 Elsevier Ltd All righ

multimillion member metagenomic library constructed from a

soil sample collected in the Anza Borrego desert of Southern

California (Feng et al., 2010). The five-member ring in the core

of 15 is derived from the oxidative excision of a single carbon

and rearrangement of a rabelomycin-like precursor (Feng et al.,

2010). Phylogenetic analysis of KSb domains can assist in struc-

ture predictions for aromatic polyketides from both cultured and

uncultured bacteria, and the novel structural families from meta-

genomic studies are honing the predictive capability of such

methods (Feng et al., 2011).

Expansion of Type I PKS DiversityA recently recognized subclass of modular type I PKSs is the

trans-acyltransferase (trans-AT) type (Piel, 2010). Trans-AT

PKSs differ from textbook (cis-AT) type I PKSs, in that a free-

standing AT domain protein acts iteratively to provide the acyl

extender units rather than each PKS module containing a dedi-

cated AT domain. Though trans-AT PKS genes or gene frag-

ments were initially identified in cultured bacteria (Huang et al.,

2001; Paitan et al., 1999; Scotti et al., 1993), the first extended

trans-AT PKS biosynthetic gene cluster attributed to a known

product was the pathway for the cytotoxin pederin (16) (Figure 4),

which is produced by an uncultured Pseudomonas sp. symbiont

of Paederus spp. rove beetles (Piel, 2002). To isolate the genes,

ts reserved

Figure 4. Polyketides Derived from Trans-AT PKS PathwaysWith the exception of leinamcyin (17), these trans-AT PKS products were derived from pathways discovered through homology-based metagenomic libraryscreening/sequencing.

Chemistry & Biology

Perspective

the ketosynthase (KS) domains of the pederin cluster were first

amplified using degenerate primers from the total DNA of ped-

erin-producing beetles. Primers from the amplified KS fragments

were then used to identify the pederin cluster from an 80,000

cosmid clone metagenomic library. The first functional evidence

for the hypothesized catalytic function of trans-AT enzymes was

provided for the leinamycin (17) PKS, showing that a single AT

loads all the expressed acyl carrier protein domains (Cheng

et al., 2003; Figure 4). From studies on trans-AT systems, it

became rapidly clear that these PKSs exhibit an exceptional

architectural and catalytic diversity. Examples of components

that are absent or rare in cis-AT PKSs include various domains

for the formation of five- and six-membered heterocyclic rings

(Gulder et al., 2011; Matilla et al., 2012; Moldenhauer et al.,

2010; Tang et al., 2004), dehydratase domains that introduce

b,g-double bonds (Kusebauch et al., 2010; Moldenhauer et al.,

2010), and enzymes and domains for diverse carbon branches

(Moldenhauer et al., 2010).

For as-yet unknown reasons, trans-AT PKSs are by far the

dominant known PKS type in uncultivated bacterial symbionts,

while they are much less frequent in free-living cultivated

bacteria. Examples include biosynthetic genes for onnamide

(18) and psymberin (irciniastatin A) (19) from marine sponges

(Figure 4), bryostatin (5; Figure 2) from a bryozoan, patellazole

(4; Figure 2) from a tunicate, and rhizoxin (20) from a fungus

(Figure 4), which all belong to diverse endosymbionts. Many

complex polyketides with exquisite bioactivities are isolated in

extremely low quantities from invertebrates but are most likely

produced by trans-AT PKSs of symbiotic bacteria. On this ratio-

nale, strategies targeting trans-AT PKS genes have been devel-

Chemistry & B

oped to provide direct access to biosynthetic pathways for such

compounds (Fisch et al., 2009; Piel et al., 2004b), which can then

be used for the development of heterologous expression sys-

tems. For example, it is evident that trans- and cis-AT PKS sys-

tems evolved separately and form distinct clades when

compared phylogenetically, allowing for targeted genome min-

ing. Structure prediction from trans-AT PKSs are challenging,

since they do not follow the same colinearity rules as textbook

cis-AT systems. However, phylogeny-based rules have been

developed to predict polyketide structures from trans-AT PKS

sequences and vice versa (Nguyen et al., 2008). These phylog-

eny-based approaches facilitated investigation of trans-AT path-

ways from both cultured and uncultured bacteria (Irschik et al.,

2010; Nguyen et al., 2008; Teta et al., 2010).

Peptide ModificationsPeptide natural products from postribosomal peptide synthesis

pathways, recently designated ribosomally synthesized and

posttranslationally modified peptides (RiPPs), are produced

by all three domains of life, and many possess potent biological

activities (Arnison et al., 2013). The biological activities of RiPPs

include cytotoxicity, antimicrobial activity, anticancer, antiviral,

and quorum sensing. Though known throughout the realm

of cultivated bacteria, functional identification of at least two

new families of RiPPs originated from metagenomic studies

(Figure 5).

Cyanobactins constitute a large class (�200 known products)

of cyclic RiPPs produced by both free-living and symbiotic cya-

nobacteria (Donia et al., 2008, 2011b). Many of these com-

pounds are cytotoxic. Common enzymatic modifications of

iology 20, May 23, 2013 ª2013 Elsevier Ltd All rights reserved 641

Figure 5. RiPPs from Pathways Elucidated through Metagenomic StudiesPosttranslational modifications are indicated by color: heterocyclization (pink); prenylation (green); C-methylation (orange); N-methylation (violet); epimerization(red); and hydroxylation (blue). Polytheonamides differ in the stereochemistry of the sulfoxide moiety.

Chemistry & Biology

Perspective

cyanobactins include N-to-C macrocyclization (Koehnke et al.,

2012; McIntosh et al., 2010a; Figure 6A) and heterocyclizations

(McIntosh et al., 2010b; McIntosh and Schmidt, 2010;

Figure 6B). Sometimes structures exhibit unprecedented preny-

lations of serine, tyrosine, or threonine residues facilitated by a

new family of O-prenyltransferases (O-PTs) (McIntosh et al.,

2011; Figure 6C). Several recent reviews discuss cyanobactins

in detail (Arnison et al., 2013; Schmidt and Donia, 2009). Though

the first cyanobactins were reported in 1980 from didemnid tuni-

cates (Aziz et al., 2008), it was only in 2005 when two groups

working independently identified and heterologously expressed

the patellamide (21; Figure 5) pathway (pat) from a cyanobacte-

rial symbiont, Prochloron didemni, of the tunicate Lissoclinum

patella (Long et al., 2005; Schmidt et al., 2005). While Long, Jas-

pars, and coworkers discovered the genes by expression

screening of an enriched P. didemni DNA library in E. coli,

642 Chemistry & Biology 20, May 23, 2013 ª2013 Elsevier Ltd All righ

Schmidt et al. analyzed the symbiont DNA first for NRPS and

then for RiPP genes. Once the putative pat pathway was identi-

fied as a small RiPP cluster, it was expressed in E. coli yielding

20 mg l�1. Knowledge of the patellamide pathway revealed

related enzymes encoded in numerous other symbiotic and

free-living cultured and uncultured cyanobacteria, with homo-

logs being present in about 30% of evolutionarily distant cyano-

bacterial taxa (Leikoski et al., 2009). In addition to the O-PTs,

mechanism of heterocyclization was elucidated from enzymes

in the pat and trunkamide (22; Figure 5) pathways (tru). Macro-

cyclization of cyanobactin RiPPs involves an Asp-His-Ser cata-

lytic triad analogous to macrocycle formation by thioesterase

domains of NRPSs. Upon recognition of a three-residue signa-

ture, a serine protease-like macrocyclase, PatG, cleaves a

C-terminal peptide sequence and facilitates the N-to-C macro-

cyclization (Figure 6A; Koehnke et al., 2012; McIntosh et al.,

ts reserved

Figure 6. Modifications of Cyanobactin Ribosomal PeptidesEnzymatic reactions for (A) heterocyclization, (B) macrocyclization, and (C) prenylation as determined through in vitro experiments.

Chemistry & Biology

Perspective

2010a). Unlike heterocyclization in nonribosomal peptide

systems, which require three distinct protein domains, the heter-

ocyclases in the pat and tru pathways involve only a single ATP-

dependent enzyme that is akin to molecular machines

(Figure 6B; McIntosh and Schmidt, 2010). Additional experi-

ments by the Schmidt group showed that combinatorial libraries

could be generated by engineering modifications in the cyano-

bactin precursor peptide (Donia et al., 2006, 2008).

The polytheonamides (23; Figure 5) are another example of

peptides originally thought to be products of an NRPS biosyn-

thetic route. Reported in 1994 from the marine sponge Theonella

swinhoei, these compounds exhibit a highly complex structure

and belong to the largest secondary metabolites known. The

polytheonamides are highly cytotoxic linear peptides comprised

Chemistry & B

of numerous nonproteinogenic amino acids, including eight

t-leucine residues, eight g-N-methylasparagines, and at least

four additional unusual amino acids (Freeman et al., 2012; Ham-

ada et al., 2005). Despite sharing structural features with nonri-

bosomal peptides, including the presence of 18 D-amino acids,

their enormous size (48 residues) merited investigation whether a

ribosomal route might be involved (Freeman et al., 2012). Using

degenerate PCR primers based on the hypothetical unmodified

precursor peptide carrying only proteinogenic amino acids, a

fragment of the precursor gene was isolated from the sponge

metagenomic DNA. The sequence information allowed isolation

of the complete gene cluster from a 980,000 clone metagenomic

library constructed from the total sponge tissue. The precursor

protein (PoyA) has an unusually long leader sequence that is

iology 20, May 23, 2013 ª2013 Elsevier Ltd All rights reserved 643

Chemistry & Biology

Perspective

homologous to nitrile hydratase-like enzymes and does not

share similarity to RiPP precursors from characterized peptide

families. However, genes for homologous precursors were found

in various sequenced genomes of cultivated bacteria, suggest-

ing the existence of a new natural product family (Haft et al.,

2010). Polytheonamides are the, so far, only attributed members

of this family, which is termed proteusins.

Proteusin pathways exhibit some highly unusual enzymology.

For biosynthesis of 23, up to 48 posttranslational modifications

are necessary, making these compounds themost heavily modi-

fied RiPPs known to date (Figure 5). For four hydroxylations, at

least 21 methylations and 18 epimerizations, only six modifying

enzyme candidates were identified. Functions for several of the

modifying enzyme candidates were assessed by in vitro coex-

pression with PoyA. Most and perhaps all epimerizations are

catalyzed by a single epimerase, PoyD. This unprecedented

enzyme is a member of the radical S-adenosylmethionine

(rSAM) superfamily and performs the epimerization in a unidirec-

tional fashion rather than generating an equilibrium. PoyF shows

homology to the dehydratase domain of bidomain lanthionine

synthetases and catalyzes dehydration of a threonine residue

that is eventually converted to the unusual N-acyl residue. One

or two rSAM methyltranferases, PoyB and C, are proposed to

perform a total of four methylations of the modified Thr, resulting

in a net tert-butylation. This intermediate is then hydrolytically

cleaved to release the leader peptide and yield the final N-acyl

moiety. A single SAM-dependent N-methyltransferase, PoyE,

iteratively methylates eight asparagine residues. The function

of the rSAM methyltransferases, PoyB and C, along with the

Fe(II)/a-ketoglutarate-dependent oxygenase, PoyI, remains to

be fully elucidated but are hypothesized to be involved in the

23C-methylations and four hydoxylations of the polytheonamide

precursor, respectively. Homologs of the polytheonamide en-

zymes, aswell as enzymeswith as-yet unknown function, are en-

coded in many other proteusin gene clusters in a wide range of

bacteria. Thus, similar to cyanobactins and other examples dis-

cussed in this perspective, this is another case in which studies

on uncultivated bacteria have set the stage for the discovery of

new structural and enzymatic diversity on a broader scope.

The Way ForwardAs described here, metagenomic approaches to natural product

drug discovery and biosynthetic understanding have not only

facilitated the functional characterization of uncultivated bacte-

ria but have also provided valuable insights into previously un-

known biochemistry in cultivatable organisms. Yet there are hur-

dles that must be overcome in order to fully maximize the

potential of metagenomics for natural product drug discovery

and development. As NGS technologies improve and become

cheaper and more routine, the amount of data an individual lab-

oratory can generate can become overwhelming fairly rapidly.

Automated genome mining tools, such as the recently devel-

oped antiSMASH (Medema et al., 2011) or ClusterMine360 (Con-

way and Boddy, 2013), will become critical when working with

multiple large data sets. However, since these methods rely on

homology-based queries, the identification of truly novel chem-

istry will likely rely on new strategies for bioinformatic genome

mining. These may include developing pattern recognition-

based algorithms designed to find conserved yet unexplored

644 Chemistry & Biology 20, May 23, 2013 ª2013 Elsevier Ltd All righ

open reading frames associated with natural product biosyn-

thesis. Additionally, genomics-guided functional screening and

heterologous expression is a logical next step in natural product

drug discovery and development as NGS sequencing technolo-

gies improve and prices drop. Host selection and engineering

based on phylogenetic and comparative genomic approaches

may assist in selecting the most appropriate universal host

for a particular environmental sample. A (meta)genome-first

approach may also provide a better understanding of compat-

ible gene regulation and resistance in order to identify and engi-

neer universal hosts that can express genes from awide range of

sources, as well as be used directly for library construction. This

will also require improved vectors that can carry and express

entire gene clusters stably and reliably.

We should not, however, turn away from cultivation. Recent

advances in cocultivation and environmental simulation have

shown that uncultured bacteria are not necessarily unculturable

(Stewart, 2012). Using semiporous culture chambers, two or

more organisms can be grown separately, yet a membrane al-

lows for the exchange of small molecules, such as nutrients

and growth factors. Other techniques, such as using natural

seawater as the culture medium (Kaeberlein et al., 2002) or the

use of microfluidics, have also proven successful for isolation

of rare bacterial taxa (Liu et al., 2009). Integration of ‘‘traditional’’

metagenomics with the latest tools in single-cell analysis (Blai-

ney, 2013; Kamke et al., 2012), metatranscriptomics (Yu and

Zhang, 2012), and metaproteomics (Muth et al., 2013) will

certainly aid in unveiling not only the biosynthetic potential of

uncultivated microbes but also how to express their genes het-

erologously and how to grow them effectively under laboratory

conditions.

ACKNOWLEDGMENTS

M.C.W. is funded by the Alexander von Humboldt Foundation.

REFERENCES

Achtman, M., and Wagner, M. (2008). Microbial diversity and the genetic na-ture of microbial species. Nat. Rev. Microbiol. 6, 431–440.

Arnison, P.G., Bibb, M.J., Bierbaum, G., Bowers, A.A., Bugni, T.S., Bulaj, G.,Camarero, J.A., Campopiano, D.J., Challis, G.L., Clardy, J., et al. (2013). Ribo-somally synthesized and post-translationally modified peptide natural prod-ucts: overview and recommendations for a universal nomenclature. Nat.Prod. Rep. 30, 108–160.

Aziz, R.K., Bartels, D., Best, A.A., DeJongh, M., Disz, T., Edwards, R.A., For-msma, K., Gerdes, S., Glass, E.M., Kubal, M., et al. (2008). The RAST Server:rapid annotations using subsystems technology. BMC Genomics 9, 75.

Bentley, S.D., Chater, K.F., Cerdeno-Tarraga, A.M., Challis, G.L., Thomson,N.R., James, K.D., Harris, D.E., Quail, M.A., Kieser, H., Harper, D., et al.(2002). Complete genome sequence of themodel actinomycete Streptomycescoelicolor A3(2). Nature 417, 141–147.

Berdy, J. (2005). Bioactive microbial metabolites. J. Antibiot. (Tokyo) 58, 1–26.

Bewley, C.A., Holland, N.D., and Faulkner, D.J. (1996). Two classes of metab-olites from Theonella swinhoei are localized in distinct populations of bacterialsymbionts. Experientia 52, 716–722.

Blainey, P.C. (2013). The future is now: single-cell genomics of bacteria andarchaea. FEMS Microbiol. Rev. 37, 407–427. http://dx.doi.org/10.1111/1574-6976.12015.

ts reserved

Chemistry & Biology

Perspective

Brady, S.F., and Clardy, J. (2000). Long-chain N-acyl amino acid antibioticsisolated from heterologously expressed environmental DNA. J. Am. Chem.Soc. 122, 12903–12904.

Brady, S.F., and Clardy, J. (2003). Synthesis of long-chain fatty acid enol estersisolated from an environmental DNA clone. Org. Lett. 5, 121–124.

Brady, S.F., and Clardy, J. (2005a). Cloning and heterologous expression ofisocyanide biosynthetic genes from environmental DNA. Angew. Chem. Int.Ed. Engl. 44, 7063–7065.

Brady, S.F., and Clardy, J. (2005b). N-acyl derivatives of arginine and trypto-phan isolated from environmental DNA expressed in Escherichia coli. Org.Lett. 7, 3613–3616.

Brady, S.F., and Clardy, J. (2005c). Systematic investigation of the Escherichiacolimetabolome for the biosynthetic origin of an isocyanide carbon atom. An-gew. Chem. Int. Ed. Engl. 44, 7045–7048.

Brady, S.F., Chao, C.J., Handelsman, J., and Clardy, J. (2001). Cloning andheterologous expression of a natural product biosynthetic gene cluster fromeDNA. Org. Lett. 3, 1981–1984.

Brady, S.F., Chao, C.J., and Clardy, J. (2002). New natural product familiesfrom an environmental DNA (eDNA) gene cluster. J. Am. Chem. Soc. 124,9968–9969.

Brady, S.F., Bauer, J.D., Clarke-Pearson, M.F., and Daniels, R. (2007). Naturalproducts from isnA-containing biosynthetic gene clusters recovered from thegenomes of cultured and uncultured bacteria. J. Am. Chem. Soc. 129, 12102–12103.

Brady, S.F., Simmons, L., Kim, J.H., and Schmidt, E.W. (2009). Metagenomicapproaches to natural products from free-living and symbiotic organisms. Nat.Prod. Rep. 26, 1488–1503.

Busch, B., and Hertweck, C. (2009). Evolution of metabolic diversity in polyke-tide-derived pyrones: using the non-colinear aureothin assembly line as amodel system. Phytochemistry 70, 1833–1840.

Chang, F.Y., and Brady, S.F. (2011). Cloning and characterization of an envi-ronmental DNA-derived gene cluster that encodes the biosynthesis of the anti-tumor substance BE-54017. J. Am. Chem. Soc. 133, 9996–9999.

Cheng, Y.-Q., Tang, G.-L., and Shen, B. (2003). Type I polyketide synthaserequiring a discrete acyltransferase for polyketide biosynthesis. Proc. Natl.Acad. Sci. USA 100, 3149–3154.

Chung, E.J., Lim, H.K., Kim, J.C., Choi, G.J., Park, E.J., Lee, M.H., Chung,Y.R., and Lee, S.W. (2008). Forest soil metagenome gene cluster involved inantifungal activity expression in Escherichia coli. Appl. Environ. Microbiol.74, 723–730.

Clarke-Pearson,M.F., and Brady, S.F. (2008). Paerucumarin, a newmetaboliteproduced by the pvc gene cluster from Pseudomonas aeruginosa. J. Bacteriol.190, 6927–6930.

Conway, K.R., and Boddy, C.N. (2013). ClusterMine360: a database of micro-bial PKS/NRPS biosynthesis. Nucleic Acids Res. 41, D402–D407.

Craig, J.W., Chang, F.Y., and Brady, S.F. (2009). Natural products from envi-ronmental DNA hosted in Ralstonia metallidurans. ACS Chem. Biol. 4, 23–28.

Craig, J.W., Chang, F.Y., Kim, J.H., Obiajulu, S.C., and Brady, S.F. (2010). Ex-panding small-molecule functional metagenomics through parallel screeningof broad-host-range cosmid environmental DNA libraries in diverse proteo-bacteria. Appl. Environ. Microbiol. 76, 1633–1641.

Donia, M.S., Hathaway, B.J., Sudek, S., Haygood,M.G., Rosovitz, M.J., Ravel,J., and Schmidt, E.W. (2006). Natural combinatorial peptide libraries in cyano-bacterial symbionts of marine ascidians. Nat. Chem. Biol. 2, 729–735.

Donia, M.S., Ravel, J., and Schmidt, E.W. (2008). A global assembly line for cy-anobactins. Nat. Chem. Biol. 4, 341–343.

Donia, M.S., Fricke, W.F., Partensky, F., Cox, J., Elshahawi, S.I., White, J.R.,Phillippy, A.M., Schatz, M.C., Piel, J., Haygood, M.G., et al. (2011a). Complexmicrobiome underlying secondary and primary metabolism in the tunicate-Prochloron symbiosis. Proc. Natl. Acad. Sci. USA 108, E1423–E1432.

Donia, M.S., Ruffner, D.E., Cao, S., and Schmidt, E.W. (2011b). Accessing thehidden majority of marine natural products through metagenomics.ChemBioChem 12, 1230–1236.

Chemistry & B

Feng, Z., Kim, J.H., and Brady, S.F. (2010). Fluostatins produced by the heter-ologous expression of a TAR reassembled environmental DNA derived type IIPKS gene cluster. J. Am. Chem. Soc. 132, 11902–11903.

Feng, Z., Kallifidas, D., and Brady, S.F. (2011). Functional analysis of environ-mental DNA-derived type II polyketide synthases reveals structurally diversesecondary metabolites. Proc. Natl. Acad. Sci. USA 108, 12629–12634.

Feng, Z., Chakraborty, D., Dewell, S.B., Reddy, B.V., and Brady, S.F. (2012).Environmental DNA-encoded antibiotics fasamycins A and B inhibit FabF intype II fatty acid biosynthesis. J. Am. Chem. Soc. 134, 2981–2987.

Fisch, K.M., Gurgui, C., Heycke, N., van der Sar, S.A., Anderson, S.A., Webb,V.L., Taudien, S., Platzer, M., Rubio, B.K., Robinson, S.J., et al. (2009). Polyke-tide assembly lines of uncultivated sponge symbionts from structure-basedgene targeting. Nat. Chem. Biol. 5, 494–501.

Freeman, M.F., Gurgui, C., Helf, M.J., Morinaka, B.I., Uria, A.R., Oldham, N.J.,Sahl, H.G., Matsunaga, S., and Piel, J. (2012). Metagenome mining revealspolytheonamides as posttranslationally modified ribosomal peptides. Science338, 387–390.

Gillespie, D.E., Brady, S.F., Bettermann, A.D., Cianciotto, N.P., Liles, M.R.,Rondon, M.R., Clardy, J., Goodman, R.M., and Handelsman, J. (2002). Isola-tion of antibiotics turbomycin a and B from ametagenomic library of soil micro-bial DNA. Appl. Environ. Microbiol. 68, 4301–4306.

Grindberg, R.V., Ishoey, T., Brinza, D., Esquenazi, E., Coates, R.C., Liu, W.T.,Gerwick, L., Dorrestein, P.C., Pevzner, P., Lasken, R., and Gerwick, W.H.(2011). Single cell genome amplification accelerates identification of the apra-toxin biosynthetic pathway from a complex microbial assemblage. PLoS ONE6, e18565.

Guan, C., Ju, J., Borlee, B.R., Williamson, L.L., Shen, B., Raffa, K.F., and Han-delsman, J. (2007). Signal mimics derived from a metagenomic analysis of thegypsy moth gut microbiota. Appl. Environ. Microbiol. 73, 3669–3676.

Gulder, T.A., Freeman, M.F., and Piel, J. (2011). The catalytic diversity of multi-modular polyketide synthases: natural product biosynthesis beyond textbookassembly rules. Top. Curr. Chem.. http://dx.doi.org/10.1007/128_2010_113

Gurgui, C., and Piel, J. (2010). Metagenomic approaches to identify and isolatebioactive natural products from microbiota of marine sponges. Methods Mol.Biol. 668, 247–264.

Haft, D.H., Basu, M.K., and Mitchell, D.A. (2010). Expansion of ribosomallyproduced natural products: a nitrile hydratase- and Nif11-related precursorfamily. BMC Biol. 8, 70–94.

Hamada, T., Matsunaga, S., Yano, G., and Fusetani, N. (2005). Polytheona-mides A and B, highly cytotoxic, linear polypeptides with unprecedentedstructural features, from the marine sponge, Theonella swinhoei. J. Am.Chem. Soc. 127, 110–118.

Handelsman, J. (2004). Metagenomics: application of genomics to unculturedmicroorganisms. Microbiol. Mol. Biol. Rev. 68, 669–685.

Hildebrand, M., Waggoner, L.E., Liu, H., Sudek, S., Allen, S., Anderson, C.,Sherman, D.H., and Haygood, M. (2004). bryA: an unusual modular polyketidesynthase gene from the uncultivated bacterial symbiont of the marine bryo-zoan Bugula neritina. Chem. Biol. 11, 1543–1552.

Huang, G., Zhang, L., and Birch, R.G. (2001). Amultifunctional polyketide-pep-tide synthetase essential for albicidin biosynthesis in Xanthomonas albilineans.Microbiology 147, 631–642.

Irschik, H., Kopp, M., Weissman, K.J., Buntin, K., Piel, J., andMuller, R. (2010).Analysis of the sorangicin gene cluster reinforces the utility of a combinedphylogenetic/retrobiosynthetic analysis for deciphering natural productassembly by trans-AT PKS. ChemBioChem 11, 1840–1849.

Jiao, J.Y., Wang, H.X., Zeng, Y., and Shen, Y.M. (2006). Enrichment for mi-crobes living in association with plant tissues. J. Appl. Microbiol. 100,830–837.

Johnston, C.W., Zvanych, R., Khyzha, N., andMagarvey, N.A. (2013). Nonribo-somal assembly of natural lipocyclocarbamate lipoprotein-associated phos-pholipase inhibitors. ChemBioChem 14, 431–435.

Kaeberlein, T., Lewis, K., and Epstein, S.S. (2002). Isolating "uncultivable" mi-croorganisms in pure culture in a simulated natural environment. Science 296,1127–1129.

iology 20, May 23, 2013 ª2013 Elsevier Ltd All rights reserved 645

Chemistry & Biology

Perspective

Kamke, J., Bayer, K., Woyke, T., and Hentschel, U. (2012). Exploring symbio-ses by single-cell genomics. Biol. Bull. 223, 30–43.

Kim, J.H., Feng, Z., Bauer, J.D., Kallifidas, D., Calle, P.Y., and Brady, S.F.(2010). Cloning large natural product gene clusters from the environment:piecing environmental DNA gene clusters back together with TAR. Biopoly-mers 93, 833–844.

King, R.W., Bauer, J.D., and Brady, S.F. (2009). An environmental DNA-derived type II polyketide biosynthetic pathway encodes the biosynthesis ofthe pentacyclic polyketide erdacin. Angew. Chem. Int. Ed. Engl. 48, 6257–6261.

Koehnke, J., Bent, A., Houssen, W.E., Zollman, D., Morawitz, F., Shirran, S.,Vendome, J., Nneoyiegbe, A.F., Trembleau, L., Botting, C.H., et al. (2012).The mechanism of patellamide macrocyclization revealed by the characteriza-tion of the PatG macrocyclase domain. Nat. Struct. Mol. Biol. 19, 767–772.

Komatsu, M., Komatsu, K., Koiwai, H., Yamada, Y., Kozone, I., Izumikawa, M.,Hashimoto, J., Takagi, M., Omura, S., Shin-ya, K., et al. (2013). EngineeredStreptomyces avermitilis host for heterologous expression of biosyntheticgene cluster for secondary metabolites (ACS Synth. Biol.). Published onlineJanuary 10, 2013. http://dx.doi.org/10.1021/sb3001003.

Kusebauch, B., Busch, B., Scherlach, K., Roth, M., and Hertweck, C. (2010).Functionally distinct modules operate two consecutive a,b—>b,g double-bond shifts in the rhizoxin polyketide assembly line. Angew. Chem. Int. Ed.Engl. 49, 1460–1464.

Kvist, T., Ahring, B.K., Lasken, R.S., and Westermann, P. (2007). Specific sin-gle-cell isolation and genomic amplification of uncultured microorganisms.Appl. Microbiol. Biotechnol. 74, 926–935.

Leikoski, N., Fewer, D.P., and Sivonen, K. (2009). Widespread occurrence andlateral transfer of the cyanobactin biosynthesis gene cluster in cyanobacteria.Appl. Environ. Microbiol. 75, 853–857.

Lenk, S., Moraru, C., Hahnke, S., Arnds, J., Richter, M., Kube, M., Reinhardt,R., Brinkhoff, T., Harder, J., Amann, R., andMußmann, M. (2012).Roseobacterclade bacteria are abundant in coastal sediments and encode a novel combi-nation of sulfur oxidation genes. ISME J. 6, 2178–2187.

Liles, M.R., Williamson, L.L., Rodbumrer, J., Torsvik, V., Goodman, R.M., andHandelsman, J. (2008). Recovery, purification, and cloning of high-molecular-weight DNA from soil microorganisms. Appl. Environ. Microbiol. 74, 3302–3305.

Liu, W., Kim, H.J., Lucchetta, E.M., Du, W., and Ismagilov, R.F. (2009). Isola-tion, incubation, and parallel functional testing and identification by FISH ofrare microbial single-copy cells from multi-species mixtures using the combi-nation of chemistrode and stochastic confinement. Lab Chip 9, 2153–2162.

Liu, X., Zhang, Q., Murata, K., Baker, M.L., Sullivan, M.B., Fu, C., Dougherty,M.T., Schmid, M.F., Osburne, M.S., Chisholm, S.W., and Chiu, W. (2010).Structural changes in a marine podovirus associated with release of itsgenome into Prochlorococcus. Nat. Struct. Mol. Biol. 17, 830–836.

Long, P.F., Dunlap, W.C., Battershill, C.N., and Jaspars, M. (2005). Shotguncloning and heterologous expression of the patellamide gene cluster as astrategy to achieving sustained metabolite production. ChemBioChem 6,1760–1765.

Lorenz, P., and Eck, J. (2005). Metagenomics and industrial applications. Nat.Rev. Microbiol. 3, 510–516.

Matilla, M.A., Stockmann, H., Leeper, F.J., and Salmond, G.P.C. (2012). Bac-terial biosynthetic gene clusters encoding the anti-cancer haterumalide classof molecules: biogenesis of the broad spectrum antifungal and anti-oomycetecompound, oocydin A. J. Biol. Chem. 287, 39125–39138.

McIntosh, J.A., and Schmidt, E.W. (2010). Marine molecular machines: heter-ocyclization in cyanobactin biosynthesis. ChemBioChem 11, 1413–1421.

McIntosh, J.A., Robertson, C.R., Agarwal, V., Nair, S.K., Bulaj, G.W., andSchmidt, E.W. (2010a). Circular logic: nonribosomal peptide-like macrocycli-zation with a ribosomal peptide catalyst. J. Am. Chem. Soc. 132, 15499–15501.

McIntosh, J.A., Donia, M.S., and Schmidt, E.W. (2010b). Insights into hetero-cyclization from two highly similar enzymes. J. Am. Chem. Soc. 132, 4089–4091.

646 Chemistry & Biology 20, May 23, 2013 ª2013 Elsevier Ltd All righ

McIntosh, J.A., Donia, M.S., Nair, S.K., and Schmidt, E.W. (2011). Enzymaticbasis of ribosomal peptide prenylation in cyanobacteria. J. Am. Chem. Soc.133, 13698–13705.

Medema, M.H., Blin, K., Cimermancic, P., de Jager, V., Zakrzewski, P., Fisch-bach, M.A., Weber, T., Takano, E., and Breitling, R. (2011). antiSMASH: rapididentification, annotation and analysis of secondary metabolite biosynthesisgene clusters in bacterial and fungal genome sequences. Nucleic Acids Res.39, W339–W346.

Miao, V., and Davies, J. (2009). Metagenomics and antibiotic discovery fromuncultivated bacteria. In Uncultivated Microorganisms, S.S. Epstein, ed. (Ber-lin Heidelberg: Springer-Verlag), pp. 217–236.

Moldenhauer, J., Gotz, D.C., Albert, C.R., Bischof, S.K., Schneider, K., Sus-smuth, R.D., Engeser, M., Gross, H., Bringmann, G., and Piel, J. (2010). Thefinal steps of bacillaene biosynthesis in Bacillus amyloliquefaciens FZB42:direct evidence for beta, gamma dehydration by a trans-acyltransferase poly-ketide synthase. Angew. Chem. Int. Ed. Engl. 49, 1465–1467.

Muth, T., Benndorf, D., Reichl, U., Rapp, E., and Martens, L. (2013). Searchingfor a needle in a stack of needles: challenges in metaproteomics data analysis.Mol. Biosyst. 9, 578–585. http://dx.doi.org/10.1039/C2MB25415H.

Nguyen, T., Ishida, K., Jenke-Kodama, H., Dittmann, E., Gurgui, C., Hochmuth,T., Taudien, S., Platzer, M., Hertweck, C., and Piel, J. (2008). Exploiting themosaic structure of trans-acyltransferase polyketide synthases for naturalproduct discovery and pathway dissection. Nat. Biotechnol. 26, 225–233.

Nikolouli, K., and Mossialos, D. (2012). Bioactive compounds synthesized bynon-ribosomal peptide synthetases and type-I polyketide synthases discov-ered through genome-mining and metagenomics. Biotechnol. Lett. 34,1393–1403.

Ouyang, Y., Dai, S., Xie, L., Ravi Kumar, M.S., Sun, W., Sun, H., Tang, D., andLi, X. (2010). Isolation of high molecular weight DNA from marine sponge bac-teria for BAC library construction. Mar. Biotechnol. (NY) 12, 318–325.

Owen, J.G., Robins, K.J., Parachin, N.S., and Ackerley, D.F. (2012). A func-tional screen for recovery of 40-phosphopantetheinyl transferase and associ-ated natural product biosynthesis genes from metagenome libraries. Environ.Microbiol. 14, 1198–1209.

Paitan, Y., Alon, G., Orr, E., Ron, E.Z., and Rosenberg, E. (1999). The first genein the biosynthesis of the polyketide antibiotic TA of Myxococcus xanthus co-des for a unique PKS module coupled to a peptide synthetase. J. Mol. Biol.286, 465–474.

Piel, J. (2002). A polyketide synthase-peptide synthetase gene cluster from anuncultured bacterial symbiont of Paederus beetles. Proc. Natl. Acad. Sci. USA99, 14002–14007.

Piel, J. (2010). Biosynthesis of polyketides by trans-AT polyketide synthases.Nat. Prod. Rep. 27, 996–1047.

Piel, J. (2011). Approaches to capturing and designing biologically active smallmolecules produced by uncultured microbes. Annu. Rev. Microbiol. 65,431–453.

Piel, J., Hui, D., Wen, G., Butzke, D., Platzer, M., Fusetani, N., and Matsunaga,S. (2004a). Antitumor polyketide biosynthesis by an uncultivated bacterialsymbiont of the marine sponge Theonella swinhoei. Proc. Natl. Acad. Sci.USA 101, 16222–16227.

Piel, J., Hui, D., Fusetani, N., and Matsunaga, S. (2004b). Targeting modularpolyketide synthases with iteratively acting acyltransferases from metage-nomes of uncultured bacterial consortia. Environ. Microbiol. 6, 921–927.

Rath, C.M., Janto, B., Earl, J., Ahmed, A., Hu, F.Z., Hiller, L., Dahlgren, M.,Kreft, R., Yu, F., Wolff, J.J., et al. (2011). Meta-omic characterization of thema-rine invertebrate microbial consortium that produces the chemotherapeuticnatural product ET-743. ACS Chem. Biol. 6, 1244–1256.

Reyes-Duarte, D., Ferrer, M., andGarcıa-Arellano, H. (2012). Functional-basedscreening methods for lipases, esterases, and phospholipases in metage-nomic libraries. Methods Mol. Biol. 861, 101–113.

Rondon, M.R., August, P.R., Bettermann, A.D., Brady, S.F., Grossman, T.H.,Liles, M.R., Loiacono, K.A., Lynch, B.A., MacNeil, I.A., Minor, C., et al.(2000). Cloning the soil metagenome: a strategy for accessing the geneticand functional diversity of uncultured microorganisms. Appl. Environ. Micro-biol. 66, 2541–2547.

ts reserved

Chemistry & Biology

Perspective

Schmidt, E.W., and Donia, M.S. (2009). Chapter 23. Cyanobactin ribosomallysynthesized peptides–a case of deepmetagenomemining. Methods Enzymol.458, 575–596.

Schmidt, E.W., Nelson, J.T., Rasko, D.A., Sudek, S., Eisen, J.A., Haygood,M.G., and Ravel, J. (2005). Patellamide A and C biosynthesis by a microcin-like pathway in Prochloron didemni, the cyanobacterial symbiont ofLissoclinum patella. Proc. Natl. Acad. Sci. USA 102, 7315–7320.

Schmidt, E.W., Donia, M.S., McIntosh, J.A., Fricke, W.F., and Ravel, J. (2012).Origin and variation of tunicate secondary metabolites. J. Nat. Prod. 75,295–304.

Scotti, C., Piatti, M., Cuzzoni, A., Perani, P., Tognoni, A., Grandi, G., Galizzi, A.,and Albertini, A.M. (1993). A Bacillus subtilis large ORF coding for a polypep-tide highly similar to polyketide synthases. Gene 130, 65–71.

Shokralla, S., Spall, J.L., Gibson, J.F., and Hajibabaei, M. (2012). Next-gener-ation sequencing technologies for environmental DNA research. Mol. Ecol. 21,1794–1805.

Siegl, A., Kamke, J., Hochmuth, T., Piel, J., Richter, M., Liang, C., Dandekar, T.,and Hentschel, U. (2011). Single-cell genomics reveals the lifestyle of Poribac-teria, a candidate phylum symbiotically associated withmarine sponges. ISMEJ. 5, 61–70.

Stewart, E.J. (2012). Growing unculturable bacteria. J. Bacteriol. 194, 4151–4160.

Sudek, S., Lopanik, N.B., Waggoner, L.E., Hildebrand, M., Anderson, C., Liu,H., Patel, A., Sherman, D.H., and Haygood, M.G. (2007). Identification of theputative bryostatin polyketide synthase gene cluster from "Candidatus Endo-bugula sertula", the uncultivated microbial symbiont of the marine bryozoanBugula neritina. J. Nat. Prod. 70, 67–74.

Chemistry & B

Tang, G.-L., Cheng, Y.-Q., and Shen, B. (2004). Leinamycin biosynthesisrevealing unprecedented architectural complexity for a hybrid polyketide syn-thase and nonribosomal peptide synthetase. Chem. Biol. 11, 33–45.

Teta, R., Gurgui, M., Helfrich, E.J., Kunne, S., Schneider, A., Van Echten-Deck-ert, G., Mangoni, A., and Piel, J. (2010). Genome mining reveals trans-ATpolyketide synthase directed antibiotic biosynthesis in the bacterial phylumbacteroidetes. ChemBioChem 11, 2506–2512.

Udwary, D.W., Zeigler, L., Asolkar, R.N., Singan, V., Lapidus, A., Fenical, W.,Jensen, P.R., and Moore, B.S. (2007). Genome sequencing reveals complexsecondary metabolome in the marine actinomycete Salinispora tropica.Proc. Natl. Acad. Sci. USA 104, 10376–10381.

Wang, G.Y., Graziani, E.,Waters, B., Pan,W., Li, X., McDermott, J., Meurer, G.,Saxena, G., Andersen, R.J., and Davies, J. (2000). Novel natural products fromsoil DNA libraries in a streptomycete host. Org. Lett. 2, 2401–2404.

Williamson, L.L., Borlee, B.R., Schloss, P.D., Guan, C., Allen, H.K., and Han-delsman, J. (2005). Intracellular screen to identify metagenomic clones thatinduce or inhibit a quorum-sensing biosensor. Appl. Environ. Microbiol. 71,6335–6344.

Wright, G.D. (2012). Antibiotics: a new hope. Chem. Biol. 19, 3–10.

Wu, D., Hugenholtz, P., Mavromatis, K., Pukall, R., Dalin, E., Ivanova, N.N., Ku-nin, V., Goodwin, L., Wu, M., Tindall, B.J., et al. (2009). A phylogeny-drivengenomic encyclopaedia of Bacteria and Archaea. Nature 462, 1056–1060.

Xing, M.N., Zhang, X.Z., and Huang, H. (2012). Application of metagenomictechniques in mining enzymes from microbial communities for biofuel synthe-sis. Biotechnol. Adv. 30, 920–929.

Yu, K., and Zhang, T. (2012). Metagenomic andmetatranscriptomic analysis ofmicrobial community structure and gene expression of activated sludge. PLoSONE 7, e38183.

iology 20, May 23, 2013 ª2013 Elsevier Ltd All rights reserved 647