Insights into an unusual nonribosomal peptide synthetase biosynthesis: Identification and characterization of the GE81112 biosynthetic gene cluster

Article abstract of DOI:10.1074/jbc.M110.146803

The GE81112 tetrapeptides (1-3) represent a structurally unique class of antibiotics, acting as specific inhibitors of prokaryotic protein synthesis. Here we report the cloning and sequencing of the GE81112 biosynthetic gene cluster from Streptomyces sp. L-49973 and the development of a genetic manipulation system for Streptomyces sp. L-49973. The biosynthetic gene cluster for the tetrapeptide antibiotic GE81112 (getA-N) was identified within a 61.7-kb region comprising 29 open reading frames (open reading frames), 14 of which were assigned to the biosynthetic gene cluster. Sequence analysis revealed the GE81112 cluster to consist of six nonribosomal peptide synthetase (NRPS) genes encoding incomplete di-domain NRPS modules and a single free standing NRPS domain as well as genes encoding other biosynthetic and modifying proteins. The involvement of the cloned gene cluster in GE81112 biosynthesis was confirmed by inactivating the NRPS gene getE resulting in a GE81112 production abolished mutant. In addition, we characterized the NRPS A-domains from the pathway by expression in Escherichia coli and in vitro enzymatic assays. The previously unknown stereochemistry of most chiral centers in GE81112 was established from a combined chemical and biosynthetic approach. Taken together, these findings have allowed us to propose a rational model for GE81112 biosynthesis. The results further open the door to developing new derivatives of these promising antibiotic compounds by genetic engineering.

Full text of DOI:10.1074/jbc.M110.146803

Genomics and Proteomics:
Insights into an Unusual Nonribosomal
Peptide Synthetase Biosynthesis:
IDENTIFICATION AND
CHARACTERIZATION OF THE
GE81112 BIOSYNTHETIC GENE
CLUSTER
Tina M. Binz, Sonia I. Maffioli, Margherita
Sosio, Stefano Donadio and Rolf Müller
J. Biol. Chem. 2010, 285:32710-32719.
doi: 10.1074/jbc.M110.146803 originally published online August 14, 2010
Access the most updated version of this article at doi: 10.1074/jbc.M110.146803
Find articles, minireviews, Reflections and Classics on similar topics on the JBC Affinity Sites.
Alerts:
•
•
When this article is cited
When a correction for this article is posted
Click here to choose from all of JBC's e-mail alerts
Supplemental material:
http://www.jbc.org/content/suppl/2010/09/01/M110.146803.DC1.html
This article cites 53 references, 13 of which can be accessed free at
http://www.jbc.org/content/285/43/32710.full.html#ref-list-1

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 285, NO. 43, pp. 32710–32719, October 22, 2010
2010 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.
©
Insights into an Unusual Nonribosomal Peptide Synthetase
Biosynthesis
IDENTIFICATION AND CHARACTERIZATION OF THE GE81112 BIOSYNTHETIC
□
S
*
GENE CLUSTER
Received for publication, May 20, 2010, and in revised form, August 5, 2010 Published, JBC Papers in Press, August 14, 2010, DOI 10.1074/jbc.M110.146803
‡1
‡
§
§
§
Tina M. Binz , Sonia I. Maffioli , Margherita Sosio , Stefano Donadio , and Rolf M u¨ ller
‡
From the Department of Microbial Natural Products, Helmholtz Institute for Pharmaceutical Research Saarland (HIPS), Helmholtz
Center for Infection Research (HZI), and Department of Pharmaceutical Biotechnology, Saarland University, Campus C2 3,
Saarbr u¨ cken 66123, Germany and Naicons Scrl, Via G. Fantoli, 16/15, Milan 20138, Italy
§
The GE81112 tetrapeptides (1–3) represent a structurally products. The former is the target of the macrolides (e.g. eryth-
unique class of antibiotics, acting as specific inhibitors of pro- romycin), the aminoglycosides (e.g. kanamycin), the strepto-
karyotic protein synthesis. Here we report the cloning and gramins, the lincosamides, tetracycline, and chloramphenicol
sequencing of the GE81112 biosynthetic gene cluster from as well as of other classes of compounds that are not in clinical
Streptomyces sp. L-49973 and the development of a genetic use. These compounds have been shown to affect protein bio-
manipulation system for Streptomyces sp. L-49973. The biosyn- synthesis at various steps (3, 4).
thetic gene cluster for the tetrapeptide antibiotic GE81112
Some aspects of protein translation are rarely targeted, e.g.
(getA-N) was identified within a 61.7-kb region comprising 29 translation initiation and polypeptide chain termination. A micro-
open reading frames (open reading frames), 14 of which were bial product screening program aimed at discovering novel inhib-
assigned to the biosynthetic gene cluster. Sequence analysis re- itors of bacterial protein synthesis revealed the new tetrapeptide
vealed the GE81112 cluster to consist of six nonribosomal pep- GE81112 compounds (1-3) (Fig. 1) to selectively inhibit the forma-
tide synthetase (NRPS) genes encoding incomplete di-domain tion of the prokaryotic 30 S initiation complex with an IC ϭ 0.9
50
NRPS modules and a single free standing NRPS domain as well ␮M (5). To date, three GE81112 congeners, A (1), B (2), and B1 (3),
as genes encoding other biosynthetic and modifying proteins. have been described from a Streptomyces sp. (Fig. 1). Extensive
The involvement of the cloned gene cluster in GE81112 biosyn- NMR and MS studies revealed the tetrapeptides to comprise
thesis was confirmed by inactivating the NRPS gene getE result- hydroxypipecolic and hydroxypentanoic acids, an (amino)histi-
ing in a GE81112 production abolished mutant. In addition, we dine, and a hydroxychlorohistidine (5). A retro-biosynthetic anal-
2
characterized the NRPS A-domains from the pathway by ysis of the GE81112 core structure suggests a NRPS origin with
expression in Escherichia coli and in vitro enzymatic assays. The additionaltailoringstepsoccurringatsomepointduringassembly.
previously unknown stereochemistry of most chiral centers in NRPS multienzymes are composed of successive catalytic units or
GE81112 was established from a combined chemical and bio- “domains” that are themselves organized into biosynthetic mod-
synthetic approach. Taken together, these findings have allowed ules which catalyze the assembly reactions in a coordinated, often
us to propose a rational model for GE81112 biosynthesis. The “co-linear” manner. Normally, each module in the assembly line
results further open the door to developing new derivatives of performs one cycle of chain extension (condensation of one resi-
these promising antibiotic compounds by genetic engineering.
due into the growing peptide chain). A typical, minimal NRPS
module consists of an adenylation (A) domain, a peptidyl carrier
protein (PCP) domain (also referred to as a thiolation (T) domain),
The emergence of multi-drug resistant microbial pathogens and a condensation (C) domain (6). However, a growing number
is driving the search for novel antibiotics with new mechanisms of gene clusters encode systems that deviate in their domain orga-
of action and natural products continue to provide original nization from the standard C-A-PCP architecture and comprise
scaffolds affecting essential bacterial targets (1, 2). As it is cur- partial modules or isolated domains acting in trans to complement
rently understood, most antibiotics act in three basic ways; 1) the functionality of the multimodular NRPSs (7–9).
inhibition of DNA replication and repair, 2) inhibition of cell
To better understand GE81112 biosynthesis and to generate
wall biosynthesis, and 3) inhibition of protein biosynthesis. Pro- structural analogues, we sought to develop a strategy for the clon-
tein translation and cell wall biosynthesis in bacteria are cur- ing and identification of the biosynthetic gene cluster. To this end,
rently the targets for the majority of antimicrobial natural the cluster was identified on two overlapping cosmids. In addition,
we expressed five A-domains from the corresponding NRPS genes
and characterized them in vitro. These results together with the
*
The work was funded in part by the Bundesministerium f u¨ r Bildung und
Forschung and the Deutsche Forschungsgemeinschaft (to R. M.).
S
assignment of the configuration of most chiral centers have
□
The on-line version of this article (available at http://www.jbc.org) contains
supplemental Table S1 and Figs. S1–S4.
Thenucleotidesequence(s)reportedinthispaperhasbeensubmittedtotheGen-
TM
2
Bank /EBI Data Bank with accession number(s) FN821996.
The abbreviations used are: NRPS, nonribosomal peptide synthetase; A,
adenylation; PCP, peptidyl carrier protein; T, thiolation; C, condensation;
1
To whom correspondence should be addressed. Tel.: 49-681-302-70201;
Fax.: 49-681-302-70202; E-mail: rom@mx.uni-saarland.de.
O-HPA, 5-hydroxy-2-aminopentanoic acid; Pip, pipecolic acid.
3
2710 JOURNAL OF BIOLOGICAL CHEMISTRY
VOLUME 285•NUMBER 43•OCTOBER 22, 2010

GE81112 Biosynthetic Gene Cluster
allowed us to delineate this unusual NRPS pathway and propose a arrays (Hybond Nϩ, Amersham Biosciences) were utilized for
biosynthetic model for the GE81112 antibiotics.
the screening of the cosmid clones, as described previously (13).
Screening of a Streptomyces sp. L-49973 Genomic Cosmid
Library—Cyclodeaminase and NRPS fragment sequences were
initially used as probes under low stringency conditions. To
create the cyclodeaminase probe, we used specific primers to
amplify two known cyclodeaminase genes, tubZ (14) (primers
Tubz_up and TubZ_down, 570-bp product)m from the tubuly-
sin cluster, and rapL (15), from the rapamycin cluster (primers
RapL_up and RapL_down, 534-bp PCR product) (see the prim-
ers in supplemental Table S1). As an alternative approach, we
designed an additional probe to identify NRPS genes. For this,
we used degenerate NRPS primers; NRPS-A1-up and NRPS-
H1-dn (supplemental Table S1). These primers amplify A-do-
mains between the structural regions A3 and A6 (16), where the
amino acid binding pocket is located, giving a 760-bp PCR frag-
ment. The amplicons were labeled with digoxigenin and used to
probe the cosmid library at low stringency (40 °C). Several cos-
mids that hybridized with both probes were subjected to PCR
analysis for the amplification of the cyclodeaminase and NRPS
A-domains. In the first screening we identified a cosmid encod-
ing a putative cyclodeaminase but not the expected GE81112
biosynthetic enzymes. A segment of this cyclodeaminase was
amplified using the primers Cyclo_probe_for and Cyclo_probe_
rev (supplemental Table S1) and used under high stringency
conditions (42 °C) for further screening of the library. Cosmids
hybridizing with this probe were analyzed by PCR using prim-
ers to amplify the cyclodeaminase and NRPS A-domains again.
The resulting PCR products were gel-purified and subcloned
into pCR2.1 TOPO vector and sequenced. Several NRPS
sequences were found from different cosmids, and the eight
critical residues responsible for substrate recognition could be
determined enabling an in silico prediction of the substrate
specificity of each cloned A-domain fragment. Cosmids har-
boring A-domains that were predicted to activate pipecolic acid
were digested with BamHI, and the restriction pattern was
compared with identify similar cosmids. End sequencing of the
cosmids was carried out using primers T4 and T7 (supplemen-
tal Table S1). One cosmid (BI11) was identified containing
a large part of the GE81112 biosynthetic gene cluster. To find a
cosmid that overlapped with the 3Ј (T7) end of the cluster, a
EXPERIMENTAL PROCEDURES
Bacterial Strains and Culture Conditions—Streptomyces sp.
Ϫ1
L-49973 was grown in INA5 (glycerol, 30 g liter ; soya extract,
Ϫ1
Ϫ1
Ϫ1
1
5 g liter ; CaCO , 5 g liter ; NaCl, 2 g liter ) or T6 (glyc-
Ϫ1 Ϫ1 Ϫ1
3
erol, 45 g liter ; soya extract, 25 g liter ; CaCO , 2 g liter
)
3
media in baffled flasks for production of GE81112. Pre-cultures
Ϫ1
were grown in V6 medium (glucose, 20 g liter ; meat extract,
Ϫ1
Ϫ1
Ϫ1
5
3
g liter ; yeast extract, 5 g liter ; peptone 5 g liter ; caseine,
Ϫ1 Ϫ1
g liter ; NaCl, 1.5 g liter , pH 7.5) and used to inoculate
production culture (1:100). The cultures were maintained at
3
0 °C and 180 rpm on a rotary incubator and harvested after 6
days. Escherichia coli DH10B, E. coli ET12567/pUZ8002, and
E. coli SURE were grown in liquid LB medium at 37 or 30 °C
with the appropriate antibiotic selection. Antibiotic concentra-
Ϫ1
tions were as follows; apramycin (60 ␮g ml ), chlorampheni-
Ϫ1
Ϫ1
col (34 ␮g ml ), kanamycin (60 ␮g/ml ), and ampicillin
Ϫ1
(
100 ␮g ml ) were used for selection in E. coli. Apramycin (60
Ϫ1
␮
g ml ) was used for selection of Streptomyces sp. L-49973
Ϫ1
recombinants. Nalidixic acid (25 ␮l ml ) was used to select
against E. coli donor after conjugation.
Molecular Biology Methods—The pET28b (ϩ) (Novagen)
and pCR2.1 TOPO (Invitrogen) cloning vectors were from
commercial sources; pKC1132 and pOJ436 were described pre-
viously (10). Restriction enzymes were purchased from MBI
Fermentas. All PCRs were carried out using Taq (MBI Fermen-
tas) or Phusion (Invitrogen) polymerase. DMSO was added to
the reaction mixture to a final concentration of 5%. Conditions
for amplification with a Peqlab ThermoCycler were as follows:
denaturation, 15–30 s at 95/98 °C; annealing, 8–20 s at
5
0–62 °C; extension, 15–60 s at 72 °C (30 cycles); final exten-
sion at 72 °C for 10 min. Oligonucleotides were obtained from
TM
Sigma. Plasmid DNA was isolated using the GeneJET
Plas-
mid Miniprep kit (Fermentas). DNA fragments from agarose
gel were isolated and purified using the NucleoSpin Extract II
kit (Macherey-Nagel). All other DNA manipulations in E. coli
(
11) and Streptomyces (10) were carried out using standard pro-
tocols. For colony hybridization analysis, digoxigenin labeling
of DNA probes, hybridization, and detection were performed
according to the manufacturer’s protocol (Roche Diagnostics).
Construction of a Streptomyces sp. L-49973 Genomic Cosmid
Library—A Streptomyces sp. L-49973 genomic cosmid library
was constructed in E. coli SURE. Genomic DNA was isolated
1
-kb fragment was amplified from the T7 end of cosmid BI11
(
primers BI11-T7end-for and BI11-T7end-rev, supplemental
Table S1) to serve as a probe. The cosmid library was then
screened with the new probe, with hybridization at 42 °C.
Among the identified cosmids, BA23 showed the smallest
with the salting out procedure (10). Then the DNA was partially extent of overlap with BI11 based on PCR analysis and restric-
digested with Sau3A, yielding fragments with an average size tion digest. Cosmids BI11 and BA23 were shotgun-sequenced
greater than 35 kb. The fragments were ligated into cosmid on both strands as described previously (17).
vector pOJ436 (12) digested with BamHI and PvuII and in vitro
Data Analysis—The annotation analysis of the sequence
packaged with the Gigapack III Gold packaging extract kit data was performed through FramePlot analysis (FramePlot
according to the manufacturer’s handbook (Stratagene). 2304 4.Obeta) (23) and data base comparison with the basic align-
colonies were transferred into six 384-well microtiter plates ment search tool (BLAST) on the server of the National Center
using a Qbot robot (Genetix) and grown overnight in 2YT for Biotechnology Information. For alignment analysis of the
medium. For storage at Ϫ80 °C, 50 ␮l of freezing solution sequence data, ClustalW on the server of EMBL-EBI was used.
(
0.076% MgSO ꢀ7 H O, 0.45% sodium citrateꢀ2 H O, 0.9% Specificity of the A-domains was determined by using the
4
2
2
NH SO , 44% glycerol, 4.7% K HPO , 1.8% KH PO ) was NRPS predictor Bioinformatics Toolbox from University of
4
4
2
4
2
4
added to each well. Robotically produced high density colony T u¨ bingen and polyketide synthase/NRPS analysis web site.
OCTOBER 22, 2010•VOLUME 285•NUMBER 43 JOURNAL OF BIOLOGICAL CHEMISTRY 32711

GE81112 Biosynthetic Gene Cluster
Conjugation and Generation of Mutant Strains of Streptomy- and cloned into pET28b (ϩ). Final plasmids were sequenced
ces sp. L-49973—DNA manipulation was carried out with and transformed into E. coli Rosetta BL21 (DE3) pLysS/RARE
E. coli DH10B as the host strain. To generate knock-out for protein expression.
mutants by means of insert-directed homologous recombina-
Expression and Purification of the A-domains—Purified
tion, a 572-bp internal fragment of gene getE was amplified with A-domain pET28b (ϩ) plasmids were transformed into E. coli
primers pipA_for and pipA_rev (supplemental Table S1). The Rosetta BL21 (DE3) pLysS/RARE competent cells for protein
fragments were cloned into pCR2.1TOPO, and the constructs production and purification. Fresh transformants harboring
were digested with EcoRI. The fragments were ligated into the constructs were grown in LB-medium (1-liter batches
knock-out vector pKC1132 (10) digested with EcoRI. The final started with 0.1% inocula from a 10-ml culture grown for 5 h at
Ϫ1
pKC1132-derived plasmids were introduced into Streptomyces 37 °C) supplemented with kanamycin (50 ␮g ml ) and chlor-
Ϫ1
sp. L-49973 by intergeneric conjugation with the methylation- amphenicol (34 ␮g ml ). All cells were grown at 37 °C to an
deficient donor strain E. coli ET12567 containing the conjuga- OD of ϳ0.8. The cells were then induced with 1 M isopropyl-
6
00
tive vector pUZ8002 (10). Mutants were analyzed by PCR using ␤-D-thiogalactopyranoside to an end concentration of 0.2 mM
appropriate control primers. One primer was designed to bind and then grown at 16 °C overnight. The cells were harvested by
to the integrated vector backbone (lacZ1, lacZ2), whereas the centrifugation (6000 rpm, 10 min, 4 °C) and resuspended in
second primer was designed to target the genome sequence buffer A (20 mM Tris-HCl, pH 7.8, 200 mM NaCl, and 10% (v/v)
either up- or downstream of the integration site (A1_for, glycerol). The cells were then lysed (2 passes at 700 p.s.i., French
A1_rev) (supplemental Table S1). PCR of the mutants yielded press, SLM Aminco), and the cell debris was removed by cen-
TM
distinct amplicons, whereas no products were detected from trifugation (21,000 rpm, 10 min, 4 °C). Prepacked HisTrap
the wild type.
HP columns were used for preparative purification of histidine-
Analysis of GE81112 Production in Streptomyces sp. L-49973— tagged recombinant proteins by immobilized metal ion affinity
TM
¨
Streptomyces strains (wild types and mutants) were cultured in chromatography on the Akta prime plus system (GE Health-
5
00-ml baffled shake flasks containing 100 ml of GE81112 pro- care). 15-ml protein lysates were filtered through a sterile filter
duction medium (INA5 or T6) at 30 °C and 180 rpm. Recombi- and loaded onto the 1-ml HisTrap column. Purification was
Ϫ1
nant strains were amended with apramycin (60 ␮g ml ). A performed as recommended in the GE Healthcare manual (His-
square of agar from a sporulating SM agar plate was used for Trap HP, Instructions 71-5027-68 AF). The desired protein was
inoculation. After 6 days of cultivation, cells were harvested by eluted from the column in a stepwise imidazole gradient with
centrifugation at 12,000 rpm for 5 min. The culture supernatant buffer B (20 mM Tris-HCl, pH 7.8, 200 mM NaCl, and 10% (v/v)
was extracted two times with ethyl acetate, evaporated, and glycerol and 60, 100, 200, 300, and 500 mM imidazole, 10-ml
redissolved in 500 ␮l of methanol. LC-coupled FT-Orbit- fractions). Fractions containing the pure target protein, as
rap-MS analysis was carried out with an Accella UPLC system determined by SDS-PAGE, were combined and concentrated
(
Thermo Electron Corp.) operating in positive ionization mode to ϳ200 ␮l by using Amicon Ultra PL-10 centricons. Then 800
at a scan range of m/z 100–2000. A Hypersil Gold column ␮l of storage buffer (50 mM Tris-HCl, pH 7.5, 50 mM NaCl, and
(
2.1 ϫ 50 mm; Thermo Fisher Scientific) was used for separa- 10% (v/v) glycerol) was added to the concentrated protein
tion with a solvent system consisting of H O (A) and acetoni- before flash-freezing in liquid nitrogen and storage at Ϫ80 °C.
2
trile (B), each containing 0.1% formic acid. A gradient of 5–95% Protein concentrations were determined using the Bradford
B was applied over 10 min. Measurements were carried out in assay (Bio-Rad). 1–3 mg/ml purified protein were obtained for
single ion mode. GE81112 compounds were identified by com- each protein per liter of culture.
32
parison to the retention times and the MS data of authentic
Determination of Substrate Specificity by ATP-[ P]PP
i
ϩ
standards (GE congener A: [MϩH] ϭ 644.21858; GE conge- Exchange Assay—To determine substrate specificity, ATP-
ϩ
ϩ
32
ner B: [MϩH] ϭ 659.22953; GE congener B1: [MϩH]
ϭ
[ P]PP reactions (100 ␮l) containing Tris-HCl (pH 7.5, 75
i
6
58.24587) in the positive ionization mode.
mM), MgCl (10 mM), dATP (5 mM), amino acid (5 mM), and
2
3
2
Construction of A-domain Overexpression Constructs—The protein (2 ␮g) were performed at 30 °C. P-Labeled tetraso-
genes encoding for the five A-domains (GetEA , GetGA , dium pyrophosphate was obtained from PerkinElmer Life Sci-
1
2
GetGA , GetJA , and GetMA ) of GE81112 were PCR-ampli- ences (NEN #NEX019). The reactions were started by the addi-
3
4
5
3
2
fied from cosmids BI11 and BA23. The forward and reverse tion of [ P]PP (0.1 ␮Ci final amount) for up to 30 min before
i
primers for amplification of A-domain fragments getEA , quenching with charcoal suspensions (500 ␮l, 1.6% (w/v) acti-
1
getGA , getGA , and getMA introduced NdeI and BamHI vated charcoal, 0.1 M Na P O , and 0.35 M perchloric acid in
2
3
5
4
2
7
restriction sites (supplemental Table S1). A-domain fragment H O). The charcoal was pelleted by centrifugation before being
2
getJA4 could not be amplified from the cosmids, probably due washed twice with the wash solution (500 ␮l, 0.1 M Na P O ,
4
2
7
to the high GC content. Therefore, the A-domain sequence was and 0.35 M perchloric acid in H O), resuspended in H O (500
2
2
synthesized, and the GC content was optimized for use in E. coli ␮l), and counted by liquid scintillation (Beckman LS6500). The
ATG::biosynthetics GmbH). The fragment was obtained in experiments were carried out in triplicate for each substrate
pBluescript SKϩ (pBSK) vector flanked by restriction sites concentration with a negative control (no amino acid).
(
NdeI and EcoRI (supplemental Fig. S1). All PCR products were
Chemical Analyses—GE81112 was purified and hydrolyzed
cloned into the digested pET28b (ϩ) vector using the corre- as described by Brandi et al. (5). Dehalogenation of GE81112
sponding NdeI/BamHI restriction sites. Fragment getJA was was carried out under H at atmospheric pressure and room
4
2
obtained after restriction of pBSK/GetJA4 with NdeI and EcoRI temperature in 10% acetic acid, with 10% palladium/carbon as
3
2712 JOURNAL OF BIOLOGICAL CHEMISTRY
VOLUME 285•NUMBER 43•OCTOBER 22, 2010

GE81112 Biosynthetic Gene Cluster
specificity revealed one cosmid (BI11) containing an A-domain
with predicted substrate specificity for proline/pipecolic acid,
as expected for the GE81112 starter unit (Table 1). Subsequent,
complete sequencing of the cosmid revealed several genes
expected for GE81112 biosynthesis. As the entire gene cluster
was not present on the cosmid, we identified overlapping cos-
mids. In total, seven cosmids were identified, and after verifica-
tion by PCR and restriction analysis, a single one (BA23) was
selected for further analysis.
FIGURE 1. Chemical structures of GE81112 tetrapeptides. The GE81112
antibiotics are three closely related NRPS tetrapeptides A (1), B (2), and B1 (3).
The absolute stereoconfiguration of each amino acid residue has now been
determined by chiral GC/MS analysis of the acid-hydrolyzed natural product.
These data indicate that the GE81112 factors 1-3 are composed of (modified)
L-amino acids.
Sequence Analysis and Organization of the get Biosynthetic
Gene Cluster—The two overlapping cosmids, BI11 and BA23,
were sequenced. The obtained sequence was analyzed for the
presence of putative open reading frames (orfs) with FramePlot
catalyst. Catalytic hydrogenation of hydroxypicolinic acid was
4
.Obeta (23), and preliminary functional assignments of indi-
performed under H atmosphere (50 p.s.i.) at room tempera-
2
vidual orfs were made by comparison of the deduced gene prod-
ture in aqueous NH OH with 10% palladium/carbon as a cata-
4
ucts with proteins of known function in the BLAST data base
lyst. Dihydroxylation of allylglycine was performed following
published procedures (18). A sample of (2R,3R)-3-hydroxy-
pipecolic acid was kindly provided from Prof. Jieping Zhu
(Table 2). Annotation of the two cosmids (BI11 and BA23)
revealed 29 orfs, of which 14, designated getA-N, are postulated
to be involved in the GE81112 biosynthetic pathway (Fig. 2A).
The first gene predicted to be involved is getA, as the proteins
encoded by the orfs upstream of getA show no homology to
proteins involved in biosynthetic pathways. getA (852 bp)
encodes a type II thioesterase. Type II thioesterases are present
in many NRPS and polyketide synthase systems, where they
perform crucial proofreading functions by hydrolyzing aber-
rant substrates from the respective carrier protein domains
(24). getA is found within an operon that also harbors the genes
getB-E. getB and getC encode proteins having homology to
known ABC transporter systems (25). The next gene, getD,
encodes the cyclodeaminase that was targeted in our library-
probing strategy. It shows 52% identity to TubZ, the cyclo-
deaminase from Angiococcus disciformis (14). The first gene
encoding a NRPS protein (a freestanding A-domain) is getE,
which likely starts with a GTG and is preceded by a putative
ribosome binding site (GGAG) 7 bp upstream of the start
codon. The next gene, getF, is oriented in the opposite direction
and encodes a protein with homology to a putative L-(2S)-pro-
line 3-hydroxylase. The following gene, getG (7212 bp), encodes
another NRPS and is the likely starting point of a new operon
which includes getH and getI. getH (1614 bp) encodes a di-do-
main NRPS (PCP-C) enzyme and again starts with a GTG,
whereas getI exhibits homology to an oxygenase. The last seg-
ment of the cluster contains 5 genes (getJ-getN) and starts with
another change in the transcription direction. It begins with
getJ (2460 bp), which encodes a NRPS di-domain (A-PCP)
enzyme. The next two genes, getK and getL, encode a GTPase
protein and a halogenase, respectively. The NRPS-di-domain
(
CNRS, Gif-sur-Yvette, France). Chiral GC-MS analyses were
performed on the methyl esters, and trifluoroacetyl derivatives
were analyzed using a Finnigan TSQ700 triple stage quadrupole
mass spectrometer interfaced with a Varian 3400 gas chro-
matographer (5). NMR and [␣] literature data for 3-hydroxy-
D
1
histidine were obtained from (19). Homonuclear H and het-
13
1
eronuclear C, H NMR experiments were recorded at 400
MHz on a Bruker Advance spectrometer in D O or in D O
2
2
acidified with trifluoroacetic acid (TFA).
RESULTS
Isolation of the GE81112 Biosynthetic Gene Cluster—To cap-
ture the GE81112 biosynthetic gene cluster (get, for GE81112
tetrapeptide) a cosmid library containing 2304 clones was
generated from the genomic DNA of the GE81112 producer
strain (Streptomyces sp. L-49973). Hybridization probes were
designed by applying a retrobiosynthetic strategy that allowed
us to predict some probable genetic elements of the get cluster
from analysis of the metabolite structures (Fig. 1). On this basis
we designed a set of probes using degenerate primers based on
A-domains of streptomycete origin using the CODEHOP soft-
ware (20); priming was targeted against the core A3 and A6
motifs. Furthermore, from the structure of 1 we predicted that
the starter unit likely involves pipecolic acid, which is known to
be formed from lysine via the action of a lysine cyclodeaminase
(
21). As cyclodeaminase genes are relatively rare in bacterial
genomes, this gene was used to design a second probe to iden-
tify the get cluster (22). We designed specific probes based on a
cyclodeaminase sequence that was identified in previous exper-
(A-PCP) encoding gene getM (1848 bp) is assumed to start with
3
iments in the same strain. Colony hybridization of the Strep-
a TTG, with a ribosome binding site (AGGG) located 6 bp
upstream. The last gene getN encodes a protein with homology
to a type I thioesterase. The involvement of orfs 1–5 and 6–15
in GE81112 biosynthesis is unlikely but cannot yet be excluded.
Further experiments will be carried out in the future to deter-
mine the exact boundaries of the gene cluster.
tomyces sp. L-49973 cosmid library with the cyclodeaminase
probe led to the identification of 7 cosmid hits. The cosmids
were then determined to contain the targeted cyclodeaminase
sequence by PCR. As we expected NRPSs to be encoded by the
identified cosmids, we used the degenerate NRPS primers to
amplify and sequence A-domain segments from these cosmids.
Sequence analysis and prediction of the A-domain substrate
Analysis of NRPS Domains—For the seven orfs with homol-
ogy to NRPS genes (Fig. 3A), the constituent domains were
assigned using the polyketide synthase/NRPS predictor (PKS
3
T. M. Binz, S. I. Maffioli, M. Sosio, S. Donadio, and R. M u¨ ller, unpublished data. Analysis Web site) and confirmed by manual inspection with
OCTOBER 22, 2010•VOLUME 285•NUMBER 43
JOURNAL OF BIOLOGICAL CHEMISTRY 32713

GE81112 Biosynthetic Gene Cluster
TABLE 1
Prediction of substrate specificity of GE81112 A-domains based on the specificity-conferring codes of A-domains
Orn, ornithine.
Position of the amino acid within the A-domain
Predicted
A-domain
Identity
amino acid
235
236
239
278
299
301
322
330
%
GetEA
Asp
Asp
Asp
Asp
Asp
Val
Ala
Ala
Ser
Ser
Gln
Tyr
Val
Ala
Ala
Tyr
Asn
Gly
Ser
Ile
Leu
Val
Ala
Gly
Gly
Ala
Ala
Gln
Leu
Glu
Glu
Glu
Val
Ile
Pro/Pip
Orn/Gln/Asp
Tyr/Trp
His
70
70
70
70
70
1
GetGA
2
3
GetGA
Val
Val
Val
GetJA
Thr
Thr
4
GetMA
Leu
His
5
TABLE 2
Predicted function of non-PKS/NRPS proteins present up- and downstream of the GE81112 biosynthetic gene clusters
No. of
Proposed function of
Identity/
Protein
Origin
Accession no.
amino acids
the homologous protein
similarity
%
Proteins encoded upstream of the GE81112 biosynthetic gene cluster
Orf1
Orf2
Orf3
Orf4
Orf5
335
502
Partitioning-binding protein
Hypothetical protein
Nocardia farcinica IFM 10152
S. scabiei 87.22
89/95
77/84
58/69
78/86
53/67
CBL93707.1
CBL93708.1
CBL93709.1
CBL93710.1
CBL93711.1
476
Hypothetical protein 5443
FtsK/SpoIIIE family protein
Serine/threonine protein kinase
S. scabiei 87.22
1203
S. scabiei 87.22
1238
Streptomyces avermitilis MA-4680
NRPS portion of the biosynthetic gene cluster
GetA
GetB
GetC
GetD
GetE
GetF
GetG
GetH
GetI
284
564
596
402
518
364
2404
538
346
820
435
585
616
288
Thioesterase
Streptomyces hygroscopicus ATCC 53653
Streptosporangium roseum DSM 43021
S. roseum DSM 43021
64/75
52/67
58/74
52/65
40/54
33/52
31/46
40/49
48/66
37/52
58/72
40/57
39/56
43/58
CBL93712.1
CBL93713.1
CBL93714.1
CBL93715.1
CBL93716.1
CBL93717.1
CBL93718.1
CBL93719.1
CBL93720.1
CBL93721.1
CBL93722.1
ABC multidrug transporter
ABC multidrug transporter
Cyclodeaminase TubZ protein
Syringopeptin synthetase
L-Proline 3-hydroxylase
Peptide synthetase protein
Peptide synthetase protein
Oxygenase
Angiococcus disciformis
Pseudomonas syringae
Ralstonia solanacearum GMI1000
R. solanacearum GMI1000
S. hygroscopicus ATCC 53653
Streptomyces rochei
GetJ
Bacitracin synthetase 3
GTP-binding protein
Halogenase
Bacillus licheniformis
GetK
GetL
GetM
GetN
S. scabiei 87.22
Microcystis aeruginosa
Bacitracin synthetase 3
B. licheniformis
CBL93723.1
CBL93724.1
Thioesterase
Myxococcus xanthus
Proteins downstream of the GE81112 biosynthetic gene cluster
Orf6
315
260
196
488
260
1194
818
285
424
486
3-Oxoacid-CoA transferase subunit B
Protocatechuate 3,4-dioxygenase ␣-subunit
Protocatechuate 3,4-dioxygenase, ␤-subunit
3-Carboxymuconate cycloisomerase
Deoxyribose-phosphate aldolase
Arthrofactin synthetase/syringopeptin synthetase
Putative NRPS
Streptomyces ghanaensis ATCC 14672
S. avermitilis MA-4680
84/90
70/81
76/83
71/78
73/79
43/54
35/47
88/92
44/60
44/59
CBL93725.1
CBL93726.1
CBL93727.1
CBL93728.1
CBL93729.1
CBL93730.1
CBL93731.1
CBL93732.1
CBL93733.1
CBL93734.1
Orf7
Orf8
Thermomonospora curvata DSM 43183
Streptomyces sviceus ATCC 29083
S. scabiei 87.22
Orf9
Orf10
Orf11
Orf12
Orf13
Orf14
Orf15
Bradyrhizobium sp. BTAi1
Streptomyces viridochromogenes DSM 40736
Streptomyces albus J1074
Translation-associated GTPase
Hypothetical protein
Streptomyces ambofaciens
Integrin-like protein
S. viridochromogenes DSM 40736
BLAST. To incorporate four amino acids into GE81112, the (Table 1 and supplemental Fig. S2) (26, 27). The first A-domain
synthetases were expected to contain a loading module fol- (GetEA ), encoded by getE as a discrete protein, was a candidate
1
lowed by three condensation modules with the standard C-A- for starter unit selection as its predicted specificity was for the
PCP arrangement. Furthermore, the simplest model predicts incorporation of proline/pipecolic acid (Table 1). The second
that the four amino acid precursors would be incorporated in a A-domain (GetGA ) showed homology to ornithine/gluta-
2
co-linear manner: pipecolic acid-ornithine/glutamine/glu- mine/asparagine-incorporating A-domains. The third A-do-
tamic acid-histidine-histidine. However, the get NRPS genes main (GetGA ) was predicted to be specific for the incorpora-
3
exhibit a non-co-linear arrangement that did not fit our tion of tyrosine/tryptophan and the two latter domains GetJA
4
expected model (Fig. 3A). Although the domain complement and GetMA for histidine (Table 1). These results correlated
5
necessary for the biosynthesis of a tetrapeptide could be iden- well with the prediction that an ornithine and two histidines are
tified, one extra A-PCP di-domain was present. Moreover, the incorporated into the GE81112 metabolites. However, it
NRPS modules show a highly split arrangement, as they remained unclear why two A-PCP di-domains (GetM and GetJ)
occurred as freestanding domains (GetE) or di-domain units are encoded by the cluster, as only one is required to give a full
(
GetH, GetJ, GetM) (Fig. 3A). Thus, from the domain assign- complement of four active modules. To check if any of the
ment alone, the overall order of subunits could not be dis- domains were inactive, the C- and PCP-domains were analyzed
cerned. To determine the substrate specificity of the A-do- as well. The get cluster encodes three C-domains that aligned
mains, the specificity-conferring code and eight conserved well with the C-domains from the rapamycin-, the gramicidin-,
motifs were identified for each A-domain, and bioinformatic and the calcium-dependent antibiotic biosynthetic clusters
analysis was employed to predict their substrate specificities (28–30). The seven conserved regions were identified in all of
3
2714 JOURNAL OF BIOLOGICAL CHEMISTRY
VOLUME 285•NUMBER 43•OCTOBER 22, 2010

GE81112 Biosynthetic Gene Cluster
the C-domains and the same analy-
sis was carried out for the PCP-
domains, revealing the signature
sequence and active serine residue,
in each case (supplemental Fig. S3).
These results demonstrate that,
in principle, all of the NRPS
domains are active. Furthermore
we annotated two discrete thio-
esterases (encoded by getA and
getN), both containing the con-
served motif GXSXG, present in
functional enzymes (16). BLAST
analysis (Table 2) revealed that the
protein GetA is more related to
type II thioesterases, whereas
GetN is more related to type I
thioesterases. This finding was
FIGURE 2. Organization of the GE81112 biosynthetic gene cluster in Streptomyces sp. L-49973. A, a sche-
matic representation of the GE81112 biosynthetic locus and flanking ORFs in Streptomyces sp. L-49973 is surprising, as GetN is a discrete
represented on two overlapping cosmids. Proposed functions for individual ORFs are summarized in Table 2.
protein (like a type II thioesterase)
B, shown is LTQ high resolution Orbitrap MS analysis of extracts of Streptomyces sp. L-49973 wild type and
Streptomyces sp. L-49973::KO1 mutant, showing a base peak chromatogram (BPC) of Streptomyces sp. L-49973 not integrated into an NRPS-like
wild type extract (m/z ϭ 100–2000) and extracted ion chromatograms (EIC) of GE81112 compound B (2) with
type I thioesterase typically found
ϩ
a molecular ion of the mass m/z ϭ 659.22953 [MϩH] from Streptomyces sp. L-49973 wild type and Strepto-
in bacterial systems.
myces sp. L-49973::KO1 mutant extracts. GE81112 production is abolished in the mutant.
FIGURE 3. A linear model of GE81112 biosynthesis. A, genetic and modular organization of the GE81112 biosynthetic gene cluster is shown. Black arrows
indicate NRPS genes, and white arrows non-NRPS genes. TEI, type I thioesterase. B, proposed biosynthesis of the fourth amino acid precursor is shown. Free
(
2S)-histidine is activated by the A-domain and loaded to the PCP of GetM. The PCP-bound histidine is chlorinated at position 6 by the halogenase GetL and
hydroxylated at the ␤-position by GetI, although one of these reactions may not occur on the GetM-bound amino acid. It is not clear if the halogenation or
hydroxylation occurs first, and it could be either way. The type II thioesterase GetA hydrolyzes the modified histidine to give the free amino acid. C, shown is the
proposed biosynthesis of GE81112 congener A. Lysine is converted to pipecolic acid by the cyclodeaminase GetD followed by hydroxylation catalyzed by GetF.
(
2S,3S)-Hydroxypipecolic acid is then activated by the A-domain GetE and loaded to the PCP of GetH. O-HPA, histidine, and 3-hydroxy-6-chlorohistidine are
activated and loaded by modules 2, 3, and 4 in the next steps, and the final tetrapeptide is released by the type I thioesterase (TEI) GetN.
OCTOBER 22, 2010•VOLUME 285•NUMBER 43
JOURNAL OF BIOLOGICAL CHEMISTRY 32715

GE81112 Biosynthetic Gene Cluster
Confirmation of the Role of the get Cluster by Gene
Inactivation—To verify the identity of the proposed GE81112
biosynthetic gene cluster, we inactivated getE, which encodes a
free-standing A-domain. For this, a knock-out construct was
designed by amplifying an internal fragment, which was then
cloned into the knock-out vector pKC1132 (10). Initial
attempts to transform Streptomyces sp. L-49973 with the
knock-out plasmid were unsuccessful, necessitating the devel-
opment of an adapted transformation method. Only by using a
10
7
larger ratio of E. coli cells (5 ϫ 10 ) to 5 ϫ 10 recipient cells
were we able to obtain several exconjugants containing the
knock-out vector, indicating that the number of donor cells is
crucial for the conjugation efficiency with this strain (31). The
resultant mutants were verified by PCR against the apramy-
cin resistance gene as well as an internal region from the
genomic DNA. The mutants were then cultivated in produc-
tion medium, and the extracts were analyzed for the pres-
ence of the GE81112 compounds by high resolution MS.
This analysis clearly showed that GE81112 production was
abolished in the mutants, confirming the role of the cloned
gene cluster (Fig. 2B).
Biochemical Analysis of Adenylation Domains—To obtain
experimental evidence for A-domain substrate specificity, we
expressed the five get A-domains as N-terminal His -tagged
6
proteins. DNA fragments coding for the adenylation domains
of getE (1 domain, GetEA , size 60.39 kDa), getG (2 domains,
1
GetGA and GetGA , sizes 61.50 and 63.13 kDa), getJ (1
2
3
domain, GetJA , size 60.0 kDa), and getM (1 domain, GetMA ,
4
5
size 63.40 kDa) were amplified from cosmids BI11 or BA23 and
cloned into pET28b (ϩ) vectors. The constructs were con-
firmed by sequencing and transformed into E. coli Rosetta BL21
(
DE3) pLysS/RARE. Cultivation was carried out at 16 °C.
Expression was induced with 0.2 mM isopropyl-␤-D-thiogalac-
topyranoside at A ϭ 0.8–1. All proteins could be obtained in
600
the soluble fraction and were used for the ATP-PP exchange
i
assay after purification.
The substrate specificity of the five purified adenylation
domains was evaluated using the established ATP-PP
i
exchange assay (32, 33). Briefly, each protein was incubated
with a panel of different amino acids, including the anticipated
substrate of each A-domain. As a control, each protein was
incubated in the absence of added amino acid. The results as
shown in Fig. 4, A–E, indicate that GetEA activated L-(2S)-Pip
1
(
100%), D-(2R)-Pip (82.9%). and L-(2S)-Pro (80.3%). The back-
ground controls were between 0.85 and 3.79%, confirming that
the measured activity reflected the true substrate preference of
the A-domain. GetGA activated L-(2S)-ornithine (100%) as
2
well as L-(2S)-Gln (62.7%) preferentially, whereas L-(2S)-Glu
and L-(2S)-Asp were activated to a minor extent. GetGA ,
3
GetJA , and GetMA all activated L-(2S)-His (100%) as well as
4
5
L-(2S)-Lys to a minor extent (between 41 and 53%). So a clear
preference for L-(2S)-His could be verified for all the three pro-
teins. Taken together, these results establish that the five pro- FIGURE 4. Relative substrate specificities of internal adenylation
domains from the GE81112 biosynthetic gene cluster. Internal adenyla-
teins exhibit the enzymatic activity of adenylation domains and
tion domains GeEA (A), GeGA (B), GeGA (C), GeJA (D), and GeMA (E) were
1
2
3
4
5
show preference for substrates consistent with the GE81112 investigated in terms of activity in the ATP-PP exchange reaction with differ-
i
ent amino acids and a control without amino acid. The highest activities were
set at 100%. The background was below 10%. The specificities of the different
domains coincide with the primary structures of the GE81112s.
structure.
Stereoconfiguration of Chiral Centers—Before this work, the
stereochemistry of the chiral centers in GE81112 was not
3
2716 JOURNAL OF BIOLOGICAL CHEMISTRY
VOLUME 285•NUMBER 43•OCTOBER 22, 2010

GE81112 Biosynthetic Gene Cluster
TABLE 3
GE81112 was hydrogenated to its dechloro analog, affording
1
13
H and C NMR chemical shift assignments of GE8112 congener A in 3-hydroxyhistidine upon acid hydrolysis. Comparison of NMR
D O at 298 K
2
and [␣] data with literature values were consistent with the
D
The assignments were made by analysis of COSY, TOCSY (two-dimensional total
correlation spectroscopy), HMQC (heteronuclear multiple quantum coherence),
and HMBC (heteronuclear multiple bond coherence) spectra. Numbering is
according to previous work (5).
(
2S,3S) configuration for the fourth amino acid (data not
shown), in agreement with the threo relative stereochemistry
suggested by the NMR experiments in D O.
Residuea
AA1
1
13C
2
Group
␦ s H
␦
These results together with the lack of epimerization
domains (or dual condensation/epimerization domains), thus,
strongly suggest that GE81112 is a tetrapeptide made of L-
amino acids (2S configuration) only and suggest the stereo-
chemistry of the other centers as shown for congener A in Fig. 1.
2
3
4
5
6
CH
CH
4.09
4.51
60.3
64.6
28.2
15.7
43.6
CH
1.82, 1.99
1.75, 1.99
3.02, 3.45
2
CH
2
2
CH
AA2
2
3
4
5
CH
4.56
50.6
33.7
65.7
68.1
DISCUSSION
CH
1.84
2
CH
3.85
Many antibiotics target the prokaryotic translational appara-
tus, but few selectively inhibit initiation. Protein translation in
prokaryotes is initiated by the binding of fMet-tRNA to the
ribosomal P-site. Recently, the GE81112 tetrapeptides were
shown to specifically inhibit this fMet-tRNA binding by block-
ing the P-site and, thus, represent a unique class of inhibitors
with a new mode of action (35). Identification and biochemical
characterization of the GE81112 biosynthetic gene cluster now
provide insights into the biosynthesis of this unique family of
CH
3.92, 4.06
2
AA3
2
3
CH
4.67
52.2
27
CH
2.98, 3.27
2
AA4
2
3
CH
CH
CH
4.50
59.3
67.8
118
5.20 (J ϭ 3 Hz)
5
Ј
6.97
a
Numbers in this column represent positions.
known. The chemical analyses described here have confirmed secondary metabolites and sets the stage for the generation of
the indication from the ATP-PP exchange assays that all amino new derivatives by genetic engineering. The identity of the
i
acids have the (2S)-configuration and provided preliminary evi- cloned gene cluster was confirmed by inactivation of getE,
dence about the other stereocenters.
which completely abolished GE81112 production. Although
NMR analyses of GE81112 dissolved in DMSO-d (5) do not many of the enzymes encoded by the get cluster are consistent
6
provide useful information about the relative stereochemistry with GE81112 biosynthesis, the NRPSs are present with a
due to generally broad signals. Improved resolution and shape highly split, non-co linear module arrangement (Fig. 3A).
of the signals were achieved using D O (as such and acidified Unusual features include two freestanding A-PCP di-domains
2
with TFA), allowing new assignments (Table 3). Under these (encoded by getJ and getM) and a stand-alone A-domain
conditions, the signal belonging to the ␤-hydrogen of chloro- (encoded by gene getE). Stand-alone A-domains have been
histidine (5.20 ppm) shows a vicinal coupling constant J of 3 Hz, identified in other nonlinear NRPS pathways, e.g. myxochelin
suggesting a threo relative stereochemistry according to litera- (36) and yersiniabactin (37), whereas free-standing A-PCP di-
ture data available for 3-hydroxy-His (19), whereas the opposite domains are found in the zorbamycin and syringomycin gene
stereochemistry is expected to have a J of 6 Hz (34). Analogous clusters (38, 39). According to bioinformatic analysis, all NRPS
indications about the 3-hydroxypipecolic acid residue were not domains are predicted to be functional (supplemental Fig. S2
achieved due to partially overlapping signals not allowing a sat- and S3). Furthermore, because the get cluster encodes five dis-
isfactory J resolution. Further experiments were, thus, neces- tinct A and PCP domains instead of the four predicted for a
sary to determine the relative and absolute stereochemistry of tetrapeptide, the order of subunits (and corresponding mod-
GE81112.
ules) could not be predicted from sequence alone. The estab-
The first three amino acids were investigated by chiral lished specificity for the five A-domains and analysis of the
GC-MS of acid-hydrolyzed congener A (data not shown). The functions predicted from the other enzymes encoded by the
first amino acid did not match (2R,3R)-3-hydroxypipecolic acid cluster allow us to draw a first model for GE81112 formation
or any of the (2S,3R) and (2R,3S) cis-diastereomers, easily (Fig. 3C).
obtained by catalytic hydrogenation of 3-hydroxy-picolinic
Accordingly, the biosynthesis of GE81112 starts with the for-
acid. We, thus, assume the first amino acid has the (2S,3S) ste- mation of (2S)-pipecolic acid from (2S)-lysine via the action of
reochemistry. For the second amino acid, protected L-allylgly- the putative cyclodeaminase GetD (Fig. 3C). Pipecolic acid is
cine was dihydroxylated to (2S)-2-amino-4,5-dihydroxy-penta- directly activated by the A-domain GetEA and then loaded
1
noic acid, yielding a mixture of the two diastereomers in the ␥ onto the adjacent PCP domain. Consistently, an ATP-PP
i
position. The same procedure, performed on racemic allylgly- exchange assay confirmed that GetEA preferentially recog-
1
cine, provided standards of the remaining two (2R)- diaste- nizes (2S)-pipecolic acid (Fig. 4A). However, the pipecolic acid
reomers. Based on the chiral analysis of the second amino acid, moiety in GE81112 is hydroxylated at the ␤-position. There is
the configuration of the ␣-carbon is assumed to be (S), and precedent for ␤-hydroxylation to occur at three different stag-
based on the published facial diastereoselection of dihydroxy- es; on the free amino acid (40), whereas the amino acid is teth-
lation (18), the ␥-stereocenter should be (S).
ered to the PCP (41), or on the mature peptide after thioesterase
The third amino acid matched (2S)-His by comparison with hydrolysis from the NRPS (42, 43). As GetEA1 could not be
commercial (S) and (R) standards. For the fourth amino acid, directly assayed with hydroxypipecolate due to the commercial
OCTOBER 22, 2010•VOLUME 285•NUMBER 43
JOURNAL OF BIOLOGICAL CHEMISTRY 32717

GE81112 Biosynthetic Gene Cluster
unavailability of this compound, we suggest that this domain tyrosine, but we could not establish whether GetGA also rec-
3
might also recognize hydroxypipecolic acid. In any case, the ognizes aminohistidine, as this amino acid is not commercially
presence of the 3-hydroxyl group on the pipecolate moiety is available. Thus, the nature and timing of histidine amination
essential for activity, as GE81112 derivatives lacking this group remains unclear.
are 3 orders of magnitude less active in the translation assay
A 3-hydroxy-6-chlorohistidine is the last amino acid to be
incorporated in GE81112. Halogenation is a common modifi-
4
than the parent compounds.
It was not obvious from which amino acid the second build- cation found in bioactive natural products (47–49), and the
ing block is derived. For GE81112 congeners A (1) and B (2), the timing of halogenation reactions has been shown to vary. For
second amino acid residue is 5-hydroxy-2-aminopentanoic example, in rebeccamycin, pre-assembly line chlorination of
acid (O-HPA), which is hydroxylated at position 4 and O-car- tryptophan occurs (50), whereas a PCP-bound threonine is
bamoylated at position 5, whereas congener B1 (3) contains a chlorinated for syringomycin (51). The putative halogenase
hydroxylated and carbamoylated ornithine residue (Fig. 1). We GetL is the likely candidate for histidine chlorination in
hypothesize that the A-domain GetGA is responsible for the GE81112. GetI shows homology to non-heme iron-dependent
2
incorporation of the second amino acid in GE81112. A recent oxygenases/hydroxylases and is, therefore, assumed to hydrox-
example in the biosynthesis of the nucleoside antibiotic ylate histidine at the ␤-position. Figs. 3, B and C, illustrate the
polyoxin shows that O-carbamoyl-polyhydroxypentanoic acid proposed mechanism for the chlorination/hydroxylation of
is generated from free (2S)-glutamate by stepwise reduction, histidine and its incorporation into GE81112. As there is evi-
O-carbamoylation, and hydroxylations (44). The complete dence from sequence analysis and in vitro experiments (supple-
building block is then attached to the nucleoside. Similarly, mental Fig. S2 and Fig. 4E) that the A-PCP domains of the
GE81112 congeners A and B could derive from activation of the seemingly superfluous GetM protein are active, this di-domain
O-HPA substrate (or of a precursor) by the A-domain GetGA
may play a role in the biosynthesis of the fourth amino acid
2
and loading on the corresponding PCP (Fig. 3C). However, we precursor. Indeed, there are a number of examples in which
could not identify candidate genes in the get cluster for O-HPA specialized A-PCP di-domains are essential for generating
biosynthesis, so they may be encoded elsewhere in the genome NRPS precursors (52–56). Here, we propose that the A-domain
(
45, 46). According to our hypothesis, the A-domain GetGA
GetMA activates (2S)-histidine, consistent with the results of
2
5
must show a broad substrate acceptance, activating O-HPA- the ATP-PP exchange assay, and tethers it to the PCP of GetM.
i
and ornithine-based amino acids to generate the different Hydroxylation and/or halogenation reactions then occur on the
GE81112 congeners. As O-HPA or any derivatives were not PCP-bound amino acid as catalyzed by GetI and GetL, respec-
commercially available, we tested the likely precursors of tively (Fig. 3B). The thioesterase GetA would then release the
O-HPA, (2S)-glutamic acid and (2S)-glutamine in addition to modified amino acid, as described for BarC from the barbamide
(
2S)-ornithine. Activation of all the three amino acids (orni- biosynthetic pathway (53). In the subsequent step, the free,
thine, Glu, and Gln) was observed (Fig. 4B) with a preference for modified histidine is activated by the A-domain GetJA and
4
(
2S)-ornithine. The reduced degree of activation of glutamic loaded onto the PCP of module 4 (Fig. 3C). This is supported by
acid and glutamine may indicate that modified versions of glu- the ATP-PP exchange assay showing that GetJA has specific-
i
4
tamic acid and glutamine are preferred, consistent with the ity for (2S)-histidine (Fig. 4D). Again, the relative timing of
hypothesis that O-HPA is formed before loading to the PCP. It chlorination and hydroxylation is unclear, as it is conceivable
should be noted that the major congeners of the GE81112 com- that one of these reactions may occur on the GetJ-tethered
5
plex produced by Streptomyces S. sp. L-49973 are A and B,
amino acid or peptide. In Fig. 3C, GetA is predicted to function
indicating that in vivo O-HPA or a precursor is the preferred as a type II thioesterase to release the modified amino acid from
substrate for the NRPS.
GetM. Alternatively, GetA could act as an aminoacyltrans-
The third amino acid incorporated into the GE81112 peptide ferase, shuttling the PCP-bound modified histidine from GetM
is either histidine (1) or aminohistidine (2 and 3). In silico anal- to GetJ. There are several examples of such PCP-to-PCP shut-
ysis of GetGA predicts specificity for tyrosine or tryptophan tling reactions (38, 55), and the aminoacyltransferases identi-
3
rather than histidine. This prediction may indicate a general fied to date have been assigned into two groups (38); one group,
preference of the A-domain for aromatic amino acid residues, comprising SyrC, CmaE, and ZmbVIId, contains a GXCXG
which might account for activation of both histidine and ami- motif at the active site, and these enzymes are predicted to act
nohistidine. However, it is not clear how the aminohistidine as acyltransferases, with the active-site cysteine shown to be
moiety is generated: “amination” might occur at the PCP- directly involved in aminoacyl transfer (52); the second group
bound histidine after release of the peptide from the NRPS, or includes the acyltransferases BarC (53) and CouN7 (54), with
alternatively, aminohistidine could derive from the cyclization an active site serine in the GXSXG motif, and are predicted to
of arginine by nucleophilic attack of the ␥-carbon. According to function as ordinary thioesterases. GetA and GetN both con-
the ATP-PP exchange assay, histidine is the preferred sub- tain active site serines, which suggest that they both function as
i
strate, and no activation was observed with arginine, trypto- normal thioesterases and not as aminoacyltransferases (supple-
phan, or tyrosine (Fig. 4C). These data suggest that aminohisti- mental Fig. S4). GetN is expected to catalyze the release of the
dine is not generated from PCP-bound arginine, tryptophan, or peptide from the assembly line (Fig. 3C).
The combination of genetic, biochemical, and chemical anal-
4
5
yses demonstrate that GE81112 is composed of L-amino acids
only. This result is consistent with the observation that it is a
S. I. Maffioli, unpublished results.
M. Sosio and S. I. Maffioli, unpublished results.
3
2718 JOURNAL OF BIOLOGICAL CHEMISTRY
VOLUME 285•NUMBER 43•OCTOBER 22, 2010

GE81112 Biosynthetic Gene Cluster
substrate of the oligopeptide permease in some bacterial spe- 22. He, M. (2006) J. Ind. Microbiol. Biotechnol. 33, 401–407
6
23. Ishikawa, J., and Hotta, K. (1999) FEMS Microbiol. Lett. 174, 251–253
cies. Many unusual features were found in the biosynthesis
24. Weissman, K. J., and M u¨ ller, R. (2008) Chembiochem 9, 826–848
25. M e´ ndez, C., and Salas, J. A. (1998) FEMS Microbiol. Lett. 158, 1–8
26. Stachelhaus, T., Mootz, H. D., and Marahiel, M. A. (1999) Chem. Biol. 6,
of this all-L-ribosome binding tetrapeptide. The NRPS modu-
lar architecture includes A-domains that incorporate unusual
amino acids, such as hydroxypipecolic acid, hydroxypentanoic
4
93–505
acid, and hydroxylchlorohistidine. Although many questions 27. Challis, G. L., Ravel, J., and Townsend, C. A. (2000) Chem. Biol. 7, 211–224
about its biosynthesis remain, the availability of the GE81112 28. Aparicio, J. F., Moln a´ r, I., Schwecke, T., K o¨ nig, A., Haydock, S. F., Khaw,
L. E., Staunton, J., and Leadlay, P. F. (1996) Gene 169, 9–16
cluster as well as tools for genetic manipulation now provide a
2
9. Turgay, K., Krause, M., and Marahiel, M. A. (1992) Mol. Microbiol. 6,
platform for attempts to decipher these issues and to generate
new GE81112 derivatives by genetic engineering.
5
29–546
3
0. Hojati, Z., Milne, C., Harvey, B., Gordon, L., Borg, M., Flett, F., Wilkinson,
B., Sidebottom, P. J., Rudd, B. A., Hayes, M. A., Smith, C. P., and Mickle-
field, J. (2002) Chem. Biol. 9, 1175–1187
Acknowledgments—We thank Dr. Shwan Rachid for helpful advice on
the construction of the cosmid library of Streptomyces sp. L-49973. We 31. Kopp, M., Irschik, H., Gross, F., Perlova, O., Sandmann, A., Gerth, K., and
thank Dr. Kira Weissman and Dr. Luke Simmons for critical reading
of the manuscript and helpful discussions. We thank M. Scharfe and
H. Bl o¨ cker for sequencing of the cosmids at the Helmholtz Centre for
Infection Research, Braunschweig. We are indebted to former col-
leagues at Vicuron Pharmaceuticals for fermentation and chemical
support.
M u¨ ller, R. (2004) J. Biotechnol. 107, 29–40
3
2. Mootz, H. D., and Marahiel, M. A. (1997) J. Bacteriol. 179, 6843–6850
3. Thomas, M. G., Burkart, M. D., and Walsh, C. T. (2002) Chem. Biol. 9,
3
1
71–184
3
3
3
3
3
3
4. Hecht, S. M., Rupprecht, K. M., and Jacobs, P. M. (1979) J. Am. Chem. Soc.
101, 3982–3983
5. Brandi, L., Fabbretti, A., La Teana, A., Abbondi, M., Losi, D., Donadio, S.,
and Gualerzi, C. O. (2006) Proc. Natl. Acad. Sci. U.S.A. 103, 39–44
6. Gaitatzis, N., Kunze, B., and M u¨ ller, R. (2001) Proc. Natl. Acad. Sci. U.S.A.
REFERENCES
98, 11136–11141
1
2
3
4
5
. Hopwood, D., Levy, S., Wenzel, R. P., Georgopapadakou, N., Baltz, R. H.,
7. Miller, D. A., Luo, L., Hillson, N., Keating, T. A., and Walsh, C. T. (2002)
Chem. Biol. 9, 333–344
Bhavnani, S., and Cox, E. (2007) Nat. Rev. Drug Discov. 6, 8–12
. Baker, D. D., Chu, M., Oza, U., and Rajgarhia, V. (2007) Nat. Prod. Rep. 24,
8. Galm, U., Wendt-Pienkowski, E., Wang, L., George, N. P., Oh, T. J., Yi, F.,
Tao, M., Coughlin, J. M., and Shen, B. (2009) Mol. Biosyst. 5, 77–90
9. Vaillancourt, F. H., Vosburg, D. A., and Walsh, C. T. (2006) Chembiochem
1
225–1244
. Walsh, C. (2003) Antibiotics: Actions, Origins, Resistance, American Soci-
ety for Microbiology, Washington, D. C.
7, 748–752
. Kohanski, M. A., Dwyer, D. J., and Collins, J. J. (2010) Nat. Rev. Microbiol.
4
4
0. Yin, X., and Zabriskie, T. M. (2004) Chembiochem 5, 1274–1277
1. Chen, H., Thomas, M. G., O’Connor, S. E., Hubbard, B. K., Burkart, M. D.,
and Walsh, C. T. (2001) Biochemistry 40, 11651–11659
8, 423–435
. Brandi, L., Lazzarini, A., Cavaletti, L., Abbondi, M., Corti, E., Ciciliato, I.,
Gastaldo, L., Marazzi, A., Feroggio, M., Fabbretti, A., Maio, A., Colombo, L.,
Donadio, S., Marinelli, F., Losi, D., Gualerzi, C. O., and Selva, E. (2006) Bio-
chemistry 45, 3692–3702
4
4
4
2. Buntin, K., Rachid, S., Scharfe, M., Bl o¨ cker, H., Weissman, K. J., and M u¨ l-
ler, R. (2008) Angew. Chem. Int. Ed. Engl. 47, 4595–4599
3. Samel, S. A., Marahiel, M. A., and Essen, L. O. (2008) Mol. Biosyst. 4,
6
7
8
9
. Finking, R., and Marahiel, M. A. (2004) Annu. Rev. Microbiol. 58, 453–488
. Hutchinson, C. R. (2003) Proc. Natl. Acad. Sci. U.S.A. 100, 3010–3012
. Wenzel, S. C., and M u¨ ller, R. (2005) Curr. Opin. Chem. Biol. 9, 447–458
. Wenzel, S. C., and M u¨ ller, R. (2007) Nat. Prod. Rep. 24, 1211–1224
3
87–393
4. Chen, W., Huang, T., He, X., Meng, Q., You, D., Bai, L., Li, J., Wu, M., Li,
R., Xie, Z., Zhou, H., Zhou, X., Tan, H., and Deng, Z. (2009) J. Biol. Chem.
284, 10627–10638
1
0. Kieser, T., Bibb, M., Buttner, M. J., Chater, K. F., and Hopwood, D. A.
4
5. Steffensky, M., M u¨ hlenweg, A., Wang, Z. X., Li, S. M., and Heide, L. (2000)
Antimicrob. Agents Chemother. 44, 1214–1222
(
2000) Practical Streptomyces Genetics pp. 562 and 566, The John Innes
Foundation, Norwich, England
4
6. Yu, T. W., Bai, L., Clade, D., Hoffmann, D., Toelzer, S., Trinh, K. Q., Xu, J.,
Moss, S. J., Leistner, E., and Floss, H. G. (2002) Proc. Natl. Acad. Sci. U.S.A.
1
1
1
1. Sambrook, J., and Russell, D. W. (2001) Molecular Cloning: A Laboratory
Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY
2. D u¨ rr, C., Schnell, H. J., Luzhetskyy, A., Murillo, R., Weber, M., Welzel, K.,
Vente, A., and Bechthold, A. (2006) Chem. Biol. 13, 365–377
99, 7968–7973
4
4
4
5
5
5
5
5
5
5
7. Neumann, C. S., Fujimori, D. G., and Walsh, C. T. (2008) Chem. Biol. 15,
9–109
8. Eust a´ quio, A. S., Gust, B., Luft, T., Li, S. M., Chater, K. F., and Heide, L.
9
3. Rachid, S., Krug, D., Kunze, B., Kochems, I., Scharfe, M., Zabriskie, T. M.,
Bl o¨ cker, H., and M u¨ ller, R. (2006) Chem. Biol. 13, 667–681
(
2003) Chem. Biol. 10, 279–288
1
1
4. Sandmann, A., Sasse, F., and M u¨ ller, R. (2004) Chem. Biol. 11, 1071–1079
5. Khaw, L. E., B o¨ hm, G. A., Metcalfe, S., Staunton, J., and Leadlay, P. F.
9. Yeh, E., Blasiak, L. C., Koglin, A., Drennan, C. L., and Walsh, C. T. (2007)
Biochemistry 46, 1284–1292
(
1998) J. Bacteriol. 180, 809–814
0. Yeh, E., Garneau, S., and Walsh, C. T. (2005) Proc. Natl. Acad. Sci. U.S.A.
1
6. Marahiel, M. A., Stachelhaus, T., and Mootz, H. D. (1997) Chem. Rev. 97,
651–2674
102, 3960–3965
2
1. Vaillancourt, F. H., Yin, J., and Walsh, C. T. (2005) Proc. Natl. Acad. Sci.
U.S.A. 102, 10111–10116
1
7. Silakowski, B., Schairer, H. U., Ehret, H., Kunze, B., Weinig, S., Nordsiek,
G., Brandt, P., Bl o¨ cker, H., H o¨ fle, G., Beyer, S., and M u¨ ller, R. (1999) J. Biol.
Chem. 274, 37391–37399
2. Strieter, E. R., Vaillancourt, F. H., and Walsh, C. T. (2007) Biochemistry 46,
7
549–7557
1
8. Girard, A., Greck, C., and Gen eˆ t, J. P. (1998) Tetrahedron Lett. 39,
3. Chang, Z., Flatt, P., Gerwick, W. H., Nguyen, V. A., Willis, C. L., and
Sherman, D. H. (2002) Gene 296, 235–247
4
259–4260
1
2
9. Dong, L., and Miller, M. J. (2002) J. Org. Chem. 67, 4759–4770
0. Rose, T. M., Schultz, E. R., Henikoff, J. G., Pietrokovski, S., McCallum,
C. M., and Henikoff, S. (1998) Nucleic Acids Res. 26, 1628–1635
1. Gatto, G. J., Jr., Boyne, M. T., 2nd, Kelleher, N. L., and Walsh, C. T. (2006)
J. Am. Chem. Soc. 128, 3838–3847
4. Garneau-Tsodikova, S., Stapon, A., Kahne, D., and Walsh, C. T. (2006)
Biochemistry 45, 8568–8578
5. Singh, G. M., Vaillancourt, F. H., Yin, J., and Walsh, C. T. (2007) Chem.
Biol. 14, 31–40
2
6. Pfeifer, V., Nicholson, G. J., Ries, J., Recktenwald, J., Schefer, A. B., Shawky,
R. M., Schr o¨ der, J., Wohlleben, W., and Pelzer, S. (2001) J. Biol. Chem. 276,
38370–38377
6
A. Maio and S. Donadio, unpublished results.
OCTOBER 22, 2010•VOLUME 285•NUMBER 43
JOURNAL OF BIOLOGICAL CHEMISTRY 32719