Thiostrepton biosynthesis: Prototype for a new family of bacteriocins

Article abstract of DOI:10.1021/ja807890a

Thiopeptide antibiotics are a group of highly modified peptide metabolites. The defining scaffold for the thiopeptides is a macrocycle containing a dehydropiperidine or pyridine ring, dehydrated amino acids, and multiple thiazole or oxazole rings. Some me

Full text of DOI:10.1021/ja807890a

Published on Web 03/05/2009
Thiostrepton Biosynthesis: Prototype for a New Family of
Bacteriocins
Wendy L. Kelly,* Lisa Pan, and Chaoxuan Li
School of Chemistry and Biochemistry and the Parker H. Petit Institute for Bioengineering and
Bioscience, Georgia Institute of Technology, Atlanta, Georgia 30332
Received October 6, 2008; E-mail: wendy.kelly@chemistry.gatech.edu
Abstract: Thiopeptide antibiotics are a group of highly modified peptide metabolites. The defining scaffold
for the thiopeptides is a macrocycle containing a dehydropiperidine or pyridine ring, dehydrated amino
acids, and multiple thiazole or oxazole rings. Some members of the thiopeptides, such as thiostrepton,
also contain either a quinaldic acid or indolic acid substituent derived from tryptophan. Although the amino
acid precursors of these metabolites are well-established, the biogenesis of these complex peptides has
remained elusive. Whole-genome scanning of Streptomyces laurentii permitted identification of a thiostrepton
prepeptide, TsrA, and involvement of TsrA in thiostrepton biosynthesis was confirmed by mutagenesis. A
gene cluster responsible for thiostrepton biosynthesis is reported, and the encoded gene products are
discussed. The disruption of a gene encoding an amidotransferase, tsrT, led to the loss of thiostrepton
production and the detection of a new metabolite, contributing further support to the identification of the tsr
cluster. The tsr locus also appears to possess the gene products needed to convert tryptophan to the
quinaldic acid moiety, and an aminotransferase was found to catalyze an early step in this pathway. This
work establishes that the thiopeptides are a type of bacteriocin, a family of genetically encoded antimicrobial
peptides, and are subjected to extensive posttranslational modification during maturation of the prepeptide.
Introduction
resistant Staphylococcus aureus and vancomycin-resistant en-
terococci, for which the available chemotherapeutic options are
Thiopeptides are widely distributed metabolites isolated from
Gram-positive bacteria of both terrestrial and marine origin.1-3
Thiostrepton A 1 (Figure 1), considered the prototype of this
family, was first isolated from Streptomyces azureus ATCC
14921 in the 1950s and later from Streptomyces laurentii ATCC
31255 (S. laurentii) in 1977.1,4,5 In both the solution and crystal
structures of thiostrepton 1, the molecule adopts a globular
conformation, with the two large ring systems of the peptide
backbone folded upon one another.6,7 The highly modified
thiopeptides are characterized by a macrocycle possessing a
central pyridine, hydroxypyridine, or dehydropiperidine (Figure
1).1 This core scaffold typically presents several dehydrated
amino acids and multiple thiazole or oxazole rings. Some
thiopeptides, such as thiostrepton 1 and nosiheptide 2, are further
modified by a quinaldic acid or indolic acid appendage,
respectively.6,8
severely limited.9,10 The thiopeptides exert their antibiotic effect
by inhibiting elongation during prokaryotic translation by one
of two general mechanisms. One group of thiopeptides, includ-
ing thiostrepton 1 and nosiheptide 2, binds directly to the large
ribosomal subunit in the region that interacts with elongation
factor G and interferes with translocation along the mRNA
transcript.11-13 The second mechanism attributed to certain
thiopeptides, such as GE2270A 3, is to bind elongation factor
Tu (EF-Tu). Upon binding of GE2270A 3 to EF-Tu, formation
of the EF-Tu ·GTP·aminoacyl-tRNA complex is prevented, and
efficient delivery of an incoming aminoacyl-tRNA to the
ribosome is impaired.14,15 The development of the thiopeptides
into clinically useful antibacterial agents has been limited by
the lipophilicity of these metabolites. There is therefore interest
(8) Benazet, F.; Cartier, M.; Florent, J.; Godard, C.; Jung, G.; Lunel, J.;
Mancy, D.; Pascal, C.; Renaut, J.; Tarridec, P.; Theilleux, J.; Tissier,
R.; Dubost, M.; Ninet, L. Experientia 1980, 36, 414–416.
(9) Pucci, M. J.; Bronson, J. J.; Barrett, J. F.; DenBleyker, K. L.; Discotto,
L. F.; Fung-Tomc, J. C.; Ueda, Y. Antimicrob. Agents Chemother.
2004, 48, 3697–3701.
Thiostrepton 1 and the other thiopeptides are potent antibacte-
rial agents against Gram-positive pathogens such as methicillin-
(1) Bagley, M. C.; Dale, J. W.; Merritt, E. A.; Xiong, X. Chem. ReV.
2005, 105, 685–714.
(10) Zhang, C.; Occi, J.; Masurekar, P.; Barrett, J. F.; Zink, D. L.; Smith,
S.; Onishi, R.; Ha, S.; Salazar, O.; Genilloud, O.; Basilio, A.; Vicente,
F.; Gill, C.; Hickey, E. J.; Dorso, K.; Motyl, M.; Singh, S. B. J. Am.
Chem. Soc. 2008, 130, 12102–12110.
(2) Shoji, J.; Hinoo, H.; Wakisaka, Y.; Koizumi, K.; Mayama, M. J.
Antibiot. 1976, 29, 366–374.
(3) Suzumura, K.; Yokoi, T.; Funatsu, M.; Nagai, K.; Tanaka, K.; Zhang,
H.; Suzuki, K. J. Antibiot. 2003, 56, 129–134.
(4) Trejo, W. H.; Dean, L. D.; Pluscec, J.; Meyers, E.; Brown, W. E. J.
Antibiot. 1977, 30, 639–643.
(11) Lentzen, G.; Klinck, R.; Matassova, N.; Aboul-ela, F.; Murchie, A. I.
Chem Biol. 2003, 10, 769–778.
(12) Porse, B. T.; Cundliffe, E.; Garrett, R. A. J. Mol. Biol. 1999, 287,
33–45.
(5) Vandeputte, J.; Dutcher, J. D. Antibiot. Annu. 1955, 3, 560–561.
(6) Anderson, B.; Hodgkin, D. C.; Viswamitra, M. A. Nature 1970, 225,
233–235.
(13) Porse, B. T.; Leviev, I.; Mankin, A. S.; Garrett, R. A. J. Mol. Biol.
1998, 276, 391–404.
(7) Hensens, O. D.; Albers-Schonberg, G.; Anderson, B. F. J. Antibiot.
1983, 36, 799–813.
(14) Anborgh, P. H.; Parmeggiani, A. EMBO J. 1991, 10, 779–784.
(15) Heffron, S. E.; Jurnak, F. Biochemistry 2000, 39, 37–45.
9
10.1021/ja807890a CCC: $40.75
2009 American Chemical Society
J. AM. CHEM. SOC. 2009, 131, 4327–4334 4327

A R T I C L E S
Kelly et al.
Figure 1. Examples of thiopeptide antibiotics.
strepton was purchased from Sigma-Aldrich. 2-Methyltryptophan
was prepared as described previously.27 High-performance liquid
chromatography (HPLC) was performed on a Beckman Coulter
System Gold equipped with a diode-array detector. Mass spectros-
copy was performed by the Georgia Institute of Technology
Bioanalytical Mass Spectrometry Facility with a Micromass Quattro
LC Mass Spectrometer.
in the generation of thiopeptide analogues of improved water
solubility by the means of biosynthetic engineering, synthesis,
or semisynthesis.9,16
The amino acid origins of the thiopeptides are well-
established,17-22 but an answer to the question of whether the
peptide backbone arises from a nonribosomal peptide synthetase
or by the posttranslational modification of a ribosomally
synthesized peptide has remained elusive. Most microbial
metabolites arise from loci on the chromosome possessing genes
encoding both biosynthetic enzymes and those for self-
resistance. This is not the case for the thiopeptides.23,24 Since
this limits the options available for biosynthetic gene cluster
identification, we adopted a genome-scanning strategy to
pinpoint the genes required for thiostrepton biosynthesis in S.
laurentii. In this work, we establish that thiopeptides do, in fact,
arise from a genetically encoded peptide and constitute a newly
identified class of bacteriocin, a ribosomally synthesized
antimicrobial peptide.
Bacterial Strains, Plasmids, and Media. Tables of strains and
plasmids used in this work are provided in the Supporting
Information. S. laurentii was obtained from American Type Culture
Collection (ATCC). pGM160 was kindly provided by Dr. Muth at
University of Tu¨bingen, Germany.28 All Escherichia coli (E. coli)
strains were grown in Luria-Bertani liquid medium or on solid
medium supplemented with the appropriate antibiotic(s). Kanamycin
(50 µg/mL), apramycin (50 µg/mL), nalidixic acid (25 µg/mL), and
chloramphenicol (30 µg/mL) were used for the selective growth
of E. coli or S. laurentii. pGM160K was constructed from pGM160,
replacing the thiostrepton resistance gene with that for kanamycin
resistance. ISP-3 agar medium was used for the growth and
sporulation of S. laurentii. MS agar was prepared as described
previously and used for intergeneric conjugation.4 The seed and
fermentation media used for thiostrepton production by S. laurentii
were the same as described previously.29 Unless specified, restric-
tion enzymes, DNA ligase, and other materials for recombinant
DNA procedures were purchased from standard commercial sources
and used as provided.
Growth of S. laurentii and Production of Thiostrepton.
Growth of S. laurentii and production of thiostrepton were
performed as described previously with only minor modifications.29
Briefly, the fermentation was carried out in a two-step process. First,
100 mL of seed medium in a 500-mL flask was inoculated with
0.1 mL of a 24 h S. laurentii culture in tryptic soy broth. After
48 h at 28 °C and 200 rpm, 5 mL of this seed culture was used to
inoculate 100 mL of fermentation medium in a 500-mL flask. The
resulting fermentation culture was incubated at 28 °C and 200 rpm
for 5 days. The whole culture was extracted twice with an equal
volume of chloroform. The chloroform layers were pooled together
and solvent removed in vacuo. The solid residue was dissolved in
1 mL of chloroform for HPLC and HPLC-MS analysis.
Experimental Section
General. All chemicals and solvents were reagent grade,
purchased from VWR or Sigma-Aldrich, and used without further
purification unless otherwise indicated. Pyrosequencing of the S.
laurentii ATCC 31255 genome and individual sequencing reactions
were performed by Eurofins MWG Operon (Huntsville, AL), and
sequencing of the shotgun fosmid libraries was performed at
Functional Biosciences, Inc. (Madison, WI). Protein-coding regions
of DNA sequences were predicted using FramePlot 4.0Beta (http://
nocardia.nih.go.jp/fp4/).25 Protein sequences were analyzed by
BLAST (Basic Local Alignment Search Tool).26 Authentic thio-
(16) Hughes, R. A.; Moody, C. J. Angew. Chem., Int. Ed. 2007, 46, 7930–
7954.
(17) De Pietro, M. T.; Marazzi, A.; Sosio, M.; Donadio, S.; Lancini, G. J.
Antibiot. 2001, 54, 1066–1071.
(18) Fate, G. D.; Benner, C. P.; Grode, S. H.; Gilbertson, T. J. J. Am. Chem.
Soc. 1996, 118, 11363–11368.
(19) Favret, M. E.; Paschal, J. W.; Elzey, T. K.; Boeck, L. D. J. Antibiot.
1992, 45, 1499–1511.
HPLC was performed using a Phenomenex Jupiter Proteo C18
column (4 µm, 250 mm × 4.60 mm). The column was developed
with a linear gradient of 0-100% of acetonitrile in water over 30
min at 1 mL/min, monitoring absorbance at 254 nm. Thiostrepton
eluted with a tR of about 22.5 min. HPLC-MS was performed with
a Phenomenex Syngeri RP column (250 mm × 2 mm) and
developed with 20% Buffer B in Buffer A for 8 min followed by
(20) Lau, R. C. M.; Rinehart, K. L., Jr. J. Am. Chem. Soc. 1995, 117, 7606–
7610.
(21) Mocek, U.; Knaggs, A. R.; Tsuchiya, R.; Nguyen, T.; Beale, J. M.;
Floss, H. G. J. Am. Chem. Soc. 1993, 115, 7557–7568.
(22) Mocek, U.; Zeng, Z.; O’Hagan, D.; Zhou, P.; Fan, L.-D. G.; Beale,
J. M.; Floss, H. G. J. Am. Chem. Soc. 1993, 115, 7992–8001.
(23) Li, Y.; Dosch, D. C.; Woodman, R. H.; Floss, H. G.; Strohl, W. R.
J. Ind. Microbiol. 1991, 8, 1–12.
(24) Smith, T. M.; Jiang, Y. F.; Shipley, P.; Floss, H. G. Gene 1995, 164,
137–142.
(27) Goss, R. J.; Newill, P. L. Chem. Commun. 2006, 4924-4925.
(28) Muth, G.; Farr, M.; Hartmann, V.; Wohlleben, W. Plasmid 1995, 33,
113–126.
(29) Priestley, N. D.; Smith, T. M.; Shipley, P. R.; Floss, H. G. Bioorg.
Med. Chem. 1996, 4, 1135–1147.
(25) Ishikawa, J.; Hotta, K. FEMS Microbiol. Lett. 1999, 174, 251–253.
(26) Altschul, S. F.; Madden, T. L.; Schaffer, A. A.; Zhang, J.; Zhang, Z.;
Miller, W.; Lipman, D. J. Nucleic Acids Res. 1997, 25, 3389–3402.
9
4328 J. AM. CHEM. SOC. VOL. 131, NO. 12, 2009

Thiostrepton Biosynthesis
A R T I C L E S
a gradient from 20-100% Buffer B over 35 min (Buffer A, 5%
acetonitrile and 0.1% formic acid; Buffer B, 95% acetonitrile and
0.1% formic acid) at 0.25 mL/min. Under these conditions,
thiostrepton elutes at a tR of about 28.4 min providing ions at m/z
of 1664.4 [M + H]+and m/z of 832.9 [M + 2H]2+. The predominant
ion for thiostrepton was the [M + 2H]2+, while the [M + H]+ was
only a minor species.
S4). A 2.3 kb PCR product was expected for wild-type tsrT versus
a 1.8 kb PCR product for the tsrT mutant. The PCR products were
cloned into pSC-B-amp/kan and confirmed by sequence analysis.
Expression and Purification of TsrV. The gene encoding TsrV
was amplified by PCR using primers TSRV1-F and TSRV1-R. The
product was digested with NdeI and XhoI, ligated into pET28b(+)
to provide pTSRV1, and confirmed by sequence analysis. E. coli
BL21(DE3) containing pTSRV1 was incubated at 37 °C overnight
in 2 × 50 mL Luria-Bertani medium supplemented with 50 µg/
mL kanamycin. The following morning, 6 × 10 mL of the overnight
culture was used to inoculate 6 × 1 L of Luria-Bertani medium
supplemented with 50 µg/mL kanamycin, and the resulting cultures
were grown at 37 °C until OD600 ) 0.4. At this time, the temperature
was reduced to 15 °C and protein expression induced with the
addition of 0.04 mM isopropyl-ꢀ-D-thiogalactopyranoside (IPTG).
Cultures were incubated for an additional 24 h. Harvested cells
were resuspended in 40 mL of lysis buffer [20 mM Tris (pH 8.0),
300 mM NaCl, 2 mM imidazole, 10% glycerol, 1 mg/mL lysozyme,
1 mM PMSF, 200 µM pyridoxal 5′-phosphate (PLP)]. The cells
were disrupted by sonication (10 10-s pulses with a 30 s pause).
The lysate was clarified by centrifugation (at 18 459g, 4 °C, 30
min). The cell-free extract was incubated with 2 mL of Ni-NTA
resin (Qiagen) for 1 h. The slurry was loaded onto a column, and
the resin was washed with 20 mL lysis buffer and then 40 mL
wash buffer [20 mM Tris (pH 8.0), 300 mM NaCl, 20 mM
imidazole, 10% glycerol, 20 µM PLP]. Protein was eluted with 12
mL of elution buffer [wash buffer containing 300 mM imidazole],
collecting 3 mL fractions. Fractions containing the protein were
pooled together and dialyzed against 2 × 1 L storage buffer [20
mM Tris (pH 8.0), 50 mM NaCl, 10% glycerol, 20 µM PLP, and
1 mM DTT]. The protein was concentrated with a Millipore-0.5
centrifugal filter. Protein concentration was determined by the
method of Bradford, using bovine serum albumin as a standard.32
The protein was flash-frozen in liquid nitrogen and stored at -86
°C.
Construction of S. laurentii Fosmid Library and Screening
for Selected Sequences. A genomic fosmid library of S. laurentii
was constructed according to the instructions of the Epicenter Copy
Control Fosmid Library Production Kit (Madison, WI). S. laurentii
genomic DNA was prepared according to protocol.30 Clones were
isolated and stored according to kit instructions. The S. laurentii
fosmid library was screened by PCR for the desired sequence. Four
rows from each 96-well plate of the fosmid library (48 clones) were
combined and the fosmid pool isolated. The two primer pairs used
in the screening were TSRA3-F/TSRA3-R and TSRA4-F/TSRA4-
R. Individual rows (containing a pool of 12 clones) from each plate
and individual clones were identified by iterations of PCR. The
PCR products were cloned into pCR-Blunt II-TOPO vector and
confirmed by sequence analysis.
Inactivation of tsrA. Inactivation of tsrA was accomplished by
PCR-targeted gene replacement31 using pGM160K. The tsrA gene
and flanking regions were amplified as a 1.8 kb fragment from
genomic DNA of S. laurentii using the primers TSRA1-F and
TSRA1-R. The resulting PCR product was cloned using the Zero
Blunt TOPO Cloning Kit, yielding pJP11 and was confirmed by
DNA sequencing. The plasmid pJP11 was digested with HindIII,
and the resulting 1.8 kb fragment ligated into pGM160K, yielding
pCL60. A 1.4 kb apramycin resistance cassette containing the
aac(3)IV resistance gene and oriT was amplified from pIJ773 using
the TSRA2-F and TSRA2-R. The resulting PCR product was used
for in-frame replacement of the tsrA gene in pJP60 by λ RED-
mediated recombination,31 generating pCL41.
Following transformation into E. coli ET12567/pUZ8002, pCL41
was introduced into S. laurentii by conjugation.31 A colony resistant
to apramycin and sensitive to kanamycin was selected following
homologous recombination between pCL41 and chromosomal DNA
and designated as tsrA-. The allelic replacement of tsrA by the
apramycin-resistance gene and oriT in tsrA- was confirmed by PCR
(Figure S2) with the primers TSRA3-F and TSRA3-R. A 1.1 kb
PCR product was expected for wild-type tsrA versus a 2.4 kb PCR
product for tsrA mutant. The PCR products were cloned into pSC-
B-amp/kan and confirmed by sequence analysis.
Inactivation of tsrT. TsrT was inactivated by a strategy similar
to that used for the inactivation of tsrA. A 1.4 kb apramycin
resistance cassette containing the aac(3)IV resistance gene and oriT
was amplified from pIJ773 using the primers TSRT1-F and TSRT1-
R. The resulting PCR product was used for in-frame replacement
of tsrT in JA8H9, generating pTSRT1. The apramycin resistance
cassette and its flanking region was amplified as a 3.5 kb fragment
from pTSRT1 using the primers TSRT2-F and TSRT2-R. The
resulting PCR product was cloned using the Zero Blunt TOPO
Cloning Kit, generating pTSRT2 and confirmed by DNA sequence
analysis. Digestion with HindIII, provided a 3.5 kb fragment, which
was then ligated into pGM160K to yield pTSRT3.
TsrV-Dependent Aminotransferase Activity Assays. Reaction
mixtures (100 µL) for detection of 2-methyl-indolepyruvate forma-
tion were prepared using 1 mM 2-methyltryptophan, 1 mM indole-
3-pyruvic acid, 100 µM PLP, and 1.5 µM TsrV in 100 mM KH2PO4
(pH 7.8). The reactions were incubated at 24 °C for 30 min. After
30 min, the reaction mixtures were frozen in liquid nitrogen and
stored at -80 °C, until ready for use. Samples were analyzed by
HPLC-MS using a Phenomenex Gemini C18 column (150 mm ×
2 mm). The column was developed at 0.2 mL/min by 100% Buffer
A for 5 min, then a gradient of 0-100% Buffer B over 40 min,
followed by 100% Buffer B for 5 min (Buffer A, 5% acetonitrile
and 0.1% formic acid in water; Buffer B, 95% acetonitrile and 0.1%
formic acid in water).
Nucleotide Sequence Accession Numbers. The sequences
reported here have been deposited into the GenBank database under
the accession number FJ652572.
Results and Discussion
Cloning and Sequencing of the Thiostrepton Biosynthetic
Gene Cluster in S. laurentii. Whole genome scanning was
employed to locate the genes required for 1 biogenesis. The S.
laurentii genome was partially sequenced to 7.26 Mb on over
3600 contiguous fragments. Sequence analysis permitted the
identification of gene products with weak similarity to lantibiotic
dehydratases. More interestingly, a gene was identified that
encoded a bacteriocin prepeptide, TsrA. Following the leader
peptide of TsrA, the 17 amino acid sequence at the C-terminal
end is identical to that predicted for a prepeptide of thiostrepton
(Figure 2).
Following transformation into E.coli ET12567/pUZ8002, pTSRT3
was introduced into S. laurentii by conjugation. A colony resistant
to apramycin and sensitive to kanamycin was selected following
homologous recombination between pTSRT3 and chromosomal
DNA and designated S. laurentii LP3tsrT-. The allelic replacement
of tsrT by the apramycin resistance gene and oriT in tsrT- was
confirmed by PCR with the primers TSRT3-F and TSRT3-R (Figure
(30) Kieser, T.; Bibb, M. J.; Buttner, M. J.; Chater, K. F.; Hopwood, D. A.
Practical Streptomyces Genetics; John Innes Foundation: Norwich,
UK, 2000.
A PCR screen of a S. laurentii genomic fosmid library
established the colocalization of tsrA and one of the genes to a
(31) Gust, B.; Challis, G. L.; Fowler, K.; Kieser, T.; Chater, K. F. Proc.
Natl. Acad. Sci. U.S.A. 2003, 100, 1541–1546.
(32) Bradford, M. M. Anal. Biochem. 1976, 72, 248–254.
9
J. AM. CHEM. SOC. VOL. 131, NO. 12, 2009 4329

A R T I C L E S
Kelly et al.
the peptide backbone. Microcin B17, the cyanobactins (e.g.,
patellamides A and C), and goadsporin are all examples of
genetically encoded peptides for which multiple thiazole rings
result from posttranslational modification.38-41 The thiazole and
oxazole rings in the E. coli metabolite microcin B17 result from
the action of the McbBCD complex.39 In goadsporin, these
heterocycles are thought to result from the combined action of
GodD and GodE.40,42 Likewise, PatD and PatG are proposed
to produce the aromatic heterocycles in patellamides A and C.41
GodD and PatD are presumed to be functional analogues of
the cyclodehydration catalyst McbB and the docking protein
McbD, fused into a single polypeptide chain.40-42 Dehydrogen-
ation to the aromatic heterocycle is mediated by the FMN-
dependent McbC for microcin B17, and likely GodE and PatG
in the case of goadsporin and the patellamides, respectively.39-41,43
It therefore seems likely that TsrH (36% similarity to GodD) is
the cyclodehydratase for the thiostrepton system and TsrF,
predicted to bind FMN, then oxidizes the dehydroheterocycle
to the aromatic species.40,42,44
The D-thiazoline at C9 is unique to dehydropiperidine
thiopeptides, and an epimerization of C9 must occur during TsrA
maturation.1,22 One possible explanation for this stereochemistry
is that heterocyclization and oxidation of all the cysteine residues
precedes introduction of the D-configuration at C9. In this
scenario, the C9 D-thiazoline is introduced following reduction
of the thiazole. This route, however, has been excluded. Both
deuterium atoms from L-[3-13C,2H]-serine remain in the C9
thiazoline of 1, suggesting that epimerization likely occurs prior
to cyclodehydration.22 Furthermore, there is precedent for the
posttranslational epimerization of an amino acid side chain in
both animal peptides and lantibiotics.45,46 Typically, only one
dehydratase is needed for the dehydration of serine and threonine
residues during lantibiotic maturation.45 If this is true for 1 as
well, then either TsrL or TsrC, similar to enzymes that abstract
a peptide’s R-protons, could operate as an epimerase rather than
a dehydratase. When introduced prior to cyclodehydration, the
C9 D configuration could prohibit subsequent dehydrogenation
Figure 2. Thiostrepton and TsrA.
(33) Frenzel, T.; Zhou, P.; Floss, H. G. Arch. Biochem. Biophys. 1990,
278, 35–40.
(34) Smith, T. M.; Priestley, N. D.; Knaggs, A. R.; Nguyen, T.; Floss,
H. G. J. Chem. Soc. Chem. Comm. 1993, 1612–1614.
(35) Zhou, P.; O’Hagan, D.; Mocek, U.; Zeng, Z.; Yuen, L.-D.; Frenzel,
T.; Unkefer, C. J.; Beale, J. M.; Floss, H. G. J. Am. Chem. Soc. 1989,
111, 7274–7276.
Figure 3. HPLC analysis of (a) thiostrepton and extracts from the cultures
of (b) wild-type S. laurentii, (c) S. laurentii tsrA-, and (d) S. laurentii
LP3tsrT-.
(36) Koponen, O.; Tolonen, M.; Qiao, M.; Wahlstrom, G.; Helin, J.; Saris,
P. E. Microbiology 2002, 148, 3561–3568.
lantibiotic dehydratase, suggestive of a biosynthetic operon for
a bacteriocin-containing dehydroalanine or dehydrobutyrine
residues. To establish the requirement of the prepeptide for
formation of 1, we constructed a mutant replacing tsrA with an
apramycin resistance cassette. The S. laurentii tsrA mutant
abrogated 1 production (Figure 3), confirming its involvement
in thiostrepton biosynthesis. The region surrounding tsrA was
then sequenced to nearly 30 kb from two overlapping fosmids,
JA8H9 and JA11E10, and found to contain 21 potential open
reading frames (Figure 4 and Table 1).
(37) Meyer, C.; Bierbaum, G.; Heidrich, C.; Reis, M.; Suling, J.; Iglesias-
Wind, M. I.; Kempter, C.; Molitor, E.; Sahl, H. G. Eur. J. Biochem.
1995, 232, 478–489.
(38) Donia, M. S.; Ravel, J.; Schmidt, E. W. Nat. Chem. Biol. 2008, 4,
341–343.
(39) Li, Y. M.; Milne, J. C.; Madison, L. L.; Kolter, R.; Walsh, C. T. Science
1996, 274, 1188–1193.
(40) Onaka, H.; Nakaho, M.; Hayashi, K.; Igarashi, Y.; Furumai, T.
Microbiology 2005, 151, 3923–3933.
(41) Schmidt, E. W.; Nelson, J. T.; Rasko, D. A.; Sudek, S.; Eisen, J. A.;
Haygood, M. G.; Ravel, J. Proc. Natl. Acad. Sci. U.S.A. 2005, 102,
7315–7320.
Posttranslational Modifications of TsrA. The proteins encoded
in this cluster are consistent with earlier feeding studies22,29,33-35
and the expectation that dehydration and heterocyclization occur
during TsrA maturation. Two proteins encoded in the tsr locus,
TsrC and TsrL, were identified having similarity to the LanB
lantibiotic dehydratases, and either one could impart the
dehydroalanine and dehydrobutyrine residues present in 1.36,37
Each cysteine residue in 1 is subjected to cyclodehydration upon
(42) Lee, S. W.; Mitchell, D. A.; Markley, A. L.; Hensler, M. E.; Gonzalez,
D.; Wohlrab, A.; Dorrestein, P. C.; Nizet, V.; Dixon, J. E. Proc. Natl.
Acad. Sci. U.S.A. 2008, 105, 5879–5884.
(43) Milne, J. C.; Roy, R. S.; Eliot, A. C.; Kelleher, N. L.; Wokhlu, A.;
Nickels, B.; Walsh, C. T. Biochemistry 1999, 38, 4768–4781.
(44) Milne, J. C.; Eliot, A. C.; Kelleher, N. L.; Walsh, C. T. Biochemistry
1998, 37, 13250–13261.
(45) Chatterjee, C.; Paul, M.; Xie, L.; van der Donk, W. A. Chem. ReV.
2005, 105, 633–684.
(46) Kreil, G. Annu. ReV. Biochem. 1997, 66, 337–345.
9
4330 J. AM. CHEM. SOC. VOL. 131, NO. 12, 2009

Thiostrepton Biosynthesis
A R T I C L E S
Figure 4. Organization of the tsr cluster.
Table 1. Deduced Functions of the Open Reading Frames in the Thiostrepton Biosynthetic Gene Cluster
gene
sizea
protein homologue and originb
identity/similarity %
proposed function
tsrV
362
putative aminotransferase (YP_871797.1); Acidothermus cellulolyticus
11B
47/62
aminotransferase
tsrU
tsrT
276
618
putative lipase (NP_631032.1); Streptomyces coelicolor A3(2)
hypothetical protein AsnH (YP_001615582.1), Sorangium cellulosum
‘So ce 56′
50/62
57/69
unknown
amidotransferase
tsrS
tsrQ
135
409
ester cyclase-like (ZP_01388898.1); Geobacter sp. FRC-32
acyl-CoA dehydrogenase, short-chain specific (YP_076186.1);
Symbiobacterium thermophilum IAM 14863
putative methyltransferase (YP_001535414.1) Salinispora arenicola
CNS-205
hypothetical protein MchlDRAFT_5537 (ZP_02058450.1);
Methylobacterium chloromethanicum CM4
putative lantibiotic precursor peptide (NP_834755.1); Bacillus cereus
ATCC 14579
36/56
34/50
unknown
oxygenase
tsrP
tsrO
tsrA
311
183
58
40/56
29/45
42/61
putative methyltransferase
unknown
thiostrepton precursor peptide
tsrB
tsrC
263
933
putative hydrolase (YP_289599.1); Thermobifida fusca YX
lanthionine biosynthesis protein (NP_834752.1); Bacillus cereus ATCC
14579
29/39
21/40
unknown
lantibiotic dehydratase
tsrD
346
lantibiotic biosynthesis protein (NP_834751.1); Bacillus cereus ATCC
14579
24/47
dihydropyridine synthase
tsrE
tsrF
368
589
unknown (ABC02780.1); Actinomadura melliaura
conserved hypothetical protein (ZP_02912150.1); Geobacillus sp.
WCH70
25/41
29/42
unknown
thiazoline dehydrogenase
tsrG
612
hypothetical protein SGR_4409 (YP_001825921.1); Streptomyces
griseus subsp. griseus
25/35
unknown
tsrH
tsrI
681
474
hypothetical protein (ABC02784.1); Actinomadura melliaura
cytochrome P450-like enzyme (YP_001102921.1); Saccharopolyspora
erythraea NRRL 2338
37/44
29/44
cyclodehydratase/docking protein
oxygenase
tsrJ
437
putative nonribosomal peptide synthetase, adenylation domain
(ABI22132.1); Streptomyces laVendulae
cytochrome P450 (YP_001275131.1); Roseiflexus sp. RS-1
hypothetical protein SGR_4412 (YP_001825924); Streptomyces griseus
subsp. griseus
28/39
adenylation enzyme
tsrK
tsrL
396
912
31/48
27/36
oxygenase
unknown/lantibiotic dehydratase/
epimerase
tsrM
tsrN
599
272
Fe-S oxidoreductase (EAY55855.1) Leptospirillium sp. Group II UBA
short-chain dehydrogenase/reductase SDR (YP_001362890.1);
Kineococcus radiotolerans SRS30216
26/39
41/52
methyltransferase
dihydropyridine reductase
a Numbers are in amino acids. b NCBI accession numbers are given in parentheses.
by TsrF, since the L stereochemistry would likely be required
of its substrate.
(Scheme 1A).20,21,47,48 A hetero-Diels-Alder strategy has, in
fact, been employed in the syntheses of this core feature of the
thiopeptides, including those for thiostrepton, GE2270A, and
other model systems.49-54 Each of these synthetic approaches,
however, necessitated modifications to the proposed biosynthetic
A defining feature of the thiopeptides is the pyridine (e.g., 2
and 3) or dehydropiperidine (e.g., 1) ring resulting from the
condensation of two serine residues at their ꢀ-carbons and the
R-carboxyl carbon an adjacent cysteine residue (Scheme 1).17-22
This likely follows generation of the dehydroalanine residues,
as suggested by Bycroft, Floss, and others.17-22,47 In 1, the
dehydropiperidine arises from S5, C13, and S14. One proposal
for the introduction of this central scaffold of the thiopeptides
suggests what is formally a hetero-Diels-Alder cyclization,
though not necessarily proceeding as a concerted cycloaddition
(48) Mocek, U.; Beale, J. M.; Floss, H. G. J. Antibiot. 1989, 42, 1649–
1652.
(49) Moody, C. J.; Hughes, R. A.; Thompson, S. P.; Alcaraz, L. Chem.
Commun. 2002, 1760–1761.
(50) Nicolaou, K. C.; Dethe, D. H.; Chen, D. Y. Chem. Commun. 2008,
2632–2634.
(51) Nicolaou, K. C.; Dethe, D. H.; Leung, G. Y.; Zou, B.; Chen, D. Y.
Chem. Asian J. 2008, 3, 413–429.
(52) Nicolaou, K. C.; Nevalainen, S. B. S.; Zak, M.; Bulat, S. Angew.
Chem., Int. Ed. 2002, 41, 1941–1943.
(47) Bycroft, B. W.; Gowland, M. S. J. Chem. Soc. Chem. Comm. 1978,
(53) Nicolaou, K. C.; Safina, B. S.; Zak, M.; Estrada, A. A.; Lee, S. H.
Angew. Chem., Int. Ed. 2004, 43, 5087–5092.
256–258.
9
J. AM. CHEM. SOC. VOL. 131, NO. 12, 2009 4331

A R T I C L E S
Kelly et al.
Scheme 1. Proposed Biosynthesis of the Dehydropiperidine Thiopeptides
Scheme 2. Proposed Biosynthesis of
4-(1-Hydroxyethyl)quinoline-2-carboxylic Acid
cyclization substrate in order to present an appropriate partner
diene and dienophile for cycloaddition. These syntheses of the
pyridine and dehydropiperidine thiopeptide scaffolds suggest
that a suitable biosynthetic intermediate for a hetero-Diels-Alder
cyclization would need to be produced as a free metabolite or
sequestered within the active site of the cyclization enzyme.
As an alternative to a hetero-Diels-Alder cycloaddition, the
cyclase facilitating formation of the dihydropyridine ring of 5
could operate by acid and base catalysis with the strategic
utilization of the adjacent thiazole as an electron sink, compa-
rable to thiamine diphosphate-dependent transformations (Scheme
1B). TsrD bears weak similarity to the C-terminal region of
lantibiotic dehydratases and could be harnessed for this cy-
clization, although alternative candidates cannot yet be ruled
out. Dehydration of 5 then leads to the dihydropyridine 8
(Scheme 1). For 1 and the reduced thiopeptides, an NAD(P)H-
dependent reductase, such as TsrN, acting upon 8 would restore
the dehydropiperidine.
Two additional modifications of the TsrA backbone are
required during thiostrepton maturation: carboxyl-terminal amid-
ation and the dihydroxylation of I10. The latter can be explained
by the action of one or both of the cytochrome P450s TsrI and
TsrK. Most carboxyl-terminal amides that are present in
peptides are introduced by the oxidative cleavage of the
carboxyl-terminal residue.55,56 This is not the case for thio-
strepton since no additional residues follow S17 of TsrA.
Instead, an amidotransferase would be required to transmit
ammonia from glutamine to an activated carboxylic acid. To
our knowledge, utilization of an amidotransferase is a novel
strategy to introduce a carboxyl-terminal amide into a peptide.
TsrT, a protein similar to asparagine synthetases, could fulfill
this role.
To address the function of the putative amidotransferase in
thiostrepton biosynthesis, tsrT was inactivated in S. laurentii.
As anticipated, this strain lost the ability to produce 1, while
accumulating a new metabolite with a similar ultraviolet-visible
absorption spectrum to thiostrepton 1 (Figure 3 and Supporting
Information). Further analysis by HPLC-MS revealed a [M +
2H]2+ ion at m/z 833.4, relative to that at 832.9 for 1, indicating
an increase in molecular weight from 1664.4 to 1665.4. This is
consistent with generation of a thiostrepton analog, 10, pos-
sessing the free carboxylic acid at the C-terminus (Figure 5).
Biosynthesis of the Quinaldic Acid Moiety. The quinaldic acid
moiety observed in 1 originates from tryptophan 11 (Scheme
2).22 There are several consistencies in the tsr cluster with earlier
studies.22,29,33,35 An early step in the conversion of tryptophan
to the free intermediate 4-(1-hydroxyethyl)quinoline-2-carboxyl-
ate 16 is a methylation to provide 2-methyltryptophan 12.33,35
Unlike a majority of S-adenosylmethionine (SAM)-dependent
methyltransferase reactions, this methylation was shown to
proceed with net retention of the methyl group configuration.35
Similar methylations, thought to be facilitated by radical SAM-
dependent enzymes, have been implicated in the biosyntheses
(54) Nicolaou, K. C.; Zak, M.; Safina, B. S.; Lee, S. H.; Estrada, A. A.
Angew. Chem. In. Ed. 2004, 43, 5092–5097.
(55) Kulathila, R.; Merkler, K. A.; Merkler, D. J. Nat. Prod. Rep. 1999,
16, 145–154.
(56) Muller, I.; Weinig, S.; Steinmetz, H.; Kunze, B.; Veluthoor, S.;
Mahmud, T.; Muller, R. ChemBioChem 2006, 7, 1197–1205.
Figure 5. Proposed structure of 10.
9
4332 J. AM. CHEM. SOC. VOL. 131, NO. 12, 2009

Thiostrepton Biosynthesis
A R T I C L E S
oftheꢀ-lactamthienamycinandtheaminocoumarinclorobiocin.57-59
One putative methyltransferase in the tsr locus, TsrP, fails to
account for the stereochemical requirements needed for the
formation of 2-methyltryptophan 12. On the other hand, TsrM,
with similarity to radical-SAM-dependent proteins, is a more
appropriate candidate to accommodate the methylation of 11
with retention of the methyl group stereochemistry.
The next step proposed for the biosynthesis of 16 is the
generation of 3-(2-methylindolyl)pyruvic acid 13, occurring
prior to ring expansion to the quinaldic acid. The R-keto acid
could be introduced either by a PLP-dependent aminotransferase,
a FAD-dependent amino acid oxidase, or an NAD+-dependent
dehydrogenase, as observed in the biosyntheses of indolmycin,
rebeccamyin, and scytonemin, respectively.60-62 One protein
in the tsr locus, TsrV, is similar to a number of putative PLP-
dependent aminotransferases and, on the basis of sequence
comparison, contains the active site lysine residue needed to
covalently bind the cofactor.63 TsrV was heterologously ex-
pressed from E. coli BL21(DE3) with an N-terminal hexahis-
tidine affinity tag. In the presence of 2-methyltryptophan 12
and the cosubstrate 3-indolylpyruvic acid 17, TsrV catalyzed
the formation of L-tryptophan 11 and 3-(2-methylindolyl)pyruvic
acid 13, as detected by HPLC-MS (Figures 6 and Supporting
Information). Further characterization of TsrV will address not
only whether 3-indolylpyruvic acid is truly the preferred R-keto
acid cosubstrate but also whether the aminotransferase is tolerant
of additional tryptophan derivatives.
It is not yet clear which gene products are responsible for
the latter steps needed to generate thiostrepton’s quinaldic acid
moiety. Expansion of the pyrrole ring could occur by a pathway
depicted in Scheme 2, leading to the quinoline ring system of
16 by a series of rearrangements and isomerizations, reminiscent
of proposals for quinoline alkaloid biosynthesis.29,64 It is
interesting to note that a distinct mechanism to cleave the C-2/
C-3 bond of tryptophan is likely utilized for the thiostrepton
quinaldic acid relative to that observed for other bacterial
quinaldic acid metabolites and the early steps of tryptophan
catabolism. Xanthurenic acid, quinoxaline-2-carboxylic acid, and
3-hydroxyquinaldic acid are constituents of quinolobactin,
echinomycin, and thiocoraline, respectively.65-68 In contrast to
the pathway proposed for the quinaldic acid 16, these latter three
tryptophan derivatives require the oxidative cleavage of the
Figure 6. Reaction catalyzed by TsrV and HPLC-MS analysis of TsrV
activity with 2-methyltryptophan. The chromatograms show the negative
ions extracted for (A) m/z 217 and (B) m/z 203. Retention times and
molecular ions are as follows. 2-Methyltryptophan 12 at tR ) 12.7 min and
a [M - H]- ion at m/z 216.9; 3-(2-methylindolyl)pyruvate 13 at tR ) 24.9
min and a [M - H]- ion at m/z 215.8; tryptophan 11 at tR ) 8.5 min and
a [M - H]- ion at m/z 202.8; 3-indolylpyruvate 17 at tR ) 22.8 min and
a [M - H]- ion at m/z 201.8.
indolebetweenC-2andC-3byaheme-dependentdioxygenase.65,67,68
Prior to introduction of the quinoxaline or quinoline ring system,
the C2 carbon is lost as formate, whereas the C2 carbon is
retained in 16.22,65,67,68
Floss et al. demonstrated the ATP-dependent activation of
16 in a cell-free extract, and this a logical function for TsrJ,
which has similarity to adenylating enzymes.29 Following
attachment of the quinaldic acid to the T12 side chain,
epoxidation by either TsrQ (a flavin-dependent enzyme) or one
of the cluster’s cytochrome P450s (TsrI or TsrK) would then
provide a suitable substrate for nucleophilic attack by the
N-terminal amine of I1 to complete the attachment of this moiety
and the generation thiostrepton’s second macrocycle. Whether
introduction of the quinaldic acid appendage in thiostrepton 1
precedes or follows any of the other posttranslational modifica-
tions of the prepeptide backbone is unknown at this time.
Prepeptide Processing and Self-Resistance. There is no
obvious gene product from the tsr locus that would be a suitable
candidate to liberate the TsrA leader peptide. TsrU and TsrB
are both similar to the R/ꢀ-hydrolase superfamily but are either
more similar to lipases than proteases or lack the conserved
motif for the nucleophilic serine, respectively.69 It is certainly
conceivable, however, that the putative lipase TsrU could be
utilized as a protease rather than a lipase. Both lantibiotics and
microcins can recruit proteases encoded from outside their
biosynthetic gene clusters; therefore, the lack of a dedicated
TsrA peptidase for thiostrepton would not be particularly
troubling.45,70-72 On a similar note, there is no evident gene
(57) Houck, D. R.; Kobayashi, K.; Williamson, J. M.; Floss, H. G. J. Am.
Chem. Soc. 1986, 108, 5365–5366.
(58) Nunez, L. E.; Mendez, C.; Brana, A. F.; Blanco, G.; Salas, J. A. Chem.
Biol. 2003, 10, 301–311.
(59) Westrich, L.; Heide, L.; Li, S. M. ChemBioChem 2003, 4, 768–773.
(60) Balskus, E. P.; Walsh, C. T. J. Am. Chem. Soc. 2008, 130, 15260–
15261.
(61) Nishizawa, T.; Aldrich, C. C.; Sherman, D. H. J. Bacteriol. 2005,
187, 2084–2092.
(62) Speedie, M. K.; Hornemann, U.; Floss, H. G. J. Biol. Chem. 1975,
250, 7819–7825.
(63) Sivaraman, J.; Li, Y.; Larocque, R.; Schrag, J. D.; Cygler, M.; Matte,
A. J. Mol. Biol. 2001, 311, 761–776.
(64) Dewick, P. M. Medicinal natural products: a biosynthetic approach,
2nd ed.; John Wiley and Sons Inc.: Chicester, England, 2001.
(65) Lombo, F.; Velasco, A.; Castro, A.; de la Calle, F.; Brana, A. F.;
Sanchez-Puelles, J. M.; Mendez, C.; Salas, J. A. ChemBioChem 2006,
7, 366–376.
(66) Matthijs, S.; Baysse, C.; Koedam, N.; Tehrani, K. A.; Verheyden, L.;
Budzikiewicz, H.; Schafer, M.; Hoorelbeke, B.; Meyer, J. M.; De
Greve, H.; Cornelis, P. Mol. Microbiol. 2004, 52, 371–384.
(67) Sheoran, A.; King, A.; Velasco, A.; Pero, J. M.; Garneau-Tsodikova,
S. Mol. Biosyst. 2008, 4, 622–628.
(69) Nardini, M.; Dijkstra, B. W. Curr. Opin. Struct. Biol. 1999, 9, 732–
737.
(68) Watanabe, K.; Hotta, K.; Praseuth, A. P.; Koketsu, K.; Migita, A.;
Boddy, C. N.; Wang, C. C.; Oguri, H.; Oikawa, H. Nat. Chem. Biol.
2006, 2, 423–428.
(70) Duquesne, S.; Destoumieux-Garzon, D.; Peduzzi, J.; Rebuffat, S. Nat.
Prod. Rep. 2007, 24, 708–734.
9
J. AM. CHEM. SOC. VOL. 131, NO. 12, 2009 4333

A R T I C L E S
Kelly et al.
product conferring a mechanism for self-resistance to 1 in the
tsr genetic locus. Previous work by Floss and co-workers
demonstrated that an rRNA methyltransferase encoded outside
the limits of a thiopeptide biosynthetic cluster imparts resistance
to thiostrepton in S. laurentii.24 The self-resistance gene for
nosiheptide 2 in the producing bacterium, Streptomyces actuo-
sus, was also found to reside outside a biosynthetic gene cluster,
suggesting that this may be common for thiopeptide antibiotics.73
across genera.42 Therefore, this report establishes the thiopeptide
antibiotics as a new class of bacteriocin.
Despite potent in vitro activity, development of a clinically
useful thiopeptide antibiotic is hampered by the poor solubility
of these agents in water. The looming crisis in antimicrobial
chemotherapy has contributed to a renewed interest in the
discovery and development of thiopeptide analogs with im-
proved water solubility. The complex architecture of thiopeptides
is a significant challenge toward de novo synthesis. Biosynthetic
engineering of thiopeptide analogues presents an attractive
alternative that can be coupled with semisynthetic modifications.
This will be facilitated by a detailed understanding of the
biosynthetic machinery required to construct the complex
thiopeptide scaffold, and it is now possible to query the
production of variants through site-directed mutagenesis of TsrA.
Conclusion
We have established that thiostrepton is derived from TsrA,
a simple, genetically encoded peptide. Disruption of tsrA and
tsrT in S. laurentii abolish production of thiostrepton, and the
tsrT mutant led to the accumulation of a thiostrepton derivative.
Initial characterization of TsrV established that it is an ami-
notransferase catalyzing a step in the conversion of tryptophan
to 4-(1-hydroxyethyl)quinoline-2-carboxylate, a substituent at-
tached to thiostrepton. Additional investigations will establish
the temporal mechanisms guiding the formation of this exten-
sively modified peptide.
Several cryptic biosynthetic clusters were recently identified
to harbor enzymatic machinery comparable to that observed here
for 1.42 Both 1 and goadsporin are peptides formed in actino-
mycetes that are processed by dehydration and heterocyclization,
and similarities between their maturation enzymes are ob-
served.40 These similarities extend to other peptide metabolites
identified by Lee et al., particularly with TsrD homologues,
suggesting a widespread occurrence of thiopeptide metabolites
Acknowledgement. This work is supported by Georgia Institute
of Technology, Camille and Henry Dreyfus Foundation, and
American Society of Pharmacognosy. L.P. was awarded a GAANN
predoctoral fellowship. David Bostwick (Georgia Institute of
Technology) is thanked for his dedicated assistance with the HPLC-
MS. Kathryn E. Roege and Tala Suidan (Georgia Institute of
Technology) assisted with genome analysis. Juan Pablo Arago´n
(Georgia Institute of Technology) is thanked for his efforts in
constructing the cosmid library. Janet Westpheling (University of
Georgia) assisted with intergeneric conjugation. Craig A. Townsend
(Johns Hopkins University), and other colleagues are thanked for
their helpful discussions.
Supporting Information Available: Additional materials and
methods, tables of strains, plasmids, and primers, and supple-
mental figures. This material is available free of charge via the
Internet at http://pubs.acs.org.
(71) Widdick, D. A.; Dodd, H. M.; Barraille, P.; White, J.; Stein, T. H.;
Chater, K. F.; Gasson, M. J.; Bibb, M. J. Proc. Natl. Acad. Sci. U.S.A.
2003, 100, 4316–4321.
(72) Yorgey, P.; Davagnino, J.; Kolter, R. Mol. Microbiol. 1993, 9, 897–
905.
(73) Li, Y.; Dosch, D. C.; Strohl, W. R.; Floss, H. G. Gene 1990, 91, 9–
17.
JA807890A
9
4334 J. AM. CHEM. SOC. VOL. 131, NO. 12, 2009