A type II polyketide synthase from the gram-negative bacterium Stigmatella aurantiaca is involved in aurachin alkaloid biosynthesis

Article abstract of DOI:10.1002/anie.200603513

A closer look: The path for the biosynthesis of aurachin alkaloids (aurachin A is shown) has been deduced by molecular biological methods, including the cloning and sequencing of the biosynthetic gene cluster in the bacterium Stigmatella aurantiaca. The gene cluster encodes a type II polyketide synthase, which was discovered in this form for the first time in a Gram-negative bacterium. (Chemical Equation Presented).

Full text of DOI:10.1002/anie.200603513

Communications
DOI: 10.1002/anie.200603513
Biosynthesis Mechanisms
A TypeII Polyketide Synthase from the Gram-Negative Bacterium
Stigmatella aurantiaca Is Involved in Aurachin Alkaloid
Biosynthesis**
Axel Sandmann, Jeroen Dickschat, Holger Jenke-Kodama, Brigitte Kunze, Elke Dittmann, and
Rolf Müller*
Dedicated to Professor Thomas Hartmann on the occasion of his 70th birthday
Over the last two decades, myxobacteria have emerged as a
rich source of secondary metabolites (SMs) with potent
biological activities.[1] These compounds often contain struc-
tural elements rarely found in SMs from other sources.[2]
Interestingly, several SM-multiproducers are known among
the myxobacteria, including Stigmatella aurantiaca Sga15,
from which the aurachins,[3] myxochelins,[4] myxalamids, and
stigmatellins[5,6] have been isolated. Whereas the molecular
details of myxochelin, myxalamid, and stigmatellin biosyn-
theses have been described in detail,[7] no such information is
available about the pathway to the aurachins.
The aurachins are rare quinoline alkaloids that also
incorporate an isoprenoid side chain. Owing to their struc-
tural similarity to vitamin K, they act as potent inhibitors of
electron transport in the respiratory chain.[2] The structurally
related isoprenylated quinoline antibiotics are potent growth
inhibitors of Helicobacter pylori,[8] while two members of the
4-hydroxy-2-alkylquinoline family (HAQs), 3,4-dihydroxy-2-
heptylquinoline (PQS)[9] and 4-hydroxy-2-heptylquinoline,
function as signaling molecules in cell–cell communication
in Pseudomonas aeruginosa.[10]
plant-type
3-deoxy-d-arabinoheptulosonate-7-phosphate
(DAHP) synthase instead of by the standard bacterial
enzyme[12]—is also the starting unit of aurachin biosynthesis.
Based on these experiments, 1 was speculated to be extended
by acetoacetate, leading to the hypothetical early-pathway
intermediate 4-hydroxy-2-methylquinoline (3; Scheme 1a).[13]
An attractive alternative mechanism, however, is that inter-
mediate 3 arises from the elongation of activated anthranilate
by two malonyl-CoA extender units, catalyzed by a typeIII
polyketide synthase (PKS). In support of this hypothesis, a
plant typeIII PKS (acridone synthase) is responsible for
formation of acridone (9), in which N-methylated anthrani-
late 8 is condensed with three units of malonyl-CoA
(Scheme 1b).[14] TypeIII PKSs, also known as chalcone
synthases (CHSs), were believed to be present only in
plants but were recently discovered in a number of bac-
teria.[15,16]
As initial experiments based on degenerate PCR did not
yield typeIII PKS products from genomic DNA of S. aur-
antiaca, we instead employed a mariner-based transposon
mutagenesis strategy to identify the genes for aurachin
biosynthesis. After screening by HPLC-MS of extracts from
a 4096-member library of S. aurantiaca Sga15 mutants grown
in microtiter plates (see the Supporting Information),[17–19] we
identified one clone (4.59) that produced significantly less
aurachin (Figure 1b) than the wild-type cells (Figure 1a).
This reduced titer was confirmed by comparisons of extracts
from 50-mL mutant cultures and from wild-type S. aurantiaca
Sga15. The chromosomal recovery of the transposon from
mutant 4.59 together with DNA sequences flanking the
transposable element in the genome[17] led to the identifica-
tion of the gene cluster for aurachin biosynthesis (see
Table 1). Sequencing of the recovery plasmid (p4.59/NotI)
revealed eleven open reading frames (ORFs) distributed
across 11191 bp of the S. aurantiaca Sga15 chromosome, with
an average guanine + cytosine content of 66.66%, which can
be regarded as typical for this bacterial species (unpublished
data; the gene cluster has been deposited in the EMBL
database under accession number AM404078).
HAQ biosynthesis has been shown to result from the
“head-to-head” condensation of anthranilic acid (1) and b-
keto fatty acids.[11] Feeding experiments with S. aurantiaca
revealed that anthranilate—which is curiously formed by a
[*] Dr. A. Sandmann, Dr. J. Dickschat,[+] Prof. Dr. R. Müller
Pharmaceutical Biotechnology, Saarland University
P.O. Box 151150, 66041 Saarbrücken (Germany)
Fax: (+49)681-302-5473
E-mail: rom@mx.uni-saarland.de
Homepage: http://www.myxo.uni-saarland.de
Dipl.-Biol. H. Jenke-Kodama, Prof. Dr. E. Dittmann
Institute of Biology, Humboldt University
Invalidenstrasse 43, 10115 Berlin (Germany)
Dr. B. Kunze
Helmholtz-Zentrum für Infektionsforschung GmbH
Inhoffenstrasse 7, 38124 Braunschweig (Germany)
[+] Current address:
Department of Biochemistry, University of Cambridge
80 Tennis Court Road, Cambridge CB21GA (UK)
Sequence analysis identified the transposition site within
orf1. However, the corresponding gene product shows
similarity to hydrolases and therefore is not thought to be
involved in the aurachin biosynthesis. Further sequence
analysis revealed additional orfs downstream of orf1. These
genes encode proteins with significant similarity to a ben-
[**] The authors thank Dr. K. Weissman and Dr. H. B. Bode for critical
review of this manuscript. This work was supported by DFG Priority
Programme 1152 “Evolution of Metabolic Diversity”.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
2712
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 2712 –2716

Angewandte
Chemie
Figure 1. HPLC-MS analysis of extracts of the wild-type S. aurantiaca
Sga15 and aua knockout mutants; a) wild type; b) mutant 4.59;
c) Sga-D_AS (the extracted-ion chromatograms of Sga-A_AS and Sga-
E_AS are the same); d) Sga-D_AS after complementation with 3. The
extracted-ion chromatograms of the aurachin derivatives from different
culture extracts are shown. 4: [M+H]+ 364; 5: [M+H]+ 380; 6:
[M+H]+ 380; 7: [M+H]+ 396.
a “chain-length factor” (CLF or KSb; AuaD). The KSa
subunit is responsible for the iterative condensation of
malonyl-CoA building blocks, whereas KSb is involved in
decarboxylation of the starter unit and determination of chain
length. AuaB and AuaC show only weak overall homology
(see Table 1) to their respective enzyme families, but the
conserved active site residues are present. In AuaD, which is
homologous to known CLFs, a threonine residue replaces the
active-site glutamine or glutamate (see Figure 3).[15,21]
In addition, a prenyltransferase (AuaA) is part of the gene
cluster (Figure 2). AuaA shows similarities to the 4-hydrox-
ybenzoate:polyprenyldiphosphate 3-polyprenyltransferase
UbiA from E. coli,[22] which prenylates 4-hydroxybenzoate
with farnesyl diphosphate. The enzyme AuaA is predicted by
the program SOSUI to contain ten putative transmembrane
helices (http://bp.nuap.nagoya-u.ac.jp/sosui). The benzoate-
CoA ligase like protein AuaE shares strong similarities with
EncN, which is part of the typeII PKS involved in enterocin
biosynthesis.[23,24] To prove the involvement of the aua gene
cluster in aurachin biosynthesis, gene-disruption experiments
were performed (see the Supporting Information).[25] As
shown by HPLC-MS analysis, disruption of auaA, auaD, or
auaE leads to abolishment of aurachin production (Fig-
ure 1c), demonstrating the essential role of the aua operon.
However, polar effects on downstream genes caused by the
Scheme 1. a) Proposed biosynthetic pathway for the formation of
aurachins D–A (4: aurachin D; 5: aurachin C, 6: aurachin B; 7: aura-
chin A). The biosynthetic steps that are catalyzed by the gene products
(AuaA–AuaE) from the aurachin core gene cluster are shown. The
intramolecular 1,2-migration of the farnesyl residue during the trans-
formation of 5 to 6 is accompanied by the loss of the labeled carbonyl
oxygen; this is compatible with a Wagner–Meerwein-type rearrange-
ment (G. Höfle, personal communication and Ref.[13]). b) Biosynthesis
of the plant metabolite acridone for comparison.
zoate-CoA ligases (AuaE), a set of proteins typical for typeII
PKS systems (AuaDCB), and a prenyltransferase (AuaA; see
Table 1). TypeII PKSs are multienzyme complexes consisting
of a single set of proteins that are used iteratively. To date,
they have been described exclusively from actinomycetes,
where they are involved in the biosynthesis of polycyclic
aromatic compounds such as actinorhodin.[20,21] The “mini-
mal” polyketide synthase is common to all typeII PKSs, and
comprises an acyl carrier protein (ACP; AuaB), and a
heterodimer consisting of a ketosynthase (KSa; AuaC) and
Angew. Chem. Int. Ed. 2007, 46, 2712 –2716
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
2713

Communications
Table 1: Proteins encoded in the sequenced region including the core gene cluster responsible for aurachin biosynthesis and their putative functions.
Protein (gene)
Size (in Da, bp)
Proposed homologue, origin (identity [%], similarity [%]; Access no.)
AuaA (auaA)
AuaB (auaB)
AuaC (auaC)
AuaD (auaD)
AuaE (auaE)
ORF1 (orf1)
ORF2 (orf2)
ORF3 (orf3)
ORF4 (orf4)
ORF5 (orf5)
ORF6 (orf6)
36605, 981
9572, 255
UbiA prenyltransferase family, Flavobacteria bacterium BBFL7 (28, 44; EAS21036)
aycl carrier protein, Enterococcus faecalis (34, 60; AAO82791)
42842, 1221
41891, 1242
53329, 1494
24354, 675
24958, 663
28683, 813
20379, 543
45410, 1286
47699, 1335
b-ketoacyl synthase, Desulfitobacterium hafniense (40, 59; YP_518890)
3-oxoacyl-ACP synthase II, Streptomyces avermitilis (31, 45; NP_824120)
benzoate-CoA ligase family, Rhodopseudomonas palustris (37, 52; YP_568017)
carboxymethylenebutenolidase, Rhodopseudomonas palustris (45, 56; NP_946545)
hypothetical protein, Flavobacterium johnsoniae (41, 59; ZP_01247630)
putative secreted hydrolase, Streptomyces coelicolor (56, 70; CAC16457)
transcriptional regulatory protein, Bradyrhizobium japonicum (40, 62; BAC47519)
protein of unknown function, delta proteobacterium MLMS-1 (28, 49; EAT04615)
protein of unknown function, delta proteobacterium MLMS-1 (28, 49; EAT04615)
D to aurachins C (5), B (6), and A (7) could not be identified,
and they must therefore be located at alternative sites in the
genome of S. aurantiaca Sga15. Although biosynthetic gene
clusters with split organization are known,[17,26] this type of
arrangement is still regarded as exceptional.[27]
In common with the aurachin typeII PKS, several other
bacterial aromatic PKSs use starter units other than acetyl-
CoA and malonyl-CoA.[28] For example, the doxorubicin and
daunorubicin PKSs are primed by propionyl-CoA,[29] whereas
frenolicin biosynthesis starts with butyryl-CoA.[30] The enter-
ocin typeII PKS from S. maritimus uses as a starting moiety
benzoyl-CoA, which is derived from benzoate in an EncN-
and ATP-dependent reaction.[31] Prior to this work, the
Figure 2. Top: Genetic organization of the gene cluster for aurachin
biosynthesis. The location and transcriptional orientation of the genes
are indicated by arrows. The flags show the sequence regions used for
plasmid integration in the different knockout mutants generated in
this work (see the Supporting Information), and the transposition site
of magellan4 in mutant 4.59 is marked with a triangle. Bottom: The
transcriptional overlap of auaE and orf1. The arrows indicate the
direction of transcription of the aurachin operon and of the additional
orfs located up- and downstream of the aurachin core cluster. Black
arrows indicate the minimal PKS, gray arrows mark the remaining
genes of the core cluster, and white arrows label the genes that encode
proteins with an unknown function. The position of the Tn5-promotor
in mutant 4.59 presumably activating transcription of the 3’ fragment
of orf1 is indicated.
enterocin synthase was the only typeII PKS known to
employ an aromatic starter unit. In order to verify the
suggested biosynthetic pathway to the aurachins, the hypo-
thetical intermediate 3 was synthesized (see the Supporting
Information) and administered to the cultures of the mutant
strains Sga-D_AS (auaDÀ) and Sga-E_AS (auaE; see the
Supporting Information). Production of metabolites 4–7 by
the mutants was restored, as determined by HPLC-MS
analysis (Figure 1d). In addition, GC-MS analysis of wild-
type extracts derivatized with N-methyl-N-trimethylsilyltri-
fluoroacetamide (MSTFA) and comparison to the synthe-
sized intermediate as a standard revealed small amounts of
intermediate 3. Intermediate 3 was not identified in extracts
of mutants Sga-D_AS and Sga-E_AS, while feeding of
intermediate 3 failed to restore aurachin production in the
mutant Sga-A_AS (auaAÀ, data not shown). Furthermore, 3
accumulated in the mutant Sga-A_AS, presumably because it
could not be processed further to the aurachins (see the
Supporting Information). Taken together, the results of these
experiments are entirely consistent with the proposed order
of transformations in aurachin biosynthesis.
In the core gene cluster, genes auaB and auaC as well as
auaD and auaE are most likely translationally coupled, as
evidenced by their overlapping start and stop codons. This
organization indicates that the aurachin core cluster forms an
operon that directs tight coexpression of the involved
enzymes. The borders of the aurachin gene cluster are defined
by orf4 and orf1, which are transcribed in the reverse
direction to the core cluster. Inactivation of these two genes
by homologous recombination did not affect aurachin pro-
duction (data not shown, see the Supporting Information).
plasmid insertions cannot be ruled out at this point. Plasmids
needed for complementation experiments are not available in
myxobacteria, and efforts made toward double-crossover
mutagenesis in this strain have not met with success (unpub-
lished results).
To our knowledge, this is the first time a typeII PKS has
been identified from a Gram-negative bacterium. Based on
the assignment of the gene cluster, a biosynthetic scheme for
aurachins can be proposed (Scheme 1). Activation of 1 to
anthraniloyl-CoA (2) by AuaE and transfer to the acyl carrier
protein AuaB results in anthraniloyl-S-AuaE. The minimal
PKS (consisting of primed AuaB plus AuaC and D) forms 3
by extension of 2 with two malonyl-CoA units, followed by
release of the carboxylic acid along with decarboxylation of
the second extender unit. Prenylation with farnesyl diphos-
phate catalyzed by AuaA would then yield aurachin D (4).
Genes presumably involved in the transformation of aurachin
2714
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 2712 –2716

Angewandte
Chemie
Unusually, the 3’-ends of orf1 and auaE transcriptionally
overlap (see Figure 2), which may explain the finding that
mutant 4.59 generated very low amounts of 4, 5, and 6, while
production of 7 was at essentially the level of the wild-type
cells (nevertheless only representing a minor fraction of the
total aurachins). In this mutant, transposition occurred in
orf1, and presumably resulted in its induction (Figure 2).
Owing to the overlap with auaE, this induction may have
decreased the expression of the aua operon. As a result,
formation of 4 likely decreased, but the expression of the
additional enzymes proposed to catalyze the transformation
from 4 to 7 would have been unaffected, leading to essentially
complete turnover of the intermediates to 7.
To discern the evolutionary relationship between the
aurachin PKS and several typeII and typeIII PKSs (including
acridone synthase), we analyzed a set of sequences from
bacteria and plants representing all major typeII and typeIII
PKS systems identified to date (Figure 3).[16] We used arch-
aebacterial FabH homologues as the outgroup and observed
two major clades. The first contains bacterial FabH sequences
and all chalcone synthase-like sequences from bacteria and
plants, while the second comprises KSa and KSb proteins of
bacterial typeII PKS. Addition of further bacterial typeII
PKS did not change the tree typology at all (data not shown).
AuaC clearly belongs to the subclade that contains well-
characterized KSa sequences from actinobacteria, although it
is only distantly related and therefore forms a separate
branch.
AuaD occupies an independent branch within the typeII
clade of the phylogenetic tree and therefore presumably
represents a new sequence type of typeII PKS CLF. Typically
the CLF active site contains a highly conserved glutamine
(Figure 3), which renders it an effective decarboxylase for
chain initiation.[32,33] WhiE from Streptomyces coelicolor is the
only known example in which the active-site glutamine is
replaced by a glutamate residue. In AuaD, the active-site
glutamine is substituted by a threonine, which may indicate
that AuaD is inactive and therefore not required for aurachin
biosynthesis. Disruption of AuaD may therefore only abolish
aurachin biosynthesis as a result of polar effects (see above).
On the other hand, the active-site glutamine may not be
required for chain-length control and only serve to mediate
protein/protein interactions between the two subunits.[34] If
AuaD is really required, it could still serve as chain-length
factor as previously discussed on the basis of the KSa/KSb
crystal structure.[34] Taken together, these observations sug-
gest that the aurachin PKS belongs to a new group of PKSs
that are undergoing an evolutionary transition from typeII
PKS into typeIII PKS and/or FABs or vice versa. Experi-
ments are currently underway in our laboratory to evaluate
this hypothesis and to verify the postulated biosynthetic
mechanism.
Figure 3. a) Phylogenetic relationship of AuaC and AuaD with selected
typeII and typeIII PKSs. For all major nodes, the Bayesian support
values are given above the branches. Branch lengths indicate the
number of inferred amino acid changes per position. b) Active-site
sequences of typical KSa and CLF proteins of typeII PKSs relative to
those of AuaC and AuaD. These alignments were generated using the
sequences of frenolicin (Fren), enterocin (Enc), oxytetracycline (Otc),
tetracenomycin (Tcm), S. coelicolor spore pigment (WhiE), and actino-
rhodin (Act) subunits.
[1]K. Gerth, S. Pradella, O. Perlova, S. Beyer, R. Müller, J.
Biotechnol. 2003, 106, 233.
Received: August 28, 2006
Revised: November 13, 2006
Published online: March 5, 2007
[2]H. Reichenbach, G. Höfle in Drug Discovery from Nature (Eds.:
S. Grabley, R. Thiericke), Springer, Berlin, 1999, p. 149.
[3]B. Kunze, G. Höfle, H. Reichenbach, J. Antibiot. 1987, 40, 258.
[4]B. Kunze, N. Bedorf, W. Kohl, G. Höfle, H. Reichenbach, J.
Antibiot. 1989, 42, 14.
Keywords: alkaloids · biosynthesis · myxobacteria ·
.
natural products · polyketides
Angew. Chem. Int. Ed. 2007, 46, 2712 –2716
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
2715

Communications
[5]K. Gerth, R. Jansen, G. Reifenstahl, G. Höfle, H. Irschik, B.
[17]A. Sandmann, F. Sasse, R. Müller, Chem. Biol. 2004, 11, 1071.
[18]M. Kopp, H. Irschik, F. Gross, O. Perlova, A. Sandmann, K.
Gerth, R. Müller, J. Biotechnol. 2004, 107, 29.
[19]M. Kopp, H. Irschik, S. Pradella, R. Müller, ChemBioChem 2005,
6, 1277.
[20]B. J. Rawlings, Nat. Prod. Rep. 1999, 16, 425.
[21]J. Staunton, K. J. Weissman, Nat. Prod. Rep. 2001, 18, 380.
[22]M. Melzer, L. Heide, Biochim. Biophys. Acta 1994, 1212, 93.
[23]C. Hertweck, B. S. Moore, Tetrahedron 2000, 56, 9115.
[24]J. Piel, C. Hertweck, P. R. Shipley, D. M. Hunt, M. S. Newman,
B. S. Moore, Chem. Biol. 2000, 7, 943.
Kunze, H. Reichenbach, G. Thierbach, J. Antibiot. 1983, 36, 1150.
[6]B. Kunze, T. Kemmer, G. Höfle, H. Reichenbach, J. Antibiot.
1984, 37, 454.
[7]H. B. Bode, R. Müller, J. Ind. Microbiol. Biotechnol. 2006, 33,
577.
[8]K. A. Dekker, T. Inagaki, T. D. Gootz, L. H. Huang, Y. Kojima,
W. E. Kohlbrenner, Y. Matsunaga, P. R. McGuirk, E. Nomura, T.
Sakakibara, S. Sakemi, Y. Suzuki, Y. Yamauchi, N. Kojima, J.
Antibiot. 1998, 51, 145.
[9]E. C. Pesci, J. B. Milbank, J. P. Pearson, S. McKnight, A. S.
Kende, E. P. Greenberg, B. H. Iglewski, Proc. Natl. Acad. Sci.
USA 1999, 96, 11229.
[25]S. Beyer, B. Kunze, B. Silakowski, R. Müller, Biochim. Biophys.
Acta 1999, 1445, 185.
[10]E. Deziel, F. Lepine, S. Milot, J. He, M. N. Mindrinos, R. G.
Tompkins, L. G. Rahme, Proc. Natl. Acad. Sci. USA 2004, 101,
1339.
[26]T. W. Yu, L. Bai, D. Clade, D. Hoffmann, S. Toelzer, K. Q. Trinh,
J. Xu, S. J. Moss, E. Leistner, H. G. Floss, Proc. Natl. Acad. Sci.
USA 2002, 99, 7968.
[11]F. Bredenbruch, M. Nimtz, V. Wray, M. Morr, R. Müller, S.
Häussler, J. Bacteriol. 2005, 187, 3630.
[12]B. Silakowski, B. Kunze, R. Müller, Arch. Microbiol. 2000, 173,
403.
[13]G. Höfle, H. Reichenbach in Sekundärmetabolismus bei Mik-
roorganismen (Eds.: W. Kuhn, H. P. Fiedler), Attempo, Tübin-
gen, 1995, p. 61.
[27]S. C. Wenzel, R. Müller, Curr. Opin. Chem. Biol. 2005, 9, 447.
[28]B. S. Moore, C. Hertweck, Nat. Prod. Rep. 2002, 19, 70.
[29]C. R. Hutchinson, Chem. Rev. 1997, 97, 2525.
[30]M. J. Bibb, D. H. Sherman, S. Omura, D. A. Hopwood, Gene
1994, 142, 31.
[31]M. Izumikawa, Q. Cheng, B. S. Moore, J. Am. Chem. Soc. 2006,
128, 1428.
[14]K. T. Junghanns, R. E. Kneusel, A. Baumert, W. Maier, D.
Gröger, U. Matern, Plant Mol. Biol. 1995, 27, 681.
[15]M. B. Austin, J. P. Noel, Nat. Prod. Rep. 2003, 20, 79.
[16]F. Gross, N. Luniak, O. Perlova, N. Gaitatzis, H. Jenke-Kodama,
K. Gerth, D. Gottschalk, E. Dittmann, R. Müller, Arch. Micro-
biol. 2006, 185, 28.
[32]J. Dreier, C. Khosla, Biochemistry 2000, 39, 2088.
[33]C. Bisang, P. F. Long, J. Cortꢀs, J. Westcott, J. Crosby, A.-L.
Matharu, R. J. Cox, T. J. Simpson, J. Staunton, P. F. Leadlay,
Nature 1999, 401, 502.
[34]A. T. Keatinge-Clay, D. A. Maltby, K. F. Medzihradszky, C.
Khosla, R. M. Stroud, Nat. Struct. Mol. Biol. 2004, 11, 888.
2716
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 2712 –2716