Multifactorial control of iteration events in a modular polyketide assembly line

Article abstract of DOI:10.1002/anie.201301322

Freedom and control: First insights into the rare programmed iteration of an individual polyketide synthase (PKS) module were obtained from the analysis and mutation of aureothin (1) synthase. The first ketosynthase (KS) domain primes the PKS, allowing intermediate retrotransfer. Addition of a designated loading module results in a complete loss of iteration. The downstream KS functions as a gatekeeper for correct chain length. Copyright

Full text of DOI:10.1002/anie.201301322

Angewandte
Chemie
DOI: 10.1002/anie.201301322
Aureothin Polyketide Synthase
Multifactorial Control of Iteration Events in a Modular Polyketide
Assembly Line**
Benjamin Busch, Nico Ueberschaar, Swantje Behnken, Yuki Sugimoto, Martina Werneburg,
Nelly Traitcheva, Jing He, and Christian Hertweck*
Complex bacterial polyketides such as polyenes, macrolides,
and polyethers are a rich source of clinically used therapeu-
tics. Despite their highly diverse structures, these valuable
compounds are assembled from simple acyl and malonyl
building blocks in an enzymatic assembly line.[1] The poly-
ketide synthases (PKSs) in charge of selecting, fusing, and
processing the building blocks are organized into modules
that consist of individual catalytic domains. Typically, there is
a unidirectional progress of chain elongation, where each
domain is only used once in the biosynthesis of one polyketide
molecule. Thus, for most synthases there is a strict colinearity
between the number and architecture of the modules and the
number of elongations and degree of b-keto processing.[2] The
beauty of this rule of colinearity is that it is possible to deduce
the structure of a polyketide metabolite from the organization
of the PKSs and vice versa. In addition, the rule of colinearity
has proven extremely valuable for rationally engineering
PKSs[3–6] and to predict structures of cryptic natural products
from genome sequences by genome mining.[7,8] However,
there are more and more deviations from this textbook
knowledge; various modular PKS single domains or entire
modules are skipped and individual modules may be used
more than once.[2] Notably, in contrast to an erratic processing
of the PKS, these events are programmed and apparently
have developed during the evolution of the assembly lines.
The first examples for the programmed iterative use of
a type I PKS module were observed in the stigmatellin,[9]
aureothin[10] (each two iterations), and borellidin[11] (three
iterations) pathways. Module fusions of the aureothin and
borrelidin PKSs proved that individual modules are respon-
sible for the iterative processes.[11,12] Recently it was shown
that iteration takes place by retrotransfer of the intermediate
from one PKS strand to the opposite PKS strand,[13] and that
a ratchet mechanism of KS–ACP interactions precludes
iterative use of any single module in colinear systems.[14]
Despite the growing number of non-colinear PKSs,[1] to date
the mechanisms that control the iteration processes in non-
colinear modular PKSs have remained obscure.[15] A possible
scenario could be that iteration is an inherent feature of the
module as, for example, in fatty acid synthases or fungal PKSs.
Alternatively, the module may be “forced” to iterate because
the downstream module only accepts an intermediate with
a particular chain length. A combination of both is possible,
too. Finally, the kinetics of iteration and de novo loading of
the ketosynthase needed to be fine-tuned to allow for an
unobstructed metabolite flux. Herein, we report the first
insights into the control of iteration in a modular PKS and
reveal that the interplay of multiple components is essential
to control the exact number of elongation cycles.
The aureothin PKS from Streptomyces thioluteus
(Scheme 1) represents a favorable model system to inves-
tigate the iteration process because the size of the gene cluster
[*] B. Busch, N. Ueberschaar, S. Behnken, Y. Sugimoto, M. Werneburg,
N. Traitcheva, J. He, Prof. Dr. C. Hertweck
Leibniz-Institut fꢀr Naturstoff-Forschung und Infektionsbiologie –
Hans-Knçll-Institut (HKI), Abteilung fꢀr Biomolekulare Chemie
Beutenbergstrasse 11a, 07745 Jena (Germany)
Scheme 1. Model of aureothin (1) biosynthesis. Non-colinear assembly
by a modular PKS. AT=acyltransferase, DH=dehydratase, ER=enoyl
reductase, KR=ketoreductase, KS=ketosynthase, TE=thioesterase.
E-mail: Christian.Hertweck@hki-jena.de
Prof. Dr. C. Hertweck
Lehrstuhl fꢀr Naturstoffchemie
Friedrich-Schiller-Universitꢁt Jena (Germany)
allows engineering and expression studies in heterologous
hosts (such as Streptomyces albus). As no aureothin homo-
logues with shortened or extended polyene stretches are
produced naturally, the iteration event in the biosynthesis of
aureothin is tightly controlled. Furthermore, a particular
feature of the aur PKS is that the iterating module is, at the
[**] We thank A. Perner for MS analysis and H. Heinecke for NMR
measurements. This work was supported by the Federal Ministry of
Science and Technology (BMBF (Germany)) and the German
Science Foundation (DFG).
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201301322.
Angew. Chem. Int. Ed. 2013, 52, 1 – 6
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

.
Angewandte
Communications
Figure 1. Investigation of the priming of the iterative module (AurA). A) The ligase AurE solely plays a role in activating the p-nitrobenzoate
(PNBA) starter unit. For iteration, KS1 must control the rates of priming and retrotransfer of the diketide. B) Aureothin biosynthesis is functional
in the absence of the N-terminus of AurA, and a DaurE mutant can be complemented by a synthetic SNAC derivative. C) Metabolic profiles of
a) S. lividans ZX1:pNT38 (DaurE mutant), b) DaurE mutant supplemented with PNBA, c) DaurE mutant supplemented with PNBA-SNAC,
d) DaurE mutant complemented with aurE (expression plasmid).
same time, the chain initiating PKS module that selects the p-
nitrobenzoate starter unit.[16,17] A prerequisite for the iterative
use of the first module is that the thiol group of the KS1
domain remains vacant. Otherwise, an occupied active site of
the KS domain loaded with the starter unit would hamper the
second elongation step. In consequence, the reconveyance of
the diketide intermediate must proceed faster than the
priming of the first module (Figure 1). For this reason, we
first investigated the loading of the aur PKS with the starter
unit. Because the aur PKS lacks a designated loading module,
we focused on the N-terminal part of the first module AurA
and the acyl-CoA ligase AurE, an enzyme that often assists in
loading a starter unit onto the PKS assembly line.[18] A mutant
lacking the gene for the acyl-CoA ligase was created by
excision of aurE from the aur gene cluster. Interestingly,
aureothin production was completely abolished in this DaurE
mutant. Genetic complementation fully restored aureothin
production, thus excluding any polar effects of the mutant.
Notably, we could also chemically restore aureothin biosyn-
thesis in the aurE mutant when adding the CoA surrogate, the
N-acetyl cysteamine thioester (SNAC) of p-nitro benzoic
acid.[19] This result clearly showed that AurE is solely involved
in starter unit activation, but does not participate in PKS
priming. To rule out the involvement of the N-terminal part of
AurA, we also tested the truncated mutant and found that the
N-terminus of AurA is not essential for priming the aureothin
PKS. These findings implied that the KS alone has full control
over priming and chain elongation events.
Next, we investigated the impact of the KS domains on the
iteration process. To exchange the KS1, two single restriction
sites flanking the gene region coding for the AT domain were
introduced in aurA. Using this construct, the gene region for
the KS1 was exchanged by the gene region coding for the
loading module and the KS1 domain of the avermectin
synthase (AVES1).[20–22] The newly generated hybrid module
was introduced into a mutant lacking wild-type aurA. HPLC-
MS analysis of the metabolic profile of the recombinant strain
revealed the formation of a new compound.
The metabolite of the ave/aur hybrid PKS, named
averpyrone, was isolated from large-scale fermentation
broth, and its structure was fully elucidated by HRMS, IR,
and both 1D and 2D NMR spectroscopy. For carbons 2–10,
the NMR data for averpyrone were quite similar to the ones
for deoxyaureothin. C11 (d 46.0) has a distinct chemical shift
compared to C11 (d 139.3) in deoxyaureothin,[23] thus
indicating a carbonyl next to this position. This is supported
by the coupling of H11 (d 3.34) as a quartet of doublets with
H11a and H10. No coupling with other protons could be
1
observed in the H NMR spectrum of H13 (d 2.05), which
suggests a connection to a quaternary carbon. The chemical
shift of C12 (d 203.1) is shifted to lower field. The HMBC
coupling of H11, H11a, and H13 with C12, along with the
deduced molecular composition, established the keto group
in position C12. Whereas averpyrone has been regioselec-
tively methylated by AurI,[24] it lacks the oxygen heterocycle
of aureothin. Apparently, 2 is not the correct substrate of the
multifunctional P-450 monooxygenase AurH, which usually
catalyzes hydroxylation–heterocyclization at C7/C9a.[23,25] As
it has been shown that AurH may install carbonyl groups into
non-natural substrates,[26–28] it is possible that it also intro-
duced the oxo group at C12 (Scheme 2).
Averpyrone features a 2-methylbutyryl side chain in lieu
of the p-nitrophenyl residue. To unequivocally prove the
origin of this moiety, 13C6-l-isoleucine was added to the
engineered hybrid PKS strain. MS analyses revealed a mass
shift of 5 Da that clearly showed the intact incorporation of
a 2-methylbutyrate unit that originates from l-isoleucine by
means of the branched chain alpha-keto acid dehydrogenase
complex (BCKD).[18] The structure of 2 and the labeling
experiments imply that 2-methylbutyl-CoA, which is loaded
onto the ave/aur hybrid PKS, undergoes only four rounds of
elongation. In the hybrid synthase, the number of modules
and the number of elongation steps are in perfect agreement
with the rule of colinearity. Clearly the first module had lost
its ability to catalyze two rounds of elongations. Conse-
quently, the genuine KS1 domain of AurA plays a crucial role
in the iteration event. The successful production of averpyr-
one has yet another implication. The size of its backbone and
the methyl branching pattern of the biosynthetic intermediate
produced by the hybrid module resemble the intermediate
produced by AurA. The (E)-2-methylpent-2-enoyl moiety is
accepted as a substrate by KS2 and all other downstream PKS
domains. We thus tested a range of alternative starter units
that are usually loaded onto the ave PKS, however, only the
branched building block was incorporated (data not shown).
It appears that the KS has specificity for substrates with
2
www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 1 – 6
These are not the final page numbers!

Angewandte
Chemie
Scheme 2. Loss of iteration in an engineered variant of AurA. Structure
of the hybrid metabolite averpyrone, and selected COSY and HMBC
correlations.
a particular substitution pattern in analogy to what has been
observed for KS from trans-AT systems.[29,30] To further
evaluate the potential gatekeeper function of the KS2, we
prepared SNAC analogues of polypropionate intermediates
with varying chain lengths and administered these probes to
the DaurA mutant.[31] Unfortunately, in all cases the synthetic
mimics were degraded. We also heterologously produced KS
domains for in vitro studies, but the soluble proteins were
unfortunately nonfunctional as standalone entities. Finally,
we chose to analyze the pathway intermediates in more detail
in vivo. Specifically, we compared intermediates produced by
AurA alone and those formed within the context of the entire
assembly line.
To promote a premature release of biosynthetic inter-
mediates, we deleted the C terminal thioesterase domain of
the aur PKS.[32–34] Initially, the mutant lacking the off-loading
machinery was cultured at 288C for 5 days, after which no
production of any biosynthetic intermediate was observed. To
rule out degradation of the biosynthetic intermediates, we
systematically varied the cultivation times. By LC-HRMS
monitoring of the 12–24 h old main culture, we were able to
detect trace amounts of metabolites. The deduced molecular
composition corresponded well with the expected biosyn-
thetic intermediates of modules 1 and 2 (Scheme 1). Only the
Figure 2. Dissection of the aur PKS and analysis of polyketide inter-
mediates by LC-HRMS; extracted ion masses are color coded to match
the synthetic references. A) Prematurely off-loaded intermediates
detected from a mutant lacking the thioesterase (TE) domain. B) Mul-
tiple products of the heterologously produced, freestanding iterative
module (AurA); multiple peaks refer to E/Z isomers.
Angew. Chem. Int. Ed. 2013, 52, 1 – 6
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

.
Angewandte
Communications
predicted intermediate of module three was not observed;
this is likely due to the instability of the b-keto acid. As
isolation of these products was not feasible owing to the low
quantities produced, we synthesized authentic references of
the postulated aureothin intermediates (see the Supporting
Information) and compared their retention times with the
observed metabolites. All proposed intermediates of AurA
and AurB could be clearly identified. However, no metabo-
lites with shorter or longer backbones were produced.
Next, we investigated the polyketides generated by the
iteratively operating module AurA alone. For this purpose,
aurA and aurE were cloned into an expression vector and
introduced into S. lividans ZX1. The resulting strain was
cultured with supplementation of PNBA, and the metabolic
profile was again monitored by LC-HRMS. As before, we
detected molecular masses that correspond to the di- and
triketide intermediates. Surprisingly, we also observed masses
that pointed to polyketides resulting from three and four
rounds of elongations (Figure 2). To rigorously determine the
identity of these compounds, we also prepared authentic
references for these metabolites and compared their masses
and retention times with the observed products by LC-
HRMS. Multiple peaks indicated E/Z-isomerizations, which
readily take place when exposing polyunsaturated polypro-
pionates to daylight.[35]
The unexpected finding of polyketide metabolites of
AurA that have undergone up to four chain elongations
indicates that the first module possesses an intrinsic ability to
work iteratively and that the repetitive usage of the first
module is not a result of its context within a multimodular
PKS. Furthermore, these results demonstrate that the first
module is, in principle, able to perform more rounds of
iteration than what is observed within the context of the
entire assembly line. Even though the triene and tetraene
products are formed only in minute amounts, it should be
highlighted that they cannot be observed in the intact
assembly line. Thus, the KS2 has specificity for a defined
chain length and readily processes the correct diene inter-
mediate, thus preventing further elongation rounds. Taken
together, the KS1 clearly mediates iteration, but the down-
stream module (KS2) serves as a gatekeeper to control the
correct assembly of the aureothin polyketide backbone
(Scheme 3).
Scheme 3. Model for the multifactorial control of non-colinear chain
elongations in aureothin biosynthesis.
provide strong experimental support for a scenario where
multiple factors are responsible for the rare iteration process.
It is clearly an inherent quality of the KS1 to mediate priming,
chain propagation, and intermediate retrotransfer, but the
downstream KS2 functions as a gatekeeper that excludes the
propagation of intermediates that are either too short or too
long. Beyond new insights into the hidden program of
a noncanonical multimodular assembly line, our findings
also provide valuable information for the rational design of
complex polyketides by pathway engineering.
Received: February 14, 2013
Published online: && &&, &&&&
Keywords: aureothin · biosynthesis · polyene ·
.
polyketide synthase · Streptomyces thioluteus
[1] C. Hertweck, Angew. Chem. 2009, 121, 4782 – 4811; Angew.
Chem. Int. Ed. 2009, 48, 4688 – 4716.
[2] S. J. Moss, C. J. Martin, B. Wilkinson, Nat. Prod. Rep. 2004, 21,
575 – 593.
[3] K. J. Weissman, P. F. Leadlay, Nat. Rev. Microbiol. 2005, 3, 925 –
936.
[4] D. H. Sherman, Nat. Biotechnol. 2005, 23, 1083 – 1084.
[5] C. Khosla, S. Kapur, D. E. Cane, Curr. Opin. Chem. Biol. 2009,
13, 135 – 143.
[6] S. Kushnir, U. Sundermann, S. Yahiaoui, A. Brockmeyer, P.
Janning, F. Schulz, Angew. Chem. 2012, 124, 10820 – 10825;
Angew. Chem. Int. Ed. 2012, 51, 10664 – 10669.
In summary, we have gained first insights into the factors
governing the iteration processes in a non-colinear PKS.
Through gene deletion and complementation with a synthetic
surrogate, we found that the KS1 of the iterating module is
exclusively in charge of priming the PKS. This is a prerequisite
to permit a vacant position for the retrotransfer of the
diketide. The model of kinetic control is fully in line with the
observed complete loss of iteration in an engineered AurA
variant harboring the designated loading module of the
avermectin PKS. Finally, through deletion of the TE domain
and the use of synthetic reference compounds, we succeeded
in detecting the polyketide intermediates of iterative and non-
iterative aur PKS modules. Whereas this experiment con-
firmed a high fidelity of the programmed events, the iterative
module alone produces polyketides with varying lengths that
correspond to one to four rounds of elongations. Thus, we
[7] G. L. Challis, J. Med. Chem. 2008, 51, 2618 – 2628.
[8] J. M. Winter, S. Behnken, C. Hertweck, Curr. Opin. Chem. Biol.
2011, 15, 22 – 31.
[9] N. Gaitatzis, B. Silakowski, B. Kunze, G. Nordsiek, H. Blçker, G.
Hçfle, R. Mꢀller, J. Biol. Chem. 2002, 277, 13082 – 13090.
[10] J. He, C. Hertweck, Chem. Biol. 2003, 10, 1225 – 1232.
[11] C. Olano, B. Wilkinson, S. J. Moss, A. F. Brana, C. Mendez, P. F.
Leadlay, J. A. Salas, Chem. Commun. 2003, 2780 – 2782.
[12] J. He, C. Hertweck, ChemBioChem 2005, 6, 908 – 912.
[13] B. Busch, N. Ueberschaar, Y. Sugimoto, C. Hertweck, J. Am.
Chem. Soc. 2012, 134, 12382 – 12385.
[14] S. Kapur, B. Lowry, S. Yuzawa, S. Kenthirapalan, A. Y. Chen,
D. E. Cane, C. Khosla, Proc. Natl. Acad. Sci. USA 2012, 109,
4110 – 4115.
[15] B. Busch, C. Hertweck, Phytochemistry 2009, 70, 1833 – 1840.
[16] J. He, C. Hertweck, J. Am. Chem. Soc. 2004, 126, 3694 – 3695.
4
www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 1 – 6
These are not the final page numbers!

Angewandte
Chemie
[17] M. Ziehl, J. He, H.-M. Dahse, C. Hertweck, Angew. Chem. 2005,
117, 1226 – 1230; Angew. Chem. Int. Ed. 2005, 44, 1202 – 1205.
[18] B. S. Moore, C. Hertweck, Nat. Prod. Rep. 2002, 19, 70 – 99.
[19] M. Werneburg, B. Busch, J. He, M. E. Richter, L. Xiang, B. S.
Moore, M. Roth, H. M. Dahse, C. Hertweck, J. Am. Chem. Soc.
2010, 132, 10407 – 10413.
[28] M. Richter, B. Busch, K. Ishida, B. S. Moore, C. Hertweck,
ChemBioChem 2012, 13, 2196 – 2199.
[29] T. Nguyen, K. Ishida, H. Jenke-Kodama, E. Dittmann, C.
Gurgui, T. Hochmuth, S. Taudien, M. Platzer, C. Hertweck, J.
Piel, Nat. Biotechnol. 2008, 26, 225 – 233.
[30] M. Jenner, S. Frank, A. Kampa, C. Kohlhaas, P. Pçplau, G. S.
Briggs, J. Piel, N. J. Oldham, Angew. Chem. 2013, 125, 1181 –
1185; Angew. Chem. Int. Ed. 2013, 52, 1143 – 1147.
[31] N. Traitcheva, H. Jenke-Kodama, J. He, E. Dittmann, C.
Hertweck, ChemBioChem 2007, 8, 1841 – 1849.
[32] J. Moldenhauer, X.-H. Chen, R. Borriss, J. Piel, Angew. Chem.
2007, 119, 8343 – 8345; Angew. Chem. Int. Ed. 2007, 46, 8195 –
8197.
[33] B. Kusebauch, B. Busch, K. Scherlach, M. Roth, C. Hertweck,
Angew. Chem. 2009, 121, 5101 – 5104; Angew. Chem. Int. Ed.
2009, 48, 5001 – 5004.
[20] H. Ikeda, T. Nonomiya, M. Usami, T. Ohta, S. Omura, Proc.
Natl. Acad. Sci. USA 1999, 96, 9509 – 9514.
[21] A. F. A. Marsden, B. Wilkinson, J. Cortꢁs, N. J. Dunster, J.
Staunton, P. F. Leadlay, Science 1998, 279, 199 – 202.
[22] L. S. Sheehan, R. E. Lill, B. Wilkinson, R. M. Sheridan, W. A.
Vousden, A. L. Kaja, G. D. Crouse, J. Gifford, P. R. Graupner, L.
Karr, P. Lewer, T. C. Sparks, P. F. Leadlay, C. Waldron, C. J.
Martin, J. Nat. Prod. 2006, 69, 1702 – 1710.
[23] J. He, M. Mꢀller, C. Hertweck, J. Am. Chem. Soc. 2004, 126,
16742 – 16743.
[24] M. Mꢀller, J. He, C. Hertweck, ChemBioChem 2006, 7, 37 – 39.
[25] M. E. A. Richter, N. Traitcheva, U. Knꢀpfer, C. Hertweck,
Angew. Chem. 2008, 120, 9004 – 9007; Angew. Chem. Int. Ed.
2008, 47, 8872 – 8875.
[26] G. Zocher, M. E. Richter, U. Mueller, C. Hertweck, J. Am.
Chem. Soc. 2011, 133, 2292 – 2302.
[34] B. Kusebauch, B. Busch, K. Scherlach, M. Roth, C. Hertweck,
Angew. Chem. 2010, 122, 1502 – 1506; Angew. Chem. Int. Ed.
2010, 49, 1460 – 1464.
[35] M. Mꢀller, B. Kusebauch, G. Liang, C. M. Beaudry, D. Trauner,
C. Hertweck, Angew. Chem. 2006, 118, 7999 – 8002; Angew.
Chem. Int. Ed. 2006, 45, 7835 – 7838.
[27] M. Henrot, M. E. Richter, J. Maddaluno, C. Hertweck, M.
De Paolis, Angew. Chem. 2012, 124, 9725 – 9729; Angew. Chem.
Int. Ed. 2012, 51, 9587 – 9591.
Angew. Chem. Int. Ed. 2013, 52, 1 – 6
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!

.
Angewandte
Communications
Communications
Aureothin Polyketide Synthase
Freedom and control: First insights into
the rare programmed iteration of an
individual polyketide synthase (PKS)
module were obtained from the analysis
and mutation of aureothin (1) synthase.
The first ketosynthase (KS) domain
primes the PKS, allowing intermediate
retrotransfer. Addition of a designated
loading module results in a complete loss
of iteration. The downstream KS func-
tions as a gatekeeper for correct chain
length.
B. Busch, N. Ueberschaar, S. Behnken,
Y. Sugimoto, M. Werneburg,
N. Traitcheva, J. He,
C. Hertweck*
&&&& — &&&&
Multifactorial Control of Iteration Events
in a Modular Polyketide Assembly Line
6
www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 1 – 6
These are not the final page numbers!