Enzymatic Polyketide Chain Branching to Give Substituted Lactone, Lactam, and Glutarimide Heterocycles

Article abstract of DOI:10.1002/anie.201407282

Polyketides typically result from head-to-tail condensation of acyl thioesters to produce highly functionalized linear chains. The biosynthesis of the phytotoxin rhizoxin, however, involves a polyketide synthase (PKS) module that introduces a δ-lactone ch

Full text of DOI:10.1002/anie.201407282

Angewandte
Chemie
DOI: 10.1002/anie.201407282
Polyketide Biosynthesis
Enzymatic Polyketide Chain Branching To Give Substituted Lactone,
Lactam, and Glutarimide Heterocycles**
Daniel Heine, Tom Bretschneider, Srividhya Sundaram, and Christian Hertweck*
Dedicated to Professor Heinz G. Floss on the occasion of his 80th birthday
Abstract: Polyketides typically result from head-to-tail con-
densation of acyl thioesters to produce highly functionalized
linear chains. The biosynthesis of the phytotoxin rhizoxin,
however, involves a polyketide synthase (PKS) module that
introduces a d-lactone chain branch through Michael addition
of a malonyl extender to an a,b-unsaturated intermediate unit.
To evaluate the scope of the branching module, polyketide
mimics were synthesized and their biotransformation by the
reconstituted PKS module from the Rhizopus symbiont
Burkholderia rhizoxinica was monitored in vitro. The impact
of the type and configuration of the d-substituents was probed
and it was found that amino-substituted surrogates yield the
corresponding lactams. A carboxamide analogue was trans-
formed into a glutarimide unit, which can be found in many
natural products. Our findings illuminate the biosynthesis of
Figure 1. A) The structure of rhizoxin (1) and a model for vinylogous
chain branching catalyzed by a specialized PKS module; 1) Michael
addition, 2) lactonization. B) The structures of isomigrastatin (2) and
glutarimide-bearing polyketides and also demonstrate the
utility of this branching module for synthetic biology.
cycloheximide (3) with the glutarimide motif highlighted. KS, keto-
synthase; B, branching domain; ACP, acyl carrier domain; AT, acyl-
transferase.
P
olyketides encompass an impressive range of natural
products, many of which are in clinical use. Their structural
diversity primarily emerges from simple head-to-tail conden-
sations of activated acyl and malonyl building blocks. The
resulting polyketide chains are enzymatically processed and
may undergo enzymatic tailoring, for example, cyclization,
oxygenation, halogenation, and glycosylation.[1] With respect
to the assembly of the carbon backbone, it is remarkable that
only a few variations to the unidirectional 1,2-condensation
scheme have been observed. However, noncanonical sub-
structures can be crucial for the biological activity of the
polyketide, as exemplified by the remarkably potent anti-
mitotic agent rhizoxin (1, Figure 1).[2] This phytotoxin is
employed in a symbiosis of the plant-pathogenic fungus
Rhizopus microsporus and its bacterial endosymbiont, Bur-
kholderia rhizoxinica.[3] The structure of the macrolide is
highly unusual because it features a d-lactone ring that is
tethered to the polyketide backbone. Notably, this chain
branch is essential for the binding of rhizoxin to the b-tubulin
subunit,[4] through which it efficiently halts cell division in the
picomolar range.[3b] Isotope-labeling experiments have
showed that the d-lactone ring integrates an acetyl unit that
branches off at a b-position relative to a former acetyl
carbonyl in the polyketide chain (Figure 1).[5]
Whereas alkyl substitutions at a-positions result from
methylation or the incorporation of substituted malonyl
units,[6] branches at b-positions typically involve enzymes
that resemble those of the isoprenoid enzymatic machinery.[7]
Surprisingly, genes for such components could not be found in
the genome of the rhizoxin producer.[8] Instead, the architec-
ture of the rhizoxin (rhi) type I polyketide synthase (PKS)[9]
and the structures of isolated pathway intermediates[10] have
suggested that the chain branch is introduced through
a Michael addition. To shed light on the enzyme mechanism,
we reconstituted the designated “branching module” of the
rhi PKS, which consists of ketosynthase (KS), branching (B),
and acyl carrier (ACP) domains, and a trans-acting acyl
transferase (AT). By means of an in vitro enzyme assay, we
successfully transformed a synthetic surrogate of the poly-
ketide intermediate into the corresponding d-lactone deriv-
ative. We thus confirmed that the pharmacophoric side chain
[*] D. Heine, T. Bretschneider, S. Sundaram, Prof. Dr. C. Hertweck
Department of Biomolecular Chemistry, Leibniz Institute for
Natural Product Research and Infection Biology (HKI)
Beutenbergstr. 11a, 07745 Jena (Germany)
E-mail: christian.hertweck@hki-jena.de
Homepage: http://www.hki-jena.de
Prof. Dr. C. Hertweck
Chair for Natural Product Chemistry, Friedrich Schiller University
Jena (Germany)
[**] We thank A. Perner and H. Heinecke for MS and NMR measure-
ments. Financial support by the Studienstiftung des deutschen
Volkes for a fellowship to D.H. is gratefully acknowledged.
Supporting information (including experimental details) for this
article is available on the WWW under http://dx.doi.org/10.1002/
anie.201407282.
Angew. Chem. Int. Ed. 2014, 53, 11645 –11649
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
11645

.
Angewandte
Communications
is introduced through a vinylogous addition of a malonyl unit
to an a,b-unsaturated thioester, followed by lactonization
(Figure 1).[11] Notably, similar PKS modules are encoded in
the biosynthetic gene clusters for glutarimide-bearing natural
products such as isomigrastatin (2),[12] an important cell
migration inhibitor (Figure 1). As in the structurally related
antibiotic cycloheximide (3), the glutarimide residue is
important for the biological activity of these compounds.[13]
Although it had been proposed that an enzymatic Michael
addition could give rise to the glutarimide structure, the roles
of these PKS modules have remained elusive and the
proposed mechanism still needs to be experimentally con-
firmed. Herein, we show that the chain-branching rhi PKS
module is more versatile than expected. Through an in vitro
multienzyme assay with synthetic polyketide mimics, we have
demonstrated that the unusual KS–B–ACP module is also
capable of forming lactams and even glutarimides.
To test the substrate specificity of the rhizoxin PKS
branching module and to explore the scope of potential
applications, we synthesized various polyketide analogues for
in vitro multienzyme biotransformation experiments. Specif-
ically, a series of d-functionalized, a,b-unsaturated N-acetyl
cysteamine thioester (SNAC) derivatives that mimic the
activated intermediates were prepared. First, we explored the
impact of the configuration of the d-hydroxy group on the
course of the reaction. For this purpose, we synthesized the R
and S enantiomers of the 5-hydroxyhexenoic acid SNAC
thioester from the corresponding protected hydroxybuta-
noates through a reduction–olefination sequence. The ste-
reoisomers 4 and 5 were individually subjected to the in vitro
enzyme assay, and the products were analyzed by HPLC–
HRMS, with the synthetic d-lactone as reference (Figure 2).
Surprisingly, in both cases formation of the branched product
was detected in only one diastereoisomeric form. From
retention-time comparisons and deductions from the natural
biosynthetic pathway, we inferred that in both cases the syn-
substituted lactone was formed.
To shed light on the preferences of the KS, we determined
the kinetic parameters of the branching reaction. For the R
enantiomer, we found a KM value of (2694 Æ 627) mm and
a vmax value of (20.9 Æ 2.4) mm minÀ1. By contrast, using the S
enantiomer gave a KM value of (1348 Æ 315) mm and a vmax
value of (2.1 Æ 0.3) mm minÀ1. Binding of the non-natural R
enantiomer to the KS is thus only lowered twofold, whereas
the transformation rate of this substrate is ten times lower
compared to the S enantiomer. This result indicates that the
KS basically does not differentiate between the configura-
tions of the d-hydroxy groups, but lactone ring formation is
drastically reduced when the non-natural isomer is applied. It
appears that this sequence of Michael addition and lactoniza-
tion exclusively yields syn-substituted lactones and that the
configuration of the hydroxy group determines the stereo-
chemical course of the reaction.
Figure 2. Vinylogous chain branching of pure R and S enantiomers of
the N-acetyl cysteamine thioester of 5-hydroxyhexenoic acid, and
subsequent lactone formation. A) Michaelis–Menten kinetics of the
biotransformation. B) MALDI analysis of ACP-bound products (peak in
middle relates to decarboxylated malonylated ACP species). C) SIM-
LC–HRMS analysis of hydrolyzed products, and comparison with the
racemic synthetic reference. SIM=selected ion monitoring.
bond. Furthermore, a phenol analogue (10) was prepared
(Figure 3). The activated malonyl units and the rhizoxin
polyketide chain mimics were individually added to the
enzyme mixture that constitutes the functional PKS module
in vitro. The reactions were monitored by MALDI analysis of
the ACP with potentially bound products, as well as by high-
resolution mass spectrometry (HRMS) of the hydrolyzed
products. Unsurprisingly, the saturated alcohol 8 did not
undergo any chain branching reaction, thus demonstrating
once again that the Michael addition precedes ester bond
formation. Furthermore, this experiment showed that the KS
does not catalyze Claisen condensation with this substrate
when a vinylogous attack is hampered. Changing the aliphatic
alcohol to a phenol nucleophile did not give rise to the
hypothetical product 11. These results indicate a high degree
of specificity for the KS–B–ACP module.
Next, we explored the possibility of replacing the nucle-
ophilic d-hydroxy substituent of the polyketide chain. It
should be highlighted that this OH group is required for the
release of the KS-bound intermediate by lactonization, which
allows the propagation of the intermediate. Nonetheless, it is
conceivable that other nucleophilic residues could also cleave
the thioester bond, which would result in the formation of
To verify the proposed course of the reaction and to
corroborate the idea that non-covalent interactions between
the substrate and the malonyl unit define the stereochemical
course of the reaction, we synthesized and tested the d-
hydroxy-substituted dihydro surrogate 8, which cannot
undergo a branching reaction because of the missing double
11646 www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 11645 –11649

Angewandte
Chemie
Figure 3. Probing the substrate tolerance of the branching module
towards dihydro and phenol variants by using synthetic polyketide
surrogates.
alternative heterocyclic systems. If the surrogate was accepted
as a substrate for the Michael addition, aminolysis of the
thioester would give a lactam ring. To evaluate the effect of an
amino group in lieu of the hydroxy group, we synthesized the
corresponding d-amino-substituted surrogate of the natural
polyketide precursor. To this end, N-Boc-protected ethyl d,l-
3-aminobutanoate was reduced with DibalH. The amino-
aldehyde was immediately subjected to a Horner–Wads-
worth–Emmons (HWE) olefination, and subsequent depro-
tection gave the desired amine 12 (see the Supporting
Information). This synthetic surrogate of the growing poly-
ketide chain was employed in the in vitro assay and product
formation was monitored by HRMS. The formation of a new
compound with m/z = 172.0966 and a molecular formula of
C8H14O3N [M+H]+ as deduced from LC–HRMS suggested
that the desired lactam compound was produced. To confirm
the identity of the lactam, we prepared a synthetic reference.
In brief, a protected d,l-b-aminoacid ester was subjected to
Figure 4. In vitro biotransformation of an amino-substituted polyketide
surrogate into a lactam moiety. A) General Scheme of the enzyme
assay. B) MALDI analysis of the ACP-bound product. C) SIM-LC–
HRMS analysis of the hydrolyzed product and comparison with the
synthetic reference (as a mixture of diastereoisomers); only one
enantiomer is shown.
desired polyketide surrogates 14 and 16 (see the Supporting
Information) through alkene metathesis of propenyl SNAC
with but-3-enoic acid and but-3-enamide, respectively. Addi-
tion of the carboxy-substituted SNAC derivative 14 to the
enzyme assay resulted in protein precipitation, and no
biotransformation took place. To exclude the possibility of
a pH effect, we repeated the experiment after accurately
adjusting the pH value to 7.0 with Tris buffer. However,
neither the respective anhydride 15 nor the hydrolyzed
product could be detected.
By contrast, the carboxamide-substituted polyketide sur-
rogate 16 was accepted as a substrate by the branching
module. Through analysis of the enzyme mixture, we detected
a MALDI signal corresponding to the ACP-bound glutari-
mide product (Figure 5). Moreover, acidic workup released
the expected product with m/z = 172.0610 and a molecular
formula of C7H10O4N (HRMS). The identity of the glutari-
mide 17 was rigorously confirmed through comparison of the
HPLC retention times and high-resolution mass signals of the
reaction product and a synthetic reference (Figure 5). A
negative control experiment with a heat-inactivated KS–B
didomain confirmed that the reaction is indeed enzyme-
catalyzed.
a
reductive Horner–Wadsworth–Emmons olefination to
obtain the substituted a,b-unsaturated ester. A cesium
carbonate catalyzed Michael addition of diethylmalonate
followed by hydrolysis yielded the lactam as a diastereoiso-
meric mixture (see the Supporting Information). Comparison
of the high-resolution mass spectra and HPLC retention times
of the enzyme products and the synthetic reference confirmed
the successful conversion of 12 into the d-lactam 13
(Figure 4). Furthermore, only the (late-eluting) syn-diaste-
reomer is formed, thus confirming that the enzymatic reaction
is highly selective, yielding only one diastereoisomer from
a mixture of enantiomeric d-amino-substituted precursors, as
has been observed for the d-lactones.
The successful biotransformation of the amino analogue
encouraged an expansion of this approach to carbonyl
substituents such as carboxylic acid and amide residues. In
principle, a carboxylate could give rise to an anhydride moiety
or the respective dicarboxylic acid, whereas the carboxyamide
would lead to the glutarimide moiety, which is found in many
natural products. Unexpectedly, the synthesis of carboxy- and
carboxamide-substituted SNAC derivatives proved to be
highly challenging. Despite many attempts, the functionaliza-
tion of glutaconic acid and HWE olefination were not
feasible, possibly owing to the highly reactive a-proton at
the allylic position. Finally, we succeeded in preparing the
Our results unequivocally show that the unusual KS–B–
ACP module of the rhi PKS is not limited to lactone
formation but is also capable of introducing branches that
Angew. Chem. Int. Ed. 2014, 53, 11645 –11649
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 11647

.
Angewandte
Communications
In conclusion, we have shed light on the scope and
versatility of a specialized PKS module that catalyzes the
noncanonical vinylogous addition of a C2 extender unit to
generate polyketide chain branches. Through the synthesis of
a series of thioester surrogates with different d-substituents,
in vitro biotransformation experiments, and product analyses
with synthetic references, we showed that the branching
module generates lactone, lactam, and glutarimide hetero-
cycles. We found that this set of domains exhibits a marked
stereochemical substrate preference and that exclusively syn-
substituted lactones and lactams are formed. However, there
is a degree of flexibility that leaves room for various
alternative conversions. Among these, the enzymatic forma-
tion of glutarimide provides first experimental support for
postulated biosynthetic pathways to known glutarimide-
bearing natural products. Although corresponding d-lactam
substructures have not yet been observed in natural poly-
ketide structures, we postulate that the biosynthesis of
polyketides with d-lactam branches would be feasible and
that such natural products await discovery. From a practical
point of view, our findings are an important addition to our
current understanding of non-terpenoid polyketide branch-
ing. Furthermore, the use of this branching module in
synthetic biology or the engineering of enzymatic pathways
could provide unique opportunities for compound diversifi-
cation.
Received: July 16, 2014
Published online: September 11, 2014
Keywords: beta-branching · biotransformation · macrolides ·
.
Michael addition · polyketide synthases
[1] C. Hertweck, Angew. Chem. Int. Ed. 2009, 48, 4688 – 4716;
Angew. Chem. 2009, 121, 4782 – 4811.
[2] J. Hong, J. D. White, Tetrahedron 2004, 60, 5653 – 5681.
[3] a) L. P. Partida-Martinez, C. Hertweck, Nature 2005, 437, 884 –
888; b) K. Scherlach, L. P. Partida-Martinez, H.-M. Dahse, C.
Hertweck, J. Am. Chem. Soc. 2006, 128, 11529 – 11536; c) L. P.
Partida-Martinez, I. Groth, M. Roth, I. Schmitt, K. Buder, C.
Hertweck, Int. J. Syst. Evol. Microbiol. 2007, 57, 2583 – 2590;
d) K. Scherlach, B. Busch, G. Lackner, U. Paszkowski, C.
Hertweck, Angew. Chem. Int. Ed. 2012, 51, 9615 – 9618;
Angew. Chem. 2012, 124, 9753 – 9756.
[4] a) M. Takahashi, S. Iwasaki, H. Kobayashi, S. Okuda, T. Murai,
Y. Sato, Biochim. Biophys. Acta Gen. Subj. 1987, 926, 215 – 223;
b) I. Schmitt, L. P. Partida-Martinez, R. Winkler, K. Voigt, A.
Einax, J. Wçstemeyer, C. Hertweck, ISME J. 2008, 2, 632 – 641;
c) B. Kusebauch, K. Scherlach, H. Kirchner, H. M. Dahse, C.
Hertweck, ChemMedChem 2011, 6, 1998 – 2001.
Figure 5. In vitro biotransformation of the amide-functionalized poly-
ketide surrogate into a glutarimide moiety. A) General Scheme of the
enzyme assay. B) MALDI analysis of the ACP-bound product. C) SIM-
LC–HRMS analysis of hydrolyzed product and comparison with the
reference. D) Proposed model for the biosynthesis of glutarimide-
bearing polyketides such as 2 and 3 (see Figure 1).
constitute lactam and glutarimide moieties. The latter are
particularly noteworthy since related modules (KS–“X”–
ACP, or KS–B–ACP) were found in PKS complexes involved
in the biosynthesis of the glutarimide-bearing polyketides
such as 9-methylstreptimidone,[14] isomigrastatin[12] and cyclo-
heximide.[15] The successful enzymatic synthesis of a glutari-
mide with the rhi PKS module for the first time confirms the
proposed pathway involving Michael addition and hetero-
cyclization.
[5] H. Kobayashi, S. Iwasaki, E. Yamada, S. Okuda, J. Chem. Soc.
Chem. Commun. 1986, 1702 – 1703.
[6] a) Y. A. Chan, A. M. Podevels, B. M. Kevany, M. G. Thomas,
Nat. Prod. Rep. 2009, 26, 90 – 114; b) M. C. Wilson, B. S. Moore,
Nat. Prod. Rep. 2012, 29, 72 – 86.
[7] a) C. T. Calderone, W. E. Kowtoniuk, N. L. Kelleher, C. T.
Walsh, P. C. Dorrestein, Proc. Natl. Acad. Sci. USA 2006, 103,
8977 – 8982; b) C. T. Calderone, Nat. Prod. Rep. 2008, 25, 845 –
853.
[8] a) G. Lackner, N. Moebius, L. P. Partida-Martinez, C. Hertweck,
J. Bacteriol. 2011, 193, 783 – 784; b) G. Lackner, N. Moebius, L. P.
11648 www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 11645 –11649

Angewandte
Chemie
Partida-Martinez, S. Boland, C. Hertweck, BMC Genomics
2011, 12, 210.
[9] a) L. P. Partida-Martinez, C. Hertweck, ChemBioChem 2007, 8,
41 – 45; b) 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.
502, 124 – 128; related highlight articles: b) C. A. Townsend,
Nature 2013, 502, 44 – 45, c) R. J. Cox, ChemBioChem 2014, 15,
27 – 29.
[12] S. K. Lim, J. Ju, E. Zazopoulos, H. Jiang, J. W. Seo, Y. Chen, Z.
Feng, S. R. Rajski, C. M. Farnet, B. Shen, J. Biol. Chem. 2009,
284, 29746 – 29756.
[10] a) B. Kusebauch, B. Busch, K. Scherlach, M. Roth, C. Hertweck,
Angew. Chem. Int. Ed. 2009, 48, 5001 – 5004; Angew. Chem.
2009, 121, 5101 – 5104; b) B. Kusebauch, B. Busch, K. Scherlach,
M. Roth, C. Hertweck, Angew. Chem. Int. Ed. 2010, 49, 1460 –
1464; Angew. Chem. 2010, 122, 1502 – 1506.
[13] S. R. Rajski, B. Shen, ChemBioChem 2010, 11, 1951 – 1954.
[14] B. Wang, Y. Song, M. Luo, Q. Chen, J. Ma, H. Huang, J. Ju, Org.
Lett. 2013, 15, 1278 – 1281.
[15] M. Yin, Y. Yan, J. R. Lohman, S. X. Huang, M. Ma, G. R. Zhao,
L. H. Xu, W. Xiang, B. Shen, Org. Lett. 2014, 16, 3072 – 3075.
[11] a) T. Bretschneider, J. B. Heim, D. Heine, R. Winkler, B. Busch,
B. Kusebauch, T. Stehle, G. Zocher, C. Hertweck, Nature 2013,
Angew. Chem. Int. Ed. 2014, 53, 11645 –11649
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 11649