Synthesis of (R)-mellein by a partially reducing iterative polyketide synthase

Article abstract of DOI:10.1021/ja304905e

Mellein and the related 3,4-dihydroisocoumarins are a family of natural products with interesting biological properties. The mechanisms of dihydroisocoumarin biosynthesis remain largely speculative today. Here we report the synthesis of mellein by a partially reducing iterative polyketide synthase (PR-PKS) as a pentaketide product. Remarkably, despite the head-to-tail homology shared with several fungal and bacterial PR-PKSs, the mellein synthase exhibits a distinct keto reduction pattern in the synthesis of the pentaketide. We present evidence to show that the ketoreductase (KR) domain alone is able to recognize and differentiate the polyketide intermediates, which provides a mechanistic explanation for the programmed keto reduction in these PR-PKSs.

Full text of DOI:10.1021/ja304905e

Communication
pubs.acs.org/JACS
Synthesis of (R)-Mellein by a Partially Reducing Iterative Polyketide
Synthase
Huihua Sun,† Chun Loong Ho,† Feiqing Ding,‡ Ishin Soehano,† Xue-Wei Liu,‡ and Zhao-Xun Liang*,†
†School of Biological Sciences, Nanyang Technological University, 60 Nanyang Drive, Singapore 637551
‡School of Mathematics and Physics, Nanyang Technological University, 21 Nanyang Link, Singapore 637371
S
* Supporting Information
dihydroisocoumarin.4,5 An immunomodulatory effect of
ABSTRACT: Mellein and the related 3,4-dihydroisocou-
marins are a family of natural products with interesting
biological properties. The mechanisms of dihydroisocou-
marin biosynthesis remain largely speculative today. Here
we report the synthesis of mellein by a partially reducing
iterative polyketide synthase (PR-PKS) as a pentaketide
product. Remarkably, despite the head-to-tail homology
shared with several fungal and bacterial PR-PKSs, the
mellein synthase exhibits a distinct keto reduction pattern
in the synthesis of the pentaketide. We present evidence to
show that the ketoreductase (KR) domain alone is able to
recognize and differentiate the polyketide intermediates,
which provides a mechanistic explanation for the
programmed keto reduction in these PR-PKSs.
ochratoxin A on a human macrophage cell line has been
established. Bacilosarcin A (7) originating from a marine
octocoral-associated bacterium exhibits potent antiplasmodial
activity.6,7 Hydrangenol (8), phyllodulcin (9) and thunberginol
A (10) are structurally related dihydroisocoumarins isolated
from the leaves of Hydrangea macrophylla var. thunbergii that
promote adipogenesis of 3T3-L1 cells and exhibit antidiabetic
properties.8,9 AI-77-B (11) isolated from the culture broth of
Bacillus pumilus AI-77 is endowed with gastroprotective and
antiulcerogenic properties without exhibiting adverse effects on
the central nervous system.10,11 However, despite the fact that
some of the naturally occurring dihydroisocoumarins have been
known for many decades, the enzymes responsible for the
biosynthesis of these compounds remain to be identified. It has
been proposed that 6-methoxymellein (4) and ochratoxin A are
synthesized by polyketide synthases (PKSs), but the sequence
and domain composition of the PKSs have not been fully
established.12−15 The only system characterized to date is the
type-I modular PKS responsible for the synthesis of the 7-
methylated dihydroisocoumarin ring of ajudazols.16
ellein (3,4-dihydro-8-hydroxy-3-methylisocoumarin, 1)
Mand other dihydroisocoumarins are widespread secon-
dary metabolites isolated from various organisms. With their
notable biological activities, some of the dihydroisocoumarins
are being considered as drug leads (see Figure 1).1 Mellein,
isolated from fungi and insect secretions (as a trail pheromone)
and first discovered in 1933, exhibits fungicidal, antibacterial,
and HCV protease-inhibitory properties.2,3 Ochratoxin A (3) is
a well-known fungal toxin that contains a chlorinated
Saccharopolyspora erythraea is a mycelium-forming actino-
mycete that produces the macrolide antibiotic erythromycin A.
The modular PKS responsible for the biosynthesis of
erythromycin A has been a model system for studying
polyketide synthesis. Sequencing of the genome of S. erythraea
NRRL23338 revealed a dozen additional PKS genes.17 The
products of many of these PKSs remain unidentified despite
extensive fermentation experiments. One of the uncharacterized
orphan PKS genes is SACE5532 (or pks8), which encodes a
single-module PKS that shares head-to-tail sequence homology
with several fungal and bacterial type-I partially reducing
iterative PKSs (PR-PKSs) for aromatic polyketide biosynthesis
(Figure 2a and Figure S1 in the Supporting Information).
Among these PR-PKSs, ChlB1, MdpB, and PokM1 synthesize
the 6-methylsalicyclic acid (6-MSA) moieties of chlorothricin,
maduropeptin and pactamycin, respectively;18−20 NcsB synthe-
sizes the 2-hydroxy-5-methyl-NPA (NPA = 1-naphthoic acid)
moiety of neocarzinostatin;21 and AziB synthesizes the 5-
methyl-NPA moiety of azinomycin B.22 Accordingly,
SACE5532 is predicted to contain a ketosynthase (KS), an
acyltransferase (AT), a thioester hydrolase (TH),23
a
ketoreductase (KR) domain, and an acyl carrier protein
(ACP) domain.
Figure 1. Isocoumarin, mellein, and representatives of naturally
occurring 3,4-dihydroisocoumarins.
Received: May 21, 2012
Published: July 13, 2012
© 2012 American Chemical Society
11924
dx.doi.org/10.1021/ja304905e | J. Am. Chem. Soc. 2012, 134, 11924−11927

Journal of the American Chemical Society
Communication
Figure 3. Production and characterization of the enzymatic product of
SACE5532. (a) HPLC analysis of the formation of the enzymatic
product by in vitro enzymatic reaction and in vivo coexpression of
Sfp/SACE5532. The structure of the product, (R)-mellein [(R)-3,4-
dihydro-8-hydroxy-3-methylisocoumarin] is shown. (b) Mass spec-
trum of the purified enzymatic product.
Figure 2. Domain organization and enzymatic activity of SACE5532.
(a) Domain organization of SACE5532 and the sequence identity/
homology of the domains shared with ChlB1 and NcsB. b) In vitro
enzymatic activity of SACE5532 analyzed by HPLC. (c) Time-
dependent depletion of NADPH and formation of enzymatic product
in the enzymatic assay. (d) Absorption spectra of 6-MSA and the
enzymatic product of SACE5532.
conjunction with the malonyl-CoA synthase MatB to produce
13C-labeled malonyl-CoA in situ for the in vitro reaction,24 the
13C-labeled product was found to exhibit a molecular mass of
186.0992 Da (m/z 186.0780 calcd) (Figure S3). The difference
of 8 Da between the labeled and non-labeled products suggests
that the product is derived from a pentaketide intermediate
with eight carbon atoms from malonic acid and the other two
from the starter unit acetyl-CoA. A large-scale in vitro reaction
was carried out, and the enzymatic product was purified by
chromatography for NMR structure determination. Together
with the mass spectrometry data, the NMR spectra established
the product as mellein (see the NMR data in Figures S4−S6).
The R configuration at C3 was established by comparing the
optical specific rotation of the enzymatic product with that of
the synthetic standard.25 A survey of the literature suggested
that (R)-(−)-mellein is the most common stereoisomer
isolated from natural sources.
On the basis of the structure of the product and the
chemistry of polyketide synthesis, we propose a biosynthetic
mechanism for the novel mellein synthase (Figure 4). With the
phosphopantetheinylated SACE5532, the AT domain loads the
acyl groups from the acyl substrates onto the KS and ACP
domains. A decarboxylative Claisen condensation catalyzed by
the KS domain follows, generating the diketide intermediate.
To identify the product of SACE5532, the recombinant
SACE5532 was cloned and coexpressed with the phospho-
pantetheinyl transferase (PPTase) Sfp in Escherichia coli to
generate the phosphopantetheinylated protein. Coexpression
with Sfp not only increases the soluble fraction of the
recombinant protein but also prevents the formation of high-
molecular-weight oligomers (data not shown). The enzymatic
activity of the recombinant SACE5532 was examined by in
vitro assay. When SACE5532 was added to the reaction buffer
containing acetyl-CoA, malonyl-CoA, and NADPH, the
absorbance of NADPH at 340 nm began to decline
immediately. We observed a faster depletion of NADPH
when the PKS was preincubated with Sfp and coenzyme A,
indicating that the protein is only partially phosphopantethei-
nylated by Sfp in E. coli cells. Product analysis by HPLC
revealed that a single new species, 1, was produced by the
enzymatic reaction (Figure 2b and Figure S3). When acetyl-
CoA or malonyl-CoA was omitted from the reaction mixture,
the new species was not detected. No product was observed
when SACE5532 was either not coexpressed or not pretreated
with Sfp. When the putative catalytic residue Cys184 in the KS
domain was replaced by an alanine residue, no product was
detected either. A time-dependent enzymatic assay showed that
the amount of product increased steadily with the consumption
of NADPH before reaching the plateau (Figure 2c). These
experiments establish that the recombinant SACE5532 is
catalytically active and produces a single enzymatic product.
The absorption spectrum of the new species indicates that the
species is neither 6-MSA nor an NPA derivative (Figure 2d),
suggesting that SACE5532 produces a novel product.
To find out whether SACE5532 generates the same product
in a cellular environment, the medium of the E. coli cells
coexpressing SACE5532 and Sfp was extracted with organic
solvent. HPLC analysis of the medium extract showed that 1
was also produced by the E. coli cells (Figure 3a). The LC−MS
experiment revealed a mass of 178.0723 Da for 1 and suggested
a molecular formula of C10H10O3 (m/z 178.0630 calcd) (Figure
3b). Moreover, when [1,2,3-13C3]malonic acid was used in
Figure 4. Proposed biosynthetic mechanism for the mellein-synthesiz-
ing SACE5532.
11925
dx.doi.org/10.1021/ja304905e | J. Am. Chem. Soc. 2012, 134, 11924−11927

Journal of the American Chemical Society
Communication
The loading of the malonyl group and condensation are
repeated four times to generate a pentaketide intermediate
covalently tethered to the ACP domain. During this process,
the KR domain selectively reduces the keto group in the
diketide and tetraketide intermediates but not the triketide and
pentaketide intermediates. The pentaketide intermediate
subsequently undergoes an aldol cyclization to furnish the
aromatic structure through dehydration, and the TH domain
catalyzes the release of (R)-mellein via a stereospecific
cyclization process.
Although SACE5532 has the same domain composition and
shares head-to-tail sequence homology with ChlB1, MdpB,
PokM1, NcsB, and AziB, it generates a pentaketide product
rather than the tetra- or hexaketide-derived 6-MSA and NPAs.
More strikingly, the mechanism depicted in Figure 4 indicates
that SACE5532 exhibits a completely different keto reduction
pattern. As illustrated in Figure 5, while the other three PKSs
Figure 6. Substrate specificity of the KR domains of SACE5532 and
NcsB. (a) HPLC analysis of the production of mellein by SACE5532
with different starter units. (b) Selective reduction of the β-keto group
of the diketide analogues by the KR domain of SACE5532. (c)
Ketoreductase activity of the KR domains of SACE5532 and NcsB
toward (left) trans-1-decalone (13), (middle) acetoacetyl-SNAC (15),
and (right) S-ethyl acetothioacetate (14).
the mellein synthase achieve such a completely different keto
reduction pattern? Multiple sequence alignment suggests that
the four KR domains contain all of the critical catalytic residues
for a typical NADPH-binding KR domain (Figure S7).
Secondary structure prediction and structural modeling also
suggest that the KR domain of SACE5532 shares a similar
overall structure with typical KR domains.26,27 To investigate
whether the KR domain is solely responsible for determining
the keto reduction pattern, we cloned and expressed the
standalone KR domains from SACE5532 (1170−1638 amino
acids) and NcsB (1152−1641 amino acids) (Figure S2). Both
KR domains are soluble and enzymatically active, as
demonstrated by the reduction of trans-1-decalone (13), a
nonspecific substrate used for assaying the activity of KR
domains (Figure 6b,c).28 Because the analogues of the tri-,
tetra-, and pentaketide are chemically labile as a result of their
tendency to form lactones,29 we used two diketide analogues to
test whether the two standalone KR domains exhibit any
substrate preference. When the analogue of the diketide
intermediate acetoacetyl-N-acetylcysteamine thioester (acetoa-
cetyl-SNAC) (15) was used as the substrate, only KRSACE5532
was capable of reducing the keto group, as evidenced by the
depletion of NADPH (Figure 6c). No reduction was observed
for KRNcsB even after prolonged incubation. Notably,
KRSACE5532 also exhibits enzymatic activity toward one of the
simplest diketide analogues, S-ethyl acetothioacetate (14),
indicating that the KR domains are able to differentiate the
diketide intermediates through the recognition of the
acetoacetyl moiety. In another control experiment, no
reduction was observed for 5-oxohexanoyl-SNAC (16) for
either KR domain (Figure 6b and Figure S9), suggesting that
the reduction of the β-keto group of the diketide by KRSACE5532
is rather specific. Together, these results suggest that the
programmed keto reduction in these PR-PKSs is achieved by
Figure 5. Keto reduction patterns for SACE5532 and the homologous
PR-PKSs.
reduce the keto groups at C4, C8, and C10, SACE5532 reduces
the keto groups at C2 and C6. In other words, while ChlB1,
NcsB, and AziB reduce the β-keto group in the tri-, penta-, and
hexaketide intermediates, SACE5532 reduces the keto group in
the diketide and tetraketide intermediates. Supporting the
proposed mechanism and diketide intermediates in mellein
synthesis is the observation that SACE5532 is able to accept
acetoacetyl-CoA or hydroxybutyryl-CoA as the starter unit to
initiate the synthesis of mellein, albeit at lower rates (Figure
6a). Direct loading of the two diketide acyl groups presumably
enables the enzyme to bypass the first condensation step.
According to the mechanisms illustrated in Figure 5, the keto
group of the diketide intermediate is reduced by SACE5532 but
not by ChlB1, NcsB, or AziB (also see Figure S8). How does
11926
dx.doi.org/10.1021/ja304905e | J. Am. Chem. Soc. 2012, 134, 11924−11927

Journal of the American Chemical Society
Communication
(10) Itoh, J.; Shomura, T.; Omoto, S.; Miyado, S.; Yuda, Y.; Shibata,
U.; Inouye, S. Agric. Biol. Chem. 1982, 46, 1255.
(11) Shimojima, Y.; Hayashi, H.; Ooka, T.; Shibukawa, M.; Iitaka, Y.
Tetrahedron 1984, 40, 2519.
(12) Kurosaki, F.; Itoh, M.; Yamada, M.; Nishi, A. FEBS Lett. 1991,
288, 219.
(13) Kurosaki, F. Phytochemistry 1996, 41, 1023.
(14) O’Callaghan, J.; Caddick, M. X.; Dobson, A. D. W. Microbiology
2003, 149, 3485.
(15) Bacha, N.; Atoui, A.; Mathieu, F.; Liboz, T.; Lebrihi, A. Fungal
Genet. Biol. 2009, 46, 77.
discrimination of the polyketide intermediates by the KR
domain alone, that is, the KR domain of SACE5532 binds only
the diketide and tetraketide intermediates in a productive
configuration to allow hydride transfer from NADPH.
In summary, the experimental results presented here
establish the polyketide origin of mellein and may facilitate
the identification of the biosynthetic gene clusters for other
dihydroisocoumarins. Since isocoumarins are usually synthe-
sized by the KR-domain-lacking nonreducing PKSs (NR-PKSs)
(Figure S10), the results indicate that a distinct iterative PR-
PKS system has evolved for the synthesis of the structurally
related dihydroisocoumarins. It is also interesting to note that
for one of the fungal species that can produce mellein and
ochratoxins, a separate PKS seems to be responsible for
assembling the ochratoxins.30 It remains to be seen whether the
ochratoxin-synthesizing PKS bears any similarity to the mellein
synthase reported here. Comparative studies of the KR
domains have yielded insight into the mechanism of the
programmed keto reduction by the PR-PKSs. The substrate
specificity exhibited by the standalone KR domains toward
diketide analogues strongly suggests that the KR domain is able
to differentiate and selectively reduce the polyketide
intermediates in its substrate-binding pocket. In conjunction
with the recent insights into the mechanism of NR-PKSs and
the determination of the methylation pattern and stereospecific
keto reduction in fungal PR-PKSs,31−35 The results further raise
the hope of reprogramming and redesigning the iterative PKSs
in the near future.
(16) Buntin, K.; Rachid, S.; Scharfe, M.; Blocker, H.; Weissman, K. J.;
̈
Muller, R. Angew. Chem., Int. Ed. 2008, 47, 4595.
̈
(17) Oliynyk, M.; Samborskyy, M.; Lester, J. B.; Mironenko, T.;
Scott, N.; Dickens, S.; Haydock, S. F.; Leadlay, P. F. Nat. Biotechnol.
2007, 25, 447.
(18) Shao, L.; Qu, X.-D.; Jia, X.-Y.; Zhao, Q.-F.; Tian, Z.-H.; Wang,
M.; Tang, G.-L.; Liu, W. Biochem. Biophys. Res. Commun. 2006, 345,
133.
(19) Van Lanen, S. G.; Oh, T. J.; Liu, W.; Wendt-Pienkowski, E.;
Shen, B. J. Am. Chem. Soc. 2007, 129, 13082.
(20) Lu, W. L.; Roongsawang, N.; Mahmud, T. Chem. Biol. 2011, 18,
425.
(21) Sthapit, B.; Oh, T. J.; Lamichhane, R.; Liou, K.; Lee, H. C.; Kim,
C. G.; Sohng, J. K. FEBS Lett. 2004, 566, 201.
(22) Zhao, Q. F.; He, Q. L.; Ding, W.; Tang, M. C.; Kang, Q. J.; Yu,
Y.; Deng, W.; Zhang, Q.; Fang, J.; Tang, G. L.; Liu, W. Chem. Biol.
2008, 15, 693.
(23) Moriguchi, T.; Kezuka, Y.; Nonaka, T.; Ebizuka, Y.; Fujii, I. J.
Biol. Chem. 2010, 285, 15637.
(24) An, J. H.; Kim, Y. S. Eur. J. Biochem. 1998, 257, 395.
(25) Islam, M. S.; Ishigami, K.; Watanabe, H. Tetrahedron 2007, 63,
1074.
(26) Keatinge-Clay, A. T.; Stroud, R. M. Structure 2006, 14, 737.
(27) Zheng, J. T.; Taylor, C. A.; Piasecki, S. K.; Keatinge-Clay, A. T.
Structure 2010, 18, 913.
ASSOCIATED CONTENT
* Supporting Information
■
S
Experimental details and supporting figures and tables. This
material is available free of charge via the Internet at http://
pubs.acs.org.
́
(28) Østergaard, L. H.; Kellenberger, L.; Cortes, J.; Roddis, M. P.;
Deacon, M.; Staunton, J.; Leadlay, P. F. Biochemistry 2002, 41, 2719.
(29) Castonguay, R.; He, W. G.; Chen, A. Y.; Khosla, C.; Cane, D. E.
J. Am. Chem. Soc. 2007, 129, 13758.
AUTHOR INFORMATION
Corresponding Author
zxliang@ntu.edu.sg
■
(30) Harris, J. P.; Mantle, P. G. Phytochemistry 2001, 58, 709.
(31) Crawford, J. M.; Thomas, P. M.; Scheerer, J. R.; Vagstad, A. L.;
Kelleher, N. L.; Townsend, C. A. Science 2008, 320, 243.
(32) Fisch, K. M.; Bakeer, W.; Yakasai, A. A.; Song, Z.; Pedrick, J.;
Wasil, Z.; Bailey, A. M.; Lazarus, C. M.; Simpson, T. J.; Cox, R. J. J.
Am. Chem. Soc. 2011, 133, 16635.
Notes
The authors declare no competing financial interest.
(33) Zhou, H.; Gao, Z. Z.; Qiao, K. J.; Wang, J. J.; Vederas, J. C.;
Tang, Y. Nat. Chem. Biol. 2012, 8, 331.
(34) Crawford, J. M.; Korman, T. P.; Labonte, J. W.; Vagstad, A. L.;
Hill, E. A.; Kamari-Bidkorpeh, O.; Tsai, S.-C.; Townsend, C. A. Nature
2009, 461, 1139.
ACKNOWLEDGMENTS
■
This work was supported by a Tier II ARC Grant to Z.X.L.
from MOE, Singapore.
(35) Liu, T.; Chiang, Y. M.; Somoza, A. D.; Oakley, B. R.; Wang, C.
C. C. J. Am. Chem. Soc 2011, 133, 13314.
REFERENCES
(1) Butler, M. S. Nat. Prod. Rep. 2008, 25, 475.
(2) Holler, U.; Konig, G. M.; Wright, A. D. J. Nat. Prod. 1999, 62,
■
114.
(3) Krohn, K.; Bahramsari, R.; Florke, U.; Ludewig, K.; Kliche-Spory,
̈
C.; Michel, A.; Aust, H.-J.; Draeger, S.; Schulz, B.; Antus, S.
Phytochemistry 1997, 45, 313.
(4) Bayman, P.; Baker, J. L.; Doster, M. A.; Michailides, T. J.;
Mahoney, N. E. Appl. Environ. Microbiol. 2002, 68, 2326.
(5) Bennett, J. W.; Klich, M. Clin. Microbiol. Rev. 2003, 16, 497.
(6) Azumi, M.; Ogawa, K.; Fujita, T.; Takeshita, M.; Yoshida, R.;
Furumai, T.; Igarashi, Y. Tetrahedron 2008, 64, 6420.
(7) Li, Y. X.; Xu, Y.; Liu, L. L.; Han, Z.; Lai, P. Y.; Guo, X. R.; Zhang,
X. X.; Lin, W. H.; Qian, P. Y. Mar. Drugs 2012, 10, 319.
(8) Matsuda, H.; Shimoda, H.; Yamahara, J.; Yoshikawa, M. Biol.
Pharm. Bull. 1999, 22, 870.
(9) Yoshikawa, M.; Ueda, T.; Shimoda, H.; Murakami, T.; Yamahara,
J.; Matsuda, H. Chem. Pharm. Bull. 1999, 47, 383.
11927
dx.doi.org/10.1021/ja304905e | J. Am. Chem. Soc. 2012, 134, 11924−11927