Engineered biosynthesis of plant polyketides: Chain length control in an octaketide-producing plant type III polyketide synthase

Article abstract of DOI:10.1021/ja053945v

The chalcone synthase (CHS) superfamily of type III polyketide syntheses (PKSs) produces a variety of plant secondary metabolites with remarkable structural diversity and biological activities (e.g., chalcones, stilbenes, benzophenones, acrydones, phloroglucinols, resorcinols, pyrones, and chromones). Here we describe an octaketide-producing novel plant-specific type III PKS from aloe (Aloe arborescens) sharing 50-60% amino acid sequence identity with other plant CHS-superfamily enzymes. A recombinant enzyme expressed in Escherichia coli catalyzed seven successive decarboxylative condensations of malonyl-CoA to yield aromatic octaketides SEK4 and SEK4b, the longest polyketides known to be synthesized by the structurally simple type III PKS. Surprisingly, site-directed mutagenesis revealed that a single residue Gly207 (corresponding to the CHS's active site Thr197) determines the polyketide chain length and product specificity. Small-to-large substitutions (G207A, G207T, G207M, G207L, G207F, and G207W) resulted in loss of the octaketide-forming activity and concomitant formation of shorter chain length polyketides (from triketide to heptaketide) including a pentaketide chromone, 2,7-dihydroxy-5-methylchromone, and a hexaketide pyrone, 6-(2,4-dihydroxy-6-methylphenyl)-4-hydroxy-2-pyrone, depending on the size of the side chain. Notably, the functional diversity of the type III PKS was shown to evolve from simple steric modulation of the chemically inert single residue lining the active-site cavity accompanied by conservation of the Cys-His-Asn catalytic triad. This provided novel strategies for the engineered biosynthesis of pharmaceutically important plant polyketides.

Full text of DOI:10.1021/ja053945v

Published on Web 08/17/2005
Engineered Biosynthesis of Plant Polyketides: Chain Length
Control in an Octaketide-Producing Plant Type III Polyketide
Synthase
Ikuro Abe,* Satoshi Oguro, Yoriko Utsumi, Yukie Sano, and Hiroshi Noguchi
Contribution from the School of Pharmaceutical Sciences and the COE21 Program,
UniVersity of Shizuoka, 52-1 Yada, Shizuoka 422-8526, Japan
Received June 15, 2005; E-mail: abei@ys7.u-shizuoka-ken.ac.jp
Abstract: The chalcone synthase (CHS) superfamily of type III polyketide synthases (PKSs) produces a
variety of plant secondary metabolites with remarkable structural diversity and biological activities (e.g.,
chalcones, stilbenes, benzophenones, acrydones, phloroglucinols, resorcinols, pyrones, and chromones).
Here we describe an octaketide-producing novel plant-specific type III PKS from aloe (Aloe arborescens)
sharing 50-60% amino acid sequence identity with other plant CHS-superfamily enzymes. A recombinant
enzyme expressed in Escherichia coli catalyzed seven successive decarboxylative condensations of malonyl-
CoA to yield aromatic octaketides SEK4 and SEK4b, the longest polyketides known to be synthesized by
the structurally simple type III PKS. Surprisingly, site-directed mutagenesis revealed that a single residue
Gly207 (corresponding to the CHS’s active site Thr197) determines the polyketide chain length and product
specificity. Small-to-large substitutions (G207A, G207T, G207M, G207L, G207F, and G207W) resulted in
loss of the octaketide-forming activity and concomitant formation of shorter chain length polyketides (from
triketide to heptaketide) including a pentaketide chromone, 2,7-dihydroxy-5-methylchromone, and a
hexaketide pyrone, 6-(2,4-dihydroxy-6-methylphenyl)-4-hydroxy-2-pyrone, depending on the size of the
side chain. Notably, the functional diversity of the type III PKS was shown to evolve from simple steric
modulation of the chemically inert single residue lining the active-site cavity accompanied by conservation
of the Cys-His-Asn catalytic triad. This provided novel strategies for the engineered biosynthesis of
pharmaceutically important plant polyketides.
Introduction
6-methyl-4-hydroxy-2-pyrone (triacetic acid lactone) (Figure
1B).2 Recent crystallographic and site-directed mutagenesis
The CHS superfamily of type III PKS enzymes is structurally
and mechanistically distinct from the modular type I and
dissociated type II PKSs of bacterial origin; the simple ho-
modimer of 40-45 kDa proteins directly utilize CoA-linked
substrates without involvement of the phosphopantetheine-armed
acyl carrier protein to catalyze complete series of polyketide
formation reactions with a single active site in an internal
cavity.1 The remarkable functional diversity of the CHS-
superfamily enzymes derives from the differences of their
selection of starter substrate, number of polyketide chain
extensions, and mechanisms of cyclization reactions. For
example, formation of 4,2′,4′,6′-tetrahydroxychalcone (narin-
genin chalcone) by CHS proceeds through sequential condensa-
tion of the C6-C3 unit of 4-coumaroyl-CoA as a starter with
three C2 units from malonyl-CoA, and the subsequent Claisen-
type cyclization of the enzyme-bound tetraketide intermediate
leads to formation of a new aromatic ring system (Figure 1A).1
On the other hand, 2-pyrone synthase (2PS) from daisy (Gerbera
hybrida) selects acetyl-CoA as a starter and carries out only
two condensations with malonyl-CoA to produce a triketide,
studies on plant2-4 and bacterial5,6 enzymes have begun to reveal
structural and functional details of the type III PKSs that share
almost the same three-dimensional overall fold and common
(2) (a) Eckermann, S.; Schro¨der, G.; Schmidt, J.; Strack, D.; Edrada, R. A.;
Helariutta, Y.; Elomaa, P.; Kotilainen, M.; Kilpela¨inen, I.; Proksch, P.;
Teeri, T. H.; Schro¨der, J. Nature 1998, 396, 387-390. (b) Jez, J. M.; Austin,
M. B.; Ferrer, J.; Bowman, M. E.; Schro¨der, J.; Noel, J. P. Chem. Biol.
2000, 7, 919-930.
(3) (a) Ferrer, J. L.; Jez, J. M.; Bowman, M. E.; Dixon, R. A.; Noel, J. P. Nat.
Struct. Biol. 1999, 6, 775-784. (b) Jez, J. M.; Ferrer, J. L.; Bowman, M.
E.; Dixon, R. A.; Noel, J. P. Biochemistry 2000, 39, 890-902. (c) Jez, J.
M.; Noel, J. P. J. Biol. Chem. 2000, 275, 39640-39646. (d) Jez, J. M.;
Bowman, M. E.; Noel, J. P. Biochemistry 2001, 40, 14829-14838. (e)
Tropf, S.; Ka¨rcher, B.; Schro¨der, G.; Schro¨der, J. J. Biol. Chem. 1995,
270, 7922-7928. (f) Suh, D. Y.; Fukuma, K.; Kagami, J.; Yamazaki, Y.;
Shibuya, M.; Ebizuka, Y.; Sankawa, U. Biochem. J. 2000, 350, 229-235.
(g) Suh, D. Y.; Kagami, J.; Fukuma, K.; Sankawa, U. Biochem. Biophys.
Res. Commun. 2000, 275, 725-730. (h) Jez, J. M.; Bowman, M. E.; Noel,
J. P. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 5319-5324. (i) Austin, M.
B.; Bowman, M. E.; Ferrer, J.-L.; Schro¨der, J.; Noel, J. P. Chem. Biol.
2004, 11, 1179-1194.
(4) (a) Abe, I.; Takahashi, Y.; Morita, H.; Noguchi, H. Eur. J. Biochem. 2001,
268, 3354-3359. (b) Abe, I.; Sano, Y.; Takahashi, Y.; Noguchi, H. J. Biol.
Chem. 2003, 278, 25218-25226. (c) Abe, I.; Utsumi, Y.; Oguro, S.;
Noguchi, H. FEBS Lett. 2004, 562, 171-176.
(5) (a) Funa, N.; Ohnishi, Y.; Fujii, I.; Shibuya, M.; Ebizuka, Y.; Horinouchi,
S. Nature 1999, 400, 897-899. (b) Funa, N.; Ohnishi, Y.; Ebizuka, Y.;
Horinouchi, S. J. Biol. Chem. 2002, 277, 4628-4635. (c) Funa, N.; Ohnishi,
Y.; Ebizuka, Y.; Horinouchi, S. Biochem. J. 2002, 367, 781-789. (d)
Izumikawa, M.; Shipley, P. R.; Hopke, J. N.; O’Hare, T.; Xiang, L.; Noel,
J. P.; Moore, B. S. J. Ind. Microbiol. Biotechnol. 2003, 30, 510-515. (e)
Austin, M. B.; Izumikawa, M.; Bowman, M. E.; Udwary, D. W.; Ferrer,
J.-L.; Moore, B. S.; Noel, J. P. J. Biol. Chem. 2004, 279, 45162-45174.
(1) For recent reviews, see: (a) Schro¨der, J. In ComprehensiVe Natural Products
Chemistry; Elsevier: Oxford, 1999; Vol. 1, pp 749-771. (b) Austin, M.
B.; Noel, J. P. Nat. Prod. Rep. 2003, 20, 79-110. (c) Staunton, J.;
Weissman, K. J. Nat. Prod. Rep. 2001, 18, 380-416.
9
10.1021/ja053945v CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 12709-12716
12709

A R T I C L E S
Abe et al.
Figure 1. Proposed mechanism for the formation of (A) naringenin chalcone from 4-coumaroyl-CoA and three molecules of malonyl-CoA by CHS, (B)
triacetic acid lactone from acetyl-CoA and two molecules of malonyl-CoA by 2PS, (C) 5,7-dihydroxy-2-methylchromone from five molecules of malonyl-
CoA by PCS, and (D) SEK4 and SEK4b from eight molecules of malonyl-CoA by A. arborescens octaketide synthase (OKS).
active-site architecture with an absolutely conserved Cys-His-
Asn catalytic triad. The polyketide formation reaction is thus
initiated by starter molecule loading at the active site Cys, which
is followed by malonyl-CoA decarboxylation, polyketide chain
elongation, and final cyclization of the polyketide intermediate.1
plant does not produce SEK4/SEK4b and their metabolites, it
is tempting to speculate that the enzyme is originally involved
in the biosynthesis of anthrones and anthraquinones in the
medicinal plant (Figure 2B).9 The physiological role of the
enzyme in the medicinal plant remains to be elucidated. Most
importantly, it was demonstrated that a single residue determines
the polyketide chain length and product specificity; the func-
tional diversity of the type III PKS was shown to evolve from
simple steric modulation of the chemically inert single residue
Gly207 (corresponding to the CHS’s active site thr197) lining
the active-site cavity.
Aloe (Aloe arborescens) is a medicinal plant rich in aromatic
polyketides such as pharmaceutically important aloenin (hexa-
ketide), aloesin (heptaketide), and barbaloin (octaketide) (Figure
2A). Therefore, in addition to regular CHSs involved in the
flavonoid biosynthesis, the presence of functionally distinct
multiple PKS enzymes that catalyze the initial key reactions in
the biosynthesis of the secondary metabolites was expected.7
Indeed, we recently succeeded in cloning a novel type III
pentaketide chromone synthase (PCS) from the medicinal plant;
PCS catalyzed condensation of five molecules of malonyl-CoA
to produce 5,7-dihydroxy-2-methylchromone (Figure 1C), a
biosynthetic precursor of the antiasthmatic furochromones,
kehellin and visnagin.7 Here we now report another novel A.
arborescens type III PKS that produces aromatic octaketides
SEK4 and SEK4b (Figure 1D), the longest polyketides known
to be synthesized by the structurally simple homodimeric type
III PKS.7 Notably, the octaketides are the products of the
minimal Type II PKS (heterodimeric complex of ketosynthase
and chain length factor) for the benzoisochromanequinone
actinorhodin (act from Streptomyces coelicolor).8 Since the aloe
Experimental Section
Chemicals. [2-14C]Malonyl-CoA (48 mCi/mmol) and [1-14C]acetyl
CoA (47 mCi/mmol) were purchased from Moravek Biochemicals
(CA). 4-Coumaroyl-CoA, cinnamoyl-CoA, and benzoyl-CoA were
chemically synthesized as described previously.10 Malonyl-CoA, acetyl-
CoA, and other aliphatic CoA esters were purchased from Sigma.
Authentic samples of SEK4/SEK4b7 and aloesone4c were obtained in
our previous works.
cDNA Cloning. As described before,7 total RNA was extracted from
young roots of A. arborescens and reverse-transcribed using oligo dT
primer (RACE 32 ) 5′-GGC CAC GCG TCG ACT AGT ACT TTT
(8) (a) Fu, H.; Ebert-Khosla, S.; Hopwood, D. A.; Khosla, C. J. Am. Chem.
Soc. 1994, 116, 4166-4170. (b) Fu, H.; Hopwood, D. A.; Khosla, C. Chem.
Biol. 1994, 1, 205-210.
(9) Dewick, P. M. In Medicinal Natural Products, A Biosynthetic Approach,
2nd ed.; Wiley: West Sussex, 2002.
(10) (a) Abe, I.; Morita, H.; Nomura, A.; Noguchi, H. J. Am. Chem. Soc. 2000,
122, 11242-11243. (b) Morita, H.; Takahashi, Y.; Noguchi, H.; Abe, I.
Biochem. Biophys. Res. Commun. 2000, 279, 190-195. (c) Morita, H.;
Noguchi, H.; Schro¨der, J.; Abe, I. Eur. J. Biochem. 2001, 268, 3759-
3766. (d) Abe, I.; Takahashi, Y.; Noguchi, H. Org. Lett. 2002, 4, 3623-
3626. (e) Abe, I.; Takahashi, Y.; Lou, W.; Noguchi, H. Org. Lett. 2003, 5,
1277-1280. (f) Abe, I.; Watanabe, T.; Noguchi, H. Phytochemistry 2004,
65, 2447-2453. (g) Oguro, S.; Akashi, T.; Ayabe, S.; Noguchi, H.; Abe,
I. Biochem. Biophys. Res. Commun. 2004, 325, 561-567.
(6) (a) Saxena, P.; Yadav, G.; Mohanty, D.; Gokhale, R. S. J. Biol. Chem.
2004, 278, 44780-44790. (b) Sankaranarayanan, R.; Saxena, P.; Marathe,
U.; Gokhale, R. S.; Shanmugam, V. M.; Rukmini, R. Nat. Struct. Mol.
Biol. 2004, 11, 894-900. (c) Pfeifer, V.; Nicholson, G. J.; Ries, J.;
Recktenwald, J.; Schefer, A. B.; Shawky, R. M.; Schro¨der, J.; Wohlleben,
W.; Pelzer, S. J. Biol. Chem. 2001, 276, 38370-38377. (d) Tseng, C. C.;
McLoughlin, S. M.; Kelleher, N. L.; Walsh, C. T. Biochemistry 2004, 43,
970-980.
(7) Abe, I.; Utsumi, Y.; Oguro, S.; Morita, H.; Sano, Y.; Noguchi, H. J. Am.
Chem. Soc. 2005, 127, 1362-1363.
9
12710 J. AM. CHEM. SOC. VOL. 127, NO. 36, 2005

Engineered Biosynthesis of Plant Polyketides
A R T I C L E S
Figure 2. (A) Structures of aromatic polyketides produced by aloe (A. arborescens). (B) Hypothetical scheme9 for involvement of OKS and a yet unidentified
ketoreductase in the biosynthesis of anthrones and anthraquinones in A. arborescens. In the absence of interactions with the tailoring enzyme, OKS just
affords SEK4/SEK4b as shunt products as in the case of the minimal type II PKS.
TTT TTT TTT TTT T-3′). The obtained cDNA mixture was used as a
template for the PCR reactions with inosine-containing degenerate
oligonucleotide primers based on the conserved sequences of known
CHSs: 112S ) 5′-(A/G)A(A/G) GCI ITI (A/C)A(A/G) GA(A/G) TGG
GGI CA-3′, 174S ) 5′- GCI AA(A/G) GA(T/C) ITI GCI GA(A/G)
AA(T/C) AA-3′, 368A ) 5′-CCC (C/A)(A/T)I TCI A(A/G)I CCI TCI
CCI GTI GT-3′, and 380A ) 5′-TCI A(T/C)I GTI A(A/G)I CCI GGI
CC(A/G) AA-3′. The number of primer indicates the amino acid number
of corresponding alfalfa (Medicago satiVa) CHS. Nested PCR was
carried out with the primer sets of 112S and 380A and then with 174S
and 368A to amplify a 544-bp DNA fragment. For the PCR 30 cycles
of reactions (94 °C for 0.5 min, 42 °C for 0.5 min, and 72 °C for 1
min) were performed each time with a 10 min final extension. The
gel-purified PCR product was ligated into pT7Blue T-Vector (Novagen)
and sequenced.
For the 3′-end amplification two specific primers (278S ) 5′-GGA
GCT CAC CAT CAT GAT GC-3′ and 327S ) 5′-CAT CTC GAC
AAT GCC ATC GG-3′) were designed based on the obtained core
sequence. First RT-PCR was carried out with the primer set of 278S
and RACE32 and the second PCR with 327S and RACE32 to amplify
a 359-bp DNA fragment. 5′-End amplification was carried out using
the Marathon cDNA Amplification kit (CLONTECH) and two specific
primers (331A ) 5′-CCA ATG ATC AGT GCA GCA GC-3′ and 232A
) 5′-CCG ATG GCA TTG TCG AGA TG-3′) to amplify a 702-bp
DNA fragment.
containing 100 µg/mL of ampicillin at 23 °C. Then, 1.0 mM isopropyl
thio-â-D-galactoside was added to induce protein expression, and the
culture was incubated further at 23 °C for 16 h.
Enzyme Purification. The E. coli cells were harvested by centrifu-
gation and resuspended in 40 mM potassium phosphate buffer, pH 7.9,
containing 0.1 M NaCl. Cell lysis was carried out by sonication and
centrifuged at 15 000 g for 40 min. The supernatant was passed through
a column of Pro-Bond resin (Invitrogen) containing Ni2+ as an affinity
ligand. After washing with 20 mM potassium phosphate buffer, pH
7.9, containing 0.5 M NaCl and 40 mM imidazole, the recombinant
OKS was finally eluted with 15 mM potassium phosphate buffer, pH
7.5, containing 10% glycerol and 500 mM imidazole. Protein concen-
tration was determined by the Bradford method (Protein Assay, Bio-
Rad) with bovine serum albumin as standard. Finally, to determine
subunit composition, the purified enzyme was applied to HPLC gel
filtration column (TSK-gel G3000SW, 7.5 × 600 mm, TOSOH), which
was eluted with 0.1 M KPB, pH 6.8, containing 10% glycerol and 0.2
M KCl at a flow rate of 1.0 mL/min. The standard molecular weight
markers used were â-amylase (200 kDa), alcohol dehydrogenase (150
kDa), bovine albumin (66 kDa), carbonic anhydrase (29 kDa), and
cytochrome c (12.4 kDa).
Site-Directed Mutagenesis. Gly207 mutants were constructed using
the QuickChange Site-Directed Mutagenesis Kit (Stratagene) and a pair
of primers as follows (mutated codons are italic): G207M sense 5′-
GAG CTC ACC ATA ATC ATG CTT CGA GGC CCT-3′, anti sense
5′-AGG GCC TCG AAG CAT GAT TAT GGT GAG CTC-3′; G207T
sense 5′-GAG CTC ACC ATA ATC ACG CTT CGA GGC CCT-3′,
anti sense 5′-AGG GCC TCG AAG CGT GAT TAT GGT GAG CTC-
3′; G207A sense 5′-GAG CTC ACC ATA ATC GCG CTT CGA GGC
CCT-3′, anti sense 5′-AGG GCC TCG AAG CGC GAT TAT GGT
GAG CTC-3′; G207L sense 5′-GAG CTC ACC ATA ATC CTG CTT
CGA GGC CCT-3′, anti sense 5′-AGG GCC TCG AAG CAG GAT
TAT GGT GAG CTC-3′; G207F sense 5′-GAG CTC ACC ATA ATC
TTC CTT CGA GGC CCT-3′, anti sense 5′-AGG GCC TCG AAG
GAA GAT TAT GGT GAG CTC-3′; G207W sense 5′-GAG CTC ACC
Expression of cDNA. A full-length cDNA was obtained using N-
and C-terminal PCR primers; 5′-CCG CCG AAT TCA TGA GTT CAC
TCT CCA AC-3′ (sense, the EcoR I site is italic) and 5′-ATA ATA
CTC GAG CAT GAG AGG CAG GCT GTG-3′ (antisense, the Xho I
site is italic). The amplified DNA was digested with EcoR I/Xho I and
cloned into the EcoR I/Xho I site of pET-22b(+) (Novagen). Thus, the
recombinant enzyme contains an additional hexahistidine tag at the C
terminal. After confirmation of the sequence, the plasmid was
transformed into E. coli BL21(DE3)pLysS. The cells harboring the
plasmid were cultured to an A600 of 0.6 in Luria-Bertani medium
9
J. AM. CHEM. SOC. VOL. 127, NO. 36, 2005 12711

A R T I C L E S
Abe et al.
ATA ATC TGG CTT CGA GGC CCT-3′, anti sense 5′-AGG GCC
TCG AAG CCA GAT TAT GGT GAG CTC-3′.
mM K-phosphate buffer, pH 7.5. Incubations were carried out at 30
°C for 30 min. The reaction products were extracted and separated by
Si-gel TLC (Merck Art. 1.11798; ethyl acetate/hexane/AcOH ) 63:
27:5, v/v/v). Radioactivities were quantified by autoradiography using
a bioimaging analyzer BAS-2000II (FUJIFILM). Lineweaver-Burk
plots of data were employed to derive the apparent KM and kcat values
(average of triplicates) using EnzFitter software (BIOSOFT).
Enzyme Reaction. As described before, the standard reaction
mixture contained 54 nmol of malonyl-CoA (and 27 nmol of other
CoA ester) and 280 pmol of the purified recombinant enzyme in a
final volume of 500 µL of 100 mM potassium phosphate buffer, pH
7.0.7 Incubations were carried out at 30 °C for 2 h and stopped by
adding 50 µL of 20% HCl. The products were then extracted twice
with 1000 µL of EtOAc and analyzed by HPLC and LC-ESIMS on a
TSK-gel ODS-80Ts column (4.6 × 150 mm, TOSOH) with a flow
rate of 0.8 mL/min. For the standard assay gradient elution was
performed with H2O and MeOH, both containing 0.1% TFA: 0-5 min,
30% MeOH; 5-17 min, 30-60% MeOH; 17-25 min, 60% MeOH;
25-27 min, 60-70% MeOH. Retention time (min): SEK4 (19.3),
SEK4b (20.6), aloesone (20.6), 6-(2,4-dihydroxy-6-methylphenyl)-4-
hydroxy-2-pyrone (16.9), 2,7-dihydroxy-5-methylchromone (22.7),
tetraacetic acid lactone (4.4), and triacetic acid lactone (6.0). For
separation of SEK4b and aloesone, gradient elution was performed with
H2O and CH3CN, both containing 0.1% TFA: 0-40 min, 15-30%
CH3CN. Retention time (min): SEK4 (17.8), SEK4b (20.3), and
aloesone (23.6). On-line HPLC-ESIMS spectra were measured with a
Hewlett-Packard HPLC 1100 series (Wilmington, DE) coupled to a
Finnigan MAT LCQ ion trap mass spectrometer (San Jose, CA) fitted
with an ESI source.
Homology Modeling. As described before,4b the model was
produced by the SWISS-MODEL package (http://expasy.ch/spdbv/)
provided by the Swiss-PDB-Viewer program.12 A standard homology
modeling procedure was applied based on the sequence homology of
residues 6-403 of A. arborescens OKS and the X-ray crystal structures
of CHS including M. satiVa CHS (1BI5A.pdb, 1BQ6A.pdb, 1CGKA.p-
db, 1CGZA.pdb, 1CHWA.pdb, 1CHWB.pdb, 1CMLA.pdb), M. satiVa
CHS C164A mutant (1D6FA.pdb), M. sativa CHS N336A mutant
(1D6HA.pdb), M. satiVa CHS H303Q mutant (1D6IA.pdb, 1D6IB.pdb),
M. satiVa CHS G256A mutant (1I86A.pdb), M. satiVa CHS G256V
mutant (1I88A.pdb, 1I88B.pdb), M. satiVa CHS G256L mutant
(1I89A.pdb, 1I89B.pdb), M. satiVa CHS G256F mutant (1I8BA.pdb,
1I8BB.pdb), and M. sativa CHS F215F mutant (1JWXA.pdb). The
corresponding Ramachandran plot was also created with Swiss PDB-
Viewer software to confirm that the majority of residues grouped in
the energetically allowed regions. Calculation of cavity volumes
(Connolly’s surface volumes) was then performed with the CASTP
program (http://cast.engr.uic.edu/cast/).
For large-scale enzyme reaction, 20 mg of purified enzyme was
incubated with malonyl-CoA (20 mg, 30 mmol) in 100 mL of 100
mM phosphate buffer, pH 7.5, containing 1 mM EDTA, at 30 °C for
18 h. The reaction was quenched by addition of 20% HCl (10 mL)
and extracted with ethyl acetate (200 mL × 3). The enzyme reaction
products (ca. 0.5-1.0 mg) were purified by reverse-phase HPLC.
Spectroscopic data of the products were as follows. 2,7-Dihydroxy-5-
methylchromone: ESIMS Rt ) 22.7 min, m/z 193 [M + H]+. UV:
Results and Discussion
A cDNA encoding the octaketide synthase (OKS) (GenBank
accession no. AY567707) was cloned and sequenced from
young roots of aloe (A. arborescens) by RT-PCR using
degenerate primers based on the conserved sequences of known
CHSs.7 A 1441-bp full-length cDNA contained a 60-bp 5′
noncoding region, a 1,212-bp open reading frame encoding a
Mr 44 568 protein with 403 amino acids, and a 169-bp of 3′
noncoding region. The deduced amino acid sequence showed
50-60% identity to those of other type III PKSs of plant origin
(Figure 3): 91% identity (368/403) with A. arborescens PCS,7
60% identity (240/403) with alfalfa (Medicago satiVa) CHS3a,
and 54% identity (216/403) with a heptaketide-producing
2-acetonyl-7-hydroxy-5-methylchromone (aloesone) synthase
(ALS) from Rheum palmatum.4c In contrast, it showed only 23%
identity (93/403) with a bacterial 1,3,6,8-tetrahydroxynaphtha-
lene synthase (RppA) from Streptomyces griseus.5a
Comparison of the sequence revealed conservation of the
catalytic triad (Cys164, His303, and Asn336) and most of the
CHS active-site residues (Met137, Gly211, Gly216, Phe215,
Phe265, and Pro375) (numbering in M. satiVa CHS);3 however,
CHS’s conserved Thr197, Gly256, and Ser338, sterically altered
in a number of divergent type III PKSs, are uniquely replaced
with Gly, Leu, and Val, respectively (Figure 3). The three
residues are also missing in the heptaketide-producing R.
palmatum ALS (T197A/G256L/S338T),4c the pentaketide-
producing A. arborescens PCS (T197M/G256L/S338V),7 and
the triketide-producing G. hybrida 2PS (T197L/G256L/S338I).2
These chemically inert residues lining the active-site cavity have
been proposed to control starter substrate selectivity and
polyketide chain length by steric modulation of the initiation/
elongation cavity.2,7 The CHS-based homology modeling pre-
dicted that A. arborescens OKS has the same three-dimensional
overall fold as M. satiVa CHS,3a with the total cavity volume
(1124 Å3) slightly larger than that of the chalcone (C15H12O5)-
λ
max 308 nm. 1H NMR (400 MHz, DMSO-d6): δ 6.53 (1H, d, J ) 2.0
Hz, H-6), 6.48 (1H, d, J ) 2.0 Hz, H-8), 5.32 (1H, s, H-3), 2.56 (3H,
s, CH3). 13C NMR (100 MHz, DMSO-d6): δ 182.3 (C-4), 164.7 (C-2),
161.8 (C-7), 160.3 (C-1a), 143.3 (C-5), 117.7 (C-4a), 114.5 (C-3), 112.5
(C-6), 100.3 (C-8), 22.7 (CH3). 6-(2,4-Dihydroxy-6-methylphenyl)-4-
hydroxy-2-pyrone: ESIMS Rt ) 16.9 min, m/z 233 [M + H]+. UV:
λ
max 304 nm. 1H NMR (400 MHz, DMSO-d6): δ 6.21 (1H, d, J ) 2.1
Hz, H-3′), 6.14 (1H, d, J ) 2.1 Hz, H-5′), 6.02 (1H, d, J ) 2.1 Hz,
H-5), 5.27 (1H, d, J ) 2.1 Hz, H-3), 2.06 (3H, s, CH3). 13C NMR (100
MHz, DMSO-d6): δ 170.5 (C-4), 166.5 (C-2), 159.1 (C-4′), 157.8 (C-
6), 156.7 (C-2′), 138.6 (C-6′), 116.8 (C-1′), 113.1 (C-5′), 105.3 (C-5),
100.1 (C-3′), 88.7 (C-3), 19.7 (CH3). The NMR assignments were
performed by comparison with those of 2-acetonyl-7-hydroxy-5-
methylchromone4c (aloesone) and 6-(2,4-dihydroxy-6-methylphenyl)-
4-methoxy-2-pyrone11 (the aglycone of aloenin).
Determination of Starter Substrate. Acetyl-CoA, resulting from
decarboxylation of malonyl-CoA, was also accepted as a starter substrate
as in the case of A. arborescens PCS7 but not so efficiently as in the
case of R. palmatum ALS.4c This was confirmed by the 14C incorpora-
tion rate from [1-14C]acetyl CoA in the presence of cold malonyl-CoA,
while the yield of the octaketides SEK4/SEK4b from [2-14C]malonyl-
CoA was almost at the same level in the presence or absence of cold
acetyl-CoA in the reaction mixture. Theoretical 14C-specific incorpora-
tion from [1-14C]acetyl-CoA should be 12.5% of those from [2-14C]-
malonyl-CoA if acetyl CoA serves as a starter unit of the octaketides
forming reaction, which was largely matched with the observed
incorporation rate.
Enzyme Kinetics. Steady-state kinetic parameters were determined
using [2-14C]malonyl-CoA (1.8 mCi/mmol) as a substrate. The experi-
ments were carried out in triplicate using five concentrations of substrate
(from 6.5 to 117.8 µM) in the assay mixture, containing 10 µg of
purified enzyme, 1 mM EDTA, in a final volume of 500 µL of 100
(11) Suga, T.; Hirata, T. Bull. Chem. Soc. Jpn 1978, 51, 872-877.
(12) Guex, N.; Peitsch, M. C. Electrophoresis 1977, 18, 2714-2723.
9
12712 J. AM. CHEM. SOC. VOL. 127, NO. 36, 2005

Engineered Biosynthesis of Plant Polyketides
A R T I C L E S
Figure 3. Comparison of the deduced amino acid sequences of A. arborescens OKS and other CHS-superfamily type III PKSs. M.s CHS, M. satiVa CHS;
A.h STS, A. hypogaea stilbene synthase; G.h 2PS, G. hybrida 2PS; R.p ALS, R. palmatum ALS; A.a PCS, A. arborescens PCS; A.a OKS, A. arborescens
OKS. The critical active-site residue 197 (in pink), the catalytic triad (Cys164, His303, and Asn336) (in red), and the residues lining the active site (Phe215,
Gly256, F265, Ser338) (in blue) are marked with # (numbering in M. satiVa CHS) and residues for the CoA binding with +.
forming CHS (1019 Å3) but much larger than that of the triketide
(C6H6O3)-producing G. hybrida 2PS (298 Å3). This suggested
that the active site of OKS is large enough to perform the seven
successive condensation reactions and accommodate the octa-
ketide products (C16H14O7).
A. arborescens OKS was heterologously expressed in E. coli
with an additional hexahistidine tag at the C-terminal. The
purified enzyme gave a single band with a molecular mass of
45 kDa on SDS-PAGE, while the native OKS appeared to be
a homodimer since it had an apparent molecular mass of 90
kDa as determined by gel filtration. Despite the 91% amino
acid sequence identity with the pentaketide-producing PCS, the
recombinant OKS did not produce 2,7-dihydroxy-2-methyl-
chromone (Figure 1C) but instead efficiently accepted malonyl-
CoA as a sole substrate to yield a 1:4 mixture of SEK4 and
SEK4b (Figure 1D and 5A), the longest polyketides known to
be generated by the structurally simple homodimeric type III
PKS.7 The octaketides SEK4/SEK4b are the shunt products of
the minimal type II PKS (heterodimeric complex of ketosynthase
and chain length factor) when expressed in the absence of the
ketoreductase.8 Since the medicinal plant does not produce
SEK4/SEK4b and their metabolites but instead produces a
significant amount of anthrones and anthraquinones (octaketides)
(Figure 2A), it is tempting to speculate that the enzyme is
originally involved in the biosynthesis of anthrones/anthraquino-
nes. However, maybe because of misfolding of the protein in
the E. coli expression system or because of the absence of
interactions with tailoring enzymes such as a yet unidentified
ketoreductase9 the recombinant OKS just afforded SEK4/SEK4b
as shunt products as in the case of the minimal type II PKS
(Figure 2B). The physiological role of OKS in the medicinal
plant remains to be elucidated.
Figure 4. Formation of a triketide and a tetraketide R-pyrone from
n-eicosanoyl-CoA and malonyl-CoA by OKS.
long-chain fatty acyl (n-hexanoyl, n-octanoyl, n-decanoyl,
n-dodecanoyl, n-tetradecanoyl, n-hexadecanoyl, n-octadecanoyl,
and n-eicosanoyl) CoA esters as a starter and carried out
sequential condensations with malonyl-CoA to produce triketide
and tetraketide R-pyrones without formation of a new aromatic
ring system (Figure 4). In particular, it is noteworthy that OKS
readily accepted the C20 chain length ester, much longer than
the previously reported Scutellaria baicalensis CHS that only
accepted up to the C12 ester,10f suggesting a difference in the
active-site structure between the two enzymes. Finally, acetyl-
CoA, resulting from decarboxylation of malonyl-CoA, was also
accepted as a starter substrate just as in the case of the previously
reported A. arborescens PCS7 (data not shown) but not so
efficiently as in the case of R. palmatum ALS.4c This was
confirmed by the 14C incorporation rate from [1-14C]acetyl CoA;
the yield of SEK4/SEK4b from [2-14C]malonyl-CoA was almost
at the same level in the presence or absence of cold acetyl-
CoA in the reaction mixture.
The recombinant OKS showed KM ) 95.0 µM and kcat
)
94.0 × 10-3 min-1 for malonyl-CoA (SEK4b forming activity)
with a pH optimum at 7.5 at 30 °C, which was comparable
with PCS (KM ) 71.0 µM and kcat ) 445 × 10-3 min-1).7 As
in the case of other type III PKSs,3h,10 A. arborescens OKS
exhibited unusually broad substrate tolerance; the enzyme also
accepted aromatic (4-coumaroyl, cinnamoyl, and benzoyl) and
9
J. AM. CHEM. SOC. VOL. 127, NO. 36, 2005 12713

A R T I C L E S
Abe et al.
As mentioned above, the octaketide-producing OKS and the
pentaketide-producing PCS share 91% amino acid sequence
identity, and the CHS’s active site Thr1973a is uniquely
substituted with small Gly207 in OKS while with bulky Met207
in PCS (Figure 3). Further, as we reported in a previous paper,
a PCS mutant in which Met207 was replaced by Gly efficiently
yielded the octaketides SEK4/SEK4b instead of the pentaketide
chromone.7 This suggested that the single residue determines
the polyketide chain length and product specificity by simple
steric modulation of the active-site cavity. To further test the
hypothesis we constructed a series of OKS mutants bearing
small-to-large changes in place of Gly207 (G207A, G207T,
G207M, G207L, G207F, and G207W mutant) and investigated
the mechanistic consequences of the point mutations.
First, when Gly207 was substituted with bulky Met as in the
case of PCS, OKS G207M mutant completely lost the octa-
ketide-forming activity but instead efficiently afforded an
unnatural pentaketide (m/z 193 [M + H]+) as a single product
(Figure 5B). The NMR (1H NMR and 13C NMR) and MS
spectra of the product obtained from a large-scale enzyme
reaction (1.0 mg from 20 mg of malonyl-CoA) were charac-
teristic of those of chromones and showed good agreement with
those of 2-acetonyl-7-hydroxy-5-methylchromone4c (aloesone),
except the signals due to the terminal acetonyl group. Accord-
ingly, the structure of the novel pentaketide was determined to
be 2,7-dihydroxy-5-methylchromone,13 which was uniquely
consistent with both biogenetic reasoning and the spectroscopic
data. The octaketide-producing OKS was thus converted to a
pentaketide synthase by single amino acid substitution. Interest-
ingly, the pentaketide is a regioisomer of the PCS product, 5,7-
dihydroxy-2-methylchromone (Rt ) 23.5 min; UV, λmax 292
nm), which is formed by a C-1/C-6 Claisen-type cyclization
(Figure 1B), while in OKS G207M a C-4/C-9 aldol-type
cyclization of the polyketide intermediate folded in a different
conformation yielded 2,7-dihydroxy-5-methylchromone (Rt )
22.7 min, UV, λmax 308 nm) (Figure 6A). Despite the high
sequence identity, OKS and PCS are not functionally intercon-
vertible by the point mutation, suggesting further subtle
structural differences of the active site between the two enzymes.
In addition, other bulky substitutions G207L and G207F also
resulted in production of the pentaketide 2,7-dihydroxy-5-
methylchromone along with a trace amount of SEK4/SEK4b
(Figure 6B).
Interestingly, unlike the wild-type OKS that efficiently
converted the C20 fatty acyl CoA into the triketide and tetraketide
R-pyrones (Figure 4), the pentaketide-producing G207M mutant
only accepted shorter chain length CoAs (up to the C12 ester)
as a starter substrate for the pyrone formation reactions (data
not shown). It is thus likely that the fatty acyl-CoAs are loaded
into the same active-site cavity where the sequential decar-
boxylative condensations of the growing polyketide chain take
place. Steric contraction of the cavity by the G207M bulky
substitution rejected the longer chain CoAs as well as octaketide
formation. Further intriguing, the G207M mutant did not afford
any reaction products from the C14 or C16 ester, while it
efficiently yielded the pentaketide chromone from malonyl-CoA
as a starter when incubated with the C18 and the longer chain
CoAs. This suggested that the C14 and C16 ester just blocked
up the active-site cavity and totally inhibited the condensation
Figure 5. HPLC elution profiles of enzyme reaction products of (A) wild-
type OKS, (B) OKS G207M mutant, (C) OKS G207A mutant, (D) OKS
G207T mutant, and (E) OKS G207W mutant. HPLC separation conditions
were as described in the Experimental Section. Note that only C was eluted
with the gradient program with H2O and CH3CN for separation of SEK4b
and aloesone.
reactions, while the C18 and larger CoAs could be no longer
loaded into the active site of the G207M mutant.
Next, when Gly207 was replaced with Ala as in the case of
the heptaketide-producing ALS, OKS G207A mutant indeed
yielded aloesone in addition to SEK4/SEK4b (Figures 5C and
6A). Notably, the heptaketide is a biosynthetic precursor of
(13) Stockinger, H.; Schmidt, U. Liebigs Ann. Chem. 1976, 1617-1625.
9
12714 J. AM. CHEM. SOC. VOL. 127, NO. 36, 2005

Engineered Biosynthesis of Plant Polyketides
A R T I C L E S
Figure 6. (A) Proposed mechanism for the formation of aromatic polyketides by OKS and its mutants. The enzymes catalyze chain initiation and elongation,
possibly initiating the first aromatic ring formation reaction at the methyl end of the polyketide intermediate (in red). The partially cyclized intermediates
are then released from the active site and undergo subsequent spontaneous cyclizations, leading to formation of the fused ring systems. (B) Distribution
pattern of aromatic polyketides produced by OKS and its mutants. Quantification of the products was calculated from the 14C incorporation rate from
[2-14C]malonyl-CoA under the standard assay condition.
aloesin (Figure 2A), the antiinflammatory agent of the medicinal
plant.4c On the other hand, when Gly207 was substituted with
Thr as in the case of CHS and most other plant type III PKSs3,
OKS G207T mutant completely lost the octaketide-forming
activity and instead efficiently afforded a hexaketide (m/z 235
[M + H]+) along with the pentaketide chromone (Figures 5D
and 6B). Spectroscopic data (1H NMR, 13C NMR, MS, and UV)
of the enzyme reaction product showed good agreement with
those of 6-(2,4-dihydroxy-6-methylphenyl)-4-methoxy-2-py-
rone11 (the aglycone of aloenin), except the signals due to the
4-methoxy group. The structure of the hexaketide was thus
determined to be 6-(2,4-dihydroxy-6-methylphenyl)-4-hydroxy-
2-pyrone (Figure 6A), which is a biosynthetic precursor of
aloenin (Figure 2A), the antihistamic agent of the medicinal
plant.11 Finally, when Gly207 was substituted with the most
bulky Trp, OKS G207W mutant just afforded a tetraketide
pyrone, 6-acetonyl-4-hydroxy-2-pyrone (tetraacetic acid lac-
tone)14 (m/z 169 [M + H]+), and triacetic acid lactone without
formation of an aromatic ring system (Figures 5E and 6A).
The small-to-large substitutions in place of Gly207 in A.
arborescens OKS (corresponding to CHS’s active site Thr197)
resulted in loss of the octaketide-forming activity and concomi-
tant formation of shorter chain length polyketides (from triketide
to heptaketide) including the hexaketide pyrone and the pen-
taketide chromone depending on the size of the side chain
(Figure 6B). It is conceivable that the point mutations cause
steric contraction of the active-site cavity where the polyketide
chain elongations take place, consequently shortening the
product chain length. Remarkably, the functional diversity of
(14) (a) Bentley, R.; Zwitkowits, P. M. J. Am. Chem. Soc. 1967, 89, 676-680.
(b) Samappito, S.; Page, J.; Schmidt, J.; De-Eknamkul, W.; Kutchan, T.
M. Planta 2002, 216, 64-71.
9
J. AM. CHEM. SOC. VOL. 127, NO. 36, 2005 12715

A R T I C L E S
Abe et al.
the type III PKS was shown to evolve from the simple steric
modulation of the active-site cavity accompanied by conserva-
tion of the Cys-His-Asn catalytic triad. An analogous result for
the chain length control has been recently described for the
minimal type II PKS (the heterodimeric complex of ketosynthase
and chain length factor).15 Presumably, OKS and the G207
mutant enzymes catalyze chain initiation and elongation and
possibly initiate the first aromatic ring formation reaction at
the methyl end of the polyketide intermediate (Figure 6A). The
partially cyclized aromatic intermediates would be then released
from the active site and undergo subsequent spontaneous
cyclizations, thereby completing the formation of the fused ring
systems.
Very recently we succeeded in crystallization of the penta-
ketide-producing A. arborescens PCS, both wild type and the
octaketide-producing M207G mutant enzyme (studies in progress).
The crystal structures at 1.6 Å resolution revealed that PCS and
CHS share the same three-dimensional overall fold, including
the CoA binding tunnel and the geometry of the catalytic triad.
Remarkably, it was clearly demonstrated that the Met207 lining
the active-site cavity indeed occupies a crucial position for the
polyketide chain elongation reactions. Substitution of Met207
with Gly opens a gate to a novel buried pocket and thus expands
a putative polyketide chain elongation tunnel, which lead to
formation of the longer octaketides. It is very likely that A.
arborescens OKS shares the similar active-site architecture and
machinery.
In conclusion, A. arborescens OKS is a novel plant type III
PKS that produces octaketides SEK4/SEK4b. We demonstrated
the mechanistic consequences of substitution of the crucial
active-site residue involved in the polyketide chain length control
and provided the structural basis for understanding the structure-
function relationship of type III PKS enzymes. These findings
revolutionized our concept for the catalytic potential of the
structurally simple plant type III PKSs and suggest novel
strategies for the engineered biosynthesis of pharmaceutically
important plant polyketides.
Acknowledgment. This work was supported in part by the
COE21 Program, Grant-in-Aid for Scientific Research (nos.
16510164, 17310130, and 17035069), by the Cooperation of
Innovative Technology and Advanced Research in Evolutional
Area (CITY AREA) from the Ministry of Education, Culture,
Sports, Science and Technology, Japan, and by a Health and
Labor Sciences Research Grant from the Ministry of Health,
Labour and Welfare, Japan.
(15) An analogous result in Type II PKS has been reported, see: (a) Tang, Y.;
Tsai, S.-C.; Khosla, C. J. Am. Chem. Soc. 2003, 125, 12708-12709. (b)
Keatinge-Clay, A. T.; Maltby, D. A.; Medzihradszky, K. F.; Khosla, C.;
Stroud, R. M. Nat. Struct. Mol. Biol. 2004, 11, 888-893.
JA053945V
9
12716 J. AM. CHEM. SOC. VOL. 127, NO. 36, 2005