derived heptaketide and a hexaketide, respectively. Spectro-
scopic data (1H NMR, 13C NMR, MS, and UV) showed good
agreement with those of naringenin chalcone and resveratrol,
respectively, except the signals due to the terminal R-pyrone
ring. The structures of the minor products were determined
to be a novel C21 heptaketide chalcone (1) (0.6 mg, 0.96%
yield) and a novel C19 hexaketide stilbene (2) (1.6 mg, 2.9%
yield), which was uniquely consistent with both biogenetic
reasoning and NMR data including HMQC and HMBC. The
C21 chalcone was thus produced by a C-7/C-12 aldol-type
cyclization of a heptaketide intermediate, while a C-6/C-11
aldol-type cyclization of the intermediate folded in a different
conformation yielded the C19 stilbene (Figure 1A).
In contrast, hexanoyl-CoA was found to be a better
substrate for the enzyme reaction;8 A. arborescens OKS
efficiently accepted hexanoyl-CoA as a starter to produce a
3:4 mixture of a novel C18 heptaketide phloroglucinol (3)
(1.5 mg, 13.0% yield) and a novel C16 hexaketide resorcinol
(4) (2.0 mg, 21.2% yield) as major products after condensa-
tions with five and six molecules of malonyl-CoA, respec-
tively (Figure 1B).8 The two products gave the parent ion
peaks [M + H]+ at m/z 333 and 291 on LC-ESIMS,
respectively, and their structures were unambiguously elu-
cidated by NMR spectroscopic analysis. Thus, as in the case
of the coumaroyl-derived products, the C18 phloroglucinol
was produced by a C-7/C-12 aldol-type cyclization of a
heptaketide intermediate, while a C-6/C-11 aldol-type cy-
clization of a hexaketide intermediate yielded the C16
resorcinol (Figure 1B). The C16 resorcinol has been previ-
ously proposed as one of the possible reaction products of
the minimal type II PKS from hexanoyl-ACP/malonyl-CoA.9
Here it should be noted that the reaction products 1-4 have
not been isolated from either the aloe plant or other natural
sources. On the other hand, as previously reported, A.
arborescens OKS also accepts long chain (C10-C20) fatty
acyl-CoAs as a starter and carries out condensations with
malonyl-CoA only to produce triketide and tetraketide
R-pyrones without formation of an aromatic ring system.6a
Figure 2. (A) Active-site cavity of A. arborescens OKS (wild-
type). The residues lining the cavity are shown with the catalytic
triad (yellow). The bottom of the pocket and N222 are highlighted
in purple. (B) Formation of SEK 15 by N222G mutant and a HPLC
profile of the enzyme reaction products.
CHS) is uniquely substituted with hydrophobic Val351,
which causes loss of the binding pocket. As a result, OKS/
PCS no longer produce the tetraketide chalcone/stilbene from
the coumaroyl starter, but instead, both the OKS and the
PCS M207G mutant utilize novel buried pockets that extend
into the traditional active-site cavity for the production of
SEK4/SEK4b (Figure 2).11b Therefore, it is likely that the
formation of the coumaroyl- and hexanoyl-derived hepta-
ketide/hexaketide product is also dependent upon the pres-
ence of the novel buried pockets.
Recently, a resolved X-ray crystal structure of A. arbore-
scens OKS at 2.6 Å resolution revealed that OKS shares
active-site architecture similar to that of the previously
reported M207G mutant of A. arborescens PCS, which also
produces SEK4/SEK4b from eight molecules of malonyl-
CoA.10,11 In the CHS/STS enzyme reaction, it has been
proposed that 4-coumaroyl-CoA first binds to the so-called
“coumaroyl binding pocket”;2 however, in OKS/PCS, the
conserved S338 of CHS/STS (numbering in Medicago satiVa
Indeed, it was clearly demonstrated that a structure-based
OKS N222G mutant, in which the buried pocket was
expanded by a large-to-small substitution of N222 at the
bottom of the polyketide chain elongation tunnel, efficiently
produced the C21 heptaketide chalcone 1 as a major product
from coumaroyl-CoA/malonyl-CoA (Figure 1A).12 Interest-
ingly, when incubated with malonyl-CoA as a sole substrate,
the OKS N222G mutant produced a C20 decaketide ben-
zophenone, SEK15 (5),7a by condensations of 10 molecules
of malonyl-CoA (Figure 2), which was confirmed by direct
comparison with an authentic compound. The decaketide
benzophenone has been previously reported as a product of
genetically engineered type II PKSs7a and is the longest
polyketide produced by the structurally simple type III
(8) Steady state kinetic analysis revealed that KM ) 42.5 µM and kcat
)
0.211 min-1 for hexanoyl-CoA for the C16 resorcinol-forming activity, while
KM ) 94.0 µM and kcat ) 0.094 min-1 for malonyl-CoA for the SEK4b-
forming activity of OKS.6a
(9) Nicholson, T. P.; Winfield, C.; Westcott, J.; Crosby, J.; Simpson,
T. J.; Cox, R. J. Chem. Commun. 2003, 686–687.
(10) Morita, H.; Kondo, S.; Kato, R.; Wanibuchi, K.; Noguchi, H.; Sugio,
S.; Abe, I.; Kohno, T. Acta Crystallogr. 2007, F63, 947–949
.
(11) (a) Abe, I.; Utsumi, Y.; Oguro, S.; Morita, H.; Sano, Y.; Noguchi,
H. J. Am. Chem. Soc. 2005, 127, 1362–1363. (b) Morita, H.; Kondo, S.;
Oguro, S.; Noguchi, H.; Sugio, S.; Abe, I.; Kohno, T. Chem. Biol. 2007,
14, 359–369. (c) Abe, I.; Morita, H.; Oguro, S.; Noma, H.; Wanibuchi, K.;
Kawahara, N.; Goda, Y.; Noguchi, H.; Kohno, T. J. Am. Chem. Soc. 2007,
(12) Steady state kinetic analysis revealed that KM ) 32.1 µM and kcat
) 0.045 min-1 for 4-coumaroyl-CoA for the C21 chalcone-forming activity,
and KM ) 54.4 µM and kcat ) 0.027 min-1 for malonyl-CoA for the SEK15-
forming activity.
129, 5976–5980
.
Org. Lett., Vol. 11, No. 3, 2009
553