Biosynthesis of the Validamycins
J. Am. Chem. Soc., Vol. 123, No. 12, 2001 2739
ments with [7-3H]-11 and [7-3H]-12, demonstrating their
incorporation into 1. When these experiments were conducted,
particular attention was paid to the possible formation of the
corresponding aminocyclitols in vivo. It was found that about
4% of the [7-3H]-12 fed to the organism was converted to
[7-3H]-3, whereas no conversion of [7-3H]-11 to [7-3H]-2 was
detected. The in vivo formation of [7-3H]-3 from [7-3H]-12 was
confirmed by regular autoradiography and by two-dimensional
co-autoradiography of the radioactive sample isolated from the
fermentation with an authentic reference sample of [7-3H]-3,
which showed a single radioactive spot. The formation within
the cells specifically of 3 from 12, but not of 2 from 11, lends
credence to the scenario of formation of validoxylamine A (13)
by reductive coupling of 3 and 11 and to the conjecture that
the negative outcome of the feeding experiments with the
tritiated aminocyclitols may be due to lack of cellular uptake.
Whereas the initial cyclization step leading to the first
carbocylic precursor seems to be the same or very similar in
the validamycin and acarbose fermentations, there are substantial
differences between the two biosynthetic pathways in the
subsequent reactions. Notably, none of the cyclitols fed to the
validamycin producer in the present study except 9 showed any
incorporation into acarbose, leading to the proposal that all of
the steps from 9 to acarviosyl-dTDP, a pseudodisaccharide
nucleotide precursor of 4 containing the valienamine moiety,
must occur on one enzyme or enzyme complex without any
free intermediates.26 In contrast, the results presented here
demonstrate that the transformation of 9 into the two cyclitol
moieties of 1 involves a series of free intermediates, and they
define the sequence of reactions and identify valienone (11) as
a proximate precursor of the two cyclitol units. The two reactions
needed to transform 9 into 11, epimerization at C-2 and
dehydration to generate the ∆5,6 double bond, both require the
carbonyl group at C-1, or an equivalent function such as an
imine, for activation. The efficient incorporation of 5-epi-
valiolone (10) demonstrates that C-2 epimerization apparently
occurs first, followed by dehydration (Scheme 5). The low
incorporation of 2-epi-valienone (15) into 1, although significant,
is best interpreted by assuming that 15 undergoes slow
spontaneous epimerization under the fermentation conditions,
producing small amounts of the pathway intermediate 11.
Alternatively the 2-epimerase may not be completely specific
for its substrate 9 but may less efficiently also epimerize 15
when presented with this compound. However, the low overall
incorporation of 15 compared to 9 and 10 argues against a
second parallel pathway from 9 to 11 via 15, suggesting that
15 may not normally be formed from 9.
Discussion
Following the earlier work of the Rinehart group19,21 which
demonstrated the biosynthetic origin of the two cyclitol moieties
of validamycin A (1) from a pentose phosphate pathway-derived
seven-carbon sugar, proposed to be either sedoheptulose 7-phos-
phate (5) or idoheptulose 7-phosphate (6), the results presented
here identify the first carbocyclic precursor of both cyclitol
moieties of 1 as 2-epi-5-epi-valiolone (9). The facts that this
cyclitol has the same configuration at all shared stereocenters
as 5 and that valiolone (7), which has the same stereochemistry
as 6, is not a precursor imply that sedoheptulose 7-phosphate
(5) must be the substrate for the cyclization reaction. These
findings parallel recent results on the biosynthesis of acarbose
(4) which have also identified 9 as the initial cyclitol precursor
of the valienamine moiety of 4.26 Furthermore, a cyclase gene
has been cloned from the acarbose biosynthetic gene cluster of
the 4-producing Actinoplanes species, and the recombinant
protein expressed from this gene has been shown to catalyze
the conversion of 5, but not 6, into 9.31 It seems very likely
that a similar enzyme catalyzes the first committed step in the
biosynthesis of 1.
Two biochemical mechanisms are known for the cyclization
of sugar phosphates to cyclitol derivatives. One is exemplified
by the cyclization of D-glucose 6-phosphate to myo-inositol
1-phosphate catalyzed by myo-inositol 1-phosphate synthase32,33
and the other by that of D-arabino-heptulosonic acid 7-phosphate
to dehydroquinic acid (DHQ) catalyzed by DHQ synthase.34
Applied to the cyclization of 5 the first mechanism would predict
2-epi-valienone (15) as the first nonphosphorylated cyclization
product, whereas the DHQ synthase-like mechanism would lead
to the observed precursor, 2-epi-5-epi-valiolone (9), as the first
cyclized compound. Several other facts support the notion that
the cyclization of 5 to 9 involves a DHQ synthase-like
mechanism: (i) The substrate 5 has the same configuration at
C-5 and C-6 as the substrate of DHQ synthase, 3-deoxy-D-
arabino-heptulosonic acid 7-phosphate, (ii) the product 9 has
the same configuration at the newly generated stereocenter, C-5,
as DHQ, the product of DHQ synthase, at the corresponding
carbon, and (iii) the cyclase from Actinoplanes generating 9
for acarbose biosynthesis shows a high degree of sequence
homology to DHQ synthases from different organisms.31
Valienone (11) as well as its precursors 9 and 10 are
incorporated into both cyclitol moieties of 1 as demonstrated
by NMR analysis of the products from feeding experiments with
the 13C- or deuterium-labeled compounds. The isotope distribu-
tion between the two cyclitol moieties was 1:1 in the experiment
with 13C-labeled 11. A slightly higher deuterium content of the
valienamine moiety than of the validamine moiety in the samples
of 1 derived from [6-2H2]-9 and [6-2H2]-10 (60 and 58%,
respectively) would be consistent with a somewhat longer
pathway to the validamine unit. It may, however, merely reflect
the loss of some deuterium from the latter moiety due to
enolization at the validone stage, the same process thought to
be responsible for the partial scrambling of deuterium between
the two methylene hydrogens at C-6′ of 1. Consistent with this
explanation the slightly lower relative deuterium enrichment in
the validamine moiety of 1 derived from [6-2H2]-9, compared
to that from [6-2H2]-10, coincides with a somewhat higher
degree of deuterium scrambling (see Figure 4).
Valienone (11) represents a branchpoint in the biosynthetic
pathway to 1. One molecule of 11 is incorporated into the
valienamine moiety of 1, the other is reduced to validone (12)
which gives rise to the validamine moiety. This follows clearly
from the fact that 12 is incorporated into 1 only half as
efficiently as 11 and labels exclusively the validamine moiety
of the antibiotic. The experiments with deuterated 9 and 10
incidentally also revealed the steric course of the reduction of
11 to 12, because both compounds give rise to [6-2H1]-11 in
the cells, which is then reduced to monodeuterated 12 and
incorporated into the validamine moiety of 1. The deuterium
NMR analysis of these samples of 1 revealed that the enzymatic
reduction of the ∆5,6 double bond in 11 involves the anti addi-
tion of two hydrogens. This steric course is commonly observed
for enzymatic reductions of C-C double bonds in R,â-un-
(31) Stratmann, A.; Mahmud, T.; Lee, S.; Distler, J.; Floss, H. G.;
Piepersberg, W. J. Biol. Chem. 1999, 274, 10889.
(32) Loewus, M. W.; Loewus, F. A.; Brillinger, G. U.; Otsuka, H.; Floss,
H. G. J. Biol. Chem. 1980, 255, 11710.
(33) Tian, F.; Migaud, M. E.; Frost, J. W. J. Am. Chem. Soc. 1999, 121,
5795.
(34) Widlanski, T.; Bender, S. L.; Knowles, J. R. J. Am. Chem. Soc.
1989, 111, 2299.