Biosynthetic studies on the α-glucosidase inhibitor acarbose in actinoplanes sp.: 2-epi-5-epi-valiolone is the direct precursor of the valienamine moiety

Article abstract of DOI:10.1021/ja991102w

The biosynthetic pathway leading to the mC₇N cyclitol (valienamine) moiety of acarbose (1) in Actinoplanes sp. strain SN 223/29 has been studied using ³H-, ²H-, and ¹³C-labeled cyclitols. These precursors were s

Full text of DOI:10.1021/ja991102w

VOLUME 121, NUMBER 30
AUGUST 4, 1999
© Copyright 1999 by the
American Chemical Society
Biosynthetic Studies on the R-Glucosidase Inhibitor Acarbose in
Actinoplanes sp.: 2-epi-5-epi-Valiolone Is the Direct Precursor of the
Valienamine Moiety
Taifo Mahmud, Ingo Tornus,† Erin Egelkrout,‡ Eckardt Wolf,§ Charmaine Uy,
Heinz G. Floss, and Sungsook Lee*
Contribution from the Department of Chemistry, Box 351700, UniVersity of Washington,
Seattle, Washington 98195-1700
ReceiVed April 7, 1999
Abstract: The biosynthetic pathway leading to the mC7N cyclitol (valienamine) moiety of acarbose (1) in
Actinoplanes sp. strain SN 223/29 has been studied using 3H-, 2H-, and 13C-labeled cyclitols. These precursors
were synthesized from D-glucose or D-mannose as starting materials. The feeding experiments demonstrated
that cyclitols having the same stereochemistry at C-2 as the valienamine moiety of acarbose; i.e., valienone,
valienamine, valiolone, valiolamine, and 1-epi-valienol, were not incorporated and thus are not plausible
intermediates in 1 biosynthesis. 2-epi-Valiolone (10b), which has the same stereochemistry as the presumed
open-chain precursor, sedoheptulose 7-phosphate, was also not incorporated. However, its C-5 epimer (10a)
was incorporated efficiently and specifically into the valienamine moiety of 1. Surprisingly, the dehydrated
form of 2-epi-5-epi-valiolone, 2-epi-valienone, was not incorporated. This suggests that 2-epi-5-epi-valiolone
must be converted directly into the pseudodisaccharide moiety of acarbose without the intervention of other
free cyclitol intermediates. This may occur by linkage to the amino group of TDP-4-amino-4,6-dideoxyglucose
to form the imine, epimerization at C-2 to the correct stereochemistry, dehydration between C-5 and C-6
aided by enamine formation, and finally reduction to the amine. It is proposed that these reaction steps all take
place on a single enzyme without free intermediates. Alternative mechanistic possibilities are also discussed.
Introduction
Acarbose (1), known as an R-glucosidase inhibitor and a
clinically useful drug for the treatment of type II insulin-
independent diabetes, is isolated from the fermentation broth
of Actinoplanes sp.1 Its structure (Figure 1) consists of an
* Corresponding author.
† Present address: Schering AG, Berlin, Germany.
‡ Present address: Department of Biochemistry, Box 7622, North
Carolina State University, Raleigh, NC 27695-7622.
§ Present address: Toronto Research Chemicals Inc., 2 Brisbane Rd.,
Toronto, ON, Canada M3J 2J8.
Figure 1. Structure of acarbose (1).
unsaturated amino-cyclitol, valienamine (7, Scheme 1), a
deoxyhexose, and a maltose. The valienamine moiety, which
(1) Truscheit, E.; Frommer, W.; Junge, B.; Mu¨ller, L.; Schmidt, D.;
Wingender, W. Angew. Chem. 1981, 93, 744.
10.1021/ja991102w CCC: $18.00 © 1999 American Chemical Society
Published on Web 07/14/1999

6974 J. Am. Chem. Soc., Vol. 121, No. 30, 1999
Mahmud et al.
Scheme 1. Original Hypotheses for the Biosynthetic
Pathway to Acarbose
metabolites.18,19 It involves the conversion of dTDP-D-glucose
into dTDP-4-keto-6-deoxy-D-glucose through a stereospecific
intramolecular hydride transfer20 catalyzed by the pyridine
nucleotide-containing enzyme dTDP-glucose 4,6-dehydratase.
The two terminal glucose moieties of acarbose are derived not
from free glucose but from maltose and maltotriose in the culture
medium.13 Using H- and H-labeled maltose and maltotriose,
it was recently found that there are two metabolic routes from
maltotriose to the maltose unit of 1. Sixty percent of the acarbose
is formed by attachment of maltose, produced by removing a
glucose from the nonreducing end of maltotriose, to the
pseudodisaccharide core unit. The other 40% of the acarbose
is formed by direct attachment of maltotriose to the core unit,
followed by loss of the terminal glucose from the reducing end.21
3
2
In contrast, little was known about the mode of formation of
the valienamine moiety of acarbose from an initial pentose
phosphate pathway-derived precursor, except that feeding
experiments with 15N-labeled precursors had demonstrated that
glutamate is the primary source of the nitrogen.22 Valienamine
(7) itself was first predicted as a direct precursor of 1, which
through reductive condensation with dTDP-4-keto-6-deoxyglu-
cose would give the pseudodisaccharide core unit 8 (Scheme
1). However, 8 could also be formed from valienone (5, also
called gabosine23), the keto analogue of 7, by connection with
dTDP-4-amino-4,6-dideoxyglucose (27). Either 5 or 7 could
presumably be derived from valiolone (4), which was considered
a plausible first cyclization product of a linear seven-carbon
sugar, sedoheptulose 7-phosphate (2) or ido-heptulose 7-phos-
phate (3). In this paper we report the synthesis of a series of
isotopically labeled cyclitols as precursors and their use in
feeding experiments to elucidate the biosynthetic pathway to
the valienamine moiety of acarbose.
is primarily responsible for the inhibition of R-glucosidase,2 is
also a structural component of other secondary metabolites, such
as the adiposins,3 amylostatins,4 salbostatin,5 trestatin,6 and
validamycins.7 Valienamine and related amino-cyclitol structures
can be viewed8 as aliphatic analogues of mC7N units which are
found in many secondary metabolites,9 particularly the ansa-
mycins10 and mitomycins.11 The mC7N unit in those metabolites
is synthesized via a branch of the shikimate pathway,12 whereas
feeding experiments with stable isotope-labeled precursors have
demonstrate that the mC7N unit of acarbose13 and validamycin
A14 is derived from the pentose phosphate pathway. Specifically,
it was predicted that either sedoheptulose 7-phosphate (2), which
has the same stereochemistry as 7 at all carbons except C-5, or
its C-5 epimer, ido-heptulose 7-phosphate (3), is a proximate
precursor of the mC7N unit of acarbose (1) and validamycins.
Genetic information15 suggests that the cyclization of the
heptulose 7-phosphate involves an enzyme resembling dehydro-
quinate (DHQ) synthase, which catalyzes the formation of DHQ
from 3-deoxy-D-arabino-heptulosonic acid 7-phosphate (DAHP),16
and this notion has since been confirmed.17
Results
Feeding Experiments with Tritiated Precursors. To probe
which, if any, of the cyclitols 4, 5, 6, and/or 7 shown in Scheme
1 are intermediates in the biosynthesis of acarbose, we prepared
tritiated versions of these compounds and evaluated their
incorporation into 1 by Actinoplanes sp. The tritiated precursors
shown in Figure 2, [7-3H]valienamine ([7-3H]-7), [7-3H]va-
lienone ([7-3H]-5), [7-3H]valiolamine ([7-3H]-6), and [7-3H]-
valiolone ([7-3H]-4), were synthesized from valienamine as
described by us previously.24 In a first experiment, [7-3H]-(7)
(40 µCi, specific radioactivity ) 100 mCi/mmol) was fed to a
production culture of Actinoplanes sp. SN223/29, and it was
found that less than 0.1% of the radioactivity was incorporated
into 1. Most of the radioactivity (75%) evaporated during the
concentration of the sample solution in preparation for the CM-
25 cation-exchange column chromatography, and the rest of it
(25%) eluted much earlier than acarbose from the CM-25
column. This observation suggested the possibility that, at high
specific radioactivity, i.e., low concentration, the labeled va-
lienamine was metabolized extensively with formation of
The biosynthesis of the deoxysugar moiety of 1 conforms to
that of other deoxyhexoses which are found in many secondary
(2) Bock, K.; Pedersen, H. Carbohydr. Res. 1984, 132, 142.
(3) Namiki, S.; Kangouri, K.; Nagate, T.; Hara, H.; Sugita, K.; Omura,
S. J. Antibiot. 1982, 35, 1234.
(4) Murao, S.; Ohyama, K.; Ogura, S. Agric. Biol. Chem. 1977, 41, 919.
(5) Vertesy, L.; Fehlhaber, H.-W.; Schulz, A. Angew. Chem., Int. Ed.
Engl. 1994, 33, 1844.
(6) Yokose, K.; Ogawa, K.; Sano, T.; Watanabe, K.; Maruyama, H. B.;
Suhara, Y. J. Antibiot. 1983, 36, 1157.
(7) Iwasa, T.; Yamamoto, H.; Shibata, M. J. Antibiot. 1970, 23, 595.
(8) Hornemann, U.; Kehrer, J. P.; Nunez, C. S.; Ranieri, R. L. J. Am.
Chem. Soc. 1974, 96, 320.
(9) Floss, H. G. Nat. Prod. Rep. 1997, 14, 433.
(10) Kibby, J. J.; McDonald, I. A.; Rickards, R. W. J. Chem. Soc., Chem.
Commun. 1980, 768.
(11) Anderson, M. G.; Kibby, J. J.; Rickards, R. W.; Rothschild, J. M.
J. Chem. Soc., Chem. Commun. 1980, 1277.
(12) Kim, C. G.; Kirschning, A.; Bergon, P.; Zhou, P.; Su, E.; Sauerbrei,
B.; Ning, S.; Ahn, Y.; Breuer, M.; Leistner, E.; Floss, H. G. J. Am. Chem.
Soc. 1996, 118, 7486.
(13) Degwert, U.; Hu¨lst, R.; Pape, H.; Herrold, R. E.; Beale, J. M.; Keller,
P. J.; Lee, J. P.; Floss, H. G. J. Antibiot. 1987, 40, 855.
(14) Toyokuni, T.; Jin, W.-Z.; Rinehart, J., K. L. J. Am. Chem. Soc.
1987, 109, 3481.
(18) Liu, H.-w.; Thorson, J. S. Annu. ReV. Microbiol. 1994, 48, 223.
(19) Kirschning, A.; Bechthold, A. F.-W.; Rohr, J. Top. Curr. Chem.
1997, 188, 1.
(20) Snipes, C. E.; Brillinger, G.-U.; Sellers, L.; Mascaro, L.; Floss, H.
G. J. Biol. Chem. 1977, 252, 8113.
(21) Lee, S.; Sauerbrei, B.; Niggemann, J.; Egelkrout, E. J. Antibiot.
1997, 50, 954.
(15) Stratmann, A., Ph.D. Thesis, Bergische Universita¨t und GH Wup-
pertal, Germany, 1997.
(16) Widlanski, T.; Bender, S. L.; Knowles, J. R. J. Am. Chem. Soc.
1989, 111, 2299.
(22) Lee, S.; Egelkrout, E. J. Antibiot. 1998, 51, 225.
(23) Bach, G.; Breiding-Mack, S.; Grabley, S.; Hammann, P.; Hu¨tter,
K.; Thiericke, R.; Uhr, H.; Wink, J.; Zeeck, A. Liebigs Ann. Chem. 1993,
241.
(17) Stratmann, A.; Mahmud, T.; Lee, S.; Distler, J.; Floss, H. G.;
Piepersberg, W. J. Biol. Chem. 1999, 274, 10889.
(24) Lee, S.; Tornus, I.; Dong, H.; Gro¨ger, S. J. Labelled Compd.
Radiopharm., in press.

Biosynthetic Path to the Valienamine Moiety of Acarbose
J. Am. Chem. Soc., Vol. 121, No. 30, 1999 6975
Scheme 2. Syntheses of [1-13C]Valiolone and
[1-13C]Valienonea
a Conditions: (a) Bu3SnH, AIBN, toluene, reflux, 70%. (b) 10% Pd-
C, H2, room temperature, 99%. (c) MsCl, Et3N, TBME, room
temperature, 85%. (d) BBr3, CH2Cl2, -40 °C, 91%.
radioactivity), were each fed to production cultures of Actino-
planes sp. SN223/29. Again, none of the compounds showed
any significant incorporation into 1. The resulting acarbose was
purified to constant specific radioactivity by cocrystallization
with carrier material, and the incorporations were determined
as 0.016% for valiolone, 0% for valiolamine, and 0.01% for
valienamine. In all cases, a large fraction (45-75%) of the
radioactivity was released into the water of the medium during
the fermentation and/or workup procedures. The fractions eluting
from the CM-25 cation-exchange column with water or 0.5 N
NH4OH were assayed by TLC and found to contain the
unchanged substrates. Totals of 14% of [7-3H]-4, 6% of [7-3H]-
6, and 36% of [7-3H]-7, respectively, were recovered from these
fractions. This is in agreement with the recovery of 30% of
unchanged tritiated valienone in the earlier experiment with
[7-3H]-5.
The negative results from the feeding experiments with the
tritiated cyclitols in the acarbose producer can be interpreted in
at least three different ways: (a) the investigated cyclitols are
not intermediates on the biosynthetic pathway to 1, (b) these
substrates are not taken up by the cells, or (c) the tritium is
completely washed out during the fermentation at the stage of
the product 1 or a biosynthetic intermediate, but not the
substrates. The second explanation seems unlikely given the
metabolic tritium exchange, leaving mainly explanations a and
c. We had considered the possibility of tritium exchange from
the precursors and judged it to be unlikely; however, exchange
from an intermediate or the product 1, e.g., through a fast
exchange process made possible by redox reactions at the C-N
linkage, was a possibility. We therefore concluded that it was
necessary to reexamine the biosynthetic pathway to 1 with stable
isotope-labeled precursors and decided to synthesize 13C-labeled
valiolone and valienone.
Figure 2. Structures of the 3H-, 2H-, and 13C-labeled cyclitols
synthesized and evaluated as acarbose precursors.
tritiated water and/or converted to a neutral compound. In an
attempt to cut down on this presumed metabolic decomposition,
the tritiated valienamine was diluted to 1 mCi/mmol and fed
again (80 µCi) to a production culture. The incorporation of
tritium into 1 was again below 0.1%. These results strongly
suggest that valienamine is not an intermediate on the biosyn-
thetic pathway to acarbose.
In view of the negative results with tritiated valienamine, we
next considered the alternative scenario, in which valienone 5,
the keto compound corresponding to 7, is connected to dTDP-
4-amino-4,6-dideoxyglucose to give the pseudodisaccharide
nucleotide 8. [7-3H]-5 was therefore tested for incorporation
into acarbose by feeding it (50 µCi, 10 mCi/mmol) to production
cultures of Actinoplanes sp. strain SN223/29. Surprisingly, the
incorporation of radioactivity into 1 was again only about 0.1%,
suggesting that valienone was not a precursor in acarbose
biosynthesis either. In the CM-25 cation-exchange column
chromatography of the extract from this feeding experiment,
some fractions preceding acarbose contained a substantial
amount of radioactivity. These fractions were lyophilized and
then analyzed by TLC and radioautography. The major radioac-
tive spot had the same Rf value as valienone, and about 15 µCi
of [7-3H]-5 (30%) was recovered from these fractions. This
provided assurance that the tritium was not completely washed
out from the precursor.
Since the feeding experiments with both tritiated valienamine
and valienone had shown no incorporation of radioactivity into
1, we next considered the possibility that the C-C double bond
in the cyclitol moiety is introduced only after the linkage to the
deoxyhexose moiety. In that case, valiolone 4 or valiolamine 6
would be plausible precursors of acarbose. 4 was a plausible
candidate to be the first intermediate resulting from the
cyclization of a heptulose 7-phosphate. 6 was a less likely choice
as a precursor because, in this compound or its condensation
product with dTDP-4-keto-6-deoxyglucose, there is no driving
force for the elimination of the tertiary hydroxy group and the
adjacent unactivated hydrogen to generate the double bond of
the cyclitol moiety. Despite this reservation, feeding experiments
were carried out with both [7-3H]-4 and [7-3H]-6. Thus, three
labeled cyclitols, [7-3H]-4 (22 µCi), [7-3H]-6 (22 µCi), and
[7-3H]-7 (28 µCi) as a reference (all at 2.5 mCi/mmol specific
Synthesis of [1-13C]Valiolone ([1-13C]-4) and [1-13C]Va-
lienone ([1-13C]-5). The synthesis started from [1-13C]glucose
(12) and followed the procedure developed by Fukase and
Horii,25 allowing the preparation of 2,3,4,7-tetra-O-benzyl-[1-
13C]valiolone (14), which was then converted to both [1-13C]-
valiolone ([1-13C]-4) and [1-13C]valienone ([1-13C]-5) in a
parallel fashion as shown in Scheme 2. The conversion of
[1-13C]-4 into [1-13C]-5 was expected to occur easily under
acidic conditions on the basis of our experience in the prepara-
tion of valiolone by oxidative deamination of valiolamine.24
Only valiolone was detected in that work when the intermediate
imine was hydrolyzed with oxalic acid at pH 4, whereas
(25) (a) Fukase, H.; Horii, S. J. Org. Chem. 1992, 57, 3642. (b) Fukase,
H.; Horii, S. J. Org. Chem. 1992, 57, 3651.

6976 J. Am. Chem. Soc., Vol. 121, No. 30, 1999
Mahmud et al.
Scheme 3. Synthesis of 1-epi-[1-13C]Valienola
valienone and lesser amounts of valiolone were formed when
the hydrolysis was conducted at a more acidic pH. However,
valiolone itself was not dehydrated under strongly acidic
conditions (pH 1) and even at elevated temperature (up to 80
°C). We then tried to synthesize [1-13C]valienone from 2,3,4,7-
tetra-O-benzyl-[1-13C]valiolone (14) by carrying out the dehy-
dration first and then attempting debenzylation of the resulting
2,3,4,7-tetra-O-benzyl-[1-13C]valienone (15) without interfering
with the R,â-unsaturated carbonyl system. Eventually, the
conversion of 14 into 15 was carried out via a single-step
dehydration using an excess of mesyl chloride and Et3N, and
the removal of the benzyl moieties from 15 was accomplished
with BBr3 in CH2Cl2 (Scheme 2).
Feeding Experiments with [1-13C]Valiolone ([1-13C]-4) and
[1-13C]Valienone ([1-13C]-5). Both [1-13C]-4 and [1-13C]-5 (15
mg each) were fed to 100-mL production cultures of the
acarbose producer. After isolation and purification by the
standard procedure, the 1 samples were analyzed by 13C NMR.
No significant enrichment of 13C was found at C-1 of acarbose
in either experiment. At this point, we suspected that the labeled
valiolone and valienone might be diluted with too much
unlabeled material in the complex medium, and we therefore
repeated the feeding experiment with [1-13C]-5 in resting cells
of Actinoplanes sp SN223/29. Parallel experiments with a known
precursor, [U-13C3]glycerol, and with [7-3H]valienamine ([7-
3H]-7) were carried out simultaneously. [U-13C3]Glycerol was
incorporated into 1 as expected,13 but again [1-13C]-5 and [7-3H]-
7 were not incorporated, and 65% of the [7-3H]-7 was recovered
unchanged from the fermentation. These results leave little doubt
that the nonincorporation into 1 of [1-13C]-5 and, by inference,
of [1-13C]-4 in the previous experiments with production cultures
was not an artifact of dilution of the labeled precursors with
nutrients from the complex medium. The negative outcome of
the experiments with the stable isotope-labeled cyclitols 4 and
5 also ruled out the possibility that the washout of tritium from
a biosynthetic intermediate or from acarbose was responsible
for the lack of incorporation of the tritiated cyclitols into 1.
Thus, it can be concluded that none of the four cyclitols tested,
i.e., valiolone, valienone, valiolamine, and valienamine, is an
intermediate on the biosynthetic pathway to acarbose. These
results seemed to cast doubt on the proposed mechanism for
the linkage of the cyclitol to the deoxysugar moiety by reduction
of an imine formed between an amine and a ketone and caused
us to consider alternatives.
a Conditions: (a) NaBH4, CeCl3, EtOH, 95% (16:17 ) 81:19). (b)
BBr3, CH2Cl2, -40 °C, 60%.
course of the concentrations of these two compounds in the
cultures indicated that valienol appeared rather late in the
fermentation, after substantial amounts of 1 had accumulated,
suggesting that it might be a degradation product of 1, whereas
9 appeared early in the fermentation, consistent with a possible
role as a precursor of 1.
Mechanistic considerations also suggested the possibility that
1-epi-valienol might be a precursor of the valienamine moiety
of 1. C-1 of 9 could be linked to the C-4′ nitrogen of TDP-4-
amino-4,6-dideoxyglucose by an SN2 displacement of the C-1
oxygen, possibly after activation (e.g., as the phosphate or
pyrophosphate), to give the correct stereochemistry of the
pseudodisaccharide unit of 1. Evidence for C-N bond formation
by such a displacement has been obtained in studies on the
biosynthesis of aristeromycin and neplanocin A in Streptomyces
citricolor.29 We therefore synthesized 1-epi-[1-13C]valienol ([1-
13C]-9) by reduction of 2,3,4,7-tetra-O-benzyl-[1-13C]valienone
(15) using NaBH4 and CeCl3, followed by debenzylation with
BBr3 in CH2Cl2 (Scheme 3). The product was fed to resting
cells of Actinoplanes sp. SN223/29. The experiment gave no
13C incorporation into 1, indicating that 1-epi-valienol is not
an intermediate in the formation of the cyclitol moiety of
acarbose and ruling out the SN2 displacement mechanism
discussed above.
Synthesis of 2-epi-5-epi-[6-2H2]Valiolone ([6-2H2]-10a),
2-epi-[6-2H2]Valiolone ([6-2H2]-10b), and 2-epi-[6-2H]Va-
lienone ([6-2H]-11). The original choice of cyclitols having the
same C-2 stereochemistry as the valienamine moiety of 1, such
as 4 or 5, as the precursors to be tested was predicated on the
assumption that either ido-heptulose 7-phosphate is the open-
chain substrate of the cyclase generating the cyclitol, or, if the
substrate is sedoheptulose 7-phosphate, it undergoes epimer-
ization at C-5 during cyclization. The nonincorporation of
valiolone, valienone, and 1-epi-valienol suggested that this
assumption may not be correct and that the immediate cycliza-
tion product generated by the cyclase may be 2-epi-valiolone
(10b). 2-epi-Valiolone has the same C-2 configuration as 2,
making 2 the plausible substrate of the cyclase; the intermediacy
of 10b would require an epimerization during its further
conversion into the valienamine moiety of 1. 10b also has the
same C-5 configuration as natural valiolamine. To test whether
10b was a precursor of 1, the synthesis of 2-epi-[6-2H2]valiolone
([6-2H2]-10b) was pursued.
Experiments with 1-epi-Valienol (9). In parallel to the
feeding experiments described above, we also tried to obtain
information on the biosynthetic pathway to 1 by analyzing the
cyclitols in the fermentation broth and cell-free extracts of
Actinoplanes sp SN223/29.26 A compound was detected by ES-
MS in both the fermentation broth and the cell-free extracts
which had the same molecular weight, 176 Da, as valienol, the
reduction product of 5 with the same C-1 stereochemistry as 7.
To identify the compound, the cell-free extracts were lyophilized
and silylated, followed by GC-MS analysis. The GC-MS
experiments revealed the presence of two compounds of
molecular weight 176, showing virtually the same mass spectral
The synthesis of [6-2H2]-10a, its C-5 isomer [6-2H2]-10b, and
its dehydration product, 2-epi-[6-2H1]valienone ([6-2H2]-11)
(Scheme 4), proceeded from the commercially available starting
material, 2,3,4,6-tetra-O-benzyl-D-mannopyranose (18), by the
same route used to prepare [1-13C]-4 from 12 (cf. Scheme 2).
Reduction of 20 with sodium borohydride yielded a mixture of
isomers of acyclic heptitol derivative 21. In contrast to the
oxidation of the acyclic heptitol derived from glucose, which
fragmentation pattern. One was identical with valienol (Tret
)
11.12 min), while the other, having a higher retention time (Tret
) 11.48 min), was identified by co-injection with silylated
synthetic [6-2H]-9 as 1-epi-valienol 9.27 Analyses of the time
(26) Lee, S.; Floss, H. G., unpublished results.
(27) [6-2H]-9 was synthesized28 from 2,3,4,7-tetra-O-benzyl-D-glucose
by adopting the procedure of Fukase and Horii.25 The deuterium was
introduced in the same way as in the preparation of [6-2H2]-24a and [6-2H2]-
24b (Scheme 4).
(28) Lee, S.; Mahmud, T.; Floss, H. G., unpublished results.
(29) Hill, J. M.; Jenkins, G. N.; Rush, C. P.; Turner, N. J.; Willetts, A.
J.; Buss, A. D.; Dawson, M. J.; Rudd, B. A. M. J. Am. Chem. Soc. 1995,
117, 5391.

Biosynthetic Path to the Valienamine Moiety of Acarbose
J. Am. Chem. Soc., Vol. 121, No. 30, 1999 6977
Scheme 4. Synthesis of 2-epi-5-epi-[6-2H2]Valiolone and
2-epi-[6-2H2]Valiolonea
3.5 Hz andJ3,4) 4 Hz) compared to the twist boat form of 24b
(J2,3) 8 Hz and J3,4) 3 Hz) (Figure 4).
Finally, both [6-2H2]-24a and [6-2H2]-24b were deprotected
to the free cyclitols by catalytic hydrogenation using wet 10%
Pd/C as the catalyst, giving [6-2H2]-10a and [6-2H2]-10b,
respectively, in quantitative yields.
Next, [6-2H]-11, a dehydrated form of both [6-2H2]-10a and
[6-2H2]-10b, was prepared from 2,3,4,7-tetra-O-benzyl-2-epi-
[6-2H]valienone (25). A single-step dehydration of both [6-2H2]-
24a and its 5-epimer [6-2H2]-24b using mesyl chloride and Et3N
produced 25 (Scheme 5). Monitoring of the reaction by HPLC
indicated that the dehydration of [6-2H2]-24a and [6-2H2]-24b
to 25 occurred immediately after the mesyl group attaches to
the C-5 tertiary hydroxyl group in both isomers. In addition,
the conversion rate of both isomers to 25 seems to be unaffected
by the stereochemical and conformational differences between
[6-2H2]-24a and [6-2H2]-24b. To maintain the olefinic moiety
in [6-2H2]-11, the debenzylation of 25 was carried out using 6
equiv of BBr3 in CH2Cl2.
Feeding Experiments with 2-epi-[6-2H2]Valiolone ([6-2H2]-
10b), 2-epi-5-epi-[6-2H2]Valiolone ([6-2H2]-10a), and 2-epi-
[6-2H]Valienone ([6-2H]-11). To evaluate the role of the 2-epi-
valiolones in acarbose biosynthesis, the synthesized [6-2H2]-
10a and its C-5-epimer [6-2H2]-10b were each fed to resting
cell cultures. A parallel experiment was carried out with [U-13C3]-
glycerol as a standard. Analysis of the resulting acarbose by
ES-MS, 1H NMR, and 2H NMR (Figure 3) showed that 2-epi-
[6-2H2]valiolone ([6-2H2]-10b) was not at all incorporated.
However, its C-5 epimer, [6-2H2]-10a, was efficiently incorpo-
rated into acarbose (15% enrichment). Recovery of [6-2H2]-
10a was relatively poor compared to that of other labeled
cyclitols, which were usually recovered to the extent of 30-
50%. These results provide strong evidence that 2-epi-5-epi-
valiolone is a key intermediate in the biosynthesis of acarbose.
This makes it likely that 10a is the first cyclized compound on
the pathway and that sedoheptulose 7-phosphate is the substrate
of the cyclase generating it.
a Conditions: (a) Ac2O, DMSO, room temperature, quantitative. (b)
LDA, CH2Cl2, -78 °C. (c) NaBH4, THF/diglyme, room temperature.
(d) TFAA, DMSO, Et3N, CH2Cl2, -78 °C, three steps, 70%. (e) Sodium
acetate, 18-crown-6, CH2Cl2, room temperature, 90%. (f) Bu3SnD,
AIBN, toluene, reflux, 80%. (g) 10% Pd-C, H2, room temperature,
quantitative.
was followed by spontaneous intramolecular aldol condensation
to give a single cyclic dichloroinosose product, the oxidation
of the diol 21 with DMSO, trifluoroacetic anhydride, and Et3N
unexpectedly gave only a stable 1,5-dioxoheptose, 22. This result
is probably due to 23a or 23b adopting a boat or a twist boat
conformation (J2,3 ) 3.5 Hz, J3,4 ) 3.5 Hz), which is higher in
energy than the chair form found in the glucose-derived
dichloroinosose derivative 13. Eventually, the intramolecular
cyclization was accomplished with sodium acetate in the
presence of 18-crown-6 but gave a mixture of diastereomers
(23a and 23b) in a 1.5:1 ratio, which could be separated by
reversed-phase HPLC. As an alternative, the mixture of 23a
and 23b was directly converted, via a free-radical reduction
using Bu3SnD/AIBN, to the deuterium-labeled 2,3,4,7-tetra-O-
benzyl-2-epi-5-epi-valiolone ([6-2H2]-24a) and its 5-epimer
[6-2H2]-24b, which showed a better separation profile on
reversed-phase HPLC than the mixture of 23a and 23b.
Further experiments to assign the C-5 configurations in 23a
and 23b, as well as [6-2H2]-24a and [6-2H2]-24b, have been
attempted, including chemical derivatizations and crystallization
of 23a, 23b, [6-2H2]-24a, and [6-2H2]-24b, all of which were
unsuccessful. In contrast to their 2-epimers derived from glucose,
which are easily crystallized, these compounds were only
obtained as syrups. Several trials to induce epimerization at C-2
via keto-enol tautomerization in 23a and [6-2H2]-24a by
treating them with LDA/THF, TEA/DMF, and t-BuOK/t-BuOH
gave only the retro-aldol products. Compound 23a was also
found to be very stable under acidic conditions. We then
attempted to assign the stereochemistry of [6-2H2]-24a and
[6-2H2]-24b using NOE experiments, reasoning that the S
configuration (24b) at C-5 would yield an NOE between 5-OH
and 2-H and/or 3-H, whereas the corresponding R configuration
would not show the enhancement. Thus, irradiation at 5-OH
(2.61 ppm, s, disappeared in CDCl3 + D2O) of 24b gave a
strong NOE effect at 3-H and 2-H, indicating that 24b has S
configuration. On the other hand, irradiation at 5-OH (4.32 ppm,
d, J ) 1.5 Hz) of 24a only showed an NOE at 4-H, which
suggested the R configuration and indicated that 24a was more
likely to exist in a half-chair conformation, as also supported
The next question was whether 10a is first dehydrated to form
2-epi-valienone 11 and then connected to TDP-4-amino-4,6-
dideoxyglucose and epimerized at C-2 to the correct stereo-
chemistry, or whether all three steps occur on one enzyme. The
nonincorporation of valiolone had already ruled out that
epimerization precedes dehydration, and reductive amination
of 2-epi-5-epi-valiolone as the first step would make epimer-
ization and dehydration mechanistically implausible. We there-
fore fed [6-2H2]-11 to the resting cells (10 mg to 60 mL).
Analysis of the resulting 1 revealed no incorporation of
deuterium, demonstrating that 2-epi-valienone is not an inter-
mediate in the formation of acarbose.
Discussion
The results of the present investigation identify 2-epi-5-epi-
valiolone 10a as the only precursor of the valienamine moiety
of 1 among all the cyclitols tested. This finding suggests that
10a is the product of the cyclase, i.e., the first cyclitol generated,
and it implies that the substrate of the enzyme must be
sedoheptulose 7-phosphate 2, which has the same stereochem-
istry as 10a. Both of these conclusions have recently been
verified by studies on the recombinant enzyme from Actino-
planes sp. catalyzing this cyclization.17 This enzyme, which has
pronounced sequence homology to dehydroquinate (DHQ)
synthases, catalyzes the cyclization of 2, but not that of ido-
heptulose 7-phosphate to 2-epi-5-epi-valiolone. In this cycliza-
tion, carbons 1-7 of 2 give rise to carbons 7, 5, 4, 3, 2, 1, and
by the coupling constants between 2-H, 3-H, and 4-H (J2,3
)

6978 J. Am. Chem. Soc., Vol. 121, No. 30, 1999
Mahmud et al.
2
1
Figure 3. 1H and H NMR spectra of acarbose isolated from feeding experiments with deuterated precursors. (A) H NMR of acarbose from
2-epi-5-epi-[6-2H2]valiolone ([6-2H2]-10a) feeding experiment. (B) 1H NMR of acarbose from 2-epi-[6-2H2]valiolone ([6-2H2]-10b) feeding experiment.
(C) 1H NMR of acarbose from 2-epi-[6-2H1]valienone ([6-2H2]-11) feeding experiment. (D) 2H NMR of acarbose from 2-epi-5-epi-[6-2H2]valiolone
([6-2H2]-10a) feeding experiment.
that, of the two C-5 epimers of 10, it is the 5R isomer 10a which
is incorporated, and not the 5S isomer 10b, was at first
unexpected because the C-5 stereochemistry of 10b matches
that of natural valiolamine, which has been isolated from the
validamycin producer, Streptomyces hygroscopicus.31 However,
the C-5 configuration of 10a matches that of dehydroquinic acid
Figure 4. NOE experiments on compounds [6-2H2]-24a and [6-2H2]-
at the corresponding carbon and is thus the stereochemical
24b.
outcome predicted for a cyclization mediated by a DHQ
Scheme 5. Synthesis of 2-epi-[6-2H]Valienonea
synthase-like enzyme. Evidently, the naturally occurring vali-
olamine is not involved in the biosynthesis of validamycin;
rather, it may be derived from a metabolic breakdown product
of validamycin, such as valienamine.
It came as a considerable surprise that none of the cyclitols
having the same C-2 stereochemistry as the valienamine moiety
of acarbose was incorporated into 1. Hence, the biosynthesis
of 1 does not proceed via any of the keto- or amino-cyclitols 4,
5, 6, or 7 as free intermediates. This is particularly remarkable
since parallel investigations on the biosynthesis of the antibiotic
validamycin have revealed incorporation of not only 10a but
also valienone 5 into its valienamine moiety.32 Evidently, there
are significant differences in the pathways from 10a to the
valienamine moieties of acarbose and validamycin. The non-
incorporation of 1-epi-valienol 9 into 1 is also noteworthy, as
it rules out a biosynthetic process in which the cyclitol and the
a Conditions: (a) MsCl, Et3N, TBME, room temperature, 90%. (b)
BBr3, CH2Cl2, -40 °C, 70%.
6, respectively, of the resulting cyclitol. The substrate stereo-
chemistry at C-5 and C-6 conforms to that of 3-deoxy-D-
arabino-heptulosonic acid 7-phosphate, the substrate of dehy-
droquinate synthase, and as in the formation of DHQ, the
configuration at C-5 is maintained, despite the fact that this
carbon undergoes transient oxidation to the ketone.30 The fact
(31) Kameda, Y.; Asano, N.; Yoshikawa, M.; Takeuchi, M.; Yamaguchi,
T.; Matsui, K.; Horii, S.; Fukase, H. J. Antibiot. 1984, 37, 1301.
(32) Dong, H., Ph.D. thesis, University of Washington, Seattle, 1998.
(30) Widlanski, T.; Bender, S. L.; Knowles, J. R. J. Am. Chem. Soc.
1989, 111, 2299 and references therein.

Biosynthetic Path to the Valienamine Moiety of Acarbose
J. Am. Chem. Soc., Vol. 121, No. 30, 1999 6979
Scheme 6. Proposed Pathway for the Biosynthesis of Acarbose
deoxysugar moieties are connected by an SN2 displacement of
the 1-OH group of 9, possibly in activated form, by the nitrogen
of a 4-amino-6-deoxysugar derivative, as precedented in the
biosynthesis of aristeromycin and neplanocin A.29 This leaves
as the most plausible alternative a process in which the two
components are connected by formation and subsequent reduc-
tion of imine intermediates.
the 1-producing Actinoplanes strain.34 We therefore propose that
the nitrogen is first introduced by transamination into 26 and
that the resulting 27 then forms a Schiff’s base with a keto-
cyclitol, which is eventually reduced to the unique saturated
C-N-C linkage found in 1. However, more evidence at the
enzymatic level is required in order to firmly establish this
pathway and to rule out a more complicated alternative three-
component process in which 26, a keto-cyclitol, and the nitrogen
are reductively condensed on a single enzyme to give the
pseudodisaccharide moiety of 1.
The conversion of 10a into the pseudodisaccharide nucleotide
8 requires the following four reaction steps: (i) epimerization
at C-2, (ii) dehydration between C-5 and C-6, (iii) condensation
between the C-1 keto group and the amino group of 27, and
(iv) reduction of the resulting imine. A key question concerns
the order in which these reaction steps take place. The
nonincorporation into 1 of valiolone 4 rules out epimerization
as the first step, and the nonincorporation of 2-epi-valienone
rules out dehydration. Thus, Schiff’s base formation between
10a and the amine component, presumably 27, must be the first
reaction step. Mechanistic considerations dictate that epimer-
ization and dehydration must occur prior to reduction of the
imine double bond, because the amine resulting from this
reduction would not provide any plausible mechanistic path for
epimerization and dehydration. Thus, we propose that steps i
and ii take place on the imine formed from 10a and 27, making
use of the imine-enamine tautomerism to promote both of these
reactions. The order of the epimerization and dehydration steps
cannot be decided without further evidence; however, epimer-
ization before dehydration would avoid the transient formation
of a cross-conjugated enol and is thus slightly preferred. The
final reduction of the imine to give 8 is likely to involve a
reduced flavin as the enzymatic cofactor.
The structures of the two compounds undergoing reductive
condensation to give the pseudodisaccharide nucleotide and the
exact nature of this process cannot yet be specified, but the
results do provide certain insights. The amino nitrogen of
glutamate has been identified as the most proximate nitrogen
source.22 This suggests, as the simplest interpretation, that the
nitrogen is first introduced into one of the keto precursors, either
a keto-cyclitol or dTDP-4-keto-6-deoxyglucose (26, Scheme 1),
followed by Schiff’s base formation with the second keto
component and reduction of the imine. The fact that neither
valiolamine 6 nor valienamine 7 is incorporated into 1, provided
it is not due to permeability barriers, rules out transamination
at the stage of the keto-cyclitols 4 or 5. Transamination of 2-epi-
5-epi-valiolone 10a cannot be ruled out as the point of first
introduction of the nitrogen, because we have not examined the
incorporation of 2-epi-5-epi-valiolamine. However, the conver-
sion of 10a into 1 requires both epimerization at C-2 and
dehydration to give the ∆-5,6 double bond. Prior transamination
would leave no plausible mechanistic path for these essential
reaction steps; hence, the intermediacy of 2-epi-5-epi-valiol-
amine seems very unlikely. Transamination of 26 to dTDP-4-
amino-4,6-dideoxy-D-glucose 27, on the other hand, is not
encumbered by such mechanistic complications. The reaction
is precedented in Escherichia coli, from which Strominger’s
laboratory partially purified an enzyme catalyzing the 4-trans-
amination of dTDP-4-keto-6-deoxy-D-glucose to its 4-amino
derivative.33 An enzyme activity catalyzing the transamination
of 26 to 27 has also been demonstrated in cell-free extracts of
The mechanism we propose for the formation of the presumed
1 precursor, the pseudodisaccharide nucleotide 8, based on the
available information is shown in Scheme 6. Whether the
(33) Matsuhashi, M.; Strominger, J. L. J. Biol. Chem. 1966, 241, 4738.
(34) Lee, S.; Egelkrout, E.; Niggemann, J., unpublished results.

6980 J. Am. Chem. Soc., Vol. 121, No. 30, 1999
Mahmud et al.
[1-13C]Valiolone and [1-13C]valienone (15 mg) were each fed to 100-
mL cultures, half at 38 h and half at 48 h, harvesting at 100 h of
fermentation.
Schiff’s bases i through iv are free intermediates in the processs
i.e., whether the four steps i-iv listed above are each catalyzed
by distinct, separate enzymes, or whether all the reactions occur
on one single enzyme or enzyme complex, with i - iv only
existing as enzyme-bound intermediatessremains a matter of
speculation. Given that the diphosphate group provides a built-
in acid catalyst for their hydrolysis and formation, compounds
i - iv may be expected to be relatively unstable in the aqueous
environment of the cell and to equilibrate with their components,
27 and the cyclitols 4, 5, and 10a, respectively. Thus, the fact
that none of the ketones 4, 5, and 11 showed even traces of
incorporation into 1 may be taken as circumstantial evidence
against the intermediacy of free i-iv in acarbose biosynthesis.
Instead we favor the hypothesis that all the reactions from 10a
plus 27 to 8 take place on a single enzyme or enzyme complex,
with Schiff’s bases i-iv as enzyme-bound intermediates.
Experiments are underway to examine this hypothesis specif-
ically and 1 biosynthesis in general at the enzymatic level.
Culture Conditions for Feeding Experiments in Resting Cells.
For the resting cell experiments, production cultures in complex medium
were incubated for 28-30 h and centrifuged (4000g for 5 min) to re-
move the supernatant. The remaining cells were washed twice with cold
100 mM potassium phosphate buffer (pH 6.9) and then suspended in
50 mM potassium phosphate (pH 6.9) containing maltose (1.5%) and
bactopeptone (0.15%) to make 75 mL of cell suspension from 50 mL
of production culture. Each 60-75 mL of resting cells suspension in
a 500-mL flask was incubated for 22-24 h on a rotary shaker at 220
rpm and 28 °C, with labeled substrates added at 0 h (one third) and
the other two thirds at 8 h. Precursors fed were [1-13C]valienone, 10
mg/75 mL; [U-13C3]glycerol, 100 mg/75 mL; [7-3H]valienamine, 60
µCi/75 mL (specific radioactivity 2.5 mCi/mmol); 1-epi-[1-13C]valienol,
15 mg/60 mL; 2-epi-5-epi-[6-2H2]valiolone, 15 mg/60 mL; 2-epi-[6-
2H2]valiolone, 15 mg/60 mL; and 2-epi-[6-2H]valienone, 10 mg/60 mL.
Isolation and Purification of Acarbose. Following incubation, the
cultures from the complex medium and resting cells were harvested
by centrifugation (4000g × 20 min). The supernatant containing
acarbose was treated with an equal volume of methanol to prevent
bumping and concentrated under reduced pressure to a syrup. The
residue was dissolved in water (20 mL) and treated with methanol (80
mL). The mixture was stirred at room temperature for 2 h and
centrifuged. The supernatant was concentrated to dryness under reduced
pressure. The residue was dissolved in 5 mL of water, centrifuged to
remove insoluble material, and applied to a CM-25 weak cation-
exchange column (1.5 × 60 cm). The column was eluted with deionized
water, and the fractions (each 7.5-8 mL) containing acarbose (usually
between fractions 65 and 80), as detected by UV monitoring at 210
nm, were pooled. The acarbose-containing fractions were evaporated
to dryness under reduced pressure for further analysis by 1H-, 2H-, and
13C NMR or electrospray mass spectrometry (ES-MS). The radioactivity
of acarbose was determined by liquid scintillation counting. The
amounts of acarbose isolated from the production cultures were about
Experimental Section
1
2
General. The H, H, and 13C NMR spectra were recorded on a
Bruker AF-300 NMR spectrometer with MacNMR 5.5 PCI as the
instrument controller and data processor. HMBC and HMQC were
recorded on a Bruker DRX-499 with SGI O2 computer. Infrared spectra
were recorded on a Perkin-Elmer 1720 infrared Fourier transform
spectrometer. Optical rotations were measured on a JASCO DIP-370
digital polarimeter. Low-resolution electrospray mass spectra were
recorded on a Kratos Profile double-focusing magnetic sector mass
spectrometer. A Fison VG Quattro II electrospray ionization mass
spectrometer was used to measure the isotope enrichment of acarbose.
FAB-MS was recorded on a Micromass 70SEQ tandem hybrid mass
spectrometer (EbqQ) with an RSX-11 operating system. A Hewlett-
Packard 5890 GC with 5971A mass-selective detector was used for
the GC-MS analysis of cyclitols. A ISF-4-V shaker, Adolf Kuhner
AG, was used for the fermentations. Tritiated samples were counted
in Bio-Safe II biodegradable counting cocktail (Research Products
International Corp.) with a Beckman LS 1801 scintillation counter.
Reactions were monitored by TLC on silica, and detection was
accomplished by UV light or by alkaline permanganate spray and Ce-
(SO4)2/H2SO4 solution.
3
30-40 mg from the incubations with H-cyclitols (50-mL cultures),
60 mg from that with [1-13C]-4, and 70 mg from the incubation with
[1-13C]-5 (100-mL cultures). In the resting cell experiments, the amounts
were 9 mg from [1-13C]-5, 8 mg from [U-13C3]glycerol (2-3%
incorporation), 4 mg from [1-13C]-9, 7 mg from [6-2H2]-10a, 6.5 mg
from [6-2H2]-10b, and 6 mg from [6-2H]-11, respectively.
Reisolation of Labeled Substrates. The earlier fractions (fractions
10-25) from the isolation of acarbose by CM-25 cation-exchange
column chromatography contained stable isotope-labeled neutral sub-
strates, i.e., [1-13C]-4, [1-13C]-5, [1-13C]-9, [6-2H2]-10a, [6-2H2]-10b,
or [6-2H]-11. These fractions were lyophilized (100 µL each) and
analyzed by both TLC and GC-MS; Rf values on TLC (n-butanol/
ethanol/water ) 9:7:4) were valiolone, 0.45; valienone, 0.64; 1-epi-
valienol, 0.54; 2-epi-valiolone and its C5-epimer, 0.56; and 2-epi-
valienone, 0.66. GC-MS (Tret in min): valiolone, 11.61; valienone,
11.51; 1-epi-valienol, 11.48; 2-epi-valiolone, 11.02; 2-epi-5-epi-vali-
olone, 11.06; and 2-epi-valienone, 10.90. In the case of radiolabeled
neutral substrates, [7-3H]-4 and [7-3H]-5, the fractions showing high
radioactivity were lyophilized, and then they were analyzed by thin-
layer autoradiography (treated with autoradiography enhancer). The
amine substrates, [1-3H]-7 and [1-3H]-6, were recovered by elution of
the CM-25 column with 0.5 N NH4OH after 1 had been eluted with
water. They were identified by radio-TLC and GC-MS: Rf values on
TLC (2-propanol/AcOH/H2O): valienamine, 0.39; valiolamine, 0.30.
GC-MS (Tret in min): valienamine, 11.33; valiolamine, 12.24.
Isolation and Identification of Valienol and 1-Epi-valienol from
Actinoplanes sp. SN223/29. (A) From the Fermentation Broth. The
culture (10 mL) from the complex medium was centrifuged (4000g ×
20 min), and the supernatant was passed through a Dowex 50 (H+ form,
1.5 × 5 cm) column. The column was eluted with water, and the eluate
was concentrated under reduced pressure. The residue was dissolved
in water (0.5 mL) and applied to a Sephadex G-15 column (1.5 × 30
cm). The column was eluted with water, and the fractions containing
valienol and 1-epi-valienol were pooled. An aliquot (100 µL) was
lyophilized and silylated with Sigma-sil A (50 µL) prior to GC-MS
Materials. Actinoplanes sp. SN223/29 was obtained from Bayer
AG, Wuppertal, Germany. Tritiated valienamine, valienone, valiolam-
ine, and valiolone were prepared in our laboratory from validamycin
A as previously described.24 All chemicals were purchased from Aldrich
or Sigma and used without further purification unless otherwise noted.
Ingredients for fermentations, soybean meal (fat free), maltzin powder,
and yeast extract, were obtained from Bayer AG, Germany, and NZ-
Amine A from ICN Biochemicals. Sephadex CM-25 was purchased
from Pharmacia and Sephadex LH-20 from Sigma. Dianion WK-100
and Amberlite IR-68 were obtained from Supelco, and Amberlite CG-
50, Dowex 50, and Dowex 1 were from Aldrich. Bio-Rad AG 1-8 was
purchased from Bio-Rad. Surface autoradiography enhancer (EN3-
HANCE Spray) was obtained from DuPont and X-ray films from
Amersham Life Science.
Culture Conditions for Feeding Experiments in Complex Me-
dium. The inoculum suspension of Actinoplanes sp. SN223/29 (1 mL)
was transferred into a 250-mL flask containing 50 mL of vegetative
medium consisting of glycerol (2.5%), soybean meal (3%, fat free),
and CaCO3 (0.2%), pH 7.2, before sterilization. After incubation of
the suspension for 3 days on a rotary shaker at 28 °C and 200 rpm, 5
mL of preculture was transferred to a 500-mL flask containing 45 mL
of complex medium, consisting of maltzin powder (7.5%), yeast extract
(0.7%), NZ-Amine A (0.3%), and K2HPO4 (0.3%), and incubated with
shaking as above. All labeled compounds were fed as filter-sterilized
solutions during the fermentation. The 3H-labeled compounds were each
added in two portions (half at 40 h, half at 50 h) to 50 mL of the
production culture, harvesting after 90-96 h of total fermentation.

Biosynthetic Path to the Valienamine Moiety of Acarbose
J. Am. Chem. Soc., Vol. 121, No. 30, 1999 6981
analysis. GC-MS (Tret in min): TMS-valienol, 11.12; TMS-1-epi-
valienol, 11.48.
mL) dropwise with stirring under argon at -40 °C, and stirring was
continued for 30 min. The reaction mixture was allowed to warm to 0
°C, MeOH (3 mL) and H2O (3 mL) were added, and the mixture was
warmed to room temperature with stirring for 15 min. The reaction
mixture was transferred to a separatory funnel and partitioned between
CH2Cl2 and H2O. The H2O layer was directly subjected to anion-
exchange chromatography (Bio-Rad AG1-X8, formate form, 20 mL),
and the product was lyophilized to dryness. The residue was dissolved
in MeOH and repurified over Sephadex LH-20 (50 mL, MeOH) to
give [1-13C]-5 (43 mg, 91%).
(B) From the Cells. The cells collected from the above culture were
resuspended in water (10 mL) and disrupted by sonication. The cell
debris was removed by centrifugation (10000g × 10 min), and the
supernatant was subjected to filtration using a Centricon-10. The filtrate
was passed through a Dowex 50 (H+ form) column, which was eluted
with water. The eluate was concentrated and subjected to Sephadex
G-15 column chromatography with elution with water. The fractions
containing valienol and 1-epi-valienol were pooled, and an aliquot (100
µL) was lyophilized and silylated with Sigma-sil A (50 µL) prior to
GC-MS analysis.
Colorless syrup, [R]D -61.7° (c ) 0.25, MeOH, 22 °C). ESI-MS:
1
m/z 198 (M + Na)+. H NMR (300 MHz, D2O): δ 3.58 (ddd, J ) 1,
Synthesis of Labeled Substrates. 2,3,4,7-Tetra-O-benzyl-[1-13C]-
valiolone (14). To a solution of 13 (3 g, 4.8 mmol) in toluene (40 mL)
were added tributyltin hydride (5 mL, 18.6 mmol) and AIBN (300 mg,
1.8 mmol), and the mixture was refluxed for 2 h and cooled to room
temperature. The product was extracted with EtOAc (50 mL) and
washed sequentially with 2 N HCl, saturated aqueous NaHCO3, and
brine. The organic solvent was evaporated under reduced pressure and
the residue purified by silica gel column chromatography (n-hexane/
CH2Cl2/MeOH ) 12:4:1) to give 14 (1.85 g, 70%).
9, 11 Hz, H-3), 4.05 (dd, J ) 4, 11 Hz, H-2), 4.26, 4.33 (ABq, J ) 19
Hz, H-7a,b), 4.36 (br d, J ≈ 9 Hz, H-4), 6.00 (dd, J ) 2, 4 Hz, H-6).
13C NMR (75 MHz, D2O): δ 59.1 (t, JC-C ) 4.8 Hz, C-7), 70.4 (d,
C-4), 74.5 (d, J
) 40.3 Hz, C-2), 75.5 (d, JC-C ) 2.5 Hz, C-3),
C-C
119.0 (d, JC-C ) 52.5 Hz, C-6), 164.6 (s, C-5), 198.1 (s, C-1).
2,3,4,7-Tetra-O-benzyl-1-epi-[1-13C]valienol (16). A suspension of
CeCl3 (138 mg, 0.56 mmol) in ethanol (3.3 mL) was cooled to -78
°C, and sodium borohydride (42 mg, 1.12 mmol) was added. The
mixture was stirred at - 78 °C for 1 h. A solution of 2,3,4,7-tetra-O-
benzyl-[1-13C]valienone (15) (200 mg, 0.37 mmol) in ethanol/THF (1:
1, 2 mL) was added dropwise with stirring under argon at -78 °C,
and stirring was continued for 1 h. The reaction mixture was allowed
to warm to 0 °C, and cold EtOAc (10 mL) and 2 N HCl (5 mL) were
added. The organic layer was washed with H2O, saturated aqueous
NaHCO3, and brine. The organic solvent was removed under vacuum,
and the residue was subjected to silica gel column chromatography
(n-hexane/EtOAc ) 5:1-4:1) to give 16 (163 mg, 81%).
14. Colorless syrup, [R]D + 44.6° (c ) 0.35, CHCl3, 22 °C). FAB-
MS: m/z 576 (M + Na)+. HR FAB-MS: m/z calcd for 12C3413C1H36O6-
Na (M + Na)+ 576.2443, found 576.2468. 1H NMR (300 MHz,
CDCl3): δ 2.37 (d, J ) 2 Hz, 5-OH), 2.46 (dd, J ) 7, 15 Hz, 6-Heq),
2.85 (bdd, J ≈ 5, 15 Hz, 6-Hax), 3.15 (d, J ) 9 Hz, 7-Ha), 3.53 (dd,
J ) 1, 9 Hz, 7-Hb), 4.01-4.15 (m, H-2, H-3, H-4), 4.38-5.00 (Ph-
CH2- × 4), 7.15-7.45 (m, C6H5 × 4). 13C NMR (75 MHz, CDCl3): δ
45.5 (t, JC-C ) 35.0 Hz, C-5), 72.7 (t, JC-C ) 3.6 Hz, C-7), 73.3, 74.1,
75.7, 76.0 (all t, Ph-CH2- × 4), 80.0 (d, C-4), 85.5 (d, JC-C ) 35.0 Hz,
C-2), 83.1 (d, JC-C ) 2.5 Hz, C-3), 127.6-128.4 (all d), 137.4, 137.7,
137.8, 138.4 (all s, C6H5- × 4), 203.9 (13C-labeled CdO).
White solid, [R]D -73.0° (c ) 0.38, CHCl3, 22 °C). FAB-MS: m/z
560 (M + Na)+. HR FAB-MS: m/z calcd for 12C3413C1H36O5Na (M +
Na)+ 560.2494, found 560.2479. 1H NMR (300 MHz, CDCl3): δ 3.55
(ddd, J ) 5, 8, 9.5 Hz, 2-H), 3.83 (ddd, J ) 2.5, 7.0, 9.5 Hz, 3-H),
3.89 (br d, J ≈ 12 Hz, 7-Ha), 4.22 (br d, J ≈ 12 Hz, 7-Hb), 4.27-4.34
(2H, m, 1-H and 4-H), 4.42-4.97 (Ph-CH2-), 5.71 (br s, H-6), 7.22-
[1-13C]Valiolone ([1-13C]-4). To a solution of 14 (55 mg, 0.1 mmol)
in 95% aqueous ethanol (5 mL) was added 10% wet Pd/C (55 mg),
and the mixture was stirred at room temperature under an H2 atmosphere
overnight. The suspension was passed through a Celite column to
remove the catalyst and subsequently was filtered through a 0.2-µm
membrane filter. The solvent was evaporated under reduced pressure
to give pure [1-13C]-4 (19.2 mg, 99%).
7.33 (m, C6H5 × 4). 13C NMR (75 MHz, CDCl3): δ 69.9 (t, JC-C
)
6.1 Hz, C-7), 71.2 (d, C-1), 79.6 (d, C-3), 83.5 (d, C-4), 83.8 (d, JC-C
) 34.2 Hz, C-2), 127.1 (d, JC-C ) 45 Hz, C-6), 135.9 (s, JC-C ) 2.44
Hz, C-5).
1-epi-[1-13C]Valienol ([1-13C]-9). To a solution of 16 (115 mg, 0.21
mmol) in CH2Cl2 (5.4 mL) was added 1.0 M BBr3 in hexane (1.28
mL) dropwise with stirring under argon at -40 °C, and stirring was
continued for 30 min. The reaction mixture was allowed to warm to 0
°C, MeOH (2.5 mL) and H2O (2.5 mL) were added, and the mixture
was warmed to room temperature with stirring for 15 min. The reaction
mixture was transferred to a separatory funnel and partitioned between
CH2Cl2 and H2O. The H2O layer was subjected to anion-exchange
chromatography (Bio-Rad AG1-X8, formate form, 20 mL), and the
product was lyophilized to dryness. The residue was dissolved in MeOH
and repurified over Sephadex LH-20 (50 mL, MeOH) to give [1-13C]-
9 (23 mg, 61%).
Colorless syrup, [R]D -32.5° (c ) 0.22, MeOH, 22 °C). ESI-MS:
m/z 200 (M + Na)+. 1H NMR (300 MHz, CD3OD): δ 3.36-3.47 (m,
2-H and 3-H), 3.81 (m), 4.35 (md, JC-H ≈ 160 Hz, H-1), 4.05-4.20
(3H, m, 4-H and 7-H), 5.57 (dd, J ) 1.5, 3.5 Hz, 6-H). 13C NMR (75
MHz, CD3OD): δ 69.3 (13C-labeled C-1).
Lactone 19. To a solution of 2,3,4,7-tetra-O-benzyl-D-mannose (18)
(5.0 g, 9.25 mmol) in DMSO (50 mL) was added acetic anhydride (35
mL), and the mixture was stirred overnight at room temperature. The
reaction mixture was poured into ice water (1 L) and extracted with
TBME (3 × 150 mL). The organic layer was washed with 2 N HCl,
H2O, saturated aqueous NaHCO3, and brine. After the solution was
dried over Na2SO4, the solvent was evaporated under vacuum. The
residue was chromatographed on a silica gel column (n-hexane/EtOAc
) 3:1), and the product was recrystallized from n-hexane/TBME to
give 19 in quantitative yield.
White needles, [R]D + 0.37° (c ) 0.10, CHCl3, 22 °C). FAB-MS:
m/z 539 (M + Na)+. HR FAB-MS: m/z calcd for C34H36O5Na (M +
Na)+ 539.2433, found 539.2408. 1H NMR (300 MHz, CDCl3): δ 3.63
(d, J ) 5 Hz), 3.77 (dd, J ) 7, 1.5 Hz), 4.05 (m), 4.24 (m), 4.23 (d,
J ) 11 Hz), 6.32 (s, J ) 11 Hz), 4.35 (d, J ) 2.5 Hz), 4.53 (br s), 4.58
Colorless syrup, [R]D + 0.30° (c ) 0.54, MeOH, 22 °C). FAB-MS:
m/z 216 (M + Na)+. HR FAB-MS: m/z calcd for 12C613C1H12O6Na
(M + Na)+ 216.0565, found 216.0557. 1H NMR (CD3OD, 500 MHz):
δ 2.36 (dd, J ) 6, 14 Hz, H-6a), 2.82 (ddd, J ) 1, 5,5, 14 Hz, H-6b),
3.30 (d, J ) 11 Hz, H-7a), 3.69 (t, J ) 9.5 Hz, H-3), 3.72 (d, J ) 11
Hz, H-7b), 3.84 (d, J ) 9.5 Hz, H-4), 4.06 (ddd, J ) 1, 3, 9.5 Hz,
H-2). 13C NMR (CD3OD, 125 MHz): δ 46.3 (t, JC-C ) 41 Hz, C-6),
66.4 (t, JC-C ) 3.9 Hz, C-7), 73.7 (d), 70.6 (d), 77.3 (d), 79.6 (d, JC-C
) 37 Hz, C-2), 207.8 (s, 13C-labeled CdO).
2,3,4,7-Tetra-O-benzyl-[1-13C]valienone (15). To a solution of 14
(425 mg, 0.76 mmol) in TBME (30 mL) were added mesyl chloride
(0.58 mL) and triethylamine (2.1 mL), and the mixture was stirred under
argon at room temperature for 4 h. The reaction was quenched by ad-
dition of saturated aqueous Na2CO3 (12 mL), and stirring was continued
for 30 min. The reaction mixture was partitioned between TBME and
H2O. The organic layer was washed with 2 N HCl, saturated aqueous
NaHCO3, and brine. After the solution was dried over Na2SO4, the
solvent was evaporated. The residue was chromatographed on a silica
gel column with n-hexane/EtOAc (6:1) to give 15 (348 mg, 85%).
Colorless syrup, [R]D -9.6° (c ) 0.49, CHCl3, 22 °C). FAB-MS:
m/z 536 (M + H)+. HR FAB-MS: m/z calcd for 12C3413C1H35O5Na (M
1
+ H)+ 536.2518, found 536.2519. H NMR (300 MHz, CDCl3): δ
3.98-4.09 (m, 2-H, 3-H), 4.08 (br d, J ≈ 15 Hz, 7-Ha), 4.26 (br d, J
≈ 15 Hz, 7-Hb), 4.37 (br d, J ≈ 7 Hz, 4-H), 4.51 (s, Ph-CH2-), 4.66-
5.13 (Ph-CH2- × 3), 6.22 (dd, J ) 1.5, 2 Hz, 6-H), 7.21-7.45 (m,
C6H5 × 4). 13C NMR (75 MHz, CDCl3): δ 68.9 (t, JC-C ) 4.9 Hz,
C-7), 73.1, 74.4, 75.6, 75.7 (all t, Ph-CH2- × 4), 79.1 (d, C-4), 83.9
(d, JC-C ) 40.3 Hz, C-2), 84.9 (d, JC-C ) 2.5 Hz, C-3), 123.9 (d, JC-C
) 51.3 Hz, C-6), 159.1 (s, C-5), 127.7-128.5 (all d), 137.4-138.0
(all s, C6H5- × 4).
[1-13C]Valienone ([1-13C]-5). To a solution of 15 (145 mg, 0.27
mmol) in CH2Cl2 (7 mL) was added of 1.0 M BBr3 in hexane (1.6

6982 J. Am. Chem. Soc., Vol. 121, No. 30, 1999
Mahmud et al.
(d, J ) 12 Hz), 4.63 (d, J ) 12 Hz), 4.83 (d, J ) 12 Hz), 5.05 (d, J
) 12 Hz), 7.06-7.10 (m), 7.27-7.37 (m).
pressure. The extract was subjected to silica gel column chromatography
(n-hexane/EtOAc ) 6:1) to give a mixture of isomers 23a and 23b
(1.0 g, 62%). Compounds 23a and 23b were separated on an analytical
scale using reversed-phase HPLC (Econosil C18 10U 250 × 10 mm,
MeOH/H2O 85:15).
2,3,4,6-Tetra-O-benzyl-1-C-(dichloromethyl)-D-mannopyranose
(20). A 2.0 M solution of lithium diethylamide (LDA) in THF/heptane/
ethylbenzene (12.5 mL) was added dropwise to a solution of 19 (4.5
g) in CH2Cl2 (17 mL) with stirring under argon at -78 °C, and stirring
was continued for 1 h at the same temperature. The mixture was poured
into ice-cooled CH2Cl2 and 2 N HCl (150 mL of each). The organic
layer was washed with H2O, saturated aqueous NaHCO3, and brine,
and then dried over Na2SO4 and concentrated in vacuo. The residue
was chromatographed on a silica gel column with n-hexane/ethyl acetate
(4:1) to give 20 (4.7 g).
Colorless syrup, [R]D -14.0° (c ) 0.50, CHCl3, 22 °C). FAB-MS:
m/z 640 (M + NH4)+. HR FAB-MS: m/z calcd for C35H40NO635Cl2
(M + NH4)+ 640.2232, found 640.2203. 1H NMR (300 MHz, CDCl3):
δ 3.41 (br s), 3.70 (br s), 3.97 (s), 3.98 (d, J ) 7 Hz), 4.13 (m), 4.28
(d, J ) 2.5 Hz), 4.49-4.61 (m, 4H), 4.76 and 4.82 (AB q, J ) 12 Hz),
4.86 (d, J ) 11 Hz), 5.13 (d, J ) 11 Hz), 6.07 (d, J ) 1 Hz), 7.18-
7.37 (m, C6H5 × 4). 13C NMR (75 MHz, CDCl3): δ 60.4 (t), 69.0 (t),
72.8 (t), 73.3 (t), 74.4 (d), 75.1 (t), 75.9 (d), 81.9 (d), 98.3 (s), 127.1-
129.7 (all d) and 138.2-138.3 (all s) (C6H5 × 4).
23a: yellowish syrup. FAB-MS: m/z 621 (M + H)+. HR FAB-MS:
m/z calcd for C35H35O635Cl2 (M + H)+ 621.1810, found 621.1804. 1H
NMR (300 MHz, CDCl3): δ 3.54 (dd, J ) 11, 2.5 Hz, 7-Ha), 3.85 (d,
J ) 11 Hz, 7-Hb), 3.94 (d, J ) 2.5 Hz, 5-OH), 4.00 (t, J ) 4 Hz,
3-H), 4.51 (d, J ) 4 Hz, 4-H), 5.19 (d, J ) 4 Hz, 2-H), 4.36-4.87
(PhCH2- × 4), 7.09-7.36 (C6H5 × 4).
23b: yellowish syrup. FAB-MS: m/z 638 (M + NH4)+. HR FAB-
MS: m/z calcd for C35H38NO635Cl2 (M + NH4)+ 638.2076, found
1
638.2057. H NMR (300 MHz, CDCl3): δ 3.71 (dd, J ) 10, 2.5 Hz,
7-Ha), 4.02 (t, J ) 3.5 Hz, H-3), 4.11 (d, J ) 4 Hz, 4-H), 4.26 (br d,
J ) ca. 10 Hz, 7-Hb), 4.45 (d, J ) 3.5 Hz, 2-H), 4.31-4.81 (PhCH2-
× 4), 6.99-7.28 (C6H5 × 4).
2,3,4,7-Tetra-O-benzyl-2-epi-5-epi-[6-2H2]valiolone ([6-2H2]-24a)
and 2,3,4,7-Tetra-O-benzyl-2-epi-[6-2H2]valiolone ([6-2H2]-24b). To
a solution of the mixture of 23a and 23b (1 g, 1.6 mmol) in toluene
(10 mL) were added tributyltin deuteride (1.72 mL, 6.4 mmol) and
AIBN (104 mg, 0.64 mmol), and the mixture was refluxed for 2 h and
then cooled to room temperature. The products were extracted with
EtOAc (50 mL), and the organic solution was washed with 2 N HCl,
saturated aqueous NaHCO3, and brine. The organic solvent was
evaporated under reduced pressure, and the extract was purified over
a silica gel column (n-hexane/EtOAc ) 6:1-3:1) to give a mixture of
[6-2H2]-24a and [6-2H2]-24b (550 mg, 62%). The products were
separated by reversed-phase HPLC (Econosil C18 10U 250 × 10 mm,
MeOH/H2O 80:20) to give [6-2H2]-24a and [6-2H2]-24b in a 1.5:1 ratio.
[6-2H2]-24a: colorless syrup, [R]D -76.4° (c ) 0.45, CHCl3, 25 °C).
IR (NaCl): νmax (cm-1) 3463 (OH), 1729 (CdO). FAB-MS: m/z 577
(M + Na)+. HR FAB-MS: m/z calcd for C35H34O6D2Na (M + Na)+
577.2535, found 577.2510. 1H NMR (300 MHz, CDCl3): δ 3.22 (d, J
) 10 Hz, 7-Ha), 3.60 (dd, J ) 10, 2 Hz, 7-Hb), 3.96 (d, J ) 4 Hz,
4-H), 4.19 (dd, J ) 3.5, 4 Hz, 3-H), 4.32 (d, J ) 1.5 Hz, 5-OH), 4.45
(d, J ) 3.5 Hz, 2-H), 4.38-4.88 (PhCH2- × 4), 7.05-7.37 (C6H5 ×
4). 13C NMR (75 MHz, CDCl3): δ 60.3 (t, C-7), 72.4, 73.6, 73.7, 74.4
(all t, PhCH2 × 4), 76.0 (d, C-4), 79.4 (s, C-5), 81.1 (d, C-3), 81.2 (d,
C-2), 127.6-128.5 (all d) and 136.8-138.1 (all s, C6H5 × 4), 204.8
(s, C-1).
(1R)- and (1S)-Isomer of 2,3,4,6-Tetra-O-benzyl-1-C-(dichloro-
methyl)-D-mannitol (21). To a solution of 20 (4.5 g, 7.2 mmol) in
diglyme/THF (1:1; 58 mL) was added NaBH4 (2.25 g, 59.5 mmol),
and the mixture was stirred overnight at room temperature. The mixture
was concentrated and partitioned between EtOAc (200 mL) and water
(100 mL). The organic layer was washed with 2 N HCl, water, saturated
aqueous NaHCO3, and brine and dried over Na2SO4. The organic solvent
was evaporated under low pressure, and the residue was chromato-
graphed on a silica gel column (n-hexane/CH2Cl2/MeOH ) 10:4:1) to
give 21 (4.77 g).
Yellowish syrup. FAB-MS: m/z 625 (M + H)+. HR FAB-MS: m/z
calcd for C35H39O635Cl2 (M + H)+ 625.2124, found 625.2100. 1H NMR
(300 MHz, CDCl3): δ 3.52 (dd, J ) 6.3, 9.7 Hz), 3.61 (dd, J ) 5.5,
9.1 Hz), 3.68 (m), 3.77 (dd, J ) 4.4, 7.3 Hz), 3.89 (m), 4.07-4.39
(m), 4.43-4.79 (Ph-CH2- × 4), 5.68 (d, J ) 7.3 Hz), 6.15 (d, J ) 1.9
Hz), 7.17-7.30 (C6H5- × 4). 13C NMR (75 MHz, CDCl3): δ 70.0,
70.7, 71.0, 73.1, 73.1, 73.2, 73.3, 74.0, 74.3, 74.7, 75.1, 75.5, 76.1,
76.5, 78.5, 79.3, 79.8, 80.1, 80.7, 126.8-128.4 (all d) and 137.2-
137.9 (all s) (C6H5 × 4).
Dioxoheptose 22. To a solution of DMSO (1.4 mL, 26 mmol) in
CH2Cl2 (15 mL) was added trifluoroacetic anhydride (1.8 mL, 13 mmol)
in CH2Cl2 (7.5 mL) dropwise at -78 °C, and stirring was continued
for 30 min. Then a solution of 21 (2 g, 3.2 mmol) in CH2Cl2 (11 mL)
was added dropwise to the reaction mixture at -78 °C, and stirring
was continued for 1 h at the same temperature. To the reaction mixture
was added triethylamine (3.6 mL, 26 mmol) in CH2Cl2 (15 mL)
dropwise at -78 °C, and stirring was continued for 30 min. The reaction
mixture was allowed to warm to 0 °C and then poured into cold CH2-
Cl2 (40 mL) and water (20 mL), and stirring was continued for 30
min. The organic layer was washed with 2 N HCl, water, saturated
aqueous NaHCO3, and brine and then dried over Na2SO4, and the
solvent was evaporated in vacuo. The extract was subjected to silica
gel column chromatography (n-hexane/EtOAc ) 4:1) to give 22.
Yellowish syrup, [R]D -19.1° (c ) 0.16, CHCl3, 22 °C). FAB-MS:
m/z 638 (M + NH4)+. HR FAB-MS: m/z calcd for C35H38NO635Cl2
(M + NH4)+ 638.2076 found 638.2049. 1H NMR (300 MHz, CDCl3):
δ 4.15 (dd, J ) 4, 4.5 Hz, 3-H), 4.30 (dd, J ) 11, 1.5 Hz, 7-Ha), 4.47
(d, J ) 11 Hz, 7-Hb), 4.36 (d, J ) 4.5 Hz, 4-H), 4.58 (d, J ) 4 Hz,
2-H), 6.40 (s, 6-H). 13C NMR (75 MHz, CDCl3): δ 68.6 (t, C-7), 74.6
(t, -CHCl2), 79.5 (d, C-3), 81.8 (d, C-2 and C-4), 73.1, 73.2, 73.9,
94.6 (all t, PhCH2 × 4), 127.5-129.0, 136.1 (s), 136.2 (s), 136.4 (s),
137.1 (s) (C6H5 × 4), 194.7, 206.2 (both s, C-1 and C-5).
[6-2H2]-24b: colorless syrup, [R]D -48.1° (c ) 0.018, CHCl3, 25
°C). IR (NaCl): νmax (cm-1) 3449 (OH), 1728 (CdO). FAB-MS: m/z
572 (M + NH4)+. HR FAB-MS: m/z calcd for C35H38NO6D2 (M +
NH4)+ 572.2981, found 572.2954. 1H NMR (300 MHz, CDCl3): δ 2.61
(s, 5-OH), 3.23 (d, J ) 9 Hz, 7-Ha), 3.49 (d, J ) 9 Hz, 7-Hb), 3.92
(dd, J ) 8, 3 Hz, 3-H), 4.10 (d, J ) 3 Hz, 4-H), 4.30 (d, J ) 8 Hz,
2-H), 4.40-4.88 (PhCH2- × 4), 7.15-7.40 (C6H5 × 4). 13C NMR (75
MHz, CDCl3): δ 60.4 (t, C-7), 72.1, 72.5, 73.2, 75.1 (all t, PhCH2 ×
4), 74.0 (d, C-4), 77.4 (s, C-5), 79.5 (d, C-3), 81.0 (d, C-2), 127.6-
128.4 (all d) and 137.2-137.9 (all s, C6H5 × 4), 206.6 (s, C-1).
2-epi-5-epi-[6-2H2]Valiolone ([6-2H2]-10a) and 2-epi-[6-2H2]Valiolone
([6-2H2]-10b). To a solution of 75 mg of either [6-2H2]-24a or [6-2H2]-
24b in 95% aqueous ethanol (7.5 mL) was added wet 10% Pd/C (75
mg), and the mixture was stirred at room temperature under an H2
atmosphere for 16 h. The suspension was passed through a Celite
column to remove the catalyst and then filtered through a membrane
filter. The solvent was evaporated in vacuo to give pure [6-2H2]-10a
or [6-2H2]-10b in quantitative yield, respectively.
[6-2H2]-10a: white solid, [R]D + 24.2° (c ) 0.23, MeOH, 20 °C).
IR (NaCl): νmax (cm-1) 3380 (OH), 1722 (CdO). FAB-MS: m/z 217
(M + Na)+. HR FAB-MS: m/z calcd for C7H10O6D2Na (M + Na)+
1
217.0657, found 217.0666. H NMR (300 MHz, CD3OD): δ 3.44 (d,
J ) 11 Hz, 7-Ha), 3.66 (d, J ) 11 Hz, 7-Hb), 4.05 (d, J ) 4 Hz, 4-H),
4.29 (t, J ) 4 Hz, 3-H), 4.62 (d, J ) 4 Hz, 2-H). 13C NMR (75 MHz,
CD3OD): δ 67.5 (t, C-7), 70.7 (d, C-4), 76.0 (d, C-2), 79.7 (d, C-3),
81.5 (s, C-5), 209.9 (s, C-1).
(5R)- and (5S)-Isomer of (2S,3S,4S)-2,3,4-Tri-O-benzyl-5-C-
[(benzyloxy)methyl]-6,6-dichloro-1-oxo-2,3,4,5-cyclohexanetetrol (23a
and 23b). To a solution of 22 (1.6 g, 2.58 mmol) in toluene (50 mL)
were added sodium acetate (1.48 g, 18 mmol) and 18-crown-6 (24 mg,
0.09 mmol). The mixture was stirred at room temperature for 3 days.
The residue was filtered, and the filtrate was washed with 2 N HCl,
water, saturated aqueous NaHCO3, and brine. The organic layer was
dried over Na2SO4, and the solvent was evaporated under reduced
[6-2H2]-10b: white solid, [R]D + 20.4° (c ) 0.23, MeOH, 20 °C).
IR (NaCl): νmax (cm-1) 3380 (OH), 1722 (CdO). FAB-MS: m/z 217
(M + Na)+. HR FAB-MS: m/z calcd for C7H10O6D2Na (M + Na)+
1
217.0657, found 217.0653. H NMR (300 MHz, CD3OD): δ 3.41 (d,

Biosynthetic Path to the Valienamine Moiety of Acarbose
J. Am. Chem. Soc., Vol. 121, No. 30, 1999 6983
J ) 11 Hz, 7-Ha), 3.63 (d, J ) 11 Hz, 7-Hb), 3.98-4.02 (m, 3-H and
4-H), 4.31 (d, J ) 3.5 Hz, 2-H). 13C NMR (75 MHz, CD3OD): δ 67.0
(t, C-7), 72.4 (d, C-4), 74.8 (d, C-2), 76.2 (s, C-5), 76.6 (d, C-3), 209.7
(s, C-1).
exchange column chromatography (Bio-Rad, AG-1 × 8, formate form)
with elution with H2
collected, lyophilized, and subsequently purified on a Sephadex LH-
20 column (MeOH) to give the product in 70% yield.
O. The fractions containing [6-2H]-11 were
2,3,4,7-Tetra-O-benzyl-2-epi-[6-2H]valienone (25). To the solution
of a mixture of [6-2H2]-24a and [6-2H2]-24b (200 mg, 0.36 mmol) in
TBME (15 mL) were added mesyl chloride (0.28 mL, 3.6 mmol) and
triethylamine (1 mL, 7.2 mmol), and the reaction mixture was stirred
under argon at room temperature for 4 h. The reaction was quenched
by addition of saturated aqueous Na2CO3, and stirring was continued
for 30 min. The reaction mixture was partitioned between TBME and
H2O. The organic layer was washed with 2 N HCl, saturated aqueous
NaHCO3, and brine and dried over Na2SO4, and the solvent was
evaporated. The residue was chromatographed on a silica gel column
with n-hexane/EtOAc (6:1) to give 25 (90%).
Colorless syrup, [R]D -31.7° (c ) 0.56, MeOH, 20 °C). IR (NaCl):
νmax (cm-1) 3404 (OH), 1679 (enone). FAB-MS: m/z 176 (M + H)+.
HR FAB-MS: m/z calcd for C7H10D1O5 (M + H)+ 176.0669, found
176.0665. 1H NMR (300 MHz, CD3OD): δ 4.16 (dd, J ) 2.5, 3.5 Hz,
3-H), 4.19 (d, J ) 3.5 Hz, 4-H), 4.21 (d, J ) 17 Hz, 7-Ha), 4.40 (d, J
) 17 Hz, 7-Hb), 4.54 (d, J ) 2.5 Hz, 2-H). 13C NMR (75 MHz, CD3-
OD): δ 63.2 (t, C-7), 69.8 (d, C-3), 73.8 (d, C-4), 76.5 (d, C-2), 162.1
(s, C-5), 199.8 (s, C-1).
Acknowledgment. The authors thank Dr. Anneliese Crueger
and her colleagues at Bayer AG for cultures, materials, and
samples and for stimulating discussions, and Mr. William N.
Howald and Dr. Ross F. Lawrence, University of Washington
Mass Spectrometry Center, for their help in carrying out the
ES-MS and FAB-MS analyses. We also thank Takeda Chemical
Co., Osaka, Japan, and Prof. Kenneth Rinehart, University of
Illinois, Urbana, for generous gifts of valienamine and valida-
mycin A. This work was supported by Bayer AG, Wuppertal,
Germany. I.T. thanks the Deutsche Forschungsgemeinschaft for
a postdoctoral fellowship.
Colorless syrup, [R]D -79.2° (c ) 0.26, CHCl3, 20 °C). FAB-MS:
m/z 536 (M + H)+. HR FAB-MS: m/z calcd for C35H34O5D1 (M +
1
H)+ 536.2547, found 536.2528. H NMR (300 MHz, CDCl3): δ 4.04
(dd, J ) 2, 4.5 Hz, 3-H), 4.08 (s, 7-H2), 4.20 (d, J ) 4.5 Hz, 4-H),
4.36 (d, J ) 2 Hz, 2-H). 13C NMR (75 MHz, CDCl3): δ 69.3 (t, C-7),
74.7 (d, C-3), 77.2 (d, C-4), 78.7 (d, C-2), 72.5, 72.6, 72.8, 74.1 (all t,
PhCH2 × 4), 127.7-128.5, 137.3 (s), 137.5 (s), 137.7 (s), 137.9 (s)
(C6H5 × 4), 195.8 (s, C-1).
2-epi-[6-2H]Valienone ([6-2H]-11). To a solution of 25 (50 mg, 0.09
mmol) in CH2Cl2 (3 mL) was added 6 equiv of 1.0 M BBr3 in hexane.
After being stirred at -40 °C for 30 min and warming to 0 °C, the
reaction mixture was quenched with MeOH/H2O (1:2) and partitioned
between CH2Cl2 and H2O. The H2O layer was subjected to anion-
JA991102W