Structure Assignment, Total Synthesis, and Antiviral Evaluation of Cycloviracin B1

Article abstract of DOI:10.1021/ja036521e

The first total synthesis of the antivirally active glycolipid cycloviracin B₁ (1) is described. The approach is based on a two-directional synthesis strategy which constructs the C ₂-symmetrical macrodiolide core of the target by an

Full text of DOI:10.1021/ja036521e

Published on Web 10/02/2003
Structure Assignment, Total Synthesis, and Antiviral
Evaluation of Cycloviracin B1
Alois Fu¨rstner,*,† Martin Albert,† Jacek Mlynarski,† Maribel Matheu,† and
Erik DeClercq‡
Contribution from the Max-Planck-Institut fu¨r Kohlenforschung,
D-45470 Mu¨lheim/Ruhr, Germany; Katholieke UniVersiteit LeuVen,
Rega Institute for Medical Research, Minderbroedersstraat 10,
B-3000 LeuVen, Belgium
Received June 5, 2003; E-mail: fuerstner@mpi-muelheim.mpg.de.
Abstract: The first total synthesis of the antivirally active glycolipid cycloviracin B1 (1) is described. The
approach is based on a two-directional synthesis strategy which constructs the C2-symmetrical macrodiolide
core of the target by an efficient template-directed macrodilactonization reaction promoted by 2-chloro-
1,3-dimethylimidazolinium chloride 14 as the activating agent. Attachment of the lateral fatty acid chains to
the lactide core thus formed features not only one of the most advanced ligand-controlled addition reactions
of a functionalized dialkyl zinc reagent to a polyfunctional aldehyde, but also a highly demanding Julia-
Kocienski olefination of a tetrazolyl sulfone bearing electrophilic and base-labile â-hydroxy ester motifs. By
virtue of the flexibility of this synthesis plan, it was possible to prepare a series of macrodiolide cores
differing only in the absolute stereochemistry at the branching points as well as a host of model compounds
for the fatty acid appendices of cycloviracin. Comparison of these derivatives with the natural product allowed
us to establish the as yet unknown absolute stereochemistry of 6 chiral centers of 1 as (3R,19S,25R,3′R,-
17′S,23′R). Thereby, the 13C NMR shifts of the anomeric position of the â-glycosides residing at those
positions turned out to be excellent probes for the absolute configuration of the attached aglycones. The
concise set of data thus obtained also makes clear that the proposed structure of the fattiviracins, a seemingly
closely related family of glycoconjugates, is not matched by the published data. Finally, the biological activity
of synthetic 1 and some of the key intermediates obtained en route to this natural product was investigated,
showing that the entire construct is necessary for appreciable and selective antiviral activity.
Introduction
immunodeficiency virus type 1 (HIV-1). Although less potent
than acyclovir4 (50% antivirally effective concentration of 1
The paucity of effective medication against viral infections
together with the increasing resistance toward approved drugs
render the search, optimization, and clinical development of
novel antiviral agents highly desirable. In this context, two
families of natural products called cycloviracins1 and fattivi-
racins2 isolated from the soil microorganisms Kibdelospo-
rangium albatum so. nov. (R761-7)3 and Streptomyces microfla-
Vus No 2445, respectively, deserve attention as potential new
lead compounds because they have been reported to exhibit
activity against the human pathogens herpes simplex virus type
1 (HSV-1), influenza A virus, varicella-zoster virus, and human
against HSV-1: ∼5 µg/mL), the known biological and physical
properties of these complex glycolipids prompt more detailed
investigations. The lack of cytotoxicity against Vero cells (50%
cytotoxic concentration, >400 mg/mL) and the lack of activity
against bacteria, yeasts, and fungi at concentrations of 100 µg/
mL suggest antiviral specificity. Preliminary data indicate that
these compounds diminish the infectivity of the viruses by
inhibiting their entry into the host cells;5 importantly, this
specific mode of action is complementary to that of most
approved antiviral drugs. Finally, the good solubility of these
glycoconjugates in aqueous media constitutes an additional
bonus.
† Max-Planck-Institut fu¨r Kohlenforschung.
‡ Katholieke Universiteit Leuven.
Although the constitutions of the cycloviracins and fattivi-
racins have been elucidated by extensive NMR spectroscopic
investigations, several structural details remain unclear and even
somewhat puzzling. As can be seen from the proposed structures
(1) (a) Tsunakawa, M.; Komiyama, N.; Tenmyo, O.; Tomita, K.; Kawano, K.;
Kotake, C.; Konishi, M.; Oki, T. J. Antibiot. 1992, 45, 1467. (b) Tsunakawa,
M.; Kotake, C.; Yamasaki, T.; Moriyama, T.; Konishi, M.; Oki, T. J.
Anitbiot. 1992, 45, 1472. (c) Tsunakawa, M.; Yamasaki, T.; Tomita, K.;
Osamu, T. EP 0 472 005 A1 (1991-07-26).
(2) (a) Uyeda, M.; Yokomizu, K.; Miyamoto, Y.; Habib, E.-S. E. J. Antibiot.
1998, 51, 823. (b) Yokomizo, K.; Miyamoto, Y.; Nagao, K.; Kumagae,
E.; Habib, E.-S. E.; Suzuki, K.; Harada, S.; Uyeda, M. J. Antibiot. 1998,
51, 1035. (c) Habib, E.-S. E.; Yokomizo, K.; Murata, K.; Uyeda, M. J.
Antibiot. 2000, 53, 1420. (d) Habib, E.-S. E.; Yokomizo, K.; Suzuki, K.;
Uyeda, M. Biosci., Biotechnol., Biochem. 2001, 65, 861.
(4) (a) Schaeffer, H. J.; Beauchamp, L.; de Miranda, P.; Elion, G. B.; Bauer,
D. J.; Collins, P. Nature 1978, 272, 583. (b) Elion G. B.; Furman, P. A.;
Fyfe, J. A.; de Miranda, P.; Beauchamp, L.; Schaeffer, H. J. Proc. Natl.
Acad. Sci. U S. A. 1977, 74, 5716. (c) Whitley, R. J.; Gnann, J. W. N.
Engl. J. Med. 1992, 327, 782.
(5) Habib, E.-S. E.; Yokomizo, K.; Nagao, K.; Harada, S.; Uyeda, M. Biosci.,
Biotechnol., Biochem. 2001, 65, 683.
(3) For information on the producing strain, see: Tomita, K.; Hoshino, Y.;
Miyaki, T. Int. J. Syst. Bacteriol. 1993, 43, 297.
9
13132
J. AM. CHEM. SOC. 2003, 125, 13132-13142
10.1021/ja036521e CCC: $25.00 © 2003 American Chemical Society

Antiviral Evaluation of Cycloviracin B
A R T I C L E S
1
Scheme 1 a
a Common structural motif of the cycloviracins and the proposed structure
of all fattiviracins known to date.
data. Finally, cycloviracin B1 itself and some of the key synthetic
intermediates obtained during its total synthesis were evaluated
for their antiviral activities. From the results obtained, it must
be concluded that neither a stripped-down macrodiolide core
nor the segments representing the lateral chains exhibit any
appreciable activity, while specific antiviral effects for the entire
construct are corroborated by these preliminary investigations.
Results and Discussion
Retrosynthetic Analysis and Strategic Considerations. In
view of the size and complexity of the targets as well as the
structural ambiguities at the outset of the project, it was clear
that only a highly convergent and inherently flexible approach
might allow to reach the objectives set forth above.
of the two prototype members 1 and 2, these compounds seem
to differ only in the length and glycosylation pattern of the lateral
fatty acid residues while sharing a common macrodiolide core.
A closer look at the reported NMR data, however, reveals a
rather suspicious detail: in contrast to the cycloviracins, which
give rise to only one set of signals for both units forming this
central motif, all members of the fattiviracin family invariably
show inequivalent subunits (cf. generalized structure in Scheme
1).6 This difference can hardly be explained in View of the
oVerall resemblance of both series. Therefore, it is an imperative
prelude to any further study to clarify this structural ambiguity
and to determine the as yet unknown absolute stereochemistry
of the six chiral centers (*) along the alkyl chains of either 1 or
2.
As part of a long-term project on bioactive glycoconjugates,7-9
we initiated an extensive program addressing these issues. Our
initial goal was the unambiguous determination of their actual
structure by a preparative approach that should ultimately enable
total synthesis. As will be shown below, this project was largely
successful. It has not only led to the elucidation of the previously
unknown stereostructure of cycloviracin B1 1, but also allowed
us to conquer this demanding glycolipid by a straightforward
and high-yielding route.10 However, the highly consistent picture
derived from these studies reveals a serious mismatch between
the proposed constitution of the fattiviracins and the published
The key design element of this plan exploits the hidden
symmetry of these glycolipids. The fact that the individual
subunits of the macrodiolide core of cycloviracin B1 are
indistinguishable by NMR at 600 MHz is best explained by
assuming a C2-symmetric structure in this part of the molecule.
This then suggests to implement a two-directional synthesis
strategy11 en route to 1 to minimize the preparative efforts and
to ensure maximum flexibility during fragment coupling (Scheme
2). Given the different length of its two lateral chains and their
discrete hydroxylation pattern, however, this plan bears a
considerable risk for the final assembly stages. While established
methodology should allow to control the configuration of the
-OH group at C-17′ generated by coupling of fragments A and
B (M ) metal), the formation of the C-C bond at the symmetry
related C-17/C-18 position joining segments A and D is highly
problematic. Any attempt to convert fragment D into a carbon
nucleophile (Y ) metal) must result in reductive elimination
with loss of the adjacent glycoside; the inverse maneuver
implying the conversion of the entire lactide A into a suitable
nucleophile (X ) metal) is similarly endangered by the presence
of electrophilic and C-H acidic sites in this molecule. More
specifically, deprotonation R to the lactones (i.e., at C-2 and/or
C-2′) easily engenders the opening of the macrocycle by
expulsion of the adjacent sugar and formation of an R,â-
unsaturated ester. Only a highly stabilized nucleophile X, if any,
might kinetically resist these self-destructive pathways.
(6) This applies to all 13 individual fattiviracins known to date, cf. refs 2 and
5.
(7) (a) Fu¨rstner, A.; Jeanjean, F.; Razon, P. Angew. Chem., Int. Ed. 2002, 41,
2097. (b) Fu¨rstner, A.; Jeanjean, F.; Razon, P.; Wirtz, C.; Mynott, R. Chem.
Eur. J. 2003, 9, 307. (c) Fu¨rstner, A.; Jeanjean, F.; Razon, P.; Wirtz, C.;
Mynott, R. Chem. Eur. J. 2003, 9, 320.
(8) (a) Fu¨rstner, A.; Mu¨ller, T. J. Am. Chem. Soc. 1999, 121, 7814. (b) Fu¨rstner,
A.; Mu¨ller, T. J. Org. Chem. 1998, 63, 424. (c) Fu¨rstner, A.; Radkowski,
K.; Grabowski, J.; Wirtz, C.; Mynott, R. J. Org. Chem. 2000, 65, 8758.
(d) Lehmann, C. W.; Fu¨rstner, A.; Mu¨ller, T. Z. Kristallogr. 2000, 215,
114.
(9) (a) Fu¨rstner, A.; Konetzki, I. J. Org. Chem. 1998, 63, 3072. (b) Fu¨rstner,
A.; Konetzki, I. Tetrahedron 1996, 52, 15071.
Despite this considerable uncertainty, the strategic discon-
nections shown in Scheme 2 are tempting for the significant
overall decrease in complexity which they entail. Thus, lactide
A as the key component of this synthesis plan might be
assembled by a template-directed cyclodimerization process12,13
if one assumes that the specific array of O-atoms in the core
(10) For preliminary communications, see: (a) Fu¨rstner, A.; Albert, M.;
Mlynarski, J.; Matheu, M. J. Am. Chem. Soc. 2002, 124, 1168. (b) Fu¨rstner,
A.; Mlynarski, J.; Albert, M. J. Am. Chem. Soc. 2002, 124, 10274.
(11) (a) Poss, C. S.; Schreiber, S. L. Acc. Chem. Res. 1994, 27, 9. (b) Magnuson,
S. R. Tetrahedron 1995, 51, 2167.
9
J. AM. CHEM. SOC. VOL. 125, NO. 43, 2003 13133

A R T I C L E S
Fu¨rstner et al.
Scheme 2
Scheme 3 a
a Conditions: [a] lithio tert-butyl acetate, THF, -78 °C, 61%; [b] [(R)-
BINAP‚RuCl2]2‚NEt3 cat., H2 (15 atm), MeOH, 70 °C, 88%; [c] TBDPSCl,
imidazole, DMF, 81%; [d] NaH, BnBr, DMF, 77%; [e] Ac2O, NaOAc,
H2SO4 cat., 95%; [f] H2NNH2‚HOAc, DMF, r.t., 81%; [g] NaH, CH2Cl2,
Cl3CCN, 74%; [h] BF3‚Et2O, MS 4 Å, CH2Cl2, -78 °C f r.t., 62%; [i] (i)
F3CCOOH, CH2Cl2; (ii) NH3/MeOH, 74%.
region endows the molecule with some degree of ionophoric
character.14 This then allows to deconvolute the target to a single
precursor E which derives from a â-selective glycosylation of
the rather simple aldol derivative F with a suitably protected
glucosyl donor G (LG ) leaving group).
The appeal of this convergent synthesis plan becomes
apparent on comparison with a stepwise and hence linear
construction of the macrodiolide core as the alternative scenario
which must come into play should any of the premises described
above prove invalid. In this context, a previous study directed
toward 1 has to be mentioned which pursued such a stepwise
route.15 It may not come as a surprise that this approach
ultimately turned out to be prohibitively lengthy to allow for
systematic variations, provided only a highly truncated version
of the cycloviracin core that cannot be elaborated any further,
and therefore remained inconclusive with respect to the absolute
stereochemistry at the branching points C-3/C-3′.15
acid chains of 1 at the glycosidic bond and the symmetry related
C-17/C-17′ positions. A ring-opening Claisen condensation of
well-accessible pentadecanolide 316 with lithio tert-butyl acetate
solves this problem very well, giving rise to â-keto ester 4 in
61% yield on a 15 g scale together with small amounts of the
corresponding tertiary alcohol formed by double addition of the
nucleophile to the lactone (Scheme 3).17 Subsequent hydrogena-
tion of 4 in the presence of Noyori’s catalyst [((R)-BINAP)-
18
RuCl2]2‚NEt3 affords the corresponding diol 5 in excellent
yield, which is regioselectively silylated at the terminal position
with TBDPSCl in the presence of imidazole to give the (R)-
configured product 6 (ee ) 98%) ready for subsequent glyco-
sylation.19
C2-Symmetrical Core Structures by Template-Directed
Macrodilactonization. To reduce the synthesis blueprint shown
in Scheme 2 to practice, it is necessary to develop an efficient
approach to the common aldol synthon F which derives from
the envisaged disconnection of the nonequivalent lateral fatty
After some experimentation with different glycosyl donors,
trichloroacetimidate 1120 was selected as a suitable precursor
(16) For a shortcut synthesis, see: (a) Fu¨rstner, A.; Langemann, K. J. Org. Chem.
1996, 61, 3942. (b) Fu¨rstner, A.; Langemann, K. Synthesis 1997, 792. (c)
Fu¨rstner, A.; Ackermann, L.; Beck, K.; Hori, H.; Koch, D.; Langemann,
K.; Liebl, M.; Six, C.; Leitner, W. J. Am. Chem. Soc. 2001, 123, 9000.
(17) Only very few examples of ring-opening Claisen condensations are known
using four- to six-membered lactones as the substrates, cf: (a) Bone, E.
A.; Davidson, A. H.; Lewis, C. N.; Todd, R. S. J. Med. Chem. 1992, 35,
3388. (b) Yamaguchi, M.; Nakamura, S.; Okuma, T.; Minami, T.
Tetrahedron Lett. 1990, 31, 3913. (c) Shimizu, M.; Ishii, K.; Fujisawa, T.
Chem. Lett. 1997, 8, 765.
(12) Fu¨rstner, A. In Templated Organic Synthesis; Diederich, F., Stang, P. J.,
Eds.; Wiley-VCH: Weinheim, 2000; pp 249-273.
(13) (a) For recent studies on the synthesis of macrodiolides by nontemplated
cyclodimerization reactions, see the following leading references and a
comprehensive compilation of pertinent literature: Su, Q.; Beeler, A. B.;
Lobkovsky, E.; Porco, J. A., Jr.; Panek, J. S. Org. Lett. 2003, 5, 2149. (b)
For the synthesis of a natural product from our laboratory by an RCM-
based cyclodimerization process, see: Fu¨rstner, A.; Thiel, O. R.; Acker-
mann, L. Org. Lett. 2001, 3, 449.
(18) For comprehensive treatises, see: (a) Noyori, R. Asymmetric Catalysis in
Organic Synthesis; Wiley: New York, 1994. (b) Ohkuma, T.; Noyori, R.
In ComprehensiVe Asymmetric Catalysis; Jacobsen, E. N., Pfaltz, A.,
Yamamoto, H., Eds.; Springer: Berlin, 1999; Vol. 1; pp 199-246.
(19) The order of silylation and hydrogenation can also be inversed. In this
case, however, the overall yield of product 6 is somewhat lower due to
partial cleavage of the terminal TBDPS group during the asymmetric
reduction step.
(14) For a study on cyclic oligosaccharides acting as crown ethers, see: Shizuma,
M.; Kadoya, Y.; Takai, Y.; Imamura, H.; Yamada, H.; Takeda, T.; Arakawa,
R.; Takahashi, S.; Sawada, M. J. Org. Chem. 2002, 67, 4795 and literature
cited therein.
(15) Velarde, S.; Urbina, J.; Pena, M. R. J. Org. Chem. 1996, 61, 9541.
(20) Wang, Y.; Mao, J.; Cai, M. Synth. Commun. 1999, 29, 2093.
9
13134 J. AM. CHEM. SOC. VOL. 125, NO. 43, 2003

Antiviral Evaluation of Cycloviracin B
A R T I C L E S
1
Table 1. Effect of Different Additives on the Product Distribution in
the Macrodilactonization Reaction of Hydroxy Acid 13. Product
Distributions Were Determined by HPLC without Correction for the
Response Factors of the Different Compounds
Scheme 4 a
no.
molarity
additive
18
15
other (total)
1
2
3
4
5
6
20 mM
20 mM
20 mM
30 mM
50mM
20 mM
28%
24%
20%
24%
23%
24%
37%
52%
75%
68%
68%
48%
35%
24%
5%
8%
9%
NaH
KH
KH
KH
Cs2CO3
28%
for the glucose units to be embedded into the lactide core. This
compound bears orthogonal protecting groups and is readily
available in only four high-yielding operations from commercial
laevoglucosane 7 as shown in Scheme 3.21 Since the benzyl
group at the C-2 position of 11 will not exert anchimeric
assistance in the glycosylation event,22,23 high selectivity in favor
of the required â-glucoside must be ensured by the proper choice
of promotor and solvent. Specifically, exposure of 6 and 11 in
CH2Cl2 at low temperature to BF3‚Et2O in the presence of
molecular sieves provides â-glucoside 12 in an appreciable 62%
isolated yield (87% based on recovered starting material), with
the R/â-ratio being 1:6; the anomers are readily separable by
flash chromatography. A similar result (68% yield, R/â ) 1:5)
is obtained by performing the reaction with TMSOTf in CH2-
Cl2/MeCN (1:1) by taking advantage of the â-directing effect
of the nitrile cosolvent.24 Subsequent cleavage of the tert-butyl
ester in 12 with trifluoroacetic acid followed by saponification
of the residual acetate with NH3/MeOH leads to hydroxy acid
13 in 74% yield over both steps and sets the stage for the
envisaged cyclodimerization reaction.
a Conditions: [a] 2-chloro-1,3-dimethylimidazolinium chloride 14, DMAP,
KH, CH2Cl2, 71%; [b] TBAF, THF, 92%; [c] tBuPh2SiCl, Et3N, CH2Cl2,
85%.
the presence of DMAP at 0 °C affords a separable mixture of
cyclic monomer 18 (28%), the desired cyclic dimer 15 (37%),
and several oligomeric byproducts (combined yield ca. 35%,
cf. Table 1, entry 1).
Preliminary experiments using DCC/DMAP as the activating
agent in dilute CH2Cl2 solution were unsuccessful. In this
context it is also important to note that Pen˜a et al. reported on
attempted cyclodimerizations of model compounds with the aid
of biocatalyts which were so unrewarding that they opted for a
stepwise approach en route to the truncated cycloviracin core
model mentioned above.15
Although this product distribution is far from optimal, it
shows that the cyclodimer can be formed in a single step.
Therefore, systematic investigations were carried out to improve
on this result by templating the cyclization reaction with the
aid of suitable additives.12 In line with our expectations, this
process was found to be highly responsive to admixed alkali
metal cations (Scheme 4 and Table 1). Gratifyingly, the addition
of KH not only led to a substantially increased reaction rate
but also to a significant improvement of the product distribution
in favor of 15; under optimized conditions, this product is
obtained in 71% isolated yield (75% HPLC, cf. Table 1, entry
3). Since the effect exerted by NaH or Cs2CO3 is much less
pronounced, this outcome is deemed to reflect the ability of
the K+ cation to preorganize the cyclization precursor 13 for
directed macrodilactonization. Representative results are sum-
marized in Table 1.
The NMR spectra of 15, the desilylated alcohol 16, as well
as the unsymmetrical derivative 17 (R1 * R2) all match those
reported for 11 very well, suggesting that the absolute stereo-
chemistry of cycloviracin B1 is (3R,3′R). To corroborate this
notion, however, it was necessary to prepare the analogous
(3S,3′S)-configured lactide 22 and to compare its spectroscopic
properties with those of the natural product. The synthesis of
compound 22 is highly straightforward. It follows the route
Despite these pitfalls, however, the original plan was pursued
further in the hope that a more efficient activator than DCC
would allow to effect the desired esterification/lactonization
tandem of the hydroxy acid 13. Encouraging observations were
made when compound 13 was exposed to 2-chloro-1,3-
dimethylimidazolinium chloride 14, a commercially available
dehydrating agent.25-27 Specifically, reaction of 13 with 14 in
(21) (a) Bourke, D. G.; Collins, D. J.; Hibberd, A. I.; McLeod, M. D. Aust. J.
Chem. 1996, 49, 425. (b) Hoch, M.; Heinz, E.; Schmidt, R. R. Carbohydr.
Res. 1989, 191, 21.
(22) General discussions: (a) Paulsen, H. Angew. Chem., Int. Ed. Engl. 1982,
21, 155. (b) Schmidt, R. R. In ComprehensiVe Organic Synthesis; Trost,
B. M., Fleming, I., Eds.; Pergamon: Oxford, 1991; Vol. 6; pp 33-61. (c)
Toshima, K.; Tatsuta, K. Chem. ReV. 1993, 93, 1503.
(23) Pertinent reviews on the trichloroacetimidate method: (a) Schmidt, R. R.
Angew. Chem., Int. Ed. Engl. 1986, 25, 212. (b) Schmidt, R. R.; Kinzy,
W. AdV. Carbohydr. Chem. Biochem. 1994, 50, 21.
(24) Schmidt, R. R.; Behrendt, M.; Toepfer, A. Synlett 1990, 694.
(25) (a) Isobe, T.; Ishikawa, T. J. Org. Chem. 1999, 64, 5832. (b) Isobe, T.;
Ishikawa, T. J. Org. Chem. 1999, 64, 6984. (c) Isobe, T.; Ishikawa, T. J.
Org. Chem. 1999, 64, 6989. (d) See also: Fujisawa, T.; Mori, T.; Fukumoto,
K.; Sato, T. Chem. Lett. 1982, 1891.
(26) For a related cyclotrimerization en route to the glycolipid anthrobacilin A,
see: Garcia, D. M.; Yamada, H.; Hatakeyama, S.; Nishizawa, M.
Tetrahedron Lett. 1994, 35, 3325.
(27) For an application of these reagents in organometallic chemistry, see:
Fu¨rstner, A.; Seidel, G.; Kremzow, D.; Lehmann, C. W. Organometallics
2003, 22, 907.
9
J. AM. CHEM. SOC. VOL. 125, NO. 43, 2003 13135

A R T I C L E S
Fu¨rstner et al.
Scheme 5 a
Scheme 7 a
a Comparison of the shifts of the anomeric carbon atoms of the (3R,3′R)
and the (3S,3′S)-configured core structures.
Scheme 6 a
a Conditions: [a] (i) [(S)-BINAP‚RuCl2]2‚NEt3 cat., H2 (15 atm), MeOH,
70 °C, 92%; (ii) TBDPSCl, Et3N, CH2Cl2, 97%; [b] imidate 11, TMSOTf
cat., MeCN/CH2Cl2 (1:1), -50 °C f r.t., 63% (+ 19% of the R-anomer
separated by flash chromatography); [c] (i) NH3 in MeOH, 83%; (ii)
F3CCOOH, CH2Cl2; [d] 2-chloro-1,3-dimethylimidazolinium chloride 14
(2.5 equiv), DMAP (3 equiv), KH (2.5 equiv), CH2Cl2, 41%.
a Conditions: [a] Diisopropyl carbodiimide (DIC), DMAP, CH2Cl2, 83%;
[b] F3CCOOH/CH2Cl2 (1:10), 77%; [c] 2,4,6-trichlorobenzoic acid chloride,
Et3N, THF; then DMAP, toluene, reflux, 89%.
Fattiviracin Problem. Preparation of the (3R,3′S)-Con-
figured Macrodiolide. In view of the foregoing, however, a
rather puzzling situation as to the actual structure of the
fattiviracins ensues. In contrast to the cycloviracins, which give
rise to only one set of signals for both units forming the
macrodiolide ring,1 all members of the fattiviracin family
invariably show inequivalent subunits (cf. Scheme 1).2,6 Specif-
ically, the anomeric positions for fattiviracin FV-8 2 as a
prototype example resonate at 104.0 and 105.0 ppm, respec-
tively, a fact that seems neither consistent with the situation
found in cycloviracin (i.e. 3R,3′R) nor with a (3S,3′S) config-
uration at the branching points. To complicate matters even
described above, simply using (S)-BINAP instead of (R)-BINAP
as the ligand in the hydrogenation of keto ester 4 (Scheme 6).
The glycosidation of the resulting aglycone 19 with trichloro-
acetimidate 11 is best achieved with TMSOTf in CH2Cl2/MeCN
(1:1) at -50 °C. As expected, the subsequent cyclodimerization
of the diastereomeric hydroxy acid 21 derived thereof occurs
smoothly in the presence of 2-chloro-1,3-dimethylimidazolinium
chloride 14,25 DMAP, and KH, albeit the isolated yield of lactide
22 is somewhat lower (41%).
Notably, however, the spectral data of 22 are significantly
different, with the high-field shift of the anomeric C-atoms at
δC ) 100.3 ppm being particularly diagnostic. These marked
shift differences of the anomeric carbon atoms in lactides 15-
17 on one hand and compound 22 on the other hand (∆δ ∼ 5
ppm, cf. Scheme 5) allow the assignment of the stereochemistry
at the branching points in the cycloViracin core as 3R,3′R.10a,28
In short, the convergent approach to the C2-symmetrical
macrodiolides outlined above is not only highly productive but
also inherently flexible and therefore allowed to determine the
previously unknown configuration of the branching points in
cycloviracin B1 1 without undue preparative efforts. It favorably
compares with the previously published route15 to a more
truncated version of the core in all relevant respects, in particular
with regard to its inherent “economy of steps” as a prime
indicator for high efficiency.29
(28) Although the closely related shift values of the glucokinase activating
macrodiolide glucolipsin A seem to suggest that this particular compounds
which contains an extra methyl branch at the adjacent positionsmight
also be (3R,3′R) configured, a recent total synthesis by our group shows
that glucolipsin incorporates in fact a (2R,3S)-configured syn aldol entity.
For the isolation of glucolipsin, see: Qian-Cutrone, J.; Ueki, T.; Huang,
S.; Mookhtiar, K. A.; Ezekiel, R.; Kalinowski, S. S.; Brown, K. S.; Golik,
J.; Lowe, S.; Pirnik, D. M.; Hugill, R.; Veitch, J. A.; Klohr, S. E.; Whitney,
J. L.; Manly, S. P. J. Antibiot. 1999, 52, 245.
(29) For a discussion, see: Fu¨rstner, A. Synlett 1999, 1523.
9
13136 J. AM. CHEM. SOC. VOL. 125, NO. 43, 2003

Antiviral Evaluation of Cycloviracin B
A R T I C L E S
1
Scheme 8 a
further, these data also seem to exclude the only remaining
option of a (3R,3′S)-configured core. Although such a stereo-
chemical hermaphrodite would be highly surprising from a
biosynthetic viewpoint as well, we saw no way except synthesis
to rigorously exclude this possibility. Since a cyclodimerization
approach is obviously unsuitable in this case, a stepwise
macrocyclization protocol had to be pursued (Scheme 7).
For this purpose, the acetate group in 20 is cleaved with NH3/
MeOH to give the (3S)-configured alcohol 23, which is esterified
in 83% yield with the (3R)-configured synthon 2430 in the
presence of DIC/DMAP. Subsequent treatment of compound
25 with F3CCOOH in CH2Cl2 leads to the concomitant cleavage
of the tert-butyl ester () R2) and the tert-butyldimethylsilyl ether
() R1) while leaving the tert-butyldiphenylsilyl group at the
terminus of one of the lipidic chains intact. The resulting
hydroxy acid 26 is then cyclized to the desired lactide 27 by a
Yamaguchi lactonization.31
The anomeric centers in the unsymmetrical lactide 27 thus
formed resonate at δ ) 99.1 and 104.5 ppm, respectively, thus
showing that the differently configured subunits can be clearly
distinguished by NMR. More important, however, is the striking
mismatch with the data reported for 2 (δ ) 104.0 and 105.0).2
They are not consistent with the proposed constitution implying
a lactide unit similar to the one found in cycloviracin 1,
independent of whether one assumes an (R,R)-, (S,S)-, or (R,S)-
configuration at the branching points. It must be concluded that
either the proposed structure of the fattiviracins or the reported
set of data are erroneous and need revision in the future.
a Conditions: [a] m-chloroperbenzoic acid, KF, CH2Cl2, cf. ref. 34; [b]
(2R)- or (2S)-hexanol, ZnCl2, THF, 32-36%; [c] (i) MeI, NaH, DMF; (ii)
H2, Pd/C, EtOH, 66-70% (over both steps).
synthesis. Since the shift of the anomeric positions of the
glucosides were found to be highly diagnostic for the config-
uration of the attached lipidic aglycones (see above), it was
envisaged to use this effect33 also for investigating the more
distal sites residing on the lateral fatty acid appendices in 1.
For this purpose, glycal epoxide 2934 was reacted with (2R)-
and (2S)-hexanol in the presence of ZnCl2 as the promoter to
give glycosides 30 and 32, respectively, albeit in moderate
yield.35 Both compounds were O-methylated under standard
conditions at the liberated O-2 site prior to hydrogenolytic
cleavage of the benzyl ether groups. As can be seen from
Scheme 8, the resulting glycosides 31 and 33 differing only in
the configuration of their aglycone are easily distinguished by
NMR. Most notable is the fact that the anomeric center of 31
linked to the (R)-configured alcohol resonates at δC ) 101.9
ppm, a shift which correlates exceedingly well with that of the
distal 2-O-methyl glucoside of cycloviracin B1 (δC ) 101.8
ppm), thus making a (25R)-configuration in the natural product
highly likely. Moreover, the comparison between 33 and 1
provides a first indication that the chiral centers at C-17′ and
C-19 of the latter might be (S)-configured.
Although a conclusive answer to this puzzle will require
further preparative and spectroscopic studies, one may speculate
about a different connectivity pattern of the individual subunits
involving an -OH group further down the fatty acid chain to
form an expanded macrolactone ring, which might perhaps
explain the observed data. Interestingly, a new family of
glycolipids with antiviral properties was recently disclosed in
the patent literature which contains such a giant entity (e.g.,
BA-2836, also called “macroviracin”).32 The possible relation-
ship between these intriguing macrocycles and the fattiviracins
is subject to further investigations in our laboratory.
Configuration of the Distal Chiral Centers in Cycloviracin.
Model Studies. After having established the absolute config-
uration of the branching points in cycloviracin as (3R,3′R),
model studies were undertaken to gain further insights into the
stereostructure of this target prior to launching the actual total
It seemed appropriate, however, to corroborate this notion
by further model studies since these stereocenters reside within
the most “symmetrical” segments of the hydrocarbon chains,
(33) For methods allowing the determination of the configuration of secondary
alcohols by using sugar derivatives as derivatizing agents, see the following
for leading references and literature cited therein: (a) Trujillo, M.; Morales,
E. Q.; Vazquez, J. T. J. Org. Chem. 1994, 59, 6637. (b) Kobayashi, M.
Tetrahedron 2002, 58, 9365.
(34) Bellucci, G.; Chiappe, C.; D’Andrea, F. Tetrahedron: Asymmetry 1995,
6, 221.
(30) The synthesis of this compound is outlined in the Supporting Information.
(31) Inanaga, J.; Hirata, K.; Saeki, H.; Katsuki, T.; Yamaguchi, M. Bull. Chem.
Soc. Jpn. 1979, 52, 1989.
(32) (a) Hyoda, T.; Tsuchira, Y.; Sekine, A.; Amano, T. Jpn. Kokai Tokkyo
Koho Jpn. Pat. 11246587 (1999-09-14). (b) See also: Takahashi, S.;
Hosoya, M.; Koshino, H.; Nakata, T. Org. Lett. 2003, 5, 1555.
(35) (a) Halcomb, R. L.; Danishefsky, S. J. J. Am. Chem. Soc. 1989, 111, 6661.
(b) Review: Danishefsky, S. J.; Bilodeau, M. T. Angew. Chem., Int. Ed.
Engl. 1996, 35, 1380.
9
J. AM. CHEM. SOC. VOL. 125, NO. 43, 2003 13137

A R T I C L E S
Fu¨rstner et al.
Scheme 9 a
Scheme 11
a Conditions: [a] (i) Mg, THF; (ii) CuCl(COD) cat., (R)-propene oxide,
70%; [b] BnBr, NaH, 73%; [c] TBAF, THF, 93%; [d] I2, imidazole, PPh3,
89%; [e] Et2Zn (excess) CuCN cat.
Scheme 10 a
material with trichloroacetimidate 41 mediated by TMSOTf in
CH2Cl2/MeCN gives the corresponding â-glycoside 42 in 92%
isolated yield (Scheme 10).
The anomeric center of this compound resonates at δC
)
102.7 ppm, while the shift of the corresponding (S)-configured
analogue 43 is δC ) 103.0 ppm. The more advanced models
44 and 45 prepared along similar lines30 can also be clearly
distinguished by NMR (∆δC ) 0.4 ppm). Although the observed
differences are small, they are significant and likely character-
istic. Most important is the fact that in all pairs of stereoisomers
investigated (31/33, 42/43, 44/45) the anomeric center of the
compound bearing the (S)-configured aglycone invariably
resonates at a shift that is (almost) identical to the corresponding
signal in 1 (δC ) 103.1).39 Moreover, an NMR inspection of a
1:1 mixture of 42/43 or 44/45 showed that the individual
compounds can be clearly discerned. Therefore, it is highly
likely that a direct comparison of the spectra of synthetic 1 with
(19S,17′S)-configuration40 with that of authentic cycloviracin
B1 will result in a visible mismatch of the patterns if the
stereochemistry at those centers does not correspond to that of
the natural product.
Biosynthetic considerations also strongly advocate an (S)-
configuration at those sites. It is known that the lipidic segments
of these antiviral glycolipids derive from acetate units via the
usual polyketide pathway.2d Therefore, the most probable
product is the one in which any leftover hydroxyl groups reside
on the same side of the aliphatic chain.40
On the basis of these results and considerations, one can (i)
assign the (R)-configuration to the branching points at C-3 and
C-3′ in cycloviracin B1, (ii) ascribe the (R)-configuration to the
distal sites at C-23′ and C-25, and (iii) make the (S)-configu-
ration for the remaining stereocenters at C-17′ and C-19 highly
probable. These conclusions set the basis for the total synthesis
of this bioactive target molecule summarized below.
a Conditions: [a] dodecanal, Ti(OiPr)4, (R,R)-39, toluene, 65%; [b]
TMSOTf cat., CH2Cl2/MeCN (1:1), 92%.
with the stereochemical determinants being fiVe CH2 units away
from the sites in question.
Specifically, reaction of the Grignard reagent derived from
the protected 4-bromo-1-butanol derivative 34 with enantio-
merically pure (R)-propene oxide36 in the presence of catalytic
amounts of CuCl(COD) leads to efficient ring opening (Scheme
9). Protection of the resulting secondary -OH group as a benzyl
ether followed by cleavage of the TBS group of 35 with TBAF
provides alcohol 36 in good overall yield, which is converted
into the primary iodide 37 under standard conditions. This
compound affords the functionalized diorganozinc derivative
38 on prolonged exposure to neat Et2Zn and catalytic amounts
(3 mol %) of CuCN.37 Addition of this donor to dodecanal in
the presence of Ti(OiPr)4 and (R,R)-39 as the controlling ligand38
furnishes the corresponding (R)-configured alcohol 40 in 65%
yield with excellent optical purity (de g 99%). Reaction of this
(36) Enantiopure propene oxide is commercially available or can be conveniently
prepared on a large scale in optically active form using Jacobsens’s excellent
HKR method, cf.: (a) Tokunaga, M.; Larrow, J. F.; Kakiuchi, F.; Jacobsen,
E. N. Science 1997, 277, 936. (b) Furrow, M. E.; Schaus, S. E.; Jacobsen,
E. N. J. Org. Chem. 1998, 63, 6776. (c) Keith, J. M.; Larrow, J. F.;
Jacobsen, E. N. AdV. Synth. Catal. 2001, 343, 5.
(37) Rozema, M. J.; Eisenberg, C.; Lu¨tjens, H.; Ostwald, R.; Belyk, K.; Knochel,
P. Tetrahedron Lett. 1993, 34, 3115.
(38) (a) Takahashi, H.; Kawakita, T.; Ohno, M.; Yoshioka, M.; Kobayashi, S.
Tetrahedron 1992, 48, 5691. (b) Takahashi, H.; Yoshioka, M.; Shibasaki,
M.; Ohno, M.; Imai, N.; Kobayashi, S. Tetrahedron 1995, 51, 12013. (c)
Knochel, P. Synlett 1995, 393.
(39) Similar effects are visible in the 1H NMR spectra which are also distinctly
different for the pairs of stereoisomers 31/33, 42/43, 44/45. Particularly
diagnostic is the shift of the terminal methyl group of the fatty acid chain
which resonates at δ ) 1.34 ppm in the 17S-configured compounds but at
δ ) 1.31 ppm in the 17R-configured products. The pertinent shift in the
natural product is δ ) 1.34 ppm.
(40) If the fatty acids are drawn in a zigzag conformation, all hydroxyl groups
are located on the same side; the fact that the centers at C-17′ and C-19
are S-configured while the other ones are R-configured simply reflects the
formalism of the Cahn-Ingold-Prelog nomenclature.
9
13138 J. AM. CHEM. SOC. VOL. 125, NO. 43, 2003

Antiviral Evaluation of Cycloviracin B
A R T I C L E S
1
Scheme 12 a
Scheme 13 a
a Conditions: [a] TBAF, THF, 92%; [b] 1-phenyl-5-mercapto tetrazole,
PPh3, DIAD, 85%; [c] (NH4)6Mo7O24‚4H2O, aq. H2O2, EtOH, 97%; [d]
LiHMDS, DME, -78 °C, then aldehyde 58, 45%.
kinetic resolution leaving the (S)-configured lactone behind
which is essentially enantiomerically pure (ee > 95%).42 For
the purpose of our study, however, an (R)-configured building
block is required.43 Since an assay of 19 different commercially
available enzymes did not show any hit for a biocatalyst with
such a reversed preference, the hydrolysis reaction was opti-
mized for the preparation of the resulting hydroxy acid. Keeping
the pH constant at 7.2 and stopping the PLE-catalyzed reaction
after 40% conversion furnishes acid 49 with an ee ) 95% in
38% yield on a multigram scale. This compound is then
activated with carbonyl diimidazole and reacted with the
magnesium carboxylate 50 developed by Masamune et al.44,45
to give keto ester 51a (R ) H) together with small amounts of
imidazolyl carbamate 51b (R ) C(O)imidazolyl); simple workup
of the crude material with aq. NaOH in THF effectively cleaves
the carbamate group and allows for a convenient isolation of
product 51a in 74% isolated yield. Asymmetric hydrogenation
of this material in the presence of [((R)-BINAP)RuCl2]2‚NEt3
18 affords diol 52 in almost quantitative yield, which was found
to be optically pure (de ) 99%). Subsequent glycosylation with
trichloroacetimidate 53 prepared as previously described by our
a Conditions: [a] LDA, THF, -78 °C, then MeI, 84%; [b] mCPBA,
CH2Cl2, 74%; [c] porcine liver esterase, pH 7.2, 40%, ee ) 95%; [d] (i)
carbonyl diimidazole, reagent 50, THF; (ii) NaOH, THF/H2O, 74%; [e] H2
(10 atm), [((R)-BINAP)RuCl2]2‚NEt3, MeOH/THF, 70 °C, 99%, de ) 99%;
[f] trichloroacetimidate 53, TMSOTf cat., MS 4 Å, CH2Cl2, 82%; [g]
LiAlH4, THF, 82%; [h] Bu3P, o-nitrophenyl selenocyanate, THF; [i] (i)
H2O2, THF, 90%; (ii) NaH, MeI, DMF, 78%; [j] O3, CH2Cl2, -78 °C,
then Me2S, 95%.
Preparation of the Missing Building Blocks and Model
Studies on Fragment Coupling. The next interim goal en route
to cycloviracin B1 was the preparation of adequate building
blocks for its glycosylated fatty acid appendices and the
assessment of proper methodology for the attachment of these
lateral segments to the central macrodiolide ring. While the
successful application of the diorganozinc derivative 38 in the
model studies strongly recommends the use of this reagent for
the construction of the C-18′-C-24′ region of the final target,
a suitable surrogate for the other terminus ranging from C-18
through C-26 remains to be developed. The chosen approach
is summarized in Scheme 12.
(42) Fouque, E.; Rousseau, G. Synthesis 1989, 661.
(43) Note again the formalism of the Chan-Ingold-Prelog nomenclature. The
use of the (2R)-configured aldehyde 58 will ultimately result in a (19S)-
configured cycloviracin derivative due to change in priorities of the
substituents at that center.
The synthesis starts with the R-methylation of cycloheptanone
4641 followed by a Baeyer-Villiger oxidation of the resulting
product 47. It is known that hydrolysis of lactone 48 thus formed
catalyzed by pig liver esterase (PLE) results in an efficient
(44) (a) Brooks, D. W.; Lu, L. D. L.; Masamune, S. Angew. Chem., Int. Ed.
Engl. 1979, 18, 72. (b) For an advanced application, see the total synthesis
of thienamycin reported by the Merck Laboratories: Salzmann, T. N.;
Ratcliffe, R. W.; Christensen, B. G.; Bouffard, F. A. J. Am. Chem. Soc.
1980, 102, 6161.
(41) (a) Liu, H.-J.; Wang, D.-X.; Kim, J. B.; Browne, E. N. C.; Wang, Y. Can.
J. Chem. 1997, 75, 899. (b) Dave, V.; Warnhoff, E. W. J. Org. Chem.
1983, 48, 2590.
(45) For a convenient preparation of the magnesium carboxylate, see: Page, P.
C. B.; Moore, J. P. G.; Mansfield, I.; McKenzie, M. J.; Bowler, W. B.;
Gallagher, J. A. Tetrahedron 2001, 57, 1837.
9
J. AM. CHEM. SOC. VOL. 125, NO. 43, 2003 13139

A R T I C L E S
Fu¨rstner et al.
Scheme 14 a
Scheme 15 a
a Conditions: [a] compound 38, Ti(OiPr)4, (S,S)-39, toluene, 87%; [b]
imidate 41, TMSOTf cat., CH2Cl2, MeCN (1:1), 87%; [c] H2 (1 atm), Pd/
C, EtOH/EtOAc (4:1), 88%.
lenocyanate in the presence of Bu3P to give the corresponding
selenide 56, which is oxidized with aq. H2O2 to the correspond-
ing selenoxide;48 this compound undergoes a rather slow but
clean elimination with formation of the corresponding alkene
in 90% yield. O-Methylation of the remaining two hydroxyl
groups on the sugar moieties under standard conditions to give
57 followed by ozonolysis of the alkene affords the desired
aldehyde 58, which constitutes a fully functional surrogate of
synthon D depicted in the retrosynthetic Scheme 2.
a Conditions: [a] 1-phenyl-5-mercapto tetrazole, PPh3, DIAD, 91%; [b]
(NH4)6Mo7O24‚4H2O, aq. H2O2, EtOH/CH2Cl2, 67%; [c] LiHMDS, DME,
-78 °C, aldehyde 58, 61% (E:Z ) 1:1); [d] H2 (1 atm), Pd/C, EtOAc,
72%; [e] TBAF, THF, 95%; [f] PCC, CH2Cl2, 83%.
As outlined in the Introduction, a successful end game of
the envisaged total synthesis critically depends on the ability
to convert the intact macrodiolide ring A into a suitable
C-nucleophile (X ) metal). Importantly, this reactive intermedi-
ate must kinetically resist a possible intramolecular attack onto
its own lactone groups as well as a destructive deprotonation
of the C-H acidic sites adjacent to those lactones. Moreover,
its reaction with aldehyde 58 must not entail any racemization
of the chiral center R to the carbonyl. Because only a highly
stabilized and weakly basic nucleophile, if any, might meet these
stringent criteria, recourse was taken to the venerable Julia
olefination49 in the powerful modification developed by Ko-
cienski.50 It was expected that the somewhat higher kinetic
group46 in the presence of TMSOTf in CH2Cl2 is not only highly
effective but also completely selective for the â-glycoside due
to the anchimeric assistance of the 2-O-acetyl group in the donor.
Interestingly, however, the inspection of the mixture shows that
the reaction initially leads to the rather selective formation of
the corresponding ortho esters, which only slowly rearrange to
the desired glycosides on prolonged reaction at -78 °C for 2
h. A similarly advanced ortho ester f â-glycoside rearrange-
ment47 was recently exploited in the final steps of our total
synthesis of woodrosin I, a macrocyclic glycolipid belonging
to the resin glycoside family.7
Further elaboration of product 54 thus formed starts with an
exhaustive LiAlH4 reduction of the three ester moieties to give
triol 55. Only its primary alcohol reacts with o-nitrophenylse-
(48) (a) Sharpless, K. B.; Young, M. W. J. Org. Chem. 1975, 40, 947. (b) Grieco,
P. A.; Gilman, S.; Nishizawa, M. J. Org. Chem. 1976, 41, 1485.
(49) (a) Julia, M.; Paris, J.-M. Tetrahedron Lett. 1973, 14, 4833. (b) Kocienski,
P. J.; Lythgoe, B.; Ruston, S. J. Chem. Soc., Perkin Trans. 1 1978, 829.
(c) Baudin, J. B.; Hareau, G.; Julia, S. A.; Ruel, O. Tetrahedron Lett. 1991,
32, 1175.
(50) (a) Blakemore, P. R.; Cole, W. J.; Kocienski, P. J.; Morley, A. Synlett
1998, 26. (b) For a recent review, see: Blakemore, P. R. J. Chem. Soc.,
Perkin Trans. 1 2002, 2563.
(46) Fu¨rstner, A.; Konetzki, I. Tetrahedron Lett. 1998, 39, 5721.
(47) (a) Kochetkov, N. K.; Khorlin, A. J.; Bochkov, A. F. Tetrahedron Lett.
1964, 5, 289. (b) Kochetkov, N. K.; Khorlin, A. J.; Bochkov, A. F.
Tetrahedron 1967, 23, 693. (c) For the use of TMSOTf as promotor in
Kochetkov-type ortho ester rearrangements, see: Ogawa, T.; Beppu, K.;
Nakabayashi, S. Carbohydr. Res. 1981, 93, C6.
9
13140 J. AM. CHEM. SOC. VOL. 125, NO. 43, 2003

Antiviral Evaluation of Cycloviracin B
A R T I C L E S
1
Table 2. Comparison of the 13C NMR Data Recorded in
stereoisomers (E:Z ) 1:1). This result shows that the ester
groups and the acidic protons adjacent to them can survive under
the chosen reaction conditions. Not surprisingly though, any
excess of the base must be strictly avoided. Moreover, it was
found that the use of either NaHMDS or KHMDS for the
deprotonation of sulfone 60 leads to significantly larger amounts
of unidentified byproducts. Although the outcome of this
experiment is not fully satisfactory in terms of yield, it suggests
that a Julia-Kocienski olefination should qualify for the delicate
assembly of cycloviracin B1 from the individual building blocks.
Completion of the Total Synthesis. Encouraged by the
results of this exploratory study, the (3R,3′R)-configured lactide
15 was desymmetrized by cleavage of both terminal silyl groups
followed by introduction of a single OTBDPS group (15 f 16
f 17, cf. Scheme 4). This protocol turned out to be more
productive than attempts to remove only one silyl ether from
15. Conversion of alcohol 17 to sulfide 62 on treatment with
commercial 1-phenyl-1H-tetrazole-5-thiol and PBu352 proceeds
with excellent yield, whereas the subsequent oxidation to the
corresponding sulfone 63 turned out to be somewhat more
difficult than expected (Scheme 14). Due to the poor solubility
of 62 in EtOH as the preferred solvent, the oxidation with H2O2
in the presence of (NH4)6Mo7O24‚4H2O53 had to be carried out
in a mixed solvent system (EtOH/CH2Cl2) which significantly
retards the reaction and occasionally leads to incomplete
conversions. Under optimized conditions, however, sulfone 63
is obtained in 67% yield.
With this compound in hand, the crucial Julia-Kocienski
olefination49-51 was investigated. Gratifyingly, addition of
aldehyde 58 to a solution of the lithio sulfone derived from 63
by deprotonation with LiHMDS in DME at -78 °C and stirring
of the mixture at that temperature for 60 min delivers the
corresponding alkene 64 in 61% yield (E:Z ∼ 1:1), which was
hydrogenated over Pd/C in EtOAc to give product 65 in order
to facilitate the analysis of the NMR spectra. We are unaware
of any precedence for Julia-type olefination reactions involving
sulfones bearing such base-labile and electrophilic â-hydroxy
ester motifs as those present in compound 63.
Pyridine-d5 for the Synthetic Sample of Cycloviracin B1 (150 MHz)
with Those Reported for the Natural Product (100 MHz).1 As Can
Be Seen, the ∆δ Is e0.1 ppm, Which Is within Experimental
Accuracy. Arbitrary Numbering Scheme as Shown in the Insert
position
1,1
2,2
3,3
19
25
26
27,25
28,26
29,27
30,28
31,29
32,30
17
22
23
24
1A
2A
3A
4A
5A
6A
2A-OMe
1B
2B
3B
4B
5B
6B
2B-OMe
1C
2C
3C
4C
5C
literature
synthetic sample
171.7
42.4
78.9
79.3
74.2
19.5
106.4
75.1
78.3
72.1
74.5
65.3
79.3
40.1
67.0
24.3
103.1
85.3
77.7
71.8
78.0
62.9
60.7
101.8
85.0
77.6
71.7
78.0
62.8
60.6
103.1
85.3
77.7
71.8
78.1
62.9
60.7
171.7
42.5
78.9
79.3
74.2
19.5
106.5
75.2
78.3
72.1
74.6
65.4
79.3
40.2
67.0
24.3
103.1
85.3
77.8
71.9
78.0
62.9
60.8
101.8
85.1
77.7
71.8
78.0
68.9
60.6
103.2
85.4
77.8
71.9
78.1
62.9
60.8
(51) See the following for leading references on the applications of Julia-
Kocienski olefination reactions to the total synthesis of complex targets:
(a) Kang, S. H.; Jeong, J. W.; Hwang, Y. S.; Lee, S. B. Angew. Chem.,
Int. Ed. 2002, 41, 1392. (b) Lee, E.; Song, H. Y.; Kang, J. W.; Kim, D.-S.;
Jung, C.-K.; Joo, J. M. J. Am. Chem. Soc. 2002, 124, 384. (c) Lee, E.;
Choi, S. J.; Kim, H.; Han, H. O.; Kim, Y. K.; Min, S. J.; Son, S. H.; Lim,
S. M.; Jang, W. S. Angew. Chem., Int. Ed. 2002, 41, 176. (d) Liu, P.;
Jacobsen, E. N. J. Am. Chem. Soc. 2001, 123, 10772. (e) Smith, A. B.;
Safonov, I. G.; Corbett, R. M. J. Am. Chem. Soc. 2001, 123, 12426. (f)
Smith, A. B.; Brandt, B. M. Org. Lett. 2001, 3, 1685. (g) Mulzer, J.; O¨ hler,
E. Angew. Chem. Int. Ed. 2001, 40, 3842. (h) Takano, D.; Nagamitsu, T.;
Ui, H.; Shiomi, K.; Yamaguchi, Y.; Masuma, R.; Kuwajima, I.; Omura, S.
Org. Lett. 2001, 3, 2289. (i) Williams, D. R.; Cortez, G. S.; Bogen, S. L.;
Rojas, C. M. Angew. Chem., Int. Ed. 2000, 39, 4612. (j) Harris, J. M.;
O’Doherty, G. A. Org. Lett. 2000, 2, 2983. (k) Williams D. R.; Clark, M.
P.; Emde, U.; Berliner, M. A. Org. Lett. 2000, 2, 3023. (l) Metternich, R.;
Denni, D.; Thai, B.; Sedrani, R. J. Org. Chem. 1999, 64, 9632. (m) Lear,
M. J.; Hirama, M. Tetrahedron Lett. 1999, 40, 4897. (n) Williams, D. R.;
Brooks, D. A.; Berliner, M. A. J. Am. Chem. Soc. 1999, 121, 4924. (o)
Smith, A. B.; Wan, Z. J. Org. Chem. 2000, 65, 3738. (p) Lautens, M.;
Colucci, J. T.; Hiebert, S.; Smith, N. D.; Bouchain, G. Org. Lett. 2002, 4,
1879. (q) Hayes, P. Y.; Kitching, W. J. Am. Chem. Soc. 2002, 124, 9718.
(r) Ahmed, A.; Hoegenauer, E. K.; Enev, V. S.; Hanbauer, M.; Kaehlig,
H.; O¨ hler, E.; Mulzer, J. J. Org. Chem. 2003, 68, 3026.
6C
2C-OMe
acidity together with the better accessibility of a terminal
tetrazolyl sulfone should allow for a selective deprotonation by
means of a sterically encumbered base without damaging the
C-H acidic groups on the lactide ring; the stabilized nature of
lithiated sulfones in general is well documented in the litera-
ture.51
To probe the viability of this concept, a model study was
carried out commencing with compound 12 (Scheme 13).
Deprotection of its terminal TBDPS-ether with TBAF in THF
is followed by conversion of the resulting primary alcohol into
sulfide 59, which is then oxidized to the required sulfone 60.50
Selective deprotonation of 60 at the sulfone site with LiHMDS
at -78 °C in DME as the best solvent followed by addition of
aldehyde 58 at that temperature furnishes the desired alkene 61
in 45% isolated yield as an inseparable mixture of both
(52) (a) Hata, T.; Sekine, M. Chem. Lett. 1974, 837. (b) Nakagawa, I.; Aki, K.;
Hata, T. J. Chem. Soc., Perkin Trans. 1 1983, 1315. (c) Fu¨rstner, A. Liebigs
Ann. Chem. 1993, 1211.
(53) Schultz, H. S.; Feyermuth, H. B.; Buc, S. R. J. Org. Chem. 1963, 28, 1140.
(54) See the following for general reviews on diorganozinc addition reactions
to aldehydes: (a) refs 18a and 38c. (b) Soai, K.; Niwa, S. Chem. ReV.
1992, 92, 833. (c) Soai, K.; Shibata, T. In ComprehensiVe Asymmetric
Catalysis, Jacobsen, E. N., Pfaltz, A., Yamamoto, H., Eds.; Springer, Berlin,
1999; Vol. 2; pp 911-922.
9
J. AM. CHEM. SOC. VOL. 125, NO. 43, 2003 13141

A R T I C L E S
Fu¨rstner et al.
Table 3. Cytotoxicity and Antiviral Activity of Various Glycolipids in HEL Cell Cultures
min virus inhibitory concb (µg‚mL-1
)
r
compd
min. cytotoxic conca (µg‚mL-1
)
HSV-1 (KOS strain)
HSV- 2 (G strain )
vaccinia virus
vesicular stomatitis virus
HSV-1 (TK- KOS ACV strain)c
1
>200
g16
>200
g400
g400
g8
24
>16
>200
>80
>80
4.8
120
>16
24
>16
120
>80
>80
4.8
>200
>16
>200
>80
>80
>8
6.4
>16
120
>80
>80
4.8
70
71
72
73
>200
>80
>80
g6.4
woodrosin (74)
a Required to cause a microscopically detectable alteration of normal cell morphology. b Required to reduce virus-induced cytopathogenicity by 50%.
c TK-: tymidine kinase deficient; ACVr: resistant to acyclovir.
Standard deprotection of the residual silyl ether in 65 followed
by oxidation of the resulting alcohol 66 with PCC affords the
rather labile aldehyde 67, which readily reacts at low temper-
ature (-50 f -20 °C) with the diorganozinc reagent 38 in the
presence of a catalyst formed in situ from Ti(OiPr)4 and the
(S,S)-configured bistriflate 39 as the controller ligand38 to give
alcohol 68 in 81% yield as a single diastereomer after flash
chromatography (Scheme 15). Beyond doubt, this transformation
constitutes one of the most advanced applications of ligand-
controlled dialkylzinc addition reactions to aldehydes known
to date and bears witness for the maturity of this method.38,54
Subsequent â-selective glucosidation of 68 with trichloro-
acetimidate 41 was promoted by TMSOTf in CH2Cl2/MeCN.
Exhaustive debenzylation of product 69 thus formed by hydro-
genolysis over Pd/C cleanly provided cycloviracin B1 1 and
completes the first total synthesis of an antiviral glycoconjugate
of this type. Not only are all analytical and spectroscopic data
in excellent agreement with those reported in the literature
(Table 2), butsmost importantlysthe pattern signature of its
NMR spectrum is perfectly superimposable to that of authentic
cycloviracin B11 recorded at the same magnetic field. Knowing
from the model studies that any diastereoisomer is clearly
discernible by high-field NMR, we therefore assign the absolute
stereochemistry of the six chiral centers residing on the fatty
acid residues as (3R,19S,25R,3′R,17′S,23′R).
Antiviral Assays. As an ultimate proof of the structural
integrity of synthetic cycloviracin B1 obtained by the route
outlined above, the cytotoxicity as well as the antiviral activity
of this compound against herpes simplex virus-1 (HSV-1, strain
KOS), HSV-1 (strain TK- KOS ACVr), HSV-2 (strain G),
vaccinia virus, and vesicular stomatitis virus have been deter-
mined. Moreover, a few selected intermediates obtained en route
to 1 were fully deprotected by standard hydrogenolysis, and
the resulting compounds 70-73 representing various subunits
of the natural product were subjected to the same antiviral
assays. Moreover, the complex glycolipid woodrosin I (74)
previously prepared by our group was also screened for antiviral
are compiled in Table 3. Compound 1 has also been evaluateds
but found inactive at subtoxic concentrationssagainst Coxsackie
B4 virus, respiratory syncytial virus, parainfluenza type 3 virus,
reovirus type 1, Sindbis virus, and Punta Toro virus.
activity.
Acknowledgment. Generous financial support by the Deut-
sche Forschungsgemeinschaft (Leibniz award to A.F.), the
Alexander-von-Humboldt Foundation (stipend to J.M.), the
Fonds der Chemischen Industrie, and the Arthur C. Cope Funds
While these data confirm a specific antiviral activity for 1
(i.e., the minimal antivirally effective concentration is g5-fold
lower than the minimal cytotoxic contentration), none of the
derivatives was found to exhibit similar effects. This also
administered by the ACS is gratefully acknowledged.
includes compound 70 representing the intact macrodiolide core
of 1 which is rather cytotoxic and not antivirally active at
subtoxic concentrations. HSV-1 (strain TK- KOS ACVr) shows
the highest sensitivity against 1 in an assay using cultured human
embryonic lung (HEL) cells, with a minimum inhibitory
concentration of about 6.4 µg‚mL-1. While woodrosin I is
effective at similar concentrations, its cytotoxicity is much more
pronounced than that of cycloviracin B1. Representative data
Supporting Information Available: Full experimental details
together with the analytical and spectroscopic data of all new
compounds. This material is available free of charge via the
Internet at http://pubs.acs.org. See any current masthead page
for ordering information and Web access instructions.
JA036521E
9
13142 J. AM. CHEM. SOC. VOL. 125, NO. 43, 2003