New efficient synthesis of resorcinylic macrolides via ynolides: Establishment of cycloproparadicicol as synthetically feasible preclinical anticancer agent based on Hsp90 as the target

Article abstract of DOI:10.1021/ja0484348

A program currently ongoing in our laboratory envisions natural macrolide radicicol-based inhibitors targeting the molecular chaperone Hsp90. Such inhibitors can be potential anticancer agents due to their ability to induce the breakdown of a variety of oncogenic proteins. In this account, we first concern ourselves with a vastly important total synthesis of such an inhibitor. We accomplished this via a new approach, which we term the "ynolide method", directed to the synthesis of resorcinylic macrolides, including cycloproparadicicol and aigialomycin D. The key features of the syntheses involve cobalt-complexation-promoted ring-closing metathesis (RCM) to generate ynolides, followed by Diels-Alder reaction with dimedone-derived bis-siloxy dienes to elaborate the benzo system. A number of interesting analogues were synthesized using this protocol. They were evaluated for their inhibitory activity against the growth of breast cancer cell line, MCF-7. The potency of their cytotoxicity was found to be consistent with their ability to degrade the oncogenic protein, Her2. From these assays, cycloproparadicicol was identified as a most promising candidate for further development.

Full text of DOI:10.1021/ja0484348

Published on Web 06/05/2004
New Efficient Synthesis of Resorcinylic Macrolides via
Ynolides: Establishment of Cycloproparadicicol as
Synthetically Feasible Preclinical Anticancer Agent Based on
Hsp90 as the Target
Zhi-Qiang Yang,† Xudong Geng,† David Solit,‡ Christine A. Pratilas,‡
Neal Rosen,‡ and Samuel J. Danishefsky*,†,§
Contribution from the Laboratory for Bioorganic Chemistry, Sloan-Kettering Institute for Cancer
Research, 1275 York AVenue, New York, New York 10021; Program in Cell Biology and
Department of Medicine, Sloan-Kettering Institute for Cancer Research, 1275 York AVenue,
New York, New York 10021; and Department of Chemistry, Columbia UniVersity,
HaVemeyer Hall, 3000 Broadway, New York, New York 10027
Received March 18, 2004; E-mail: s-danishefsky@ski.mskcc.org
Abstract: A program currently ongoing in our laboratory envisions natural macrolide radicicol-based inhibitors
targeting the molecular chaperone Hsp90. Such inhibitors can be potential anticancer agents due to their
ability to induce the breakdown of a variety of oncogenic proteins. In this account, we first concern ourselves
with a vastly important total synthesis of such an inhibitor. We accomplished this via a new approach,
which we term the “ynolide method”, directed to the synthesis of resorcinylic macrolides, including
cycloproparadicicol and aigialomycin D. The key features of the syntheses involve cobalt-complexation-
promoted ring-closing metathesis (RCM) to generate ynolides, followed by Diels-Alder reaction with
dimedone-derived bis-siloxy dienes to elaborate the benzo system. A number of interesting analogues
were synthesized using this protocol. They were evaluated for their inhibitory activity against the growth of
breast cancer cell line, MCF-7. The potency of their cytotoxicity was found to be consistent with their ability
to degrade the oncogenic protein, Her2. From these assays, cycloproparadicicol was identified as a most
promising candidate for further development.
Introduction
The heat shock protein 90 (Hsp90) is a molecular chaperone
that mediates the stabilization and folding of various oncogenic
proteins, such as Raf1 and Her2.1 Recently, Hsp90 has attracted
extensive attention as a novel, potential antitumor target. The
natural product family which has attracted by far the greatest
attention as a potential inhibitor of the action of Hsp90 are
congeners of geldanamycin (3).2-4 Indeed, a derivative of
geldanamycin, 17AAG (4), is the most advanced of drug
candidates based on Hsp90 and is currently in phase II clinical
trials (Figure 1).1,5
Figure 1. Structures of Hsp90 inhibitors.
natural product targeting Hsp90, i.e., radicicol. We postulated
that radicicol-derived drug candidates could in the long run out
perform “geldanamycinoids” in that the latter would carry
potential liabilities from the ansa bridged quinone substructure.
Indeed, in a potentially important comparison, radicicol is
significantly less hepatotoxic than 17AAG.9 However, radicicol
was found to be ineffective in vivo animal models. We theorized
Though our laboratory has been involved in attempting to
build upon 17AAG as a discovery lead centering around
Hsp90,6-8 we hoped to evaluate the potentialities of another
† Laboratory for Bioorganic Chemistry, Sloan-Kettering Institute for
Cancer Research.
‡ Program in Cell Biology and Department of Medicine, Sloan-Kettering
Institute for Cancer Research.
§ Department of Chemistry, Columbia University.
(1) Banerji, U.; Judson, I.; Workman, P. Curr. Cancer Drug Targets 2003, 3,
385-390.
(6) Kuduk, S. D.; Harris, C. R.; Zheng, F. F.; Sepp-Lorenzino, L.; Ouerfelli,
O.; Rosen, N.; Danishefsky, S. J. Bioorg. Med. Chem. Lett. 2000, 10, 1303-
1306.
(7) Zheng, F. F.; Kuduk, S. D.; Chiosis, G.; Munster, P. N.; Sepp-Lorenzino,
L.; Danishefsky, S. J.; Rosen, N. Cancer Res. 2000, 60, 2090-2094.
(8) Kuduk, S. D.; Zheng, F. F.; Sepp-Lorenzino, L.; Rosen, N.; Danishefsky,
S. J. Bioorg. Med. Chem. Lett. 1999, 9, 1233-1238.
(9) Chiosis, G.; Lucas, B.; Huezo, H.; Solit, D.; Basso, A.; Rosen, N. Curr.
Cancer Drug Targets 2003, 3, 371-376.
(2) Delmotte, P.; Delmotte-Plaquee, J. Nature 1953, 171, 344.
(3) Ayer, W. A.; Lee, S. P.; Tsuneda, A.; Hiratsuka, Y. Can. J. Microbiol.
1980, 26, 766-773.
(4) Roe, S. M.; Prodromou, C.; O’Brien, R.; Ladbury, J. E.; Piper, P. W.; Pearl,
L. H. J. Med. Chem. 1999, 42, 260-266.
(5) DeBoer, C.; Meulman, P. A.; Wnuk, R. J.; Peterson, D. H. J. Antibiot.
1970, 23, 442-447.
9
10.1021/ja0484348 CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 7881-7889
7881

A R T I C L E S
Yang et al.
Scheme 1. Our First Total Synthesis of Cycloproparadicicol
Scheme 2. New Ynolide Approach for Cycloproparadicicol
that the epoxide both in radicicol and in radicicol oxime (which
is active in in vivo models10) could well be a source of
nonspecific cytotoxicity, which could narrow the exploitable
margin of therapeutic index. Furthermore, the potential chemical
vulnerability of the dienyl epoxide raised concerns about drug
shelf stability as well as pharmacokinetics. With a view to
molecular editing, of the oxido function in a setting of minimal
conformational perturbation of the radicicol lead, we were drawn
to analogue 2 in which the epoxide linkage is replaced by a
cyclopropane.
based on early investigations of biological profiles, cyclopro-
paradicicol now emerges as a candidate for full scale preclinical
evaluation.
We had previously shown that compound 2 has an in vitro
biological profile comparable to that of 1.11 Moreover, we
brought to bear an important line of evidence that 2 and 1 were
closely related in their interactions with their biotargets. Thus
changes in peripheral stereogenic centers in both 1 and 2 bring
about the same consequences in biological function. In other
words, both structures as shown are optimized from a stereo-
chemical perspective. Moreover, and of considerable potential
advantage for radicicol-based inhibitors, they also display
cytotoxicity against Rb (retinoblastoma)-negative cells known
to be resistant to 17AAG.11
These promising findings inevitably raised issues as to the
availability of cycloproparadicicol. Clearly, our first synthesis
of this compound,11 while touching on several issues of
academic interest in organic chemistry, did not seem promising
for producing more than token amounts of the now interesting
2.
As seen, the first synthesis from our lab relied on the
appropriate sequenced Mitsunobu esterification, dithiane alky-
lation, and ring closing olefin metathesis. While highly con-
vergent and concise, this first generation pathway to radicicol
suffered from several low yielding steps which did not improve
following attempted optimization. In particular, the low yields
associated with the dithiane alkylation and ring-closing metath-
esis (RCM) steps sharply curtailed access to cycloproparadicicol
for evaluation.11
Indeed, a new strategy has been formulated and reduced to
practice, resulting in a much improved second generation total
synthesis of cycloproparadicicol.12 In addition, the new route
shows promise of applicability to a broad range of resorcinylic
macrolides.13 Herein, we disclose the full details of this new
approach and its application to reach cycloproparadicicol as well
as aigalomycin D. The synthesis and biological activities of a
number of interesting analogues of cycloproparadicicol will be
described. Based on the much improved route of synthesis, and
Results and Discussions
Overall Strategy. The defining element of our second
generation strategy was the building of the aromatic sector of
the resorcinylic marcrolide by Diels-Alder reaction of a new
type of dienophile, i.e., an “ynolide”. This cycloaddition route
to the benzo-fused macrolactone represents a substantial de-
parture from the usual mode of synthesis in which one starts
with an aromatic ring and appends to it suitable arms to close
the macrolactone ring (cf. Schemes 1 and 2). We hoped that
the new ynolide approach, if successful, would be highly
convergent and would allow for rapid access to a broad family
of resorcinylic macrolides. Since it has been our experience that
monoactivated acetylenic dienophiles are surprisingly weakly
reactive in Diels-Alder reactions,14 the success of a projected
Diels-Alder cycloaddition aromatization sequence could not
have been anticipated with confidence.
Synthesis of Acyclic Alkynoic Ester. To facilitate progress
to the key issues of the plan, we sought to develop an efficient
synthesis route to reach a seco-alkynoate ester, which, following
RCM, would afford the ynolide dienophile. In practice, the
synthesis commenced with commercial 2,4-hexadienal (sorbal-
dehyde, 6, Scheme 3). Reformatsky-like condensation of pro-
pargyl bromide (5) with 6 led to the expected carbinol.
Following â-esterification, alkyne precursor 7 was in hand in
good yield.15,16 Treatment of 7 with n-butyllithium generated
an alkynide ion, which was carboxylated with dry ice, to provide
acid 8, in very high yield.17 Following reaction of racemic 8
and the known optically pure and defined alcohol 918 under
Mitsunobu conditions,19 ester 10 was obtained as a diastereo-
meric mixture at the future C2′.
(14) Danishefsky, S.; Etheredge, S. J. J. Org. Chem. 1979, 44, 4716-4717.
(15) Friedrich, L. E.; De Vera, N.; Hamilton, M. Synth. Commun. 1980, 10,
637-643.
(16) Barrett, A. G. M.; Pena, M.; Willardsen, J. A. J. Org. Chem. 1996, 61,
1082-1100.
(10) Agatsuma, T.; Ogawa, H.; Akasaka, K.; Asai, A.; Yamashita, Y.; Mizukami,
T.; Akinaga, S.; Saitoh, Y. Bioorg. Med. Chem. 2002, 10, 3445-3454.
(11) Yamamoto, K.; Garbaccio, R. M.; Stachel, S. J.; Solit, D. B.; Chiosis, G.;
Rosen, N.; Danishefsky, S. J. Angew. Chem., Int. Ed. 2003, 42, 1280-
1284.
(12) Yang, Z.-Q.; Danishefsky, S. J. J. Am. Chem. Soc. 2003, 125, 9602-9603.
(13) Geng, X.; Danishefsky, S. J. Org. Lett. 2004, 6, 413-416.
(17) Fuganti, C.; Pedrocchi-Fantoni, G.; Sarra, A.; Servi, S. Tetrahedron:
Asymmetry 1994, 5, 1135-1138.
(18) Garbaccio, R. M.; Stachel, S. J.; Baeschlin, D. K.; Danishefsky, S. J. J.
Am. Chem. Soc. 2001, 123, 10903-10908.
(19) Romo, D.; Rzasa, R. M.; Shea, H. A.; Park, K.; Langenhan, J. M.; Sun,
L.; Akhiezer, A.; Liu, J. O. J. Am. Chem. Soc. 1998, 120, 12237-12254.
9
7882 J. AM. CHEM. SOC. VOL. 126, NO. 25, 2004

Synthesis of Resorcinylic Macrolides via Ynolides
A R T I C L E S
Scheme 3. Synthesis of the Acyclic Alkynoic Ester
Scheme 4. Attempted Ring-Closing Metathesis Reactions of 10
stable complexes, in which the alkyne functions are, in effect,
protected.24,25 Moreover, the geometry of cobalt-complexed
alkynes optimizes at approximately 140°.26 Such a departure
from linearity could well favor cyclization.27 We thus hoped
that cobalt protection of the alkyne could serve to our advantage
in the problematic macrocyclization.
To probe the feasibility of this hypothesis, model ester 14
was first treated with cobalt dicarbonyl (Scheme 6). Complex
formation proceeded readily at room temperature in toluene to
provide 15 (in 87% yield) after silica gel chromatography. When
complex 15 was subjected to the conditions of ring closing olefin
metathesis, with catalyst, cyclization proceeded smoothly at
room temperature. The cobalt-protected ynolide 16 was obtained
in 71% using 25 mol % of 11 and high dilution (0.2 mM) in
methylene chloride.
To retrieve the free acetylene of the ynolide, the cobalt
complex had to be “demetalated.” This goal was readily
achieved by treatment of 16 with excess ceric ammonium nitrate
in acetone at low temperature.28 Ynolide 17 was obtained in
excellent yield (Scheme 6).
Macrocyclization via Ring Closing Olefin Metathesis. The
next key reaction was the formation of our hoped macrocyclic
ynolide through ring-closing olefin metathesis. For this purpose,
we made recourse to the powerful metalo carbene catalysts,
innovated by Grubbs and associates, based on ligand 11.20
However, as shown in Scheme 4, all efforts to form the desired
ynolide 12 using catalytic amounts of 11 provided the recovered
starting ester 10. The Lewis acid, Ti(Oi-Pr)4, was used in the
hope of preventing formation of an inactive complexing locus
which included the ester group. However, this attempt at reaction
steering was also not successful. Attempted increase of the
amounts of 11 to stoicheometric levels eventually led to
decomposition of 10.
The failure of probe structure 10 to cyclize could be ascribed
to conformational rigidities associated with the trans-disubsti-
tuted cyclopropane and the linear acetylene “linker”. Accord-
ingly, a more flexible model ester 14 was prepared (in 59%
yield) using standard protocols. Compound 14 was subjected
to RCM conditions (Scheme 5). Unfortunately, no formation
of cyclic product was observed.
Cobalt Complexation to Promote RCM. At this stage, we
reasoned that the failure of olefin metathesis might be due to
the presence of the acetylene group. Aside from the constraint
to cyclization imposed by its linear character, the cyclization
could further be complicated by nonproductive coordination of
the acetylene to the RCM catalytic machinery. The latter
possibility could not be dismissed out of hand, since the enyne
metathesis reactions are well established.21-23 Accordingly,
protection of the alkyne could well be helpful. It is well-known
that reactions of dicobalt carbonyl with acetylenes can lead to
We next applied this strategy to the target system 10.
Gratifyingly, as was the case with the model compound 15,
RCM reaction of cobalt-complexed 18 proceeded smoothly at
room temperature in CH2Cl2 to afford cyclic product 19 in 57%
yield. Postmetathesis removal of the cobalt, however, was
nontrivial. Treatment of 19 with CAN in acetone led primarily
29
to decomposition. Other oxidizing agents such as I2 or Me3-
NO30 improved the yields significantly but were not fully
reproducible when conducted in large scales. Eventually, it was
found that when the solution of 19 was buffered with 2,6-di-
tert-butyl pyridine prior to the treatment with CAN, the desired
product 12 was isolated in 50% yield in two steps from 18
(Scheme 7).31 At this stage, the two stereoisomers of 12 became
separable, and they were advanced individually.
Cycloaddition-Aromatization Reactions. Having achieved
an efficient synthesis of the required ynolide, we directed our
attentions to the next challenge, i.e., the fashioning of the desired
resorcinylic macrolides using a Diels-Alder elimination se-
quence. We appreciated that acetylenic dienophiles, where only
one of the sp carbons is activated with a typical activating group
such as an ester can be rather unreactive.14 These monoactivated
acetylenes must be clearly distinguished at the planning level
(24) Greenfield, H.; Sternberg, H. W.; Friedel, R. A.; Wotiz, J. H.; Markby, R.;
Wender, I. J. Am. Chem. Soc. 1956, 78, 120-124.
(25) Nicholas, K. M.; Pettit, R. Tetrahedron Lett. 1971, 37, 3475-3478.
(26) Dickson, R. S.; Fraser, P. J. AdV. Organomet. Chem. 1974, 12, 323-377.
(27) Recently, a similar protocol was reported: Young, D. G. J.; Burlison, J.
A.; Peters, U. J. Org. Chem. 2003, 68, 3494-3497.
(28) Magnus, P. Tetrahedron 1994, 50, 1397-1418.
(29) Tanaka, S.; Tsukiyama, T.; Isobe, M. Tetrahedron Lett. 1993, 34, 5757-
5760.
(20) Scholl, M.; Ding, S.; Lee, C. W.; Grubbs, R. H. Org. Lett. 1999, 1, 953-
956.
(30) Jones, G. B.; Wright, J. M.; Rush, T. M.; Plourde, G. W., II.; Kelton, T.
F.; Mathews, J. E.; Huber, R. S.; Davidson, J. P. J. Org. Chem. 1997, 62,
9379-9381.
(21) Kinoshita, A.; Mori, M. Synlett 1994, 1020-1022.
(22) Kinoshita, A.; Mori, M. J. Org. Chem. 1996, 61, 8356-8357.
(23) Mori, M.; Sakakibara, N.; Kinoshita, A. J. Org. Chem. 1998, 63, 6082-
6083.
(31) Magnus, P.; Eisenbeis, S. A.; Fairhurst, R. A.; Iliadis, T.; Magnus, N. A.;
Parry, D. J. Am. Chem. Soc. 1997, 119, 5591-5605.
9
J. AM. CHEM. SOC. VOL. 126, NO. 25, 2004 7883

A R T I C L E S
Yang et al.
Scheme 5. Olefin Metathesis Reaction of Model Ester 14
Scheme 6. RCM of Cobalt-Protected Model Ester
Scheme 8. Attempted Diels-Alder Reactions with Acyclic Dienes
Scheme 7. Synthesis of Ynolide 12
Scheme 9. Diels-Alder Reaction of 12 with Cyclic Diene 25
and diene 21 were heated to 160 °C, no products were isolated.
The ynolide 12 had decomposed. In the hope of lowering the
temperature required for the cycloaddition, we turned to the use
of Lewis acids. Considering the extreme acidic liability of these
dienes, we chose mild reagents Eufod35-37 in the context of
projected Diels-Alder reactions of model ynolide 17 with diene
21. Unfortunately, no product formation was observed.
We next directed our attentions to the use of the dimedone-
derived cyclic diene 2538,39 (Scheme 9). The latter could, in
principle, react with ynolide diastereomers 12 to afford desired
product 22 after elimination of isobutylene.40,41 The possible
advantages of this diene over tri-oxygen-substituted acyclic
possibilities, such as 20 or 21, are that the cyclic diene could
from diactivated acetylenes such as diesters of acetylenedicar-
boxylic acid which are powerful dienophiles.
At first, trioxygen-substituted dienes of the type 2032 and
2114,33,34 seemed to be appropriate choices to attempt cycload-
dition with ynolides 12. Ideally, reaction of 12 with diene 20
followed by elimination of MeOH would, upon desilylation,
lead to the desired free phenol product 22. The use of diene 21
would require an additional step to deprotect the phenolic methyl
ether of the product 23 (Scheme 8). Such deprotections had
been found to be nontrivial in related systems.18
To our surprise, when 12 and diene 20 were heated to 70
°C, no aromatic products were isolated. Desilylated 12 was
recovered. Since 20 is prone to undergo a 1,5-silyl shift at
temperatures above 70 °C,26 we directed our efforts to the
thermally more stable diene 21. Unfortunately, even when 12
(35) Danishefsky, S.; Harvey, D. F.; Quallich, G.; Uang, B. J. J. Org. Chem.
1984, 49, 392-393.
(36) Castellino, S.; Sims, J. J. Tetrahedron Lett. 1984, 25, 2307-2310.
(37) Lopez, R.; Carretero, J. C. Tetrahedron: Asymmetry 1991, 2, 93-96.
(38) Ibuka, T.; Mori, Y.; Aoyama, T.; Inubushi, Y. Chem. Pharm. Bull. 1978,
26, 456-465.
(32) Yamamoto, K.; Suzuki, S.; Tsuji, J. Chem. Lett. 1978, 649-652.
(33) Banville, J.; Brassard, P. J. Chem. Soc., Perkin Trans. I 1976, 1852-1856.
(34) Danishefsky, S.; Singh, R. K.; Gammill, R. B. J. Org. Chem. 1978, 43,
379-380.
(39) Langer, P.; Schneider, T.; Stoll, M. Chem.sEur. J. 2000, 6, 3204-3214.
9
7884 J. AM. CHEM. SOC. VOL. 126, NO. 25, 2004

Synthesis of Resorcinylic Macrolides via Ynolides
A R T I C L E S
Scheme 10. Oxidation in the Presence of Free Phenols
major isomer of 25 (80% yield from the minor isomer). The
stereochemistry of two isomers of 25 could be determined by
a modified Mosher’s analyses (see Supporting Information for
details).49 The acetate groups were cleaved under mildly basic
conditions (5% NaHCO3/MeOH, 1:1) in excellent yield.18
Finally, regioselective chlorination provided cycloproparadicicol
(2) in 70% yield. Unfortunately, this product was accompanied
by the formation of an isomeric chlorination product 26 in 27%
yield.18 Although the issues governing such chlorinations have
not necessarily been optimized, this level of regioselectivity
seems to be quite general for such resorcinylic compounds.50,51
In summary, we have developed a new concise synthesis for
cycloproparadicicol, in 6% yield following 13 steps.
Synthesis of Analogues and Formulation of a Preliminary
SAR Pattern in the Cycloproparadicicol Series. The new
synthesis developed here not only allowed us to generate
sufficient quantities of cycloproparadicicol to support our
biological studies on the compound itself but also allowed us
to fashion a series of interesting analogues to study structure-
activity relationships. To evaluate the biological impact of
modification of the ketone group, two alcohol isomers 27R,â
were prepared (Scheme 12). Using SO2Cl2, the chloride deriva-
tive 28R was synthesized in one step from the major isomer,
27R. The methyl ether derivative of 27R, 29R was also prepared
under neutral conditions (MeI/Ag2O).52 Oxime analogues of
cycloproparadicicol [(Z)- and (E)-30] were synthesized as well.10
well be more reactive (due to a locking in of the s-syn
conformation by the six-member ring) and more thermally
stable. Happily, when diene 25 and dienophile 12 were heated
to 160 °C, followed by desilylation, in the course of silica gel
chromatography, the desired product, 22, was obtained in 78%
yield. Thus a new, highly efficient and convergent method to
assemble the resorcinylic macrolide 22, needed for cyclo-
proparadicicol, had been achieved.
Completion of the Synthesis. To complete the synthesis of
cycloproparadicicol (2), two remaining steps were required.
Oxidation of the 2° alcohols to the corresponding ketone and
regioselectve introduction of a chlorine had to be accomplished
in some as yet unspecified order. We first attempted to carry
out the oxidation of the secondary homobenzylic alcohol in the
presence of the free phenolic hydroxyls (Scheme 10). Following
the removal of the TBS ether group (using HF/pyridine) in 74%
yield, we surveyed a variety of oxidation conditions. Unfortu-
nately, in the best case, we obtained only 30% yield through
the use of activated MnO2.42,43 The use of other oxidizing agents
examined resulted in the decomposition of starting material,
22.44-48
Synthesis of Difulorocycloproparadicicol. Our laboratory
recently discovered that a trifluoro-derivative of epothilone
exhibits a far superior in vivo profile.53 Accordingly, we sought
to synthesize difluorocycloproparadicicol 31, hoping to identify
even better druglike molecules targeting Hsp90.
The difluorocyclopropane group was introduced to the known
conjugated ester 32, using reagents developed by Dolbier and
co-workers, thereby affording difluoroesters 33 as a 1:1 mixture
of stereoisomers (Scheme 13).54 Our goal here was to achieve
the target rapidly using the methodology already described in
this account. Hence, the mixture of stereoisomers was carried
forward through the required synthetic steps. As outlined in
Scheme 13, the key ynolide 36 was prepared in reasonable yields
following the now familiar protocols described above. The
subsequent Diels-Alder reaction with diene 25, however,
proved to be very difficult, affording only 18% of product.
Increasing the reaction temperature further eroded the yield,
presumably due to the thermal instability of the difluorocyclo-
propane-containing ynolide 35. Nevertheless, using the post-
Accordingly, we were obliged to protect the two free phenolic
hydroxyl groups. Treatment of 22 with acetic anhydride in DMF
and cat. amount of N,N-dimethylamino pyridine (DMAP)
provided diacetate 24 (87% from the major isomer of 22 and
76% from the minor isomer of 22) (Scheme 11). Subsequent
removal of its TBS ether gave alcohol 25. This deprotection
was followed by oxidation using Dess-Martin periodinane, to
afford the desired ketone intermediate in 68% yield from the
Scheme 11. Completion of the Synthesis of 2
9
J. AM. CHEM. SOC. VOL. 126, NO. 25, 2004 7885

A R T I C L E S
Yang et al.
Scheme 12. Synthesis of Cycloproparadicicol Analogues
set out to synthesize the lactam version of the cycloproparadici-
col (see compound 39).
To introduce the functionality required to reach 39, a sequence
involving Staudinger reduction of an azide was considered.55
However, Mitsunobu type methodology,56 including Thomp-
son’s protocol,57 failed to deliver the corresponding azide of
compound 9. This is probably due to the instability of the azide
compound. Fortunately, it was found that the Nosyl approach
recently developed by Fukuyama58 afforded the desired sul-
fonamide 40 in 61% yield (Scheme 14). After the removal of
the sulfonyl group, the resulting amine, generated in situ,
coupled with alkynoic acid 8 to afford amide 41. Application
of the protocol described above (cobalt complexation f RCM
f oxidative decomplexation) to the case at hand furnished
“ynelactam” 44. The Diels-Alder reaction of 44 with diene
25, however, proved to be highly problematic. The best yield
realized for reaching product 45 was only around 25%, and that
proved to be difficultly reproducible. It is likely that replacement
of the ester group by an amide function attenuates the elec-
tronegative pull on the acetylene, thereby eroding the reactivity
of this linkage as a dienophile. Currently, investigations of other
types of more reactive dienes are actively underway in our
laboratory to deal with the problem of poorly reactive acetylenic
dienophiles. Success on this front would allow us to operate
more effectively with weakly reactive dienophiles such as 44.
Pending solution of the synthetic problem, studies directed to
the synthesis of 39 have been deferred.
Diels-Alder sequence described previously, adequate amounts
of 31 and 38 were obtained to support in vitro studies. It will
be noted that, after Dess-Martin oxidation, ketone 38 was
isolated as a single isomer. Seemingly, this difluoroyclopropyl
isomer of 38 appears to have the same cyclopropane centers
(7′R, 8′R) as does cycloproparadicicol itself, based on 1H NMR
and optical rotation comparisons (see Supporting Information
for details). Presumably, the alternate trans cyclopropyl dias-
tereomer fares poorly in terms of yield in going through the
steps from 35 to the end.
Synthetic Approach to the Cycloproparadicicol Lactam.
One of the structural features of radicicol that might have
contributed to its instability in vivo, in addition to the 7′-8′
epoxide, is the macrolactone. The latter could be subject to
metabolic hydrolysis by esterases. With this in mind, we also
Extension of the Ynolide Approach to the First Total
Synthesis of Aigialomycin D. Nature has provided a variety
of biologically significant natural products which share the same
structural motif as radicicol, i.e., a 14-member resorcinylic
macrolide. We sought to test the scope and limitations of our
newly developed ynolide-DA approach to such ring systems
by applying it to an interesting member of this family,
agialomycin D (46) (Figure 2). Recently, this compound was
isolated from the marine mangrove fungus Aigialus parVus
BCC5311.59 It exhibited potent antimalarial activity (IC50: 6.6
µg/mL against P. falciparum) and antitumor activity (IC50: 3.0
µg/mL against KB cells).59 We saw agialomycin D as an
attractive candidate expanding upon our new “ynolide” strategy.
An important issue to be overcome would be the proper
emplacement of the E 1′, 2′ double bond requested to reach
Scheme 13. Synthesis of Difluorocycloproparadicicol
9
7886 J. AM. CHEM. SOC. VOL. 126, NO. 25, 2004

Synthesis of Resorcinylic Macrolides via Ynolides
A R T I C L E S
Scheme 14. Synthesis of Cycloproparadicicol Lactam
Scheme 15. Synthesis of Alkyne 56
aigialomycin D. Also, the route must accommodate the hydroxy
bearing stereogenic centers at carbons 5′ and 6′.
furnish key aldehyde 52, as shown in Scheme 15. This
compound was subjected to nucleophilic propargylation fol-
lowed by protection of the resultant alcohol to afford TBS ether
53. The pivaloyl group of 53 was then removed, and a vinyl
(43) Trost, B. M.; Caldwell, C. G.; Murayama, E.; Heissler, D. J. Org. Chem.
1983, 48, 3252-3265.
(44) Huang, S. L.; Omura, K.; Swern, D. J. Org. Chem. 1976, 41, 3329-3231.
(45) Hauser, F. M.; Prasanna, S.; Combs, D. W. J. Org. Chem. 1983, 48, 1328-
1333.
(46) Welch, S. C.; Levine, J. A.; Arimilli, M. N. Synth. Commun. 1993, 23,
131-134.
(47) Tatsuta, K.; Takano, S.; Sato, T.; Nakano, S. Chem. Lett. 2001, 172-173.
(48) Ley, S. V.; Norman, J.; Griffith, W. P.; Marsden, S. P. Synthesis 1994,
639-666.
Figure 2. Structure of aigialomycin D.
(49) Ohtani, I.; Kusumi, T.; Kashman, Y.; Kakisawa, H. J. Org. Chem. 1991,
56, 1296-1298.
As shown in Scheme 15, the total synthesis commenced from
the naturally derived D-2-deoxy-ribose 47, which already has
the desired future syn 5′,6′-diol functionality of aigialomycin
D in place. Compound 47 was processed forward via diol
protection, Wittig olefination, hydroboration, and oxidation, to
(50) Elix, J. A.; Barclay, C. E.; Lumbsch, H. T.; Wardlaw, J. H. Aust. J. Chem.
1997, 50, 971-975.
(51) Elix, J. A.; Crook, C. E.; Hui, J.; Zhu, Z. N. Aust. J. Chem. 1992, 45,
845-855.
(52) Tanis, S. P.; Robinson, E. D.; McMills, M. C.; Watt, W. J. Am. Chem.
Soc. 1992, 114, 8349-8362.
(53) Chou, T.-C.; Dong, H.; Rivkin, A.; Yoshimura, F.; Gabarda, A. E.; Cho,
Y. S.; Tong, W. P.; Danishefsky, S. J. Angew. Chem., Int. Ed. 2003, 42,
4762-4767.
(40) Uchiyama, M.; Kimura, Y.; Ohta, A. Tetrahedron Lett. 2000, 41, 10013-
10017.
(41) Morrison, C. F.; Burnell, D. J. Tetrahedron Lett. 2001, 42, 7367-7369.
(54) Tian, F.; Kruger, V.; Bautista, O.; Duan, J.-X.; Li, A.-R.; Dolbier, W. R.,
Jr.; Chen, Q.-Y. Org. Lett. 2000, 2, 563-564.
(42) Gritter, R. J.; Wallace, T. J. J. Org. Chem. 1959, 24, 1051-1056.
9
J. AM. CHEM. SOC. VOL. 126, NO. 25, 2004 7887

A R T I C L E S
Yang et al.
Scheme 16. Completion of the Synthesis of Aigialomycin D
Table 1. IC50 (nM) Values of Cycloproparadicicol Analogues and Aigialomycin D
compound
2
26
27r
27â
28R
29R
(Z)-30
(E)-30
31
38
46
IC50
54
>500
150
>500
390
>10 000
98
282
10 000
3000
>10 000
group was installed through oxidation and Wittig reactions (see
compound 56).
reaction was carried out using the Martin sulfurane reagent.60
Finally, global deprotection under acidic conditions completed
the total synthesis of aigialomycin D 46. The physical data (1H
and 13C NMR, Mass spectrum, optical rotation, and IR)59 in
the context of the synthetic sequence and earlier spectral data
serve to establish, independently, the structure of synthetic
agialomycin D as shown. Moreover, the data for the final
synthetic material are fully consistent with those reported for
the natural product.59
Following chemistry of the type described above, compound
56 was advanced to ester 58 through carboxylation of the alkyne
function and Mitsunobu esterification using the commercially
available (2R)-4-penten-2-ol. We again investigated the pos-
sibility of closure on 58, itself containing the free alkyne moiety
as attempted above (see 10 and 14). Again, starting material
(58) was recovered following attempted RCM. Fortunately, we
found that our newly discovered protocol (cobalt complexation
followed by RCM and then decomplexation) proceeded very
efficiently delivering ynolide 61.
The two benzylic stereoisomers of 61 (2′-(R) and 2′-(S)) were
separated by silica gel chromatography, and their absolute
configurations were determined by modified Mosher’s ester
methodology.49 As now expected, Diels-Alder reaction between
62 and diene 25 proceeded with high efficiency to afford the
desired macrolide 63 in good yield. Following the protection
of the phenol hydroxyl groups as the corresponding MOM
ethers, the 2′-silyl group was removed and the dehydration
The accomplishment of the total synthesis of aigialomycin
D serves to suggest the generalizability of our newly developed
protocol. We are hoping to utilize this enhanced methodology
to discover resorcinylic derivatives, with significant biological
activity.
Biological Evaluation of Cycloproparadicicol Analogues.
As discussed earlier, it appears that the inhibition of Hsp90
induces the proteasomal degradation of various oncogenic
proteins, thereby leading to inhibition of tumor cell growth. We
hoped to evaluate our synthetic analogues for their growth
inhibition activity against MCF breast cancer cells. The IC50
values of our synthetic compounds were determined after
treatment of the cells with each drug listed in Table 1 for 72 h.
Cycloproparadicicol is the most potent of the new compounds.
Interestingly, chloride regioisomer 26 was found to be inactive.
Of the two alcohol isomers, 27R displayed an IC50 value of
(55) Stachel, S. J.; Lee, C. B.; Spassova, M.; Chappell, M. D.; Bornmann, W.
G.; Danishefsky, S. J.; Chou, T. C.; Guan, Y. J. Org. Chem. 2001, 66,
4369-4378.
(56) Mitsunobu, O. Synthesis 1981, 1-28.
(57) Thompson, A. S.; Humphrey, G. R.; DeMarco, A. M.; Mathre, D. J.;
Grabowski, E. J. J. J. Org. Chem. 1993, 58, 5886-5888.
(58) Kurosawa, W.; Kan, T.; Fukuyama, T. J. Am. Chem. Soc. 2003, 125, 8112-
8113.
(59) Isaka, M.; Suyarnsestakorn, C.; Tanticharoen, M.; Kongsaeree, P.; Thebt-
aranonth, Y. J. Org. Chem. 2002, 67, 1561-1566.
(60) Martin, J. C.; Arhart, R. J. J. Am. Chem. Soc. 1971, 93, 4327-4329.
9
7888 J. AM. CHEM. SOC. VOL. 126, NO. 25, 2004

Synthesis of Resorcinylic Macrolides via Ynolides
A R T I C L E S
Figure 3. Her2 degradation assay.
150 nM, and 27â is inactive. Further modifications of 27R,
however, decreased its potency. Surprisingly, introduction of
the chlorine function (28R) resulted in raising the IC50 to 390
nM. Compound 29a, methyl ether derivative of 27R, is
completely inactive. The (Z)-oxime, (Z)-30, was found 3 times
more potent that the corresponding (E)-isomer. Aigialomycin
D was significantly less active than the cycloproparadicicol
derivatives.
To confirm that the cytotoxicity was indeed due to the
inhibition of Hsp90, we set out to investigate the degradation
of Her2, one of the most sensitive client proteins of this
chaperone. The expression levels of Her2 of drug-treated MCF-7
cells were analyzed using immunoblotting (Figure 3). P85, used
as a control, is not a client protein of Hsp90. As expected, the
growth inhibition activities of the synthetic active analogues
are reflected in their capacity to degrade Her2. The three most
active compounds, 2, 27R, and (Z)-30, all were able to degrade
Her2 at 0.3 µM. (E)-30 is less effective. Nonchlorinated difluoro
compound 38 degraded Her2 at 1 µM, and its chlorinated
counterpart 31, at 10 µM. Aigialomycin D (46) did not degrade
the protein even at 10 µM.
It is, however, appropriate to point out that the sluggish
dienophilicity of monoactivated acetylenes is still a problem
preventing full generalizability of the method. For instance, in
the work described above, i.e., in the cases of the yne lactam
44 and the ynolide 36 containing the sensitive difluorcyclopropyl
group, the yields of the Diels-Alder step are low.
The biological activities of our synthetic analogues shown
here, in conjunction with the earlier demonstration that cyclo-
proparadicicol binds to Hsp90 at ca. 160 nM,11 support the
notion of the latter as the likely target for this new group of
anticancer agents. In vivo evaluations of cycloproparadicicol
are in progress in various settings and will be described upon
completion of full preclinical studies.
Acknowledgment. This work was supported by the National
Institutes of Health (CA-28824). Z.Q.Y. gratefully acknowledges
a postdoctoral fellowship from Mr. William H. Goodwin and
Mrs. Alice Goodwin and the Commonwealth Cancer Foundation
for Research and The Experimental Therapeutics Center of
Memorial Sloan-Kettering Cancer Center. The U.S. Army Breast
Cancer Foundation postdoctoral fellowship support is gratefully
acknowledged by X.G. (BC022120). We thank Ms. Anna
Dudkina and Ms. Sylvi Rusli (NMR Core Facility, CA-02848)
for mass spectral analyses, Dr. George Sukenic for NMR
analyses, and Dr. William Berkowitz for helpful discussions.
Conclusions and Future Directions
In summary, we have developed a new and efficient synthesis
enabling the full evaluation of cycloproparadicicol as a feasible
candidate for further advancement. The highlights of the new
synthesis are cobalt-complexation-promoted RCM to generate
ynolides, followed by Diels-Alder reaction with dimedone-
derived bis-siloxyl diene to assemble the benzofused macrolides.
Given this pathway, we can generate gram quantities of
cyloproparadicicol for biological evaluations. The generality of
the synthesis plan has been demonstrated by its application to
the first total synthesis of aigialomycin D.
Supporting Information Available: Experimental procedures
and characterization data. This material is available free of
charge via the Internet at http://pubs.acs.org.
JA0484348
9
J. AM. CHEM. SOC. VOL. 126, NO. 25, 2004 7889